The effects of UV radiation on the growing ability and induction of apoptosis was studied in diverse cellular contexts, in both human and mouse cell lines with normal, defective or no p53. Regardless of the different genetic backgrounds, UVC radiation of 254 nm appeared to be an efficient inducer of transitory growth arrest in every cell line studied.
G1- and S-phase arrests after UV
The effect of 50 J/m2 UVC radiation on DNA replication was first examined in exponentially growing mouse NIH3T3 fibroblasts. As measured by 5-bromo-2´-deoxyuridine (5-BrdUrd) nucleotide analog incorporation, cells underwent a transient growth arrest between 6 h and 12 h after irradiation followed by a gradual increase in replication from 12 h onwards (I). However, flow cytometry analyses from 6 h and 12 h timepoints indicated that despite inhibition of DNA replication cell cycle phase distribution was identical in radiated and unradiated cell populations, demonstrating that instead of gathering into G1 the radiated cells arrested in the phase they were at the time of radiation (unpublished observation).
To examine the kinetics of UV-induced growth arrest - as well as activation of tumor suppressors pRB and p53 - in specific cell cycle phases, NIH3T3 cells were synchronized by contact inhibition and serum starvation to the G0/G1 phase, replated, and stimulated with normal medium thus allowing cells to progress from G1 to S. Flow cytometry analyses revealed that if the cells were radiated in G1 or the G1/S border, they arrested in that phase for 10 to 12 h before entering back into the cell cycle. However, if the cells were exposed to UV while in S or early G2/M, the arrest period was considerably longer and cells failed to recover from the insult during follow-up unlike G1-phase cells (I).
An even more detailed study of growth arrest kinetics was carried out with NIH3T3 cells that were synchronized to G0/G1 and replated in the presence of hydroxyurea (HU) to inhibit the progression to S. Synchronized cells were treated with 50 J/m2 of UV either at this artificial G1/S border or, after release from HU-block, in different S-phase stages (II). Information combined from 5-BrdUrd incorporation analyses and flow cytometry profiles indicated that during the 6 h incubation time after radiation the cells were able to minimally progress in the S phase. Accordingly, DNA replication did not cease instantly but decreased gradually becoming negligible during the last incubation hour. HU is an inhibitor of ribonucleotide reductase depleting the nucleotide pool available and thus inhibiting the DNA replication. When it was added to the medium, cell cycle progression and DNA replication of all UV-radiated cells were inhibited (II).
Diverse DNA damaging conditions including UV radiation lead to growth arrest of cells in order to provide sufficient time for repair of the damaged DNA and maintenance of genomic stability. As found by us and others, UV-induced inhibition of growth is rapid, beginning almost immediately after the insult. Pyrimidine dimers and photoproducts caused by UV radiation impair the replication by distorting the DNA conformation Taylor et al., 1988; Taylor et al., 1990) and, at the same time, activate regulatory pathways involving damage recognition and checkpoint control. Inhibition of DNA replication upon high dose UV comprises inhibition of replicon initiation and reduction in the rate of chain elongation, the former being inhibited by 50% within the first hour after exposure to UV (reviewed by Frieberg, 1995).
Cell cycle phase synchronization is absolutely essential in order to separate the cell cycle responses of exponentially growing cells representing all cycle phases. As suggested by results obtained from exponentially growing cells (I), we demonstrated that after UV exposure the cell cycle can be halted in either G1 or S phases depending on the original stage of cells, and after insult the cells progressed in the cycle only little if at all (I, II). Moreover, S phase arrest appeared longer than G1 arrest as if more time was required for repair of S-phase lesions. Contrary to what might be expected, the apoptosis rate was not significantly higher in S-arrested cells (unpublished observation).
It has been demonstrated years ago that mutations due to UV radiation are considerably fewer if cells are prevented from entering the S phase, for example by HU (Stone-Wolff and Rossman, 1982). Due to active replication in S, repair of UV lesions may be less efficient in this phase leaving more mutations unrepaired. In addition to impaired repair, mutations are also fixed to the genome during subsequent DNA replication. Studies with synchronized normal fibroblasts have demostrated that the longer the period between UV and entry into S, the less the mutations accumulate (Konze-Thomas et al., 1982). The disappearance of premutagenic lesions before S is probably based on nucleotide excision repair, since XP cells deficient in NER display the same rate of mutations regardless of the length of G1 phase preceding S. Consistent with our findings, final cell survival is not, however, affected by the time point of UV radiation in either cell lines, suggesting that other repair mechanisms may in part overtake the functions of NER (Konze-Thomas et al., 1982). When regarding mutagenesis, a wave length of 254 nm has proven to be most efficient for cells in the S phase or its vicinity (Kaufmann, 1989).
Growth arrest and apoptosis in the absence of functional p53
The dependency of UV-induced growth arrrest on p53 functions was examined in multiple cell lines with different p53 status. Besides normal human and mouse fibroblasts, growth arrest upon UV treatment was observed also in cell lines with mutant p53 (SW480) or with p53 inactivated by SV40 large T antigen (RB4.6pVU-0) with at least 60% inhibition of DNA replication (I).
p53 coding exons of seven melanoma cell lines were sequenced, and three cell lines were found to carry mutant p53, one of them expressing a truncated form of p53 protein (IV). As determined by 5-BrdUrd incorporation, all melanoma cell lines underwent a growth arrest after 50 J/m2 of UVC with at least 60% inhibition of DNA replication. Consistent with findings in NIH3T3 cells, the arrest responses of melanoma cell lines peaked at 6 to 16 h after irradiation, after which two alternative responses were found: three cell lines (A-375, WM239, and SK-MEL-2) remained arrrested for 24 h of incubation, while four lines (G361, Malme-3M, SK-MEL-28, and RPMI-7951) restored the cellular growth, RPMI-7951 with truncated p53 protein did so only partially. Melanoma cell lines permanently arrested and incapable of entering cell cycle also harbored markers of apoptosis. Melanocytes, used as control cells and an example of untransformed primary skin cells, took a course of their own; although lacking detectable DNA replication at 24 h after radiation they were able to recover at 30 h postirradiation and displayed only a few apoptotic cells. Most interestingly, neither growth arrest kinetics nor apoptosis pattern were dependent on p53 status. It should be remembered, however, that the UV response of wild-type p53 melanoma cell lines cannot be necessarily regarded as normal, since their p53 pathway may be defective due to accumulated wild-type p53. Nevertheless, UVC was shown to result in growth arrest in melanoma cell lines irrespective of functional p53 (IV).
