Antimyeloma Effects of a Sesquiterpene Lactone Parthenolide

Despite intensive chemotherapy, multiple myeloma (MM) remains an incurable, yet chronic, blood cancer with over 14,000 patients diagnosed and 45,000 patients treated yearly in the United States (1). The interaction of MM cells with the bone marrow microenvironment contributes to the heterogeneous treatment response and drug resistance.

The nuclear factor-κB (NF-κB) family of transcription factors consist of various dimeric complexes of the Rel protein family members, including c-Rel, p65 (Rel A), RelB, NF-κB1 (p50 and its precursor p105), and NF-κB2 (p52 and its precursor p100). The p65/p50 heteromer is the most prevalent form in cancers, including MM (2). NF-κB is central to the pathogenesis of MM both within the MM cells, as well as the biological consequence of the interaction between MM cells and the bone marrow milieu. Upon adhesion of MM cells to bone marrow stromal cells (BMSC), NF-κB activation in the BMSCs up-regulates interleukin-6 (IL-6) and vascular endothelial growth factors, which further enhance MM cell growth both directly and, in the case of vascular endothelial growth factor, also indirectly via promoting angiogenesis (3). In addition, NF-κB–mediated secretion of receptor activator of NF-κB ligand by the BMSCs activates osteoclasts, resulting in bone resorption and further tumor expansion (4). Growth factors and cytokines in the bone marrow milieu, including tumor necrosis factor-α (TNF-α), IL-1, and insulin-like growth factor I (IGF-I), activate NF-κB within the MM cells, resulting in the up-regulation of downstream target genes (5). Many of these genes have been implicated in resistance to anticancer drugs, including stress-response enzymes, such as Gadd45 and manganese superoxide dismutase, cell adhesion molecules, such as I-CAM and V-CAM, and antiapoptotic proteins, such as Bcl-2, Bcl-X, cIAP1, X-linked inhibitor of apoptosis protein (XIAP), and cellular FADD-like IL-1β–converting enzyme inhibitory protein (c-FLIP) (5). Targeting NF-κB is therefore rational, as it will affect both the tumor cells and the supporting microenvironment. Strategies that target NF-κB have primarily involved the sequestration of NF-κB in the cytoplasm by blocking degradation of the inhibitor of κB (6, 7). One such strategy involves the use of inhibitors of IKK, a protein which phosphorylates inhibitor of κB leading to its proteosomal degradation. However, early-generation IKK inhibitors suffer from a lack of specificity. Specific and potent IKK inhibitors are being developed (8). A second strategy involves the inhibition of proteins which comprise the proteosome itself. Bortezomib, a Food and Drug Administration–approved proteosome inhibitor for the treatment of MM, blocks NF-κB DNA binding, among its other properties (9). However, the proteosomal inhibition in normal cells can lead to significant side effects due to the accumulation of toxins, which would otherwise be removed by a fully functional proteosome. NF-κB targeting, using an agent with less off-target effects, which might lower toxicities while maintaining efficacy, would be appealing for clinical development.

Parthenolide, a major sesquiterpene lactone from the plant Tanacetum parthenium, inhibits NF-κB both indirectly by blocking IKK and directly by inhibiting p65 at the cysteine residue in its activation loop (10, 11). Due to its antiinflammatory properties and low toxicity, parthenolide has been used to treat migraines and rheumatoid arthritis (12). More recently, an antitumor effect of parthenolide has been observed in various cancers, including breast, lung, prostate, cholangiocarcinoma, and acute myeloid leukemia (13–15). In addition, parthenolide augments the cytotoxicity of chemotherapy, oxidative stress, and TNF-related apoptosis-inducing ligand (TRAIL) when used in combination (13, 14). Novel analogues of parthenolide, with a high oral bioavailability, have now been developed and are undergoing preclinical testing. These compounds are active on their own and yield serum parthenolide levels of over 10 μmol/L after oral gavage in animal models (16).

The potential significance of parthenolide in MM therapy is largely unexplored despite its antitumor activity against a broad range of cancers and the well known central role of NF-κB in MM. In this study, we show the cytotoxic effect of parthenolide in MM cell lines and primary samples with or without the protective effects of bone marrow cytokines and BMSC adhesion. In addition, we explored its mechanism of action and potential use in combined treatment.

