γ-Tocotrienol induces apoptosis in pancreatic cancer cells by upregulation of ceramide synthesis and modulation of sphingolipid transport

Pancreatic cancer is the fourth leading cause of cancer-related deaths in the United States [1]. Since available treatment options are limited, novel therapeutic agents that demonstrate the ability to inhibit signaling pathways implicated in the proliferation and survival of pancreatic cancer cells need to be evaluated for drug development. Promising agents may uncover new targets and alternate treatment strategies that have the potential to contribute to the understanding of pancreatic carcinogenesis and progression. It has been known for a long time that ceramides can inhibit cell proliferation and induce apoptosis in cancer cells via various stress stimuli such as tumor necrosis factor-α and platelet-activating factor [2, 3]. Furthermore, recent studies in the synthesis and metabolism of ceramides suggest that changes in the expression levels of these compounds may contribute to metastasis and resistance to therapy [4]. The syntheses of ceramides occur through multiple pathways that include: de novo synthesis from serine and palmitoyl-CoA substrates, salvage, from sphingosine [5] and from the hydrolysis of sphingomyelin by acid sphingomyelinase (ASM). The de novo synthesis is initiated in the cytoplasmic face of the endoplasmic reticulum by serine palmitoyl transferase (SPT), to form 3-keto-sphinganine, which is subsequently reduced to sphinganine (SA). Ceramide synthase (CerS) acetylates SA followed by desaturation by ceramide desaturase (DES) to form ceramide [6, 7]. There are six CerSs that regulate ceramide synthesis to produce a variety of compounds with di-and tri-hydroxy long-chain bases linked to fatty acids of variable length [8] and with C16 and C24 ceramides being most abundant in mammalian cells. These highly hydrophobic molecules can displace cholesterol and disrupt lipids rafts that may be associated with signaling molecules, thus affecting their function [9]. Moreover, the biophysical properties of ceramides may influence lipid reorganization in the membrane and cause destabilization, efflux and fusion. Hence, their expression levels and localization are tightly controlled.

Tocotrienols are members of the vitamin E family that unlike tocopherols possess an unsaturated isoprenoid side-chain [10]. These compounds have shown cytotoxic activity on pancreatic cancer cells via a multi-pronged mechanism. We had previously shown that γ-tocotrienol (GT3) is cytotoxic to pancreatic cancer cells, and is significantly more potent in its ability to inhibit cell viability as compared to alpha-tocopherol [11]. The ability of tocotrienols to selectively inhibit the PI3 kinase/Akt pathway, Ras/Raf/Erk signaling [11], HMG CoA reductase, and transcription factor NF-κB [12], are contributors to these properties. In pancreatic cancer, the oncogenic process is frequently driven by aberrant K-Ras. We have shown that GT3 can cause inhibition of cellular proliferation and survival in pancreatic cancer cells regardless of their K-Ras status [11]. However, the mechanism of action is not completely understood. It has been reported that vitamin E isoforms other than tocotrienols can increase cellular ceramide and dihydroceramide levels. Alpha-TEA, a modified form of alpha tocopherol, can increase membrane ceramide levels in mammary cancer cells [13], and γ-tocopherol has a similar effect on prostate cancer cells [14]. In vivo, pharmacokinetics studies have demonstrated the bioavailabilty of tocotrienols in humans [15]. These studies led us to determine whether the observed apoptotic effects in pancreatic cancer cells dosed with GT3 involved changes in ceramide transport and levels in K-Ras mutated cells as compared to wild type. Here we show that GT3 causes an increase in the levels of certain ceramides at the plasma membrane by the upregulation of enzymes involved in both the de novo pathway and the hydrolysis of sphingomyelin, and the modulation of ceramide transporters regardless of K-Ras status. The apoptotic nature of these changes is confirmed by the clustering of death receptor 5 at the membrane and confirming previous observations of the mechanism of action by which GT3 inhibits cell proliferation and survival in pancreatic cancer cells.

