The Antiproliferative Response of Indole-3-Carbinol in Human Melanoma Cells Is Triggered by an Interaction with NEDD4-1 and Disruption of Wild-Type PTEN Degradation Free

Human melanomas, which arise from melanocytes of neuro-ectodermal origin and can contain mutations in one or more tumor suppressors and/or oncogenes, are the most aggressive form of malignant skin cancers and can be remarkably resistant to conventional cancer therapies (1, 2). Natural dietary phytochemicals represent a promising but largely untapped source of chemotherapeutic agents that has the potential to control the formation, growth, and metastasis of human cancers with minimal side effects, especially during prolonged treatments (3–6). Cellular, physiologic, and epidemiologic studies have implicated indole-3-carbinol (I3C), an indolecarbinol compound derived from hydrolysis of glucobrassicin produced in cruciferous vegetables of the Brassica genus, including cabbage, broccoli, and Brussels sprouts, and its natural diindole condensation product 3,3′-diindolylmethane (DIM) as promising anticancer phytochemicals with negligible toxicity (5–10). Depending on the human cancer cell type, I3C triggered cell-cycle arrest, apoptosis, and disruption of cell migration, and modulated hormone receptor signaling is mediated by the selective regulation of transcriptional, metabolic, and cell signaling cascades (10–18; reviewed in refs. 6–9). Many of these antiproliferative responses are selectively controlled by I3C-activated pathways that are unaffected by DIM, which suggests the presence of I3C target proteins that activate specific signaling pathways in different types of cancer cells.

We recently established that I3C and its more potent and stable derivative 1-benzyl-I3C act as direct noncompetitive inhibitors of the proteolytic activity of neutrophil elastase, the first such identified target protein for I3C (19–21). The I3C inhibition of elastase enzymatic activity directly prevents cleavage of the CD40 member of the tumor necrosis factor receptor gene family, which causes CD40 signaling to switch from activating cell survival pathways to triggering antiproliferative cascades in human breast cancer cells (20, 21). Relatively little is known about the responsiveness of human skin cancers to I3C beyond the observations that I3C increases the sensitivity to UV-induced apoptosis and enhances cytotoxic responses in human melanoma (22, 23) and squamous cell carcinomas (24), respectively. Conceivably, the sensitivity of human melanoma cells to I3C may be to due to the expression and function of melanoma-expressed indolecarbinol target proteins that can activate antiproliferative signaling cascades and/or disrupt cell survival pathways.

A variety of genetic alterations in the tumor-suppressor PTEN (phosphatase and tensin homolog detected on chromosome 10) have been detected in human primary and metastatic melanomas (25, 26). However, many melanomas express very low to nearly undetectable levels of the wild-type PTEN due to the loss of heterozygosity, promoter methylation, and/or alterations of protein stability (27–30). PTEN dephosphorylates phosphatidylinositol 3,4,5-triphosphate (PIP3) and phosphatidylinositol 3,4-bisphosphate (PIP2) at the cell membrane (31). PIP3 generates membrane-docking sites for both phosphotidylinositol-dependent kinase 1 (PDK1) and for the serine/threonine protein kinase AKT-1 through their pleckstrin homology domains, where PDK1 phosphorylates and activates AKT-1 (31–33). Therefore, low levels of wild-type PTEN ensures maintenance of AKT-1–mediated cell survival networks, evasion of apoptosis, and enhanced cell invasion properties of human melanoma cells (31–33).

The targeted increase in PTEN level and/or activity in melanoma cells should potentially disrupt the PDK-1–mediated activation of AKT-1 and thereby negatively regulate AKT-1 cell survival signaling (33). Steady-state levels of PTEN protein are highly regulated by the E3 ubiquitin ligase NEDD4-1, which specifically targets PTEN for proteasomal degradation (34). In the present study, we demonstrate that I3C selectively stabilizes PTEN protein to induce an apoptotic response in human melanoma cells that express wild-type PTEN protein. We further show that I3C disrupts the NEDD4-1–dependent ubiquitination and degradation of PTEN protein, and directly interacts with purified NEDD4-1 protein. Our study implicates this E3 ubiquitin ligase as a biologically significant I3C target protein in human melanoma cells and further suggests that by stabilization of PTEN protein, indolecarbinol-based compounds could potentially be developed in new therapeutic strategies for treatment of human melanoma.

