Ultraviolet radiation-induced interleukin 6 release in HeLa cells is mediated via membrane events in a DNA damage-independent way.

Evidence exists that ultraviolet radiation (UV) affects molecular targets in the nucleus or at the cell membrane. UV-induced apoptosis was found to be mediated via DNA damage and activation of death receptors, suggesting that nuclear and membrane effects are not mutually exclusive. To determine whether participation of nuclear and membrane components is also essential for other UV responses, we studied the induction of interleukin-6 (IL-6) by UV. Exposing HeLa cells to UV at 4 degrees C, which inhibits activation of surface receptors, almost completely prevented IL-6 release. Enhanced repair of UV-mediated DNA damage by addition of the DNA repair enzyme photolyase did not affect UV-induced IL-6 production, suggesting that in this case membrane events predominant over nuclear effects. UV-induced IL-6 release is mediated via NFkappaB since the NFkappaB inhibitor MG132 or transfection of cells with a super-repressor form of the NFkappaB inhibitor IkappaB reduced IL-6 release. Transfection with a dominant negative mutant of the signaling protein TRAF-2 reduced IL-6 release upon exposure to UV, indicating that UV-induced IL-6 release is mediated by activation of the tumor necrosis factor receptor-1. These data demonstrate that UV can exert biological effects mainly by affecting cell surface receptors and that this is independent of its ability to induce nuclear DNA damage.

Ultraviolet radiation (UV) and, in particular, UVB with a wave length range between 290 and 320 nm represents one, if not the most, important environmental factor of inducible health hazards for mankind, which include the induction of skin cancer (1), suppression of the immune system (2), and chronic skin damage including premature skin aging (3). Similar to chemical agents, UV has the ability to alter mammalian gene expression (4 -6). Elucidation of the mechanisms by which UV affects gene expression is crucial for understanding how UV exerts its biological effects and how it develops its pathogenic properties. In this context, one of the most frequently but also controversially discussed issues is whether the cellular UV response is initiated at the cell membrane or in the nucleus (reviewed in Refs. 7 and 8).
The biological effects of UV are multiple and include the release of soluble mediators, the induction of apoptosis, and alterations of surface molecule expression, just to name a few. To exert these biological effects, UV must first be absorbed by a chromophore within the cell, which then transduces energy into a biochemical signal. A number of chromophores have been identified, e.g. porphyrins, aromatic amino acids, urocanic acid, and DNA. Among these, DNA has been regarded as the most important one, since the wavelength dependences of various UV effects match that of DNA absorption (9). In addition, removal of UV-induced DNA damage, e.g. by enhancing DNA repair, reduces or even inhibits some of the biological UV effects (10 -13). Finally, lower UV doses induce some of the biological effects at the same magnitude in DNA repair-deficient cells as in cells with normal DNA repair (14). Considering these observations, it is understandable that for quite a long time DNA was regarded as the only molecular target for UV, and why the dogma existed that any biological effect must be a direct consequence of DNA damage.
On the other hand, a variety of groups have provided convincing evidence that UV may exert biological effects without the need of a nuclear signal. Utilizing enucleated cells, Devary et al. (15) demonstrated that activation of the transcription factor nuclear factor B (NFB) 1 does not require a nuclear signal. This was confirmed by the observation that UV exposure of cytosolic extracts containing NFB in its inactive form supplemented with cellular membranes causes activation of NFB (16). In addition, growth factor receptors appear to be involved in the UV response since UV activates the epidermal growth factor receptor by directly initiating tyrosine phosphorylation (17). Rosette and Karin (18) reported for the first time that UV and osmotic shock, respectively, can activate cell surface receptors by inducing their oligomerization. Triggering of receptors in such a way by UV takes place without the binding of any ligand and independently of DNA damage (18,19).
