Down-regulation of Tumor Necrosis Factor α Expression by Activating Transcription Factor 2 Increases UVC-induced Apoptosis of Late-stage Melanoma Cells*

To identify mechanisms whereby activating transcription factor 2 (ATF2) alters the radiation resistance of human melanoma cells, we examined the possible role of ATF2 in UVC-induced apoptosis. Forced expression of full-length or truncated (Δ1–195 amino acids) forms of ATF2 in LU1205, a late-stage human melanoma cell line, elevated the levels of UVC-induced apoptosis. At the same time, either truncated or full-length forms of ATF2 reduced UVC-induced activation of the tumor necrosis factor-α (TNFα) promoter and decreased expression of TNFα. Forced expression of c-Jun in ATF2-expressing melanoma cells restored TNFα expression, suggesting that both forms of ATF2 sequestered transcription factors that positively regulate TNFα expression in response to UV irradiation. Antagonistic antibodies to Fas, but not to TNFR1, efficiently suppressed UVC-induced apoptosis, suggesting that the Fas pathway mediates the primary apoptotic signal in melanoma cells whereas the TNFR1 pathway elicits a survival signal. Indeed, treatment of melanoma cells with TNFα before UVC irradiation partially suppressed UVC-induced apoptosis, further supporting the protective role of TNFα in UVC-treated melanoma cells. Taken together, our findings suggest that ATF2 contributes to UVC-induced apoptosis through transcriptional silencing of TNFα, which balances Fas-mediated cell death in melanoma.

Programmed cell death (apoptosis) is a common cellular response to stress caused by environmental challenges (1). Altered expression of apoptosis-related proteins, which coincides with decreased or absent apoptosis, is commonly observed in various tumor types and is of fundamental importance in tumor resistance to host defenses as well as to clinical therapy (2). Indeed, reduced ability of tumors to undergo apoptosis is often associated with elevated drug resistance and poor clinical outcomes. Malignant melanoma is a primary example of a cancer that responds poorly to various treatments, including chemotherapy and ␥-irradiation (3). Despite the alarming increase in the incidence of this tumor in the past decade, the molecular mechanisms of its progression as well as the regulation of apoptosis in human melanoma remain largely unknown (4).
The mechanisms underlying apoptotic signaling have been intensively studied in recent years. Two cell surface molecules, Fas and TNFR1, 1 represent the main death signaling receptors (1). Receptor trimerization by ligands FasL and TNF␣, respectively, initiate receptor interaction with the intermediate signaling molecules FADD and FLICE (caspase-8), followed by activation of the caspase cascade (1,5,6). In addition to its role in apoptosis, TNFR1-mediated signaling is also linked with the regulation of survival functions, including induction of cytokines, growth factors, and cell proliferation and differentiation (7)(8)(9). Furthermore, TNF␣ appears to be one of the primary growth regulators of metastatic cancer cells (10,11), including late-stage melanoma cells (12)(13)(14). Hence, suppression of TNF signaling may enhance the apoptotic stimulus to tumor cells. Key to understanding the interplay between TNF␣ and Fas signaling is the nature of their upstream transcriptional regulators and downstream targets (i.e. inhibitors of apoptosis and survival proteins) in concordance with, or independent of, NF-B-dependent signaling (15)(16)(17). Expression of the human TNF␣ gene is controlled by the AP-1, ATF2, Erg-1, C/EBP␤, and NF-B transcription factors (18 -22). Transcriptional regulation of Fas and FasL genes was shown to be mediated by NF-B, AP-1, NF-AT, ATF2, and Egr3 (23)(24)(25)(26). As a dynamic process, programmed cell death is dependent, in many cases, on new gene expression (27), and is tightly regulated at the transcriptional level (28 -30).
