Peroxisome Proliferator–Activated Receptor α Activation Decreases Metastatic Potential of Melanoma Cells In vitro via Down-Regulation of Akt

The significant need to develop new therapeutic and preventive approaches for melanoma treatment focuses attention of health service and basic science all over the world. Malignant melanoma incidence has increased dramatically in recent years (1–3) whereas the mortality rate in melanoma patients with the vertical tumor growth phase is very high due to tumor cell penetration in the surrounding tissues, intravasation into blood or lymphatic vessels, and rapid formation of metastases. Only very thin melanoma lesions (<1 mm in thickness) in radial growth phase when subjected to surgical removal are curable in almost all cases (4). The therapy of the advanced stages is extremely difficult and rarely successful (5, 6) due to the prominent metastatic ability of melanoma. Despite trials with a variety of chemotherapeutic regimens, the prognosis for metastatic melanoma patients remains very poor (5, 7).

Therefore, strong emphasis has recently been put on chemoprevention, which involves the use of synthetic or natural substances to reverse the transition of premalignant lesions into invasive cancer. Recent reports indicate the potential role of lipid-lowering drugs, fibrates and statins, in melanoma chemoprevention (8–10). One candidate for melanoma chemoprevention is fenofibrate (22). It has been widely used to lower the plasma levels of triglycerides and cholesterol, improve low-density lipoprotein/high-density lipoprotein ratio, and prevent development of arteriosclerosis mainly through regulation of apolipoprotein genes expression (11, 12). Fenofibrate is also a potent ligand for peroxisome proliferator–activated receptor α (PPARα), a nuclear receptor, which, together with two other PPAR isoforms, β and γ, belongs to the steroid hormone receptor superfamily (13). These ligand-modulated transcription factors were first discovered to regulate glucose, lipid, and amino acid metabolism (14–16). Recent research, however, revealed their broader function as differentiation inducers, inflammatory response modulators, and potential anticancer agents (17–19).

Cancer progression is often associated with prolonged activation of signal transduction pathways, such as phosphatidylinositol 3-kinase/Akt pathway. This constitutive activation contributes to cellular transformation, mainly through promotion of cell survival and proliferation. Akt phosphorylates and inactivates numerous proapoptotic proteins, such as BAD and procaspase-9, and simultaneously facilitates nuclear factor-κB translocation to the nucleus, where it activates antiapoptotic genes (20). The progression from a benign in situ tumor to an invasive disseminating form requires profound changes in the tumor cell phenotype (e.g., gaining the ability to migrate, to intravasate and survive in the anchorage-independent conditions in blood vessels, and to penetrate distant organs). Indeed, active Akt predisposes fibrosarcoma cells to an aggressive phenotype by increasing their migration, invasion, and metalloproteinase production (21). Therefore, the inhibition of phosphatidylinositol 3-kinase/Akt pathway could be involved in antimetastatic activity of fenofibrate.

We have recently reported that fenofibrate inhibits metastatic spread from s.c. melanoma tumors in vivo (22). The experiments reported here were designed to elucidate molecular mechanisms underlying the antimetastatic action of fenofibrate.

Here we attempt to evaluate whether cell migration and anchorage-independent growth, the hallmarks of metastatic behavior, are affected by fenofibrate. We are also asking whether fenofibrate acts in a PPARα-dependent or PPARα-independent manner and which signaling molecules are involved in its antimetastatic action. We have selected B16F10 invasive mouse melanoma because this cell line is recognized as a standard model for assessing metastatic ability of melanoma (23, 24). Additionally, some experiments were done in the SkMel 188 human melanoma cell line to confirm that the observed fenofibrate inhibitory effects are not restricted to the mouse model. Our results document that fenofibrate inhibits melanoma cell migration and anchorage-independent growth and that this inhibitory action of fenofibrate is associated with the attenuation of Akt constitutive phosphorylation. These findings may suggest a novel therapeutic approach in which nontoxic PPARα activators by itself or in combination with proapoptotic agents could be more effective in eliminating malignant melanoma cells in vivo.

