In vitro and in vivo anti-leukemic effects of cladoloside C2 are mediated by activation of Fas/ceramide synthase 6/p38 kinase/c-Jun NH2-terminal kinase/caspase-8

We previously demonstrated that the quinovose-containing hexaoside stichoposide C (STC) is a more potent anti-leukemic agent than the glucose-containing stichoposide D (STD), and that these substances have different molecular mechanisms of action. In the present study, we investigated the novel marine triterpene glycoside cladoloside C2 from Cladolabes schmeltzii, which has the same carbohydrate moiety as STC. We assessed whether cladoloside C2 could induce apoptosis in K562 and HL-60 cells. We also evaluated whether it showed antitumor action in mouse leukemia xenograft models, and its molecular mechanisms of action. We investigated the molecular mechanism behind cladoloside C2-induced apoptosis of human leukemia cells, and examined the antitumor effect of cladoloside C2 in a HL-60 and K562 leukemia xenograft model. Cladoloside C2 dose- and time-dependently induced apoptosis in the analyzed cells, and led to the activation of Fas/ceramide synthase 6 (CerS6)/p38 kinase/JNK/caspase-8. This cladoloside C2-induced apoptosis was partially blocked by specific inhibition by Fas, CerS6, and p38 siRNA transfection, and by specific inhibition of JNK by SP600125 or dominant negative-JNK transfection. Cladoloside C2 exerted antitumor activity through the activation of Fas/CerS6/p38 kinase/JNK/caspase-8 without showing any toxicity in xenograft mouse models. The antitumor effect of cladoloside C2 was reversed in CerS6 shRNA-silenced xenograft models. Our results suggest that cladoloside C2 has in vitro and in vivo anti-leukemic effects due to the activation of Fas/CerS6/p38 kinase/JNK/caspase-8 in lipid rafts. These findings support the therapeutic relevance of cladoloside C2 in the treatment of human leukemia.


INTRODUCTION
Leukemia is a heterogeneous clonal disorder characterized by defects in cell differentiation and the death of hematopoietic progenitor cells. Despite major advances in drug development, leukemia treatment remains limited by resistance to chemotherapeutic agents in many patients [1,2]. Thus, there remains a need for new therapeutic agents and strategies to improving the leukemia cure rate.

Research Paper www.impactjournals.com/oncotarget
The tumor-suppressor lipid ceramide reportedly exhibits potent growth inhibition effects in a variety of cell types [3]. Ceramide can be generated by either ceramide synthases or sphingomyelinases [4,5]. Sphingomyelinases (SMases) are categorized as acid, neutral, or alkaline [6,7], according to the pH at which they show maximum activity [8]. Many anticancer agents increase ceramide levels to varying extents in different types of cancer cells [9]. Thus, the pharmacological modulation of sphingolipid metabolism to enhance ceramide in tumor cells represents a novel therapeutic approach.
Marine triterpene glycosides exhibit a wide range of biological activities, including antitumor activity [10,11]. We previously demonstrated that stichoposide C (STC) ( Figure 1A)-a hexaoside containing quinovose as the second monosaccharide unit-induces apoptosis of leukemia cells by generating ceramide. The mechanisms for ceramide generation include activation of acid SMase after activating caspase-8, and the activation of neutral SMase resulting from glutathione depletion and increased ROS production [12]. Another study reported that stichoposide D (STD) ( Figure 1A)-a STC structural analog that contains glucose instead of quinovose in its carbohydrate chains-induces leukemia cell apoptosis through the activation of ceramide synthase 6 (CerS6) [13]. These studies suggest that marine triterpene glycosides, particularly hexaosides containing a quinovose as the second monosaccharide, are strong candidates for anti-leukemic agents.
Cladoloside C 2 , a novel marine triterpene glycoside with a quinovose as the second monosaccharide ( Figure 1A), has been isolated from the holothurian Cladolabes schmeltzii (Subfamily Cladolabinae, Family Sclerodactylidae, Order Dendrochirotida). In our present study, we investigated the molecular mechanism underlying the anti-leukemic potential of cladoloside C 2 in K562 and HL-60 cells, and mouse leukemia xenograft models. Our data provide the first evidence that cladoloside C 2 induces apoptosis of human leukemic cells through an extrinsic, but not an intrinsic pathway. We further demonstrated that the in vitro and in vivo anti-leukemic effects of cladoloside C 2 occur through a mechanism involving the activation of Fas/CerS6/p38 kinase/c-Jun-NH 2 -terminal kinase (JNK)/ caspase-8 in lipid rafts.