One of the most important biological effects of UV is the induction of apoptotic cell death (20). Convincing data exist that nuclear events especially UV-induced DNA damage determine whether a cell undergoes apoptosis or not (21). On the other hand, it was clearly demonstrated that activation of death receptors such as CD95 (Fas/APO-1) on the cell surface by UV induces the apoptotic machinery without being connected with DNA damage (22,23). In addition, Sheikh et al. (19) proposed that UV-stimulated ligand independent activation of the tumor necrosis factor receptor plays a major role in mediating the apoptotic effects of UV. Considering these data, which on first glance appear conflicting, we recently tried to determine the relative contribution of nuclear and membrane effects in UV-induced apoptosis (24). Removal of UV-induced DNA damage by enhancing DNA repair using the repair enzyme photolyase significantly reduced the apoptosis rate in HeLa cells. On the other hand, exposure of HeLa cells to UV at 4°C, which prevents death receptor clustering (18,22), also reduced the apoptosis rate, although to a lesser extent. It is important to mention that neither of these strategies alone was able to completely prevent UV-mediated apoptosis. However, when both strategies were combined, i.e. when cells were exposed to UV at 4°C and DNA damage removed by photolyase, UV-induced apoptosis was completely inhibited (24). Hence, these data indicated that, although nuclear effects are predominant in comparison to membrane events, both are necessary to obtain the complete apoptotic response.
Inspired by these observations, we were interested to determine whether the participation of both nuclear and membrane components is specific for UV-induced apoptosis or is also essential for other UV responses. Here, we studied the induction of the release of the inflammatory cytokine interleukin 6 (IL-6) in HeLa cells by UV. Using this system, we demonstrate that removal of DNA damage by enhancing DNA repair does not cause reduction of IL-6 release, implying that UV-induced DNA damage is not an important intermediate in this type of UV response. On the other hand, prevention of triggering cell surface receptors by maintaining HeLa cells at 4°C during UV exposure resulted in complete inhibition of UV-mediated IL-6 secretion. Using dominant negative mutants, we provide evidence that NFB is involved in this signaling process and that the tumor necrosis factor type 1 receptor (TNF-R1) seems to be a major target at the cell membrane in this UV response. Together, these data indicate that UV can also exert its biological effects by exclusively acting on the cell membrane without the necessity of a nuclear signal. In addition, our findings suggest that the multiple biological effects of UV on mammalian cells do not only differ in their final outcome but are also dependent on how they are generated.

MATERIALS AND METHODS
Cells and Reagents-The human epithelial carcinoma cell line HeLa (American Tissue Culture Collection) was cultured in RPMI 1640 with 10% FCS. Human recombinant tumor necrosis factor ␣ (TNF␣) was obtained from Endogen (Woburn, CA). IL-6 release from cells was measured by subjecting supernatants (10 l each) to an IL-6 ELISA kit (Diaclone, Besancon, France). Measurements were performed according to the manufacturer's guidelines. The proteasome inhibitor MG132 was purchased from Calbiochem (San Diego, CA). The plasmid allowing overexpression of a mutated IB variant was kindly provided by K. Schulze-Osthoff (University of Mü nster, Mü nster, Germany) (25), the plasmid overexpressing a dominant negative mutant of TRAF-2 was kindly provided by David Goeddel (Tularik Inc., San Francisco, CA) (26).
Treatment of Cells-UV irradiation was performed as described previously with slight modifications (24). Briefly, subconfluent cells were washed with PBS and exposed to UV light through colorless medium without FCS. For UV irradiation, we used a bank of six TL12 fluorescent bulbs (Philips, Eindhoven, The Netherlands), which emit most of their energy within the UVB range (290 -320 nm) with an emission peak at 313 nm. Throughout this study, a dose of 400 J/m 2 was used. Control cells were subjected to the identical procedure without being exposed to UV. UV irradiation at low temperature was carried out by keeping cells at 4°C for 10 min before UV exposure and during exposure, which lasted 40 s. Cells were kept at 4°C for another 20 min before incubation at 37°C for 16 or 24 h.
Osmotic shock was induced by incubating cells with 1 M sorbitol (Sigma, Munich) in FCS-free medium for 30 min either at 37°C or at 4°C (18). Thereafter cells were washed with PBS, supplemented with normal RPMI medium, and incubated for 16 h or 24 h at 37°C.
Stimulation of cells with TNF␣ at low temperature was carried out by adding TNF␣ (100 ng/ml) to cells that had been kept at 4°C for 10 min. Cells were kept at 4°C for another 20 min and then cultured at 37°C for 16 or 24 h.
