Purification and characterization of UDP-glucose:ceramide glucosyltransferase from rat liver Golgi membranes.

We present a method for solubilizing and purifying UDP-Glc:ceramide glucosyltransferase (EC 2.4.1.80; glucosylceramide synthase (GCS) from a rat liver and present data on its substrate specificity. A Golgi membrane fraction was isolated, washed with N-lauroylsarcosine, and subsequently treated with 3[3-cholamidopropyl)-dimethylammonio]-2-hydroxy-1-propanesulfonate to solubilize the enzyme. GCS activity was monitored throughout purification using UDP-Glc and a fluorescent ceramide analog as substrates. Purification of GCS was achieved via a two-step dye-agarose chromatography procedure using UDP-Glc to elute the enzyme. This resulted in an enrichment > 10,000-fold relative to the starting homogenate. The enzyme was further characterized by sedimentation on a glycerol gradient, I labeling, and SDS-polyacrylamide gel electrophoresis. which demonstrated that two polypeptides (60-70 kDa) corresponded closely with GCS activity. Purified GCS was found to require exogenous phospholipids for activity, and optimal results were obtained using dioleoyl phosphatidylcholine. Studies of the substrate specificity of the purified enzyme demonstrated that it was stereospecific and dependent on the nature and chain length of the N-acyl-spingosine or -sphinganine substrate. UDP-Glc was the preferred hexose donor, but TDP-glucose and CDP-glucose were also efficiently used. This study provides a basis for molecular characterization of this key enzyme in glycosphingolipid biosynthesis.

Glycosphingolipids (GSLs) 1 are amphipathic molecules that contain the hydrophobic moiety, ceramide, and a hydrophilic oligosaccharide residue. They are found in the plasma membranes of all eukaryotic cells and play important roles in cell recognition, cell proliferation and differentiation, immune recognition, and signal transduction (for reviews, see Refs. [1][2][3][4][5]. The biochemical pathways for GSL synthesis are well established (reviewed in Refs. 6 -9), but not all of the enzymes involved in GSL synthesis have been purified and/or cloned (for review, see Ref. 10). One such enzyme is UDP-Glc:ceramide glucosyltransferase (EC 2.4.1.80; glucosylceramide synthase (GCS)), which catalyzes the formation of glucosylceramide (GlcCer) from ceramide and UDP-Glc (11). This enzyme is of particular interest for a number of reasons. First, it has been shown that there is a correlation between tumor progression and cell surface GSLs (1,12). Since many complex acidic and neutral GSLs are derived from GlcCer, regulation of GCS activity could have a profound effect on cell growth activity. Indeed, Radin and colleagues (13) have developed a GCS inhibitor (1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP)) and demonstrated some remarkable effects of this compound both in vitro and in vivo. Recently, a defect in GCS activity was also characterized in a mutant melanoma cell line and associated with altered growth properties (14). Second, GCS activity is concentrated at the Golgi complex (15,16), but its precise distribution is not known. Unlike sphingomyelin synthesis, which occurs principally at the cis/medial Golgi apparatus, GlcCer synthesis is more widely distributed within the Golgi. Thus, it is of interest to learn what parts of the GCS molecule are responsible for its unique intracellular distribution. Finally, it should be noted that GlcCer synthesis occurs at the cytosolic surface of intracellular membranes (15)(16)(17). However, formation of complex GSLs from GlcCer is believed to occur by glycosylation reactions taking place on the lumenal surface of the Golgi apparatus. Thus, GlcCer synthesis must be accompanied by transbilayer movement or "flip-flop" of the lipid during or after its synthesis.
To further study some of these problems, it will be necessary to purify, characterize, and eventually clone and sequence GCS. In the present paper, we describe methods for the solubilization and purification of GCS from rat liver Golgi membranes as a step toward this goal. We also provide novel information about the enzymatic properties of GCS.