Reagents. Parthenolide (Sigma) was dissolved in ethanol at a stock concentration of 10 mmol/L and stored at −20°C until use. Recombinant human IL-6 and IGF-I were purchased from PeproTech, Inc. The pan-caspase inhibitor Z-VAD-FMK was purchased from Axxora, LLC, and dexamethasone and doxorubicin were purchased from EMD Biosciences.

Cell culture. Dexamethasone-sensitive MM.1S and its dexamethasone-resistant counterpart MM.1R cell lines were kindly provided by Dr. Alan Lichtenstein (University of California at Los Angeles). NCI-H929, ARH-77, IM-9, and U266 human multiple myeloma cells were obtained from American Type Culture Collection. RPMI-8226 and the doxorubicin-resistant RPMI-8226 Dox6 cells were kindly provided by Dr. Maria Baer (Roswell Park Cancer Institute). All MM cell lines were cultured in RPMI 1640 containing 10% fetal bovine serum, 2 mmol/L l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. All experiments were done in at least three different MM cell lines and in triplicate. Figures are representative examples in one of the cell lines.

Primary multiple myeloma cells, BMSCs, and peripheral blood mononuclear cells. Primary MM cells were isolated from patients' bone marrow aspirates after Ficoll-Hypaque gradient centrifugation using CD-138–positive selection magnetic-assisted column sorting (Miltenyi) as per the manufacturer's instruction. The purity of the MM cells was confirmed to be >90% by flow cytometric analysis using an anti-CD138 antibody (Miltenyi). CD138-negative mononuclear cells were also used to establish long-term BMSCs as previously described (17). BMSCs were between three and five passages in all experiments. The study was approved by the Indiana University Scientific Review Committee and Institutional Review Board. Buffy coats from healthy volunteers (Indiana Blood Center) were used to isolate peripheral blood mononuclear cells (PBMC) using Ficoll-Hypaque gradient centrifugation.

Proliferation assays. MTS assays (Promega) were used to assess the antiproliferative effect of parthenolide, dexamethasone, and TRAIL, as per manufacturer's instructions. For the coculture experiment, cellular proliferation was done using the BrdUrd cell proliferation ELISA kit (Roche Diagnostics). MM cells were cultured in the BMSC-coated 96-well plates for 48 h, with the indicated concentrations of parthenolide or ethanol. Cells were pulsed with BrdUrd during the last 8 h of 48-h culture.

Apoptosis assays. Apoptosis and cell death were measured using the staining of the fluorescent inhibitor of caspases (FLICA), FAM-VAD-fmk, and propidium iodide as previously described (18, 19).

Lysate preparation and Western blotting. Cells were harvested, washed, and lysed in 50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 1% Nonidet P-40, 5 mmol/L EDTA, 5 mmol/L NaF, 1 mmol/L Na₃VO₄, 1 mmol/L phenylmethylsulfonyl fluoride, 5 μg/mL leupeptin, and 5 μg/mL aprotinin. Whole-cell lysates were subjected to SDS/PAGE, transferred to a polyvinylidene difluoride membrane (GE Healthcare), and immunoblotted with the following antibodies: anti-Bim, BID, and BAX (Cell Signaling Tech); anti–Bcl-X, Bik, caspase-8, XIAP, and TRAF2 (BD Biosciences); anti–Mcl-1, caspase-9, and glyceraldehyde-3-phosphate dehydrogenase (Santa Cruz Biotech); anti–Bcl-2 (Upstate USA, Inc.); anti-NOXA (ProSci, Inc.); anti–caspase-3 (Calbiochem); and anti–c-FLIP (Axxora, LLC).

Electrophoretic mobility shift assays. Cell extracts were subjected to electrophoretic mobility shift assays with NF-κB and Oct-1 probes (Promega) as described previously (20).

Caspase activity assays. Caspase-3, caspase-8, and caspase-9 activities were measured using the fluorogenic substrates Ac-DEVD-AMC, Ac-IEPD-AMC, and Ac-LEHD-AMC, respectively (Axxora, LLC; ref. 21).

IL-6 assay. Cell culture supernatants were analyzed for IL-6 using an ELISA (Quantikine; R&D Systems) with a lowest detection limit of 0.7 pg/mL. Concentrations of IL-6 (pg/mL) are presented as corrected for 10⁵ cells.

