Two death pathways induced by sorafenib in myeloma cells: Puma-mediated apoptosis and necroptosis
Abstract
Purpose Sorafenib is a multikinase inhibitor that targets the MAPK pathway and is currently used for the treatment of hepatocellular and renal carcinoma. Recently, it has been shown that sorafenib is also cytotoxic to multiple myeloma (MM) cells. Here, we have further analyzed the mechanism of sorafenib-induced death in MM cells.
Methods Cell death induced by sorafenib in MM cell lines and in plasma cells from MM patients was evaluated by analysis of gene expression by RT-MLPA and quanti- tative PCR, protein levels and functionality by Western blot and flow cytometry and gene silencing with siRNA.
Results Cell death was characterized by phosphatidylser- ine exposure, DWm loss, cytochrome c release and caspase activation, hallmarks of apoptosis. DL50 at 24 h ranged from 6 to 10 lM. Ex vivo treatment with 20 lM sorafenib induced apoptosis in around 80 % myeloma cells from six multiple myeloma patients. Sorafenib induced caspase- dependent degradation of Bcl-xL and Mcl-1 proteins, destabilizing the mitochondria and speeding up the devel- opment of apoptosis. Sorafenib treatment increased levels of Puma at mRNA and protein level and gene silencing with siRNA confirmed a relevant role for Puma in the induction of apoptosis. Co-treatment with the pan-caspase inhibitor Z-VAD-fmk prevented cell death to a variable degree depending on the cell line. In RPMI 8226 cells, Z-VAD-fmk prevented most of sorafenib-induced death. However, death in MM.1S was only prevented by co-incubation with both Z-VAD-fmk and the RIP1K inhibitor necrostatin-1, indi- cating that under conditions of inefficient caspase activa- tion, sorafenib induces death by necroptosis.
Conclusion Our results demonstrate a key role for Puma in the triggering of sorafenib-induced apoptosis and that this drug can also induce death by necroptosis in multiple myeloma cells.
Keywords : Sorafenib · Puma · Mcl-1 · Bcl-xL · Apoptosis · Necroptosis
Introduction
Multiple myeloma (MM) is a plasma cell malignancy that accounts for around 10 % of all hematological cancers with more than 22,000 new cases expected in 2013 in the USA [1]. It remains generally incurable, with most patients experiencing relapse after first-line treatment. Median survival after treatment has increased in last years to more than 5 years [2] owing to the introduction of new drugs, such as the proteasome inhibitor bortezomib. However, MM remains incurable in most patients and therefore, new drugs are still needed to improve the anti-myeloma arsenal. A potentially useful new drug to treat myeloma could be the multikinase inhibitor sorafenib, currently used for the treatment of hepatocellular carcinoma and advanced renal carcinoma [3]. Although initially described as an inhibitor of the Ser/Thr-kinase Raf1, further studies revealed that sorafenib can also inhibit B-Raf, p38 and the Tyr kinases c-kit, Flt3, Ret, VEGFR and PDGFR, which are involved in cell proliferation, differentiation, survival and angiogenesis [4]. The interaction between MM cells and stromal cells in the bone marrow microenvironment provides survival signals mediated by several cytokines such as VEGF [5], IL-6 [6] and IGF-1 [7] that activate the Ras/Raf/MEK/Erk pathway, which has been shown to be essential for the survival of myeloma cells [8]. Stroma protects myeloma cells from cell death induced by antitumor agents and inhibition of the MAPK pathway with sorafenib could become a therapeutic strategy to fight chemotherapy resistance [9]. In this sense, sorafenib has been shown to be cytotoxic for human myeloma cells in vitro and in in vivo mouse xenografts [10, 11] and to synergize with bortezo- mib in human myeloma cell lines [10]. Also, it is known that sorafenib induces apoptosis in MM cell lines [5], although the key determinants for apoptosis induction as well as the possible existence of alternative death pathways have not been fully determined. Available data indicate that sorafenib-induced apoptosis in human leukemia cell lines is regulated by the Bcl-two family of proteins, that includes antiapoptotic (Bcl-2, Bcl-xL or Mcl-1) and proa- poptotic members (Bax, Bak, Bim, Puma, Noxa, Bid and Bad among others). Proteasome-mediated degradation of the antiapoptotic protein Mcl-1 [12] or inhibition of its mRNA translation [13] has been proposed to be key event for sorafenib-induced apoptosis. Recently, it has been reported that sorafenib can induce apoptosis in AML cells through Bim up-regulation [14]. Puma (p53-upregulated modulator of apoptosis), another proapoptotic protein of the same family, has been implicated in sorafenib-induced apoptosis in colorectal and hepatoma cell lines [15, 16], but there are not data about its role in hematological malig- nancies. We have studied here in detail the mechanism of sorafenib-induced death in MM cells. First, analysis of the
cell death mechanism revealed that Puma is upregulated by sorafenib and has a role in the cytotoxic effect of this compound on myeloma cells. Degradation of Mcl-1 and Bcl-xL also participates in the regulation of apoptosis in this model, but the latter seems to be more important for cell fate upon sorafenib treatment. In addition, we also show that sorafenib can induce necroptosis in MM cells when caspases are inhibited.
