Expression of Ceramide Synthase 6 Transcriptionally Activates Acid Ceramidase in a c-Jun N-terminal Kinase (JNK)-dependent Manner*

Background: Ceramide is important for cellular signaling. Results: Increasing the expression of ceramide synthase 6 (CerS6) results in transcriptional activation of acid ceramidase independent of catalytic CerS6 activity. Conclusion: Modulation of a single member of the ceramide synthase family impacts on sphingolipid composition and ceramide metabolizing enzymes. Significance: Understanding how CerS impacts gene expression and signaling is important for the development of novel therapeutic approaches. A family of six ceramide synthases with distinct but overlapping substrate specificities is responsible for generation of ceramides with acyl chains ranging from ∼14–26 carbons. Ceramide synthase 6 (CerS6) preferentially generates C14- and C16-ceramides, and we have previously shown that down-regulation of this enzyme decreases apoptotic susceptibility. In this study, we further evaluated how increased CerS6 expression impacts sphingolipid composition and metabolism. Overexpression of CerS6 in HT29 colon cancer cells resulted in increased apoptotic susceptibility and preferential generation of C16-ceramide, which occurred at the expense of very long chain, saturated ceramides. These changes were also reflected in sphingomyelin composition. HT-CerS6 cells had increased intracellular levels of sphingosine, which is generated by ceramidases upon hydrolysis of ceramide. qRT-PCR analysis revealed that only expression of acid ceramidase (ASAH1) was increased. The increase in acid ceramidase was confirmed by expression and activity analyses. Pharmacological inhibition of JNK (SP600125) or curcumin reduced transcriptional up-regulation of acid ceramidase. Using an acid ceramidase promoter driven luciferase reporter plasmid, we demonstrated that CerS1 has no effect on transcriptional activation of acid ceramidase and that CerS2 slightly but significantly decreased the luciferase signal. Similar to CerS6, overexpression of CerS3–5 resulted in an ∼2-fold increase in luciferase reporter gene activity. Exogenous ceramide failed to induce reporter activity, while a CerS inhibitor and a catalytically inactive mutant of CerS6 failed to reduce it. Taken together, these results suggest that increased expression of CerS6 can mediate transcriptional activation of acid ceramidase in a JNK-dependent manner that is independent of CerS6 activity.

A family of six ceramide synthases with distinct but overlapping substrate specificities is responsible for generation of ceramides with acyl chains ranging from ϳ14 -26 carbons. Ceramide synthase 6 (CerS6) preferentially generates C 14 -and C 16 -ceramides, and we have previously shown that down-regulation of this enzyme decreases apoptotic susceptibility. In this study, we further evaluated how increased CerS6 expression impacts sphingolipid composition and metabolism. Overexpression of CerS6 in HT29 colon cancer cells resulted in increased apoptotic susceptibility and preferential generation of C 16 -ceramide, which occurred at the expense of very long chain, saturated ceramides. These changes were also reflected in sphingomyelin composition. HT-CerS6 cells had increased intracellular levels of sphingosine, which is generated by ceramidases upon hydrolysis of ceramide. qRT-PCR analysis revealed that only expression of acid ceramidase (ASAH1) was increased. The increase in acid ceramidase was confirmed by expression and activity analyses. Pharmacological inhibition of JNK (SP600125) or curcumin reduced transcriptional up-regulation of acid ceramidase. Using an acid ceramidase promoter driven luciferase reporter plasmid, we demonstrated that CerS1 has no effect on transcriptional activation of acid ceramidase and that CerS2 slightly but significantly decreased the luciferase signal. Similar to CerS6, overexpression of CerS3-5 resulted in an ϳ2-fold increase in luciferase reporter gene activity. Exogenous ceramide failed to induce reporter activity, while a CerS inhibitor and a catalytically inactive mutant of CerS6 failed to reduce it. Taken together, these results suggest that increased expression of CerS6 can mediate transcriptional activation of acid ceramidase in a JNK-dependent manner that is independent of CerS6 activity.
Sphingolipids are important signaling molecules and can significantly impact on cellular function. Ceramide, the central molecule in sphingolipid biosynthesis, can be generated through the action of ceramide synthases (CerS) 3 in the de novo or the salvage pathway (1). CerS comprise a family of six enzymes that preferentially conjugate a fatty acyl-CoA moiety to the sphingoid base, thereby generating ceramides with fatty acid side chains ranging from 14 to 26 carbons. Recent studies have demonstrated associations between specific ceramide species and cellular responses (2).