Cladoloside C 2 induces apoptosis of leukemic cells through extrinsic pathway activation
To examine whether cladoloside C 2 can induce apoptosis of K562 and HL-60 cells, K562 and HL-60 cells were treated with various cladoloside C 2 concentrations for different time periods, and co-stained with PI and FITC-conjugated annexin V. Cladoloside C 2 treatment dose-and time-dependently increased the proportions of apoptotic cells ( Figure 1B).
In contrast, the concentrations of cladoloside C 2 that were used in this study (0.1-1.0 μM) did not increase apoptosis of normal human hematopoietic progenitor cells (CD34 + cells) compared to control, as further confirmed by annexin-V/PI staining (data not shown).
To assess mitochondrial pathway activation by cladoloside C 2 treatment, we measured the mitochondrial membrane potential (MMP) and examined mitochondrial protein expression in the cytosol using western blot analysis. Cladoloside C 2 -treated K562 and HL-60 cells showed no MMP loss ( Figure 1E), as well as no cytoplasmic release of cytochrome c, Smac/DIABLO, or AIF ( Figure 1F). These findings indicate that cladoloside C 2 treatment of K562 and HL-60 cells activated extrinsic apoptotic pathways, but not intrinsic pathways. To explain this phenomenon, we further investigated how cladoloside C 2 treatment affected the levels of the antiapoptotic proteins myeloid cell leukemia-1 (Mcl-1), B-cell lymphoma-2 , and B-cell lymphoma extra large (Bcl-xL); and the proapoptotic protein Bcl-2-associated X protein (Bax). Interestingly, cladoloside C 2 led to increased expressions of Mcl-1 and Bcl-xL, decreased expression of Bax, and no change in Bcl-2 expression ( Figure 1C) . The decreased Bax expression and increased expressions of Mcl-1 and Bcl-xL may support mitochondrial preservation.

