IFN-α and Bortezomib Overcome Bcl-2 and Mcl-1 Overexpression in Melanoma Cells by Stimulating the Extrinsic Pathway of Apoptosis

Bortezomib is a novel antitumor compound that specifically and reversibly inhibits the 26S proteasome. The proteasome plays a critical role in the ordered, temporal degradation of transcription factors, cyclins, and cyclin-dependent kinase inhibitors required for cell cycle progression in malignant and normal cells (1). Bortezomib has in vitro activity against a variety of tumor cell types (2–5). As a single agent, bortezomib has an acceptable toxicity profile and has shown activity in patients with advanced multiple myeloma, non–Hodgkin's lymphoma, mantle cell lymphoma, and non–small cell lung cancer (6–9). Bortezomib has also shown some clinical activity in other solid tumors, including advanced renal cell carcinoma (RCC; refs. 10, 11). Single-agent bortezomib was tested in a phase II study of malignant melanoma; however, no responses were achieved when the drug was administered at 1.5 mg/m² i.v. twice weekly for 2 weeks out of every 3 (12). Based on these studies, current research efforts in solid tumors are now focused on the use of bortezomib in combination with other proapoptotic agents (13–16).

Recombinant IFN-α has been used in the treatment of malignant melanoma and RCC and mediates the regression of metastatic disease in about 10% to 15% of patients (17, 18). Studies investigating the proapoptotic effects of IFN-α in tumor cell lines indicate that this cytokine can activate both the intrinsic and extrinsic pathways of apoptosis (19–24). Of note, IFN-α has been shown to increase the expression of cell cycle regulatory proteins (e.g., p21) in malignant cells and proteins involved in the death receptor cascade [Fas and tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL)], thereby sensitizing cells to apoptotic stimuli (25, 26). These observations suggested that IFN-α therapy might enhance the proapoptotic effects of proteasome inhibition in the setting of melanoma.

In the present study, we have shown that treatment of melanoma cells with bortezomib and IFN-α synergistically induced apoptotic cell death via FADD-dependent activation of caspase-8. Importantly, this treatment combination could effectively induce apoptosis in cells that overexpressed Bcl-2 or Mcl-1, two relevant pathways of cellular survival and resistance to apoptosis in melanoma cells.

Cell lines. The B16F1 (murine), HT144 and A375 (human) melanoma, Kasumi-3 lymphoma, and COLO-205 colorectal carcinoma cell lines were purchased from the American Type Culture Collection (ATCC). The 1259 MEL, 18105 MEL, and MEL 39 human melanoma cell lines were obtained from Dr. Soldano Ferrone (Roswell Park Cancer Institute, Buffalo, NY). The JB/MS murine melanoma cell line was obtained from Dr. Vincent Hearing (National Cancer Institute, Bethesda, MD). RCC cell lines RC-45 and RC-54 were obtained from Dr. Charles Tannenbaum (Cleveland Clinic Foundation, Cleveland, OH).

Reagents. Murine IFN-α was purchased from Access Biochemical. Recombinant human IFN-α was obtained from Schering-Plough, Inc. Recombinant human IFN-β was purchased from R&D Systems, Inc. Bortezomib (Velcade, PS-341) was obtained from Millennium Pharmaceuticals, Inc. The irreversible proteasome inhibitor MG-132 was purchased from Calbiochem, Inc. The pan-caspase inhibitor Z-VAD-FMK, caspase-8 inhibitor Z-IETD-FMK, caspase-9 inhibitor Z-LEHD-FMK, and negative control compound Z-FA-FMK were purchased from R&D Systems. The pcDNA3-Bcl-2 vector was provided by Dr. A. Letai (Dana-Farber Cancer Institute, Boston, MA). The pCR3.1-Mcl-1 vector was a gift from Dr. H. Rabinowich (University of Pittsburgh Cancer Institute, Pittsburgh, PA). The dominant-negative FADD (FADD-DN) vector that expresses a truncated form of the FADD protein was provided by Dr. A. Taghiev (University of Iowa, Iowa City, IO; ref. 27). Fas-specific small interfering RNA (siRNA) and negative control siRNA constructs were purchased from Santa Cruz Biotechnology, Inc.