Upregulation of ceramide via the de novo pathway and the hydrolysis of sphingomyelin by GT3 in pancreatic cancer cells regardless of their Ras status. Previous studies had shown that apoptosis is induced by GT3 in both wild type and mutated K-Ras pancreatic cancer cell lines via a mechanism that involves disruption of signaling of receptor tyrosine kinase ErbB2. To test whether changes in ceramide expression levels occur, we probed the ceramide de novo synthesis pathway by analyzing the expression of enzymes SPT, CERS-6, DEGS1 at 2, 4, and 6 hours after dosing with GT3, K-Ras mutated MIA PaCa-2 (Fig. 1a-f) and Panc 1 (Fig. 2a-f), and wild type BxPC3 pancreatic cancer cells (Fig. 2g-l). The expression levels of all enzymes tested increased in a time-dependent manner in MIA PaCa-2, Panc 1, and BxPC3 pancreatic cancer cells. A 1.5 fold increase of SPT is apparent 2 hours after dosing MIA PaCa -2 cells with GT3 as compared to the control (Fig. 1a and b). Similarly, DEGS1 levels increase in all cell lines tested with up to 3.0 fold increase in MIA PaCa-2 cells (Fig. 1e and f). A robust increase in the CERS6 levels at 4 and 6 hours after treatment with GT3 also support the activation of this pathway (Fig. 1c and d). Many recent studies have demonstrated the central role of ASM in the apoptotic process via the formation of ceramide-enriched membrane domains, specifically by gamma radiation [18, 19], UV light [20, 21], and chemotherapeutic agents such as cisplatin [22] and doxorubicin [23]. To determine whether GT3 activates the hydrolysis of sphingomyelin, we probed the levels of ASM. Our results indicate that GT3 may also favor the hydrolysis of sphingomyelin to produce ceramide by increasing the concentration of ASM by 3.4 fold in MIA PaCa-2 cells after treatment with GT3 (Fig. 1g and h). A similar trend on the levels of ASM was observed in Panc 1 and BxPC3 pancreatic cancer cells (results not shown). These results suggest that the activation of ceramide synthesis by GT3 is likely to increase the ceramide levels in the treated cells. Conversely, in non-cancerous human pancreatic ductal epithelial cells (HPDE-E6E7), GT3 has no significant effects on any of the enzymes tested, as compared to the control (Fig. 2m-r).

Pancreatic cancer cells MIA PaCa-2 have higher total ceramide content than non-malignant pancreatic HPDE-E6E7 epithelial cells and GT3 cause a further increase in ceramides and dihydroceramides in MIA PaCa-2 cells. Ceramide is an important regulator of cellular homeostasis, involved in signaling pathways of apoptosis, senescence, and differentiation [24]. However, its expression levels are normally low and upregulation of ceramide concentration is tightly controlled. Increased levels of ceramides are known to cause apoptosis and are also involved in oncogenesis. To test whether pancreatic cancer MIA PaCa-2 cells have higher ceramide levels than epithelial pancreatic HPDE-E6E7 cells, the total content of ceramides was analyzed by LCMS. The results obtained are in agreement with previous studies in other cancers, and show that MIA PaCa-2 cells have approximately three-fold expression levels than non-malignant (HPDE-E6E7) pancreatic cells (Fig. 3a). Further accumulation of ceramides in the cell membrane may activate their apoptotic function, and support the observations in studies that have shown inhibition of cell viability by GT3. LCMS analysis showed that GT3 cause a further increase of 72.9%±1.82 total membrane ceramides in MIA PaCa-2 cells, confirming the observations on the activation of synthesis pathways of ceramide (Fig. 3b). Dihydroceramides are present at significantly lower concentrations as compared to ceramides in both MIA PaCa-2 and HPDE-E6E7 cells. Unlike ceramides, dihydroceramide levels in the membrane are three fold higher in HPDE-E6E7 than in MIA PaCa-2 cells. These compounds show an increase of 80.5 %±1.16, and 42.97%±3.47 after treatment with GT3 of MIA PaCa-2 and HPDE-E6E7 cells, respectively (Fig. 3b).

Membrane sphingomyelin is significantly decreased in MIA PaCa-2 and HPDE-E6E7 cells treated with GT3. Besides the de novo pathway, ceramide synthesis can occur via the hydrolysis of sphingomyelin. In order to determine whether there is a significant change in sphingomyelin content, LCMS analysis was conducted on cell lysates of GT3 treated and untreated MIA PaCa-2 and HPDE-E6E7 cells. In untreated cells, non-malignant cells have only slightly higher levels of sphingomyelin than pancreatic cancer cells (Fig. 3c). A significant decrease in membrane sphingomyelin of 26.54%±0.071 in MIA PaCa-2, and 70.05% ±11.17 in HPDE-E6E7 was observed after treatment with GT3, supporting our previous data on the activation of the ceramide synthesis pathway via sphingomyelin hydrolysis (Fig. 3c).