Inspired by our previous findings we explored the growth arrest of cells totally lacking p53. Unsynchronized p53-/- MEFs from p53 knockout mice and p53 positive normal MEFs were exposed to 25 and 50 J/m2 of UVC. Both cell lines were arrested with similar kinetics, maximal inhibition of DNA replication occurring 6 h after insult correlating with that seen in NIH3T3 cells. However, p53-/- cells were permanently arrested, and a considerable fraction of cells (approximately 15% with 25 J/m2 and 35% with 50 J/m2) underwent apoptosis, whereas p53+/+ cells reentered the cell cycle by 24 h after UV radiation with less than 3% of cells exibiting apoptotic features (III). These findings involving both melanoma cells and p53-/- MEFs demonstrated that functional p53 is not required for UVC-induced growth arrest.
In contrast to UV radiation, functional p53 is required for successful growth arrest after g-irradiation (Kastan et al., 1991; Kastan et al., 1992; Kuerbitz et al., 1992; Dulic et al., 1994). However, several lines of evidence have demonstrated that it is not necessarily needed for apoptosis induced by g-radiation or by other signals (Clarke et al., 1993; Kelley et al., 1994). Although p53 has been suggested to be essential for growth arrest after UV radiation as well, direct evidence of its requirement has been missing. On the contrary, UVC has been shown to induce G1 growth arrest in p53 deficient Li-Fraumeni cells (Loignon et al., 1997) being in accordance with our results for p53-/- MEFs (III). Recently it was reported that HaCaT cells lacking p53 ceased to proliferate when exposed to UVC during G0/G1 but not during S or G2/M suggesting that in the absence of p53 cells failed to undergo G2/M arrest but maintained the capacity for G1 arrest (Merryman, 1999). Our findings do not support the role of p53 in immediate, UV-triggered growth arrest, since growth inhibition was intact also in cells with defective or absent p53 (I, III, IV). However, after immediate inhibition of DNA replication, p53 -/- and +/+ cells followed different pathways. In contrast to normal MEFs, radiated p53 -/- MEFs were unable to re-enter cell cycle and were prone to apoptosis (III). Since apoptosis in -/- MEFs was more prominent with higher UV doses, programmed cell death can be speculated to occur due to defective DNA repair and accumulation of unrepaired DNA lesions in cells lacking p53. On the contrary, however, DeFrank et al. (1996) found no difference in survival between p53 null and p53+/+ cells. Similarly, apoptosis in UV-exposed melanoma cell lines (IV) and hepatocytes (Bellamy et al., 1997) displayed no correlation with p53 function.
Although p53 failed to be vital for successful growth arrest after UV radiation, some other cell cycle regulators - though closely related to the p53 pathway - were found to correlate with growth inhibition temporally; most of these are discussed in separate sections and are only listed here. One of our first observations was that G1 growth arrest was associated with pRB dephosphorylation and not with p53 accumulation (I). In melanoma studies, GADD45 mRNA induction was simultaneous with growth arrest and occurred in all melanoma cell lines regardless of p53 status. Moreover, high levels of GADD45 mRNA was associated with prominent apoptosis (IV). Another target gene of p53, p21, was induced in synchronized mouse fibroblasts during G1 growth arrest (II), and p21 mRNA and protein were overexpressed in UV-radiated p53-/- MEFs (III) suggesting a role for p21 in the transmission of growth arrest. However, in melanoma cell lines growth arrest failed to parallel p21 induction (IV). Depending on cell type and genetic backround, several pathways are undoubtedly involved in UV-induced growth arrest.
Since p53 was known to accumulate in response to g- and UV radiation (Lu and Lane, 1993; Zhan et al., 1993), we examined whether pRB, an other tumor suppressor and cell cycle regulator, has any role in immediate cell response to UVC. As expected, p53 protein accumulated within 6 h after radiation with 50 to 100 J/m2 in exponentially growing mouse (NIH3T3) and human fibroblasts (CCL-137, WI-38) (I). At the same time, pRB in radiated cells was found exclusively in the dephosphorylated, active form while unradiated control cells contained also hyperphosphorylated forms of pRB. Besides fibroblasts, dose-dependent pRB hypophosporylation upon UV insult was detected in all cell lines studied, including cells with mutant or inactive p53 (I). Hypophosphorylation of pRB correlated temporally with growth arrest kinetics in unsynchronized, UV-treated NIH3T3 cells. In addition, synchronized NIH3T3 cells radiated in G1 exhibited only hypophosphorylated pRb, and, to our initial surprise, no p53 accumulation occurred during the subsequent G1 proliferative arrest. Consistent with general growth arrest in unsynchronized cells, specific G1 arrest and pRB dephosphorylation attenuated with identical kinetics (I).
pRB hypophosphorylation is frequently detected in cells growth arrested by diverse genotoxic or growth restricting agents, but our study was one of the first to report its involvement in UV-induced growth arrest (I; Medrano et al., 1995). All DNA damage types do not lead to pRB hypophosphorylation; g-irradiation failed to dephosphorylate pRB in p53-/- fibroblasts suggesting that the nature of the damage influences pRB phosphorylation changes (Dulic et al., 1994). This approach with synchronized cells allowed us to study pure G1 or S phase arrest and the functions of cell cycle regulators pRB and p53 in those phases.