We have previously shown that RNAi-mediated down-regulation of CerS6 results in a specific decrease in C 16 -ceramide and increased resistance to the death receptor ligand TRAIL whereas overexpression of CerS6 increased susceptibility to TRAIL (3). CerS6 has also been implicated to contribute to apoptosis induced by 17AAG (4) and MDA-7 (5), the combination of sorafenib and vorinostat (6,7), celecoxib-mediated chemoprevention of colon cancer (8), and efficacy of photodynamic therapy (9). These studies suggest that CerS6 activity contributes to the efficacy of existing therapies and might be a potential biomarker to predict responsiveness.
Evidence that CerS can have opposing and tissue-specific roles is also emerging. Thus while several studies have found links between CerS6/C 16 -ceramide and apoptosis (10), overexpression of CerS2, which generates C 24 -ceramides can promote proliferation and offer protection against radiation-induced cell death (11,12). Generation of CerS-deficient mice is revealing tissue-specific effects as well. For example, knock out of CerS2, which is expressed at similar levels in the mouse liver and kidney, results in liver abnormalities while kidney function remains normal (13,14). CerS6 also appears to have tissuespecific effects as decreased expression induces ER stress and apoptosis in head and neck squamous carcinoma cells but not in A549, MCF-7, or SW480 cells, which were derived from lung, breast, and colon cancer, respectively (3,15,16).
Altered ceramide distributions and/or changes in ceramide synthase expression are beginning to be associated with specific diseases (17). For example, elevated expression of CerS6 has recently been suggested to play a role in the onset of disease in chronic experimental autoimmune encephalomyelitis, a model of multiple sclerosis (18). CerS6 is highly expressed in the intestinal tract (19), and we have therefore focused our investigations on the role of CerS6 in colon cancer. In HCT-116 colon cancer cells overexpression of CerS6 induced spontaneous apoptosis (11). In our hands elevated expression of CerS6 in SW620 colon cancer cells did not result in spontaneous apoptosis but increased susceptibility to apoptotic stimuli (3). However, the consequence of CerS6 expression was not previously investigated in detail, and we hypothesized that due to the highly dynamic nature of sphingolipid metabolism (13,14,20,21), alterations in the expression of CerS6 may have impacts beyond increased generation of C 16 -ceramide. In support of this hypothesis, the current study reveals a novel connection between ceramide synthases and acid ceramidase.
The adenovirus expressing CerS6 was generated using the AdEasy system (ATCC) (25). Briefly, the EcoRI and HindIII fragment from pCerS6-IRES-GFP was subcloned into the Shuttle vector, which was then recombined with the AdEasy vector in the Escherichia coli BJ5183 strain. Following screening by PacI digestion, DNA of positive recombinants was transiently transfected into HEK293A cells using Lipofectamine 2000 and CerS6 expression verified by Western blot. A positive recombinant was transfected into HEK293A cells, and cultures were observed for formation of viral plaques (25). Crude viral lysate was provided to Vector Biolabs (Philadelphia, PA, for amplification and determination of titer). The control adenovirus was also obtained from Vector Biolabs.
Flow Cytometry-Flow cytometry (LSRFortessa) and cell sorting (MoFlo) was performed in the MUSC flow cytometry shared resource.
Transfections and Transductions-HT29 transfectants expressing CerS6 or GFP only were generated by transfection of pCerS6-IRES-GFP (3) or pIRES-GFP plasmids using Lipofectamine 2000 (Invitrogen, Grand Island, NY) followed by selection and maintenance in 1.5 mg/ml neomycin (Fisher Scientific). Experiments with stably transfected mass clones were performed within 25 passages. HT29 cells expressing pGL3-ASAH1 were generated by co-transfection with a G418 resistance plasmid and selected as above. SW480 were transfected with pLKO-Tet-On shRNA-CerS6 followed by selection in 1 g/ml puromycin.
For adenoviral transductions, cells were plated overnight and the next day adenovirus added to the culture medium at the indicated concentration. Luciferase activity was measured 4 days post-infection.
Transient transfections of HEK293A cells were performed in 96-well plates using 200 ng DNA and 0.5 l Lipofectamine per well according to the manufacturer's instructions.
