|Home | About | Journals | Submit | Contact Us | Français|
The mammalian Golgi apparatus is composed of multiple stacks of cisternal membranes organized laterally into a ribbon-like structure, with close apposition of trans Golgi regions with specialized endoplasmic reticulum (ER) membranes. These contacts may be the site of ceramide transfer from its site of synthesis (ER) to sphingomyelin (SM) synthase via ceramide transfer protein (CERT). CERT extracts ceramide from the ER and transfers it to Golgi membranes, but the role of overall Golgi structure in this process is unknown. We show here that localization of CERT in puncta around the Golgi complex requires both ER and Golgi binding domains of CERT. To examine how Golgi structure contributes to SM synthesis, we treated cells with Golgi perturbing drugs and measured newly synthesized SM. Interestingly, disruption of Golgi morphology with nocodazole, but not ilimaquinone inhibited SM synthesis. Decreased localization of CERT with a Golgi marker correlated with decreased SM synthesis. We propose that some Golgi structural perturbations interfere with efficient ceramide trafficking via CERT, and thus SM synthesis. The organization of the mammalian Golgi ribbon together with CERT may promote specific ER-Golgi interactions for efficient delivery of ceramide for SM synthesis.
The Golgi complex is a central organelle in the secretory pathway. In mammals, the organization of the Golgi apparatus is complex. It is composed of multiple stacks of polarized cisternal membranes that are held in a juxtanuclear position by microtubules (reviewed in 1). These stacks of cisternal membranes are aligned laterally in a ribbon-like structure, which is unique to mammalian cells. However, the importance of this ribbon-like Golgi structure in mammalian cells remains unknown.
The Golgi apparatus plays an important role in protein processing and sorting. Proteins destined for secretion move from the cis face to the trans face of the Golgi, undergoing post-translational modifications and sorting for transport to target membranes. In addition to protein processing and sorting, the Golgi also plays an important role in sphingolipid biosynthesis. Enzymes that convert ceramide to glucosyl ceramide (GlcCer), sphingomyelin (SM), lactosyl ceramide and other sphingolipids, all reside in Golgi membranes. Sphingolipids are integral components of cell membranes, and their metabolites are important second messengers that regulate cell proliferation, differentiation and survival (reviewed in 2, 3). Ceramide (and ceramide metabolite) levels are regulated by a number of enzymes including sphingomyelinase, ceramidase, glucosyl ceramide synthase, and sphingomyelin synthase, which control production and degradation (reviewed in 2, 4).
Ceramide is made in the endoplasmic reticulum (ER) by condensation of serine and palmitate (reviewed in 2, 4). For sphingomyelin synthesis, ceramide must be transported to the Golgi. In HeLa cells, SM synthase 1 is enriched in the trans Golgi membranes (5). Interestingly, when vesicular trafficking from the ER to the Golgi was blocked, SM synthesis was unaltered (6, 7), suggesting that ceramide can be transported to Golgi membranes by a vesicle-independent mechanism. It has been suggested that the pool of ceramide required by GlcCer synthase is delivered to cis Golgi by vesicular transport while the pool of ceramide required by SM synthase is delivered directly to the trans Golgi by a non-vesicular mechanism (8–10). In addition, high voltage electron microscope tomography and three-dimensional reconstructions have revealed the existence of ER membranes called ‘trans ER’ that lie in close apposition to trans Golgi membranes (11). These special trans ER-trans Golgi membrane contact sites have been speculated to function in non-vesicular lipid trafficking from the ER directly to the trans Golgi (11). Taken together, these findings point towards the existence of a non-vesicular trafficking route for ceramide.
Recently, Hanada et al. identified a protein called ceramide transfer protein (CERT) in a screen for genes that could restore SM synthesis in the LY-A cell line, where ceramide delivery to the Golgi is defective (8). In the LY-A cells, GlcCer synthesis occurrs normally, but the conversion of ceramide to SM is defective (8). CERT is a cytoplasmic protein of 68kDa, which is identical to a splicing variant of Goodpasture antigen-binding protein of unknown function (8). It has an N-terminal pleckstrin homology (PH) binding domain that binds phosphatidylinositol 4-monophosphate (PI4P)-rich membranes, an FFAT domain (2 phenylalanines in an acidic tract) that binds vesicle-associated membrane protein-associated protein (VAP) that is anchored on ER membranes, and a C-terminal steroidogenic acute regulatory protein-related domain that forms a lipid binding pocket that binds and extracts ceramide (8, 10).