In theory, pRB dephosphorylation in radiated cells could have been a secondary change caused by p53-mediated G1 growth arrest. However, two findings in our studies favored the p53-independent role of pRB in growth inhibition: first, pRB hypophosphorylation - and growth arrest - occurred also in cell lines harboring mutant or inactivated p53, and secondly, the kinetics of growth arrest in unsynchronized NIH3T3 cells correlated with that of pRB hypophosphorylation (I). Rephosphorylation of pRB 12 h after UV treatment, when DNA replication was still dampened, preceded, not followed, the entry of cells to the cycle. Moreover, in synchronized UV-treated NIH3T3 cells G1-phase growth arrest endured as long as pRB remained hypophosphorylated, but p53 accumulation was delayed until the cells recovered from G1 arrest and entered S phase (I).
These results thus not only demonstrate the role of pRB in UV-induced growth arrest in G1 but also question the participation of p53 in that particular phase (I). Our observation of pRB as a central mediator of the G1/S checkpoint proved to be in accordance with recent findings that exposure of RB-/- MEFs to various DNA damaging stimuli including UV, ionizing radiation, and chemotherapeutic agents resulted in defective cell cycle arrests (Harrington et al., 1998). It is particularly interesting that also g-irradiation, which has been traditionally strongly coupled to p53 function, failed to cause G1/S arrest in the absence of pRB despite increased levels of p53 and p21 (Harrington et al., 1998). On the other hand, polyamine analog treatment has been shown to raise the level of hypophosphorylated pRB, correlating temporally with G1 growth arrest as in our studies, but that the rise was preceded by p53 accumulation (Kramer et al., 1999). Thus, the nature of DNA damage caused by numerous growth suppressive agents is one of the most important determinants when cell cycle responses are interpreted.
NIH3T3 cells synchronized to and radiated in S phase remained arrested in S, and contained both hypo- and hyperphosphorylated pRB, whereas pRB in S-phase control cells consisted solely of hyperphosphorylated forms (I). Theoretically, it seems unlikely that pRB could influence S-phase arrest in S-phase cells, because phosphorylation events inactivating pRB have already taken place in late G1 or G1/S transition during the previous cycle. However, the existence of both phosphorylation forms of pRB in radiated S-phase cells as well as existence of only hypophosphorylated pRB in exponentially growing cells including cells in S (I), raises the question of UV-induced dephosphorylation of pRB during the S phase. Indeed, it was recently found that underphosphorylated pRB was present also in cells growth arrested by sodium butyrate in the S or G2/M phases (Yen and Sturgill, 1998), and in all cell cycle phases after hypoxia treatment (Danielsen et al., 1998), suggesting that the growth controlling capability of RB may not be limited to the G1 phase alone.
p53 accumulation in cell lines with wild-type p53
In response to a UVC of 50 J/m2, accumulation of p53 in exponentially growing NIH3T3 and CCL-137 fibroblasts was observed by immunoblotting and immunofluoresence staining between 4 and 30 h after radiation, after which elevated levels of p53 gradually decreased (I; data not shown). Similarly, p53 accumulation occurred within 6 h after UV treatment in normal MEFs (III) and in melanoma cell lines expressing wild-type p53 (IV). On the contrary, cells harboring mutant (IV) or inactivated p53 (I) already expressing high levels of stabilized p53 failed to further accumulate it in response to UV. In most experiments cells were exposed to a UVC of 50 J/m2, representing a sublethal dose with less than 5% of NIH3T3 cells undergoing apoptosis after 36 h incubation (unpublished data). Doses as high as 200 J/m2 seriously damaged cells resulting in enhanced apoptosis and lower accumulation of p53 (I).
Accumulation of p53 upon UV radiation was first detected by Maltzman and Czyzyk as early as 1984 (Maltzman and Czyzyk, 1984). The kinetics of p53 accumulation in NIH3T3 was in agreement with findings of others (Lu and Lane, 1993; Zhan et al., 1993), and differed from that seen after ionizing radiation. g-irradiation causing direct double-strand breaks is a more rapid and more efficient inducer of p53 accumulation than UV radiation (Lu and Lane, 1993). In general, agents leading to DNA strand breaks, such as topoisomerase I inhibitors, g-irradiation, drugs intercalating to DNA, and DNA nucleases are most potent inducers of p53 (Nelson and Kastan, 1994) . The C-terminal domain of p53 is known to bind to single-stranded DNA ends (Bakalkin et al., 1994) and DNA lesions (Lee et al., 1995) leading to an activation of transcriptional activity of p53. Since UV radiation causes DNA strand breaks only during repair processes, it is understandable that p53 response to UV radiation is kinetically slower but longer lasting than to ionizing radiation. Compared to ionizing radiation, a more extended accumulation of p53 after UV may correlate with the longer persistence of DNA lesions and ongoing repair processes and may hence be dependent on UV dose.
p53 accumulation requires replication of damaged DNA
Although G1-arrested cells lacked p53 stabilization, this was, however, detected in exponentially growing fibroblasts (I), indicating that cells in cell cycle phases other than G1 were responsible for elevated p53 levels. Indeed, in synchronized NIH3T3 cells p53 stabilization after UV radiation of 50 J/m2 was detected in two circumstances, both involving the S phase. While absent during G1 arrest, p53 accumulation was detected when G1-arrested cells recovered from UV insult and entered the S phase (I). Alternatively, if cells were UV treated while in the G1/S-phase border or S, p53 stabilized without delay and correlated with S-phase arrest (I, II). These observations are very suggestive that not the stage in the cell cycle when the insult is encountered but the replication of damaged DNA could be a prerequisite for p53 stabilization. Furthermore, regarding p53 accumulation, p53 seemed to play an essential part rather in S and G2/M arrest than in G1 arrest.