Induction of DNA Repair via Photoreactivation-Photolyase was encapsulated into liposomes (Photosomes ® , AGI Dermatics, Freeport, NY) at a concentration of 1.2 mg/ml (27). Liposomes consisted of the lipids egg phosphatidylcholine, egg phosphatidyl trans-ethanolamine, oleic acid, and the membrane stabilizer cholesterol hemisuccinate. Empty liposomes were used as negative controls, referred to as liposomes. For photoreactivation, HeLa cells were irradiated as described above and either Photosomes ® or liposomes (40 l/ml each) were added. Cells were incubated at 37°C for 1 h in the dark, followed by illumination with photoreactivating light. As a light source for photoreactivating light, UVA fluorescent bulbs (TL09, Philips) filtered through a 6-mm glass plate with peak emission at 365 nm were used. Cells were exposed for 20 min, which corresponds to a photoreactivating light fluence of 12 kJ/m 2 . After photoreactivation, cells were supplemented with normal RPMI medium containing 10% FCS and incubated for 24 h at 37°C.
Detection of Cell Death-16 h after stimulation cells were detached from dishes, and apoptosis analyzed by a cell death detection ELISA (Cell Death Detection ELISA PLUS , Roche Molecular Biochemicals). The enrichment of mono-and oligonucleosomes released into the cytoplasm of cell lysates is detected by biotinylated anti-histone-and peroxidasecoupled anti-DNA antibodies and is calculated as follows: absorbance of sample cells/absorbance of control cells. Unless otherwise stated, this factor was used as a parameter of apoptosis and is given as the mean Ϯ S.D. of three independently performed experiments.
Semiquantitative Reverse Transcription Polymerase Chain Reaction (RT-PCR)-At 4 h after stimulation, total RNA was extracted from cells according to the protocol described by Chomczynski and Sacchi (28). Total RNA (1 g) was reverse transcribed with SuperScript RNase H reverse transcriptase (Life Technologies, Inc.). The amount of template needed was titrated by ␤-actin PCR in a 20-l reaction utilizing the RedTaq polymerase system from Sigma and evaluated densitometrically. A hIL-6-amplimer set from CLONTECH (Palo Alto, CA) was used as primers for IL-6 PCR.
Western Blot Analysis-Western blot analysis was performed as recently described (29). Briefly, cells were lysed in lysis buffer (50 mM Hepes pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl 2 , 1 mM EGTA, 100 mM NaF, 10 mM pyrophosphate, 0.01% NaN 3 , and Complete protease inhibitor mixture) for 15 min on ice. After centrifugation, supernatants were collected, and the protein content measured by Bio-Rad protein assay kit. The protein samples were subjected to 12% SDS-polyacrylamide gel electrophoresis, blotted onto nitrocellulose membranes, and incubated with antibodies directed against IB (Upstate Biotechnology, Inc., Lake Placid, NY). Equal loading was determined by reprobing membranes with an antibody directed against ␣-tubulin (Calbiochem, San Diego, CA). Signals were detected with an ECL kit (Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom).

RESULTS
Keeping HeLa Cells at Low Temperature during Exposure to UV Irradiation or Osmotic Shock Reduces Apoptosis to Different Degrees-Both nuclear and membrane events have been shown previously to contribute independently to UV-induced apoptosis (24). Since both components are essential to obtain the complete apoptotic response, inhibition of aggregation of death receptors expressed on the cell surface, such as CD95, by keeping cells at 4°C during UV exposure only partially inhibits UV-induced apoptosis (Fig. 1). Osmotic shock was recently demonstrated to induce receptor aggregation similar to UV (18). However, in contrast to UV, osmotic shock does not induce DNA damage and can thus be used as a stimulus that acts on the cell membrane exclusively. Rosette and Karin (18) predicted that any receptor whose activation mechanism involves multimerization should be activable by UV or osmotic shock. Hence, osmotic shock should also activate the CD95 receptor and thus induce apoptosis. As predicted, exposure of HeLa cells to osmotic shock caused apoptotic cell death (Fig. 1). When HeLa cells were kept at 4°C during osmotic shock, which prevents receptor clustering (18), apoptosis was completely inhibited. The complete prevention of apoptosis by inhibiting receptor aggregation thus confirms that osmotic shock, in contrast to UV, acts exclusively at the cell membrane when inducing apoptosis.