Analysis of apoptosis via Annexin V/propidium iodide staining. Phosphatidylserine exposure was assessed in tumor cells by flow cytometry using APC-Annexin V and propidium iodide (PI; BD PharMingen) as previously described (28). Each analysis was performed using at least 10,000 events.

Immunoblot analysis. Immunoblots were prepared as previously described and probed with antibodies specific for FADD, Fas, FasL, Bax (Santa Cruz Biotechnology), human Bid (a BH3-only member of the Bcl-2 family), Bcl-2, Mcl-1, Bcl-xL, caspase-3, caspase-7, caspase-8, cleaved caspase-7, poly(ADP-ribose) polymerase (PARP; Cell Signaling Technology), caspase-9 (Upstate Cell Signaling Solutions), Noxa (Calbiochem), or β-actin (Sigma; ref. 29). Following incubation with the appropriate horseradish peroxidase–conjugated secondary antibody, immune complexes were detected using the enhanced chemiluminescence Plus detection kit (Amersham Biosciences) and analyzed by quantitative densitometry using Optimas 6.51 image analysis software (Media Cybernetics).

Proliferation assays. The effects of bortezomib and IFN-α on cell proliferation were tested in murine melanoma cell lines (B16F1 and JB/MS), human melanoma cell lines (HT144 and MEL 39), and human RCC cell lines (RC-45 and RC-54). Cell proliferation was measured as absorbance at 570 nm using the 3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide (MTT) Cell Proliferation Assay kit according to the manufacturer's instructions (ATCC). All assays were performed in triplicate as previously described (19).

Murine tumor models and treatments. An i.p. tumor model of malignant melanoma was used to determine whether treatment with the combination of bortezomib and IFN-α was superior in antitumor activity to either agent alone (19). Immune-competent female C57BL/6 mice (n = 11 mice per group; 6–8 wk of age; Taconic Farms, Inc.) were injected i.p. on day 0 with 10⁶ B16F1 murine melanoma cells. Beginning the next day, mice received PBS, bortezomib (1.0 mg/kg twice weekly, i.p.), murine IFN-α (2 × 10⁴ units/d, i.p.), or IFN-α and bortezomib combined and monitored for survival. Human xenograft studies were conducted similar to other published reports investigating in vivo activity of bortezomib in athymic mice (5, 16). Briefly, female BALB/c^nu/nu (athymic) mice (Taconic Farms, Inc.) were injected s.c. in the right flank with 2 × 10⁶ human A375 melanoma cells (day 0). Once tumors were palpable, mice were randomized to one of four treatment groups: (a) PBS, (b) IFN-α2b (2 × 10⁴ units/d, i.p.), (c) bortezomib (1.0 mg/kg twice weekly, s.c. peritumoral), and (d) IFN-α and bortezomib combined. Bidimensional tumor measurements were obtained twice weekly using microcalipers. The weight of all animals receiving bortezomib or IFN-α and bortezomib combined was monitored for dosing and to assess toxicity throughout the study. Histologic analysis of apoptosis in formalin-fixed paraffin-embedded (FFPE) tumor xenografts was conducted using terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) as recommended by the manufacturer (Chemicon, Inc.). CD31 immunostaining was also used to evaluate angiogenesis in FFPE tumor xenografts. All samples were analyzed in a blinded fashion by an experienced pathologist (A.R.C.), and the number of cells positive for each marker of interest was quantitated per 10 high-power fields.

