Signalling responses linked to betulinic acid-induced apoptosis are antagonized by MEK inhibitor U0126 in adherent or 3D spheroid melanoma irrespective of p53 status

Abstract

MEK1/2 inhibitors like U0126 can potentiate or antagonize the antitumor activity of cytotoxic agents such as cisplatin, paclitaxel or vinblastine, depending on the drug or the target cells. We now investigated whether U0126, differentially regulates melanoma signaling in response to UV radiation or betulinic acid, a drug lethal against melanoma. This report shows that U0126 inhibits early response (ERK) kinase activation and cyclin A expression in wt p53 C8161 melanoma exposed to either UV radiation or betulinic acid. However, U0126 does not protect from UV damage, but counteracts betulinic acid-mediated apoptosis in the same cells. Protection from the latter drug by joint treatment with U0126 was also evident in wt p53 MelJuso melanoma and mutant p53 WM164 melanoma. The latter cells were the most responsive to betulinic acid, showing a selective decline in the cdk4 protein, without a comparable change in other key cell cycle proteins like cdc2, cdk2, cdk7 or cyclin A, prior to apoptosis-associated PARP fragmentation. Laser scanning cytometry also showed that betulinic acid induced a significant increase in chromatin condensation in WM164 melanoma irrespective of whether they were in adherent form or as multicellular spheroids. All these betulinic acid-induced changes were counteracted by U0126. Our data show for the first time that (a) cdk4 protein is an early target of betulinic acid-induced apoptosis and (b) unrestricted ERK signaling favours betulinic acid-induced apoptosis, but this is counteracted by U0126, partly through counteracting chromatin condensation and restoring Akt activation decreased by betulinic acid treatment. © 2005 Wiley-Liss, Inc.

Constitutive activation of p42/p44 MAPK (mitogen-activated protein) kinases or early response (ERK) kinases occurs in tumor progression in prostate carcinoma and melanoma.1, 2 On the basis of the importance of constitutive ERK activation in tumors, pharmacological inhibitors of MAPK kinases MEK1/2 that activate MAPK/ERK have attracted attention since some of them inhibit ERK activation and block human colon carcinoma growth in vivo.3, 4 Pharmacological inhibition of the MAPK kinases MEK1/2 was recently reported as a useful strategy for killing melanoma cells.5 However, others have shown that the MEK inhibitor U0126 can normalize the morphology of adherent Ki-ras-transformed rat fibroblasts,6 inhibit invasion of A375 melanoma through Matrigel7 and promote B16 melanoma differentiation,8 without exerting a cytotoxic effect on these tumor cells. Moreover, pharmacological inhibition of ERK activation by U0126 or by PD98059 was reported to strongly attenuate apoptosis induced by high dose etoposide, adryamicin, or UV radiation.9 Enforced activation of ERK by overexpression of MEK-1/Q56P also sensitized cells to DNA damage-induced apoptosis.9 In contrast, in vivo studies involving injection of TPras melanoma cells into SCID mice resulted in the development of invasive tumors. Application of the PI3K inhibitor Ly294002 significantly reduced tumor size, without a comparable effect in mice treated with the MEK1/2 inhibitor U0126.10 In nonadherent myeloid leukemia ML-1 cells, vinblastine-mediated apoptosis was enhanced by concurrent ERK inhibition using 50 μM PD98059. However, the vinblastine-PD98059 combination that potentiated apoptosis in ML-1 cells did not sensitize other lymphoid cells like U937 and Jurkat, to the combination of vinblastine and PD98059.11 These results implied that MEK inhibitors may potentiate lethality of particular cell types in response to some specific drugs. To confirm this interpretation, it was reported that apoptosis induced by paclitaxel in ML-1 cells was not potentiated by the MEK inhibitor PD98059.11 In contrast, inhibition of MEK/ERK with 10 μM U0126 enhanced paclitaxel induced apoptosis in breast (BT474) and lung (H358) carcinoma cell lines.12 Moreover, ERK activation was induced by cisplatin in ovarian carcinoma, and cisplatin cytotoxicity was potentiated by pharmacological MEK inhibitors.13 However, in cervical carcinoma HeLa cells, cisplatin also activated the ERK pathway but MEK1/2 inhibitors blocked cisplatin-induced apoptosis.14 All these contradictory results involving the use of the same MEK inhibitors against different tumors suggest that response to MEK inhibitors may protect, kill or fail to act on tumors, depending on the apoptotic inducer, microenvironment and genetic programming of the tumor cell. Hence we now used the human C8161 melanoma15 to investigate whether inhibition of ERK activation by U01266, 7 is comparably effective against apoptosis induced by UV radiation16, 17 or betulinic acid.18 The latter is a potentially useful antitumor drug with limited toxicity toward normal cells19 and with a preferential cytotoxicity toward metastatic C8161 melanoma variants.20 Betulinic acid induces the formation of reactive oxygen species, which is essential for its apoptotic effect21; UV radiation also causes oxidative imbalance.17 One of the consequences of oxidative stress is induction of ERK/MAPK phosphorylation,16 which is the reason why it was important to examine if MEK inhibition modulated the effects of betulinic acid and UV radiation against some melanoma cells.