Materials and methods
Materials
Sorafenib was kindly provided by Bayer (Germany). Stock solutions were made at 10 mM in DMSO. Final DMSO concentration in cultures was lower than 0.01 %. Caspase inhibitors Z-VAD-fmk, Z-IETD-fmk and Z-DEVD-fmk were from Bachem (Germany) and Z-LEHD-fmk from BD Biosciences (Spain). Necrostatin-1 was from Sigma (Spain). Tetramethylrhodamine ethyl ester (TMRE) and 3,30-dihexyloxacarbocyanine iodide (DiOC6(3)) were from Invitrogen (Spain).
Cell lines and patient samples
MM.1S, U266, H929, RPMI 8226 and OPM-2 multiple myeloma were routinely cultured at 37 °C in RPMI 1640 medium supplemented with 10 % fetal calf serum (FCS), L-glutamine and penicillin/streptomycin. Myeloma cell lines overexpressing Mcl-1 (MM.1S/Mcl-1 and 8226/Mcl- 1) or Bcl-xL (MM.1S/Bcl-xL and 8226/Bcl-xL) were gen- erated in our laboratory from MM.1S and RPMI 8226 cells, as previously described [17–19]. Bone marrow mononu- clear cells (BMMC) were obtained from bone marrow aspirates of seven patients from the Hospital Cl´ınico Uni- versitario (Zaragoza), either at diagnosis or after relapse from therapy. BMMC were purified by Ficoll-Paque den- sity centrifugation and cultured in RPMI 1640 medium supplemented with 15 % fetal bovine serum (FBS), 100 IU/ml IL-6, 100 U/ml penicillin and 100 lg/ml streptomycin. In some patients, CD138+ cells were purified from BMMC by positive selection with CD138 Micro- Beads (Miltenyi Biotec, Spain). Viability of the MM cell- enriched fractions was [98 %.
Cytotoxicity assays
Cells (3–5 × 105 cells/ml) were treated in flat-bottom, 24-well (1 ml/well) with sorafenib (1–20 lM) in complete medium for the indicated times. For death inhibition assays, cells were pre-incubated for 1 h with 100 lM of the general caspase inhibitor Z-VAD-fmk, or with the selective inhibitors Z-IETD-fmk (caspase-8), Z-DEVD-fmk (cas- pase-3) or Z-LEHD-fmk (caspase-9) prior to the addition of sorafenib. In some experiments, the RIPK1 inhibitor ne- crostatin-1 (30 lM) was also used. Sorafenib cytotoxicity was also determined in BMMC isolated from myeloma patients. BMMC were incubated at 37 °C with sorafenib (5–20 lM) in RPMI 1640 complete medium, in 6-well plates for 24–48 h. To discriminate between apoptosis in myelomatous plasma cells and non-myeloma cells in BMMC samples, we used a combination of annexin
V-FITC and anti-CD38 and anti-CD45 mAbs. Cells were incubated for 15 min at room temperature in the dark with 0.5 ll annexin V-FITC (Immunostep) and a mixture of anti-CD38-PC5 and anti-CD45-PE (BD Biosciences). Plasma cells were identified as CD38++/CD45- cells and lymphocytes as CD45++/CD38- cells. The percentage of annexin V-FITC positivity in each sub-population was determined. In some cases, purified MM plasma cells (3 × 105 cells/ml) were treated with sorafenib for 24 h and cell lysates analyzed by Western blot.