Melanoma cell lines G-361, SK-MEL-28, SK-MEL-30, RPMI-7951, and normal human primary epidermal melanocytes were all purchased from American Type Culture Collection (ATCC), and were authenticated according to the ATCC guidelines. I3C and MG-132 were purchased from Sigma-Aldrich. Two different PTEN siRNAs, two distinct NEDD4-1 siRNAs and the corresponding scrambled siRNAs, and the HiPerFect reagents were purchased from Qiagen. Antibodies to PTEN, NEDD4-1, ubiquitin, Bcl-2, and MDM-2 were obtained from Santa Cruz Biotechnology, the antibody to Hsp90 was purchased from BD Biosciences and antibodies to PARP, cleaved PARP, cleaved caspase-3, p53, phosphor-MDM2, phospho-AKT-1, and AKT-1 were acquired from Cell Signaling Technology. The secondary anti-mouse and anti-rabbit antibodies conjugated with horseradish peroxidase (HRP) were obtained from Bio-Rad Laboratories. Secondary antibodies conjugated to fluorescent probes were purchased from Molecular Probes/Invitrogen. All primers used in RT-PCR reactions were synthesized by IDT Technologies.

The G-361 melanoma cells were cultured in Modified McCoy's 5A cell media supplemented with 10% fetal bovine serum (Gemini Bio Products), 2 mmol/L l-glutamine, and 2.5 mL of 10,000 U/mL penicillin/streptomycin mixture (Gibco, Life Technologies). SK-MEL-28 and SK-MEL-30 cells were cultured in DMEM with 4.5 g/L glucose supplemented with 10% fetal bovine serum, 2 mmol/L l-glutamine, 2.5 mL of 10,000 U/mL penicillin/streptomycin mixture in addition to 1× of MEM nonessential amino acid (Gibco, Life Technologies). RPMI-7951 cells were cultured in DMEM containing 4.5 g/L glucose, 114 mg/L sodium pyruvate, and 2 mmol/L l-glutamine, supplemented as described above. Normal human primary epidermal melanocytes were cultured in Dermal Cell Basal Medium supplemented with the Melanocyte Growth Kit (ATCC) and 2.5 mL of 10,000 U/mL penicillin/streptomycin mixture. The cells were incubated in tissue culture dishes (Nalgene Nunc) at 37°C with controlled humidity and 5% CO₂ air content. All treatment conditions used cells at approximately 80% confluency. I3C was dissolved in 99.9% high-performance liquid chromatography (HPLC) grade DMSO (Sigma-Aldrich) and the final dilution was performed in the media aliquots used for treatment. For use of the MG-132 proteasome inhibitor, G-361 and SK-MEL-30 melanoma cells were treated with or without 200 μmol/L I3C for 48 hours, and the cell cultures were incubated in the presence or absence of 10 μmol/L MG-132 for the last 5 hours of the treatment. The DNA content of propidium iodide–stained nuclei from 48-hour I3C-treated and untreated cells was determined by flow cytometry as previously described (21).

Western blot analyses of samples electrophoretically fractionated on 8% to 10% acrylamide gels were carried out as previously described (20, 21). Enhanced chemiluminescence (ECL) lightening reagents were used to visualize the primary antibody–bound protein bands in nitrocellulose membranes and the results were captured on the ECL Autoradiography Film (GE Healthcare). Immunoprecipitations were carried out as previously described (20). The cell lysates were immunoabsorbed using either mouse PTEN-specific antibodies or nonimmune antibodies bound to protein-G–coupled precleared Sepharose beads. Of note, 1% by volume of total protein extracts were separately analyzed as the input control before adding beads and antibodies. Immunoabsorbed and the input material were fractionated in a 12% polyacrylamide gel, transferred to nitrocellulose membranes, and then the samples were probed with ubiquitin- and mouse PTEN-specific antibodies. HSP90 was used as a gel-loading control for the input samples. Membranes containing the immunoprecipitated material were stripped and reprobed with a rabbit PTEN-specific antibody as a loading control for total PTEN in the immunoprecipitated samples. Negative control samples containing IgG lacked any immunoprecipitated proteins.

Total RNA was extracted using the RNeasy Extraction Kit for mammalian cells (obtained from Qiagen) and spectrophotometrically quantified by absorbance at 260 nm. Reverse transcription (RT) reactions were carried out using RT-MMLV reverse transcriptase (Invitrogen) and the cDNA was used for PCR reactions using 10 pmol/L of the following primers: PTEN forward, CCACCAGCAGCTTCTGCC ATCTCT and reverse, CCAATTCAGGACCCACACGACGG; and GAPDH forward, TGAACGGGAAGCTCACTGG and reverse, TCCACCACCCTGTTGCTGTA. PCR conditions were as follows: 30 seconds at 94°C, 30 seconds at 55°C, and 30 seconds at 72°C for 28 cycles. PCR products were electrophoretically fractionated in 1.5% agarose gels containing 0.01% Gel Red (Biotium) for DNA staining along with 1 kb plus DNA ladder and further visualized by a UV transilluminator.