TNF␣ is known to induce apoptosis via activation of the TNF receptor 1 (TNF-R1) (31). To determine the effect of low temperature on TNF␣-induced apoptosis, HeLa cells were exposed to TNF␣, maintained at 4°C during the first 30 min of exposure, and then kept at 37°C. 16 h later apoptosis was measured. Under these conditions, low temperature had no effect on TNF␣-induced apoptosis, since the death rate was the same irrespective of whether the cells were kept at 4°C for the first 30 min or at 37°C throughout the entire incubation period (Fig. 1).
Low Temperature Inhibits UV-induced IL-6 Release-To further determine whether the involvement of both nuclear and membrane events is unique for UV-mediated apoptosis or also relevant for other biological effects caused by UV, we studied UV-mediated release of IL-6 by HeLa cells. HeLa cells do not constitutively secrete IL-6. However, UV irradiation resulted in a significant secretion of IL-6 ( Fig. 2). Exposure of HeLa cells to TNF␣, a well known inducer of IL-6, also caused enhanced IL-6 levels in the supernatants. Exposing HeLa cells to osmotic shock also induced IL-6 production, which may best be explained by activation of TNF-R1 (18). Accordingly, osmosisinduced IL-6 release was drastically reduced, when cells were kept at 4°C during exposure to osmotic shock (Fig. 2). Likewise, when cells were stimulated with TNF␣ at 4°C for 30 min, TNF␣ removed after that period by medium change and cells cultured for another 24 h at 37°C, no induction of IL-6 was observed (data not shown). In contrast, keeping TNF␣-stimulated cells at 4°C for 30 min did not have any inhibitory impact on TNF␣-induced IL-6 release provided that TNF␣ was not washed off but left in the medium for the rest of the incubation period of 24 h at 37°C (Fig. 2). This confirms that keeping cells at low temperature for such a limited period by itself does not cause reduced IL-6 production by inhibition of molecular processes within the cell, e.g. transcription, but just interferes with membrane receptor activation (18,22). Surprisingly, when HeLa cells were kept at 4°C during UV exposure, UVinduced IL-6 release was strongly reduced close to base-line levels. Since inhibition of UV-induced IL-6 release was almost as pronounced as the inhibition of IL-6 release induced by osmotic shock, a purely membrane-located event, this implies that UV-induced IL-6 release appears to be primarily mediated via membrane and not nuclear events.
Enhancement of DNA Repair Does Not Affect UV-mediated IL-6 Release-To confirm the above presented observations, we tested the effect of accelerated removal of UV-mediated DNA damage by enhancing DNA repair on UV-mediated IL-6 release. This approach was used previously to demonstrate the importance of DNA damage in mediating UV effects (10 -13, 24). To induce repair of UV-mediated DNA damage, we utilized the photoreactivating enzyme photolyase. Photolyase binds to UV-induced cyclobutane pyrimidine dimers in DNA and catalyzes its splitting by electron transfer from absorbing wave-FIG. 1. Low temperature differentially affects apoptosis induced by UV or osmotic shock. HeLa cells were exposed to UV (400 J/m 2 ), osmotic shock (Osmo, 1 M sorbitol) or TNF␣ (100 ng/ml) either at 37°C or after being kept at 4°C for 30 min. After treatment, cells were incubated at 37°C for 16 h. Control cells (Co) were incubated at 4°C for 30 min or kept at 37°C only. 16  lengths above 320 nm (photoreactivating light) (32). To enable uptake of the enzyme into the cells, photolyase was encapsulated into liposomes (Photosomes ® ) (27).
HeLa cells were irradiated with 400 J/m 2 UV. Immediately thereafter, Photosomes ® or empty liposomes were added and cells kept in the dark for 1 h, followed by exposure to photoreactivating light. As demonstrated previously (24), the combination of Photosomes ® and photoreactivating light significantly reduces UV-induced DNA damage in HeLa cells. However, enhancement of DNA repair by Photosomes ® had no effect on UV-induced IL-6 release, implying that in this case DNA damage is not an important mediator (Fig. 3). Likewise, addition of empty liposomes did not affect UV-stimulated IL-6 secretion. As already demonstrated in Fig. 2, the most effective way to inhibit UV-induced IL-6 release was keeping cells at 4°C during UV exposure. The combination of inhibiting aggregation of cell surface receptors by keeping cells at low temperature and enhancement of DNA repair by adding Photosomes ® did not result in further inhibition of UV-mediated IL-6 release (Fig.  3), although UV-induced pyrimidine dimers were reduced by 50% to 70%, as demonstrated by Southwestern dot blot analysis using an antibody directed against pyrimidine dimers (24). Taken together, these data indicate that DNA damage might not be of importance for mediating IL-6 release following UV exposure, further suggesting that membrane events may predominate over nuclear events concerning UV-induced IL-6 release in HeLa cells.