To examine the temporal correlation between p53 stabilization and DNA replication in more detail, a double block system was used to synchronize NIH3T3 cells either to G1, the G1/S border, early- or mid-S phase (II). The 5-BrdUrd incorporation percentage of cells UV treated in G1 or the G1/S border and incubated for 6 h correlated with accumulation of p53 analyzed by both immunofluoresence staining and western blotting; p53 level was low or undetectable when less than 5% of radiated cells replicated DNA, and p53 level was high (50% of cells p53 immunopositive) when 70% of cells were replicating at the time of UV insult (II). To further demonstrate that p53 accumulation was dependent on DNA replication rather than on the S phase itself, HU was added to the medium after exposure to UV. HU treatment was able to abolish the p53 accumulation in S-phase cells, but interestingly, it failed to decrease p53 levels in cells radiated at the G1/S border; on the contrary, even higher amount of p53 was detected in the presence of HU in these cells (II).
Studies examining the cell cycle phase-dependent regulation of p53 accumulation by UV treatment had not been carried out before. We demonstrated that p53 accumulation after UV was cell cycle phase-dependent occurring only in cells radiated in S or G1/S phases or entering S after G1 arrest, and that accumulation correlated with DNA replication (I, II). Hypoxia, an other effective inducer of p53, has been later shown to induce p53 accumulation more efficiently in the S than in G1 or G2/M phases (Danielsen et al., 1998). The absence of stabilization of p53 during G1-phase growth arrest was unexpected (I), because p53 has been considered to be a G1 checkpoint controller. Assuming that p53 accumulation is dependent on recognition of DNA strand breaks, UV-type DNA lesions in G1-arrested cells may not be properly recognized by p53. In such a situation G1 arrest is perhaps maintained by other cell cycle regulators, such as pRB.
Replication of damaged DNA either in cells UV radiated in S or entering it after G1 arrest seemed to be a precondition for p53 accumulation. The replication dependency of p53 accumulation by UV is consistent with previous findings (Nelson and Kastan, 1994), but detailed analysis of p53 accumulation in different cell cycle stages had not been reported previously. Besides UV radiation, topoisomerase I inhibitor camptothecin requires DNA replication for p53 stabilization (Nelson and Kastan, 1994) suggesting that if a DNA damaging agent does not directly cause DNA strand breaks, the lesions should be at least exposed during replication. Damage signals evoked by UV in G1 may be recognized by p53 only in replicating DNA when it is processed to single-stranded DNA.
p53 accumulation in cells irradiated at the G1/S border was different from that seen in G1- or S-phase cell populations. Irradiated G1/S phase cells displayed accumulated p53, but in contrast to S-phase cells, p53 levels remained high in the presence of HU. However, G1/S time point fails to represent a normal cell cycle stage, because G1 cells that would otherwise continue to S are forced by HU to stop at the threshold of S. At the time of radiation these cells have opened replication forks, but a lack of nucleotides inhibits DNA replication but not NER. At this stage p53 stabilization independent of replication may be based on interaction of p53 with other cellular proteins. In fact, it has been later shown that p53 accumulation in G1/S, as well as in other phases, requires protein synthesis, since treatment with cycloheximide, an inhibitor of protein synthesis, totally prevented p53 accumulation in the presence and absence of HU (Pitkänen et al., 1998). It should be pointed out that up to the present, these findings apply only to these UVC-radiated mouse fibroblasts.
Regulation of p53 transactivation in UV-radiated cells in different cell cycle phases turned out to be different from that of p53 accumulation. As in earlier studies, NIH3T3 mouse fibroblasts were exposed to 50J/m2 of UVC in diverse stages of the cell cycle, and transactivation determinants, such as DNA binding activity, transcriptional activity as determined by chloramphenicol acetyltransferase (CAT) assay, and induction of target gene mRNAs, were compared to growth arrest kinetics and p53 accumulation. First, however, protein levels of p21 CKI were determined in cells irradiated in the G0, G1, G1/S, and S phases; G1/S represented the time point where G1 cells were inhibited by HU from entering S. In contrast to p53 protein which lacked accumulation in the G0 and G1 phases, p21 protein was induced in these and in all other stages (II). After UV radiation HU treatment abolished p21 protein induction in S but not at G1/S, a situation identical to that of p53 accumulation (II).
Since p21 protein was induced both in the presence and absence of accumulated p53, mRNA levels of two p53 target genes, p21 and GADD45, were measured in different cell cycle phases with and without HU. p21 and GADD45 mRNA levels were induced by UV radiation in all cell cycle phases. Compared to non-irradiated cells maximum induction of p21 mRNA ranged from 8-fold in G1 to 3-fold in S, GADD45 induction being 3-fold and 2-fold, respectively. Inhibition of replication by HU abolished mRNA inductions of both p21 and GADD45, including G1/S-phase cells (II).
The capability of p53 to specifically bind to the p53-responsive element and induce transcription was explored using p53-responsive reporter constructs. NIH3T3 cells were stably transfected with the reporter plasmid construct PG13-CAT, and the effect of UV on transcriptional activity of p53 was measured as an increase in CAT activity. It reached on average a 6-fold increase in G0/G1 UV-treated cells and 3-fold increase in G1/S- and S-phase cells without any inhibitory effect by HU. Transfection with MG15-CAT, containing mutant p53-binding sites, resulted in negligible CAT activities (II).
Finally, the sequence specific DNA-binding activity of p53 was assayed by EMSA. UV-activated p53 protein binds to a radiolabeled oligonucleotide containing consensus p53-binding site from p21 promoter, and the number of complexes containing active p53 was estimated after separation in polyacrylamide gel. EMSA results were consistent with the findings in CAT-assay: specific DNA binding activity of p53 was increased by UV in all cell cycle phases despite the presence of HU (II).