Transfected G-361 cells were seeded on 6-well plates at 40% confluency, and after 24 hours of incubation, the cells were treated with or without I3C for 48 hours. The cells were collected in 1-mL sterile PBS and 1 μL of quenched NucView 488 probe (35) was added to each sample protected from light and incubated on ice for 30 minutes. The fluorescence emitted by the caspase-3 cleaved NucView 488 probe was analyzed using a Coulter Elite flow cytometer (Beckman-Coulter) and quantified by EPICS software.

Transfections with siRNA were carried out according to the Qiagen HiPerFect transfection protocol as we previously described (19, 20). Briefly, G-361 cells were plated at 50% confluency on 6-well plates in full cell culture media, and 150 ng of siRNA was mixed with 100 μL of media without serum or antibiotics. HiPerFect transfection reagent was then added to the siRNA suspensions. After vortexing, each mixture was incubated for 10 minutes at room temperature, and added to cells in a dropwise manner while gently agitating to ensure even distribution of the siRNA. The cells were incubated for 24 hours in full growth media, and after aspirating the media, the cells were treated in the presence or absence of 200 μmol/L I3C in full-growth media for 48 hours, with the media being replaced every 24 hours of treatment.

Approximately 1 million G-361 cells in a total volume of 0.1 mL of Matrigel were injected subcutaneously into each lateral flank of NIH III athymic nude mice. The resulting tumors were allowed to grow to a mean starting volume of 146 ± 10 mm³. The animals were then randomized into two groups, each containing 5 mice; a vehicle control group that was treated with DMSO and an I3C-treated group that were injected daily in their scruff with 200 mg/kg body weight of I3C. The resulting tumor volumes were measured every alternate day for a period of 19 days using calipers. The tumor volumes were calculated using the standard formula: (width² × length)/2, and changes in tumor volumes were calculated using the formula: 100 + {(T_f − T_i)/T_i × 100}, where T_f is the final mean tumor volume and T_i is the initial mean tumor volume. The data represent the mean ± SEM (**, P < 0.01) for 5 mice per group, each with two tumors, one in each flank. At terminal sacrifice, tumor xenografts were harvested, and a portion of each tumor was fixed in 4% paraformaldehyde for 1 hour at room temperature, followed by a PBS wash and subsequently immersed in 3% sucrose overnight at 4°C. The resulting tissues were embedded in optimal cutting temperature (OCT) and 10-μm thin sections were taken for immunofluorescence studies using primary antibodies to PTEN.

Structures of NEDD4-1 protein domains were obtained from the Protein Data Bank (PDB; http://www.rcsb.org/pdb/home/home.do) with accession numbers 2ONI and 2NSQ for the HECT ubiquitination domain and C2 domain, respectively. The PRODRG server was used to produce the topology files for modeling the I3C structure. NEDD4-1 domain and I3C structures were loaded into the Hex Protein Docking program (http://www.csd.abdn.ac.uk/hex_server/), and binding simulations were performed using shape and electrostatics as restrictive parameters (36). Before docking the structures, all water molecules and hetero molecules were manually removed by editing each PDB file. Modeling results were visualized using the PyMol program (37). The schematic diagrams that illustrate the pattern of interactions between the 3-D coordinates of the protein and bound I3C molecule were generated using the LigPlot program (38). Using Hex Protein Docking software, the molecular dynamics and thermostability were calculated using free energy simulations. Further validation of the simulation revealed that 97.59% of the amino acids were in the preferred regions for Phi and Psi angles and 0% of the amino acids were in disallowed regions. A total of 2,000 solutions were analyzed with a grid dimension of 0.6, and the most favorable models were then selected on the basis of E values. NEDD4-1 amino acid residues that were within 3.5 Å from indolecarbinol-predicted binding clusters were identified and the α-carbon numbers labeled.

Isothermal titration calorimetry experiments were performed at 4°C with a reference power of 4 μcal/s using an ITC200 instrument (MicroCal GE Healthcare Bio-Sciences). The purified NEDD4-1 protein and I3C were stored in buffer containing 25 mmol/L Tris–HCl at pH 8.0, 100 nmol/L NaCl, and 2 mmol/L MgCl₂. Each titration in the reaction cell used 400 μL of protein at a concentration of 10 μmol/L and 130 μL of I3C up to a concentration of 100 μmol/L. Thirteen 2-μL injections of I3C into the cell containing NEDD4-1 were performed with a duration of 4 seconds per injection and 180-second spacing. To extract the thermodynamic parameters (binding enthalpy and dissociation constant), titration curves were fit by a nonlinear, least-squares method. From these data, the changes in enthalpy and free energy were calculated using the following equations: ΔG = −RT Ln K; ΔG = ΔH − TΔS. The binding curve yielded an equilibrium dissociation constant of approximately 88.1 nmol/L.