The existence of a lipid-binding domain in addition to Golgi and ER-binding domains suggests that CERT can directly shuttle ceramide between the ER and Golgi. The model predicts that CERT interacts with VAP on ER membranes where it extracts ceramide, and delivers it to the trans Golgi where SM synthesis occurs (10, 12). However, it remains unclear if CERT can simultaneously interact with the two membranes at the trans ER-trans Golgi contact sites. Here, we show by immunofluorescence studies that CERT is localized in an area surrounding the Golgi region. This localization requires both ER- and Golgi- binding domains on CERT, suggesting that CERT may interact with both ER and Golgi membranes simultaneously. Unexpectedly, we found that some acute Golgi structural perturbations reduced SM synthesis, while others did not. Reduced SM synthesis correlated with decreased Golgi association of CERT. We propose that close association or direct binding of ER and Golgi membranes is required for efficient delivery of ceramide to SM synthase 1, and that the organization of the mammalian Golgi ribbon together with CERT promotes these ER-Golgi interactions.
To assess CERT’s localization in HeLa cells, we analyzed the distribution of endogenous CERT protein with Golgi and ER markers by indirect immunofluorescence microscopy. Endogenous CERT protein was enriched at the Golgi with additional staining surrounding the Golgi region (Figure 1). Some of the CERT staining surrounding the Golgi region colocalized with the ER marker (Figure 1, arrows), suggesting that CERT interacts with both Golgi and ER membranes. Since CERT possesses both ER-binding (FFAT) and Golgi-binding (PH) domains in addition to its ceramide-binding domain, we examined the contribution of both ER- binding and Golgi-binding to the steady state localization of CERT. To this end, we mutated either the Golgi or ER-binding domains of CERT. CERT was Myc-tagged and mutant versions of CERT were constructed to inactivate the FFAT domain (FF321, 322AA; CERT-FFATmut) or to inactivate the PH domain (G67E; CERT-PHmut). Myc-tagged CERT protein was concentrated at the Golgi complex and also localized to distinct punctate structures near the Golgi region (Figure 2A), similar to the distribution of endogenous CERT. The mutated CERT proteins localized differently than the Myc-tagged wild type protein. CERT-FFATmut localized exclusively to the Golgi, without additional puncta (Figure 2B). However, CERT-PHmut localized exclusively to the ER (Figure 2C and data not shown). These results suggest that in HeLa cells, wild type CERT protein can interact with both Golgi and ER membranes in a localized region. Our data demonstrate that disrupting either of these membrane interacting domains affects the steady state localization of CERT, an observation that is consistent with the possibility that CERT may bind both Golgi and ER membranes simultaneously.
We hypothesized that perturbation of Golgi structure would disrupt the trans ER-trans Golgi contacts through which CERT may mediate non-vesicular transport of ceramide from the ER to the Golgi. In order to determine if Golgi morphology is important for efficient sphingomyelin synthesis, we treated cells with different Golgi perturbing drugs including ilimaquinone, nocodazole and brefeldin-A for 1 h. Ilimaquinone disrupts the Golgi by forming small vesicles from Golgi membranes, seen as punctate structures scattered through the cytoplasm by immunofluorescence microscopy (13). Nocodazole destabilizes microtubules and induces formation of small Golgi mini-stacks that retain their cis/trans polarity (but lack the lateral association between the stacks of the Golgi ribbon), also seen as punctate structures scattered through out the cytoplasm (14). Brefeldin-A, however, causes Golgi membranes to fuse with the ER, resulting in Golgi and ER membrane mixing (15). Ilimaquinone and brefeldin-A block secretion of cargo proteins, whereas secretion is normal in nocodazole-treated cells after a brief delay (13–15). Thus, the complex structure of the mammalian Golgi apparatus is disrupted by each of these treatments in different ways.
Cells were treated with ilimaquinone, nocodazole, brefeldin-A or vehicle control and labeled with 3H-serine for 1 h at 37°C. Lipids were extracted and analyzed by thin layer chromatography (TLC) and phosphorimaging. Figure 3A depicts the amount of newly synthesized SM in the presence of Golgi disrupting drugs as a percentage of their respective vehicle controls. Cells treated with ilimaquinone showed a slight (but not statistically significant) increase in the amount of newly made SM, while those treated with brefeldin-A showed a dramatic increase in newly made SM during the 1 h treatment. However, the amount of newly made SM decreased by about 50% when cells were treated with nocodazole. An approximate 50% reduction in newly made SM was also observed when a higher concentration of nocodazole was used, or when a longer nocodazole treatment (3 h) was used to allow full recovery of protein secretion (14) (data not shown).
