BMS-345541 Targets Inhibitor of κB Kinase and Induces Apoptosis in Melanoma: Involvement of Nuclear Factor κB and Mitochondria Pathways

Metastatic melanoma is the most deadly form of skin cancer. The effects of systemic therapy for aggressive metastatic melanoma have been minimal. Immunotherapy, or even biochemotherapy that combines chemotherapeutic drugs and interleukin-2 or IFN-α, has also failed in large phase III clinical trials (1), implying that the resistance mechanisms in melanoma are complex. Through inactivation of apoptosis, melanoma cells obtain both advanced survival capacity and resistance to chemotherapeutic agents (2). The apoptotic machinery can be regulated in part by the transcription factor, nuclear factor κB (NF-κB), which regulates transcription of the Bcl-2 family members, which in turn, control the mitochondrial release of cytochrome c and the activation of certain caspases (3). New therapeutic approaches might take advantage of the intrinsic cell death machinery to “trick” the tumor cells into committing programmed death.

The inhibitor of κB kinase (IKK) complex is a key regulator of NF-κB signaling. It consists of two catalytic subunits, IKKα and IKKβ, and a regulatory component, IKKγ (4). The kinase activity of IKKα and IKKβ can be induced with cytokine challenge, resulting in consequent phosphorylation, ubiquitination, and degradation of substrate IκB proteins. IκB is composed of a family of inhibitory proteins (IκBα, IκBβ, IκBε, IκBγ, or Bcl-3) that largely retain the transcription factor NF-κB in the cytoplasm by masking the nuclear localization signal of NF-κB (5). NF-κB represents a family of five Rel proteins, c-Rel, RelA/p65, RelB, NF-κB1 (p50 and its precursor, p105), and NF-κB2 (p52 and its precursor, p100; ref. 6). In many cancers, NF-κB is persistently activated, which protects developing tumor cells from death and thereby contributes to tumorigenesis (7) and cancer therapy resistance (8).

In previous studies, we have presented evidence that IKK is constitutively active in human melanoma cells, which leads to NF-κB activation and results in aberrant overexpression of chemokines such as CXC ligand 1 (CXCL1) and/or CXCL8 (9). These chemokines have been implicated in melanocyte transformation and melanoma tumor progression both in vivo and ex vivo (7, 10). We have shown that the CXCL1 chemokine could induce activation of IKK in normal human melanocytes (9) and potentiate melanoma formation in a transgenic mouse model (11). Because IKK is a key molecular complex specifically regulating IκB proteins and subsequently targeting NF-κB, we speculated that IKK would be a good therapeutic target for malignant melanoma. A novel compound, BMS-345541, was identified as a highly selective IKKβ inhibitor (12). To determine whether BMS-345541 manipulates the apoptotic machinery through targeting the highly active IKK complex in melanoma cells, we delivered the IKK inhibitor, BMS-345541, to human melanoma cells in vitro and in vivo. We investigated in detail the effect of BMS-345541 on IKK/NF-κB signaling pathway, chemokine production, tumorigenesis, and apoptosis. The results presented here show that inhibition of constitutive IKK activation in human melanoma by BMS-345541 directly reduced NF-κB activity and induced mitochondria-mediated apoptosis largely through caspase-independent pathways, resulting in antitumorigenic effects.

Reagents and cell culture. BMS-345541 [4(2′-aminoethyl) amino-1,8-dimethylimidazo (1,2-α)-quinoxaline]-4,5-dihydro-1,8-dimethylimidazo(1,2-α)quinoxalin-4-one-2-carboxylic acid, was prepared by the described procedure (12) and supplied by the Bristol-Myers Squibb Pharmaceutical Research Institute (New York, NY). BMS-345541 was dissolved in DMSO to produce a 50 mmol/L stock solution for in vitro experiments or stock solutions of BMS-345541 (10, 25, and 75 mg/10 mL) were dissolved in water and the pH value was adjusted to 7.0 for in vivo experiments. The super-repressor form of human IκBα (S32 and 36A) resistant to degradation and mutant IKKβ (K44M) were kindly provided by Dr. Javier Piedrafita (Sidney Kimmel Cancer Center, University of California, San Diego School of Medicine, San Diego, CA). Antibodies to IKKα (H-744), IKKβ, Bcl-2, Bax and apoptosis-inducing factor (AIF) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Dihydroethidine, 3,3dihexyloxacarbocyanine iodide (DiOC₆) and pan-caspase inhibitor (Z-VAD-fmk) were purchased from Molecular Probes (Eugene, OR). Normal human epidermal melanocytes were provided by the Skin Disease Research Center in Vanderbilt University School of Medicine (Nashville, TN). Normal human epidermal melanocytes were cultured in 154 medium with 1× human melanocyte growth supplement (Cascade Biologics, Inc., Portland, OR). The melanoma cell lines, SK-MEL-5, A375, and Hs 294T were purchased from American Type Culture Collection (Manassas, VA) and were cultured in DMEM/Ham's F-12 medium containing 10% fetal bovine serum, 2 mmol/L of l-glutamine, 100 μmol/L of MEM nonessential amino acids (Invitrogen Corporation, Carlsbad, CA), and 1 mmol/L of sodium pyruvate (Sigma-Aldrich, St. Louis, MO).