Statistical analysis. ANOVA was used to determine if the two drugs (bortezomib and IFN-α) reduced cell proliferation in a synergistic manner. Regression diagnostics (i.e., residual plots and descriptive statistics) were used to check model assumptions. Median-effect analysis was used to analyze the interaction between bortezomib and IFN-α as described by Chou and Talalay (30). The combination index (CI) evaluates dose information for each drug alone and, in combination, at a specified effect level to identify whether two drugs act to induce apoptosis in a synergistic manner (CI < 1), an additive manner (CI = 1), or an antagonistic fashion (CI > 1). CalcuSyn software from Biosoft was used for this multiple drug-effect analysis. The analyses for synergistic apoptosis were conducted using data obtained from Annexin V/PI assays (% apoptotic cells) and were performed under the assumption that these drugs had independent modes of action and were therefore mutually nonexclusive. Median survival time in murine experiments was estimated by the Kaplan-Meier method and comparisons overall and between groups were made using the log-rank test. The Holm procedure is a modified Bonferroni type procedure that maintains the experiment-wise error rate and was used to make adjustments for multiple comparisons between dose groups. For human xenograft studies, tumor volume data were log transformed before statistical analyses to stabilize the variance of the measures and improve normality. Tumor volume data between the four groups on day 25 were compared via ANOVA with a Tukey-Kramer adjustment for multiple comparisons. A mixed-effects model was used to analyze the repeated volume measurements while adjusting for the correlation within mice. All analyses were performed in Statistical Analysis System (version 9.1; SAS Institute). Adjusted P values were considered significant at the 0.05 level and all tests were two sided.

Treatment of melanoma and RCC cell lines with bortezomib and IFN-α leads to enhanced apoptotic cell death. The ability of both bortezomib and IFN-α to induce tumor cell apoptosis when administered as single agents led us to examine the effects of combined therapy (3, 20). Human and murine melanoma cells were treated for 48 h with PBS, bortezomib, IFN-α, or both agents combined. Bortezomib in combination with IFN-α led to significantly increased levels of apoptosis as measured by Annexin V/PI staining compared with either agent alone in human melanoma (HT144, A375, 1259 MEL, 18105 MEL, and MEL 39), RCC (RC-45 and RC-54), colorectal carcinoma (COLO-205), lymphoma (Kasumi-3), and multiple murine melanoma cell lines (JB/MS and B16F1; Fig. 1A; Supplementary Fig. S1A; data not shown; P < 0.05). An irreversible proteasome inhibitor (MG-132) also induced synergistic apoptosis when combined with IFN-α (data not shown). Time course experiments conducted in the B16F1 cell line revealed that apoptosis began at ∼24 h and reached maximal levels between 48 and 72 h (Supplementary Fig. S1B; data not shown). The order of treatment did not seem to be critical as cells treated with IFN-α separately followed by bortezomib separately or, conversely, bortezomib followed by IFN-α displayed comparable cell death (Supplementary Fig. S1C). Consistent with our previous observations, cells treated simultaneously with both agents displayed the greatest level of apoptosis. Bortezomib plus IFN-α induced apoptosis in melanoma cell lines with both functional p53 (A375, 18105 MEL, and HT144) and in cell lines with a p53-mutant phenotype (SK-MEL-2) and did not induce apoptosis in normal peripheral blood mononuclear cells (data not shown). Treatment of human (HT144 and MEL 39) and murine (B16F1 and JB/MS) melanoma cell lines with bortezomib plus IFN-α also led to a significant reduction in cell proliferation compared with either agent alone (Fig. 1B; ANOVA interaction P < 0.001). Of note, enhanced apoptosis of human melanoma cells was observed following treatment with bortezomib and IFN-β but not in response to bortezomib and IFN-γ (Supplementary Fig. S2A–C).

Apoptosis induced by bortezomib and IFN-α is synergistic. Using the median-effect method, we next assessed if bortezomib and IFN-α induced synergistic apoptosis (31). HT144 melanoma cells were treated for 48 h with various doses of bortezomib, IFN-α, or both agents combined (five dose combinations) and then analyzed for apoptosis via Annexin V/PI staining. Dose-effect plots indicated that increasing the dosage of IFN-α had little effect on the induction of apoptosis, whereas bortezomib treatment resulted in dose-dependent increases in apoptosis (Fig. 1C). At each paired dose, the proapoptotic effect was further enhanced following combined treatment with bortezomib plus IFN-α (Fig. 1C). Dose combinations of 5 nmol/L bortezomib and 5 × 10³ units/mL IFN-α or greater resulted in CIs consistently below 1.0, suggesting a dose-dependent, synergistic effect on apoptosis of HT144 melanoma cells (Fig. 1C and D).