Material and methods

Cells

wt p53 human melanomas: C8161 and Mel Juso cells.20

Functional wt p5315 was identified in these cells by its lack of immune precipitation with Pab 240 (SC-99) under nondenaturing immune precipitation and reactivity with monoclonal antibody DO-1 (SC-126) specific for wt p53 under comparable conditions.15 Immune precipitation was aided by addition of Protein A/G Plus agarose (SC-2003). Subsequently, immune precipitated proteins were subjected to SDS-PAGE electrophoresis, bidirectionally blotted onto nitrocellulose membranes15 and p53 identified by the DO-1 (SC126) monoclonal antibody, which recognizes both wt and mut p53 after SDS-PAGE electrophoresis and immune blotting

Mutant p53 WM164 melanoma.22

Expression of p53 mutant protein in these cells was verified by immune precipitation of mutant p53 with monoclonal antibody Pab240 (SC-99), which is mutant p53 specific under nondenaturing conditions. These experiments confirmed selective immune precipitation of mutant p53 with Pab 240. Both Pab 240 and DO-1 antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA).

Morphological and DNA fragmentation assays

This was carried out in 24-well tissue-culture plates seeded in triplicate with 15,000 cells each. One day after cell attachment, betulinic acid, U0126 or vehicle (DMSO) were added, and after 48–60 hr, cells were examined under an inverted microscope followed by addition of 7 M guanidinium hydrochloride and 0.1 M β-mercaptoethanol, to promote cell lysis. Subsequently, viscous samples were electrophoresed in 1.5% agarose gels, for nonradioactive visualization of DNA damage20 with Sybr Gold 1:10,000 (Molecular Probes) in a Dark Reader (Clare Chemical Research, Denver, Colorado).

Immune blotting.

SDS-PAGE and bidirectional immune blots were carried out by lysing the cells as previously described (70 μg protein/lane),15 including prestained MW markers (Life Technologies, Gaithersburg, MD, USA) during SDS-PAGE electrophoresis to asses specific changes in protein expression. After blocking nonspecific sites on nitrocellulose membranes with 5% skimmed milk containing 0.05% Tween 20 in Tris-buffered saline (pH 7.5) for 2 hr at room temperature, membranes were reacted overnight with specific antibodies in the same blocking solution. Antibodies used for specific immune blotting like ERK # 9107, P-ERK # 9106, Akt # 9272 and P-Akt #9271, were from Cell Signalling (Boston, MA). Other antibodies like those against cdk4 (SC-601), cdk2 (SC-163), cdk7 (SC-529), cdc2 (SC-8395), cyclin A (SC-751), and PARP (SC-7150) were from Santa Cruz Biotechnology (Santa Cruz, CA). After extensive washing with Tris-buffered saline containing 0.05% Tween 20, membranes were reacted with anti-mouse IgG-peroxidase for DO-1 monoclonal antibody detection of p53 or with Protein A-peroxidase for detection of rabbit antibodies against all other proteins. Finally, detection was achieved by super signal-mediated chemiluminescence.

Laser Scanning Cytometry.