Our interest in exploring mechanisms underlying key melanoma phenotypes led us to identify CREB/ATF transcription factors as regulatory proteins that play important roles in determining both melanoma's resistance to radiation and its metastatic potential (31,32). ATF2, a member of the ATF/ CREB family of basic region leucine zipper (bZIP) DNA-binding proteins (33)(34)(35), was shown to modulate melanoma's resistance to UVC irradiation but not to contribute to its metastatic potential (36). ATF2 is among several transcription factors, including AP-1, CREB, and Rel/NF-B, which have been found to be UV-inducible. ATF2 induction in response to stress occurs at the levels of transcription-translation and posttranslational modification, the latter of which is mediated by JNK/p38 kinases (37,38). ATF2 has been implicated in the transcriptional control of various stress-responsive genes, including c-Jun (39), interferon-␤ (40), urokinase (41), TGF-␤2 (42), TNF␣ (18,20), and DNA polymerase ␤ (43). Differential splicing of ATF2 creates several isoforms which exhibit different transcriptional outputs (44). Full-length ATF2 is transcriptionally inactive in its native form as a result of intramolecular interaction of its DNA-binding domain with the amino-terminal transactivation domain (33). ATF2 as well as a related factor, ATFa, not only can form DNA-binding homodimers but also can efficiently heterodimerize with members of the ATF/CREB and Jun/Fos families, thus providing an important transcriptional control mechanism (41,(45)(46)(47). Duration and magnitude of ATF2 transcriptional output are also regulated at the level of phosphorylation and stability. While stable in its native non-active form, ATF2 is a short-lived protein after homo-or heterodimerization. ATF2 instability is further facilitated by heterodimerization with certain partners, which provide additional targeting molecules for ubiquitination through their own docking sites, as shown for c-Jun (48).
To identify mechanisms underlying ATF2's ability to modulate radiation resistance, we studied the possible role of ATF2 in apoptosis of human melanoma cells. We demonstrate that ATF2 expression increases UV-mediated apoptosis through suppression of TNF␣ expression, which elicits survival signals in tumors of this type.
Transient Transfection and GFP Assay-Melanoma cells (6 ϫ 10 6 ) were transiently cotransfected with either ATF-2-or TAM-67 expression vectors together with marker plasmid encoding green fluorescent protein, pGFP (CLONTECH), 10 and 2.5 g, respectively, by electroporation. Twenty-four hours after transfection cells were irradiated by UVC (60 J/m 2 ) and 18 h after treatment were stained with PI and analyzed using flow cytometry. The ratio of GFP ϩ PI ϩ to total GFP ϩ cells was used to measure frequencies of death in the transfected cells.
Transient Transfection and X-gal Staining-Melanoma cells (6 ϫ 10 6 ) were transiently cotransfected with ATF2 expression vectors and pCMV-␤-gal (10 and 2.5 g, respectively) by electroporation. Twentyfour hours after transfection, cells were irradiated with UVC (60 J/m 2 ) as described previously (32). To identify ␤-galactosidase activity, cells were fixed 18 h after UVC treatment and stained with X-gal (28). The ratio of blue cells with apoptotic morphology to the total number of blue cells served as the indicator of the level of apoptosis in transfected cells.
Annexin-V-FITC staining in the presence of PI was performed for detection of early apoptosis levels using TACS Annexin-V-FITC kit (Trevigen, Gaithersburg, MD). Flow cytometric analysis was performed on an Epics-Profile II flow cytometer (Coulter, Hialeah, FL) with Elite software 4.01 by analyzing 10 4 cells/sample, using wide scatter gates to include late apoptotic cells.
DNA fragmentation analysis was performed as described previously (54). Cells were pelleted and resuspended in 0.5 ml of hypotonic buffer with 0.1% Triton X-100 containing propidium iodide (PI) (40 g/ml) and DNase-free RNase A (1 mg/ml). Cells were incubated at 37°C for 30 min and analyzed on an Epics-Profile II flow cytometer (Coulter). The percentage of cells to the left of the diploid G 0/1 peak, diagnostic of hypodiploid cells that have lost DNA, was taken as the percentage of apoptotic cells. The analysis was performed without light scatter gating.