Small interfering RNA. Pooled small interfering RNAs (siRNA) against c-FLIP (short and long isoform; c-FLIP_S/L) and nontargeting scrambled siRNAs were purchased from Santa Cruz Biotech. Cells were transfected with either c-FLIP or scrambled siRNAs (500 nmol/L) using Nucleofection (Amaxa).

Statistical analyses. Values presented are the mean ± SEs from at least triplicate experiments. IC₅₀ values for the proliferation inhibition of parthenolide, dexamethasone, and TRAIL were calculated as the concentration of drugs resulting in a 50% of reduction in viable cells compared with untreated controls by MTS or BrdUrd assays. The IC₅₀ values and the combination effect of multiple drugs were determined using the Calcusyn software program for Windows (Version 1.2, 1996; Biosoft). The levels of protein expression by Western blotting were quantified by densitometry. Comparisons between two groups were done using Student's t test; three or more groups were compared by ANOVA. P values of <0.05 were considered significant.

Parthenolide selectively inhibits the growth of MM cell lines and patient cells compared with BMSCs and normal PBMCs. After a 72-hour treatment of MM cells with parthenolide (0-10 μmol/L), we observed a dose-dependent inhibition of proliferation in all cell lines, including the dexamethasone-resistant MM.1R cells and the doxorubicin-resistant RPMI-8226/Dox6 cells (Fig. 1A). In addition, parthenolide inhibited the proliferation of the primary MM cells tested (Fig. 1B). The IC₅₀s for all cell lines and primary cells ranged from 1 to 3 μmol/L. In contrast, parthenolide at these doses had a marginal effect on the growth of the PBMCs and BMSCs, suggesting that the parthenolide cytotoxicity is selective to MM cells (Fig. 1B).

Parthenolide attenuates the protective effect of the bone marrow microenvironment on MM cells. We further explored the cytotoxicity of parthenolide on MM cells with and without adhesion to BMSCs, using a BrdUrd incorporation assay. Parallel experiments using MM cells alone or BMSCs alone were done as controls. BMSCs showed a low BrdUrd uptake, suggesting a low proliferation rate (Fig. 1C). Parthenolide marginally decreased the BrdUrd uptake of BMSCs at the doses that significantly inhibited the BrdUrd uptake of MM cells. The coculture with BMSCs strongly increased the BrdUrd uptake of NCI H929 cells (2.4-fold, P = 0.000), whereas a more modest increase was observed with the RPMI 8226 cells (1.2-fold, P = 0.04). This observation suggests that the interaction of MM cells and BMSCs promotes MM cell proliferation. Furthermore, the coculture with BMSCs partially protected the MM cells from parthenolide cytotoxicity as higher concentrations of parthenolide were required to cause 50% inhibition of proliferation in the presence of BMSCs (IC₅₀ = 3.0 and 7.5 for NCI H929 and RPMI 8226, P = 0.003; Fig. 1C).

Parthenolide overcomes the protective effect of IL-6 and IGF-I. As the adhesion of MM cells to BMSCs increases IL-6 and IGF-I secretion from the BMSCs (3), we explored whether these cytokines may protect MM cells from parthenolide cytotoxicity. Both cytokines promoted MM cell growth; however, neither one protected MM cells against parthenolide cytotoxicity. Specifically, the IC₅₀s of parthenolide-mediated proliferation inhibition remained unchanged in the presence of these cytokines: IC₅₀ MM.1S, 1.66 μmol/L; IC₅₀ MM.1S + IL-6, 1.68 μmol/L; IC₅₀ MM.1S + IGF-I, 1.81 μmol/L (P = 0.11 and 0.11, respectively; Fig. 2B). This indicates that, in addition to IL-6 and IGF-I, other factors, upon adhesion to BMSCs, contribute toward the protection of MM cells against parthenolide.

Parthenolide inhibits NF-κB DNA binding activity and prevents TNF-α–induced NF-κB activation. Electrophoretic mobility shift assays were done to explore the effect of parthenolide on the DNA binding capacity of NF-κB. TNF-α treatment represented a positive control. TNF-α strongly enhanced the NF-κB DNA binding, as expected. Parthenolide inhibited the NF-κB–DNA binding and further reversed the effect of TNF-α–induced NF-κB activation (Fig. 2A).