These results indicate that p53 transactivation is dissociated from p53 accumulation, since transcriptional activition was observed by all three methods in the G0 and G1 phases in the absence of stabilized p53. Moreover, transactivation appeared to be clearly independent of the cell cycle phase, and, according to CAT-assay and EMSA, independent of DNA replication, because HU treatment in the S phase had no effect on DNA binding activity of p53. However, p21 and GADD45 mRNA as well as p21 protein were not induced in S in the presence of HU suggesting a role for accumulated p53 in these processes. On the other hand, neither the absence of replication nor p53 accumulation impaired the increase in p21 and GADD45 mRNA levels in G1 phase (II).
The three ways employed to analyze the transcativation function of p53 have their limitations. Mobility shift assay displays only the amount of p53 protein bound to the p53 consensus binding site. Although being highly specific for p53, it fails to measure activation of transcription. Increase in CAT activity, on the other hand, has been preceded by DNA binding and the activation of transcription by p53. Inductions of target gene mRNA and protein are close to physiological signals, but they are subject to diverse posttranscriptional and posttranslational modifications independent of p53. Thus, EMSA and CAT-assay are highly specific for p53 but may be too simplistic, whereas evaluation of target gene induction does not require artificial interventions such as transfections, but may in turn be a too complicated method to explore p53 function alone. Nevertheless, information combined from these assays provide indications of the regulation of transcriptional activity of p53.
Traditionally, p53 accumulation has been considered to be a consequence of p53 activation due to inhibition of protein degradation, and accumulation and transactivation have been regarded as simultaneous and interdependent events. However, coinciding with our report, discordance has been found between transcriptional activity and accumulation of p53 at low UV doses; 10 J/m2, but not higher doses, resulted in enhanced transcriptional activity without an increase in p53 protein (Lu et al., 1996). Although p53 accumulation after UV radiation seems to require replication of damaged DNA, other signals must be involved for its transcriptional activation. Regulation of p53 accumulation may depend on the ability of mdm2 to bind and target p53 for destruction, whereas other events, particularly protein modification by phosphorylation, may regulate the transcriptional activity of p53. In fact, mice lacking DNA-PK, a kinase phosphorylating Ser15 and 37, showed normal p53 accumulation but no activation of p53 by ionizing radiation (Woo et al., 1998). Discordance between accumulation and transactivation suggests that p53 accumulation is involved in recognition of unrepaired DNA lesions in the S phase. Supporting this hypothesis, NER-deficient cells accumulated p53 at much lower UV doses and exhibited a more prolonged response compared to normal cells or cells with restored NER activity, suggesting that p53 accumulation does not correlate with the NER process itself but with the presence of unrepaired cyclobutane dimers (Dumaz et al., 1997; Dumaz et al., 1998). The finding that p53 is transcriptionally active - although not accumulated - throughout the cycle, support its role as a checkpoint controller in all phases of the cycle including G1. Whether p53 accumulation alone has a function independent of p53 transactivation still lacks direct evidence and remains to be clarified.
Regulation of p21 by UV radiation in synchronized NIH3T3 cells showed many interesting aspects. Contrary to p53 accumulation, p21 protein was induced in all cell cycle phases (II) suggesting its participation in both G1- and S-phase arrests via its Cdk-inhibiting or PCNA-binding functions (Chen et al., 1995; Luo et al., 1995). Although the DNA-binding activity of p53 was not dependent on DNA replication, p21 protein, its mRNA and GADD45 mRNA were upregulated in the S phase only in replicating cells (II). It seems that target gene induction requires more than binding of activated p53 to its responsive element and that these other events, whether posttranscriptional modifications or stabilization of p53, are prevented by HU. Alternatively, UV radiation can induce p21 and GADD45 independently of p53, but p53 may still augment this effect. p53-independent transcriptional regulation of p21 gene is frequently observed, although most commonly in conditions other than genotoxic stress (Chin et al., 1996; Somasundaram et al., 1997; Li et al., 1998; Moustakas and Kardassis, 1998). Additionally, cells radiated at the G1/S border showed peculiar features. HU treatment prevented p21 mRNA induction but it had no effect on p21 protein induction (II). This dissociation of p21 protein and mRNA induction has been also demonstrated in breast epithelial cells after ionizing radiation (Gudas et al., 1995) and in growth arrested fibroblasts after serum stimulation (Macleod et al., 1995), where high p21 protein levels have been detected without an increase in p21 mRNA.
The observed discordance between the induction of p21 and activation or accumulation of p53 (II) were suggestive of p53-independent regulation of p21 after UVC radiation. Similarly, a melanoma cell line carrying mutant p53 was able to induce p21 mRNA in response to UV (IV). Since mutant p53 proteins may have retained some functions of wild-type p53, p21 regulation was studied in the total absence of p53 in p53-/- MEFs (III).
As described in the context of growth arrest, UVC responses of low passage p53-/- and p53+/+ MEFs were compared with respect to growth arrest and p21 induction. UVC doses of 10 to 50 J/m2, that are known to be sufficient for induction of p53 transcriptional activity, resulted in rapid increase in p21 protein in both cell lines. p53 +/+ cells, expressing higher p21 basal level than -/- cells, retained high p21 protein induction with 50 J/m2, whereas the same dose in -/- MEFs yielded an attenuated p21 induction. The p53 accumulation pattern in p53+/+ MEFs followed the course observed in NIH3T3 cells and paralleled p21 induction (I, III).
Because elevation of p21 protein levels can be caused by transcriptional activation or by posttranslational modifications (Timchenko et al., 1996), mRNA levels in both cell types were evaluated. p21 mRNA was induced during the first 6 h after UV also in the absence of p53. When compared to unirradiated cells an over 5-fold induction was obtained within 16 h in both cell lines (III).