Western Blotting Analysis-Total cell extracts were resolved on 10% SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and processed by standard methods. The polyclonal anti-PARP serum (Biomol, Plymouth Meeting, PA) was used at a 1:1000 dilution. Polyclonal Abs against ATF2 and phosphoATF2 (New England Biolabs) were used at a 1:1000 dilution. Monoclonal Abs against human TNF␣ (clone Mab1), Fas (clone G254 -267) (PharMingen), and TNFR1(R&D System), and polyclonal Abs against FasL and IB␣ (Santa-Cruz Biotechnology) were used at the same dilution. The secondary Abs were goat anti-rabbit IgG or goat anti-mouse IgG conjugated to horseradish peroxidase (dilution 1:5000). Signals were detected using the ECL system (Amersham Pharmacia Biotech).

Characterization of UVC-induced Apoptosis in Human
Melanoma Cells-To study regulation of UVC-induced apoptosis in melanoma, we used the LU1205 cell line, which represents a late stage melanoma that harbors wild type p53. These cells exhibit higher resistance to irradiation than early stage melanoma cells (55). The finding that expression of a truncated ATF2 form (⌬1-195) reduced radiation resistance of late-stage melanoma cells (36) prompted our interest in elucidating mechanisms by which ATF2 may elicit such changes. Since radiation resistance may be inversely correlated with the degree of programmed cell death, we studied the possible role of ATF2 in UV-induced apoptosis of these melanoma cells.
Several markers of apoptosis were used to monitor and characterize UVC-mediated programmed cell death in melanoma cells. To distinguish between early and late phases of apoptosis, annexin-V-FITC and PI staining, which detect early and late phases of apoptosis (56), respectively, were used. Eighteen hours after UVC irradiation, 20% of LU1205 melanoma cells were annexin-V ϩ , of which 11% were annexin-V ϩ PI Ϫ and 9% annexin-V ϩ PI ϩ , reflecting the early and late phases of apoptosis, respectively (Fig. 1A). In several independent experiments, we found that 25 Ϯ 6% LU1205 cells were annexin-V ϩ (apoptotic) 18 h after UVC irradiation, compared with 5 Ϯ 3% for nontreated cells. DNA fragmentation analysis, performed by staining cell nuclei with PI and determining the percentage of apoptotic cells with hypodiploid DNA content (Fig. 1B), revealed that 36% of the total cell population had apoptotic nuclei with hypodiploid DNA content 40 h after UVC irradiation, compared with 2% for nontreated cells (Fig. 1B). Within 3 days after treatment the apoptosis level reached 70%; the vast majority of melanoma cells died 4 days after irradiation (data not shown). Cycloheximide (Fig. 1C) as well as actinomycin D (data not shown) partially suppressed UVC-induced apoptosis of melanoma cells, indicating that UVC-induced gene expression is required if maximal levels of apoptosis are to be achieved.
Since caspase-dependent proteolysis has been shown to play an important role in different types of programmed cell death (5), we determined the possible role of activation of the caspase cascade following UVC irradiation of melanoma cells. A 50 M quantity of zVAD-fmk, a caspase inhibitor, substantially suppressed UVC-induced apoptosis (Fig. 1, A and B). To monitor cleavage of poly(ADP-ribose) polymerase (PARP), a typical target of caspase-3 and an indicator of early apoptotic commitment (57), immunoblot analysis of nuclear proteins from UVCtreated melanoma cells was performed with the aid of anti-PARP Ab, revealing the presence of the 85-kDa cleavage product 18 h after UVC treatment, but not in the presence of zVAD-fmk (Fig. 1D).
The role of proteasome-dependent degradation in UVC-mediated apoptosis of melanoma cells was examined using the proteasome inhibitor lactacystin (58). Whereas 10 M lactacystin did not induce apoptosis in nonstressed melanoma cells (data not shown), a substantial increase in UVC-induced apoptosis was seen in the presence of 10 M lactacystin, indicating a protective role of proteasome-dependent protein degradation in UVC-induced apoptosis (Fig. 1A). Increased PARP cleavage was seen in lactacystin-treated cells following UVC irradiation (Fig. 1D). These observations suggest that all three major cell regulatory processes (new gene expression, proteasome-dependent protein degradation, and caspase proteolysis) contribute to UVCinduced apoptosis of melanoma cells.