Viability and Luciferase Activity Assays-Viability was measured using the CellTiterBlue substrate and luciferase activity was determined using the Steady-Glo kit. Both kits were purchased from Promega (Madison, WI). Signals were quantified using a BMG Optima plate reader.
Sphingolipid Analysis and Metabolic Labeling-Sphingolipid analysis was performed as described previously (27). The cell pellets were stored at Ϫ80°C until processing for sphingolipid analysis by liquid chromatography/mass spectrometry (LC-MS/MS) in the MUSC Lipidomics facility (28). An aliquot of the lipid extract was used to carry out lipid phosphate estimation using Bligh Dyer extraction and a colorimetric assay (29). For sphingolipid metabolic labeling, cells were incubated with 1 M 17 C-sphingosine (Avanti Polar Lipids) for 30 min. The LC-MS/MS analysis was modified to detect only 17 C-sphingolipids (30).
In Vitro Acid Ceramidase Activity Assay-Cells were plated at a density of 1 ϫ 10 7 cells on 150-mm dishes and allowed to form a monolayer for 2 days. At harvest, cells were counted and lysed in an acidic buffer (50 mM sodium acetate, 5 mM magnesium chloride, 1 mM EDTA, and 0.5% Triton X-100, pH 4.5) to determine acid ceramidase activity as previously (26). The assay was carried out in duplicate using an equal amount of protein lysate (close to 200 g), and results displayed as pmol palmitate liberated per hour per mg protein.
Statistical Analysis-Differences in viability and lipid composition were determined in the unpaired Student's t test using the GraphPad software. To evaluate differences in fold-change of luciferase activity in transient transfections experiment in HEK293A cells, we used a hierarchical linear mixed effects regression model that was fit with the main effects of marker (CerS1, CerS2, etc.) and nested random effects for experiment and well over 10 cycles (10 min/cycle) of luciferase activity. Regression coefficients and their standard errors were used for making inferences regarding statistical significance at the ␣ ϭ 0.05 level. Fold-change and S.E. were based on model results.

Cells with Elevated CerS6 Expression Have Increased Susceptibility to TRAIL and 5-Fluorouracil Chemotherapy-HT29
colon carcinoma cells were stably transfected with pCerS6-IRES2-EGFP or pIRES2-GFP plasmids to generate HT-CerS6 and HT-GFP cells, respectively. Analysis of mRNA from G418resistant mass clones indicated that total CerS6 transcript lev-els increased ϳ3-fold in the HT-CerS6 transfectants compared with the HT-GFP cells (Fig. 1A). Levels of endogenous CerS6 mRNA were not significantly changed (Fig. 1A). These results suggest that the increase in total CerS6 mRNA resulted from expression of the transgene. Based on GFP expression, at least 20% of cells within mass clones expressed the transgene (Fig. 1B). An increase in CerS6 protein in the mass clone was observed when GFP-positive cells were analyzed by Western blot following sorting by flow cytometry (Fig. 1C). Consistent with our previous findings in SW620 cells, elevating CerS6 increased susceptibility of HT29 cells to the death ligand TRAIL (Fig. 1D). Sensitivity to 5-fluorouracil, a chemotherapeutic agent frequently used in colorectal patients, was also enhanced in HT-CerS6 cells compared with HT-GFP (Fig. 1D).
Up-regulation of CerS6 Expression Increases C 16 -ceramide Generation at the Expense of Very Long Chain Ceramides-Although cells expressing elevated CerS6 are more susceptible to cell death, the exact impact of CerS6 expression on sphingolipids in stably transfected cells has not previously been determined. Therefore, we next used a cell-based assay, in which 17 C-sphingosine serves as a metabolic label, to determine how increased expression of CerS6 impacts on the incorporation of sphingosine into ceramides. HT-GFP and HT-CerS6 cells were incubated with 1 M 17 C-sphingosine for 30 min followed by analysis of the 17 C-sphingolipid profile. Total 17 C-ceramide levels were similar between the GFP and the CerS6 transfectants ( Fig. 2A), indicating that the total cumulative activity of CerS, at least in the salvage pathway in which sphingosine is recycled into ceramides, is comparable between HT-GFP and HT-CerS6 cells. The majority of 17 C-sphingosine was incorporated into 17 C 16 -, 17 C 22 -, and 17 C 24 -ceramides ( Fig. 2B) but CerS6 expression influenced the distribution of ceramide species. Compared with HT-GFP, HT-CerS6 cells contained significantly more 17 C 16 -and 17 C 24:1 -ceramides and significantly less 17 C 22:0 -, 17 C 24:0 , and 17 C 26:0 -ceramides. These results suggest that increased generation of C 16 -ceramide occurs at the expense of very long chain (C 22 -C 26 ) saturated but not unsaturated ceramides.