Cladoloside C 2 generates ceramide through activation of ceramide synthase 6 following Fas activation in human leukemic cells
We previously demonstrated that quinovose-containing STC induces apoptosis through ceramide generation via activation of acid and neutral SMase [12]. We expected that cladoloside C 2 and STC would induce apoptosis through the same mechanism. Thus, we performed immunofluorescence staining, and found that cladoloside C 2 increased ceramide generation (Figure 2A). To evaluate whether cladoloside C 2 -induced apoptosis was mediated by acid SMase, neutral SMase, or ceramide synthase, we incubated cells for 1 h with www.impactjournals.com/oncotarget the acid SMase inhibitor desipramine, the neutral SMase inhibitor GW4869, the serine palmitoyl transferase inhibitor myriocin, or the ceramide synthase inhibitor fumonisin B 1 , followed by treatment with cladoloside C 2 . Cladoloside C 2induced apoptosis was partially blocked by pretreatment with myriocin or fumonisin B 1 , and was not blocked by pretreatment with desipramine or GW4869 ( Figure 2B). Cladoloside C 2 -induced ceramide generation in K562 and HL-60 cells was also blocked by pretreatment with myriocin or fumonisin B 1 (data not shown). We also examined the expressions of CerS4, CerS5, and CerS6 during cladoloside C 2 -induced apoptosis. Cladoloside C 2 treatment markedly induced expression of CerS6, but not CerS4 or CerS5 (Supplementary Figure 1A, 1B). Thus, we focused on the role of CerS6 in cladoloside C 2 -mediated cell death. Western blot analysis and immunofluorescence staining revealed that cladoloside C 2 treatment increased CerS6 expression in K562 and HL-60 cells ( Figure 2C).
To verify the essential role of CerS6 activation in cladoloside C 2 -mediated apoptosis, we transfected K562 and HL-60 cells with siRNA against CerS6 or nonspecific control siRNA. Western blot analysis and immunofluorescence staining confirmed CerS6 knockdown ( Figure 2D, 2E). The extent of apoptosis was monitored in cladoloside C 2 -treated transfected cells. CerS6 knockdown by siRNA partially protected cells from cladoloside C 2induced apoptosis ( Figure 2F). We observed similar results in CerS6 shRNA-silenced K562 and HL-60 cells (Supplementary Figure 2A-2C).
To establish the sequence of events accompanying cladoloside C 2 -induced cell death, we monitored cladoloside C 2 -induced activation of Fas, caspase-8, and caspase-3 in CerS6 siRNA-transfected K562 and HL-60 cells. CerS6 siRNA transfection reversed the activation of caspase-8 and caspase-3, but not Fas ( Figure 2D). These data suggest that CerS6 activation occurred downstream of Fas activation and upstream of caspase-8 and caspase-3.

Fas activation is involved in cladoloside C 2 -induced CerS6 activation and apoptosis
Since cladoloside C 2 treatment activated Fas ( Figure 1C), we evaluated the functional significance of this activation by performing Fas knockdown with Fas siRNA in K562 and HL-60 cells. Western blot analysis and immunofluorescence staining confirmed Fas knockdown ( Figure 3A, 3B), and the extent of apoptosis was monitored in cladoloside C 2 -treated transfected cells. Fas knockdown partially protected cells from cladoloside C 2 -induced apoptosis ( Figure 3C), and reduced cladoloside C 2 -induced CerS6 activation and ceramide generation ( Figure 3A, 3B). Additionally, Fas siRNA transfection inhibited cladoloside C 2 -induced activation of caspase-8 and caspase-3 ( Figure 3A).