Approaches of drug delivery and tumor measurement. Animal experimentation was undertaken according to protocols approved by the Institutional Animal Care and Use Committee at Vanderbilt University. BMS-345541 solution at 10 mL/kg body weight was orally administered to the mouse using a modified dull 19G1_1/2-gauge needle connected to a 1 mL syringe. Tumor size was measured with an electronic digital caliper. Tumor volume was calculated by width² × length × 0.52 and expressed as mean ± SD mm³.

Immunoprecipitation and kinase assay and Western blot analysis. Immunoprecipitation for IKK proteins and IKK activity assays were done as previously described (9). Experimental protocol for Western blotting of proteins was carried forth as we have previously described (13).

Transfection and luciferase reporter activity assay and ELISA. Melanoma cells were transiently transfected with NF-κB-responsive luciferase reporter gene and the luciferase activity was determined according to the protocol described earlier (12). CXCL1 and CXCL8 chemokine levels in vitro and in vivo were quantified following the methods as previously described (9).

NF-κB/p65 nuclear translocation assay. SK-MEL-5 cells were transfected with a green fluorescent protein (GFP)-p65 expression vector using FuGENE 6 Transfection Reagent (Roche Diagnostics Corporation, Indianapolis, IN). The next day, transfected cells were treated with DMSO vehicle as control, or 10 μmol/L of BMS-345541 1 hour prior to visualization by fluorescence microscopy. Time-lapse video microscopy was used to follow the localization of GFP-p65 in live cells. Images were captured at 1-hour intervals over a 7-hour observation period. The cytosol versus nuclear green fluorescent density was quantified by the Openlab software program. In addition, the NF-κB super-repressor (cyan fluorescent protein-IκBαAA) and the yellow fluorescent protein-p65 expression vectors were cotransfected in SK-MEL-5 cells at a 9:1 ratio. Nuclear translocation of yellow fluorescent protein–fusion protein was monitored as described above.

Caspase-3 activity assay. The activity of caspase-3 was detected with the Caspase-3 Cellular Activity Assay kit PLUS obtained from Biomol (Plymouth Meeting, PA) according to the manufacturer's protocol with slight modifications. Briefly, cell extracts were incubated with N-acetyl-Asp-Glu-Val-Asp-p-nitroaniline (Ac-DEVD-pNA) substrate at 37°C for 10 minutes and the released p-NA was measured at a test wavelength of 405 nm and a reference wavelength of 595 nm. p-NA standard (50 μmol/L) was used for calibration of the specific caspase-3 activity in the samples. Experiments were repeated twice and duplicate samples were analyzed for each experiment.

Preparation of subcellular fractions. The Mitochondria Isolation kit (Pierce, Rockford, IL) was used for isolation of mitochondria from cultured melanoma cells according to the manufacturer's instruction for the reagent-based method. The purity of mitochondria was determined by Western blotting with β-actin, and β-actin negative mitochondria preparations were determined to be pure. The procedures for preparation of whole cell lysates and nuclei were as previously described (9).

Quantification of cell death. To quantify cell death, cells were stained with propidium iodide (PI) and the DNA fragmentation in sub-G₀/G₁ phase was quantitated by fluorescence-activated cell sorting analysis. Apoptotic cells were identified as a population of cells with a lower DNA content (14). Cells to the left of the G₀/G₁ peak were therefore considered to be dead. Briefly, SK-MEL-5 cells were treated with BMS-345541 at different concentrations or for different time periods. The cells were collected by trypsinization, fixed in 70% ethanol for 2 hours on ice and stained with PI solution (PBS containing 2 μg/mL PI, 0.1% Triton X-100, and 125 units/mL RNase A) at 37°C for 30 minutes. Cell fluorescence was measured by flow cytometry with 488 nm excitation and 620 nm emission filters and resulting data were analyzed using the software program MultiCycle (Phoenix Flow Systems, San Diego, CA).