To examine if Golgi perturbing drugs had any affect on synthesis of ceramide (the precursor of SM), we also analyzed the levels of ceramide after treatment with Golgi perturbing drugs. Figure 3B depicts the amount of newly synthesized ceramide in the presence of the Golgi disrupting drugs as a percentage of their respective vehicle controls. None of the drugs affected newly synthesized ceramide levels significantly, indicating that ceramide synthesis occurred normally in the ER in treated cells. Thus, the decrease in newly synthesized SM in nocodazole treated cells could not be attributed to a decrease in ceramide.
The increase in SM synthesis in brefeldin-A-treated cells probably reflects the mixing of substrate and enzyme in the ER (16–19). However, the opposite effects of ilimaquinone and nocodazole on SM synthesis were unexpected, and warranted further investigation.
We next followed the localization of CERT when Golgi morphology was disrupted by Golgi perturbing drugs. We used cyan fluorescent protein (CFP)-fused to an ER localization signal KDEL (CFP-KDEL) to label the ER and galactosyl transferase tagged with DsRed (galT-DsRed) to label Golgi membranes. Transfected cells were treated with ilimaquinone, nocodazole, brefeldin-A or vehicle control for 1 h. After drug treatment, the cells were immunostained and confocal imaging was used to assess CERT’s localization with respect to the ER and Golgi markers.
In control cells, Myc-tagged CERT localized with galT-DsRed at the Golgi and in a few punctate structures near the Golgi region (Figure 4A). With ilimaquinone and nocodazole treatments, the Golgi apparatus dissociated into punctate structures scattered throughout the cytoplasm as shown by the galT-DsRed pattern (Figures 4B and C). The punctate Golgi structures in ilimaquinone-treated cells were smaller and more abundant than in nocodazole-treated cells. However, CERT puncta associated more closely with the Golgi structures formed by ilimaquinone treatment (Figure 4B, arrows) than those formed by nocodazole treatment (Figure 4C). In nocodazole-treated cells, CERT localization was more diffuse than punctate, and overlapped little with Golgi ministacks. Thus, CERT appeared to associate with ilimaquinone-induced vesiculated Golgi membranes more efficiently than with the ministacks formed with nocodazole treatment. To confirm this observation, we quantified the overlap of both Myc-tagged CERT (Figure 5) and endogenous CERT (Figure S1) with the fragmented Golgi structures and the ER. Confocal microscopy data were used to measure colocalization by determining the overlap coefficient as described in Materials and Methods. The overlap of Myc-tagged CERT and the Golgi marker galT-DsRed (Figure 5) paralleled the SM synthesis data (Figure 3A), with a similar reduction in overlap and newly synthesized SM in nocodazole-treated cells. Interestingly, the overlap of CERT and the ER marker increased with both ilimaquinone and nocodazole treatments, suggesting that the ability to associate with ER was not inhibited by disrupting the Golgi with these drugs. In addition, the overlap of endogenous CERT with both the Golgi and ER markers (Figure S1) was similar to that of the Myc-tagged protein for all treatments tested.
We asked if the defective SM synthesis and localization of CERT in nocodazole-treated cells was reversible. HeLa cells were treated with nocodazole for 1 h after which nocodazole was removed and the Golgi was allowed to re-assemble over a period of 2 h. Cells were labeled with 3H-serine during the last hour of washout as described in Materials and Methods. Vehicle-treated and nocodozole-treated cells were used as controls. Lipids were extracted and analyzed by TLC and phosphorimaging. Newly synthesized SM levels recovered to control levels during the 2 h nocodazole washout procedure (Figure 6A) while newly synthesized ceramide levels remained unaffected (Figure 6B), suggesting that the affect of nocodazole on newly synthesized SM levels was reversible.
We also looked at the localization of both endogenous and Myc-tagged CERT proteins along with Golgi and ER markers. In nocodazole-treated cells, localization of both endogenous and Myc-tagged CERT proteins was diffuse and overlapped little with Golgi ministacks (Figures 4C, ,55 and S2). However, in nocodazole-washout cells, the ministacks re-assembled to form a compact Golgi structure and CERT re-localized to the Golgi region (Figure S2), as seen in vehicle control cells (Figures 4A and S2). This suggests that the effect of nocodazole on CERT localization and SM synthesis is fully reversible, further confirming the correlation between efficient SM synthesis and Golgi localization of CERT.
Our lipid synthesis data and analysis of CERT localization suggested that transfer of ceramide from the ER to Golgi ministacks in nocodazole-treated cells was inefficient. However, there are several alternate explanations for this result. To test if nocodazole directly inhibits SM synthase activity, we used an in vitro assay. Lysates from HeLa cells were incubated with nitro-2-1, 3-benzoxadiazol-4-yl (NBD)-C6-ceramide (NBD-C6-ceramide) in the absence or presence of nocodazole. As shown in Figure 7A, nocodazole did not inhibit SM synthase activity; instead a slight (but not statistically significant) increase in enzyme activity was seen.