We next determined whether transcriptional suppression by ATF2 alters the degree to which melanoma cells undergo apoptosis in response to UV irradiation. Apoptosis studies were performed by cotransfection of ATF2 expression vectors with either pGFP or pCMV-␤-gal encoding green fluorescent protein and ␤-galactosidase, respectively, as marker genes. To determine the frequency of death in ATF2-transfected cells, UVCirradiated and nontreated cells were stained with PI and analyzed by flow cytometry, allowing quantification of GFP ϩ PI ϩ cells in the overall population of GFP ϩ cells. Expression of either full-length or truncated forms of ATF2 notably increased (1.8-and 1.3-fold, respectively) the frequency of cell death following UVC irradiation of LU1205 cells (Fig. 2B). Parallel analysis using X-gal staining of ␤-galactosidase-expressing cells enabled visual detection of blue cells with apoptotic morphology in the total population of blue cells (28), revealing a similar increase in apoptosis levels after UVC irradiation in ATF2-transfected cells (data not shown).
To further investigate ATF2-dependent regulation of UVCinduced apoptosis of human melanoma cells, we established LU1205 cell lines stably transfected with either full-size ATF2 cDNA (LU1205/ATF2 cells) or the truncated form, GAL4-ATF2⌬1-195 (LU1205/ATF2⌬1-195), as well as the control cell line transfected with the empty vector and neo marker plasmid (LU1205/neo). Western blot analysis confirmed elevated expression of the respective forms of ATF2. Cells that constitutively express the full-size ATF2 revealed increased levels of 68-kDa ATF2 as well as spliced isoforms, before and after UVC irradiation (Fig. 3A). However, immunoblot with anti-phospho-ATF2 (Thr71) Ab revealed the same level of phosphorylated forms of ATF2 after UVC irradiation of LU1205/ATF2 cells, indicating a relative decrease in the ratio of phospho-ATF2 to total ATF2 (Fig. 3A). Immunoprecipitation with anti-GAL4 and subsequent Western blotting with anti-ATF2 Ab further confirmed the presence of the fused truncated form of ATF2 in LU1205/ATF2⌬1-195 cells (data not shown), as was observed previously with MeWo melanoma cells transfected with .
To confirm the role of ATF2 in regulation of TNF␣ expression in the melanoma cells, LU1205 cells were transiently cotransfected with the Ϫ615 TNF-Luc reporter construct and either full-size or truncated ATF2 expression vectors. Both ATF2 constructs reduced the degree of UVC-mediated transactiva-tion by the TNF␣ promoter in a dose-dependent manner (Fig.  5B). That full-length (although hypophosphorylated upon UV irradiation; Fig. 3A) and truncated (without transactivator domain) forms of ATF2 elicit the same silencing effect on TNF␣ transcription could be attributed to squelching of a positive factor as a result of efficient heterodimerization with other bZIP family members. Among the primary transcription factors that associate with ATF2 is c-Jun (39,47), which was also shown to increase TNF␣ transcription via the CRE/AP-1 site, following stimulation of monocytes by lipopolysaccharide (21). We therefore tested the possible effect of c-Jun on TNF␣ transcription in ATF2-expressing melanoma cells. To this end transcriptional activity of c-Jun was blocked by using a dominant   FIG. 6. A, c-Jun transactivates TNF␣ promoter activity in LU1205 cells. Effect of full-length and truncated forms of ATF2 on c-Jun-dependent transactivation of TNF␣ promoter activity. LU1205 cells were transiently cotransfected with Ϫ615 TNF-Luc (5 g) in the presence or absence of 10 g of pECE-ATF2 and pECE-ATF2⌬1-195 and indicated quantities of the