Mitochondria membrane potential (Δψm) assay. To determine mitochondrial Δψm, the purified mitochondria were resuspended in buffer [225 mmol/L mannitol, 75 mmol/L sucrose, 10 mmol/L KCl, 10 mmol/L Tris-HCl, 5 mmol/L KH₂PO₄ (pH 7.2)] and incubated with 80 nmol/L DiOC₆ for 15 minutes at 37°C, followed by fluorescence-activated cell sorting analysis with 488 nm excitation and 530 emission filters. To evaluate the generation of mitochondrial reactive oxygen species, cells treated or not treated with BMS-345541 were incubated in 5 μmol/L dihydroethidine at 37°C for 15 minutes, harvested by trypsinization, and washed with cold PBS solution thrice. Reactive oxygen species were determined by fluorescence-activated cell sorting analysis with 488 nm excitation and 585 emission filters.

Microscopy. For electron microscopy, SK-MEL-5 cells treated with BMS-345541 were pelleted by centrifugation and fixed for 1 hour in cold glutaraldehyde (2%) in cacodylate buffer and postfixed for 1 hour with OsO₄. The pellet was embedded in epoxy resin, sectioned, and collected on nickel grids. The sections were stained with uranyl acetate for 10 minutes followed by lead citrate staining for 15 minutes prior to electron microscopy observation. For confocal microscopy, SK-MEL-5 cells treated with BMS-345541 were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, Fort Washington, PA) and blocked with 2% bovine serum albumin in 1× PBS containing 0.5% Triton X-100. Cells were incubated with the anti-AIF polyclonal antibody at a dilution of 1:800. After 2 hours of incubation, cells were washed twice with 1× PBS and incubated with the secondary antibody conjugated with FITC for 1 hour. Nuclei were stained with PI (0.5 μg/mL) for 10 minutes. After washing thrice, coverslips were mounted onto microscope slides. The slides were analyzed using a confocal laser-scanning microscope.

Terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling (TUNEL) assay. The DeadEnd Fluorimetric (TUNEL) system from Promega Corporation (Madison, WI) was used to detect apoptosis in the cultured melanoma cells and paraffin-embedded tumor tissues following the manufacturer's protocol. These sections were subsequently stained for nuclei with PI (500 ng/mL). The samples were immediately analyzed under a fluorescence microscope using a standard fluorescent filter set to view the green fluorescence at 520 nm and red fluorescence of PI at >620 nm. The slides were imaged using an Openlab controlled microscope.

Small interfering RNA transfection. RNA interference of AIF was done using siGENOME SMARTpool small interfering RNA (siRNA; Dharmacon, Inc., Chicago, IL). A nonrelated siRNA sequence (GGCTACGTCCAGGAGCGCACC) that targets the GFP was used as a control. For transfection, SK-MEL-5 cells were seeded in six-well plates and transfected with 100 nmol/L siRNA using FuGENE 6 transfection reagent according to the manufacturer's recommendations. The protein level of AIF was detected in the whole cell lysate by Western blotting. Three successive rounds of transfection were done until 90% of the AIF protein was knocked down. Each transfection lasted for 24 hours in 10% fetal bovine serum medium along with an interval of 12 hours between transfections to allow time for cells to recover from the potential toxicity of transfection reagent.

Statistical analysis. Results are expressed as mean ± SD from three independent experiments. Statistical analysis was done using the unpaired Student's t test. P < 0.05 was considered statistically significant.

BMS-345541 inhibits growth of melanoma cells in vitro and melanoma tumors in vivo. Aberrant activation of NF-κB has been associated with carcinogenesis, and constitutively high IKK activity has been detected in many tumor types, including human melanoma (9, 15). IKK is a major regulator of the NF-κB pathway and therefore represents an attractive therapeutic target in advanced cancers. In the present study, we examined the effect of BMS-345541 as a highly selective IKKβ inhibitor on melanoma tumorigenesis. The SK-MEL-5, A375, and Hs 294T cell lines used in this study were established from human metastatic melanoma and exhibit high constitutive IKK activity and CXCL1 secretion (9). SK-MEL-5, A375, and Hs 294T cells (1 × 10⁵ cells/well) were cultured in medium with BMS-345541 at 0, 0.1, 1, and 10 μmol/L concentrations. After 3 days of culture, the cell viability was determined by hemocytometer counting (Table 1). The data from this experiment suggest that BMS-345541 treatment resulted in a concentration-dependent inhibition of melanoma cell proliferation. Cells treated for 14 hours with 10 or 100 μmol/L of BMS-345541 showed apoptotic features as revealed by TUNEL staining and nuclear condensation (Fig. 1A).