Recent studies have shown that CERT can exist as both phosphorylated and non-phosphorylated forms (20, 21). CERT has many potential phosphorylation sites, but the major phosphorylation sites were identified at a serine-repeat motif, next to the PH domain (20, 21). It has also been shown that phosphorylation of CERT causes inactivation of the PH domain and ceramide transfer activities of CERT, thus preventing CERT from binding to PI4P-enriched Golgi membranes and downregulating the transport of ceramide from the ER to the Golgi (20, 21). To rule out the possibility that nocodazole promotes inactivation of CERT by phosphorylation, we examined the effect of nocodazole on the phosphorylation status of CERT. Lysates from control cells or those treated with nocodazole for 1 h were incubated with or without phosphatase and analyzed by Western blotting. The migration pattern of CERT is normally a doublet of approximately 75 kDa (Figure 7B). After phosphatase treatment, the slowest migrating band was shifted to a single band at 73 kDa. Nocodazole treatment did not significantly affect the amount of phosphorylated and non-phosphorylated forms of CERT when compared to the control (Figure 7B). This suggests that nocodazole does not affect the phosphorylation status of CERT and in turn the ability of CERT to transfer ceramide from the ER to the trans Golgi.
Finally, to directly test if nocodazole interferes with interaction of CERT with Golgi membranes, we asked if CERT-FFATmut (which exclusively binds the Golgi) could associate with Golgi membranes after nocodazole treatment. CERT-FFATmut localized more strongly with Golgi ministacks in nocodazole-treated cells (Figure 7C), than to wild type CERT in control cells. This suggests that nocodazole does not eliminate CERT’s ability to interact with Golgi membranes. This is consistent with another study demonstrating that FAPP2, a GlcCer transfer protein with a PH domain similar to that of CERT, is able to interact with Golgi membranes after nocodazole treatment (22). Although nocodazole treatment somewhat decreased the overlap of CERT-FFATmut with the Golgi marker when compared to untreated cells expressing CERT-FFATmut, the overlap still exceeded that of wild type CERT in control cells. Interestingly, wild type CERT associated weakly with Golgi ministacks in nocodazole-treated cells and instead showed a diffuse staining pattern (Figure 7C). Thus, in the presence of an intact FFAT binding domain, wild type CERT was unable to efficiently localize to the Golgi in nocodazole-treated cells.
Our results suggest that the reorganization of the Golgi complex in nocodazole-treated cells is incompatible with efficient delivery of ceramide to Golgi membranes by wild type CERT resulting in a decrease in SM synthesis.
Proteins and most lipids made in the ER are transported to other subcellular organelles by vesicular transport. However, some lipids can move directly from one compartment to another by non-vesicular routes (reviewed in 9). Recently, Hanada et al. (8) proposed that ceramide can be transported from the ER to the Golgi by the ceramide transfer protein, CERT. Different domains of CERT protein can interact with ER and Golgi membranes. However, CERT’s mechanism of action is unclear. CERT could (a) function by extracting ceramide anywhere from the ER, traverse the cytoplasm and transport it to trans Golgi membranes, (b) interact with both membranes simultaneously at the trans ER-trans Golgi contact sites, or (c) shuttle between the two organelles across the ~10 nm cytosolic junction at these membrane contact sites. In HeLa cells, CERT was enriched at the Golgi, and present in punctate structures dispersed around the Golgi region that partially overlapped with an ER marker (Figures 1 and and2).2). It is possible that the CERT-containing puncta near the Golgi region represent trans ER-trans Golgi contact sites, although we did not test this directly. However, when the ER or Golgi binding domains were mutated, CERT localized tightly to the Golgi or the ER, respectively, suggesting that wild type CERT protein requires both Golgi- and ER-binding domains for proper localization (Figure 2). CERT may simultaneously interact with both ER and Golgi membranes, or shuttle ceramide between these two organelles in an area confined to the Golgi region. Our data are consistent with the ‘neck swinging’ mechanism of the CERT lipid binding domain that has been previously proposed (8, 23). Our results are also consistent with the recent biochemical finding that dephosphorylation of CERT is VAP-A dependent and that dephopshorylation of CERT enhanced the association of CERT with both ER and Golgi membranes (24), lending support to the localized function of CERT.