GT3 cause an increase in membrane ceramides and dihydroceramides C16, C24:1, and C24 and a decrease in sphingomyelins in MIA PaCa-2 cells Ceramides with a different fatty acid chain length, have different cellular functions. It has been previously reported that increase in the levels of C16, C24:1 and C24 ceramides, can induce cell death in lymphoma cells after treatment with cannabinoids [25], and the specific upregulation of C16 ceramide in leukemia and colon cancer cells can cause apoptosis [26]. To test whether membrane C16, C24:1, and C24 ceramides are altered by GT3 treatment of MIA PaCa-2 and HPDE E6E7 cells, we analyzed these compounds by LC/MS. Our data show that in untreated cells, C16 is present at 2.7-fold higher levels in MIA PaCa-2 than in HPDE E6E7 cells. Similarly, C24:1, and C24 ceramides, are also present at 7.5-fold and 3.5 fold higher levels in MIA PaCa-2 than in HPDE E6E7 cells. Treatment with GT3 produce a significant change only in C16 ceramide, with an increase of 32.36% as compared to the control (Fig. 4a). Conversely, in non-malignant HPDE-E6E7 cells, the only membrane ceramide that is observed to change significantly is also C16 with a 41.47% decrease as compared to the control (Fig. 4b). Dihydroceramides are present at significantly lower concentrations than ceramides in MIA PaCa-2 cells. However, these compounds may contribute to the total content of cellular ceramides. Our results show that GT3 cause an increase in dihydroceramides of 79.08% 87.34%, and 83.44% for C16, C24:1, and C24, respectively as compared to the control (Fig. 4c). HPDE-E6E7 cells have higher concentrations of dihydroceramides at the cellular membrane than MIA PaCa-2 cancer cells. Our results show that only C16 display a significant increase of 59.52% (Fig. 4d). All sphingomyelins tested displayed a significant decrease in MIA PaCa-2 cells after treatment with GT3. LCMS analysis of sphingomyelin in MIA PaCa-2 cells treated with GT3 showed a significant decrease of 29. 2% (C16), 24.44 (C24:1), and 45.0% (C24). (Fig. 4e). Similarly to sphingomyelins in MIA PaCa-2 cells, GT3 cause a significant decrease in HPDE-E6E7 -E6E7 cells of 79.5% (C16), 74.8% (C24:1), and 79.3% (C24) (Fig. 4f).

To determine whether the activation of the pathways of ceramide synthesis cause an effect on the expression levels of ceramide transporters, we probed ceramide transport protein CERT and ARV-1. The function of the ceramide transport protein CERT, is to transport newly synthesized ceramide from the ER to the Golgi [27], where the latter is hydrolyzed to sphingomyelin by sphingomyelinase [28]. To determine whether treatment of MIA PaCa-2, BxPC3, and Panc 1 cancer cells with GT3 favor the transport of newly synthesized ceramide to the Golgi, the levels of CERT and its phosphorylated form were probed by immunoblot. As shown in Fig. 5a and b, there is a significant decrease of both, the activated form and total CERT levels in GT3 treated MIA PaCa-2 cells as compared to the control. We also probed ARV1, a sterol/ceramide transport protein that interacts with genes involved in GPI anchor synthesis; it has been shown that GPI assembly is required for ceramide transport from the ER [29]. As shown in Fig. 5c, d and f, ARV-1 displays a dose-dependent increase in expression levels in MIA PaCa-2, BxPC3 and Panc 1 cells treated with GT3 at 5, 10, 20, and 40μM, with ARV-1 expression levels reaching a 3-fold increase at the highest concentration tested in MIA PaCa-2 cells. A similar trend was observed in BxPC3 and Panc 1 cells, suggesting a decrease in sphingomyelin biosynthesis from newly synthesized ceramide (Fig. 5g).This trend is also evident in rtPCR analysis suggesting that GT3 effect on ARV-1 may occur at the transcription level (Fig. 5e).