To initially determine whether human melanoma cells are sensitive to the antiproliferative effects of I3C, four human melanoma cell lines, G-361, SK-MEL-30, SK-MEL-28, and RPMI-7951, with distinct mutational profiles (described in Fig. 1), as well as human primary epidermal melanocytes, were treated with or without 200 μmol/L I3C for 48 hours, and potential cell-cycle effects were analyzed by flow cytometry of propidium iodide–stained nuclear DNA. This concentration of I3C is optimally effective in other human cancer cell lines (12, 14) as the functional intracellular concentration of this indolecarbinol entering cells is less than 0.3% of the final concentration in the cell culture media (39). As shown in Fig. 1, I3C induced a G₁ cell-cycle arrest of G-361 and SK-MEL-30 melanoma cells, both of which express wild-type PTEN and wild-type p53. In contrast, SK-MEL-28 and RPMI-7951 melanoma cell lines, which display A499G-mutant PTEN or null PTEN genotypes, respectively, were relatively unaffected by 48-hour I3C treatment (Fig. 1). SK-MEL-28 melanoma cells did show some sensitivity to I3C at time points beyond 48 hours (data not shown). Importantly, normal melanocytes were resistant to the I3C (Fig. 1), suggesting that this indolecarbinol compound triggers a melanoma-selective response.

Because the strong I3C antiproliferative response correlated with the wild-type PTEN status of the melanoma cells lines, the potential effects on PTEN expression was assessed in G-361 and SK-MEL-30 cells treated over a concentration range of I3C for 48 hours. Western blot and RT-PCR analysis of PTEN protein and mRNA levels revealed that I3C strongly upregulated PTEN protein levels without altering the level of PTEN transcripts in both wild-type PTEN-expressing cell lines (Fig. 2A). The HSP90 protein and GAPDH transcript gel-loading controls remained unchanged throughout the dose response. The maximal I3C response in the G361 cells occurred at somewhat lower concentrations compared with the SK-MEL-30 melanoma cells. Densitometry of the Western blot analyses revealed that in G361 cells, treatment with 200 μmol/L I3C stimulated over a 2-fold increase in PTEN protein levels compared with the untreated cells with no increase in PTEN transcript levels (Fig. 2B). In SK-MEL-30 cells, the fold stimulation in PTEN protein was much higher, with 200 μmol/L I3C inducing an almost 10-fold increase in PTEN protein levels with no effect on PTEN transcript levels (Fig. 2B). In comparison with G361 and SK-MEL-30 cells, I3C had no effect on the levels of the mutant A499G-PTEN expressed in SK-MEL-28 melanoma cells, whereas PTEN expression remained ablated in RPMI-7951 cells, which display a null PTEN genotype (Fig. 2C). Normal primary melanocytes express wild-type PTEN, however, I3C had no effect on PTEN protein levels (Fig. 2C), suggesting a selective effect of I3C in melanoma cells compared with their normal cell counterpart.

To initially determine whether I3C upregulated PTEN protein levels by preventing its ubiquitin-mediated proteasomal degradation, 48-hour I3C-treated and untreated G-361 and SK-MEL-30 melanoma cells were incubated in the presence or absence of 10 μmol/L MG-132, a 26S-specific proteasome inhibitor, during the last 5 hours of the incubation. PTEN protein was assessed in total cell lysates by Western blot analysis. As shown in Fig. 2D, in the presence of MG132, PTEN protein levels were observed to increase in cells treated with or without I3C, and the combination of MG-132 and I3C resulted in an enhanced accumulation of PTEN protein. To further determine whether I3C stabilizes PTEN protein through downregulation of its ubiquitination, cells extracts from I3C-treated and untreated cells exposed to MG-132 during the last 5 hours of the incubation were immunoprecipitated with either PTEN-specific or nonimmune antibodies and Western blot analyses of electrophoretically fractionated samples probed with anti-ubiquitin antibodies. As shown in Fig. 2E, I3C strongly downregulated the level of ubiquitinated PTEN under conditions in which the level of immunoprecipitated PTEN protein increased compared with cells not treated with I3C. PTEN-associated ubiquitin was not detected in immunoprecipitations carried out with nonimmune antibodies, demonstrating the fidelity of the assay.

The I3C-mediated stabilization of PTEN protein predicts that I3C should disrupt cell survival pathways and potentially induce apoptosis of melanoma cells in a PTEN-dependent manner. One of the key downstream effectors of PTEN is AKT-1, as PTEN dephosphorylates PIP3, and thereby prevents the PDK-1 phosphorylation and activation of AKT-1 at the plasma membrane (31–33). G-361 melanoma cells were treated with a range of I3C concentrations for 48 hours, and the levels of cleaved poly(ADP-ribose) polymerase (PARP) molecule, which corresponds to apoptosis-related DNA repair activity, were compared with the levels of phosphorylated and total AKT-1 protein. As shown in Fig. 3A, Western blot analysis of total cell extracts using primary antibodies that detect both full-length and cleaved PARP showed that PARP cleavage was maximally induced by treatment with concentrations of I3C that stabilize PTEN and downregulate phosphorylated AKT-1. The level of total AKT-1 protein and the HSP90 gel-loading control remained relatively unchanged under these conditions.