Keeping Cells at Low Temperature during UV Exposure Inhibits IL-6 mRNA Expression-Next, we addressed whether blocking UV-mediated receptor oligomerization by keeping cells at low temperature inhibits induction of IL-6 mRNA transcription. Therefore, semiquantitative RT-PCR utilizing primers amplifying parts of the IL-6 gene was performed. HeLa cells were UV-irradiated at 37°C or at 4°C, as described before, or alternatively exposed to osmotic shock at either 37°C or 4°C. In addition, cells were stimulated with TNF␣ at 37°C or 4°C. After incubating cells at 37°C for another 4 h, RNA was extracted and semiquantitative PCR performed on reverse transcribed templates. Inhibition of UV-or osmosis-induced receptor clustering by low temperature significantly reduced IL-6 mRNA expression (Fig. 4). In contrast, TNF␣ induced equal levels of IL-6 transcripts, irrespective of whether cells were kept at 37°C or 4°C during the initial period of stimulation. These data indicate that membrane effects, most likely receptor aggregation, induced either by UV or osmotic shock cause induction of IL-6 mRNA transcription. In addition to the ex-periments described above, these results exclude the unlikely possibility that low temperature inhibits UV-induced IL-6 release by interfering with protein translation or secretion.
UV-induced IL-6 Release Is Mediated via Activation of NFB-TNF␣ stimulates HeLa cells to produce enhanced amounts of IL-6 via activation of TNF-R1. TNF-R1 belongs to the group of receptors that are only biologically active when trimerized. Since UV directly induces receptor oligomerization independently of the respective ligands (18,19,22,23), UVinduced IL-6 release may be due to direct activation of TNF-R1 by UV light. One consequence of triggering TNF-R1 is activation of the transcription factor NFB. In addition, the IL-6 promoter contains several NFB binding sites (34). Hence, we postulated that, if activation of TNF-R1 is the initial signaling step in UV-or osmosis-induced IL-6 release, NFB should be involved in the signaling cascade. Activation of NFB is associated with degradation of the inhibitory protein IB by the proteasome pathway. Upon activation, IB is phosphorylated at two serine residues (Ser-32 and Ser-36); this acts as a signal for ubiquitination, followed by its degradation (35, 36). Since

FIG. 4. Low temperature inhibits both UV-and osmosis-induced induction of IL-6 mRNA expression.
HeLa cells were exposed to UV (UVB, 400 J/m 2 ), osmotic shock (OSMO, 1 M sorbitol), or TNF␣ (100 ng/ml) either at 37°C or after being kept at 4°C for 30 min. After treatment, cells were incubated at 37°C for 4 h. RNA was extracted and RT-PCR performed using primers for IL-6 and ␤-actin, respectively. Co, control.

FIG. 3. Enhancement of DNA repair does not affect UV-induced IL-6 release.
After UV irradiation, Photosomes ® were applied to the medium and cells incubated at 37°C for 1 h. Subsequently photoreactivation was carried out by irradiating cells with photoreactivating light. For control purposes, cells were either exposed to UV only or left untreated (Co). In addition, cells were incubated with empty liposomes followed by exposure to photoreactivating light. Experiments were performed either at 37°C (black bars) or at 4°C (white bars). 24 h later amounts of IL-6 were measured using an IL-6 ELISA. IL-6 concentrations (pg/ml) are shown on the y axis (mean Ϯ S.D. of triplicate samples). Data presented show one representative experiment of three independently performed experiments. IB is degraded by the 26 S proteasome, NFB activation can be blocked by proteasome inhibitors (36). Thus, we tested whether the proteasome inhibitor MG132 inhibits IL-6 release induced by UV and osmotic shock, respectively. Therefore, HeLa cells were stimulated with UV, osmotic shock, or TNF␣ at 37°C either with or without pretreatment with 5 M MG132 for 1 h. 24 h later supernatants were harvested and analyzed for their IL-6 levels. As demonstrated in Fig. 5, MG132 inhibited IL-6 release, irrespective of whether cells were stimulated with UV, osmotic shock, or TNF␣. This indicates that NFB is involved in the signaling of all three stimuli and may also imply that they use the same signaling pathway.