GT3 treatment of ARV-1 inhibited MIA PaCa-2 cells causes upregulation of the expression levels of caveolin-1. To test whether the effect of GT3 on ARV-1 occurs mainly at the messenger level, protein expression levels were analyzed by immunoblot in MIA PaCa-2 cells transduced with ARV-1 shRNA and treated with GT3. As shown in Fig. 6a and b, ARV-1 is effectively downregulated by ARV-1 shRNA (Fig. 6a and b). GT3 has no significant effect on the expression of ARV-1 (Fig. 6a and b). One of the functions of ARV-1 is to regulate sterol trafficking and plasma membrane structure. It is also known that ceramide is present in caveolae in cancer cells. To test whether GT3 has an effect on caveolin-1, we probed the expression levels of this protein in the presence and in the absence of ARV-1 in MIA PaCa-2 cells. Data shown in Fig. 6c and d suggest that in the presence of ARV-1, there is no significant change in the expression levels of this molecule. However, substantial upregulation is observed in the absence of ARV-1 regardless of GT3 treatment. Caveolin-1 is present in caveolae and is a major component of the vesicular transport system of the trans-Golgi network [27]. To study the role of ARV-1 in the localization of ceramide and caveolin in GT3 treated cells, ARV-1 shRNA transduced cells were processed for immunofluorescence. In the absence of ARV-1, caveolin (green channel) and ceramide (red channel) are localized at the plasma membrane and intracellularly in GT3 treated cells (Fig. 6e).

Ceramide-rich caveolae favor clustering of DR5 in GT3 treated pancreatic cancer cells. It has been reported that a high presence of ceramide in caveolae in human colon cancer cells, favors apoptosis via clustering of Death Receptor 5 [28]. We have previously shown that GT3 has a potent inhibitory effect on ErbB2 phosphorylation [11]. Since ErbB2 is co-localized with caveolin [30], we probed whether the increase of membrane ceramides and the onset of apoptosis by GT3 include the clustering of DR5, by determining the presence and localization of the receptor via immunofluorescence. In untreated and treated cells, ceramide and DR5 are localized at the membrane, red and green channels respectively, Fig. 7a xy and xz planes. Intensity correlation analysis of 14 fields, each with an average of 12.5 cells, showed no significant change with a percentage of colocalization of 74.6% in untreated cells (Fig. 7b) and 73.8% in treated cells (Fig. 7d). However, there was a substantial increase in fluorescence intensity on treated cells, of 3-fold for DR5, and 1.5-fold for ceramide (Fig. 7e) as compared to the control (Fig. 7c). To confirm these observations, analysis of mean intensities of the red and green channels was conducted in these cells (Fig. 8a) GT3 treated cells showed a significant increase in intensities of both channels as compared to untreated cells (Fig. 8b). These results suggest that the activation of apoptosis by GT3 in pancreatic cancer cells may include the participation of DR5 and an increase of ceramide expression levels at the cell membrane suggesting the presence of ceramide micro domains that have been described previously [31].