The activity of caspase-3, an upstream regulator of PARP processing, was examined using the highly sensitive NucView488 live-cell assay in which the NucView488 probe releases its fluorescent moiety from a quenched state upon specific cleavage by active caspase-3 (35). Treatment of G-361 cells with 200 μmol/L I3C for 48 hours produced a 5-fold induction of caspase-3 activity compared with vehicle control DMSO-treated cells (Fig. 3B). The effect was slightly attenuated by 72 hours, a time point when more cells displayed cytotoxic effects (blebbing, rapid detachment from the plate surface), associated with later stages of apoptosis. To determine the role of I3C-induced stabilization of PTEN protein in melanoma cell apoptosis, initially G-361 cells were transfected with either PTEN siRNA or with scrambled siRNA, and caspase-3 activity was assessed in cells treated with or without I3C for 48 hours. As shown in Fig. 3C, siRNA knockdown of PTEN prevented the I3C-induced apoptotic response, whereas transfected scrambled siRNA had no effect on the indolecarbinol-induced apoptosis. Western blot analyses demonstrated the efficiency of the PTEN siRNA knockdown in I3C-treated cells compared with the scrambled siRNA controls.

Because of the possibility of off-target effects of the PTEN siRNA, siRNA transfections in both G361 and SK-MEL-30 cells were carried out with a different set of PTEN-specific and scrambled siRNAs. As shown in Fig. 3D, I3C stimulated an increase in PTEN protein in cells transfected with the scrambled siRNA, whereas transfection with PTEN siRNA efficiently knocked down PTEN protein levels. In both melanoma cell lines, knockdown of PTEN disrupted the I3C regulation of phosphorylated Akt and prevented the increase in the apoptotic response based on production of the cleaved caspase-3 (Fig. 3D). The production of total Akt-1 was unaffected by either I3C treatment or the PTEN siRNA. Thus, in human melanoma cells expressing wild-type PTEN, production of PTEN is required for the I3C apoptotic response.

PTEN protein stability can be regulated by the E3 ubiquitin ligase NEDD4-1, a member of neural precursor cell expressed, developmentally downregulated 4 E3 molecules, which attaches ubiquitin moieties to the PTEN protein on residues Lys 289 and Lys 48, effectively targeting it for proteasomal degradation (34). Regardless of the PTEN status, human melanoma cells as well as normal melanocytes express generally similar levels of NEDD4-1 protein in the presence or absence I3C, although the SK-MEL-30 cells did display a reduced level of NEDD4-1 in I3C-treated cells (Fig. 4A). Because I3C downregulated PTEN ubiquitination in wild-type PTEN-expressing melanoma cells, we explored potential connections between I3C and NEDD4-1 that could regulate PTEN protein stability. To initially test the role of NEDD4-1 in the I3C-induced stabilization of PTEN protein, NEDD4-1 expression in G-361 cells was knocked down by transfection of NEDD4-1–specific siRNA. A control set of transfections was carried out with scrambled siRNA. Western blot analysis of electrophoretically fractionated cell extracts revealed that siRNA knockdown of NEDD4-1 triggered an accumulation of PTEN protein in I3C-treated and untreated cells similar to that observed after I3C treatment of cells transfected with scrambled siRNA (Fig. 4B). Thus, knockdown of the NEDD4-1 E3 ubiquitin ligase mimics the effects of I3C on PTEN protein stabilization.

Several downstream effectors of PTEN signaling that are involved in cell survival and/or apoptotic pathways were examined in melanoma cells transfected with either NEDD4-1 siRNA or scrambled siRNA. An increase in PTEN protein would be predicted to reduce the cellular level of phosphorylated AKT-1 (31, 33). As also shown in Fig. 4B, siRNA knockdown of NEDD4-1 downregulated the level of phosphorylated AKT-1 in the presence or absence of I3C to a level that closely approximates that observed in I3C-treated control cells. Activation of AKT-1 and its other isoforms has been implicated in cell survival pathways in human cancer cells through its phosphorylation of the ubiquitin ligase MDM2 (40), which in turn promotes the MDM2-mediated ubiquitination and degradation of p53, which disrupts the p53 intrinsic apoptotic program (41). Western blot analysis further revealed that in G-361 cells transfected with scrambled siRNA, I3C induced the downregulation of Ser 166 phosphorylated MDM2 with a corresponding moderate accumulation of p53 protein. Both effects were mimicked in either I3C-treated or untreated cells transfected with NEDD4-1 siRNA (Fig. 4B). Consistent with these effects on Ser 166 phosphorylated MDM2, I3C downregulated the level of the Bcl-2 antiapoptotic proteins (Fig. 4B), which is a downstream target of p53 (42) that has been connected to melanoma pathogenesis. Knockdown of NEDD4-1 attenuated Bcl-2 protein levels that was further reduced after I3C treatment (Fig. 4B).