General reagents. Cell culture reagents used were DMEM, RPMI 1640; PBS; fetal bovine serum (FBS); sodium pyruvate; antibiotics penicillin and streptomycin, G418; Lipofectamine 2000, ultrapure agarose (Invitrogen, Grand Island, NY); bovine serum albumin; Tris; DMSO; trypan blue; HEPES; EGTA; sodium chloride; phenylmethylsulfonyl fluoride; sodium ortovanadate; Triton X-100; Igepal NP40; fenofibrate (Sigma, St. Louis, MO); staurosporine (Upstate, Lake Placid, NY); GW9662, U0126 (Biomol, Plymouth Meeting, PA); magnesium chloride; glycerol; and methanol (Fischer Scientific, Pittsburgh, PA); protease inhibitor cocktail (Roche, Indianapolis, IN).

Cell culture. B16F10 mouse melanoma cells (kindly provided by Dr. Martyna Elas, Jagiellonian University, Krakow, Poland) and SkMel 188 cells (a gift from Dr. Andrzej Slominski, University of Tennessee, Memphis, TN) were maintained in RPMI 1640 supplemented with 10% FBS, 50 units/mL penicillin, and 50 ng/mL streptomycin. For migration assay, serum-free medium (SFM) was used: RPMI 1640 with penicillin and streptomycin supplemented with 0.1% bovine serum albumin. Cell growth and survival in various treatment conditions, described in the figure captions, were evaluated by trypan blue exclusion test. Cells were treated with fenofibrate, GW9662, staurosporine, and U0126, as indicated in the figures. Cells in control groups were treated with DMSO, which was used as a solvent for fenofibrate, and with GW9662 and U0126.

B16F10/myrAkt cell line was obtained by stable transfection (using Cell Line Nucleofector kit R, Amaxa, Gaithersburg, MD) of parental B16F10 cells with pUSEamp vector encoding activated (myristoylated) Akt1 (Upstate; a kind gift from Dr. Ashwani Malhotra, Department of Medicine, New Jersey School of Medicine and Dentistry, Newark, NJ). Stable cell clones were selected with G418.

Small interfering RNA (siRNA)–mediated Akt1 gene silencing was done by transfection of B16F10 cells with siGenome Smart pool mouse Akt1 siRNA (Dharmacon, La fayette, CO) using Lipofectamine 2000. Anti–lamin A siRNA was used in control experiments.

Luciferase assay. B16F10 cells were transfected with PPAR responsive element reporter plasmid, J3TKpGl3 (a generous gift from Dr. Alicja Jozkiewicz, Jagiellonian University), and Renilla luciferase expression vector (Promega, Madison, WI) as a control of transfection efficiency. J3TKpGl3 vector contains three copies of J site from apo-AII gene promoter (25, 26) cloned into pGl3 reporter plasmid (Promega). Four hours after transfection, cells were treated for 40 hours with fenofibrate (25 μmol/L) or DMSO as a control. Luciferase activity measurements were done using Dual-Reporter Luciferase assay system (Promega).

Western blot. Western blot assays were done as described elsewhere (27). Briefly, cells were lysed in the lysis buffer [50 mmol/L HEPES (pH 7.5), 150 mmol/L NaCl, 1.5 mmol/L MgCl₂, 1 mmol/L EGTA, 10% glycerol, 1% Triton X-100, 1 mmol/L phenylmethylsulfonyl fluoride, 0.2 mmol/L sodium ortovanadate, supplemented with protease inhibitors]. Liver and intrascapular brown adipose tissue samples were isolated from C57BL/6 mice (Temple University Animal Care and Use Procedure approval no. 1237), chopped, washed thrice in sterile PBS, and homogenized in the glass-teflon homogenizer with addition of TNN buffer [50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 0.5% NP40, protease inhibitors].

Aliquots of 50 μg of protein extracts were separated in a 4% to 15% gradient SDS-PAGE (Bio-Rad, Hercules, CA) and transferred onto nitrocellulose membranes. The resulted blots were blocked in 5% nonfat milk and probed with the following antibodies: anti-PPARα (Chemicon, Temecula, CA), anti–phospho-Akt(Ser⁴⁷³), anti-Akt, anti–phospho-extracellular signal–regulated kinase (Erk) 1/2 (p42/44) (all from Cell Signaling Technologies, Danvers, MA), anti-Erk1 (Santa Cruz Biotechnology, Santa Cruz, CA), anti-GRB2 (Transduction Laboratories, Lexington, KY), and anti–lamin A (Santa Cruz Biotechnology). To quantitate Western blot results, densitometric analysis was done for phospho-Akt, phospho-Erk1/2, and phospho-glycogen synthase kinase 3β (GSK3β) blots using Scion Image freeware software.