Preparation of Golgi Membranes
Golgi-enriched membranes from rat liver were prepared as described (21,22) with the following modifications. Livers were minced with a razor blade in homogenization buffer (0.25 M sucrose, 50 mM Tris/HCl, pH 7.4, 25 mM KCl containing 1 g/ml each of antipain and leupeptin, 25 M amidinophenylmethanesulfonyl fluoride, and 10 g/ml aprotinin). Additional homogenization buffer was added to make a 20% (w/v) suspension, and the minced livers were then homogenized by passing the suspension through a ball bearing homogenizer four times (Berni-Tech Engineering, Saratoga, CA) with a clearance of 0.0054 inches (23). The homogenate was adjusted to 1.07 M sucrose by addition of 2.0 M sucrose. The homogenate (19 ml) was loaded into ultracentrifuge tubes and overlaid in succession with 4 ml of 1.02 M sucrose and a 14-ml linear gradient of 1.02 M to 0.2 M sucrose. After centrifugation (Beckman SW28 rotor, 83,000 ϫ g, 4°C, 210 min), a prominent white band at ϳ0.75 M sucrose was harvested. These Golgi-enriched fractions were flash frozen in liquid nitrogen and stored at Ϫ80°C until use. Under these conditions, GCS activity was stable for several months.
Detergent Solubilization of Membrane Proteins-Golgi-rich fractions from rat liver were thawed and mixed with an equal volume of 0.1% N-lauroylsarcosine in buffer A (50 mM Hepes, pH 7.4, 20% (v/v) glycerol, and 0.02% NaN 3 ) with 25 mM KCl and 2 mM dithiothreitol. The mixture was stirred for 30 min and centrifuged for 1 h at 200,000 ϫ g (Beckman 50.2Ti rotor). GCS activity was recovered in the N-lauroylsarcosineinsoluble pellet. The pellet was suspended at ϳ1 mg of protein/ml in buffer A supplemented with 1% CHAPSO, 1 mM UDP-Glc, and 1 mM dithiothreitol and stirred for 1 h. The CHAPSO-soluble supernatant and insoluble pellet were separated after centrifugation for 1 h at 100,000 ϫ g.
Dye-Agarose Chromatography-Preliminary studies showed that GCS bound to dye-agarose but was eluted from the dye column in the presence of UDP-Glc. These observations suggested a two-step procedure to enrich for GCS. First, CHAPSO-soluble supernatant (ϳ50 ml) was loaded (flow rate of 12 ml/h) onto a dye-agarose column (1 ϫ 20 cm), which had been equilibrated with buffer B (buffer A supplemented with 0.5% CHAPSO and 1 mM dithiothreitol) plus 1 mM UDP-Glc. The unbound fraction (42 ml) from the first column containing ϳ70% of the total applied GCS activity was concentrated to 4 ml by using Centriprep-50 concentrators (Amicon, Inc.). UDP-Glc was then removed from the unbound fraction by gel filtration through Biogel P2 (Bio-Rad) columns (1 ϫ 10 cm) equilibrated with buffer B using a centrifugation method (24). The unbound fraction was then slowly applied (12 ml/h) onto a second dye-agarose column (1 ϫ 4 cm) equilibrated in buffer B. The column was washed with 10 ml of buffer B and sequentially eluted with 20 ml of 0.15 M KCl in buffer C (buffer B plus 0.1 mM 1,2-dioleoyl phosphatidylcholine (DOPC)), 15 ml of 20 mM UDP-Glc, and 20 mM NADH in buffer C followed by 15 ml of 1 M KCl in buffer C. 2-ml fractions were collected, and aliquots (20 l) of each fraction were assayed for GCS activity.
Glycerol Density Gradients-Purified GCS samples were first chromatographed through a Sephadex G-25 column (1 ϫ 5 cm) equilibrated with 1% glycerol in buffer D (50 mM Hepes, pH 7.4, 0.6% CHAPSO, 0.5 mM UDP-Glc, and 100 mM KCl) as described above for the gel filtration procedure to reduce glycerol concentration. The fractions (250 l) were then loaded on top of 6 -25% (v/v) linear glycerol gradients in 5 ml of buffer D and centrifuged in a Beckman SW 55Ti rotor at 250,000 ϫ g for 12 h. After centrifugation, 30 fractions of 170 l were collected starting from the top of the gradient and were assayed for enzymatic activity. The sedimentation coefficient of the native enzyme was estimated as described previously (25) using catalase, cytochrome C, malate dehydrogenase, and lactate dehydrogenase as calibrating enzymes. For polypeptide visualization, an aliquot of the enzyme fraction was prelabeled with 125 I (see below) and added as a tracer to the sample loaded onto the glycerol gradient. The radioactive polypeptides in the glycerol gradient fractions were then analyzed by SDS-PAGE (see below).

