Uncoupling Ceramide Glycosylation by Transfection of Glucosylceramide Synthase Antisense Reverses Adriamycin Resistance*

Previous work from our laboratory demonstrated that increased competence to glycosylate ceramide conferred adriamycin resistance in MCF-7 breast cancer cells (Liu, Y. Y., Han, T. Y., Giuliano, A. E., and M. C. Cabot. (1999) J. Biol. Chem. 274, 1140–1146). This was achieved by cellular transfection with glucosylceramide synthase (GCS), the enzyme that converts ceramide to glucosylceramide. With this, we hypothesized that a decrease in cellular ceramide glycosylation would result in heightened drug sensitivity and reverse adriamycin resistance. To down-regulate ceramide glycosylation potential, we transfected adriamycin-resistant breast cancer cells (MCF-7-AdrR) with GCS antisense (asGCS), using a pcDNA 3.1/his A vector and developed a new cell line, MCF-7-AdrR/asGCS. Reverse transcription-polymerase chain reaction assay and Western blot analysis revealed marked decreases in both GCS mRNA and protein in MCF-7-AdrR/asGCS cells compared with the MCF-7-AdrR parental cells. MCF-7-AdrR/asGCS cells exhibited 30% less GCS activity by in vitro enzyme assay (19.7 ± 1.1 versus 27.4 ± 2.3 pmol GC/h/μg protein,p < 0.001) and were 28-fold more sensitive to adriamycin (EC50, 0.44 ± 0.01 versus12.4 ± 0.7 μm, p < 0.0001). GCS antisense transfected cells were also 2.4-fold more sensitive to C6-ceramide compared with parental cells (EC50= 4.0 ± 0.03 versus 9.6 ± 0.5 μm,p < 0.0005). Under adriamycin stress, GCS antisense transfected cells compared with parental cells displayed time- and dose-dependent increases in endogenous ceramide and dramatically higher levels of apoptotic effector, caspase-3. Western blotting showed that adriamycin sensitivity, introduced by asGCS gene transfection, was independent of P-glycoprotein and Bcl-2 expression. In summary, this work shows that transfection of GCS antisense tempers the expression of native GCS and restores cell sensitivity to adriamycin. Therefore, limiting the potential to glycosylate ceramide, which is an apoptotic signal in chemotherapy and radiotherapy, provides a promising approach to combat drug resistance.

mide production is one cause of cellular resistance to apoptosis induced by either ionizing radiation or tumor necrosis factor-␣ and adriamycin (2)(3)(4)(5)(6)(7). Accumulation of glucosylceramide (GC), 1 a simple glycosylated form of ceramide, is a characteristic of some multidrug-resistant cancer cells and tumors derived from patients who are less responsive to chemotherapy (8,9). The study of GC metabolism, as a molecular determinant of the drug-resistant phenotype, has been a subject of recent attention. Modification of ceramide metabolism by blocking the glycosylation pathway has been shown to increase cancer cell sensitivity to cytotoxics (10 -12). Further, drug combinations that enhance ceramide generation and limit glycosylation have been shown to enhance kill in cancer cell models (11,12). Other work has shown that ceramide toxicity can be potentiated in experimental metastasis of murine Lewis lung carcinoma and human neuroepithelioma cells by inclusion of a glucosylceramide synthase inhibitor (13,14). These findings assign biological significance to ceramide metabolism as it relates to circumvention of resistance to antineoplastic agents.
The increased capacity for ceramide glycosylation in GCStransfected human breast cancer cells conferred resistance to adriamycin and to tumor necrosis factor-␣ (7,15). Both agents are known to activate ceramide generation and potentiate apoptosis (1,2,7,15). From this, we hypothesized that transfection of asGCS, to limit cellular ceramide glycosylation, would overcome adriamycin resistance. By introducing asGCS to modulate GCS activity in adriamycin-resistant human breast cancer cells, we successfully decreased native GCS expression and restored cellular sensitivity to adriamycin and to C 6 -ceramide. The present study shows further that ceramide generation is a major factor in the cytotoxicity of adriamycin and suggests that asGCS would be a novel force to overcome adriamycin resistance.