Parthenolide inhibits IL-6 secretion in a bone marrow stroma and MM cell coculture system. Adhesion of MM cells to BMSCs up-regulates NF-κB–dependent IL-6 expression in BMSCs (22). We determined the effects of parthenolide on IL-6 secretion in the BMSCs cocultured with MM cells. To differentiate the effect of parthenolide on IL-6 secretion from its cytotoxicity, we used sublethal doses of parthenolide (0.5 μmol/L for NCI H929 cells and 1 μmol/L for RPMI 8226 cells) and a short treatment time (12 h). At these parthenolide doses and treatment duration, parthenolide did not inhibit MM cell proliferation measured by the MTS assay (data not shown). The IL-6 levels in the supernatant of NCI-H929 were nearly undetectable, compared with that of the BMSCs (Fig. 2B). IL-6 levels increased when BMSCs were cultured with MM cells (means ± SDs, 279.3 ± 4.0 versus 459.3 ± 2.5 ng/mL, respectively; P = 0.00004). Parthenolide inhibited IL-6 secretion by the BMSCs alone (means ± SDs, 279.3 ± 4.0 versus 240.0 ± 5.0 ng/mL; P = 0.00022) and in the coculture system (means ± SDs, 459.3 ± 2.5 versus 363.3 ± 9.9 ng/mL; P = 0.0019). A similar finding was also observed with RPMI 8226 cells (data not shown). This suggests that the decrease of IL-6 levels is due to the inhibition of its secretion rather than a function of cell death.

Parthenolide triggers a dose-dependent, caspase-mediated apoptosis of MM cells with a minimal effect on cell cycle progression. As proliferation inhibition can be a function of cell death or cell cycle interference, we next did apoptosis assays and cell cycle analyses of parthenolide-treated MM cells. Figure 3A represents the apoptosis induction in NCI-H929 cells. Using flow cytometric analysis of FLICA and propidium costaining, we observed a dose-dependent rapid accumulation of FLICA positive cells, which represent cells with activated caspases, occurring in <12 hours after parthenolide treatment. At the higher parthenolide doses, an accumulation of cells dually labeled with FLICA and propidium was observed, suggesting late apoptosis. Using propidium staining of NCI-H929 and RPMI-8226 cells, we observed no definitive effect of parthenolide on cell cycle progression. Specifically, the percentage of cells in G₀-G₁, G₂-M, and S phases did not differ after parthenolide treatment compared with ethanol-treated controls (data not shown). However, we observed a rapid increase in the sub–G₀-G₁ fraction representing apoptotic cells, supporting the results from the FLICA-PI assay. In addition, pretreatment of the MM cells with the pan caspase inhibitor ZVAD-fmk (25 μmol/L) partially inhibited parthenolide-induced apoptosis, indicating that the proliferation inhibition by parthenolide in MM is, at least in part, due to caspase-dependent apoptosis induction, rather than cell cycle interference (Fig. 3B).

Parthenolide rapidly activates both the extrinsic and intrinsic apoptotic pathways. Two major pathways of apoptosis include the extrinsic pathway which is characterized by surface death receptors binding to their ligands followed by the activation and cleavage of procaspase-8 and the intrinsic pathway which is characterized by loss of mitochondrial integrity followed by the release of cytochrome c and the activation of caspase-9 (23). The rapid caspase activation from the FLICA assay led us to explore the effect of parthenolide on the different caspases using Western blot analysis. We observed a rapid accumulation of cleaved products of caspase-8 and caspase-3, whereas the accumulation of cleaved caspase-9 products was less impressive (Fig. 3C). To further characterize the kinetics and objectively quantify the extent of caspase activation, we did colorimetric assays for the activity of caspase-8, caspase-3, and caspase-9. Corresponding to our Western blot data, the caspase-8 and caspase-3 activities increased early at 2 hours and peaked at 31.7-fold and 26.5-fold, respectively, over the controls 8 hours after parthenolide treatment. Meanwhile, caspase-9 activity only peaked at 6-fold of control (Fig. 3D).