We reasoned that the Golgi ribbon structure contributes to maintaining trans ER-trans Golgi contact sites, and that disrupting Golgi structure might interfere with ceramide trafficking and SM synthesis. Surprisingly, we found that vesiculation of Golgi membranes with ilimaquinone resulted in a slight increase in newly synthesized SM, while induction of Golgi ministacks with nocodazole resulted in a significant decrease in newly synthesized SM (Figure 3A). The large increase in newly synthesized SM in brefeldin-A-treated cells was not unexpected, since the substrate and enzyme would be mixed together in the ER (16–19). However, the opposite effects of ilimaquinone and nocodazole on newly made SM levels were unexpected, and warranted further investigation.
We found that association of CERT with Golgi membranes correlated with SM synthesis in drug-treated cells (Figures 4, ,55 and S1). In ilimaquinone-treated cells, CERT associated efficiently with both vesiculated Golgi membranes and the ER, and SM synthesis was normal or slightly increased. This could indicate that trans ER-trans Golgi contact sites can be maintained or regenerated after vesiculation with ilimaquinone, or that the large number of Golgi vesicles allows efficient ceramide transport in the absence of contact sites. In nocodazole-treated cells, we propose that the reduction in SM synthesis is due to decreased ceramide trafficking, since nocodazole did not directly inhibit SM synthase activity, alter the phosphorylation of CERT, or eliminate CERT’s ability to associate with Golgi membranes (Figure 7). Although treatment with nocodazole reduced the overlap of CERT-FFATmut with the Golgi marker somewhat when compared to DMSO-treated cells (Figure 7C), this would be unlikely to account for the 50% reduction in newly synthesized SM that we observed. Also, the effects of nocodazole were completely reversible with washout, as both SM synthesis and CERT localization were restored (Figures 6 and S2). Finally, it is unlikely that ceramide trafficking by CERT requires microtubules, since ilimaquinone disrupts microtubules in addition to vesiculation of Golgi membranes (25).
We also found that in drug-treated cells, the levels of newly synthesized ceramide remained largely unaffected (Figure 3B). The finding that newly synthesized ceramide levels did not mirror differences in newly synthesized SM suggests that ceramide levels are regulated independently in terms of synthesis and metabolism. Thus, while vesiculated Golgi membranes formed by ilimaquinone support ceramide trafficking and SM synthesis, ministacks formed by nocodazole treatment do not.
In this study we only analyzed ceramide and SM levels with respect to Golgi structure. However, it would be interesting to see if newly synthesized GlcCer and other glycosphingolipid levels are also affected by Golgi structural perturbations, especially with the recent finding that FAPP2 is a non-vesicular carrier of GlcCer that is responsible for trafficking the lipid from early Golgi membranes to late Golgi membranes for glycosphingolipid synthesis (5, 26). It would also be interesting to see if localization of FAPP2 and another PH domain containing protein, the oxysterol-binding protein (OSBP) to Golgi membranes is affected when Golgi structure is perturbed. OSBP was shown to be involved in regulation of CERT activity and SM synthesis (27). Also, all our studies were performed on non-synchronized HeLa cells. Previous data suggest that SM synthesis (but not GlcCer synthesis) is reduced during mitosis when the Golgi vesiculates (28). However, these studies were performed in cells that were synchronized by nocodazole treatment (28), so it is unclear if the reduction in SM is solely due to the mitotic rearrangement of Golgi structure.
Our data suggest that certain Golgi morphologies (normal Golgi ribbon and ilimaquinone-induced vesiculated Golgi membranes) promote efficient SM synthesis while others (nocodazole-induced ministacks) do not. Taken together, our data indicate that the organization of the mammalian Golgi structure may play an important role in efficient ceramide trafficking and in SM synthesis. We speculate that the complex organization of the Golgi apparatus facilitates rapid transport of ceramide from the ER via a non-vesicular mechanism, resulting in increased SM synthesis to support diverse biological processes. Acute perturbation of Golgi structure during cellular stress could thus impact ongoing SM synthesis, and could initiate signaling pathways induced by sphingolipid metabolites. We plan to further examine the role of Golgi structure by carrying out knock-down studies on Golgi structural proteins in order to understand the cellular mechanism and functional basis of the complex ribbon-like structure of the mammalian Golgi apparatus in terms of sensing and maintaining ceramide and sphingolipid levels in cells.
HeLa cells were maintained in Dulbecco’s modified Eagle’s medium (Gibco Invitrogen, Carlsbad, CA) supplemented with 10 % fetal calf serum (Atlanta Biologicals, Norcross, GA) and 0.1 mg/ml Normocin (Invivogen, San Diego, CA) at 37°C in 5% CO2.