To quantitate subG1 DNA damage, chromatin condensation and cell cycle profiles, we used laser scanning cytometry (LSC). Cellular DNA in fixed cells23, 24 was stained with propidium iodide and simultaneously analyzed for maximal pixel DNA-associated fluorescence (a sensitive marker of chromatin condensation) and DNA integral fluorescence23 with an LSC-2 cytometer (Compucyte, Cambridge, MA, USA).

Results

U0126 decreases ERK activation and cyclin A expression but does not protect C8161 p53 wt melanoma from UV radiation

Pharmacological inhibition of the MAPK kinases MEK1/2 was recently reported as a useful strategy for killing melanoma cells.5 Since C8161 melanoma were not used in such study, and others reported that the MEK1 inhibitor PD98059 is not cytotoxic but can sensitize C8161 melanoma cells to cisplatin-induced apoptosis,25 we now investigated whether U0126, which is a more updated MEK inhibitor26 is cytotoxic to C8161 melanoma, or whether it protects these cells from UV radiation-mediated toxicity.16, 17

Prior to studying the ability of U0126 to regulate the effect of UV radiation on C8161 melanoma, we investigated whether these agents affected ERK phosphorylation (P-ERK) relative to ERK expression and cyclin A expression, when normalized to actin. This revealed a marked UV-induced increase in P-ERK relative to total ERK, effect that was partly counteracted by concurrent treatment by MEK inhibition with U012626. In contrast, UV did not significantly change cyclin A levels, but U0126 decreased cyclin A expression irrespective of whether cells were UV-irradiated or not (upper Fig. 1). This effect of U0126 of lowering cyclin A in adherent C8161 melanoma was paralleled by change in the G1 population of about 29% in DMSO-treated control cells compared to 57.9% of G1 cells in those treated with 10 μM U0126 (not shown), results compatible with the U0126-mediated decline in cyclin A (upper Fig. 1). Since PARP fragmentation is a hallmark of apoptosis,27 we investigated whether PARP cleavage induced by UV radiation is counteracted by 10 μM U0126. This revealed partial PARP fragmentation 18 hr after 5 J/m² of UV irradiation and extensive PARP cleavage after a similar interval with 10 J/m² of UV in C8161 melanoma. No PARP fragmentation was detected in unirradiated cells even when treated with U0126, and no protection from UV irradiation was evident by a 2-hr pretreatment with 10 μM U0126 (lower Fig. 1).

Figure 1. U0126 decreases UV-induced ERK activation but does not protect C8161 p53 wt melanoma from UV radiation. Upper : ERK activation (P-ERK) and cyclin A are reduced by U0126 in nonirradiated and UV-irradiated cells. Lower: PARP fragmentation is increased in response to UV radiation and these changes are not antagonized by U0126. Cells were examined 18 hr after UV irradiation. Similar results were obtained in 3 independent experiments.

U0126 counteracts PARP fragmentation and ERK activation induced by Betulinic acid in C8161 melanoma

Since exposure of melanoma cells to betulinic acid induces oxidative stress21 linked to P-ERK,16 this was inhibited with U012626 to ask whether it could modulate melanoma responses to betulinic acid. Morphological studies with C8161 melanoma exposed for 4 days to 5 μg/ml betulinic acid showed increased cell detachment, suggestive of cytotoxicity. However, this was counteracted by a 2-hr pretreatment with 10 μM U0126 (upper Fig. 2). We also investigated whether the cytotoxic effect of betulinic acid correlated with the induction of PARP cleavage.26 This showed that betulinic acid induced PARP fragmentation reciprocal with increased cyclin A expression and greater P-ERK. However, P-ERK and cyclin A expression were decreased by U0126, concomitantly with its effect in antagonizing betulinic acid-mediated PARP cleavage (lower Fig. 2). This suggests that MEK inhibitors like U0126, which decrease cyclin A (Fig. 1) required for S + G2 progression,28, 29, 30, 31 may protect C8161 melanoma from betulinic acid by increasing G1 accumulation, as shown by others.28, 29

Figure 2. U0126 counteracts PARP fragmentation and ERK activation induced by Betulinic acid in C8161 melanoma. Upper : Morphological observations indicating betulinic acid cytotoxicity and its antagonism by U0126. Cells were examined 4 days after drug treatment. Lower: PARP fragmentation, P-ERK and cyclin A expression normalized to ERK levels are increased in betulinic acid-treated cells and these changes are antagonized by U0126. Comparable data were seen in 4 separate experiments.