RT-MLPA and real-time PCR
Total RNA from 5 × 106 cells was isolated with the RNeasy Mini Kit and QiaShredder columns (Qiagen). Cellular mRNA was analyzed by reverse transcriptase multiplex ligation-dependent probe amplification (RT- MLPA), using SALSA MLPA KIT R011 Apoptosis mRNA from MRC-Holland (Amsterdam, The Netherlands), which contains probes for mRNAs of 39 apoptosis-related genes, as described previously [19]. The mRNA levels of each gene were expressed as a normalized ratio of the peak area divided by the peak area of a housekeeping control gene (b-microglobulin). For real-time PCR, total RNA (2 lg) was retro-transcribed using random primers and Super- script II kit (Invitrogen). mRNA expression was analyzed in triplicate by quantitative reverse transcription–PCR on the 7500 real-time PCR sequence detection system using the TaqMan One Step PCR Master Mix kit and predesigned assay-on-demand primers and probes (Applied Biosystems). ®-glucuronidase (GUS) was used as an endogenous control.
RNA interference assays
Cells were electroporated with siRNA oligonucleotides using a Nucleofector system (Lonza) as described previ- ously [19]. Cells (3 × 106) were electroporated in 100 ll solution V containing 90 pmol ds–siRNAs and using the O23 (MM.1S) or G-016 (RPMI 8226) programs. Trans- fected cells were cultured for 24 h prior to sorafenib addition and further incubated for another 24 h with the drug. The sense strand sequences used, with two 30-dTdT overhangs, were as follows: Bim-siRNA: 50-GAC- CGAGAAGGUAGACAAUUG-30; Bcl-xL-siRNA: ON-TARGETplus SMARTpool BCL2L1; Puma-siRNA: 50- GCCTGTAAGATACTGTATA-30; Lamin A/C-siRNA: 50-CTGGACTTCCAGAAGAACA-30 (negative control). Oligonucleotides for Bim, Bcl-x and Lamin A/C were synthesized by Thermo-Dharmacon and those of Puma were from Qiagen.
Statistical methods
Statistical analysis was performed by analysis of variance (ANOVA) or t test using the GraphPad Prism 4.0 software (GrandPath Software, Inc., San Diego, CA, USA).
Results
Characteristics of cell death induced by sorafenib
As previously reported [11], sorafenib induced apoptosis in MM cell lines (Fig. 1a, b) and in plasma cells (Fig. 1c) from MM patients, in a dose- and time-dependent way. All myeloma cell lines were sensitive to sorafenib and DL50 ranged between 6 lM (MM.1S, RPMI 8226 and NCI- H929) and 10 lM (OPM-2) approximately. Cells from patients were clearly sensitive to doses of 15 and 20 lM. (Fig. 1c). Apoptosis was characterized by nuclear con- densation (not shown), DWm loss (Fig. 1c) and PS expo- sure (Fig. 1b, c). Sorafenib treatment of MM cells induced both Bax and Bak proapoptotic conformational changes, cytochrome c release and caspase-3 activation (Fig. 3a). However, besides canonical apoptosis, sorafenib induced another death pathway since co-treatment with the general caspase inhibitor Z-VAD-fmk prevented death at a dif- ferent degree depending on the cell line (Fig. 2a). Death was almost completely blocked in RPMI 8226 cells (71.5 ± 1.5 vs. 25.5 ± 11.5, p \ 0.005), partially in U266 (50.0 ± 17.0 vs. 22.6 ± 8.4, p \ 0.05), H929 (32.4 ± 2.6 vs. 20.8 ± 2.3, p \ 0.01) and OPM-2 (52.6 ± 2.6 vs. 29.0 ± 2.9, p \ 0.005) and only slightly in MM.1S cells (44.6 ± 5.9 vs. 38.5 ± 8.1). Co-treatment of cells with sorafenib and more selective caspase inhibitors again prevented in part death in U266 and RPMI 8226 cells but these peptide inhibitors were ineffective to prevent death in MM.1S cells (Fig. 2b). Indeed, the selective caspase-8 inhibitor Z-IETD-fmk even potentiated death induced by sorafenib in these cells (Fig. 2b). To gain insight into the differences in caspase dependence for cell death between RPMI 8226 (highly dependent) and MM.1S (barely dependent) cells, we treated these myeloma cells with sorafenib in the presence or absence of Z-VAD-fmk and/ or necrostatin-1, an inhibitor of RIP1 kinase (RIP1K), an essential mediator of necroptosis [21]. Results indicate that Z-VAD-fmk alone slightly attenuated sorafenib- induced death in MM.1S cells and necrostatin-1 alone had no preventive effect. However, the combination of Z-VAD-fmk and necrostatin-1 efficiently prevented so- rafenib-induced cell death in these cells (Fig. 2c), from 56.3 ± 5.6 to 14.7 ± 3.2. However, necrostatin-1 did not increase the high degree of protection offered by Z-VAD- fmk on sorafenib-induced death of RPMI 8226 cells (Fig. 2c). Analysis of RIP1 and RIP3 levels revealed dif- ferences between MM.1S and RPMI 8226 cells (Fig. 2d). RIP3 was downregulated by sorafenib in MM.1S; the combination of both inhibitors avoided both loss of RIP3 and cell death induced by sorafenib in these cells. No significant changes in the levels of RIP3 were observed in RPMI 8226 cells.