Changes in ceramide composition observed in the metabolic labeling assay were also reflected at the steady state, i.e. the very long chain saturated ceramides C 24:0 -ceramide and C 26:0 -ceramide were significantly decreased while C 16 -ceramide and C 24:1 -ceramide were increased (Fig. 3A). Similar results were observed in SW620 cells upon adenoviral expression of the CerS6 transgene (data not shown), which suggests that CerS6mediated alterations in ceramide species distribution is not cell  line specific. Since ceramide can be further metabolized into complex sphingolipids, sphingomyelin (SM) composition was also analyzed. We found that expression of CerS6 increased the C 16 -SM content from 43% in HT-GFP cells to 51% in HT-CerS6 cells (Fig. 3B). This significant increase in C 16 -SM was accompanied by an almost 50% decrease in C 24:0 -SM. The increase in C 24:1 -Cer in HT-CerS6 cells corresponded to an increase in C 24:1 -SM (Fig. 3B). These results suggest that changes in ceramide composition as a consequence of increased CerS6 expression are also reflected in complex sphingolipids such as sphingomyelin.
HT-CerS6 Cells Have Elevated Acid Ceramidase Expression and Activity-In addition to being incorporated into complex sphingolipids, ceramide can also be hydrolyzed to sphingosine by ceramidases. We found that HT-CerS6 cells contained nearly twice as much sphingosine as HT-GFP cells (Fig. 4A). Five ceramidases, including ASAH1, ASAH2, and ACER1-3, have been identified. The gene threshold for ASAH2 and ACER1 was high (Ͼ30) indicating that the relative expression levels of these genes are low in HT29 cells. Of the remaining ceramidases, mRNA levels were increased only for acid ceramidase (ASAH1) (Fig. 4B). Western blot analysis of two colon cancer cells confirmed that ASAH1 expression is elevated when CerS6 is overexpressed (Fig. 4C). An in vitro assay confirmed that acid ceramidase activity is also higher in HT-CerS6 cells than in HT-GFP cells (Fig. 4D).
Sphingosine serves not only as substrate for ceramide synthases in the salvage pathway but can also be further metabolized to sphingosine-1-phosphate (S1P) through the action of sphingosine kinases. Steady state levels of intracellular S1P were below detection in our analysis. We therefore used 17 Csphingosine as the substrate for metabolic labeling and found that HT-CerS6 had a significantly reduced capacity to generate S1P compared with HT-GFP cells (Fig. 4E).
Our results suggested that increasing CerS6 results in elevated expression of acid ceramidase. To investigate if decreasing CerS6 reduces acid ceramidase expression we chose SW480 cells, which express higher levels of CerS6 than the isogenic SW620 cells used in overexpression studies (3). SW480 cells were transfected with an inducible shRNA against CerS6 and analysis for acid ceramidase performed in two individual clones. As shown in Fig. 4F, knockdown of CerS6 did not appear to decrease acid ceramidase expression.

Increased Expression of Acid Ceramidase in Response to CerS6 Expression Occurs via a JNK-AP1-dependent Mechanism and Is Important for Survival-We hypothesized that increased expression and activity of acid ceramidase occurs in response to
increased generation of C 16 -ceramide. Ceramide stress has been shown to activate the JNK pathway (31-33) and more recently it has been shown that radiation-induced ceramide stress and subsequent up-regulation of acid ceramidase occurs in an AP-1-dependent manner (34). Therefore, we treated HT-CerS6 cells with the JNK inhibitor SP600125 or curcumin, a natural compound with anti-tumor activity that has been shown to directly interfere with DNA binding at the AP-1 transcription factor (35,36), and examined the impact on phosphorylation of c-jun and acid ceramidase expression. Compared with HT-GFP cells, HT-CerS6 cells had increased phosphorylation of c-jun on Ser-63 (Fig. 5A). Treatment with the JNK inhibitor SP600125 decreased phosphorylation of c-jun to undetectable levels. In contrast to SP600125, curcumin did not decrease phosphorylation of c-jun, which is consistent with its function of inhibiting AP-1 DNA binding downstream of c-jun. Treatments with either SP600125 or curcumin also diminished levels of acid ceramidase (Fig. 5A). Morphological assessment of treated cultures suggested that HT-CerS6 cells were beginning to die when exposed to inhibitors of the JNK pathway. Analysis of PARP, a marker of apoptosis, confirmed that this protein was cleaved when HT-CerS6 cells were treated with SP600125 or curcumin.