Giemsa staining was performed as described (17). Cells were seeded in 60-mm dishes (10 5 cells/dish) in 10% FBS RPMI 1640 medium and grown for 2 days at 37°C. After rinsing with PBS, cells were fixed with 50% methanol/PBS, followed by methanol, and stained with KaryoMAX Giemsa stain stock solution (Life Technologies, Inc.). Following washing with deionized water, cells were photomicrographed. The population doubling time of each cell line was measured. Briefly, cells were seeded in 24-well plates (10 4 cells/well) in 10% FBS RPMI 1640 medium and grown for 24-, 48-, 72-, and 96-h periods. After rinsing with PBS, cells were dispersed with trypsin/EDTA, suspended in medium, and counted by hemocytometer. pcDNA 3.1/his A-asGCS and pcDNA 3.1/his-GCS Expression Vectors and Transfection-pCG-2, a Bluescript II KS containing GlcT-1(Ref. 18; terminology for GCS) in the EcoRI site, was kindly provided by Dr. Shinichi Ichikawa and Dr. Yoshio Hirabayashi (The Institute of Chemical and Physical Research, Saitama, Japan). The full-length cDNA of human GCS was subcloned into the EcoRI site in the pcDNA 3.1/His A with Xpress TM tag peptide (Invitrogen) in the upstream region. Xpress tag fuses at the N terminus of the cloned gene; therefore, GCS will be expressed as Xpress-GCS. The antisense and sense orientation of GCS cDNA was analyzed with Vector NTI 4.0 and doubly checked by restriction digestion. When MCF-7-AdrR cells reached 20% confluence, pcDNA 3.1-asGCS or pcDNA 3.1-GCS (10 g/ml, 100-mm dish) was introduced by co-precipitation with calcium phosphate (Mammalian Transfection Kit, Stratagene, La Jolla, CA). The transfected cells were selected in RPMI 1640 medium containing 10% FBS and 400 g/ml G418. Each G418-resistant clone, isolated utilizing cloning cylinders, was propagated and later screened by GCS enzyme assay. pcDNA 3.1/his A plasmid, without GCS DNA, was used in control transfection.
Glucosylceramide Synthase Assay-To determine the levels of GCS in the G418-resistant clones, a modified radioenzymatic assay was utilized (7,19). Cells were homogenized by sonication in lysis buffer (50 mM Tris-HCl, pH 7.4, 1.0 g/ml leupeptin, 10 g/ml aprotinin, 25 M phenylmethylsulfonyl fluoride). Microsomes were isolated by centrifugation (129,000 ϫ g, 60 min). The enzyme assay, containing 50 g of microsomal protein, in a final volume of 0.2 ml, was performed in a shaking water bath at 37°C for 60 min. The reaction contained liposomal substrate composed of C 6 -ceramide (1.0 mM), phosphatidylcholine (3.6 mM), and brain sulfatides (0.9 mM). Other reaction components included sodium phosphate buffer (0.1 M), pH 7.8, EDTA (2.0 mM), MgCl 2 (10 mM), dithiothreitol (1.0 mM), ␤-NAD (2.0 mM), and [ 3 H]UDPglucose (0.5 mM). Radiolabeled and unlabeled UDP-glucose were diluted to achieve the desired radiospecific activity (4,700 dpm/nmol). To terminate the reaction, tubes were placed on ice, and 0.5 ml of isopropanol and 0.4 ml of Na 2 SO 4 were added. After brief vortex mixing, 3 ml of t-butyl methyl ether was added, and the tubes were mixed for 30 s. After centrifugation, 0.5 ml of upper phase, which contained GC, was withdrawn and mixed with 4.5 ml of EcoLume for analysis of radioactivity by liquid scintillation spectroscopy.