Antiapoptotic proteins of the Bcl-2 family, Bcl-xL or Mcl-1 for instance, can block mitochondrial permeabili- zation by interacting with Bax and Bak or with proteins of the BH3-only subfamily [22]. Furthermore, Mcl-1 degra- dation has been proposed to be a key process for sorafenib- induced apoptosis. We decided to evaluate the effect of Mcl-1 and Bcl-xL overexpression on the sensitivity to so- rafenib. According to the expected, overexpression of Bcl- xL significantly blocked sorafenib-induced apoptosis both in RPMI 8226 and MM.1S cells (Fig. 3b). Overexpression of Mcl-1 had a lower protective effect than Bcl-xL in both cell lines (Fig. 3b).
Effect of sorafenib on levels of mRNA and proteins involved in apoptosis
RT-MLPA analysis indicated that sorafenib, at doses near the LD50, induced moderate to small changes in mRNA levels of many apoptosis-related genes. The most signifi- cant variations detected in all cell lines were the decrease of survivin mRNA levels and a significant increase of the relative Puma mRNA levels (Fig. 4a). The increase in Puma mRNA transcripts in MM.1S cells was confirmed by real-time PCR analysis (Fig. 4b). Sorafenib at 7 lM induced a 1.6-fold increase in Puma mRNA. Lower doses of sorafenib did not induced significant changes in Puma mRNA. The levels of Bim mRNA, another gene of the BH3-only subfamily, only increased at the lower dose (Fig. 4b), which induces cell death in only a small per- centage of cells at 24 h (Fig. 1), suggesting that this protein could contribute to an early response to sorafenib. At the protein level, Western blot analysis revealed that sorafenib treatment induced a time-dependent decrease of antiapoptotic proteins Mcl-1, Bcl-xL and survivin and an increase in proapoptotic Puma in MM.1S, U266 and H929 cells (Fig. 5a). The decrease in Mcl-1 and Bcl-xL levels roughly correlated with an acceleration of apoptosis at 16–24 h (Fig. 1b). Time-course experiments in RPMI 8226 and MM.1S cells incubated with sorafenib revealed that the decrease of Mcl-1, Bcl-xL and Puma was caspase-depen- dent, since it was reverted by co-treatment with Z-VAD- fmk (Fig. 5b). Indeed, in MM.1S and RPMI 8226 cells treated with sorafenib and Z-VAD-fmk, a clear increase of Puma levels, more conspicuous in MM.1S cells (Fig. 5b), was observed. XIAP levels also decreased after 24 h in MM.1S cells after 24-h incubation with sorafenib while Bim levels remained roughly constant or slightly increased, depending on the cell line (Fig. 5a). Western blot analysis of purified CD138+ MM cells from three patients indicated that sorafenib-induced apoptosis was also accompanied by a decrease in Mcl-1 and Bcl-xL levels (Fig. 5c), as found in MM cell lines. Puma levels remained steady or slightly increased at the lower concentrations of sorafenib tested and decreased at higher doses (Fig. 5c), probably due to cell death and secondary necrosis.
Gene-silencing experiments
To verify the role of Puma and other Bcl-2 family proteins in sorafenib-induced apoptosis, we reduced their expres- sion by transfection with the corresponding siRNA. Transfection with Puma-siRNA significantly decreased levels of Puma-a and, to a lesser extent those of Puma-b in MM.1S and RPMI 8226 cells (Fig. 6a). When sensitivity to sorafenib was analyzed in Puma-silenced cells, we observed a partial protection, in accordance with the partial reduction in Puma protein levels. Bim gene silencing with Bim-siRNA significantly reduced Bim protein levels but did not alter the amount of sorafenib-induced death in MM.1S cells (Fig. 6c). Reducing Bcl-xL levels in MM.1S cells by treatment with the corresponding siRNA greatly increased sensitivity to sorafenib (Fig. 6b).