Our previous work indicated that GT3 has potent anti-proliferative activity in both k-Ras wild type and k-Ras mutated pancreatic cancer cells, but not on non-cancerous pancreatic duct epithelial HPDE-E6E7 cells. The mechanism of action included inhibitory effects on transcription factor NF-κB [12], ErbB2, and the PI3K, MAPK, and HMG CoA reductase pathways, as well as induction of apoptotic signaling through Foxo3 and GSK-3β [11]. However, the coordination of these multi-pronged effects is not completely understood. The results presented in this study suggest that the modulation of expression and localization of ceramide is a relevant factor contributing to the previously observed effects on cell signaling pathways and cell survival. The increase of expression of ASM, responsible for the production of ceramide from sphingomyelin, lends support to this notion. ASM localizes in sphingolipid and cholesterol-enriched membrane domains [32], and the ceramide molecules thus generated, freely associate with each other forming strongly stabilized [33] ceramide-enriched membrane domains [25] by the selective displacement of cholesterol [9]. These changes in membrane composition alter the relatively rapid lateral diffusion characteristic of cholesterol-rich liquid-ordered state [26], and cause destabilization in the structure and function of membrane domains where ErbB2 and other signaling molecules are known to be localized. We had previously shown that cholesterol depletion result in the loss of ErbB2 activation and subsequent inhibition of the ERK pathway [17] It has also been shown that ceramide can cause apoptosis via inactivation of the Ras-Raf/MEK pathway [34]. Thus the exclusion of cholesterol from the plasma membrane, and the downregulation of cholesterol synthesis through inhibition of the HMG CoA reductase pathway by GT3, are likely contributors to the profound changes at the cell membrane imposed by the upregulation of ceramide synthesis. Furthermore, ASM activity also leads to cholesterol displacement from the plasma membrane [35‐38] supporting a shift in membrane composition and structure, and subsequently altering signaling function. The resulting ceramide-enriched membrane domains [31], may then favor the presence and clustering of death receptors [28, 39], in agreement with our observation of DR5 localization and clustering at the plasma membrane. These events are required for the formation of death-inducing signaling complexes [40, 41] and are indicative of the activation of the extrinsic apoptotic (JNK) pathway [42], which lend support to our previous results on the upregulation and activation of c-JUN and caspase 8 by GT3 [11]. The role of ceramide in the activation of the JNK pathway and initiation of apoptosis has been demonstrated in several studies [43, 44]. Additionally, DR5 is a component of the TRAIL apoptotic signaling, important for its selective toxicity towards tumor cells [45] and its relation to resistance and sensitization to chemotherapy agents [46]. Our results suggest that GT3 cytotoxic effects carried out via various signaling molecules, are facilitated by the upregulation of expression levels of enzymes involved in the synthesis of ceramide and the subsequent shift in cellular membrane composition and function. However, this set of finely orchestrated events that may obviate the aberrant survival mechanisms present in pancreatic cancer cells, would not be functional without the upregulation of ARV-1 [47] and downregulation of CERT. ARV-1 is involved in sterol transport [48] as well as sphingolipid [49] and glycosylphosphatidyl- inositol [29] biosynthesis. Thus suggesting that GT3 may exert an effect on multiple targets related to control of transcription and molecular synthesis, as well as lipid trafficking and distribution. Loss of membrane cholesterol stimulate sphingolipid transport via CERT [36], and changes in membrane composition by the increase in ceramide content, cause displacement of cholesterol [9]. However, GT3 effectively downregulates CERT, thus causing inhibition of sphingomyelin synthesis in the Golgi from ceramide by diminishing the transport of newly produced ceramide from the ER [50, 51]. Additionally, upregulation of sterol transporter ARV-1, and ceramide synthesis enzymes cause redistribution and upregulation of ceramide at the cellular membrane strongly promoting a profound change in membrane composition and function. Thus, GT3 displays remarkable efficacy in inhibiting the survival mechanisms present in pancreatic cancer cells by its direct involvement in multiple pro-apoptotic events. Specifically, the inhibition of the MAPK and PI 3K/AKT pathways affecting proliferation and survival, the activation of GSK3β and Foxo3 translocation to the nucleus causing G1-phase cell cycle arrest, and activation of the JNK apoptotic pathway via upregulation of and clustering of DR5 at the cell membrane, with subsequent phosphorylation of c-JUN and activation of caspases (10). Additionally, the increase of synthesis of ceramides, specifically, C:16 ceramide a pro-apoptotic molecule. Since most of the newly synthesized ceramides in the ER are used in the Golgi for the synthesis of sphingomyelin and glycosphingolipids [52], the inhibition of CERT [50] causes accumulation of ceramides at the ER, favoring cellular stress and apoptosis. Previous studies have shown increased cytotoxic effects by upregulation of ceramide synthesis in cancer cells.

However, tumor cells treated with drugs that target one or two of the pathways, are likely to fail. Eight approaches to raise ceramide levels have been identified, and many studies suggest that a multiple approach is best to attain a successful intervention [53]. This study has shown that GT3 effectively addresses four of these: stimulation of ceramide synthesis by the de novo pathway and sphingomyelin hydrolysis, inhibition of glycoceramide and sphingomyelin synthesis from ceramide. The latter is due to the inhibition of CERT and results in the downregulation of the MDR1 [54] suggesting the involvement of ceramide with the multidrug resistance pathway. The mechanism of action of several current chemotherapeutic agents include the upregulation of cellular levels of ceramide. Treatment with daunorubicin activate the de novo ceramide synthesis pathway [55], and may like cisplatin, stimulate sphingomyelin hydrolysis (32, 33).

Ceramide is a recognized target in oncotherapy. However, in order to obviate the effects of abnormal signaling pathways that result in various mechanisms of resistance in cancer cells, it has been determined that activation of several pathways conducive to raising ceramide levels is necessary for successful activation of apoptosis. The concerted effects on ceramide synthesis, localization and transport by GT3 and its effects on membrane receptors may provide important insight into the targeted pathways necessary to lead pancreatic cancer cells to initiate the apoptotic process and guide the design and development of new therapies for the treatment of pancreatic cancer.