To evaluate the in vivo specificity of induction of melanoma death by BMS-345541, SK-MEL-5, A375, and Hs 294T melanoma cells (2 × 10⁶) were s.c. inoculated into nude mice and melanoma tumor formation was followed. When the tumor volume reached ∼20 mm³, BMS-345541 (0, 10, 25, and 75 mg/kg body weight) was administered to mice orally once daily for a duration of 21 days. Figure 1B shows that BMS-345541 effectively inhibits SK-MEL-5 tumor growth in a dose-dependent manner. Tumor-bearing mice treated with 75 mg/kg of BMS-345541 showed effective inhibition of growth of SK-MEL-5, A375, and Hs 294T tumors by 86 ± 2.8%, 69 ± 11% and 67 ± 3.4%, respectively, when compared with control animals treated with vehicle alone (Fig. 1C). BMS-345541 treatment was well tolerated by the mice. However, an average of 3%, 5%, 9%, or 14% loss of body weight was observed in mice receiving 0, 10, 25, or 75 mg/kg BMS-345541, respectively. Tumor-bearing mice treated daily with an oral dose of the DNA alkylating agent, temozolomide (60 mg/kg; ref. 16), showed 81% inhibition of tumor growth and a 21% reduction in body weight. Moreover, i.p. injection of tumor-bearing mice with bortezomib (1.25 mg/kg twice a week) resulted in a 71% inhibition of tumor growth and an 11% loss of body weight (Fig. 1C). These data show that BMS-345541 is reasonably well tolerated by mice even at a high dose, and is approximately as effective as high doses of temozolomide or bortezomib.

Because BMS-345541-induced growth inhibition of human melanoma grafts in nude mice was dose-dependent, we further examined the status of the drug distribution in the tissues (serum and tumor tissue) of the tumor-bearing mice. Eight hours after mice received oral doses of 10, 25, or 75 mg/kg of BMS-345541, serum and tumor tissue samples were taken from the tumor-bearing mice and the BMS-345541 concentrations in these samples were determined as previously described (12). Results in Fig. 1D show that BMS-345541 concentrations in serum reached 0.90 ± 0.25, 3.1 ± 1.1, or 8.2 ± 1.1 μmol/L in mice receiving 10, 25, or 75 mg/kg BMS-345541, respectively. Tumor tissue levels of BMS-345541 were 3.2 ± 0.97, 12.5 ± 3.28, or 28.1 ± 2.65 nmol/g tissue protein in tumor-bearing mice orally given respective doses of 10, 25, or 75 mg/kg of BMS-345541. To determine levels of BMS-345541 over time in mice, blood and tumor tissues were collected at 1.5, 3, 5, 8, and 24 hours after an oral dose of 75 mg/kg. Results show that concentrations of BMS-345541 in serum were 6.6 ± 0.31, 18 ± 1.3, 5.2 ± 0.19, 4.1 ± 0.22, 1.1 ± 0.065 μmol/L and concentrations in tumor tissues were 19 ± 2.5, 44 ± 1.5, 30 ± 1.0, 29 ± 2.5, and 11 ± 0.83 nmol/g tissue protein corresponding to time points of 1.5, 3, 5, 8, and 24 hours after drug administration (Fig. 1E).

Inhibition of IKK activity correlates with suppression of both NF-κB activation and CXCL1 production. NF-κB is a regulator of expression of several antiapoptotic, tumorigenic, and angiogenic factors (6, 17). IKK is a key mediator of the NF-κB signaling pathway (18). For example, in melanoma cell lines, IKK activity is 3- to 14-fold higher and NF-κB activity is 3.4- to 6.8-fold higher than in normal human epidermal melanocytes (9). To determine whether BMS-345541 inhibition of melanoma cell proliferation is associated with the IKK/NF-κB signaling pathway, SK-MEL-5 melanoma cells were transfected with a NF-κB-dependent luciferase reporter gene and treated with BMS-345541 for an additional 36 hours. The experiments revealed that the constitutive NF-κB activity in melanoma cells, as monitored by NF-κB-dependent luciferase activity, was dramatically reduced by BMS-345541 in a concentration-dependent manner (Fig. 2A). BMS-345541 treatment (0.1, 1, and 10 μmol/L) resulted in 36%, 75%, and 95% reduction in NF-κB-dependent luciferase activity, respectively. To determine whether inhibition of constitutive NF-κB activity was the result of down-regulation of the upstream IKK activity by BMS-345541, SK-MEL-5 melanoma cells were treated with BMS-345541, and an IKK-activity assay was done. The high activity of IKK in SK-MEL-5 melanoma cells, as noted by incorporation of radiolabeled ³²P into GST-IκBα, was reduced by BMS-345541 in a concentration-dependent fashion. As indicated in Fig. 2B, the IKK activities were reduced by 16%, 31%, and 76% in response to BMS-345541 treatment of 0.1, 1, and 10 μmol/L, respectively.