pCMV-c-Jun expression construct and 5 g of ␤-galactosidase. Luciferase and ␤-galactosidase activities were determined after UVC treatment, as described in the legend to Fig. 2. B, c-Jun partially suppressed UVC-induced apoptosis, which was accelerated by ATF2⌬1-195 expression. LU1205 cells were cotransfected with either pECE-ATF2⌬1-195 (20 g) or with this vector and pCMV-c-Jun-HA (10 g or 20 g) in the presence of GFP marker plasmid. Two days after transfection, cells were UVC irradiated. Apoptosis levels of GFP-positive cells were detected using flow cytometry after PI staining. negative c-Jun construct, TAM-67, which lacks the first 200 amino acids (52,53), similar to the truncated ATF2 form used in the present study. Expression of TAM67 in LU1205 suppressed TNF promoter-dependent luciferase activity (Fig. 5B) as well as increased apoptosis (data not shown). Conversely, transfection of wild type c-Jun resulted in a dose-dependent increase of TNF␣ promoter activity in both non-treated and UVC-irradiated cells (Fig. 6A). Furthermore, forced expression of wild type c-Jun partially abrogated the negative effect of both full-length and truncated forms of ATF2 on TNF␣ promoter activity (Fig. 6A). c-Jun expression in LU1205 cells (cotransfected with the GFP marker plasmid) also blocked the increase in UVC-induced apoptosis caused either by ATF2⌬1-195 (Fig. 6B) or full-length ATF2 overexpression (data not shown). These observations suggest that the ability of ATF2 to increase the degree of UVC-induced apoptosis is mediated via suppression of positive transcription factors, such as c-Jun.
Analysis of TNF␣ promoter activity in stably transfected LU1205/ATF2 cells further confirmed the data obtained by transient transfection. LU1205/ATF2 cells exhibited reduced (35-40%) TNF␣ promoter activity after UVC treatment (without changes in basal levels) (Fig. 7A). Decrease in TNF␣ transcription coincided with down-regulation of TNF␣ protein levels. Western blot analysis revealed a UVC-dependent increase in the level of TNF␣ protein (p17) in LU1205/neo cells, whereas this increase was substantially suppressed in LU1205/ATF2 cells (Fig. 7B). Expression of TNFR1 (whose transcription is positively regulated by TNF␣) was also decreased in LU1205/ ATF2 cells prior to and after UV irradiation compared with control cells while TRAF2 (60) levels were relatively stable (data not shown).
Role of TNFR and Fas-mediated Signaling in UVC-induced Apoptosis of ATF2-expressing Melanoma Cells-To determine which death-signaling pathway mediates UVC-induced apoptosis in control and ATF2-expressing melanoma cells, antagonistic mAbs to either TNFR1 or Fas that block the relevant signaling cascade were added to cell cultures 1 h before UVC irradiation. Addition of anti-TNFR1 mAb further increased the degree of apoptosis after UVC irradiation of control and ATF2 (both full-length and truncated forms) transfected cells. Conversely, anti-Fas mAb substantially suppressed apoptosis in these melanoma cells (Fig. 8A). These results point to the pro-apoptotic role of Fas signaling and anti-apoptotic role of TNFR-mediated signaling for LU1205/neo, LU1205/ATF2, and LU1205/ATF2⌬1-195 cells following UVC irradiation. To confirm the protective role of TNFR1-mediated signaling, melanoma cells were pretreated with recombinant TNF␣ before UVC irradiation. The presence of TNF␣ (2-4 ng/ml) in the medium partially reduced the degree of UVC-mediated apoptosis of control and ATF2-transfected melanoma cells (Fig. 8B). These observations demonstrate that TNF␣ activates a survival signal in melanoma cells, which is suppressed upon ATF2 expression.