Activation of p38 kinase and JNK occurs downstream of Fas and CerS6 activation, and may contribute to cladoloside C 2 -induced apoptosis in human leukemic cells
Ceramide activates multiple signaling pathways, including those involving mitogen-activated protein kinases (MAPKs) [14][15][16][17][18]. MAPKs-such as the extracellular signal regulating kinase (ERK), p38 kinase, and JNK-are centrally involved in stress-induced cell death, and in apoptotic signaling of ceramide. We explored the involvement of ERK, p38 kinase, and JNK in cladoloside C 2 -induced apoptosis by treating K562 and HL-60 cells with cladoloside C 2 for various time periods, and then measuring MAPK protein levels by western blot analysis. Cladoloside C 2 time-dependently activated all MAPKs ( Figure 4A). To ascertain the roles of activated ERK, p38 kinase, and JNK in cladoloside C 2 -induced cell death, we used specific inhibitors of ERK (PD98059), p38 kinase (SB203580), and JNK (SP600125) and measured the extent of apoptosis after 6 h of cladoloside C 2 treatment. Apoptosis was significantly reduced by inhibition of p38 kinase and JNK, but not ERK ( Figure 4B). Inhibition of p38 kinase and JNK also significantly inhibited cladoloside C 2 -induced caspase-8 activation, but not Fas activation, CerS6 activation, or ceramide generation ( Figure 4C, 4D).
To further confirm the crucial role of p38 kinase in cladoloside C 2 -induced apoptosis, we transfected K562 and HL-60 cells with p38 kinase siRNA. Western blot analysis confirmed p38 kinase knockdown ( Figure 5A), The blot is representative of three separate experiments. (D) Functional involvement of caspases in cladoloside C 2 -induced apoptosis of K562 and HL-60 cells. Cells were pretreated for 1 h with the pan-caspase inhibitor Z-VAD-FMK (25 μM), the caspase-8 inhibitor Z-IETD-FMK (20 μM), the caspase-9 inhibitor Z-LEHD-FMK (20 μM), or the caspase-3 inhibitor Z-DEVD-FMK (50 μM), followed by treatment with 0.2 or 0.1 μM cladoloside C 2 for 6 h. The extent of apoptosis was measured by flow cytometry after annexin V staining. These data represent the mean ± SD of three independent experiments. ** P < 0.01; *** P < 0.001 vs. cladoloside C 2 -treated cells. (E) Left panel: K562 and HL-60 cells were treated with 0.2 or 0.1 μM cladoloside C 2 for 6 h. The extent of apoptosis was measured by flow cytometry after annexin V staining. These data represent the mean ± SD of three independent experiments. *** P < 0.001 vs. control cells. Right panel: K562 and HL-60 cells were treated with 0.2 or 0.1 mM cladoloside C 2 for 4 h or 2 h. The cells were stained with DiOC 6 , and the reduction in Δφ m was determined by monitoring the DiOC 6 uptake using flow cytometry. Low Δφ m values are expressed as the percentage of cells exhibiting diminished mitochondrial potential. The values obtained from the DiOC 6 assays represent the mean ± SD of three independent experiments. (F) Western blot for the mitochondrial proteins AIF, Smac/DIABLO, cytochrome oxidase IV, and cytochrome c. Cytochrome oxidase IV (COX IV) was used as a mitochondrial marker. Protein lysates were prepared and subjected to western blot analysis using corresponding antibodies. β-actin was used as a loading control. The blot is representative of three separate experiments. and the extent of apoptosis was monitored in cladoloside C 2 -treated transfected cells. Knockdown of p38 kinase partially protected cells from cladoloside C 2 -induced apoptosis ( Figure 5B), but did not reduce cladoloside C 2induced Fas and CerS6 activation ( Figure 5A, 5C). These siRNA experiments also revealed inhibition of caspase-8 and caspase-3 activation ( Figure 5A).
To ascertain whether JNK activation was necessary for cladoloside C 2 -mediated apoptosis, K562 cells were transiently transfected with a dominant-negative JNK expression vector (DN-JNK) expression vector or empty vector. Western blot analysis confirmed JNK inhibition ( Figure 5D), and the extent of apoptosis was monitored in cladoloside C 2 -treated transfected cells. JNK inhibition partially protected cells from cladoloside C 2 -induced apoptosis ( Figure 5E), but did not reduce cladoloside C 2 -induced activation of Fas, CerS6, or p38 kinase ( Figure 5D). DN-JNK transfection also inhibited activation of caspase-8 and caspase-3 ( Figure 5D).

Clustering of Fas and its downstream signaling molecules in lipid rafts during cladoloside C 2induced apoptosis
The above-described observations suggest that cladoloside C 2 activated Fas and CerS6, leading to ceramide generation, followed by the activation of p38 kinase, JNK, and caspase-8. We next examined the clustering of these signaling molecules (Fas, CerS6, ceramide, p38 kinase, JNK, and active caspase-8) in lipid rafts during cladoloside C 2 -induced apoptosis. K562 and HL-60 cells were pretreated with the cholesterol-depleting agent methyl-βcyclodextrin (MβCD) and nystatin for 1 h, followed by cladoloside C 2 treatment. Then we assessed the extent of apoptosis and the activation of Fas, caspase-8, caspase-3, and CerS6. Incubation of K562 and HL-60 cells with MβCD and nystatin inhibited cladoloside C 2 -induced apoptosis, as well as the activation of Fas, caspase-8, caspase-3, and CerS6 ( Figure 6A-6C).