Lower Akt activation induced by Betulinic acid is counteracted by U0126

Some specific extracellular signals associated with survival and apoptosis often result in reciprocal or simultaneous activation of the MEK-ERK and Akt pathways. Enforced activation of Akt or inhibition of the MEK-ERK pathway promoted differentiation and therefore myotube formation.32 However, the cross-talk was only observed in postdifferentiation myotubes and not in myoblasts, where Akt activity or its inhibition had no influence on the MEK-ERK pathway.32 This suggests a regulation of the cross-talk, in this particular case with a cell stage specificity. On the other hand, exposure to UV radiation (60 J/m²) induced Akt phosphorylation in JB6 P⁺ mouse epidermal cell line, Cl 41, but pretreatment with PD98059 did not inhibit UV-induced Akt phosphorylation, though it blocked UV-induced P-ERK. Reciprocally, overexpression of a dominant negative Akt mutant, Akt-T308A/S473A, did not affect UV-induced Erk activation, but attenuated UV-induced increase of Akt phosphorylation.33 The authors concluded that there is no cross talk between the Erk and Akt pathway in “UV responses.”31 Since suppression of ERK activation seemed important in protecting C8161 melanoma from betulinic acid, we tried to learn whether changes in Akt activation were involved in the effects of betulinic acid and U0126. For this analysis, we used the same extracts used in Figure 2. SDS-PAGE and immune blotting with antibodies to phospho Akt and Akt protein showed loss of Akt activation in betulinic acid–treated cells compared to control cells. However, this loss of Akt activation was counteracted by U0126 (Fig. 3a).

Figure 3. (a) U0126 counteracts loss of Akt activation induced by betulinic acid in C8161 melanoma. Betulinic acid-induced loss of Akt activation (P-Akt) normalized to Akt expression is counteracted by U0126. Similar results were obtained in 3 separate instances. (b) Differential Effect of U0126 and Ly294002 in counteracting toxicity induced by Betulinic acid in WM164 melanoma. Morphological observations after 4 days of treatment of WM164 cells with 3.5 μg/ml betulinic acid, 10 μM U0126 or the same concentration of Ly294002 : note greater cell detachment, suggestive of cytotoxicity in betulinic acid-treated cells and in those treated with this drug and Ly294002. Similar results were seen in 2 different experiments.

This implied that C8161 melanoma survival from 5 μg/ml betulinic acid by U0126 inhibition of ERK signaling32, 33 was partly helped by Akt signaling.

To investigate further whether other melanoma cells were comparably protected from betulinic acid-mediated cell death by inhibitors of MEK or Akt signaling, we used mutant p53 WM 164 melanoma.22 These cultures were pretreated with DMSO (control cells) and 10 μM of either U0126 in DMSO, to inhibit MEK signaling,26 or the same concentration of LY 2900402 in DMSO, to inhibit Akt signaling,32 prior to treatment with 3.5 μg/ml betulinic acid. These studies showed that only 10 μM U0126 protected WM164 melanoma from morphological damage associated with cell rounding by 4 days with 3.5 μg/ml betulinic acid. However, no comparable protection was afforded by the same concentration of LY2900402, with no obvious damage in WM164 melanoma treated with either LY 2900402 or U0126 in the absence of betulinic acid (Fig 3b). Moreover, evidence of toxicity and cell rounding was partly accelerated in betulinic acid–treated WM164 cells pretreated with LY2900402 (not shown). This suggests that inhibition of ERK signaling by U0126 is more critical than inhibition of Akt signaling by LY2900402 to protect against betulinic acid.