RNA Analysis-Cellular mRNA was purified using a mRNA isolation kit (Roche Molecular Biochemicals). Equal amounts of mRNA (5.0 ng) were used for RT-PCR. Under upstream primer (5Ј-CCTTTCCTCTC-CCCACCTTCCTCT-3Ј) and downstream primer conditions (5Ј-GGTT-TCAGAAGAGAGACACCTGGG-3Ј), a 302-base pair fragment in the 5Ј-terminal region of the GCS gene was produced using the ProSTAR HF single-tube RT-PCR system (High Fidelity, Stratagene) in a thermocycler (Mastercycler Gradient, Eppendorf). mRNAs were reverse transcribed using Moloney murine leukemia virus reverse transcriptase at 42°C for 15 min. DNA was amplified with TaqPlus Precision DNA polymerase in a 40-cycle PCR reaction, using the following conditions: denaturation at 95°C for 30 s, annealing at 60°C for 30 s, and elongation at 68°C for 120 s. RT-PCR products were analyzed by 1% agarose gel electrophoresis stained with ethidium bromide. ␤-Actin (Life Technologies, Inc.) was used as control for even loading.
Cytotoxicity Assay-Assays were performed as described previously (7,11). Briefly, cells were seeded in 96-well plates (2,000 cells/well) in 0.1 ml RPMI 1640 medium containing 10% FBS and cultured at 37°C for 24 h before addition of drug. Drugs were added in FBS-free medium (0.1 ml), and cells were cultured at 37°C for the indicated periods. Drug cytotoxicity was determined using the Promega 96 Aqueous cell proliferation assay kit (Promega, Madison, WI). Absorbance at 490 nm was recorded using a Microplate Fluorescent Reader, model FL600 (Bio-Tek, Winooski, VT).
Analysis of Ceramide-Analysis was performed as described previously (7,8). Cells were seeded in 6-well plates (60,000 cells/well) in 10% FBS RPMI 1640 medium. After 24 h, cells were shifted to 5% FBS medium with or without adriamycin and grown for the indicated times. Cellular lipids were radiolabeled by adding [ 3 H]palmitic acid (2.5 Ci/ml culture medium) for 24 h. After removal of medium, cells were rinsed twice with PBS (pH 7.4), and total lipids were extracted as described (8). The resulting organic lower phase was withdrawn and evaporated under a stream of nitrogen. Lipids were resuspended in 100 l of chloroform/methanol (1:1, v/v), and aliquots were applied to TLC plates. Ceramide was resolved using a solvent system containing chloroform/acetic acid (90:10, v/v). Commercial lipid standards were cochromatographed. After development, lipids were visualized by iodine vapor staining, and the ceramide area was scraped into 0.5 ml of water. EcoLume counting fluid (4.5 ml) was added, the samples were mixed, and radioactivity was quantitated by liquid scintillation spectrometry.
Caspase-3 Assay-Caspase-3 activity was assayed by DEVD-AFC cleavage, using the ApoAlert Caspase-3 assay kit (CLONTECH, Palo Alto, CA). The assay was performed as described previously (15). Cells were seeded in 100-mm dishes (500,000 cells/dish) in 10% FBS RPMI 1640 medium. After 24 h, cells were shifted to 5% FBS RPMI 1640 medium without or with adriamycin and grown for 24 and 48 h. Following harvest, cells (10 6 /vial) were lysed on ice for 10 min with 50 l of lysis buffer, and cell debris was removed by centrifugation at 4°C at 10,000 ϫ g for 5 min. The soluble fraction was incubated with 50 M conjugated substrate DEVD-AFC in a 100-l reaction volume at 37°C for 60 min. The free AFC fluoresce was measured at excitation 400 nm and emission 505 nm using a FL600 Microplate Fluorescence Reader. The caspase-3 inhibitor, acetyl-Asp-Glu-Val-Asp-aldehyde, was used to exclude nonspecific background in the enzymatic reaction.
Statistics-All data represent the means Ϯ S.D. Experiments were repeated two or three times. Student's t test was used to compare mean values.