Cladoloside C 2 induces antitumor activity through the activation of Fas, CerS6, p38 kinase, and JNK in K562 and HL-60 xenograft mouse tumor models
Finally, we observed that cladoloside C 2 significantly inhibited tumor growth in both HL-60 and K562 mouse xenograft models ( Figure 7A). Tumors from control mice displayed the typical histologic appearance of leukemic cells. After 21 d, tumors from cladoloside C 2 -treated mice showed mean volumes over 75% smaller than the volumes of tumors in vehicle-treated K562 and HL-60 xenograft mice (control group: 3174.18 ± 262.39 mm 3 , 3922.49 ± 241.42 mm 3 ; cladoloside C 2 group: 640.27 ± 159.74 mm 3 , 685.51 ± 154.62 mm 3 ). We used stable CerS6 shRNA-silenced K562 and HL-60 xenograft models to investigate the involvement of CerS6 in the in vivo antitumor activity of cladoloside C 2 . In parallel, CerS6-silenced cells and nonspecific control (NC) cells were subcutaneously inoculated into 6-week-old nude mice. Next, cladoloside C 2 or vehicle was injected into each mouse. The anti-tumor effect of cladoloside C 2 was significantly inhibited in CerS6 shRNA-silenced K562 and HL-60 xenograft models ( Figure 7B), with 75.6% and 84.1% inhibition of tumor growth in NC-shRNA-1 and NC-shRNA-3 xenograft models, respectively, vs. 15.6% and 25.6% inhibition of tumor growth by cladoloside C 2 in CerS6-shRNA-1 and CerS6-shRNA-5 xenograft models, respectively. As expected, western blot analysis of tumors from vehicle-treated NC-shRNA-1 and NC-shRNA-3 control mice revealed weak expressions of Fas, CerS6, p-p38, and p-JNK ( Figure 7C), and we observed  Figure 7D). In contrast, western blot and immunohistochemical analysis of tumors from cladoloside C 2 -treated NC-shRNA-1 and NC-shRNA-3 shRNA mice revealed up-regulation of Fas, CerS6, ceramide, p-p38, and p-JNK ( Figure 7C, 7D). The expression and staining of p38 kinase and JNK did not differ between control and cladoloside C 2 -treated tumors ( Figure 7C, 7D). These data were consistent with our in vitro findings.