The constitutive expression of CXCL1 by melanoma cells was speculated to produce an “autocrine loop” to promote melanoma tumorigenesis because CXCL1 induces IKK activity in normal human melanocytes (9). To determine whether BMS-345541-mediated inhibition of IKK resulted in reduction of CXCL1 secretion, CXCL1 levels were determined by ELISA assay from the culture medium of SK-MEL-5 melanoma cells and from the serum of melanoma tumor–bearing mice treated with BMS-345541. The results in Fig. 2C and D show that CXCL1 expression in melanoma cell cultures or mouse sera was dramatically suppressed in a concentration-dependent manner by BMS-345541 treatment. In contrast, CXCL8 expression in SK-MEL-5 cells was relatively low (Fig. 2C) and barely detectable in the sera of melanoma tumor-bearing mice (data not shown). These data are also in agreement with the decrease of NF-κB activity in response to the BMS-345541 treatment shown in Fig. 2A.

Blocking NF-κB/p65 nuclear translocation leads to melanoma cell apoptosis. Nuclear localization and activation of the NF-κB complex is often crucial for tumorigenesis (19–21), and in particular for human melanoma (9). We therefore hypothesized that inhibition of IKK activity in melanoma cells by BMS-345541 would stabilize the p65/p50 IκB complex in the cytoplasm and prevent NF-κB nuclear translocation and activation of gene expression. To test the above hypothesis, SK-MEL-5 melanoma cells were transfected with GFP-p65 expression vector DNA and treated with BMS-345541. Technically, the GFP-p65 fusion protein shows the same abilities as wild-type p65 for nuclear translocation, NF-κB-dependent promoter binding, and functional transactivation of reporter genes (22). As expected, results in Fig. 3 shows that treatment of SK-MEL-5 cells with 10 μmol/L of BMS-345541 resulted in retention of GFP-p65 in the cytoplasm. Moreover, transfection with the NF-κB super-repressor (cyan fluorescent protein-IκBαAA) blocked yellow fluorescent protein-p65 nuclear translocation similar to that of cells treated with 10 μmol/L BMS-345541 (data not shown).

Inhibition of human melanoma tumor growth by BMS-345541 is through apoptotic pathways. In melanoma, constitutive activation of NF-κB confers tumor survival capacity and escape from apoptosis (7). We hypothesized that BMS-345541 induced apoptosis of melanoma tumor cells through suppression of NF-κB activation. To characterize the mechanism for BMS-345541 suppression of melanoma tumor growth, cultured melanoma cells were exposed to 10 μmol/L BMS-345541 for 24 hours. The cells showed a typical feature of apoptosis (Fig. 4B) along with altered mitochondria and nuclear chromatin condensation, in contrast to normal cells (Fig. 4A) under electron microscopy. A similar morphologic phenomenon was observed in vivo. Melanoma tumor sections from mice treated with BMS-345541 were stained with H&E (Fig. 4C and D) and for nuclear apoptosis using the DeadEnd Fluorometric TUNEL reagent staining (Fig. 4E and F). In contrast to the melanoma tumor from mice treated with vehicle alone (Fig. 4C and E), the melanoma tumors from mice treated with 25 mg/kg BMS-345541 (Fig. 4D and F) showed obvious nuclear condensation and DNA fragment staining with the fluorescent TUNEL reagent. Moreover, when mice were treated with 75 mg/kg of BMS-345541, TUNEL staining appeared extensively over the tumor sections (data not shown). To further quantitatively examine the drug-induced apoptosis in melanoma cells, SK-MEL-5 melanoma cells were cultured in medium with BMS-345541 treatment for 36 hours. Apoptotic cells were labeled using DeadEnd Fluorometric TUNEL reagent and analyzed by flow cytometry. Application of 1, 10, or 100 μmol/L of BMS-345541 to SK-MEL-5 cell cultures resulted in 44%, 73%, and 87% apoptotic cells, respectively. Thus, the induction of apoptosis in melanoma cells by BMS-345541 seems to be concentration-dependent (Fig. 4G). To confirm that modulation of the IKK pathway leads to apoptosis in melanoma cells, DNA from dominant-negative expression vectors for either IKKβ (K44M) or IκBα (S32 and 36A) were independently transfected into SK-MEL-5 cells. Thirty-six hours after transfection, cells were labeled with TUNEL reagent and analyzed by flow cytometry. In contrast to mock-transfected cells (vector-transfected control cells), the percentage of apoptotic melanoma cells in IKKβ (K44M) and IκBα (S32 and 36A) transfected cells was 55.5% and 96.3% greater, respectively (data not shown). To examine nuclear apoptosis, SK-MEL-5 cells were exposed to 10 μmol/L BMS-345541 for different time points. Induction of DNA fragmentation, a hallmark of apoptosis, was evaluated by staining DNA with PI followed by fluorescence-activated cell sorting analysis. Induction of DNA fragmentation by BMS-345541 was both concentration-dependent (data not shown) and time-dependent (Fig. 4H).