Betulinic acid induction of DNA damage is counteracted by U0126 in human melanoma cells irrespective of p53 status

Since results in Figure 3 revealed that the lethal effects of betulinic acid on C8161 melanoma were counteracted by 10 μM U0126, we examined whether this effect was evident in mutant p53 WM 164 melanoma22 and wt p53 Mel Juso melanoma.20 In contrast to C8161 melanoma in which exposure for 4 days to 5 μg/ml betulinic acid led to morphological evidence of cytotoxicity (Fig. 2), this became evident only after 70 hr in WM164 melanoma and took about 90 hr in Mel Juso melanoma. To ask whether these effects were associated with genomic damage, both floating and adherent cells were harvested, DNA was solubilized with 7 M guanidinium hydrochloride and 0.01 M β-mercaptoethanol, and genomic fragmentation was assesed after agarose gel electrophoresis (see Methods). This revealed fast-migrating DNA indicative of genomic cleavage only in cells treated with betulinic acid and its prevention by U0126 (Fig. 4). Such results suggest that the attenuating effect of U0126 on betulinic acid-mediated cell death are not restricted to C8161 melanoma since they are seen in 2 other different melanomas, irrespective of their p53 status (Figs. 2 and 4).

Figure 4. Betulinic acid induction of DNA damage is counteracted by U0126 in human melanoma cells irrespective of p53 status. Wt p53 Mel Juso cells and mutant p53 WM164 cells were exposed to betulinic acid, U0126 or both, whenever indicated. When cells showed early morphological signs of response to betulinic acid, DNA was prepared for electrophoretic analysis in 1.5% agarose gels. Note DNA fragmentation irrespective of p53 status in betulinic acid-treated cells and its antagonism by U0126. These experiments gave similar results in 2 different experiments.

Betulinic acid induction of chromatin condensation is preferential in G2 cells and is counteracted by U0126 in adherent WM 164 human melanoma

Chromatin condensation and G2/M arrest are observed at low concentration (5 nM) of paclitaxel in NPC carcinoma, which subsequently convert to a sub-G1 apoptotic form, suggesting that persistent or transient G2/M arrest is a prerequisite step for apoptosis elicited by low doses of paclitaxel.29 Since WM 164 melanoma responded more rapidly to betulinic acid than C8161 or Mel Juso melanomas, we used LSC24 to investigate whether chromatin condensation and G2/M accumulation also occurred after treatment with betulinic acid in these cells. Chromatin condensation reflected by an increase in DNA maximal pixel relative to DNA integral was measured by propidium iodide staining using LSC.23, 24 In Figure 5, chromatin condensation coincided with G2 accumulation (66.8% in quadrant 2) in cells treated with betulinic acid. In contrast, DMSO-treated cells showed only 23.0% compared to 14.8 and 12.0% in the same quadrant 2, in cells exposed to U0126 by itself or in combination with betulinic acid. With reference to quadrant 1, which represents G2/M cells with lower maximal pixel, DMSO-treated cells showed 47.5% in this region vs 10.6% in cells treated with betulinic acid, 15.6% in cells treated only with U0126 and 29.7% in cells jointly treated with both drugs. All this indicates that concomitant chromatin condensation and G2 accumulation are induced by betulinic acid, that such changes are counteracted by U0126, and that in cells jointly treated with both drugs, there is an increase in G2 cells with low DNA condensation (maximal pixel) in contrast to that seen in cells treated only with betulinic acid (Fig. 5).

Figure 5. Betulinic acid induction of DNA condensation is preferential in G2 cells and is counteracted by U0126 in WM 164 human melanoma monolayers. LSC was used to quantitate DNA condensation in chromatin and cell cycle profile in propidium iodide stained fixed cells after 60 hr of treatment.25 Note that DNA condensation in betulinic acid-treated cells is preferential for G2/M cells (quadrant 2). These effects are counteracted by U0126 cells, which show much lower DNA condensation (quadrant 2). These experiments were reproducible in 2 different experiments.