Migration assay. Cell migration was assessed in Transwell chambers (Corning Corporation, Corning, NY) with polycarbonate filters (6.5-mm diameter, 8.0-μm pore size). Cells were suspended in 200 μL of SFM supplemented with the treatment agents, as indicated in the figures, and plated in the upper chambers. The bottom wells were filled with SFM. After 48 hours, the inserts were washed with PBS, the nonmigratory cells were wiped out with cotton swabs, and the filters were stained and fixed with crystal violet/carbol/25% methanol (1:1:2) mixture for 20 minutes. The blue-stained cells were counted under a light microscope.

Soft-agar colony-forming assay. Cells were suspended in 0.4% agarose/DMEM with 10% FBS, penicillin, streptomycin, 1 mmol/L sodium pyruvate, and the treatment agents, as indicated in the figures, and plated onto 0.8% agarose/DMEM–covered 35-mm dishes. The dishes were kept in the cell culture incubator for 10 days, whereupon the colonies >50 μm were counted under a light microscope.

Clonogenic assay. Cells were plated in six-well plates (1,000 per well) in regular culture medium with fenofibrate (25 μmol/L), staurosporine (10 nmol/L), or both. After 1 week, cells were washed with PBS, fixed, and stained in crystal violet/carbol/25% methanol (1:1:2) mixture for 20 minutes. Colonies with diameter >0.5 mm were counted.

Statistical analysis. Statistical significance of the differences between groups was tested using the Student's t test for homogeneous or heterogeneous variances, as appropriate, or one-way ANOVA (significance level, 5%).

Active PPARα is expressed in B16F10 melanoma cells. We previously reported the detection of PPARα in hamster melanoma cell lines and the attenuation of metastatic spread of these cells by a synthetic PPARα agonist, fenofibrate (22). To investigate the cellular and molecular mechanisms underlying the action of fenofibrate in melanoma, we used a highly metastatic murine melanoma cell line, B16F10 (23, 24). The results in Fig. 1A show that B16F10 cells express PPARα protein at a level comparable to the control brown adipose tissue and much lower than in liver. The evaluation of PPARα transcriptional activity in B16F10 cells showed >2-fold stimulation of PPAR responsive elements by fenofibrate (Fig. 1B), further confirming the presence of an active PPARα signaling system in these malignant cells.

Fenofibrate decreases migration and anchorage-independent growth of mouse and human melanoma cell lines. The ability to migrate, cell survival, and anchorage-independent proliferation are the hallmarks of malignant transformation. To determine whether PPARα activation affects those properties of B16F10 cells, we evaluated cell migration in Transwell chambers and colony formation in soft agar after the treatment with fenofibrate. As shown in Fig. 2A and C, fenofibrate remarkably decreased the migration of B16F10 and SkMel cells. Reductions of >4-fold and >2-fold in the number of migratory B16F10 and SkMel cells, respectively, were observed in the presence of 25 μmol/L fenofibrate. In B16F10 cells, migration in the presence of fenofibrate and GW9662, a synthetic PPARγ inhibitor, was tested. In higher concentration (10 μmol/L), GW9662 also blocks PPARα (28) and efficiently restores cell migration, indicating that fenofibrate action on cell migration is PPARα dependent. Because cell proliferation could affect cell counts in the migration assay, we evaluated the contribution of this variable in our experimental setting. As shown in Fig. 2A (right), the increase in cell number was minimal in the condition in which cell migration was assessed (SFM and SFM + F), providing further confirmation that fenofibrate inhibited the ability of B16 cells to cross through the pores of the polycarbonate membrane. Importantly, fenofibrate partially inhibited also colony formation in soft agar. Results in Fig. 2B and D show 50% and 75% inhibition of colony formation in the presence of 25 μmol/L fenofibrate for B16F10 and SkMel cells, respectively. This statistically significant inhibition in B16F10 was restored when the fenofibrate treatment was accompanied by the PPAR inhibitor GW9662.