BMS-345541 induces mitochondria-mediated apoptosis. Mitochondria play a crucial role in the regulation of programmed cell death (23). The release of proteins from the intermembrane space of mitochondria is one of the pivotal events in the initiation of the apoptotic process (24). To investigate alteration of the apoptotic proteins in mitochondria during BMS-345541-induced apoptosis, the ratio of Bcl-2 and Bax distributed in mitochondria was analyzed in SK-MEL-5 cells using Western blot. We observed a decreased ratio of Bcl-2 to Bax with BMS-345541 treatment at 24 hours (Fig. 5A). Either decreased Bcl-2 or increased Bax in mitochondria results in the instability of the mitochondrial membrane. The change in mitochondrial membrane potential (Δψm) was determined by fluorescence-activated cell sorting analysis after the mitochondria were stained with DiOC₆. SK-MEL-5 cells showed obvious loss of mitochondrial membrane potential by 16 ± 1.5% and 87 ± 5.4%, when exposed to 10 μmol/L BMS-345541 for 24 and 48 hours, respectively (Fig. 5B). Reduction in the intact mitochondrial population occurred by 48 hours of BMS-345541 treatment (Fig. 5C). Inhibition of NF-κB by overexpression of either IKKβ (K44M) or the IκBα (S32, 36A) super-repressor in SK-MEL-5 melanoma cells resulted in a respective 32 ± 16% or 52 ± 2.8% loss of mitochondrial potential, in contrast to vector controls, confirming that NF-κB inhibition is the cause for mitochondrial dissipation. Moreover, 24 hours of treatment with 10 μmol/L BMS-345541 resulted in cytochrome c release into the cytoplasm and AIF translocation from the mitochondria to the nucleus (Fig. 5D). To rule out the direct action of BMS-345541 on mitochondria, mitochondria isolated from SK-MEL-5 were treated with 10 μmol/L BMS-345541, and the release of cytochrome c into the supernatant was analyzed by Western blot (25). In contrast to the 200 μmol/L Ca²⁺-positive control, no direct action of 10 μmol/L BMS-345541 on mitochondria was observed, based on a failure to see release of cytochrome c into the supernatant (Fig. 5E).

BMS-345541-induced apoptosis is largely caspase-independent and AIF-dependent. To examine whether BMS-345541 induction of apoptosis in melanoma cells is through a caspase-dependent pathway, SK-MEL-5 melanoma cells were treated with 10 μmol/L BMS-345541 for 36 hours. Cytosolic extracts were tested for caspase activity with the caspase-3 substrate, Ac-DEVD-pNA. The specific caspase-3 activity in response to treatment with vehicle DMSO, 10 μmol/L BMS-345541 or in combination with 50 μmol/L Z-VAD-fmk was 0.27 ± 0.075, 1.6 ± 0.092, and 0.44 ± 0.035 pmol/min/μg protein, respectively (Fig. 6A). These results indicate that BMS-345541 is able to induce caspase-3 activity in cytosolic extracts and that this induction could be blocked 87.2% with the pan-caspase inhibitor Z-VAD-fmk. The cell-permeable pan-caspase inhibitor, Z-VAD-fmk, is able to block caspase-dependent apoptosis in a number of melanoma cell lines (26). However, when SK-MEL-5 cells were treated with 50 μmol/L Z-VAD-fmk for 48 hours, it protected only 11% (P < 0.05) of cells from nuclear apoptosis induced by 10 μmol/L BMS-345541 (Fig. 6B). These data indicate that BMS-345541-induced apoptosis is not prevented by Z-VAD-fmk and therefore functions through a pathway that is largely caspase-independent.

AIF is a mitochondria-localized flavoprotein and is known to be involved in apoptosis via a caspase-independent pathway. To determine whether AIF was involved in BMS-345541-induced apoptosis, the release of AIF and cytochrome c into the cytosol was evaluated at different time points after exposure of SK-MEL-5 cells to 10 μmol/L BMS-345541. Data in Fig. 6C show an initiation of AIF mitochondrial release after only 4 hours of BMS-345541 treatment, whereas cytochrome c was not released until 14 hours after BMS-345541 treatment. These data indicate the involvement of AIF in the initiation of the apoptotic signal, which is followed by cytochrome c release. To rule out the possibility that AIF is also involved in a caspase-dependent apoptotic pathway (27), SK-MEL-5 cells were treated with 10 μmol/L BMS-345541 and/or 50 μmol/L Z-VAD-fmk, and AIF nuclear translocation was examined by Western blot and confocal microscopy. As shown in Fig. 6D and E, Z-VAD-fmk failed to influence AIF translocation to the nucleus and the subsequent nuclear condensation. Thus, BMS-345541 induced AIF nuclear translocation and nuclear apoptosis in a caspase-independent manner. Moreover, siRNA targeted to AIF resulted in >90% knockdown of AIF protein levels after three successive transfections of 100 nmol/L of siRNA and effectively reduced the percentage of apoptotic cells in BMS-345541-treated cells (10 μmol/L for 24 hours) from 24.3 ± 6.8% to 10.3 ± 3.8% (Fig. 7A). In addition, siRNA targeting of endonuclease G (EndoG) using the same approach knocked down >90% of total EndoG protein, which resulted in a 16% (P < 0.05) attenuation of BMS-345541-induced cell death (Fig. 7B) in contrast to 58% (P < 0.01) apoptotic attenuation by AIF siRNA (Fig. 7A). Importantly, the translocation of mitochondrial AIF to the nucleus is a critical step for BMS-345541-induced caspase-independent cell death in SK-MEL-5 melanoma cells. However, ∼14% of apoptosis induced by BMS-345541 has not been accounted for in our experiments.