DISCUSSION
Marine triterpene glycosides represent promising candidate anticancer agents. However, their detailed molecular mechanisms have not been clearly defined. We previously reported that triterpene glycosides from Thelenota anax induce leukemic cell apoptosis through ceramide generation. However, molecular mechanisms and potencies differ among structurally different triterpene glycosides, and a previous study suggests that STC has more potent anti-leukemia cell activity than STD. Here we investigated the novel nonsulfated triterpene glycoside cladoloside C 2 , which contains the same carbohydrate chain as STC but with a different aglycone moiety. Our present results indicated that cladoloside C 2 is more potent than STC. Moreover, unlike STC, cladoloside C 2 apparently activates extrinsic apoptosis pathways, but not intrinsic pathways.
Interestingly, cladoloside C 2 did not seem to affect mitochondria, as we observed increased expressions of Mcl-1 and Bcl-xL and decreased Bax expression. Many cancer cells are apt to be resistant to chemotherapeutic agents due to overexpression of prosurvival factors, such as Mcl-1, Bcl-2, and Bcl-xL. Therefore, these proteins are important targets for the development of new anti-cancer agents [19][20][21]. Since cladoloside C 2 induces apoptosis despite enhancing the expressions of Mcl-1 and Bcl-xL, it may be useful as an anti-leukemic agent in leukemic cells that overexpress Mcl-1 and Bcl-xL. In particular, Mcl-1 overexpression reportedly leads to etoposide resistance [22], and we observed that cladoloside C 2 sensitized K562 cells to etoposide and Ara-C (Supplementary Figure 3). Further studies are required to confirm this sensitizing effect in various leukemic cells that are resistant to chemotherapeutic agents.
Since cladoloside C 2 and STC share the same carbohydrate chain structure, we expected that cladoloside C 2 , like STC, would induce apoptosis through the activation of acid and neutral SMases. Unexpectedly, we found that cladoloside C 2 induced apoptosis through CerS6 activation, following to Fas activation, and the subsequent activation of p38 kinase/ JNK/caspase-8. Similar to the mechanism of STD, Fas activation by cladoloside C 2 was not reversed by transfection with CerS6 siRNA. However, CerS6 activation and ceramide generation were reversed by Fas siRNA transfection.
In ceramide-induced apoptosis, MAPKs are important signaling molecules. Several previous studies have demonstrated that ceramide causes ERK dephosphorylation, and induces p38 kinase and JNK phosphorylation [14][15][16][17][18]. Ceramide-activated p38 kinase and JNK generally contribute to the induction of cell apoptosis through mitochondrial damage and caspase activation. Our present results demonstrated that cladoloside C 2 -generated ceramide activated ERK, p38 kinase, and JNK, but did not induce mitochondrial damage. Since JNK and p38 kinase signaling target the anti-apoptotic Mcl-1 and Bcl-2 proteins [23][24][25], we examined how p38 kinase and JNK inhibition (by chemical inhibitors or p38 kinase siRNA or DN-JNK) influenced Mcl-1 and Bcl-2 expressions, caspase-8 activation, and apoptosis. Interestingly, inhibition of p38 kinase by SB203580 and of JNK by SP600125 led to reduced Mcl-1 expression, and reversed capase-8 activation and apoptosis, but did not change Bcl-2 expression. Similar results were obtained from silencing p38 kinase using p38 kinase siRNA, and inhibiting JNK via DN-JNK transfection. Thus, it appears that the activities of p38 kinase and JNK lead to enhanced Mcl-1 expression and are involved in cladoloside C 2 -induced apoptosis.
Lipid rafts play crucial roles in the Fas receptor death pathway [26,27]. Based on the recently demonstrated role of plasma membrane lipid rafts in STD-induced apoptosis [13], here we wed MβCD and nystatin to investigate the roles of the Fas death receptor pathway and ceramide-enriched membrane domains in cladoloside C 2 -induced cell death. Cladoloside C 2 induced Fas clustering in lipid rafts. We also detected colocalization of the Fas downstream signaling molecules CerS6, ceramide, p38 kinase, JNK, and caspase-8 in lipid rafts upon cladoloside C 2 stimulation. MβCD blocked the cladoloside C 2 -induced clustering of Fas and the downstream signaling molecules, we well as apoptosis. These results suggest that the clustering of Fas and downstream signaling molecules in lipid rafts was essential for cladoloside C 2 -induced apoptosis. The samples were then stained with PE-conjugated antibodies against Fas, CerS6, ceramide, cleaved caspase-8, p-p38 kinase, p38 kinase, p-ERK, ERK, p-JNK, or JNK, as well as Alexa 488-labeled cholera toxin B antibody. The pictures are representative of three separate experiments. (C) The culture medium was changed, and cells were incubated for 6 h with or without cladoloside C 2 . The percentage of apoptotic cells was determined by annexin V-FITC/PI staining. Upper panels: Representative of three independent experiments in each cell line. Lower panels: Mean ± SD of three independent experiments. ** P < 0.01, cells treated with cladoloside C 2 versus cells transfected with Fas siRNA and treated with cladoloside C 2 . www.impactjournals.com/oncotarget Compared to controls, cladoloside C 2 treatment significantly inhibited tumor growth in mouse HL-60 and K562 leukemic xenograft models. Cladoloside C 2 treatment also led to up-regulation of Fas, CerS6, and ceramide and activation of p38 kinase, JNK, and caspase-8. The antitumor effect of cladoloside C 2 was significantly prevented in CerS6 shRNA-silenced xenograft models. These results are consistent with the in vitro data. Moreover, cladoloside C 2 showed a lack of toxicity towards normal hematopoietic progenitor cells and in mice, supporting it as a promising potential candidate for therapeutic use.
Future studies are needed to explore the antitumor activity of cladoloside C 2 in other types of leukemia, including chemotherapy-resistant leukemia cells, and in other types of cancer. There also remains a need to investigate other molecular mechanisms that may be involved in cladoloside C 2 -induced apoptosis.
In summary, this study provides the first evidence that cladoloside C 2 induces apoptosis of human leukemic cells via activation of Fas/CerS6/p38 kinase/JNK/caspase-8 in lipid rafts (Figure 8). Cladoloside C 2 also exhibited in vivo antitumor activity through the activation of Fas/CerS6/p38 kinase/JNK/caspase-8. Our results suggest that cladoloside C 2 may be a useful candidate for the treatment of human leukemia overexpressing Mcl-1 and Bcl-xL.