PARP fragmentation is preceded by a specific cdk4 down-regulation

Analysis of WM164 cells at 70 hr after exposure to betulinic acid revealed PARP fragmentation, which was antagonized by U0126. However, within 48 hr of treatment, U0126 decreased cyclin A, suggesting that inhibition of P-ERK by U0126 prevents S+G2 progression (Fig. 6), as shown by others.27, 28, 31 Similar results were observed after a 82-hr treatment with the same agents in Mel Juso cells (not shown). Since G2 accumulation was observed in WM164 cells exposed to betulinic acid (Fig. 5), we investigated whether such changes were associated with down-regulation of specific cell cycle controlling cyclin-dependent kinases like cdc2 (cdk1), cdk2, cdk4, and cdk-activating cdk7. Immune blotting for these proteins showed a selective decline in cdk4 protein within 48 hr with betulinic acid, (Fig. 6) prior to any detectable PARP cleavage, which coincided with a late decrease in cdk2 (Fig. 6). Alterations in cdk4 or cdk2 proteins and PARP cleavage induced by betulinic acid were all antagonized by U0126. However, the loss of cyclin A after exposure to U0126 was also partly counteracted by betulinic acid, implying that this drug acts by trying to overcome the U0126-mediated G1 block (Fig. 6).

Figure 6. Cdk4 down-regulation precedes apoptosis-associated PARP fragmentation in betulinic acid-treated WM164 melanoma monolayers. Immune blotting of bidirectionally transferred blots was used for repeated reaction and erasure with antibodies indicated in each case. Note an early cdk4 loss prior to PARP fragmentation in betulinic acid-treated cells, without a concomitant change in cdk2, cdc2 cdk7 or cyclin A. Note a late decrease in cdk2 together with PARP fragmentation in betulinic acid-treated cells, and reversal of these changes by U0126, which decreases cyclin A, effect partly antagonized by betulinic acid. Similar results were obtained in 3 separate instances.

DNA condensation induced by betulinic acid is counteracted by U0126 in WM164 spheroids

Since multicellular spheroids are more drug resistant than the corresponding adherent tumor monolayers,24, 34 and greater chromatin condensation in G1 may be suggestive of apoptosis or delay to reenter G1,23, 24 we investigated whether betulinic acid increased DNA condensation in WM 164 spheroids and if this effect was counteracted by joint treatment of this drug with U0126. In contrast with the adherent 2D cultures of WM164 melanoma, in which cells treated with betulinic acid showed greater chromatin condensation coinciding with G2 accumulation (66.8% in quadrant 2) (Fig. 5), 3D spheroids of the same tumor cells similarly treated with betulinic acid showed 12% of their population in quadrant 2 compared to 11% of a similar population in DMSO–treated control spheroids (Fig. 7). However, betulinic acid treatment of spheroids doubled the G1/S population with greater chromatin condensation (quadrant 4) compared to that seen in the same quadrant in DMSO-treated control spheroids and in those treated with U0126 even when this was added together with betulinic acid (Fig. 7). In summary, the most obvious effect of betulinic acid against spheroids was to decrease the G1/S population (quadrant 3), and this effect was completely reversed by U0126 (Fig. 7).

Figure 7. Betulinic acid induction of DNA condensation in G1/S in WM 164 melanoma spheroids is counteracted by U0126. LSC was used to quantitate DNA condensation in chromatin and cell cycle profile in propidium iodide stained fixed cells after 60 hr of treatment.25 Note that DNA condensation in betulinic acid-treated cells is preferential for G1/S cells (quadrant 4). These effects are counteracted by U0126 cells, which show much lower DNA condensation (quadrant 4). These results were reproducible in 2 different experiments.