Combined treatment with bortezomib and IFN-α induces processing of effector caspases and PARP. Enhanced processing of the major effector caspases (caspase-3 and caspase-7) to their active forms and cleavage of PARP (a target of activated effector caspases) was observed at 48 h following treatment of the HT144 human melanoma cell line with bortezomib plus IFN-α (Fig. 2A). This pattern of caspase activation and cleavage of PARP was reproducible in multiple human melanoma cell lines (1259 MEL, A375, HT144, 18105 MEL, and MEL 39; Supplementary Fig. S3; data not shown). Enhanced cleavage of caspase-3 in the human HT144 and murine B16F1 melanoma cell line following dual treatment was confirmed by flow cytometry (Fig. 2B; data not shown).

Bortezomib and IFN-α induce caspase-dependent apoptosis. We next sought to determine the dependence of cell death on caspase activation. HT144 human melanoma cells were treated with PBS, IFN-α, bortezomib, or both agents combined in the presence of either the pan-caspase inhibitor Z-VAD-FMK (50 μmol/L) or Z-FA-FMK (negative control). The pan-caspase inhibitor Z-VAD-FMK significantly inhibited apoptotic cell death at the 48-h time point in response to bortezomib plus IFN-α (mean % of Annexin V–positive cells = 10.08 ± 3.37%; n = 3 experiments; P = 0.04) compared with similarly treated cells cultured in the presence of the negative control compound (mean % of Annexin V–positive cells = 35.84 ± 13.85%; Fig. 2C; Supplementary Fig. S4). These data suggest that the synergistic death induced by bortezomib and IFN-α is dependent on caspase activation.

Bortezomib and IFN-α can overcome elevated expression of prosurvival proteins and promote apoptosis in melanoma cells. The Bcl-2 family of mitochondrial proteins inhibits apoptosis by stabilizing the mitochondrial membrane (32) and down-regulation of these proteins has been shown to contribute to the induction of caspase-mediated apoptosis in response to bortezomib (33). Mcl-1 promotes cell survival through its antagonistic interactions with the proapoptotic protein Bax, whereas Bcl-2 and Bcl-xL have been shown to directly antagonize Bak (34). The Bcl-2 and Mcl-1 proteins are often overexpressed in melanoma and have been shown to affect the susceptibility of melanoma cells to apoptotic stimuli in other systems (35–37). Decreased levels of Bcl-2 and Mcl-1 were also observed following treatment of melanoma cell lines with bortezomib and IFN-α (Fig. 3A). In contrast, Bcl-xL and Bax expression was not affected by treatment with bortezomib or bortezomib plus IFN-α (Fig. 3A). The ability of this treatment combination to stimulate processing of effector caspases and down-regulate levels of prosurvival Bcl-2 and Mcl-1 proteins initially suggested that bortezomib and IFN-α might promote the intrinsic pathway of apoptosis by modulation of protein targets within the mitochondria. To further test the importance of the intrinsic pathway in mediating apoptosis, and to determine whether this treatment combination could induce apoptosis in the presence of high Bcl-2 levels, Bcl-2 was overexpressed in the human A375 melanoma cell line. Twenty-four hours following transient transfection with either the pcDNA3-Bcl-2 vector or pcDNA3 control vector, cells were treated for an additional 48 h with PBS, bortezomib, IFN-α, or both agents combined. Cells transfected with the unmodified pcDNA3 vector displayed 60% cell death and a characteristic decrease in Bcl-2 protein expression 48 h following treatment with bortezomib plus IFN-α (Fig. 3B). However, melanoma cells transfected with the pcDNA3-Bcl-2 construct remained sensitive to the proapoptotic effects of bortezomib plus IFN-α, as indicated by the cleavage of PARP (arrow; lanes 4 and 8). In addition, cell death as measured by trypan blue exclusion was only slightly reduced at 48 h, despite the presence of high levels of Bcl-2 at all time points in all four conditions.