Discussion
The Raf–MEK–ERK mitogen-activated protein kinase (MAPK) pathway is a key in supporting survival of mye- loma cells [8, 9] and thus, an attractive target for molecular therapy [23]. This pathway is activated by IL-6 and other myeloma growth factors [6, 24]. In addition, the MAPK pathway can also be activated in MM cells by oncogenic Ras mutations, which can occur in about 35 % of plasma cells [25, 26]. Sorafenib, a drug initially designed as a B- and C-Raf inhibitor, indeed inhibits VEGFR and a number of other kinases [27]. Sorafenib is currently used in the mutant of RIP3 leads to caspase-independent cell death. In the case of RPMI 8226 cells, Z-VAD-fmk offered an almost complete protection, indicating that in these cells the necroptotic pathway is not activated by sorafenib as efficiently as in MM.1S cells. Further studies will be nee- ded to understand the molecular determinants of this dif- ferent behavior.
Bcl-2 proteins are the main regulators of the intrinsic, or mitochondrial, pathway of apoptosis. Present results and previous reports clearly demonstrate that sorafenib acts through this pathway, characterized by activation of Bax and Bak and release of mitochondrial apoptogenic proteins, in MM cells. However, BH3-only proteins that cause Bax/Bak activation in this model have not been determined. A role for Bim was discarded by Kharaziha et al. [11] since a MM cell line lacking Bim, LP1, is still sensitive to sorafenib and silencing of this protein also suggests that it is dispensable for sorafenib toxicity on MM cells. Among BH3-only proteins, Puma is a potent apoptosis inducer and a contributor to cell death induced by several antitumor agents [33]. We found Puma levels increased during sorafenib treatment at the mRNA and protein level. In other cellular model it has been shown that Puma induction occurs after sorafenib-mediated MAPK pathway inhibition, as a consequence of the reactivation of GSK3b, which in turns activates the transcription factor NF-jB [15]. Moreover, our results indicate that Puma is an important player in the induction of cell death by sorafenib, as demonstrated by gene- silencing experiments. Puma can directly elicit Bax and Bak activation [22] and the subsequent release of cyto- chrome c from mitochondria and caspase-3 activation, leading to the proteolysis of key cellular substrates. One of these substrates is Puma protein itself, and so, Puma increase is only clearly appreciable in the presence of the pan-caspase inhibitor Z-VAD-fmk. Puma up-regulation could contribute to caspase-independent cell death observed in MM.1S cells, as described in other models [34, 35]. Mitochondrial permeabilization in the intrinsic pathway is negatively regulated by antiapoptotic members of the Bcl-2 family that bind and inhibit Bax, Bak and the BH3-only proteins [22]. Mcl-1 is a short-life antiapoptotic protein and a decrease in its levels is usually observed upon activation of the intrinsic pathway by several stim- uli, including sorafenib [10, 11]. Our present data indicate that Mcl-1 downregulation due to sorafenib treatment is a caspase-dependent process. In MM.1S cells, cell death is observed in the presence of Z-VAD-fmk despite keeping high Mcl-1 expression. Furthermore, we have observed that Mcl-1 overexpression offers very limited protection against sorafenib toxicity, both in MM.1S and RPMI 8226 cells. In contrast, Bcl-xL, another antiapoptotic protein of the same family, protected MM.1S and RPMI 8226 from
sorafenib toxicity more efficiently than Mcl-1 and decreasing Bcl-xL levels by treatment with siRNA greatly potentiates sorafenib-induced apoptosis. These observa- tions would explain the potentiating effect of ABT-737, a BH3-mimetic that inhibits Bcl-2 and Bcl-xL, on sorafenib- induced cell death [11]. Altogether, these results suggest that caspase-mediated degradation of Bcl-xL and Mcl-1 contributes to accelerate cell death by liberating Puma and other proapoptotic proteins that permeabilize mito- chondria. The relative contribution of this amplification loop for apoptosis varied among cell lines and it probably depends on the extent of initial destabilization of mito- chondria caused by Puma. This would explain why cas- pase inhibition prevents most of sorafenib-induced death in RPMI 8226 but barely in MM.1S cells.
In summary, our data, obtained using also cells from MM patients, indicate that sorafenib could potentially be a useful drug for MM therapy, since in addition to Puma- mediated apoptosis, it can also induce non-apoptotic cell death. Phase I clinical trials results [36] support future assays on the use of this compound for the treatment of multiple myeloma, specially in combination with other anti-myeloma drugs.