Discussion

This paper has attempted to clarify whether the conflicting effects of pharmacological inhibitors of MEK activation are cell specific or stimulus dependent. Most studies to date have used either PD98059 or U0126, but we now used the latter because of its higher affinity to all MEK forms.26 The prior conflicting results with MEK inhibitors included enhancement of vinblastine-mediated apoptosis by ERK inhibition with 50 μM PD98059 in myeloid leukemia ML-1 cells, without a comparable effect on U937 and Jurkat lymphoid cells.11 The same report showed that apoptosis induced by paclitaxel was not potentiated by the MEK inhibitor PD98059 in ML-1 cells.11 In contrast, a similar MEK inhibition enhanced paclitaxel-induced tumor apoptosis in carcinoma cells.12 Another heterogeneous response to U0126 was evident in a recent report using 8 melanoma lines, of which 3 (MALME 3M, M14-MEL and SK-MEL2) were highly susceptible to U0126. In contrast, 5 (LOX-IMVI, SK-MEL28, SK-MEL 5, UACC-257 and UACC-62) were far less susceptible to this agent.5 However, the authors concluded that the inhibition of the MAPK pathway is a useful strategy for killing melanoma, without explaining the reason for the nonuniform melanoma apoptotic response to U0126.5 Regulation of cisplatin response is another example of the contradictory effect of U0126, since ERK activation is induced by cisplatin, and its cytotoxicity is potentiated by pharmacological MEK inhibition in ovarian carcinoma.13 However, in cervical carcinoma Hela cells, cisplatin also activated the ERK pathway but MEK1/2 Inhibitors blocked cisplatin-induced apoptosis.14 This implies that although cisplatin activates the ERK pathway in different cells, MEK inhibition with U0126 gives a cell-specific response to this drug. Others reported that the MEK1 inhibitor PD98059 is not cytotoxic but can sensitize C8161 melanoma cells to cisplatin-induced apoptosis.25 To define whether U0126 exerts different effect in different cells or whether it exerts unequal responses depending on the stimuli, we now used the same 2 hr preincubation and dose of U0126 with human C8161 melanoma, and subsequently exposed these cells to either 5 or 10 J/m² of UV radiation or 5 μg/ml of betulinic acid. Our data suggest that the contradictory results with MEK inhibitors are partly due to the fact that it gives a stimulus-dependent response in the same cells, and it may give cell-specific responses because of unequal extent of differentiation, culture conditions and genetic background of target cells exposed to these agents. In contrast to our studies showing lack of attenuation in UV-induced PARP fragmentation by U0126 in human C8161 melanoma, pharmacological inhibition of ERK activation by 50 μM PD98059 was reported to attenuate apoptosis induced by UV radiation (100 mJ/cm²) in mouse NIH3T3 fibroblasts.9 Although these authors showed that etoposide induced apoptosis determined by TUNEL and sub G1 DNA was counteracted by PD98059, their conclusions indicating PD 98059-associated attenuation of apoptosis by UV9 was based on %viability, measured with WST-1, 2-(4-Iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, a variant of the MTT assays. Others have reported that decreased MTT activity might be a reflection of lower mitochondrial activity, leading to either growth arrest or cell death,35 the reason why this method is inconclusive35 to claim a U0126 protection from UV-mediated “apoptosis.”9 In our experiments, we measured apoptosis-associated PARP fragmentation27 in response to 5 or 10 J/m² of UV, revealing a dose-dependent PARP cleavage irrespective of pretreatment with U0126 (Fig. 1). In contrast, in the same C8161 cells, U0126 prevented PARP fragmentation induced by betulinic acid (Fig. 2) and DNA damage induced by betulinic acid in wt p53 Mel Juso20 and mut p53 WM 164 cells.22 A difference in the inability of U0126 to protect from acute apoptotic stimuli like UV radiation, which causes rapid cell death within 18 hr of treatment, may be the superinduction of P-ERK without increasing cyclin A in a comparable manner in C8161 melanoma (Fig. 1). In contrast, U0126 protects the same cells from betulinic acid-induced slower apoptosis that takes 4–6 days at 5 μg/ml in these cells, leading to comparable P-ERK and cyclin A increases (Fig. 2). Perhaps a more plausible reason for the different ability of U0126 to protect C8161 melanoma from UV radiation or betulinic acid may be the reported inability of UV radiation to promote a cross-talk between the ERK and Akt pathways.32, 33 In contrast, we now show that inhibiting ERK activation with U0126 counteracted betulinic acid cytotoxicity, partly by restoring Akt phosphorylation in C8161 melanoma (Fig. 3a). This not only suggests that allowing unrestricted ERK signaling favors betulinic acid-induced apoptosis, but also implies that U0126 partly protects from this stress by restoring Akt activation.32, 33 Additional observations pretreating WM164 melanoma either with LY 2900402, an Akt signaling inhibitor,32 or with U0126, an ERK signaling inhibitor,26 prior to betulinic acid treatment revealed better protection with U0126 (Fig. 3b). This suggests that inhibition of ERK signaling by U0126 is more critical to protect against betulinic acid-mediated cell death than inhibition of Akt signaling by LY2900402. Besides increasing Akt activation (Fig. 3a), the protective effect of U0126 in the presence of betulinic acid includes its ability to prevent cell cycle progression,29 since we now demonstrate that it causes an accumulation of G1 cells, in agreement with other studies with this drug.30, 31