RESULTS
Expression of GCS Antisense-The structure of pcDNA 3.1/ his A-asGCS is shown in Fig. 1A. The GCS antisense was cloned into the EcoRI site, just downstream from the anti-Xpress tag sequence in pcDNA 3.1/his A. This plasmid was introduced into MCF-7-AdrR cells by calcium phosphate coprecipitation. G418 was used to select transfectants. We found that the number of G418-resistant clones in MCF-7-AdrR as-GCS transfected cells was much lower than in MCF-7-AdrR cells transfected with pcDNA3.1/his A vector (54/10 6 versus 251/10 6 ). G418-resistant clones were further selected by meas-uring GCS activity using the cell-free radioenzymatic assay. In all, fifty-four G418-resistant clones of MCF-7-AdrR asGCStransfected cells were obtained, and we identified one clone that exhibited a stable 30% decrease in GCS activity (Fig. 1B). Compared with 27.4 Ϯ 2.3 pmol of GC synthesized by MCF-7-AdrR parental cells, GCS activity in MCF-7-AdrR/asGCS was decreased to 19.7 Ϯ 1.1 pmol of GC (Fig. 1B, p Ͻ 0.001). There were no differences in GCS activities between the pcDNA 3.1/ his A vector-transfected cells and parental MCF-7-AdrR cells (Fig. 1B).
The asGCS-transfected and parental MCF-7-AdrR cells were stained with Giemsa. Representative photomicrographs are shown in Fig. 1C. MCF-7-AdrR/asGCS cells, including nuclei, are flatter and larger than the dome-shaped, more stellate MCF-7-AdrR cells. The asGCS cell line is also more cuboidal with less dense cytoplasm. The population doubling times for both cell lines were similar, 32 and 30 h for MCF-7-AdrR/ asGCS and MCF-7-AdrR cells, respectively.
Consistent with diminished GCS activity, GCS mRNA and GCS protein were reduced in MCF-7-AdrR/asGCS cells, compared with MCF-7-AdrR cells. Total mRNA was isolated from both cell lines and reverse transcribed and amplified through RT-PCR. A representative RT-PCR gel electropherograph is shown in Fig. 2A. As with that revealed by densitometric scanning, the mRNA in MCF-7-AdrR/asGCS cells was reduced 3-fold compared with that in MCF-7-AdrR cells (25.4% versus 77.5% of ␤-actin). GCS protein in cell lysates was resolved by SDS-polyacrylamide gel electrophoresis and identified using GCS antiserum. Western blotting showed that the total amount of GCS protein in MCF-7-AdrR/asGCS cells decreased by 32% compared with MCF-7-AdrR parental cells (77,520 and 112,860 optical density units, respectively) (Fig. 2B, right and  center bands). However, MCF-7-AdrR cells that were transfected with pcDNA 3.1/his A-GCS expressed greater amounts of GCS (Fig. 2B, left band, AdrR/GCS). MCF-7-AdrR/GCS cells were developed by stable transfection of sense orientation pcDNA 3.1/his A-GCS vector in MCF-7-AdrR cells. This GCStransfected cell line displays 80% higher GCS activity than MCF-7-AdrR cells as measured by radioenzymatic assay. After transfection with pcDNA 3.1/his A-GCS vector, although the expressed GCS was fused with Xpress tag (-Asp-Leu-Tyr-Asp-Asp-Asp-Lys-), the upward shift in molecular mass (about 800 daltons) was undetectable by Western blot (Fig. 2B). To evaluate the expression of transfected GCS antisense gene, we employed a Xpress antibody to detect the production of Xpress-GCS fused protein (Fig. 1A). We did not find the GCS-Xpress tag in either MCF-7-AdrR or MCF-7-AdrR/asGCS cells (Fig.  2C). However, the tag protein was highly expressed in MCF-7-AdrR GCS transfected cells (Fig. 2C, center band). In MCF-7-AdrR/asGCS cells, what appears to be the Xpress-asGCS protein (Fig. 2C, faint band) had a higher molecular mass compared with Xpress-GCS protein of MCF-7-AdrR/GCS and was present at only 15% the level of the latter (Fig. 2C, center band).