The transcriptional activation of p21 by UV radiation was further examined by studying the activation of p21 promoter. p53-/- cells were transiently transfected with p21 promoter deletion constructs containing luciferase gene as a reporter. Luciferase activity was measured from radiated (25 J/m2) and control cells 24 h after UV. Transfection with full length p21 2.4 kb promoter construct, as well as with three shorter constructs lacking either one or two p53 binding sites, resulted on the average in 6-fold induction of luciferase activity by UVC confirming the p53-independent transcriptional activation of p21 by UV radiation (III).
As p53-/- cells are incapable of activating p53 binding sites we examined whether wild-type p53 in NIH3T3 cells was able to further increase the p21 promoter activity. UVC treatment led to 4-fold increase in reporter gene activity with both full-length p21 promoter and with a construct devoid of major p53 consensus binding site. Moreover, a construct with 400 bp 5´ deletion resulted in even greater induction of over 6-fold suggesting a negative regulatory element in region between -2.25 kb and -2.0 kb relative to the trancription initiation site. Deletion constructs containing DNA 500 to 110 bp upstream of the transcription initiation site showed a somewhat decreased UV inducibility, while only the shortest construct containing the last 61 bp totally abolished the UV-induced luciferase activity (III).
UV-induced transcriptional activation of p21 appeared to be most strongly regulated by a p21 promoter region between -110 bp and -61 bp relative to the transcription initiation site. Since the -110 bp area comprises five Sp1-binding sites, Sp1-binding site mutation constructs containing 93 bp were developed (Kivinen et al., 1999) and used to explore the effect of these sites on p21 induction. Of five Sp1-binding sites mutation of sites 2 and 4 decreased the UV inducibility, while the combined mutations of the two sites resulted in a complete loss of UV response. EMSA studies confirmed the sequence specific binding of Sp3 transcription factor to Sp1-binding site. In conclusion, UV-induced transcriptional activity of p21 was linked to two Sp1-binding sites in the vicinity of transcription initiation site, while a weaker regulation area was present between -1300 and -500 bp relative to transcription start. (III)
Hence, these studies indicated that UVC radiation induces p21 protein and mRNA expression and transcriptionally activates p21 promoter constructs also in the absence of p53. While examples of p53-independent induction of p21 are abundant (see Review of the literature), only a few reports have connected it to UV radiation. p53 independency of p21 promoter activity was demonstrated by two ways: firstly, p21 promoter activity was found to be induced by UV in p53-/- cells, and secondly, deletion constructs devoid of both p53 binding sites showed normal promoter activity in NIH3T3 cells carrying normal p53 (III).
The induction of p21 by ionizing radiation, drugs, and other agents that cause double-strand breaks depends on intact p53 (Dulic et al., 1994; El-Deiry et al., 1994; Macleod et al., 1995). While UVB/A radiation has been shown to enhance p21 expression only in the presence of p53 (Liu and Pelling, 1995), UVC has been able to induce p21 protein in Li-Fraumeni cells (Loignon et al., 1997), in melanoma cells expressing mutant p53 (IV), and in p53-/- fibroblasts (III). Additionally, our study extended the findings to the transcriptional level, and was the first to determine the p21 promoter elements involved in p21 regulation after UV treatment (III). However, in contrast to our findings, UVC was recently found to be unable to induce p21 protein or mRNA in p53-/- MEFs, and, additionally, p21 induction in normal cells occurred through stabilization of p21 mRNA (Gorospe et al., 1998). Although Gorospe et al. (1998) failed to observe elevation of p21 protein levels in p53-/- cells, p21 mRNA was induced 3-4 fold, as compared to 25-fold increase in p53+/+ MEFs, the authors disregarded this relatively low level of induction in p53-/- cells. In our experiments both cell types showed equally strong p21 mRNA induction (maximum fold 5.2 and 6.7 in +/+ and -/- cells, respectively). Moreover, using the same p21 promoter luciferase construct, the authors observed only about 2-fold induction in luciferase activity while in our study the induction was 5-fold, but, most importantly, the induction was the same in -/- and +/+ cells (Gorospe et al., 1998). So, due to high p21 mRNA induction and low p21 promoter activity in normal cells the authors concluded that p21 induction resulted from stabilization of p21 mRNA rather than transcriptional activation (Gorospe et al., 1998). The stabilization of p21 mRNA is not an unusual phenomenon (Akashi et al., 1999), even if the regulation of the stability of mRNA is poorly understood. Of course, it cannot be ruled out that mRNA stabilization may participate in p21 induction in our cells as well, although p21 mRNA and promoter activity were equally induced (III); studies determining the halflife of mRNA would illuminate this aspect.
The regulatory element in the p21 promoter that totally abolished the UV response when absent was mapped between -110 bp and -61 bp proximal of the transcription start and was linked to two Sp1 binding sites (sites 2 and 4) within the region (III). This area seems to be of special importance in transcriptional activation of p21 promoter, since numerous p21 inducers, including TGF-ß (Datto et al., 1995), progesterone (Owen et al., 1998), okadaic and phorbol ester (Biggs et al., 1996) , nerve growth factor (Yan and Ziff, 1997) and ras (Kivinen et al., 1999) are regulated by these same or neighboring Sp1-binding sites. A negative regulatory element -2.0 kb distal relative to transcription initiation site was found in NIH3T3 but not in p53-/- cells suggesting also a p53-mediated negative regulation of p21 promoter activity.