Parthenolide down-regulates the levels of c-FLIP and TRAF2. The dramatic parthenolide-mediated caspase-8 activation and the known ability of parthenolide to block NF-κB binding led us to focus on the NF-κB–regulated proteins that may modulate caspase-8 or death receptor–mediated apoptosis.

c-FLIP modulates caspase-8 activity by competing with caspase-8 for binding at the death enhancing domain, where procaspase-8 is cleaved (24). c-FLIP is overexpressed in various cancers, including MM (25). c-FLIP has two alternatively spliced isoforms; short (c-FLIP_S) and long (c-FLIP_L). In NCI-H929 cells, c-FLIP_L is more abundant than c-FLIP_S. Both isoforms gradually decreased at 2 hours and completely disappeared at 8 hours after parthenolide treatment (Fig. 4A).

The inhibitors of apoptosis family genes cIAP1 and cIAP2 and the TNF receptor–associated factor family genes TRAF1 and TRAF2 are NF-κB–regulated proteins that, when overexpressed concurrently, can block caspase-8 activation and TNF-mediated apoptosis (26). We observed that only TRAF2, but not cIAP1, cIAP2, or TRAF1, levels changed significantly. Namely, TRAF2 levels decreased ∼8 hours after parthenolide treatment (Fig. 4A). The promptness of c-FLIP down-regulation, even before the decrease in TRAF2 levels, raises the possibility that it might be an initial event leading to procaspase-8 cleavage.

Direct targeting of c-FLIP using siRNA inhibited MM cell growth and decreased the threshold of parthenolide-mediated growth inhibition. We next explored the potential significance of c-FLIP down-regulation in parthenolide cytotoxicity. c-FLIP siRNA rapidly decreased the c-FLIP protein levels in U266 MM cells (Fig. 4B) and was sufficient to inhibit the proliferation of U266 cells by ∼30%. In addition, c-FLIP siRNA decreased the IC₅₀ for proliferation inhibition in U266 cells (IC₅₀ c-FLIP siRNA of 1.83 μmol/L versus IC₅₀ control siRNA of 3.06 μmol/L; combination indices, 0.76-0.98; Fig. 5C). These data confirm the importance of c-FLIP in MM cell survival and also support the hypothesis that c-FLIP down-regulation is potentially an important mechanism for parthenolide cytotoxicity.

Parthenolide induced caspase-dependent cleavage of MCL-1 and XIAP. We next examined other antiapoptotic proteins which are NF-κB regulated, as they may play roles in parthenolide toxicity. Among the BCL-2 antiapoptotic family members, the levels of Bcl-2 and Bcl-X(L) were unaffected by parthenolide (data not shown), but we observed a time-dependent cleavage of MCL-1, as evidenced by decreasing full-length MCL-1 levels and a gradual accumulation of cleaved MCL-1 (Fig. 5A). MCL-1 is up-regulated by IL-6 potentially via NF-κB regulation; hence, we explored the effect of parthenolide on MCL-1 in the presence of IL-6. IL-6 caused an expected increase in the MCL-1 levels; however, parthenolide could effectively cleave MCL-1, suggesting that this may be a mechanism that allows parthenolide to overcome the antiapoptotic effect of IL-6 (Fig. 5B).

In addition to MCL-1 cleavage, we also noted a time-dependent cleavage of the XIAP (Fig. 5A). XIAP is a potent inhibitor of downstream effector caspase-3, caspase-7, and caspase-9. Furthermore, XIAP overexpression has been related to resistance to chemotherapy and poor prognosis in MM (27). Because both MCL-1 and XIAP have multiple caspase cleavage sites, we explored whether parthenolide-induced caspase activation leads to the cleavage of MCL-1 and XIAP. The pan-caspase inhibitor Z-VAD-fmk completely blocked MCL-1 cleavage and partially blocked XIAP cleavage, indicating caspase dependency (Fig. 5A).

As mitochondrial stability depends on the balance between the proapoptotic and antiapoptotic members of the Bcl-2 family, we also examined the effect of parthenolide on the levels of these proteins. Although we did not observe a change in the levels of the multidomain proapoptotic Bcl-2 proteins (Bax, Bak, and Bok) and the BH3-only proteins (Bad, Bim, and Noxa), we observed a time-dependent cleavage of BID (Fig. 6A). Interestingly, we did not observe an effect of Z-VAD-fmk on BID cleavage, suggesting a possible caspase-independent mechanism.