To assess the potential effects of NEDD4-1 knockdown on melanoma cell apoptosis, G-361 cells were transfected with either scrambled siRNA or NEDD4-1 siRNA, and then incubated with the NucView488 fluorescent probe to assay caspase-3 activity in live cells. Knockdown of NEDD4-1 induced a strong apoptotic response in the absence or presence of I3C (Fig. 4C, NEDD4-1 siRNA) that was approximately equivalent to the I3C-nduced caspase-3 activity in control cells transfected with scrambled siRNA (Fig. 4C).

As a complementary functional approach, NEDD4-1 production was knocked down in the absence or presence of PTEN knockdown by transfection of the corresponding siRNA in both G361 and SK-MEL-30 melanoma cells, and apoptotic signaling in cells examined in comparison those transfected with scrambled siRNA. As shown in Fig. 4D, transfection of the PTEN- and NEDD4-1–specific siRNA was highly efficient in knocking down production of the corresponding proteins. In both melanoma cell lines, knockdown of NEDD4-1 stabilized PTEN levels in the presence or absence of I3C, although compared with cells receiving scrambled siRNA, SK-MEL-30 cells transfected with NEDD4-1 siRNA expressed a reduced level of PTEN protein. Also, similar to untransfected SK-MEL-30 cells (Fig. 4A), I3C reduced the level of total NEDD4-1 protein in SK-MEL-30 cells transfected with scrambled siRNA (Fig. 4D). The I3C-stimulated production of cleaved caspase-3 in both melanoma cell lines was completely blocked in PTEN knockdown cells, whereas in NEDD4-1 knockdown cells cleaved caspase-3 was produced in I3C-treated or untreated cells (Fig. 4D). Importantly, cotransfection of siRNA for PTEN along with siRNA for NEDD4-1 in both cell lines resulted in low levels of cleaved caspase-3 in cells treated with our without I3C (Fig. 4D). Thus, I3C-regulated apoptotic signaling through NEDD4-1 requires the presence of the wild-type PTEN protein.

The essential role of the NEDD4-1 ubiquitin ligase in the I3C-induced accumulation of PTEN protein and apoptotic response in human melanoma cells suggests the possibility of a direct functional interaction between I3C and NEDD4-1. Computer-aided molecular binding simulation, which takes into account the known 3-D crystal structure of NEDD4-1 and the chemical structure of I3C, was used to evaluate potential binding sites for I3C within the NEDD4-1 protein structure. A Hex server protein-ligand docking analysis was used for the initial simulations because of the ability to scan the large NEDD4-1 protein surface for potential docking sites with a small ligand such as I3C (36). In silico molecular modeling revealed a potential I3C binding site on the catalytic HECT domain of NEDD4-1, which mediates the ubiquitin ligase activity of the molecule. The electrostatic potential map of the HECT domain shows that I3C binds to a negative surface and forms interactions that are within 4 Å of residues on the HECT domain (Fig. 5A). Upon further inspection of the model using the LigPlot program, it is evident that I3C makes Van der Waals interactions (within 3.5 Å) with residues Ser 652, Phe 656, Ile 693, Val 696, Phe 692, Phe 608, Glu 654, and Leu 651 (Fig. 5B). Calculated force field energy values for the in silico defined I3C interaction with the NEDD4-1 HECT domain yielded negative values (E value = −170.9 kJ/mol), demonstrating that the structures are predicted to be stable. Simulations that used the C2 domain structure of NEDD4-1 displayed no binding interactions with I3C (data not shown), indicating that I3C likely does not interact with this region of the NEDD4-1 molecule.

Isothermal titration calorimetry was used to directly test the thermodynamics and binding of I3C to purified NEDD4-1 protein. Increasing concentrations of I3C were titrated into a solution of NEDD4-1, resulting in an enthalpically favorable exothermic interaction, which is indicative of an I3C interaction with NEDD4-1. To extract the thermodynamic parameters of binding enthalpy and dissociation constant, titration curves were fit by a nonlinear, least-squares method. Analysis of the binding of increasing concentrations of I3C to NEDD4-1 protein revealed that this indolecarbinol protein interacts with NEDD4.1 with a calculated equilibrium dissociation constant of approximately 88.1 ± 13.0 μmol/L (Fig. 5B).

