Glucosylceramide Synthase and Its Functional Interaction with RTN-1C Regulate Chemotherapeutic-induced Apoptosis in Neuroepithelioma Cells¹

GCS⁴ catalyzes the first glycosylation step in the biosynthesis of GSLs by transferring the glucose from UDP-glucose to ceramide (1, 2). After its translocation to the Golgi lumen, GC can be additionally metabolized to higher GSLs, which are major constituents of the outer leaflet of the plasma membrane in eukaryotic cells. GSLs play an essential role in many biological processes, including development, cell death, tumor progression, and pathogen/host interaction (3, 4, 5). GCS is a type III integral protein, localized in the cis/medial Golgi, which has a single membrane spanning region near its NH₂ terminus, whereas most of the protein, including the catalytic site, faces the cytoplasm (2, 6).

Although previous studies have suggested different ways by which GCS is regulated (7, 8), it is not clear how the enzyme is modulated in the Golgi apparatus and consequently the molecular mechanism(s) at the basis of its pleiotropic effect on cellular functions.

Compelling evidence has been presented indicating that GCS is constitutively expressed in a variety of tissues (6) and that the synthesis of GSLs is vital for embryonic development (9). Furthermore, treatment of cells with different GCS inhibitors affects basic cellular functions, including growth, death, and adhesion (10).

Recent studies have demonstrated a direct correlation between the development of multidrug resistance and increased levels of GC (11, 12), hence, GCS has been suggested as a candidate target for cancer therapy. However, it remains controversial as to how the inhibition of GCS could represent a way to sensitize transformed cells to chemotherapeutic agents (13). In fact, in previous studies, we reported that the apoptotic response of GCS antisense clones to various p53-dependent anticancer drugs (doxorubicin, etoposide, and cisplatin) was not increased (14). We used the synthetic retinoid fenretinide to elucidate whether GCS might be involved in a p53-independent drug-induced apoptosis pathway. In fact, fenretinide has been shown to induce apoptosis in a variety of cancer cell lines (15, 16) through a p53-independent pathway involving the induction of ER-response gene GADD153 and the generation of reactive oxygen species (17).

We show here that GCS is involved in fenretinide-dependent apoptosis of cancer cells and its action is specifically modulated by the interaction with a member of the RTN family, RTN-1C. These results suggest that the p53-independent pathway of fenretinide-induced apoptosis is regulated by the GCS/RTN-1C complex at the ER/Golgi interface.

CHP-100 human neuroepithelioma were kindly donated by Gerry Melino (University of Rome “Tor Vergata,” Italy). C₆-NBD was from Molecular Probes (Eugene, OR). PDMP and the anti-β-tubulin mouse monoclonal antibody were from Sigma Chemical Company (St. Louis, MO). 4-HPR was from Janssen-Cilag Ltd. (Saunderton, United Kingdom) as described previously (17). The anti-GFP rabbit polyclonal antibody was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The anti-RFP rabbit polyclonal antibody was from Clontech (Palo Alto, CA). The anti-calnexin and anti-membrin antibodies were from Stressgene (Victoria, British Columbia). The anti-β-CopI rabbit polyclonal antibody was from Vinci-Biochem (Lausen, Switzerland). The chemiluminescence ECL detection system was from Amersham Corp. (Burck, United Kingdom). High-performance thin layer chromatography slice gel 60 plates were from Merck (Darmstadt, Germany).

Peptides with the sequence corresponding to amino acid 5–14 of human GCS or amino acid 6–20 of human RTN-1C were synthesized and used as antigens for immunization in rabbits. To generate specific antisera peptides were covalently linked to the carrier protein cBSA (Pierce, Rockford, IL). After 5 weekly inoculations with the immunogen the antisera fram rabbits were tested and then the antibodies were purified using a protein A-Sepharose CL-4B column (Amersham, Burck, United Kingdom).