CERT was cloned from HeLa cell mRNA by reverse-transcriptase polymerase chain reaction using sense primer 5 ′ GAATTCACCATGTCGGATAATCAG 3′ and antisense primer 5 ′ GCGGCCGCGAACAAAATAGGCTTTCC 3′ and inserted into pcDNA 3.1/Myc-His vector (Invitrogen) at the EcoRI and NotI restriction sites to insert a Myc tag at the C-terminus. The sequence was confirmed by dideoxy sequencing. Two mutant versions of CERT were also made by site directed mutagenesis using QuikChange (Stratagene, La Jolla, CA). In one construct, the FFAT ER-binding domain of CERT was inactivated (FF321,322AA) while in the other construct, the Golgi-binding PH domain of CERT was inactivated (G67E). To construct galactosyl transferase (galT) fused to the coral Discosoma striata red fluorescent protein (DsRed), cDNA encoding residues 1–47 of β1,4-galactosyl transferase were excised by digestion with NheI and BamHI from pECFP-Golgi (Clontech, CA), and ligated with cDNA encoding a monomeric version of DsRed fluorescent protein (excised with Bgl2 and NotI from pST10 (from B. Glick, Univ. of Chicago)) into pcDNA3.1 (Invitrogen) at the NheI and NotI sites. The sequence was confirmed by dideoxy sequencing. A construct carrying cyan fluorescent protein (CFP)-tagged to an ER localization signal KDEL (pCFP-ER) was obtained from Clontech.
Affinity purified anti-golgin-160 antibodies recognizing residues 60–139 and 140–311 were used in a ratio of 1:1. Briefly, purified glutathione-S-transferase (GST) was cross-linked to glutathione beads using disuccinimidyl suberate and incubated with rabbit antiserum raised to GST-fused to golgin-160(1–393) to deplete the serum of anti-GST antibodies. The flow through (depleted of anti-GST) was then incubated with glutathione beads crosslinked to GST-tagged golgin-160(60–139) or golgin-160(140–311). Bound antibodies were eluted with 4 M MgCl2 and dialysed. Mouse anti-GM130 was obtained from BD Transduction (San Diego, CA) and monoclonal anti-Myc antibody (clone 9E10) was from Roche Molecular Biochemicals (Indianapolis, IN). Chicken anti-CERT IgY (IgG) polyclonal antibody was from GenWay Biotech, Inc (San Diego, CA). Polyclonal rabbit anti-GFP, Alexa-488 conjugated goat anti-rabbit and Alexa-488 conjugated donkey anti-mouse antibodies were from Molecular Probes, Inc (Eugene OR). Texas Red-conjugated goat anti-rabbit (IgG), Cy5-conjugated donkey anti-mouse (IgG) and Cy5-conjugateddonkey anti-chicken IgY (IgG) were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). Horseradish peroxidase-conjugated donkey anti-mouse (IgG) antibodies were obtained from GE Healthcare Bio-Sciences Corp. (Piscataway, NJ).
HeLa cells were grown to confluence on 6 cm dishes at 37°C. The cells were treated with 2.5 μg/ml ilimaquinone (from V. Malhotra, UCSD, La Jolla, CA), 5 μg/ml and 2.5 μg/ml nocodazole, 1 μg/ml brefeldin-A (Sigma, St Louis, MO), or ethanol or DMSO (Burdick and Jackson, Muskegon, MI) vehicle controls for 1 h at 37°C in minimum essential medium (MEM) (Gibco Invitrogen). During drug treatments, 50 μCi/ml 3H-serine (GE Healthcare Bio-Sciences Corp.) was added to label newly synthesized sphingolipids. The treatments were performed in the presence of 10 μg/ml cycloheximide to reduce incorporation of 3[H]-serine into proteins. Cells were rinsed twice in phosphate-buffered saline (PBS) and resuspended in 500 μl of PBS. The resuspended cells were transferred to glass tubes and kept on ice. 50 μl of the cells from each treatment were used to assay for total protein.