In conclusion, fenofibrate treatment leads to impaired cell migration and the attenuation of anchorage-independent growth. The next experiments were designed to test whether alterations in specific cell signaling pathways could account for the changes in cellular behavior after fenofibrate treatment.

Fenofibrate-activated PPARα interferes with Akt and Erk1/2, but not GSK3β, signaling cascades. In the next step, we evaluated the phosphorylation status of Akt, Erk1/2, and GSK3β, three signaling molecules known to play a role in malignant transformation. We observed a time-dependent down-regulation of Ser⁴⁷³ phoshorylation of Akt after incubation with fenofibrate (25 μmol/L) without a change in total Akt protein level (Fig. 3A). B16F10 cells possess a high basal level of Akt phosphorylation (control), which was not affected by DMSO used as a vehicle. A similar down-regulation in phosphorylation after fenofibrate treatment was noted for Erk1/2, but not for GSK3β. Quantitatively, fenofibrate induced >3-fold decrease in Akt phosphorylation and 24-fold decrease in Erk1/2 phosphorylation from the basal level over the 48-hour incubation time.

To prove that these effects were PPARα dependent, we treated the cells with fenofibrate (25 μmol/L) together with the PPAR inhibitor GW9662 (10 μmol/L) and compared the Akt and Erk1/2 phosphorylation status at corresponding time points. As shown in Fig. 3B, PPARα inhibition totally abolished the action of fenofibrate. These results show that activated PPARα can interfere with both Akt and Erk1/2 signaling cascades. The next set of experiments was designed to determine whether impairment of signal transduction via Akt or Erk1/2 is responsible for fenofibrate-induced inhibition of melanoma cell migration and growth.

The decrease of B16F10 migration and anchorage-independent growth by fenofibrate is Akt dependent but not Erk1/2 dependent. The results depicted in Fig. 4A show that Akt is continuously phoshorylated when B16F10 cells are incubated in SFM. To further assess the involvement of Akt in PPAR-mediated inhibition of B16F10 migration and anchorage-independent growth, we developed a stable B16F10 cell line expressing constitutively activated (myristoylated) Akt (B16/myrAkt). In such cells, the level Akt is several times higher than in parental cells and its phosphorylation is strongly promoted by constitutive membrane association of the myristoylated protein (29). The migration assay of B16/myrAkt cells showed that fenofibrate was no longer able to inhibit either migration or colony formation in soft agar (Fig. 4B and C, respectively) when the constitutively active Akt was expressed. Importantly, the inhibitory effect of fenofibrate on cell migration was efficiently imitated by silencing the Akt expression with specific siRNA (Fig. 4D). Irrelevant siRNA against lamin A and mock-transfected cells served as controls. These results confirm that active Akt is important for B16F10 migration, which is in agreement with other reports on melanoma invasive behavior (30). This also supports that the observed action of fenofibrate on B16F10 migration is, at least in part, Akt dependent.

In contrast to Akt phosphorylation, serum starvation of B16F10 cells resulted in significant down-regulation of Erk1/2 phosphorylation (Fig. 5A). Because the migration assays were done in serum-free conditions, the involvement of Erk1/2 activity in this process was probably less pronounced. To evaluate such a possibility, we used the synthetic inhibitor of mitogen-activated protein kinase/Erk kinase, U0126, which blocks Erk1/2 phosphorylation (31). Indeed, in B16F10 cells, U0126, even in a concentration as low as 0.5 μmol/L, abolished Erk1/2 phosphorylation (Fig. 5B). U0126 at 0.5 μmol/L was therefore used in the migration and soft-agar colony formation assays of B16F10 and B16/myrAkt cells (Fig. 5C and D, respectively). In contrast to Akt, Erk1/2 inhibition did not decrease migration and colony formation of B16F10 cells. Therefore, it is reasonable to assume that although Erk1/2 phosphorylation is inhibited by fenofibrate, it does not contribute to the fenofibrate effects on migration and colony formation.