The yeast two-hybrid screen was performed with yeast strain PJ69–4A MATa trp1–901 leu2–3, 112 ura3–52 his3–200 gal4Δ gal80Δ LYS2::GAL1-HIS3 GAL2-ADE2 mt2::GAL7-lacZ (18). The plasmid used as bait was constructed by fusion of GAL4BD to human GCS coding sequence into the EcoRI site of the vector p21.29 AmpR ColE1ORI, TRP1, CEN6, GAL4BD (amino acid 1–147). The human brain cDNA library was cloned into pACT2 vector AmpR ColE1ORI, LEU2, GAL4AD (amino acid 768–881; Clontech). The positive clones were selected on synthetic medium lacking tryptophan, leucine, adenine, and histidine. Healthy colonies that also displayed β-galactosidase activity were isolated and sequenced.

The full length of human GCS, RTN-1C, or RTN-3 were obtained from cDNA human brain library by PCR amplification using specific primers. Fragments were cloned into BglII/BamHI sites of pDsRed1-N1 or HindIII/BamHI sites of pEGFP-C1 vectors (Clontech, Palo Alto, CA), respectively. CHP-100 were grown in RPMI 1640 (Invitrogen, Carlsbad, CA) as described previously (14). CHP-100 cells (80% confluent) were transiently transfected by lipofection according to the manufacture’s specifications (DMRIE-C Transfection Reagent; Invitrogen, United Kingdom).

Cells were washed twice with PBS and scraped into RIPA assay buffer [150 mm NaCl, 1% NP40, 0.5% sodium deoxycholate, 0,1% SDS, 50 mm Tris-HCl (pH 7.5)] with freshly added protease inhibitors. After an incubation for 30 min on ice and a brief sonication, the lysate was centrifuged at 14,000 rpm for 10 min at 4°C to remove the insoluble cell debris. The anti-GCS antibody was coupled to Dynabeads protein G (Dynal, Oslo, Norway) using dimethylpimelimidate coupling according to the manufacture’s instructions. Equal amounts of cellular protein were incubated for 2 h with antibody-linked beads at 4°C with continuous rocking. After washing in PBS buffer, the beads were boiled in SDS buffer, and samples were resolved by 12% SDS-PAGE, transferred overnight at 25 mA onto nitrocellulose paper, and analyzed by Western blotting as described previously (19).

Cells were fixed in 4% paraformaldehyde for 20 min at room temperature and permeabilized with 0.1% Triton-1X-100 in PBS for 20 min at room temperature. Cells were then washed with PBS, and nonspecific binding was blocked by incubating the cells with 5% BSA in PBS for 30 min at room temperature. Indirect immunofluorescence was performed incubating the cells with anti-β-CopI, diluted 1:2000 in 1% BSA (blocking solution) for 1 h at room temperature. Washes were followed by incubation with the antirabbit antibody (Alexa-Fluor, Burlingame, CA, 568 diluted 1:1000 in blocking solution for 1 h at room temperature). Fluorescence was then evaluated with a confocal microscope (Nikon Instruments Spa, Eclipse TE200) equipped with EZ2000 software for PCM2000.

Cells were incubated with 5 μm C₆-NBD for 2 h, and total lipids were extracted from cells and culture medium according to the method of Bligh and Dyer (20) After partition, the chloroformic phase was collected and lipids resolved by High-performance thin layer chromatography in chloroform/methanol/water (65:25:4, vol/vol). GlcC₆-NBD was detected under UV illumination, extracted from the silica-gel in 3 ml of ethanol and its fluorescence measured by a Perkin-Elmer LS-5 luminescence spectrometer. Quantitation was carried out referring to a GlcC₆-NBD fluorescence calibration curve.

Cells were detached by trypsinization and centrifuged at 300 × g for 5 min; pellets were washed with PBS (pH 7.4), placed on ice, and fixed with ice-cold 70% ethanol for 1 h. After washing in PBS, pellets were incubated with 100 μg/ml RNAsi (Roche, Basel, Switzerland) for 20 min at room temperature and stained with propidium iodide (50 μg/ml) at 4°C for 30 min, before analysis by flow cytometry using a FACScan Flow Cytometer (Becton-Dickinson, Franklin Lakes, NJ) at 565 and 605 nm as described previously (21).