Parthenolide augments the growth inhibition of MM cells treated with dexamethasone or TRAIL. The preferential activation of the extrinsic apoptotic pathway by parthenolide differentiates it from other common treatments for MM including conventional chemotherapy, radiation, and glucocorticosteroids, which target the intrinsic apoptotic pathway (28). Because an additive effect, or synergy, may be expected when both apoptotic pathways are activated concurrently (29), we investigated the cytotoxicity of parthenolide combined with dexamethasone. The combination yielded an additive effect in MM cell lines. Figure 6A shows a representative example of the additive cytotoxicity of the combination in MM.1S cells (combination indices, 0.75-1.04). Because c-FLIP overexpression has been related to TRAIL resistance and c-FLIP down-regulation can sensitize cancer cells to TRAIL-induced apoptosis (30, 31), we next tested the cytotoxicity of the combination of parthenolide and TRAIL. The combination yielded an additive effect in NCI H929, MM1.S, and U266, and a synergistic effect in RPMI 8226 cells. Figure 6B shows a representative example of the synergistic cytotoxicity of the parthenolide and TRAIL combination in RPMI 8226 cells (combination indices, 0.64-0.75).

In this study, we report the ability of parthenolide to inhibit NF-κB, induce apoptosis in MM cell lines and primary cells, and overcome the protective effect of the bone marrow microenvironment.

Our results clearly show the selective cytotoxicity of parthenolide against MM cells compared with normal BMSCs and PBMCs. These data are in line with previous findings in an acute myeloid leukemia model wherein leukemic stem cells are more susceptible to parthenolide compared with normal hematopoietic progenitor cells (14). Parthenolide overcomes the proliferative effects of IL-6 and IGF-I. However, adhesion to BMSCs partially protects the MM cells from parthenolide, suggesting that the protective effect may be related to other factors, such as direct cell-cell contact or a combination of various cytokines. We also showed that parthenolide not only targets NF-κB signaling in the MM cells but also deprives MM cells of the NF-κB–mediated IL-6 secretion by BMSCs.

The biological effects of parthenolide in MM cells are multifaceted and distinct from other cancers. Firstly, in prostate and breast cancer models, parthenolide induces a sustained c-Jun-NH₂ kinase activation resulting in cell cycle arrest. In MM cells, caspase activation is very rapid, followed by apoptosis, without significant cell cycle perturbation. In addition, we did not observe a change in c-Jun-NH₂ kinase levels or its phosphorylation status (data not shown). In cholangiocarcinoma and leukemia cells, the cytotoxicity of parthenolide depends on the status of p53 (32, 33). This is unlikely to be important in MM cells as parthenolide was equally cytotoxic in RPMI-8226 and ARH77cells with mutant p53 compared with MM cell lines with wild-type p53, and we did not observe a change in p53 levels or phsophorylation status after parthenolide treatment.

We observed that parthenolide cytotoxicity is likely partly caspase-dependent, as the pan-caspase inhibitor Z-VAD-fmk could partially protect MM cells from parthenolide cytotoxicity. Previous work in other settings that supports our observation include the following: (a) parthenolide simultaneously induces apoptosis and necrosis in jurkat and HL60 leukemia cells (34); (b) in some cancer models, the accumulation of reactive oxygen species after parthenolide treatment leads to the loss of mitochondrial membrane potential and increases the permeability of the lysosomal membranes, resulting in caspase-independent autophagic cell death (35).