To assess the in vivo effects of I3C on melanoma cell growth and regulation of PTEN protein levels, G-361 cell–derived tumor xenografts in NIH III athymic mice were first allowed to grow to an average volume of approximately 150 mm³, and then the mice were injected subcutaneously with either I3C (200 mg/kg body weight) or with the DMSO vehicle control over an 18-day time course. This dose of injected I3C closely corresponds to the level of I3C administered in clinical trials. In vehicle control treated animals, the G-361 cell tumor xenografts showed robust growth, whereas I3C strongly suppressed the growth rate of G-361cell–derived tumor xenografts (Fig. 6A). The resulting tumors from vehicle control-treated animals displayed highly concentrated gross tumor vascularization and were dense, which is consistent with the rapid growth of cells within the tumor (Fig. 6A, micrograph inset). In contrast, the tumor xenografts from I3C-treated animals were smaller in size and appeared less vascularized (Fig. 6A, micrograph inset). Immunofluorescence analysis of 10-μm cryostat sections from tumors excised from I3C-treated and untreated animals revealed a strong increase in PTEN protein staining in selective regions of the tumor xenografts from I3C-treated animals compared with tumors from vehicle control-treated animals (Fig. 6B, left). The areas in the tumors with the most prominent PTEN protein staining were proximal to structures that resembled blood vessels (Fig. 6B, right). Analyses of the merged images show that in I3C-treated animals, a significant fraction of the tumor cells express higher levels of PTEN protein. In independent sets of tumor xenografts from I3C-treated and untreated animals, the PTEN, NEDD4-1, and phosphorylated Akt (Akt-p) protein densities were analyzed. As shown in Fig. 6C, in vivo treatment with I3C induced an increase in PTEN protein and inhibited the presence of phosphorylated Akt without altering NEDD4-1 protein levels.

The therapeutic enhancement of cellular PTEN levels or lipid phosphatase activity represents a potential strategy to effectively target melanoma cells that express the wild form of this tumor-suppressor protein (25, 26, 31, 32). Relatively little was known about the mechanism by which indolecarbinol compounds mediate their antiproliferative effects in human melanoma cells. We functionally established that I3C triggers an apoptotic response in human melanoma cells by increasing the level of wild-type PTEN protein resulting from its stabilization due to the disruption of the NEDD4-1–dependent ubiquitination and 26S proteasomal degradation of PTEN protein. For example, siRNA knockdown of PTEN prevented and knockdown of NEDD4-1 mimicked the I3C stimulation of caspase-3 activity and apoptotic response, and this NEDD4-1 apoptotic response required the presence of wild-type PTEN. Furthermore, we observed that I3C directly interacts with purified NEDD4-1 protein, which identifies this E3 ubiquitin ligase as a new biologically significant I3C target protein that is critical for the control of AKT-1–mediated cell survival cascades in human melanoma cells.

We propose a direct functional connection between I3C apoptotic signaling and the disruption of NEDD4-1 E3 ubiquitin ligase targeting of PTEN protein due to the loss of NEDD4-1 activity and/or protein–protein interactions. In silico molecular modeling revealed a potential I3C binding site in the catalytic HECT domain of NEDD4-1, which is responsible for the ubiquitin ligase activity of the molecule. Consistent with the computer simulations, isothermal titration calorimetry directly demonstrated that I3C binds to purified NEDD4-1 protein with a dissociation constant of approximately 88 μmol/L. The C2 and HECT domains of NEDD4-1, which mediate the membrane localization and ubiquitin ligase activity, respectively, bind to each other, resulting in autoinhibition of NEDD4-1 activity (43). Conceivably, I3C binding to the HECT domain could potentially inhibit NEDD4-1 activity by stabilizing the C2-HECT autoinhibitory intramolecular interactions. Furthermore, NEDD4-1 activity can be stimulated by specific activators such as the NEDD-family interacting proteins, NDFIP1 and NDFIP2, which function by binding to the WW domains of NEDD4-1 via their PY motifs, and thereby abolish the autoinhibitory interactions between the C2 and HECT domains of NEDD4-1 (43, 44). A recent study showed that a p34 protein interacts with the WW1 domain NEDD4-1 to regulate NEDD4-1 stability and thereby PTEN ubiquitination (45). Consistent with study, in SK-MEL-30 cells, but not the G361 cells, we observed that I3C treatment causes a moderate downregulation of NEDD4-1 protein. Hence, it is conceivable that I3C interactions with NEDD4-1 may potentially alter or disrupt protein binding to NEDD4-1 in a way that downregulates NEDD4-1 activity and/or protein levels and thereby stabilizes PTEN protein. In contrast, one study suggests that NEDD4-1 activity is dispensable for the regulation of PTEN stability and localization in mouse cells and tissue (46), although melanoma or other skin-derived transformed cells were not tested. Conceivably, different regulatory pathways could be acting on PTEN depending on the transformation state or tissue origin of the cells. Studies are currently under way to determine the precise I3C interaction site in NEDD4-1 and to access the functional consequences of the interaction.