Ceramide Generation and Caspase-3 Activity under Adriamycin Stress-To further elucidate the dynamics of ceramide metabolism in drug sensitivity, we measured ceramide generation in the two cell lines. We found that adriamycin exposure dramatically elevated ceramide levels in GCS antisense-transfected cells. As shown in Fig. 4, adriamycin treatment increased the levels of ceramide in MCF-7-AdrR/asGCS cells in a time-and dose-dependent manner. At 24 and 48 h post-treatment, ceramide levels in MCF-7-AdrR/asGCS cells increased 200 and 250%, respectively (Fig. 4A). In sharp contrast, adriamycin treatment did not greatly modify ceramide levels in MCF-7-AdrR cells, which at 48 h increased only 16% above control. The result of increasing adriamycin dose on ceramide metabolism in the cell lines is shown in Fig. 4B. Adriamycin at 0.5, 1.0, and 2.5 M enhanced ceramide levels by 181, 188, and 246%, respectively, in MCF-7-AdrR/asGCS cells (Fig. 1B), whereas MCF-7-AdrR cells displayed minimal response over the same dose range.
In mammalian cells, ceramide induces apoptosis directly through effector caspases, such as caspase-3 (21,22). To identify whether an alteration in ceramide metabolism in asGCS cells is related to adriamycin sensitivity via signal cascades, we analyzed caspase-3 activity in the parental and transfected cell lines. The data demonstrate that increased effector caspase-3 activity is consistent with changes in ceramide metabolism. At 10 M adriamycin, the EC 50 in MCF-7-AdrR cells, caspase-3 activity in MCF-7-AdrR/asGCS increased 290 and 980% over control, at 24 and 48 h, respectively (Fig. 5). In contrast, adriamycin treatment increased caspase-3 by 160% in MCF-7-AdrR cells, albeit only at 48 h (Fig. 5). In summary, caspase-3 activity in the GCS antisense-transfected cells was 3-and 6-fold greater in response to adriamycin treatment than observed in parental cells (p Ͻ 0.0001). This suggests that impaired GCS activity permits cells to maintain high levels of ceramide under adriamycin stress, activating caspase-3 for progression of programmed cell death.
After transfection with pcDNA 3.1/his A-asGCS plasmid, we found that MCF-7-AdrR/asGCS cells expressed lower levels of GCS, based upon both mRNA and protein (Fig. 2). GCS enzy- matic activity was also found to be lower in MCF-7-AdrR/ asGCS cells (Fig. 1B). Because of markedly decreased expression of Xpress-asGCS tag (Western blot, Fig. 2C), it is likely that binding of asGCS mRNA to native GCS mRNA blocks GCS translation and diminishes GCS protein in the antisense transfected cells. It is noteworthy that the EC 50 for adriamycin was reduced 28-fold (Fig. 3), whereas in the cell-free enzyme assays GCS activity was reduced by only 30% in MCF-7-AdrR/asGCS cells (Fig. 1B). Similarly, in previous work, we have shown that GCS transfection by an inducible expression system conferred adriamycin resistance in MCF-7 cells (7). In MCF-7-GCStransfected cells, GCS activity was enhanced 4-fold, and the EC 50 of adriamycin increased 11-fold compared with MCF-7 cells (7). Other factors including the existence of GCS isoforms, substrate specificities, and enzyme compartmentalization may also play a role in GCS effects on adriamycin sensitivity. For example, GCS catalyzes ceramide glycosylation, the first step in the biosynthesis of glycosphingolipids (28). A recent GCS knockout study showed that embryonic lethality was the consequence of homozygosity, revealing a vital role for GCS during development and differentiation in mice (29). In present study, G418 survival of the asGCS-transfected clones was minimal compared with survival of the asGCS-free plasmid transfectants. This implies that GCS antisense blocks ceramide glycosylation that is essential for cell development, and only the partially blocked clones are able to survive the selection conditions. In addition, molecular specificity of ceramide has been demonstrated, as some species, C 16 -ceramide for example, are more prevalent in apoptosis signaling (30). Cellular ceramide response to DNA damage has been shown to rely on mitochondrion-dependent caspases (31).