Treatment of malignant melanoma poses a great challenge today due to its resistance to conventional chemotherapeutics and radiation (28). Our understanding of the intricate mechanisms leading to resistance has undergone tremendous conceptual advances and the strategies to intercept and disturb these resistance pathways are becoming more specific and target-selective. There are central regulatory networks that serve as a common denominator in melanoma. One such network includes the transcription factor NF-κB. NF-κB is induced by chemokines, cytokines, and anticancer agents. Activation of NF-κB is thought to confer resistance to cytotoxic therapies and escape from apoptosis. Studies in our lab have shown that the persistent activation of NF-κB results in the production of chemokines such as CXCL8 and CXCL1 that are angiogenic and promote tumorigenesis in melanoma (7, 29). We have also shown that the constitutive activation of NF-κB is mainly due to the constitutive activation of IKK, a major regulator of the NF-κB pathway (9). This critical event could endow cells with the potential for unabated proliferation, insensitivity to death-inducing signals, and enhanced metastatic potential. However, the mechanism of NF-κB regulation of apoptotic machinery is poorly understood and whether IKK can serve as a therapeutic target in melanoma is still unclear. In the present study, the compound BMS-345541, a highly selective inhibitor of IKKβ, was used to examine this possibility and explore the role of NF-κB in the network of apoptosis. The IKK inhibitory activity of BMS-345541 was confirmed through IKK activity assays as well as NF-κB luciferase reporter gene assays. The results showed a concentration-dependent inhibition of IKK activity that correlated with suppression of NF-κB activation as measured through transcription reporter assay. In addition, we showed that BMS-345541 treatment of melanoma in vitro and in vivo results in a significant suppression of transcription of a NF-κB target gene, chemokine CXCL1. Indeed, the suppression of CXCL1 expression by BMS-345541 was consistent with inhibition of the nuclear translocation of NF-κB and its accumulation in the cytosol of cancer cells, as illustrated with GFP-p65 cytosolic versus nuclear localization. As a result, BMS-345541 exhibited a concentration-dependent inhibition of melanoma cell survival in vitro and a dose-dependent suppression of melanoma tumor growth in vivo. Furthermore, melanoma cells were more sensitive to BMS-345541 than normal human epidermal melanocytes.

In recent years, there has been remarkable progress in understanding the NF-κB signaling pathway and its role in the control of programmed cell death (30, 31). NF-κB activation can maintain tumor cell viability, and inhibition of NF-κB alone is sufficient to induce apoptosis (32, 33). In the present study, the TUNEL assay revealed that inhibition of NF-κB by BMS-345541 resulted in a dramatic induction of melanoma cell apoptosis, which was in agreement with the morphologic alterations observed under electron microscopy. NF-κB is most commonly involved in oncogenesis by regulating the transcription of antiapoptotic genes (34). Therefore, chemotherapeutic resistance of melanoma cells was generally characterized by a high Bcl-2/Bax ratio and the important role of this ratio was confirmed by stably transfecting Bcl-2 into melanoma cells, which resulted in resistance to chemotherapy (35). Moreover, apoptosis induced by inhibiting NF-κB activation is associated with down-regulation of Bcl-2 via a mechanism of caspase-3 activation that mediates Bcl-2 cleavage (36). Although we failed to observe a decrease in the total Bcl-2 protein level in the melanoma cells 24 hours after exposure to BMS-345541, based on other studies showing that NF-κB regulates the transcription of antiapoptosis Bcl-2 members as well as the other proapoptosis members (37), we predict that BMS-345541 inhibition of NF-κB-mediated transcription of these various apoptotic genes contributes to the apoptotic process in these melanoma cells. Constitutive NF-κB activation and overexpression of Bcl-2 protein occurs in many types of tumors (38) and these events are associated with metastasis in certain cancers (39). Bcl-2 indirectly protects mitochondrial membranes from Bax via Bcl-2 family members only containing the BH3 domain such as the Bcl2-antagonist of cell death (Bad; ref. 40). Bad promotes apoptosis via the BH3 domain to interact with diverse antiapoptotic Bcl-2 family proteins and thus inhibits their survival functions (41). We showed here that BMS-345541 induces mitochondria-mediated apoptosis via a cascade of events, whereby inhibition of IKKβ results in reduction of the Bcl-2/Bax ratio in mitochondria and dissipation of membrane potential, leading to release of AIF and cytochrome c from the mitochondria. Although BMS-345541 targeting IKK mediates Bcl-2/Bax redistribution in mitochondria, the interactions between BMS-345541 and Bcl-2 inhibitor proteins require further investigation.