In this study UV-induced p21 expression paralleled growth arrest kinetics, its role as proliferation controller being supported by a rapid response regulated at transcriptional level both in the presence and absence of p53 (III). Accordingly, flavone-induced G1 growth arrest induced by flavone was shown to depend on p53-independent transcriptional activation of p21 (Bai et al., 1998). It is possible that in the absence of p53 p21 may in part compensate for the function otherwise accomplished by p53. On the other hand, it has been assumed that despite p53 independent regulation, p21 induction may be enhanced by p53. For example, PMA was shown to induce p21 independent of p53 but the induction was increased manyfold in the presence of p53 (Akashi et al., 1999) . For these reasons it was surprising that p21 was not only induced in the absence of p53, but that the level of protein and mRNA induction was completely independent of p53. However, two events were affected by a lack of p53: firstly, p53-/- cells were not able to reenter the cell cycle and many underwent apoptosis possibly reflecting deficient repair, and secondly, high UVC doses (50 J/m2) caused attenuated induction of p21 protein. These findings suggest that p53 may be needed for full p21 induction after massive DNA damage and that p21 may function as a preventor of apoptosis.
The main results of UV-radiation induced activities of p53, pRB, and p21 in mouse fibroblasts, particularly their cell cycle phase and replication dependence, are summarized in the following figure.
The low frequency of p53 mutations (Hartmann et al., 1996) and high basal expression of wild-type p53 often encountered in melanomas (Lassam et al., 1993; Floerens et al., 1994; Floerens et al., 1995) aroused our interest in exploring the UVC response of seven melanoma cell lines. Based on previous reports (Montano et al., 1994; Bae et al., 1996) as well as on our analysis of all p53 coding exons by RT-PCR and direct sequencing (IV), four melanoma cell lines (A-375, Malme-3M, WM239, and G361) appeared to harbor wild-type p53 and three cell lines mutant p53, including a nonsense mutation leading to a truncated p53 protein (RPMI-7951), and two mutations in p53 DNA-binding domain (SK-MEL-2, SK-MEL-28). Compared to melanocytes expressing low or negligible levels of p53 protein, all untreated melanoma cell lines, regardless of the p53 status, expressed high levels of p53 protein with the exception of RPMI-7951 which lacked detectable p53. Accumulation of both wild-type and mutant p53 was detected by western blotting and immunostaining of fixed cells, the latter revealing also the nuclear localization of p53 protein. p21 protein was expressed only in melanoma cells with wild-type p53, but its level failed to correlate with that of p53. Similarly, although mdm2 mRNA levels on the average were higher in cells expressing wild-type than in cells expressing mutant p53, high mdm2 levels were not associated with high p53 protein levels in individual melanoma cell lines (IV).
UVC treatment of 50 J/m2 resulted in further accumulation of p53 only in wild-type cell lines. Although the kinetics of p53 stabilization in melanoma cell lines was consistent with that in NIH3T3 cells being detectable within 6 h incubation, the elevation in most wild-type cell lines was quite modest and less than in melanocytes. As previously shown (I), mutant p53 was not subject to further stabilization (IV).
To demonstrate transcriptional activity of p53, UV-responsiveness of three p53 target genes was examined. Induction of p21 protein was observed only in melanoma cell lines expressing normal p53, whereas induction of p21 mRNA occurred also in one mutant cell line (SK-MEL-28) lacking an increase in p21 protein. Interestingly, p21 protein and mRNA levels were not elevated until 16 to 24 h after exposure to UV reflecting a slower kinetics than seen in human and mouse fibroblasts (I, II). The GADD45 response, on the other hand, was rapid and simultaneous with growth arrest. GADD45 mRNA induction was detected in all UV-treated melanoma cells studied, maximal fold induction ranging from 2.8 to 8.8 independent of p53 status and p21 induction pattern. Moreover, high induction of GADD45 mRNA (4-fold or more) correlated with prominent apoptosis, which, like growth arrest, occurred irrespective of p53 status (IV). The UV response of mdm2 mRNA was consistent in all melanoma cells, with the exception of RPMI-7951: an initial decrease in mdm2 levels 6 h after UV was followed by slight increase 24 h after UV treatment. In RPMI-7951 cells this decrease was absent and maximal induction took place already 16 h after UV. The inductions were small, and no significant difference was detected between cells expressing wild-type or mutant p53. Thus, regulation of target gene inductions by UV - though different from each other - appeared to be independent of p53 function (IV).
Since DNA damage can stimulate p53 effector genes also in a p53-independent manner, EMSA analysis was undertaken to examine the UV-induced specific DNA-binding activity of p53. Mutant cell lines were also studied in order to find out whether mutant p53 could induce p21 mRNA expression detected in SK-MEL-28. The specificity of binding of p53 to DNA was confirmed by using oligonucleotides containing specific p53-binding sites from either p21 or GADD45 promoters and by detecting supershifts of DNA-p53-antibody -complexes in the presence of p53 antibodies. As expected, UV-stimulated DNA binding activity of p53 to both p21 and GADD45 probes was detected in A-375, Malme-3M, and WM239 cell lines expressing wild-type p53 and was absent in SK-MEL-28, SK-MEL-2, and RPMI-7951 cells carrying mutant p53 (IV; data not shown). Surprisingly, however, G361 cells exhibiting high levels of wild-type p53 completely lacked UV-stimulated DNA-binding activity with all probes even in the presence of the antibody PAb421 that is known to stimulate the binding. This suggests that the UV response in G361 occurs independent of p53, or at least independent of its transactivation property. It should be also noted that when observed, DNA-binding activity was stimulated already 6 h after DNA damage thus hardly contributing to p21 induction detected only after 16 h incubation (IV).
The expression of some other cell cycle regulators, that are known to be involved in the pathogenesis of malignant melanoma, were also studied. pRB expression was observed in all melanoma cell lines, whereas only cell lines carrying mutant p53 showed p16 protein and mRNA expression (unpublished observation); the lack of p16 expression in wild-type p53 cell lines is due to premature stop codon in A-375 (Ohta et al., 1994), or deletion of exon a in WM239 and G361 cell lines (Ohta et al., 1994). In addition to the p53 mutation, SK-MEL-28 cell line harbors Cdk4 mutation (Tsao et al., 1998). It can be speculated that in the absence of p53 mutation, inactivating mutations of p16 or other targets of RB pathway have been selected.