It has recently been shown that fenretinide-induced apoptosis of cancer cells requires the induction of the ER-stress response gene GADD153 (15, 16, 17, 22). It has also been proposed that ceramide metabolism may play a role in this pathway because fenretinide leads to the accumulation of high levels of ceramide from de novo synthesis (16). Considering the role of GCS in multidrug resistance, we investigated the possibility that the enzyme may be involved in fenretinide-induced apoptosis. We first analyzed the effect of GCS inhibition (by using GCS antisense cells) on the apoptotic response elicited by fenretinide in CHP-100 neuroepithelioma cells. Our CHP-100-derived GCS antisense cells display a reduction of GCS expression resulting in a decrease of its specific activity of ∼60% (14). Control cells (wild-type and CHP-Neo) and GCS antisense-expressing cells (CHP-AS) were treated with fenretinide for 24 h, and apoptosis was evaluated by flow cytometric analysis (Fig. 1,A). Interestingly, we detected a decrease of the apoptotic response to fenretinide in GCS antisense cells, paralleled by a drastic reduction of the expression of GADD153 (Fig. 1,B), which is known to mediate fenretinide-dependent apoptosis (17). To confirm the involvement of the enzyme activity in this pathway, CHP-100 cells were treated with fenretinide in the presence or absence of PDMP, a specific inhibitor of GCS (Fig. 1 C). As suggested by the antisense approach, the apoptotic response to the retinoid was also suppressed by the inhibition of GCS activity, thus demonstrating that the functional enzyme is essential for fenretinide-induced apoptosis. These results are particularly interesting considering that GCS is not involved in the p53-mediated apoptosis induced by chemotherapeutic drugs (14). In fact, fenretinide-induced apoptosis occurs via a p53-independent pathway and its effect is synergistic with that of DNA-damaging agents (15). Nevertheless the precise mechanism of action of fenretinide is presently not clear.

In the attempt to better understand the regulation of GCS, we searched for potential GCS interactors by the yeast two-hybrid approach using the full-length human GCS as bait to screen a human adult brain library. We analyzed 6 × 10⁶ library clones and, among the 160 clones that were able to grow on synthetic medium in the absence of histidine and adenine, 13 were positive for β-galactosidase staining. Sequence analysis of these 13 clones resulted in the identification of RTN family members as the only GCS-interacting proteins; 12 clones were identified as RTN-1C (four different clusters; Fig. 2, A and B) and 1 as RTN-3. RTNs are neuroendocrine-specific proteins encoded by genes that are known to produce different products by alternative splicing (23, 24). In particular, the RTN-1 gene encodes three proteins (RTN-1A, RTN-1B, and RTN-1C) with a conserved COOH-terminal region (∼70% identical with the other members of RTN family) containing two short hydrophobic domains that are predicted to be membrane spanning domains. In fact, RTN-1 is integral membrane proteins tightly associated with the ER membrane (Ref. 25; Fig. 2, C and D). All of the RTN clones identified contain the lumenal loop and the conserved COOH-terminal tail, localized in the cytoplasm; thus, it is likely that these two regions might be involved in the interaction between GCS and the RTN proteins.

To verify whether the GCS-RTN interaction, highlighted by the two-hybrid method, takes place also in mammalian cells, we have carried out coimmunoprecipitation experiments of the native RTN-1C and RTN-3 with GCS in CHP-100 cell extracts using an anti-GCS antibody. As shown in Fig. 3,A, the Western blot analysis of the immune complexes revealed that the endogenous RTN-1C coimmunoprecipitated with GCS; by contrast, interaction between GCS and RTN-3 could not be detected (data not shown). These results demonstrated that although RTN-3 and RTN-1C share high homology, the interaction with GCS is very specific and restricted to RTN-1C. To verify the specificity of this interaction, an unrelated integral membrane protein of the Golgi apparatus (Membrin; Ref. 26) and a resident ER transmembrane protein (Calnexin; Ref. 27) were also tested. Although, these two proteins were detected in the cell lysate, they were not present in the GCS/RTN-1C complexes (Fig. 3 A). These results confirm and extend the yeast two-hybrid findings indicating that RTN-1C specifically interacts with GCS in mammalian cells.