Suppression of UVC-induced TNF␣ promoter activity by both the truncated and the full-length forms of ATF2, in transient transfection as well as in LU1205/ATF2 cells (followed by down-regulation of TNF␣ expression), coincided with increased levels of UV-mediated cell death and pointed to a causal linkage between ATF2 and regulation of cell death in melanoma. It is important to note, however, that full-length ATF2 was more efficient in elevating the degree of UVC-induced apoptosis of melanoma cells than the truncated form of ATF2. Since the full-length form of ATF2 is also expected to bring about positive transcriptional output, we performed experiments to determine whether the apoptotic signal in melanoma cells, namely the Fas pathway, may be also affected by ATF2. Melanoma cell lines, including LU1205, express both the death receptor and its ligand, Fas and FasL (Ref. 61 and Fig. 9D). The human FasL promoter contains two NF-AT sites at positions Ϫ279 and Ϫ140 (23, 62), MEKK-RE (at Ϫ336), which binds ATF2-Jun heterodimers (63), and AP-1-and NF-B-binding sites at positions Ϫ1088 and Ϫ1050, respectively (23,24). Transfection of nested deletions of 5Ј FasL promoter constructs (Ϫ1.3 kilobase pairs, and Ϫ453, Ϫ318, Ϫ237 base pairs) revealed activation of a UVC-dependent promoter. Increased FasL promoter activity after UV irradiation is likely to be mediated by target sequences located within the Ϫ453 and Ϫ318 region, as the Ϫ237 construct no longer exhibited strong activation by UV irradiation. Further support for the role in FasL activation of the MEKK-RE site located within this comes from analysis of 5Ј promoter sequences mutated within the NF-B or AP1 sites; neither was able to completely abrogate UV responsiveness (Fig. 9B). These results point to the role of MEKK-RL and its associated proteins in mediating increased FasL transcription. Indeed, cells expressing the ATF2 construct exhibited a higher basal level of FasL promoter activity. After UV irradiation, a further increase of 2-3-fold was noted in the LU1205ATF2 cells (Fig. 9C). The overall level of FasL protein was also higher in LU1205/ATF2 cells before and after UVC treatment than in control cells (Fig. 9D). These observations suggest that ATF2 plays a positive role in the regulation of FasL transactivation/ expression in melanoma (Fig. 9, A-D), as previously observed in other cell systems (63). Because of blockage by different inhibitory proteins, including Bcl-2, the presence of both Fasreceptor and Fas ligand in melanoma cells is necessary but not sufficient to initiate death signaling. UVC-induced degradation of Bcl-2 ( Fig. 9D) correlates with the progression of apoptosis seen in these cells. The elevated expression level of FasL in LU1205/ATF2 cells may explain the higher level of apoptosis seen after UV irradiation when compared with the apoptosis seen in cells expressing the truncated form of ATF2.
Melanoma cells are resistant to apoptotic signaling induced by TNF␣, which, as noted, elicits a survival signal in these cells. Since NF-B has been implicated in protection against TNF-mediated apoptosis (64 -66), we next examined the possible effects of ATF2 overexpression on NF-B DNA binding activity and transactivation before and after UVC treatment of melanoma cells. LU1205 cells contain relatively high basal NF-B activity (Fig. 10A). An additional increase in DNA binding activity of NF-B p65-p50 (Fig. 10B, the upper band of the specific NF-B DNA-binding complex) and transactivation of NF-B were observed in ATF2-and ATF2⌬1-195-transfected cells, both at the basal and at the UVC-induced levels (Fig. 10,  A and B). Increased levels of NF-B activity after UVC irradiation of ATF2-overexpressing cells, when compared with the control cells, are due to down-regulation of IB␣ levels in these cells 18 h after UVC treatment (Fig. 10C), possibly as a result of ATF2-dependent regulation of IB expression. Since UVC treatment causes prolonged activation and nuclear translocation of NF-B (0.5-20 h) when compared with the short NF-B activation in response to TNF␣ treatment (10 -90 min), it is not possible to distinguish between primary and secondary TNFmediated activation of NF-B in melanoma cells. However, the abundance of NF-B in these cells, especially in ATF2-overexpressing cells, is likely to represent an additional factor that promotes Fas-versus TNF-mediated apoptosis of late-stage melanoma cells. DISCUSSION The functional role of ATF2 in the regulation of cell death and survival in human melanoma was evaluated by overexpression of its full-size and truncated (without transactivation domain) forms. Here we provide evidence that both forms of ATF2 serve as potent accelerators of UVC-induced cell death in metastatic human melanoma LU1205 cells. Apoptosis following UVC treatment and its enhancement by overexpression of both forms of ATF2 were demonstrated by characteristic changes in cell morphology, redistribution of phosphatidylserine in the plasma membrane (annexin-V-FITC staining), activation of caspase cascade, PARP cleavage, and DNA fragmentation.