On the other hand, the approach using proteasome inhibitors provides only indirect evidence that induction of IL-6 by UV, osmotic shock, or TNF␣ is due to activation of NFB since one cannot exclude that inhibitors like MG132 may affect other pathways as well (37). Thus, we determined whether activation of NFB is involved in IL-6 induction by overexpressing a super-repressor form of IB. In this mutant form, two point mutations (Ser-32 3 Ala, Ser-36 3 Ala) prevent phosphorylation and subsequent proteasomal degradation of IB (25). As a consequence, NFB release, nuclear translocation, and functional DNA binding are prevented. Although HeLa cells transfected with the empty CMV vector were not impaired in their IL-6 release upon stimulation with UV or osmotic shock, cells transiently transfected with the IB dominant negative mutant exhibited a significant reduction in the release of IL-6 upon these two stimuli (Fig. 6). Together, these data indicate that activation of NFB is involved in signaling IL-6 release induced either by UV or by osmotic shock.
Keeping Cells at Low Temperature Inhibits Activation of NFB Induced by UV, Osmotic Shock, or TNF␣-It has previously been reported that UV activates NFB independent of a nuclear signal (15,16). Hence, we postulated that if activation of NFB by UV or osmotic shock is due to direct activation of surface receptors, e.g. TNF-R1, activation of NFB should be prevented when receptor aggregation is inhibited by keeping cells at low temperature. Therefore, HeLa cells were exposed to UV, osmotic shock, or TNF␣ either at 37°C or at 4°C. Protein extracts were prepared 30 min later and subjected to Western blot analysis using an antibody directed against IB (Fig. 7). At 37°C all three stimuli caused degradation of IB, UV being the weakest stimulus. In contrast, when cells were exposed to osmotic shock or UV at 4°C, degradation of IB was significantly reduced, indicating that activation of NFB under these conditions is inhibited. Taken together, these data confirm previous findings localizing activation of NFB by UV close to the membrane (15, 16).
Inhibition of Signaling of TNF-R1 by a Dominant Negative Mutant of TRAF-2 Reduces UV-induced IL-6 release-The data FIG. 5. IL-6 release induced by UV, osmotic shock, or TNF␣ is inhibited by the proteasome inhibitor MG132. HeLa cells were pretreated with MG132 (5 M) for 1 h and thereafter exposed to UV (400 J/m 2 ), osmotic shock (Osmo, 1 M sorbitol), or TNF␣ (100 ng/ml). 24 h later amounts of IL-6 were measured using an IL-6 ELISA. IL-6 concentrations (pg/ml) are shown on the y axis (mean Ϯ S.D. of triplicate samples). Data show mean values (Ϯ S.D.) of three independently performed experiments. Co, control.
FIG. 6. NFB is involved in UV-, osmosis-, and TNF␣-induced IL-6 release. HeLa cells were transiently transfected with a plasmid allowing overexpression of the super-repressor form of the NFB inhibitor IB (pRc IB D/N) or with the empty control vector (pRc CMV) only. Cells were exposed to UV (400 J/m 2 ), osmotic shock (Osmo, 1 M sorbitol) or TNF␣ (100 ng/ml). 24 h later amounts of IL-6 were measured using an IL-6 ELISA. IL-6 concentrations (pg/ml) are shown on the y axis. so far suggest that UV induces IL-6 release in HeLa cells by primarily acting on the cell membrane rather than affecting the nucleus. Reduction of IL-6 release upon UV exposure at 4°C implies that UV radiation activates a cell surface receptor, which ultimately induces IL-6 release. TNF-R1 appears to be the best candidate for several reasons; (i) TNF induces IL-6 release in HeLa cells, (ii) triggering of TNF-R1 activates NFB (26,38,39), and (iii) UV or osmotic shock cause clustering of TNF-R1 (18). To finally prove whether TNF-R1 is involved in UV-mediated induction of IL-6 release, a dominant negative mutant for the TNF-R1 signaling protein TNF-R-associated factor-2 (TRAF-2) was used (26). Upon activation, TNF-R1 trimerization is followed by the binding of the adapter protein TNF-R-associated death domain protein, subsequent recruitment of the signaling protein TRAF-2 and