Human melanoma represents one of the few malignancies that seldom select for p53 mutations (Hartmann et al., 1996). Although primary melanomas rarely contain p53 mutations, they are more frequently detected in melanoma cell lines (Albino et al., 1994; Bae et al., 1996) corresponding to mutation frequency of 43% among our seven randomly selected melanoma cell lines. On the other hand, all cell lines expressing wild-type p53 showed abnormal stabilization of normal p53. Accumulation of wild-type p53 was neither due to extranuclear localization nor due to overexpression of mdm2, since mdm2 levels failed to systematically accompany p53 accumulation although they were generally higher in the wild-type than in mutant p53 cells (III, see also " Review of the literature"). Similarly, although all untreated melanoma cell lines harboring normal but stabilized p53 were expressing p21 protein, the levels of p53 and p21 did not correlate (IV). This is in contrast to non-Hodgkin´s lymphomas where overexpression of wild-type p53 was associated with accumulation of p21 (Maestro et al., 1997). Teratocarcinomas, which never contain p53 mutations, also exhibit high levels of nuclear wild-type p53. This protein is transcriptionally inactive until activated by DNA damage or by agents inducing differentiation which leads to transcriptional induction of target genes and p53-mediated apoptosis (Lutzker and Levine, 1996).
Our results indicate that despite the wild-type sequence of p53 its response to UVC radiation differed in several aspects from that of normally functioning p53. Irrespective of this, or due to it, the response patterns of mutant and wild-type p53 expressing melanoma cell lines did not show any major differences. Although wild-type p53 accumulated in response to UV as expected, the p53 pathway downstream of accumulation displayed several specific features: induction of p21 protein and mRNA expressions by UV radiation were delayed, 2) inductions of the target genes p21, GADD45, and mdm2 were dissociated from p53 regulation, 3) one wild-type p53 melanoma cell line had no UV-stimulated DNA-binding activity, and 4) of the cell cycle regulators studied only GADD45 expression correlated with growth arrest and apoptosis.
UVC-stimulated p21 protein and mRNA inductions appeared to be independent of p53 function, as accumulation and DNA-binding of p53 took place considerably earlier than p21 induction, and p21 mRNA elevation occurred with the same kinetics also in SK-MEL-28 cell line harboring mutant p53. However, p53 may still play a role in the full induction of p21, since p21 protein was not induced in SK-MEL-28 cells despite an elevation in the level of mRNA encoding it. This finding is consistent with the fact that p53-/- MEFs, but not p53+/+ MEFs, had an attenuated p21 protein response to high UVC dose (III), providing evidence that p53 activity is required for some aspects of p21 regulation. The requirement of p53 for growth arrest and p21 induction is certainly affected by the nature of DNA damage. Polyamine analogs usually cause G1 and G2/M arrests in p53 proficient cells, but treatment of SK-MEL-28 cells carrying mutant p53 have been shown to abolish G1 arrest and p21 induction (Kramer et al., 1999).
In contrast to stimulation of p21 expression, GADD45 mRNA induction occurred rapidly within 6 h after UV treatment in all melanoma cell lines irrespective of p53 status. p53-independent activation of GADD45 in response to UV has been detected previously (Kearsey et al., 1995). Since activation of GADD45, and not of p53, correlated with both growth arrest and apoptosis, GADD45 appeared to serve as a rapid UV response target (IV). Contrary to our results, however, GADD45 has been proposed to protect cells from UV-induced apoptosis (Smith et al., 1996). Interestingly, GADD45 induction in g-radiated melanoma cells expressing wild-type, accumulated p53 was very different from that after UV radiation. Ionizing radiation resulted in strong G1 growth arrest, prolonged p21 induction, and low or absent GADD45 induction (Bae et al., 1996). Taken together this suggests that GADD45 activation would be vital in UV response and regulated independent of p53, whereas it can be neglected after ionizing radiation and is probably under the control of intact p53.
Of the p53 effector genes mdm2 has recently received most interest due to its role as a negative regulator of p53, a function which is affected by several different p53 activating pathways. The regulation of mdm2 by UV radiation elicits some special features. Compared to p53 accumulation, low UV doses lead to rapid mdm2 induction, while with high UV doses mdm2 response is delayed, the decrease being p53-independent and the increase p53-dependent (Perry et al., 1993; Wu and Levine, 1997). The mdm2 response of melanoma cells studied here seemed to follow that pattern, but the increase in mdm2 mRNA levels after an initial decrease was detected also in two mutant p53 cell lines, one of them (RPMI-7951) displaying the highest mdm2 induction of all. Thus, UV-stimulated induction of mdm2, as well as that of p21 and GADD45, appeared to be independent of functional p53.
Although accumulation and DNA-binding activity of p53 were upregulated with rapid kinetics in all melanoma cell lines expressing wild-type p53 except for G361, none of the UV responses - growth arrest, apoptosis, or target gene inductions of p21, GADD45, or mdm2 - appeared to be regulated by p53. Despite the DNA-binding activity of p53 observed in the majority of cell lines, this suggests that abnormally stabilized wild-type p53 in melanoma cell lines is functionally inactive or possesses alternative functions not characterized here. The functional inactivity downstream of sequence-specific DNA binding would explain tolerance to high p53 levels and why mdm2 levels are not raised to downregulate p53. As in the case of p53 mutation, a cell with inactive p53 is not able to sense the effects of p53 and responds by decreasing p53 degradation leading to accumulation of inactive, though wild-type, p53 protein.