Among the important determinants influencing the degree of ATF2-mediated transactivation are: (i) alternative promoter usage of the ATF2 gene and differential splicing, which create several ATF2 isoforms that exhibit different transcriptional output (fulllength ATF2 in its native form is transcriptionally inactive as a result of inhibitory intramolecular interaction, whereas a 42-kDa spliced form (⌬150-248) is constitutively active; Refs. 33  ; and (iv) interaction with the transcriptional coactivator p300 (dependent on phosphorylation of Ser 121 ), which contributes to ATF2 transcriptional activity (70). This tight regulation suggests that any of these variables (hypophosphorylation, altered ratio between different full-length and spliced forms, a change in the nature of the heterodimers) may suppress ATF2-dependent transcription. Indeed, silencing effects were described for different ATF2/CREB family members. For example, binding of CREB and CREM to the AP-1 site inhibited activation by c-Jun (71), ATF2 down-regulated hepatitis B virus X promoter activity by competition for the AP-1 binding site and formation of ATF2-Jun heterodimers (72), and CREB and ATF1 were shown to be potent inhibitors of several transcription factors (73). The dominant negative form of CREB efficiently reduced radiation resistance (32), and inhibited tumor growth and metastatic potential of human melanoma through squelching of other CREB-associated proteins (31,74). The relationship between changes in ATF2-dependent transactivation after UVC irradiation and the degree of apoptosis, as demonstrated in the present study, suggest that ATF2 tightly regulates expression of genes that control apoptosis.
One of the common causes of cell death is down-regulation of expression and decrease in the concentration within the medium of essential growth factors, for example, death of the IL-2-dependent CTLL-2 cell line upon deprivation of IL-2 (75). Late-stage metastatic melanoma cells are characterized by autocrine growth stimulation as a result of production of several cytokines, including bFGF, IL-6, IL-8, and TNF␣ (59). Numerous observations indicate that TNF␣ promotes the metastatic potential of melanoma cells (12)(13)(14) as well as of some other FIG. 10. A, NF-B activity in ATF2transfected LU1205 cells. LU1205/neo, LU1205/ATF2, and LU1205/ATF2⌬1-195 cells were transiently transfected with either 4xNF-B-CAT or with the empty vector pBL4-CAT in the presence of pCMV-␤-gal. Two days after transfection, cells were irradiated with UVC and protein extracts were prepared 18 h later. The protein concentration of extracts was normalized based on ␤-gal activity. Relative CAT activity is indicated. B, electrophoretic mobility shift assay of nuclear extracts of control and LU1205/ATF2 cells 3 h after UVC irradiation was performed as described previously (29). The positions of NF-B DNA-binding complexes are indicated by arrows; the upper bands represent the p65-p50 heterodimer, the intermediate band reflects the p50-p50 complex, and the lower band, nonspecific binding (ns). C, Western blot analysis of IB␣ levels in the respective melanoma cell lines 18 h after UVC treatment.