activation of the NFB pathway (38). The TRAF-2 dominant negative mutant plasmid is a truncated version of TRAF-2 lacking its N-terminal RING finger domain (26). HeLa cells were transiently transfected with this mutant encoding plasmid and exposed to UV or osmotic shock. Transient transfection with TRAF-2 dominant negative mutant resulted in significant reduction of IL-6 release caused by UV or osmotic shock (Fig. 8). Taken together, these data indicate that induction of IL-6 release by UV is initiated at the cell membrane by direct activation of TNF-R1. As a consequence of TNF-R1 triggering, NFB is activated, which ultimately results in induction of IL-6 gene transcription and its secretion. DISCUSSION The observations that removal of UV-mediated DNA damage by enhancing DNA repair results in the inhibition of certain biological UV effects strongly supported the idea that UVinduced DNA damage is crucially involved in mediating these effects (10 -13). On the other hand, numerous studies provided clear evidence that UV can also affect targets at the cell membrane, suggesting UV responses being independent of nuclear events (reviewed in Refs. 7 and 8). Bender et al. (40) recently reported that UV activates NFB sequentially in a DNA damage-independent and -dependent way. The discovery of a dual mechanism finally solved the apparent discrepancy between previous studies favoring either nuclear or non-nuclear pathways only. Along this line, using UV-induced apoptosis as the biological read-out-system, we currently observed that nuclear and membrane events are not mutually exclusive but that both components are essential to obtain the complete apoptotic response to UV (24). To determine whether the participation of both nuclear and membrane components is specific for UVinduced apoptosis or is also essential for other UV responses, we used UV-induced release of IL-6 in HeLa cells as an additional biological read-out-system. Surprisingly, the data obtained provide evidence that induction of IL-6 release in HeLa cells by UV is predominantly due to membrane events, whereas UV-induced DNA damage, if at all, might be of minor importance. These conclusions are based on several experimental approaches. First, keeping HeLa cells at 4°C during UV exposure drastically reduced IL-6 release. Keeping cells at low temperature prevents multimerization of cell surface receptors and thus inhibits their activation (18,22). The mechanisms by which low temperature inhibits receptor aggregation is not clear, but changes in membrane fluidity have been suggested (18).
The finding that UV-induced apoptosis is completely prevented only by inhibition of death receptor aggregation and removal of DNA damage by enhanced DNA repair (24) clearly indicates that in UV-induced apoptosis both membrane and nuclear events are critically involved. In contrast, induction of apoptosis by osmotic shock, which acts exclusively at the cell membrane and does not induce DNA damage, was completely prevented when cells were kept at low temperature during exposure to 1 M sorbitol (Fig. 1). Osmotic shock also induced IL-6 release presumably via triggering cell surface receptors, since IL-6 induction was profoundly inhibited by keeping cells at 4°C. Surprisingly, low temperature also drastically reduced UV-mediated IL-6 production. Inhibition of IL-6 release by low temperature is not due to induction of stress proteins induced by cold shock (22). In addition, keeping cells at low temperature does not interfere with transcription in general, since TNFinduced IL-6 expression was not affected when HeLa cells were kept at 4°C for 30 min and then cultured at 37°C in the presence of TNF␣ for another 24 h ( Fig. 2; Ref. 24). It is also unlikely that low temperature interferes with UV-induced DNA damage, since generation of cyclobutane pyrimidine dimers is not prevented in the cold (41). Together, these findings suggest that receptor aggregation may be the predominant event in UV-mediated IL-6 release, whereas UV-induced DNA damage may be of minor importance in this type of response.