We additionally characterized the GCS/RTN-1C interaction by analyzing their intracellular localization. To this aim, NH₂-terminal GFP-tagged RTN-1C and RFP-tagged GCS expression plasmids were constructed and used to transiently cotransfect CHP-100 cells (Fig. 3,B). We firstly confirmed the interaction between tagged GCS and RTN-1C in CHP-100 transfected cells by coimmunoprecipitation (Fig. 3,C) and subsequently analyzed their intracellular localization by confocal microscopy (Fig. 3,D). Although RTN-1C is mainly detected in the ER (green fluorescence), its localization is also detected on the Golgi apparatus overlapping with the GCS (red fluorescence). The Golgi was identified by staining the cells with an anti-β-CopI antibody (28). These findings support the above described interaction between GCS and RTN-1C by demonstrating that the two proteins colocalize in vivo in the Golgi apparatus or in regions of contact (anchorage) between the ER and the Golgi itself (Fig. 3,D). On the basis of these results, we hypothesized that the GCS/RTN-1C interaction might play a role in the regulation of their respective functions. To verify this hypothesis CHP-100 cells were transiently transfected with RTN-1C expression plasmid and the levels of RTN-1C overexpression analyzed by Western blot analysis (Fig. 4,B). Transfected cells were then incubated with C₆-NBD, a cell permeable synthetic fluorescent substrate of GCS (29). In cells overexpressing RTN-1C, we detected a significant increase of GC levels compared with control cells (Fig. 4 A). By contrast, changes in the levels of sphingomyelin were not observed (data not shown), thus indicating that RTN-1C-induced increase of GC is very likely because of its direct modulation of GCS enzymatic activity. These results provide the first evidence for a GCS regulatory mechanism based on protein-protein interaction.

Considering the involvement of GCS in feneretinide-induced apoptosis previously demonstrated in this study, we next investigated the possibility that the GCS/RTN-1C interaction might play a role in the retinoid-induced apoptosis. We analyzed whether the RTN-1C-dependent increase of GCS activity may affect fenretinide-induced apoptosis. CHP-100 cells were transiently transfected with RTN-1C or RTN-3 expression plasmid (used as control) and treated with fenretinide for 24 h. As shown in Fig. 4, C and D, we detected a specific increase of the apoptotic response to fenretinide in RTN-1C-transfected cells as compared with controls. Interestingly, the RTN-1C effect on fenretinide is also suppressed by GCS inhibition (Fig. 4, D and E), thus additionally indicating the involvement of GCS. Taken together, these results indicate that GCS/RTN-1C interaction may mediate signals between Golgi and ER compartments, including the cellular response to apoptotic stimuli. In keeping with these findings Tagami et al. (30) identified RTN-1C as a Bcl-x_L-interactor and suggested a proapoptotic role for this protein.

Because apoptosis induction by ceramide requires its conversion to GSLs (31), the potentiation of fenretinide-induced apoptosis observed in RTN-1C-transfected cells may be attributable to the GCS-dependent conversion of preexisting ceramide into higher order GSLs or gangliosides. It has also been reported that ceramide levels increase in response to fenretinide treatment, as a consequence of de novo synthesis; however, pretreatment of cells with fumonisin B₁, an inhibitor of ceramide synthase, does not affect fenretinide-induced cytotoxicity (16). In keeping with these findings, fenretinide-induced apoptosis may be dependent on ceramide metabolism, in particular on GSLs and gangliosides synthesis mediated by GCS. It is worth mentioning that in the GCS-antisense CHP-100 cells used in this study, the expression of gangliosides of higher order than GM3 is markedly reduced (14), thus supporting the notion of a strict correlation between GCS activity, gangliosides biosynthesis, and cell death. It has been suggested that GD3 is the major GSL candidate to mediate apoptosis; thus, it is likely that the increased apoptotic response to fenretinide observed in RTN-1C-transfected cells might be mediated by an increase of GD3 levels (32, 33).

Finally, the data presented in this study demonstrate that GCS plays a key role in fenretinide-induced apoptosis, and this effect is influenced by a member of the RTN family, RTN-1C, which provides a functional link between Golgi and ER in this response. These results may form the basis to develop a targeted therapy for cancer using a combination of p53-dependent and independent pathways.

We thank Giuseppe Bertini and Giancarlo Cortese and Pierino Piccoli (SAFU-IRE) for technical assistance. We thank Marco Ranalli for technical assistance with the confocal microscopy.