Experiments using the pCR3.1-Mcl-1 vector indicated that melanoma cells overexpressing Mcl-1 also remained sensitive to the proapoptotic effects of bortezomib and IFN-α compared with cells transfected with the negative control vector (Fig. 3C). Interestingly, Bcl-2 and Mcl-1 levels did not decrease when cells were treated with bortezomib and IFN-α in the presence of a pan-caspase inhibitor (Fig. 3D). These results indicated that the observed reductions in levels of Bcl-2 and Mcl-1 were likely due to the cleavage of these proteins by activated caspases and were not due to a direct modulation of Bcl-2 or Mcl-1 by bortezomib and IFN-α. Taken together, these data suggest that bortezomib and IFN-α may be an effective means of promoting cell death in cells that have developed resistance to apoptosis due to elevated expression of prosurvival proteins.

Bortezomib and IFN-α induce apoptosis via FADD-mediated caspase-8 activation. Our previous data suggested that bortezomib and IFN-α may induce apoptosis of melanoma cells through a mechanism that is independent of the mitochondrial, intrinsic pathway of apoptosis. We therefore investigated the role of the extrinsic pathway of apoptosis in mediating bortezomib-induced and IFN-α–induced cell death. Although caspase-9 is activated via the mitochondrial pathway of apoptosis, the extrinsic pathway of apoptosis involves processing of procaspase-8 and subsequent cleavage of Bid (a proapoptotic BH3-only protein) in response to the binding of death receptors by their cognate ligands (e.g., TRAIL, TNF-α, and FasL; ref. 38). Immunoblot analysis revealed that treatment of human melanoma cell lines with bortezomib plus IFN-α led to enhanced processing of initiator caspase-8 and caspase-9 and reduced levels of native Bid, a proapoptotic BH3-only protein (Fig. 4A). The caspase-9 inhibitor Z-LEHD-FMK did not significantly inhibit apoptosis of melanoma cells in response to bortezomib and IFN-α (data not shown); however, the caspase-8 inhibitor Z-IETD-FMK significantly inhibited cell death at the 48-h time point (mean % of Annexin V–positive cells = 20 ± 5.4%; n = 3 experiments) when compared with cells cultured in the presence of a negative control inhibitor (mean % of Annexin V–positive cells = 71 ± 4.2%; P = 0.004; Fig. 4B). These data suggest that caspase-8 is activated first in response to treatment with bortezomib and IFN-α and that caspase-9 is processed in a secondary fashion by activated caspase-8. Interestingly, no detectable levels of TNF-α, TRAIL, or FasL protein were evident in cell culture supernatants from melanoma cell lines following treatment with bortezomib and IFN-α (data not shown). Although FasL and Fas proteins were detectable in whole-cell lysates from untreated melanoma cell lines, treatment with bortezomib and IFN-α did not alter the expression level of these proteins in any cell line tested (data not shown).

Importantly, immunoprecipitation of cell lysates with an anti-Fas antibody followed by immunoblot analysis with an anti-FADD antibody revealed that there was a significant increase in the association between these two proteins before the onset of cell death following a 16-h cotreatment with bortezomib and IFN-α (Fig. 4C). The association of Fas and FADD and subsequent activation of caspase-8 and induction of apoptosis can occur via both FasL-dependent and FasL-independent mechanisms (39). Importantly, bortezomib and IFN-α were significantly less effective at inducing apoptosis in melanoma cells transfected with a FADD-DN construct that lacks the death effector domain, and therefore cannot bind procaspase-8 (FADD-DN; Fig. 4D). Similarly, siRNA-mediated reduction of cellular Fas inhibited cell death mediated via bortezomib and IFN-α (Supplementary Fig. S5). We also observed a decrease in the basal expression of FADD in cells treated with bortezomib and IFN-α at the 48-h time point. This event is likely due to nonspecific cleavage of intracellular proteins by activated caspases as was observed for Bcl-2 and Mcl-1 (Fig. 3D). Together, these data suggest that bortezomib and IFN-α activate caspase-8 and induce apoptosis following the association of Fas and FADD.