The role of NF-κB inhibition in the disruption of mitochondrial membrane potential and induction of apoptosis in melanoma cells in our studies was confirmed by transfection of melanoma cells with either mutant IKKβ or super-repressor IκBα. This method of reducing NF-κB activity created a similar dissipation in the mitochondrial membrane potential and subsequent apoptosis as BMS-345541. Moreover, BMS-345541 dissipation of mitochondrial membrane potential required intact cells because the drug failed to act on mitochondrial membrane preparations directly. These data indicate that BMS-345541 does not directly disrupt the mitochondrial membrane.

Caspase-3 activity might be induced by cytoplasmic cytochrome c, however, caspase activity accounted for only 11% of the induction of the nuclear fragmentation produced by BMS-345541, based on the limited effect of pan-caspase inhibitor (Z-VAD-fmk) on the cultured melanoma cells. These data led us to further explore the mechanism for caspase-independent apoptosis in BMS-345541-treated melanoma. Mammalian AIF is normally localized to the mitochondria and might participate in scavenging reactive oxygen species (42). Following exposure of cells to apoptotic stimuli, mitochondrial AIF translocates to the nucleus, where it binds to DNA (43) and alters the structure of chromatin resulting in large-scale DNA fragmentation, chromatin condensation and nuclear shrinkage (44, 45). This “nuclear apoptosis” cannot be prevented by caspase inhibitors in certain types of cell death (44), but can be prevented by inhibition of cysteine proteases (46). Our findings revealed that although BMS-345541 seems to elevate caspase-3 activity in human melanoma SK-MEL-5 cells, the mitochondria-nuclear translocation of AIF is not attenuated by the pan-caspase inhibitor, indicating that AIF translocation is a caspase-independent process in response to the IKK inhibitor. Moreover, it has been reported that cytosolic AIF is able to disrupt the mitochondrial potential, leading to the release of cytochrome c and activation of caspases. The late caspase activity after AIF release may also contribute to apoptosis (44). Consistent with these findings, our data showed that AIF was released from mitochondria at 4 hours, whereas cytochrome c was not released until 14 hours. Based on the release time of these apoptotic proteins from mitochondria, AIF seems to be involved in the early and late progress of apoptosis, whereas caspases might only contribute to the late stage of cell death triggered by BMS-345541. Thus, the highly selective IKK inhibitor–initiated apoptosis in melanoma cells is largely caspase-independent, AIF-dependent cell death program. Furthermore, siRNA knockdown of the AIF protein in melanoma cells effectively rescued the BMS-345541-induced nuclear apoptosis. Thus, attenuation of the BMS-345541-induced apoptosis by the AIF siRNA underscores that AIF activity is required for BMS-345541-induced programmed cell death in human melanoma cells. Moreover, a serine protease called HtrA2/Omi (47) and an apoptotic DNase called EndoG (45) have also recently been identified as potential caspase-independent apoptotic mediators. However, the relationships among AIF, EndoG, and HtrA2/Omi remain elusive.

In summary, this study highlights the importance of an aberrant IKK/NF-κB pathway in the progression and growth of melanoma and validates this pathway as a potential target for the treatment of malignant melanoma. We have elucidated the mechanism of BMS-345541-induced apoptosis in human melanoma cells. BMS-345541 targeting IKK regulates the expression of antiapoptotic proteins of Bcl-2 family in the mitochondria, and contributes to a reduction in the mitochondrial membrane potential and the release of mitochondrial AIF to initiate the cell death program in melanoma cells. In addition, caspases and EndoG seem to be involved in the late stage of apoptosis induced by BMS-345541. One of the earliest events we observed here in response to BMS-345541 was the release of AIF from the mitochondria, suggesting that AIF as well as NF-κB are key modulators of melanoma apoptosis. Altogether, our data show great promise for BMS-345541 or related compounds as potential therapeutic agents against melanoma tumors with constitutive IKK activity.

We thank Javier Piedrafita for the expression vectors of human mutant IκBα and IKKβ, Jiqing Sai and Yingchun Yu for excellent technical assistance, and Linda Horton and Elizabeth Yang for critical reading of the manuscript.