tumors (10, 76 -78). Down-regulation of c-KIT levels in melanoma cells has been implicated in the enhancement of melanoma tumorigenicity and metastasis (79). Although the effect of TNF␣ on negative regulation of c-KIT expression in melanoma has not been determined, it has been described for normal bone marrow progenitor cells and myeloid leukemia cells (80,81). In general, TNF␣ plays a dual role in the regulation of cell death and survival. This cytokine is an essential growth factor not only for metastatic, but also for normal, cells in certain circumstances, e.g. in immature thymocytes (82). On the other hand, this cytokine induces apoptosis of some tumor cell lines and elicits an unusually wide range of biological responses (7)(8)(9). Expression of the TNF␣ gene is regulated by different transcription factors, including AP-1, ATF2, NFAT, Erg-1, C/EBP␤, and NF-B (18 -22). The present study demonstrated the ability to modulate TNF␣ transcription using both full-length and truncated forms of ATF2, resulting in increased susceptibility of UV-treated melanoma cells to programmed death. The dominant negative form of c-Jun (TAM 67) (53) also had a similar effect, further indicating that down-regulation of TNF␣ transcription could be due to heterodimerization of inactive ATF2 with c-Jun. In contrast, overexpression of functionally active c-Jun allowed the negative effect of ATF2 on TNF␣ expression to be overcome and partially rescued UV-and ATF2-mediated apoptosis, suggesting that ATF2 silences the c-Jun contribution to TNF␣ expression.
The present study also demonstrates that UVC-induced apoptosis of LU1205 as well as ATF2-transfected LU1205 cells is mediated by Fas death signaling. The role of Fas signaling in radiation-induced apoptosis was described in several normal and tumor cell systems (83)(84)(85)(86)(87). It was also demonstrated that UVC irradiation by itself may initiate trimerization of the Fas receptor and induction of signaling, substituting FasL (83,87). However, it is still unresolved whether UVC-induced Fas trimerization per se is sufficient to mediate apoptosis, as it is difficult to exclude other cellular changes elicited by UV irradiation. Melanoma cells expressing full-length ATF2 were found to express higher levels of FasL protein, which may also contribute to Fas-mediated apoptosis, as was shown in UVinduced apoptosis of human lymphocytes (63,84). Elevated FasL expression may explain the higher degree of apoptosis seen after UV irradiation of melanoma cells that express fulllength ATF2 versus truncated form, as, in addition to silencing TNF, by both forms of ATF2, only the full-length form appears capable of mediating positive transcriptional signals. The ability of full-length ATF2 to elicit a negative signal in the case of TNF␣ and a positive signal in the case of FasL may be due to its heterodimerized partner, the overall contribution of neighboring transcription factors on a given promoter, and a possible requirement for the contribution of co-activators. Clearly, the effects of full-length ATF2 as a transcription factor are impaired due to its poor phosphorylation. We did not, however, complete analysis of all spliced forms of ATF2 expressed in these melanoma cells and thus cannot exclude further interplay between themselves and other bZIP partners.
The resistance of cells against TNF-mediated killing is linked to NF-B expression (9, 64 -66). Metastatic melanoma LU1205 cells, used in the present study, are also characterized by moderate NF-B activity that may provide resistance against TNF␣-mediated apoptosis. Indeed, inhibition of NF-B activity by IB⌬ transformed TNF␣ from a protective to a death factor. 2 Our data extend previous findings regarding down-regulation of radioresistance in human melanoma MeWo cells by truncated forms of CREB and ATF2 (32,36). We have shown in the present study that the truncated form of ATF2 increased UVC-induced apoptosis. Thus, an elevated level of apoptosis coincides with reduced resistance to UV irradiation. Surprisingly, comparison of ATF2⌬1-195 with full-length ATF2 overexpression revealed that while both elicit equally efficient silencing of TNF␣ expression, full-length ATF2 was more efficient in elevating UV-mediated apoptosis, possibly because of its ability to increase expression of FasL, which boosts death signaling in these melanoma cells. This is unlike the effect of full-length ATF2 in early stage melanoma cells DM3211, i.e. an increase radiation resistance (36), suggesting that the impact of ATF2 on cell survival depends on its proper activation and interaction with other factors that have been altered during melanoma progression. Indeed, the balance between various ATF2 forms, c-Jun, and other members of the CREB/ATF and AP-1 families is expected to change during melanoma development as was previously shown in the B16 mouse melanoma model (88).
In summary, we established an inverse correlation between TNF␣ expression as a survival factor and the consequences of Fas-mediated death signaling in melanoma cells. Our findings provide the foundation for design of ATF2-based reagents that could be useful in altering the rate of irradiation-mediated programmed death of melanoma cells.