Cell preparations
The human leukemic cell lines K562 and HL-60 were obtained from the Korean Cell Line Bank (Seoul National University, Seoul, Korea), and cultured in RPMI1640 medium supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin. Human hematopoietic progenitor CD34 + cells were purchased from STEM CELL Technologies (Vancouver, BC), and cultured in Hematopoietic Progenitor Expansion Medium DXF with cytokine mix E (PromoCell, Heidelberg, Germany).

Reagents
Cladoloside C 2 was isolated and purified following the procedure published by Silchenko et al. [28], and was dissolved in sterilized distilled water. Annexin V was obtained from BD Biosciences Clontech (Palo Alto, CA, USA). Anti-Fas, anti-procaspase 8, anti-procaspase-3, anti-procaspase 9, and anti-cytochrome c antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Antibodies against poly (ADP-ribose) polymerase (PARP), JNK, and p-JNK were purchased from Cell Signaling Technology (Beverly, MA, USA). The antiβ-actin antibody was obtained from Sigma (St. Louis, MO, USA). Unless otherwise stated, all other chemicals were purchased from Sigma.

Apoptosis analysis
The extent of apoptosis was evaluated using annexin V-FITC and flow cytometry as previously described [29].

Separation of the cytosolic and mitochondrial proteins
Cytosolic and mitochondrial proteins were extracted from cells treated with sterilized water or with cladoloside C 2 for the indicated times, and were separated as previously described [30,31].

Western blot analysis
Cell lysis and western blot analysis were performed as described previously [29], using 30 μg protein for immunoblotting. β-actin was used as the loading control.

Immunofluorescence staining
Cells were fixed and permeabilized with 1% formaldehyde/methanol in PBS for 10 min at room temperature. Next, the cells were washed, and a series of antibodies was used as indicated, followed by staining with FITC-or PE-conjugated goat anti-mouse and antirabbit IgG (Calbiochem, San Diego, CA). The samples were then mounted using glycerol, and analyzed by (A) K562 and HL-60 cells were treated with cladoloside C 2 for the indicated times. Protein lysates were prepared and subjected to western blot analysis. β-actin was used as a loading control. The blot is representative of three separate experiments. (B) K562 and HL-60 cells (1 × 10 5 cells/well) were pretreated with the p38 kinase inhibitor SB203580, the JNK inhibitor SP600125, or the ERK inhibitor PD98059, followed by 6 h of treatment with 0.2 or 0.1 μM cladoloside C 2 . After treatment for the indicated times, the percentage of apoptotic cells was determined by annexin V-FITC/PI staining. Upper panel: Representative of three independent experiments. Lower panel: Mean ± SD of three independent experiments. * P < 0.05; ** P < 0.01; *** P < 0.001, versus cells treated with cladoloside C 2 in the absence of SB203580 or SP600125. (C and D) K562 and HL-60 cells (1 × 10 5 cells/well) were pretreated with the p38 kinase inhibitor SB203580 (C) or the JNK inhibitor SP600125 (D), followed by treatment with 0.2 or 0.1 μM cladoloside C 2 for 6 h. Protein lysates were prepared and subjected to western blot analysis using corresponding antibodies. Western blots are each representative of three separate experiments. β-actin was used as a loading control. Densitometry results are expressed above the bands.
confocal microscope (Carl Zeiss LSM 510; Carl Zeiss, Thornwood, NY) with a 40× C-Apochromat objective. Negative control staining was performed using only secondary antibodies.