In contrast, betulinic acid does not prevent progression into G2 or cyclin A expression, but rather increases the G2 population with increased chromatin condensation in adherent WM164 melanoma (Fig. 5). Moreover, betulinic acid may prevent reentry into G1, since it specifically down-regulates cdk4 in adherent WM164 melanoma (Fig. 6).35, 36, 37 A possible basis for the preferential effect of this drug against melanoma compared to melanocytes19 is that normal cells in G(0) or G(1) or G1-arrested tumor cells exposed to U012627, 31 may avoid the toxicity induced by betulinic acid19 exerted vs cycling tumor cells attempting to reenter G1 but lacking cdk4. Others have shown that hyposmosis dramatically decreased cdc2, cdk2 and cdk4.37 However, in contrast to hypoosmosis, betulinic acid did not decrease cdc2, cdk7 or cdk2 in a comparable manner, indicating that it acts in a more targeted manner against cdk4. This early betulinic acid effect on cdk4 may produce a tumor inability to respond to mitogenic survival signaling, like that seen in quiescent CDK4(−/−) fibroblasts, which show a substantial delay in S-phase entry after serum stimulation.37, 38 Since multicellular spheroids are more drug resistant than the corresponding adherent tumor monolayers and show little response to conventional chemotherapy,24, 34 we also investigated whether WM 164 melanoma spheroids responded to betulinic acid. A comparison of results in Figures 5 and 7 shows that WM 164 melanoma exhibits a different response in 2D adherent monolayers compared to 3D spheroids, which are a more representative model of in vivo tumors.34 In both 2D and 3D tumor cell cultures, betulinic acid increased chromatin condensation although this occurred mostly in G2 for 2D cultures and preferentially in G1 for 3D spheroids. In spite of these unequal responses, inhibition of MEK signaling with U0126 counteracted the effects of betulinic acid, irrespective of tumor cell anchorage (Figs. 5 and 7).

In summary (Table I), our results indicate that U0126 counteracts betulinic acid mediated tumor cell death in 3 different adherent human melanomas, irrespective of p53 status, indicating that pharmacological inhibition of MEK activation may not always be helpful against melanomas.5 The complexity of targeting constitutive ERK activation as a general melanoma therapy5 is also exemplified by a recent study of activated ERK1/2 in a panel of superficial spreading and nodular primary and metastatic melanomas, which found that 54% of primary and 33% of metastatic melanomas expressed variable levels of activated ERK1/2, concluding that activation of ERK1/2 did not have any impact on relapse-free or overall survival in melanoma patients.39 These studies and those in prostate carcinoma,40 in which advanced disease is accompanied by Akt activation and Erk inactivation, argue that pharmacological inhibition of the MAPK kinases MEK1/2 may not be an unqualified anticancer strategy.4 The efficacy of inhibiting ERK signaling as part of anticancer therapy may depend on cellular background, disease stage and on whether there is cross talk between contradictory PKB/Akt and MEK-ERK signaling32 in response to other agonist signals.

Table I. Effects of Betulinic Acid and U0126 On Adherent WM 164 Melanoma

	G1/G2 accumulation	cdk4	Chromatin condensation	Cyclin A	Akt
Control cells		Intermediate	Low	High	Phosphorylated
Cells treated with U0126	G1 accumulation	High	Low	Low	Phosphorylated
Cells treated with betulinic acid	G2 accumulation	Low	High	High	Dephosphorylated
Cells treated with U0126 and betulinic acid	No G2 accumulation	Intermediate	Lower	Intermediate	Phosphorylated

Acknowledgements

We thank Dr. M. Herlyn and the Wistar Institute (Philadelphia, PA) for providing WM164 human melanoma cells. Partial support from the (CIC-CCA Caracas)-Terry Fox Run for Cancer Research is also gratefully acknowledged.