To further examine the role of DNA damage in this system, DNA damage was reduced by enhancing DNA repair. For this purpose, we used the repair enzyme photolyase, which has already been successfully utilized for such purposes (24). As demonstrated previously, delivery of photolyase to UV-exposed HeLa cells via the liposome route followed by exposure of cells to photoreactivating light successfully enhances DNA repair, resulting in significantly reduced amounts of thymine-thymine cyclobutane pyrimidine dimers (24). In contrast to UV-induced apoptosis, UV-mediated IL-6 release by HeLa cells was not affected at all by accelerated removal of pyrimidine dimers. These observations are in contrast to a recent publication by Petit-Frere et al. (9), in which they reported that photoreactivation led to a reduction of IL-6 release in KB cells and normal human keratinocytes. However, in that case inhibition of UVinduced IL-6 release was not complete, implying that other pathways may be involved as well. In addition, photoreactivating light itself in the absence of photolyase reduced UV-induced IL-6 secretion in KB cells, although no reduction in pyrimidine dimers was observed under these conditions, a finding that needs further investigation. Furthermore, the relevant experiments in the study by Petit-Frere et al. were done with UVC, whereas we have used the biologically and physiologically more relevant UVB spectrum.
The failure of enhanced DNA repair to reduce IL-6 release on the one hand, and the marked inhibition of IL-6 production by keeping cells at 4°C on the other hand, suggested that UV might mediate IL-6 release via activation of cell surface receptors rather than via induction of DNA damage. In this scenario, activation of TNF-R1 appears to be of critical importance. TNF-R1 was among the receptors initially found to be aggregated by UV (18). Examining the role of TNF-R1 as a mediator of the activation of Rel/NFB proteins in keratinocytes, Tobin et al. (33) demonstrated ligand-independent activation of TNF-R1 by UV leading to recruitment of the transducer TRAF-2. Based on these findings and our own observation that TNF␣ induces the release of IL-6 in HeLa cells, we postulated that UV-induced IL-6 release may be due to direct activation of TNF-R1.
Upon activation, TNF-R1 becomes trimerized, which leads to the induced association of the intracellular TNF-R-associated death domain protein and subsequent recruitment of further proteins, including TRAF-2 and Fas-associated death domain protein. Although Fas-associated death domain protein mediates the apoptotic pathway, TRAF-2 is responsible for activation of NFB (38,39). Transfection of HeLa cells with a dominant negative mutant for TRAF-2 significantly reduced UVinduced IL-6 release; the same inhibition was obtained when osmotic shock was used as the inducing stimulus. Hence, these observations support our hypothesis that UV-induced IL-6 release in HeLa cells is mediated via direct activation of TNF-R1. However, it is important to mention that inhibition of IL-6 release upon transfection with the dominant negative TRAF-2 mutant was not complete. This may be due to the fact that only transiently but not stably transfected cells were used. Second, Rosette and Karin (18) predicted that any receptor whose activation mechanism involves multimerization should be activable by UV. Hence, we cannot exclude that other putative receptors may be involved as well. However, TNF-R2, which also uses TRAF-2 for signaling, does not play a role since, like keratinocytes and other epithelial cells (33,42,43), HeLa cells do not express TNF-R2 (data not shown). In addition, activation of TNF-R1 appears to be a direct effect of UV and not due to autocrine release of TNF␣, since TNF␣ was not found increased in the supernatants of UV-exposed HeLa cells (data not shown).
UV-induced association of TNF-R1 with TRAF-2 was shown to mediate Rel/NFB proteins. This pathway may be involved in UV-induced IL-6 release by HeLa cells, as addition of the proteasome inhibitor MG132, which blocks the activation of NFB (36), drastically suppressed UV-induced IL-6 release. In addition, inhibition of NFB by overexpression of a dominant negative mutant of IB reduced UV-enhanced IL-6 secretion in HeLa cells. It was previously demonstrated that NFB can be activated by UV independently of DNA damage (15,16). Direct triggering of TNF-R1 by UV may represent one, but certainly not the only, mechanism by which NFB can be activated independently from a nuclear signal. However, in this context, it is important to mention that UV was also found to be able to activate NFB independent of serine phosphorylation (40,44).
In summary, our data demonstrate that UV radiation can act on HeLa cells via activation of TNF-R1 and recruitment of TRAF-2 to elicit NFB-DNA binding, which results in enhanced transcription of IL-6. The signaling is independent of the ligand but also of DNA damage, since inhibition of TNF-R1 activation by low temperature results in pronounced inhibition of the IL-6 induction by UV. Hence, UV-mediated IL-6 release appears to represent an UV response that does not need nuclear signaling. Together, these findings indicate that UV can exert biological effects by acting exclusively on the cell membrane. Furthermore, these findings suggest that the multiple biological effects of UV on mammalian cells not only differ in the final processes they induce within the cell but also in the ways they are generated.