Ceramide can be generated by de novo biosynthesis and sphingomyelin degradation via the action of sphingomyelinases (1,32,33). Intracellular levels of ceramide are elevated by a variety of stimuli and/or agents that induce apoptosis, including Fas ligand engagement of CD95, ionizing radiation, ultraviolet radiation, chemotherapeutic drugs and genotoxic chemicals, and several cytokines (1-7, 15, 33-35). Ceramideinduced cellular death is one mechanism of adriamycin-induced toxicity (7,8,12,14). Cellular ceramide impacts a variety of signaling molecules and pathways (33). Of these various effects, ceramide induction of the stress-activated protein kinase cascade and inhibition of complex III activity in the mitochondrial respiratory chain have been linked to the induction of apoptosis (36 -38). Capspase-3, one of the effector caspases in the stress-activated protein kinase apoptotic signaling pathway, is activated by cell-permeable ceramide as well as endogenous ceramide generated in response to extracellular stimuli (15,39,40). In present study, adriamycin treatment increased cellular ceramide with activation of caspase-3 in the GCS antisense transfected cells but not in parental cells. Therefore, the diminished capacity for glycosylation promotes adriamycin-induced cytotoxicity via ceramide-linked activation of caspase-3.
P-glycoprotein, a well characterized drug resistance mechanism (41), is highly expressed in MCF-7-AdrR cells (18). In previous work on the conversion of cells toward drug resistance, increased expression of P-glycoprotein in MCF-7 cells transfected with GCS sense was not observed (7). Much in line, in the present study we did not observe decreased expression of P-glycoprotein in chemosensitive MCF-7-AdrR/asGCS cells (Fig. 6). This suggests that the reversal of adriamycin resistance conferred by asGCS is not related to P-glycoprotein. Bcl-2 in dephosphorylated form is a strong anti-apoptosis effector involved in ceramide-induced apoptosis signaling pathways (42)(43)(44). We did not find that increased Bcl-2 in GCS modulates MCF-7 cells (7), nor in this study was altered Bcl-2 expression found in GCS antisense-transfected MCF-7-AdrR cells. These data reinforce the idea that up-regulation and down-regulation of GCS regulates adriamycin sensitivity by a mechanism divorced from Bcl-2.
In keeping with our previous report (7), the GCS gene knockout data presented here further demonstrate that GCS is one cause of adriamycin resistance. This positions antisense technology as a promising tool for reversal of certain forms of chemotherapy resistance.
FIG. 5. Caspase-3 activity under adriamycin stress. Cells were treated without or with adriamycin (10 M) for 24 and 48 h. After harvest, the soluble fraction obtained after cell lysis (10 6 cell/tube) was incubated with DEVD-AFC substrate at 37°C for 60 min as detailed under "Experimental Procedures." The fluorescence of cleaved AFC was measured at 505 nm. *, p Ͻ 0.0001, compared with MCF-7-AdrR cells treated with adriamycin for each corresponding treatment period.
FIG. 6. P-glycoprotein and Bcl-2 expression in MCF-7-AdrR and MCF-7-AdrR/as GCS cells. Detergent-soluble cellular protein was isolated from the respective cell lines and subjected to SDS-polyacrylamide gel electrophoresis (50 g/lane). Protein was transferred to nitrocellulose, and the immunoblot was incubated with the specified antibody. A, P-glycoprotein Western blots. C219 monoclonal antibody was used to recognize P-glycoprotein. B, Bcl-2 Western blots. Ab-1 monoclonal antibody was utilized to blot Bcl-2 protein. MCF-7 cells were used as a positive control for Bcl-2.