The cells were subjected to lipid extraction by the standard Bligh and Dyer method (29). Briefly, chloroform/methanol (1:2) was added to each tube. The tubes were vortexed and the cells were sonicated in a water bath for 30 seconds. The tubes were incubated on ice for 30 min after which chloroform/water (1:1) were added to each tube and spun at 720 × g for 5 min at 4°C. The upper aqueous phase was discarded while the lower organic phase was transferred to new tubes and concentrated to a final volume of 20 μl of chloroform/methanol (19:1) under N2 gas and stored at −20°C. The samples were run on high performance-thin layer chromatography (HP-TLC) silica gel 60 plates (EMD Chemical Inc., Gibbstown, NJ) along with sphingomyelin (SM) (Sigma, St Louis, MO), ceramide, phosphatidylcholine (Avanti polar-lipids Inc., Alabaster, AL), phosphatidylethanolamine and phosphatidylserine (Matreya Inc., Pleasant Gap, PA) non-radioactive standards in two different solvents; chloroform/methanol/acetic acid/formic acid/water (17.5:7.5:3:1:0.5) and hexane/diethyl ether/acetic acid (20:5:0.25), sequentially. The plates were dried and the side with the standards was sprayed with a solution of 3% cupric acetate and 8% phosphoric acid, dried and charred over a heater to develop the non-radioactive standard bands. The plates were then exposed to a K/tritium phosphorimager screen (Bio-Rad Laboratories, Inc) for 2 days. The bands were subjected to analysis using Molecular Imager FX (Bio-Rad Laboratories, Inc) and Quantity One software (Bio-Rad Laboratories, Inc). The amount of each lipid measured was normalized to the amount of protein in each sample.
HeLa cells were were transiently transfected for approximately 24 h at 37°C with 0.5–1 μg DNA per 3.5 cm dish with Fugene 6 transfection reagent (Roche diagnostics, Indianapolis, IN) according to the manufacturer’s instructions. Cells were then either untreated, or subjected to treatment with ilimaquinone (2.5 μg/ml), nocodazole (2.5 μg/ml) brefeldin-A (1 μg/ml) or DMSO vehicle for 1 h at 37°C in MEM medium containing cycloheximide to mimic the lipid labeling experiments. Immunofluorescence studies were carried out to assess the localization of proteins of interest as described previously (30). Briefly, cells were either fixed with 3% paraformaldehyde in PBS for 10 min at room temperature and then permeabilized with 0.5% Triton X-100 in PBS containing 10 mM glycine (Gly) for 3 min (Figures 1, ,22 and S2 (endogenous CERT)), or fixed and permeabilized with methanol for 15 min at 4°C (Figures 4, ,7C7C and S2 (Myc-tagged CERT)). Coverslips were washed in PBS/Gly and were then incubated with primary antibodies diluted in PBS/Gly containing 1% bovine serum albumin for 20 min. This was followed by a 20 minute incubation with appropriate secondary antibodies. The coverslips were rinsed and then mounted onto glass slides in glycerol containing 0.1 M N-propyl gallate. Cells were visualized using an Axioskop microscope (Zeiss, Thornwood, NY) with an attached Sensys charge-coupled device camera (Photometric, Tucson, AZ). IP Lab imaging software (Signal Analytics, Vienna, VA) was used to collect and analyze the images. Cells expressing low levels of protein were selected for imaging. Confocal imaging wasperformed with a single point laser scanning confocal microscope (Zeiss Axiovert 200 microscope with 510-Meta confocal module from Zeiss, Thornwood, NY) to visualize Alexa-488 and DsRed spectral emissions. The meta detector was used to visualize the Cy5 spectral emission. Z-stacks were obtained using the Zeiss LSM image examiner software. Volocity software (Improvision, Inc., Lexington, MA) was used to create a single image from confocal sections and to obtain colocalization values by analyzing overlap coefficient for each cell. Overlap coefficient describes overlap of signals and thus the degree of colocalization (31). The values are defined between 0 and 1, with 1 denoting perfect colocalization. Individual cells (8–22) were analyzed for each treatment. Statistical measurements were performed by averaging data points for each cell and normalizing them to the average value of the vehicle-control in order to compensate for day-today variations in fixation, protein expression, staining and image capture. Standard error of the mean was calculated on the normalized data for each treatment.
For the 2 h nocodazole washout procedure, cells were treated with nocodazole (2.5 μg/ml) for 1 h at 37°C in MEM medium containing cycloheximide to mimic the drug treatment conditions. After 1 h nocodazole treatment, the cells were rinsed in PBS and incubated in DMEM for 1 h. During the 2nd hour of washout, the cells were once again rinsed in PBS and then incubated in MEM medium containing cycloheximide and 3H-serine. Lipids were extracted, run on TLC and phosphorimaged as described above. When imaging cells by confocal microscopy, all conditions were kept exactly the same, except for the addition of 3[H]-serine.