The in silico–predicted I3C interaction site on NEDD4-1 possesses striking structural and chemical similarities with the I3C binding pocket that we characterized in neutrophil elastase, the first identified indolecarbinol target protein (21). Antiproliferative responsiveness of I3C in human breast cancer cells is mediated by the ability of I3C to act as a noncompetitive inhibitor of elastase enzymatic activity (19–21). In contrast to breast cancer cells, antiproliferative signaling by I3C in melanoma cells is not mediated by the disruption of elastase cleavage of CD40 (data not shown), which implicates NEDD4-1 as a critical I3C target protein in melanoma cells expression of wild-type PTEN. Thus, an emerging concept from our previous and current studies is that I3C antiproliferative responsiveness can be triggered in different cancer cell types by the differential expression of distinct I3C target proteins.

Disruption of PTEN causing PI3Kinase/AKT activation and NRAS/BRAF mutations leading to constitutive activation of signaling via the MAPK pathway are considered two of the key drivers of melanomagenesis (1, 2, 47). Therefore, targeting components of these pathways with either single agents or combinations have been a developing therapeutic strategy in recent years. The BRAF-mutant melanomas showed high dependency on B-RAF kinase activity and so inhibition of the oncogenic B-RAF resulted in striking reduction in tumor volume in preclinical and clinical trials (48). On the basis of these observations, the FDA approved PLX4032/vemurafenib (Plexxicon) and dabrafenib (GlaxoSmithKline) to treat metastatic melanoma, which are two selective and potent inhibitors of the oncogenic BRAF(V600E). However, this treatment has been associated with inevitable development of rapid or acquired tumor resistance within months of the initial treatment (49). The therapeutic potential of selective inhibitors of BARF(V600E) has also been restricted by their ability to paradoxically stimulate growth in tumors harboring a wild-type BRAF (50). Also, it has been recently reported that loss of PTEN could confer intrinsic resistance to BRAF inhibitors in melanoma cells (51), suggesting the importance of taking into account the activities of different cell signaling cascades in the treatment of human melanomas.

Our study suggests that I3C-based compounds have the potential to be developed in new therapeutic strategies for treatment of human melanoma either alone or in combination with other therapeutic compounds to target different aberrant signaling pathways. Furthermore, given the significant degree of intratumoral heterogeneity in melanoma (1, 2), combinational therapies could conceivably be used to simultaneously target multiple cell subpopulations within the tumor. In this regard, a combination of BRaf and MEK inhibitors (dabrafenib/trametinib) is now FDA approved for patients with melanoma. Human primary epidermal melanocytes that express wild-type PTEN were not responsive to I3C, which indicates that I3C should not adversely affect normal skin cell proliferation. The precise mechanism by which normal melanocytes remain resistant to I3C are not known, and we have observed that in athymic mice, injections of I3C over 3 weeks have no apparent effects on animal weight or behavior. Clinical trials for I3C in women at risk for breast cancer have provided a valuable assessment of clinical safety and tolerated doses of I3C (52, 53). Cytotoxicity results from this clinical trial have shown that patients are capable of receiving as high as 800 mg/kg/d I3C without any adverse side effects, which is a concentration that approximates both the 200 μmol/L I3C used in the cultured melanoma cells and the level of I3C used in the in vivo tumor xenograft experiments. Our findings implicate I3C as a potential therapeutic for melanomas that express low levels of wild-type PTEN protein in a background of either wild-type BRAF or BRAF(V600E) by disrupting the NEDD4-1–mediated ubiquitination of PTEN and enhancing the total levels of this tumor-suppressor protein. The development of structural studies on I3C interactions with its potential molecular target NEDD4-1 may provide valuable insight for the design of highly customized I3C derivatives that would more efficiently stabilize PTEN at lower doses and maximize the potential therapeutic effects.

No potential conflicts of interest were disclosed.

Conception and design: I. Aronchik, A. Kundu, G.L. Firestone

Development of methodology: A. Kundu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Aronchik, A. Kundu, J.G. Quirit

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Aronchik, A. Kundu, J.G. Quirit, G.L. Firestone

Writing, review, and/or revision of the manuscript: I. Aronchik, A. Kundu, G.L. Firestone

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Kundu, G.L. Firestone

Study supervision: G.L. Firestone

The authors thank Kathleen A. Durkin for her help with the in silico modeling during the early stages of the work and Kevin Poindexter, Lauren Meyer, Michelle Khouri, and Andrew Gabrielson for their helpful suggestions during the writing of this article.

This study was supported by NIH Public Service grant CA164095 awarded from the National Cancer Institute. I. Aronchik was a postdoctoral trainee supported by a National Research Service Grant (CA-09041) awarded by the NIH.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.