Parthenolide induces both the extrinsic and intrinsic apoptotic pathways and activates caspase-8, caspase-3, and caspase-9. However, the levels of caspase-8 and caspase-3 activation are more pronounced than caspase-9. Exploring the NF-κB–regulated genes with known functions in modulating caspase-8, we found that c-FLIP is rapidly down-regulated. c-FLIP is overexpressed in MM, and direct silencing of c-FLIP using siRNA has been shown to increase caspase-8 recruitment to the death enhancing domain and induce apoptosis in some cancer models (24). We hence hypothesize that parthenolide-mediated c-FLIP down-regulation may be an initial step leading to parthenolide cytotoxicity. We found that c-FLIP siRNA inhibits the proliferation of MM cells, confirming the importance of c-FLIP in MM cell survival. c-FLIP siRNA also decreases the parthenolide dose needed for cell kill, suggesting that c-FLIP down-regulation is important for parthenolide cytotoxicity. c-FLIP overexpression has been shown to confer TRAIL resistance (30). It is therefore plausible that the parthenolide-mediated sensitization to TRAIL in our study is secondary to c-FLIP down-regulation. TRAIL is the leading death receptor ligand in clinical development with selective activity against cancer cells. However, its success is limited due to clinical resistance. Our data suggest that parthenolide may help to overcome the TRAIL resistance mediated by c-FLIP overexpression. c-FLIP can also regulate the activation of NF-κB and provide a positive feedback loop (36); hence, another potential benefit of c-FLIP down-regulation in our study may include the depletion of a potential activator of NF-κB.

Parthenolide activates both the extrinsic and intrinsic apoptotic pathways, whereas other anti-MM treatments, including conventional chemotherapy, radiation, and glucocorticoid, primarily activate the intrinsic apoptotic pathway. We hypothesize that simultaneous activation of both apoptotic pathways may enhance cytotoxicity. Our finding of an additive cytotoxicity with the parthenolide and dexamethasone combination supports this concept. The strong extrinsic pathway activation of parthenolide may also have an additional benefit in that it could bypass the common mechanisms of resistance to the intrinsic pathway, such as Bcl-2 overexpression. Our evidence that parthenolide is equally toxic to MM cell lines with high basal levels of Bcl-2, including U266 and RPMI 8226 cells, supports this hypothesis (37).

We observed cleavage, but not transcriptional, down-regulation of MCL-1 and XIAP, although the expression of these genes can be regulated by NF-κB in some settings (data not shown; ref. 38). The expression of these proteins in MM could possibly be regulated by other transcription factors. Examples may include the control of MCL-1 expression by the phosphoinositide 3-kinase/Akt-1 or signal transducers and activators of transcription 3 pathways in macrophages of patients with rheumatoid arthritis (39) and the control of XIAP by mitogen-activated protein kinase in small cell lung cancer (40). XIAP overexpression has been shown to sufficiently confer TRAIL resistance (38, 41); hence, its cleavage may provide another explanation for the augmented cytotoxicity of the TRAIL and parthenolide combination in our study. MCL-1 is important for MM cell survival because down-regulation of MCL-1 using antisense oligonucleotides causes apoptosis, whereas that of Bcl-2 and Bcl-X(L) do not (42–44). Because MCL-1 binds a broad array of proapoptotic BCL-2 members, down-regulation of MCL-1 may free up the proapoptotic BCL-2 members for their translocation to the mitochondria causing apoptosis (45). Cleaved MCL-1 products also have a proapoptotic function (46), perhaps further contributing to the cytotoxicity of parthenolide.

In addition to MCL-1 and XIAP cleavage, parthenolide also induces caspase-independent BID cleavage. Other proteins that can cleave BID include cathepsins, calpains, and granzyme B. One possible mechanism for parthenolide-induced BID cleavage may be by the accumulation of ROS (14), which can destabilize the lysosomal membranes and cause the release of hydrolytic enzymes, including cathepsins (47). MM cells overcome BID-mediated apoptosis by overexpressing casein kinase II, which phosphorylates and protects BID from caspase cleavage (48). The caspase-independent BID cleavage by parthenolide, hence, may have a possible advantage by overcoming casein kinase II overexpression.

Collectively, parthenolide cytotoxicity may depend on the relative contributions of c-FLIP, XIAP, MCL-1, and BID down-regulation, but the degree to which each factor is required remains unclear and warrants further investigation. Novel analogues of parthenolide with a high-oral bioavailability have now been developed and are undergoing preclinical testing (16). The doses of parthenolide required for the cytotoxic effect in our study are within the achievable range observed by using the parthenolide analogue LC-1 in a mouse model (49).

In this study, we show that parthenolide is effective against MM cells in the context of the bone marrow microenvironment and that its mechanisms of action are both caspase-dependent and independent. In combined therapy, parthenolide is additive and synergistic with dexamethasone and TRAIL, respectively. Our findings provide a rationale for the clinical development of parthenolide.