To more directly quantify the effect of SP600125 and curcumin on viability, HT-GFP and HT-CerS6 cells were cultured in the absence or presence of these agents. As shown in Fig. 5B, HT-CerS6 cells were significantly more susceptible to SP600125 than control cells. Viability in the presence of curcumin was also reduced but did not reach statistical significance.
Taken together, the results suggested that inhibition of acid ceramidase expression by either SP600125 or curcumin may be responsible for induction of PARP cleavage and preferentially decreased viability of HT-CerS6 cells. To directly test how inhibition of acid ceramidase impacts on viability, HT-GFP and HT-CerS6 cells were treated with the acid ceramidase inhibitor LCL-521 (26). As shown in Fig. 5C, HT-CerS6 cells were more susceptible to inhibition of acid ceramidase than HT-GFP cells.
Transcriptional Activation of Acid Ceramidase by Ceramide Synthases-The increase in acid ceramidase (ASAH1) mRNA suggested that CerS6 may transcriptionally activate expression of this enzyme. To further explore this possibility, we stably transfected HT29 cells with pGL3-ASAH1, a plasmid in which luciferase expression is under control of the full-length acid ceramidase promoter (23). Stable transfectants were then transduced with an adenovirus-expressing CerS6 (AdCerS6). As shown in Fig. 6A, the control virus did not alter ASAH1-driven luciferase activity whereas AdCerS6 increased luminescence 2-3-fold. Next, we asked whether the ability to transcriptionally activate acid ceramidase expression is unique to CerS6. HEK293A cells were co-transfected with the acid ceramidase reporter construct (pGL3-ASAH1) and plasmids expressing CerS1-6. Expression of CerS3-6 transcriptionally activated the ASAH1 promoter as evidenced by a significant ϳ2-fold increase in luciferase reporter gene activity (Fig. 6B). Expression of CerS1 did not significantly alter ASAH1 reporter activity while expression of CerS2 slightly but significantly decreased luciferase activity. Since treatment with SP600125 or curcumin decreased acid ceramidase expression in HT-CerS6 cells, we next investigated how these inhibitors affect transcriptional activation of ASAH1. We verified based on GFP expression that neither inhibitor interfered with transfection efficiency. When CerS6 was expressed in the presence of SP600125, ASAH1 reporter activity was not significantly different from GFP-transfected cells (Fig. 6C). We found that similar to inhibition of JNK, treatment with 25 M curcumin prevented the CerS6-mediated transcriptional activation of acid ceramidase (Fig. 6C).
Next, we investigated whether the increase in ASAH1 transcription upon elevated CerS6 expression is a direct consequence of increased generation of C 16 -ceramide. HEK293A cells were transfected with pGL3-ASAH1 and after 20 h, C 16ceramide was exogenously added for 6 h. In the presence of 10 M C 16 -ceramide we did not observe an increase in pGL3-ASAH1 reporter activity (Fig. 6D). At higher concentrations of exogenous C 16 -ceramide luciferase reporter activity declined, and cells began to lose viability (Fig. 6, D and E). Similarly exogenous C 6 -ceramide, which is preferentially metabolized into C 16 -ceramide but was less toxic under our assay conditions failed to alter luciferase reporter gene activity (Fig. 6, D and E). Furthermore, the ceramide synthase inhibitor fumonison B1 did not prevent the CerS6-induced increase luciferase reporter activity, suggesting that enzymatic activity does not significantly contribute to increased acid ceramidase expression (Fig.  6F). Finally, to substantiate this observation, we utilized a CerS6 mutant in which the catalytic domain has been inactivated through a histone to alanine substitution at residue 212 (15). As shown in Fig. 6F, mutant H212A CerS6 retained the ability to significantly increase pGL3-ASAH1 reporter activity.