Fenofibrate reveals synergy with staurosporine against B16F10 melanoma growth. The crucial role of Akt phosphorylation and its down-regulation by fenofibrate led us to speculate that fenofibrate could be additive to the antimelanoma activity of selected cytotoxic agents such as staurosporine. Shutdown of the phosphatidylinositol 3-kinase/Akt pathway, which usually confers a powerful prosurvival signaling cascade, could potently increase the effect of apoptosis-inducing chemotherapeutics, such as staurosporine. Staurosporine and its derivatives, which are protein kinase C inhibitors, have already been used to treat advanced metastatic melanoma in phase I and II of clinical trials (32, 33).

Cell survival in monolayer cultures in Fig. 6A shows dose-dependent decrease in cell proliferation for fenofibrate (left) and decrease of both cell proliferation and cell survival for staurosporine (right), evaluated 48 hours after the treatment. Fenofibrate, even in high concentrations (50 and 100 μmol/L), did not cause cell loss, but rather blocked proliferation. Staurosporine, however, was cytotoxic in concentrations exceeding 100 nmol/L. We chose fenofibrate concentration of 25 μmol/L (used in the experiments described above) and a mildly effective staurosporine concentration of 10 nmol/L to check their potentially synergistic effect on B16F10 cell proliferation and survival. Untreated B16F10 grew exponentially in the presence of 10% FBS whereas both staurosporine and fenofibrate inhibited cell proliferation in these culture conditions (Fig. 6B). The most remarkable growth inhibition was noted for combined fenofibrate and staurosporine 96 hours after the treatment. In comparison with fenofibrate alone, the inhibition was 4-fold, and in comparison with staurosporine, the inhibition was >8-fold greater. Synergistic effects of low concentration of fenofibrate and staurosporine on B16F10 growth were even more spectacular when tested in clonogenic assay. Figure 6C shows complete inhibition of colony formation following single-dose treatment of B16F10 cells with 10 nmol/L staurosporine and 25 μmol/L fenofibrate.

In this article, we show that mouse (B16F10) and human (SkMel 188) melanoma cell lines express PPARα, and its activation by fenofibrate causes inhibition of migration and anchorage-independent growth. It is consistent with our previous report that fenofibrate administered orally to melanoma-bearing hamsters significantly decreases the metastatic spread from the primary site (22). We found that fenofibrate inhibits both Erk1/2 and Akt phosphorylation in a PPARα-dependent fashion, although only the latter event is functionally linked with fenofibrate-mediated growth repression and inhibition of migration.

Both phosphatidylinositol 3-kinase/Akt and mitogen-activated protein kinase/Erk1/2 pathways play important roles in normal and malignant cell migration (34–37). PPARs have been reported to interfere with both these pathways. PPARγ and PPARα ligands, including fenofibrate, inhibit vascular endothelial growth factor– and basic fibroblast growth factor–induced endothelial cell migration, which is accompanied by a decrease in Akt phosphorylation (38, 39). In contrast, some other authors report a rapid but transient increase of Erk1/2 and Akt phosphorylation caused by PPARα and PPARγ ligands in cell culture conditions (40–42). However, in these studies, changes in the phosphorylation status were detected shortly (10-30 minutes) after the treatment and therefore cannot be ascribed to canonical activity of PPARs as transcription factors. Such rapid, “nongenomic” effects are most likely PPAR independent (42, 43). In our case, fenofibrate induced PPARα transcriptional activity and the biological consequences are long-term and stable for several days after the initial administration. This time scale draws the attention to potential therapeutic applications and relevance to cancer treatment.

Akt, a cellular homologue of a retroviral oncoprotein v-Akt, is a kinase recruited during signal transduction from growth factor receptors and intracellular pathways. Its activity is regulated by association with the plasma membrane, where it interacts with phosphatidyl inositide phosphate residues, and by subsequent phosphorylation on Thr³⁰⁸ and Ser⁴⁷³ by phosphoinositide-dependent kinases 1 and 2, recently identified as DNA-dependent protein kinase (44, 45). Akt has maximal kinase activity only when phosphorylated at both sites (20, 45).

We have shown that overexpression of a constitutively active Akt mutant rescues cells from fenofibrate-induced inhibition of migration and anchorage-independent growth. This mutant contains the myristoylation signal from src protein, which is responsible for the membrane localization, and possesses several-fold higher activity than wild-type Akt due to hyperphosphorylation (29). Apparently, this hyperphosphorylation makes Akt resistant to fenofibrate. Most likely, the inhibitory action of PPARα on Akt does not involve a direct interaction between these two proteins but rather requires at least one of the proteins regulated by PPARα at the transcriptional level. Phosphatase and tensin homologue is known to inhibit Akt and block its recruitment to the membrane (46) but there are no reports of this protein being regulated by PPARα; indeed, we did not observe any change in phosphatase and tensin homologue level after fenofibrate treatment (not shown).