siRNA transfection
We purchased pre-designed siRNA targeted to human CerS6-1 mRNA (catalog number SI02758245; ID 253782), and AllStars negative control siRNA (catalog number 1027280) from Qiagen (Hilden, Germany). The siRNA sequence used for targeted silencing of CerS6 has been previously described [13]. Fas siRNA was obtained from Santa Cruz Biotechnology, and p38 kinase siRNA from Dharmacon (L-003512-00-0005; Thermo Scientific, Chicago, IL, USA), and their sequences have been previously published [13].
Cells were resuspended in PBS at 1.3 × 10 7 cells/0.5 mL, and then mixed with 200 nM anti-CerS6 siRNA, anti-p38 kinase siRNA, anti-Fas siRNA, or non-silencing siRNA. This mixture was added to an electroporation cuvette with a 0.4-cm electrode gap, for transfection at 300 V and 950 μF in a Gene Pulser Xcell Electroporation System (Bio-Rad, Richmond, CA, USA). After electroporation, the cells were cultured for 48 h in RPMI1640 supplemented with 10% FBS, then treated with sterilized water or cladoloside C 2 for the indicated times. These cells were then analyzed using annexin-V staining, immunofluorescence, and western blot.

Plasmids and transfection
K562 and HL-60 cells (1 × 10 6 ) were transfected with 6 μg of DN-JNK expression vector or empty vector (pUSEamp) (Upstate Technology, Lake Placid, NY, USA) using Lipofectamine following the manufacturer's protocol. After transfection, cells were cultured for 24 h in RPMI-1640 supplemented with 10% FBS, and then treated for 6 h with sterilized water or cladoloside C 2 . These cells were analyzed by annexin-V staining, immunofluorescence, and western blot.

Establishment of HL-60 and K562 leukemia xenograft models
All animal procedures and care were approved by the Institutional Animal Care and Usage Committee of Dong-A University. To determine the in vivo activity of cladoloside C 2 , K562 cells (1.5 × 10 7 /100 μL PBS per mouse) and HL-60 cells (2 × 10 7 /100 μL PBS per mouse), which were confirmed to be viable by trypan blue staining, were injected into the right flanks of 6-week-old female Balb/c nude mice (n = 5 mice per group; Orient Bio Inc., Korea), as previously described [12,29]. To confirm the essential role of CerS6 in the in vivo antitumor activity of cladoloside C 2 , K562 and HL-60 cells expressing NC construct and K562 and HL-60 cells expressing CerS6 shRNA were injected, at the above-mentioned concentrations, into the right flanks of 6-week-old female Balb/c nude mice (n = 5 mice per group; Orient Bio Inc., Korea). When the average subcutaneous tumor volume reached 60-100 mm 3 , the mice were assigned  to either the cladoloside C 2 treatment or control group, receiving 7.19 μg/kg cladoloside C 2 or vehicle via the tail vein every 3 days. Tumor size was measured using a caliper: calculated volume = shortest diameter 2 × longest diameter/2. Mice were followed for tumor size and body weight, and were sacrificed on the 14th or 21st day. Tumors were resected, weighed, and frozen or fixed in formalin and paraffin embedded for western blot or immunohistochemical analyses.