Discussion
Several groups including ours have shown that decreased expression of CerS6 results in a specific decrease in C 16 -ceramide (3,37). The current study was initiated to understand how overexpression of CerS6 impacts on sphingolipid composition and signaling. We show that cells with increased expression of CerS6 preferentially generate C 16 -ceramide, which is consistent with previously observed activity of the enzyme in vitro (1). Incorporation of sphingosine into ceramide was comparable between HT-GFP and HT-CerS6 cells, suggesting that the increase in C 16 -ceramide occurred at the expense of saturated very long chain ceramides (C 22:0 , C 24:0 , and C 26:0 ) (Fig. 2). Similar results have been observed in models that modulate other CerS family members. For example, in CerS2-deficient mice and in SMS-KCNR neuroblastoma cells treated with CerS2 RNAi, long chain ceramides such as C 16 -Cer compensated for the decrease in C 24 -and C 24:1 -ceramides (13,38). Mullen et al. also showed that down-regulation of individual CerS in MCF7 breast cancer cells can transcriptionally impact the expression of non-targeted CerS (37). The distribution of ceramide species was mirrored in sphingomyelin composition (Fig. 3), suggesting that incorporation of ceramides into complex sphingolipids occurs without preference for a specific ceramide species.
In addition to altered ceramide composition, HT29 cells expressing CerS6 also contained increased intracellular sphin-gosine, which suggested the possibility that CerS6 expression also increases ceramidase expression and activity. Our results indicate that expression of CerS6 can stimulate expression of acid ceramidase, resulting in increased mRNA, protein expression and activity of the enzyme (Fig. 4, A-C). The increase in acid ceramidase following CerS6 expression was not unique to HT29 cells and was also observed in SW620 colon cancer cells that overexpress CerS6 (previously described in (3)) as well as in PPC1 prostate cancer cells transduced with an adenovirus expressing CerS6 (34). These data suggest that the increase in acid ceramidase following CerS6 expression is not a cell line or tissue-specific response.
Interestingly, both HT29 and SW620 cells transfected with CerS6 express higher levels of acid ceramidase, yet are more susceptible to apoptotic stimuli (3,6). This is in contrast to prostate cancer cells in which elevated acid ceramidase expression has been associated with apoptosis resistance and relapse following radiation therapy (34,39). It was previously observed that an increase in acid ceramidase, which is a lysosomal enzyme, resulted in elevated lysosomal density and increased levels of autophagy (40). Autophagy has been demonstrated to serve as a cellular mechanism to limit ceramide levels in the liver (41) and also occurs as a consequence of increased sphingolipid synthesis in RAW264.7 cells following TLR4 stimulation (42). In contrast to these studies, we did not detect an increase in overall ceramide synthesis but rather a shift in composition (Fig. 2) and using lysotracker staining we were unable to detect any differences in lysosomal density. 4 One possibility for the discrepancy in apoptotic responsiveness between prostate cancer cells and our model system are differences in subsequent metabolism of sphingosine that is generated as a consequence of increased acid ceramidase expression. Sphingosine holds a unique position in sphingolipid metabolism in that it can be further metabolized to sphingosine-1-phosphate (S1P) by sphingosine kinases or serve as a substrate for ceramide synthases in the salvage pathway (10). Irradiation of prostate cancer cells increased the pro-apoptotic sphingolipids ceramide and sphingosine but also elevated sphingosine-1-phosphate (S1P), indicating that sphingosine was further metabolized by sphingosine kinases (34). In contrast, in HT29 colon cancer cells intracellular S1P generation is greatly diminished upon CerS6 expression (Fig. 4E). It is possible that ceramide synthases and sphingosine kinases compete for the sphingosine substrate, although this idea would need to be reconciled with subcellular localization of the enzymes involved. Ceramide synthases are primarily localized to the ER and have also been detected in mitochondria, while sphingosine kinases have been localized to cytosol/plasma membrane (SK1) or the nucleus (SK2) (43). Therefore, although sphingosine is a soluble product of sphingolipid metabolism, it is unlikely that ceramide synthases and sphingosine kinases directly compete for the substrate in the same compartment. While details remain to be investigated, the increased apoptotic susceptibility of HT-CerS6 or SW620-CerS6 cells, despite elevated acid ceramidase expression, may offer an explanation for recent studies in ovarian and 4 C. Voelkel-Johnson, unpublished data.
We also observed that HT-CerS6 cells were more susceptible to treatment with the acid ceramidase inhibitor LCL-521 (Fig.  5C), which suggests higher levels of acid ceramidase maybe important for cells to maintain viability when CerS6 expression is elevated. However, as a consequence of elevated acid ceramidase activity, intracellular levels of sphingosine increase, which may result in heightened susceptibility to apoptotic signals either in a therapeutic setting or endogenously through the immune system. The dynamic nature of sphingolipids and the complexity of crosstalk between sphingolipid metabolic pathways, suggests it may be very difficult to utilize a single sphingolipid enzyme such as acid ceramidase as a biomarker for prognosis or therapy responsiveness.