A good candidate to play a role in PPARα-mediated Akt inhibition is TRB3, a mammalian homologue of Drosophila tribbles. TRB3 blocks insulin signaling pathway and its main physiologic function is to promote glucose output from liver and switch to fatty acid oxidation under fasting conditions (47). It does this task in tight cooperation with PPARα and PPARγ coactivator-1 (48).

TRB3 inhibits Akt phosphorylation on both Ser⁴⁷³ and Thr³⁰⁸ sites by direct binding to Akt, and therefore prevents its activation (47). TRB3 promoter contains numerous functional PPAR responsive element consensus sites and activation of PPARα induces transcription of TRB3 and hepatic glucose production in vivo as a result of Akt inhibition by TRB3 (48). It remains to be verified if PPARα-induced Akt inhibition in B16F10 cells is TRB3 mediated. Interestingly, interactions between Akt and PPARα in regulation of glucose metabolism are mutual because in mice with cardiac overexpression of myristoylated Akt, PPARα and PPARγ coactivator-1 transcription is severely down-regulated (49). A similar situation might be in B16F10/myrAkt transfectants and it could explain why this cell line was resistant to fenofibrate.

Akt is a key member of a signaling cascade coordinating prosurvival and antiapoptotic processes, as well as cell cycle progression (20, 50). Direct involvement of Akt in tumor cell invasion has been reported (21). Constitutive activation of Akt is observed in various tumors and is correlated with their progression and level of malignancy (50–52). In melanoma cells, a high basal level of Akt phosphorylation is frequently observed and may be essential for invasion and metastatic spread (30, 53). This is the case in B16F10 cells, which were selected as a highly invasive variant of the B16 murine melanoma (23, 24). Decrease of high basal level of Akt phosphorylation by fenofibrate in cancer cells can have a therapeutic application in both early and advanced stages of malignant melanoma. The great advantage of fenofibrate is its low toxicity and the fact that it is routinely used to treat hyperlipidemia and hypercholesterolemia. Besides its beneficial effects in reducing the risk of cardiovascular disease and atherosclerosis progression, fibrate drugs have been reported to lower the incidence of various types of cancer, including melanoma (8–10). These observations have led the authors to hypothesize that lipid-lowering drugs could decrease the risk of melanoma development and implied the potential chemopreventive role for these compounds (8, 9). In this article, we asked whether fenofibrate could enhance the effect of chemotherapeutic drugs through inhibition of Akt. We chose staurosporine, a protein kinase C inhibitor with a strong proapoptotic activity, which is currently tested in clinical trials of patients with various tumors including metastatic melanoma (32, 33). Indeed, we observed that fenofibrate and staurosporine are much stronger inhibitors of the proliferation and clonogenic growth of B16F10 melanoma cells when administered together rather than separately. This synergistic effect can be attributed to the fact that staurosporine inhibits phosphoinositide-dependent kinase 1 and, consequently, Thr³⁰⁸ but not Ser⁴⁷³ phosphorylation of Akt (54). Fenofibrate, in parallel, down-regulates Ser⁴⁷³ phosphorylation (Fig. 3A) and therefore enhances the effect of staurosporine. This synergy in antitumor action between fenofibrate and staurosporine is a novel observation and could be applied to develop new therapeutic regimens for patients with metastatic disease.

The fenofibrate concentration used in our in vitro experiments is comparable to plasma concentrations of fenofibric acid, an active metabolite of fenofibrate, detected in patients during standard hyperlipidemia treatment: 300 mg regular or 250 mg slow-release capsules daily. In such patients, plateau-phase plasma concentration of fenofibric acid is ∼10 to 12 μg/mL (28-33 μmol/L; ref. 55). Therefore, as a systemic chemotherapy regimen, the combination of fenofibrate and staurosporine also holds a promise for tolerability and low toxicity compared with current regimens, of which many are highly toxic to the patient.