Combined administration of bortezomib and IFN-α prolongs survival in a murine model of malignant melanoma. The effects of combined treatment with bortezomib and IFN-α were studied in an i.p. murine model of B16 malignant melanoma in which immune-competent mice routinely succumb to fatal disease burden after 12 to 14 days (19). Daily i.p. administration of IFN-α alone led to a significant improvement in survival of tumor-bearing mice compared with mice treated with PBS alone {PBS median survival = 15 days [95% confidence interval (95% CI), 14–15]; IFN-α median survival = 21 days (95% CI, 17–28+; P = 0.02)}. Mice receiving bortezomib as a single agent (twice weekly) also displayed a significant improvement in survival compared with PBS-treated mice [bortezomib median survival = 21 days (95% CI, 17–21; P < 0.01)]. However, the median survival of mice treated with bortezomib and IFN-α combined was significantly enhanced when compared with mice treated with PBS (P < 0.001), IFN-α (P = 0.02), or bortezomib alone [P < 0.001; bortezomib plus IFN-α median survival = 27 days (95% CI, 22–30); Fig. 5A]. No remarkable alterations in the weight or behavior of tumor-bearing mice were evident when IFN-α was administered with bortezomib (Supplementary Fig. S6).

A tumor xenograft model of human A375 melanoma in BALB/c^nu/nu mice was also used to evaluate the antitumor effects of this treatment combination in vivo. Administration of IFN-α2b or bortezomib as single agents led to a significant inhibition of tumor growth compared with control mice treated with PBS (P < 0.05). Coadministration of bortezomib and IFN-α2b to A375-bearing animals also led to a statistically significant inhibition of tumor growth compared with either agent alone (P = 0.047 versus bortezomib; P = 0.0004 versus IFN-α; Fig. 5B). Histologic analysis revealed significantly greater apoptosis in tumors derived from mice treated with bortezomib and IFN compared with tumors from PBS-treated mice as determined by TUNEL staining (P = 0.038, ANOVA; Fig. 5C and D). No significant differences in the percentage of TUNEL-positive cells were observed in tumors from combination-treated mice compared with mice treated with IFN-α or bortezomib as single agents (P > 0.05). Prior studies have shown that combined treatment with bortezomib and IFN-α2b led to inhibition of angiogenesis in athymic mice bearing human bladder cancer tumors (5). However, no difference in the microvessel density of tumors as detected by CD31 immunostaining was observed in any treatment group of the present study (Supplementary Fig. S7).

We have shown that bortezomib and IFN-α induce synergistic apoptosis in human melanoma cell lines and that cell death was dependent on caspase activation. The proapoptotic effects of this treatment combination were significantly reduced following inhibition of caspase-8 and in the presence of a dominant-negative form of the FADD protein or Fas-specific siRNA. Combined treatment with bortezomib and IFN-α also led to a significant prolongation of survival in a murine model of melanoma and significantly reduced tumor growth in a xenograft model of human melanoma in athymic mice compared with either agent alone. The present study shows that this treatment combination may have utility for the treatment of melanoma.

Our initial experiments showed that treatment of melanoma cell lines with bortezomib and IFN-α led to processing of procaspase-9 and decreased levels of Bcl-2 and Mcl-1. These data suggested that the mitochondrial pathway could be involved in promoting apoptosis in response to this treatment combination (40). Previous studies by Nencioni and colleagues (33) have eloquently shown that apoptosis induced by proteasome inhibition and TRAIL may involve both the cleavage of Bcl-2 to a fragment with putative proapoptotic activity and the elimination of antiapoptotic Mcl-1. However, Bcl-2 and Mcl-1 protein levels remained stable in the presence of a caspase inhibitor and overexpression of these prosurvival proteins or caspase-9 inhibition did not protect melanoma cells from the apoptotic program induced by bortezomib plus IFN-α. Therefore, reduced levels of these proteins in this panel of cell lines after treatment were likely the result of nonspecific cleavage by activated caspase proteins rather than an initiating event of the apoptotic program. The activation of caspase-9 and decreased expression levels of Bcl-2 and Mcl-1 in this system likely serve to amplify the process of apoptosis once it has been initiated.