Using an ASAH1-promoter driven luciferase reporter plasmid, we confirmed that CerS6 induced transcriptional up-regulation of acid ceramidase (Fig. 6). Experiments with pharmacological inhibitors suggest that transcriptional activation of acid ceramidase by CerS6 occurs in a JNK/AP-1-dependent manner (Figs. 5 and 6). These results are consistent with previous studies that show activation of the JNK pathway following ceramide stress as well as with ceramide/AP-1-dependent transcriptional activation of acid ceramidase following radiation therapy in prostate cancer cells (31)(32)(33)(34). JNK belongs to the larger group of mitogen-activated protein kinases and responds to a variety of signals including cytokines, radiation, heat shock, and autophagy (46,47). How CerS6 overexpression impacts on signaling pathways other than apoptosis largely remains to be determined.
Curcumin, which also inhibited CerS6-induced transcriptional induction of acid ceramidase, affects numerous signaling pathways. In addition to inhibiting AP-1 binding, it has recently been shown to stimulate ceramide synthase activity through increased CerS dimer formation (48). Although it is unclear whether CerS dimer formation can occur at the concentration of curcumin used in our study (25 M versus 50 M), these studies suggest an impact of curcumin on sphingolipid signaling may possibly occur through both down-regulation of acid ceramidase expression and increased CerS dimer formation.
To investigate if the ability to enhance acid ceramidase transcription is unique to CerS6, we extended our study to all members of the CerS family. CerS1 was the only CerS family member that did not alter acid ceramidase reporter activity, which was not completely unexpected, as CerS1 is phylogenetically the most distant family member and appears to have functions that are distinct from CerS6 (49). For example, CerS1 but not CerS6 has been shown to induce mitophagy (50). Transfection of the CerS2 plasmid slightly but significantly decreased ASAH1-promoter-driven reporter activity (Fig. 6B). Whether the opposing effects of CerS2 and CerS6 on transcriptional activation of acid ceramidase is a consequence of differential heterodimer composition remains to be investigated. CerS3, -4, and -5 shared the capacity of CerS6 to stimulate transcriptional activation of acid ceramidase. Somewhat surprisingly, there was no correlation between the ability to stimulate ASAH1-luc reporter activity and the fatty acid specificities of the CerS isoforms (19). For example, both CerS1 and CerS4 can generate C 18 -ceramide, yet transcriptional activation of ASAH1-luc was only observed with CerS4 and not CerS1. A similar discrepancy was observed between CerS2 and CerS3/4, which have overlapping abilities to generate very long chain ceramides (19).
To further investigate the requirement for ceramide synthase activity for induction of acid ceramidase transcription, we used three different approaches: exogenously added ceramide, the CerS inhibitor FB1, and a CerS6 mutant that lacks catalytic activity. Exogenous ceramide failed to induce ASAH1-luc reporter activity, while FB1 and the CerS6 H212A mutant failed to significantly reduce it. Taken together, these results suggest that transcriptional activation of acid ceramidase in our model is not mediated by ceramide itself and does not depend on the catalytic activity of CerS. Recently it has been demonstrated that ectopically expressed Bcl2L13 binds to CerS6 (51). Therefore, one possibility is that elevated expression of acid ceramidase in response to CerS6 overexpression occurs as a consequence of altered protein-protein interactions with non-CerS proteins. This hypothesis would also explain why knockdown of CerS6 did not decrease acid ceramidase expression in SW480 cells (Fig. 4F). Future studies to investigate CerS binding partners may elucidate the exact mechanism by which overexpression of CerS6 (or CerS3, 4, 5) mediates a transcriptional increase in acid ceramidase.

Conclusions
This study shows how altered expression of a single ceramide synthase has profound effects on the sphingolipid network, impacting both sphingolipid composition as well as acid ceramidase expression, which leads to alterations in signaling pathways and cell death susceptibility.
Furthermore, our results suggest that the catalytic activity of CerS6 is not required for the ability to transcriptionally activate acid ceramidase expression, thereby revealing a new level of complexity by which ceramide synthases can impact cellular responses.