HeLa cells were grown on 6 cm dishes. SM synthase activity was measured as previously described with modifications (32). Cells were resuspended in 400 μl of lysis buffer (250 mM sucrose and 5 mM Hepes, pH 7.4) containing protease inhibitors. Lysates were made by homogenizing in a Dounce homogenizer and the homogenate was spun at 1000 × g for 10 min at 4°C. 150–200 μg of protein from the post-nuclear supernatant were aliquoted for each reaction. 5 nmol of nitro-2-1,3-benzoxadiazol-4-yl (NBD)-C6-ceramide (NBD-C6-Cer) (Molecular Probes Inc.) complexed to bovine serum albumin was added to each reaction along with nocodazole (2.5 μg/ml) or DMSO. The reaction was carried out at 30°C for 30 min, in the dark. The reaction was terminated with addition of 200 μl of chloroform and methanol (1: 1). Lipids were extracted by the standard Bligh and Dyer method and analyzed by TLC, as described above. The newly synthesized fluorescent NBD-sphingomyelin made from the exogenously added NBD-C6-Cer substrate was analyzed by VersaDoc Imaging System (Bio-Rad Laboratories, Inc) and quantified by Quantity One software (Bio-Rad Laboratories, Inc). The amount of each lipid measured was normalized to the amount of protein in each sample.
HeLa cells were grown on 6 cm dishes and transiently transfected with a plasmid encoding Myc-tagged CERT as described above. The cells were then incubated with nocodazole (2.5 μg/ml) or DMSO for 1 h at 37°C in MEM medium containing cycloheximide to mimic the lipid labeling experiments. The cells were washed twice with PBS and resuspended in lysis buffer (10mM Hepes, pH 7.4, 2mM EDTA, 1% Nonidet P-40 and 5mM DTT) containing a cocktail of protease inhibitors (Sigma). The lysate was incubated in absence or presence of 400 units of lambda protein phosphatase (New England Biolabs Inc., Ipswich, MA) for 30 min at 30°C and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE).
Samples were resuspended in sample buffer (50mM Tris-HCl, pH 6.8, 2% sodium dodecyl sulfate (SDS), 20% glycerol, 0.025% bromophenol blue) containing 3.75% 2-mercaptoethanol and run on a 12.5% polyacrylamide denaturing gel. The proteins were electrophoretically transferred to a polyvinylidene difluoride membrane (Milipore Corporation, Bilerica, MA) and the membrane was blocked with 5% non fat dry milk in Tris buffer saline (TBS) (150mM NaCl, 10mM Tris-HCl, pH 7.4) containing 0.05% Tween 20 (TBS-T) for 1 h at room temperature. The membrane was incubated in primary antibody in TBS-T containing 4% bovine serum albumin overnight at 4°C followed by incubation with secondary antibody in TBS-T containing 5% non-fat dry milk at room temperature for 1 h. After incubation with enhanced chemiluminescence reagents (GE Healthcare BioSciences Corp., NJ) bands were detected by using autoradiography film (Denville Scientific Inc., NJ).
Figure S1. Decreased overlap of endogenous CERT with Golgi membranes in nocodazole-treated cells. Overlap of CERT with the Golgi (galT-DsRed) and ER (CFP-ER) markers was quantified for 9–13 individual cells for each treatment as described in Materials and Methods, and overlap coefficient was calculated. The graph indicates average overlap coefficient values for each treatment after normalization to the DMSO control (which was set to 1.0). Open columns indicate overlap coefficient of endogenous CERT with the Golgi marker, and closed columns indicate overlap coefficient of endogenous CERT with the ER marker. Standard error of the mean was calculated for each treatment and is represented on the graph.
Figure S2. CERT is re-localized to the Golgi region after nocodazole washout. HeLa cells expressing galT-DsRed (Golgi marker) and CFP-ER (ER marker) were treated with (A) DMSO (vehicle control for nocodazole) or (B) nocodazole (2.5 μg/ml) for 1 h at 37°C in presence of 10 μg/ml cycloheximide. Nocodazole washout (C) was performed as described in Materials and Methods. The cells were fixed and stained with commercially available chicken anti-CERT IgY (IgG) and Cy5 conjugated anti-chicken IgY (IgG) antibodies (for endogenous CERT) or with anti-Myc and Cy5 conjugated anti-mouse (IgG) (for cells expressing Myc-tagged CERT). Confocal microscopy was performed and 3-D images were created using Volocity software. Bars, 10 μm.
We greatly appreciate the gifts of ilimaquinone from Dr. Vivek Malhotra (UCSD, La Jolla, CA) and DsRed plasmid from Dr. Benjamin Glick (Univ. of Chicago, IL). We would like to acknowledge David Zuckerman (Johns Hopkins University School of Medicine, Baltimore, MD) for making affinity purified golgin160(60–311) and (140–311) antibodies. We would also like to thank the Johns Hopkins University School of Medicine Microscope Facility for their assistance with confocal microscopy. We finally thank the members of the Machamer Lab and Dr. Dan Raben for useful comments on the manuscript. This work was supported by National Institutes of Health Grant GM42522 (to C. E. M).