We subsequently hypothesized that this treatment combination might lead to increased production of death receptor ligands (i.e., TRAIL, TNF-α, and FasL) by the melanoma cells. Previous reports have shown that either bortezomib or IFN-α can activate the extrinsic pathway of apoptosis in this manner (41). In addition, Papageorgiou and colleagues (5) have shown that combined therapy with IFN-α and bortezomib enhances apoptosis of human bladder cancer cell lines by a TRAIL-associated mechanism. Although treatment with a caspase-8 inhibitor led to a significant decrease in cell death in the present study, the endogenous expression of TRAIL, TNF-α, or FasL was not observed in either culture supernatants or lysates from melanoma cells treated with bortezomib and IFN-α (data not shown). Furthermore, previous studies have shown that TRAIL expression is not induced in A375 melanoma cells following treatment with type I IFNs (42), but A375 cells remained sensitive to the proapoptotic effects of combined therapy with bortezomib and IFN-α.

Despite the absence of a defined death receptor ligand, we observed an increased association between Fas and FADD in multiple melanoma cell lines before the onset of cell death. Furthermore, transfection of A375 melanoma cells with a FADD-DN construct or Fas-specific siRNA significantly inhibited the proapoptotic effects of bortezomib and IFN-α. Although increased binding between Fas and FasL can activate caspase-8 and initiate the apoptotic process (43), other reports have suggested that processing of procaspase-8 can be a downstream consequence of Bid cleavage (44). However, the timing of Bid cleavage in the present model does not support this conclusion. Our data suggest that caspase-8 activation and subsequent apoptosis is initiated by Fas trimerization and its association with FADD independent of FasL (39). The precise molecular events leading to an association between Fas and FADD proteins in the present model are under investigation.

The antitumor properties of bortezomib have been attributed to inhibition of IκB degradation, which leads to inactivation of nuclear factor-κB (NF-κB), a prosurvival transcription factor that is constitutively activated in many melanoma cell lines (16, 45, 46). However, our preliminary data indicated that bortezomib did not decrease the levels of NF-κB DNA binding or transcriptional activity in melanoma cell lines as measured by gel shift assay or analysis of NF-κB reporter activity in cells transfected with the 3xκB-Luc NF-κB promoter luciferase construct (data not shown). In addition, other groups have recently shown that inhibition of NF-κB activity accounts for only a portion of the antitumor activity of bortezomib (47, 48). Previous studies by Qin and colleagues (49) and Fernandez and colleagues (50) have also described a role for the BH3-only protein Noxa in mediating the proapoptotic effects of proteasome inhibition in melanoma cell lines. However, we did not detect an increase in Noxa transcript or protein in these particular melanoma cell lines following treatment with bortezomib plus IFN-α compared with bortezomib alone (Supplementary Fig. S8). These observations suggest that the proapoptotic effects of bortezomib may proceed by different mechanisms depending on the melanoma cell line under study and are augmented in a unique manner when coadministered with IFN-α.

Together, these preclinical data suggest that combined treatment with bortezomib and IFN-α represents a novel treatment strategy for inducing a direct, proapoptotic effect on tumor cells. We are currently evaluating the safety and tolerability of this treatment combination in an investigator-initiated phase I clinical trial at our institution. Elucidation of the pathways of cell death induced by this treatment combination may help to identify key molecular targets that govern the viability of melanoma cells that can be further manipulated to provide an antitumor effect.

W.E. Carson III: financial support from Millennium Pharmaceuticals, Inc. to help support a phase I clinical trial of bortezomib and IFN-α2b in melanoma. The other authors disclosed no potential conflicts of interest.

Grant support: The Harry J. Lloyd Charitable Trust; The Melanoma Research Foundation; The Valvano Foundation for Cancer Research Award (G.B. Lesinski and W.E. Carson III); NIH grants CA84402, K24-CA93670 (W.E. Carson III), P30-CA16058, and P01-CA95426 (M.A. Caligiuri); Millennium Pharmaceuticals, Inc.; and Johnson & Johnson Pharmaceutical Research & Development, L.L.C.

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.

We thank John Byrd, M.D. for critical review of this manuscript and The Ohio State University Comprehensive Cancer Center Analytical Cytometry, Histology and Biostatistics Shared Resources.