GCS Assay
GCS activity was assayed as described previously (16), with the following modifications. Enzyme fractions were preincubated in screw cap tubes in 50 mM Hepes (pH 7.4), 25 mM KCl, 5 mM MnCl 2 , 5 mM UDP-Glc in a final volume of 0.5 ml for 5 min at 37°C. The enzyme reaction was initiated by addition of 5 nmol of C 6 -NBD-Cer⅐bovine serum albumin complex prepared as described (18,26). After 15 min at 37°C with constant stirring, the reaction was stopped by addition of 3 ml of chloroform/methanol (1:2 (v/v)). Lipids were extracted (27) and separated by thin layer chromatography (26). Individual spots on the TLC plates were identified by comparison with fluorescent standards and quantified by image analysis as described (28). In some cases, the reaction tubes were precoated with phospholipids (usually 0.1 mM DOPC) by first adding phospholipid dissolved in chloroform to screw cap tubes and evaporating the chloroform under a stream of N 2 . For the experiment shown in Fig. 5, the reaction tubes were also precoated with D-threo-PDMP using an ethanolic stock solution of the inhibitor.

Analytical Methods
The polypeptide composition of various fractions was assessed by SDS-PAGE (29). In some cases, the enzyme fractions were concentrated by precipitation with trichloroacetic acid in the presence of sodium deoxycholate prior to electrophoresis (30). Polypeptides were visualized using a silver staining kit (Bio-Rad) as described (31). To visualize the polypeptides in samples with very low protein concentrations, samples were prelabeled with 0.5-1 mCi of 125 I by the chloramine-T method (32). Radioactive gels were analyzed using a phosphorimager (Molecular Analyst GS-363, Bio-Rad) and by autoradiography. Immunoprecipitation of rat albumin was performed as described (33) using a polyclonal anti-rat albumin antiserum from Cappell/Organon Teknika (Durham, NC). The Golgi marker enzyme, galactosyltransferase, was measured as described (35). Electron microscopy of Golgi membranes was performed as described (21). Protein concentrations were measured using Coomassie Blue dye reagent (Bio-Rad) with bovine serum albumin as a standard (36).

Purification of GCS
Preliminary studies showed that GCS in Golgi membranes could be solubilized with CHAPSO (1% (w/v)) with recovery of ϳ70% of the activity in the extract (data not shown). We also found that addition of 0.1-1.0 mM DOPC and 0.1 mM UDP-Glc to the solubilized GCS improved its stability at 4°C (data not shown). Thus, these protective agents were added during some purification steps. Finally, we noted that after chromatographic procedures, GCS had specific phospholipid requirements for optimal activity (see below). Thus, most activity assays of solubilized GCS fractions were performed in the presence of 0.1 mM DOPC.
We used rat liver Golgi membranes as a starting material for the purification of GCS because our previous studies showed that these membranes were highly enriched in GCS activity (16). The Golgi fractions prepared in the present work were enriched 60 -80-fold in galactosyltransferase, a trans-Golgi marker protein, with ϳ35% recovery of this enzyme. In addition, electron microscopy of the Golgi samples revealed numerous intact stacks comprised of 4 and 5 cisternae (data not shown). Thus, these Golgi preparations were similar in enrichment and morphology to highly purified Golgi fractions prepared by established methods (for review, see Ref. 37).
Golgi membrane fractions were enriched Ͼ60-fold in GCS activity compared to crude rat liver homogenate with a yield of Ͼ50% (Table I). Since our enzyme assay used a fluorescent ceramide analog, which is also a substrate for sphingomyelin synthase (18,38,39), we were also able to determine that the Golgi membranes were enriched 35-40-fold in sphingomyelin synthase activity with a recovery of ϳ40% (data not shown). This observation is in agreement with the differential location of GCS and sphingomyelin synthase in the Golgi apparatus (16,17,21,40).
Detergent Solubilization-We found that pretreatment of Golgi-enriched fractions with 0.05% (w/v) N-lauroylsarcosine followed by ultracentrifugation resulted in the removal of 75% of the total Golgi protein in the supernatant, while most GCS activity (80%) was recovered in the pellet (Table I). The enrichment achieved by this step was ϳ200-fold relative to homogenate (Table I) or 3.15-fold compared to Golgi membranes. GCS activity in the pellet was then solubilized with 1% CHAPSO. The solubilized GCS was not apparently further enriched (Table I). However, activity of the solubilized enzyme was tested in the presence of CHAPSO (final concentration, 0.05%), a condition that suppresses GCS activity in intact Golgi membranes by ϳ30% (data not shown). Thus, the % recovery and enrichment values shown in Table I are underestimates. Two-step Dye-Agarose Column Chromatography-We attempted to purify GCS by affinity chromatography using various bound ligands (UDP-hexanolamine, UDP-GlucUA, ceramide, sphingosine, and PDMP), but these efforts were unsuccessful. We next explored dye adsorption chromatography and found that GCS activity could be released from a dye-agarose column by UDP-Glc, while most other dye-binding proteins were elutable only with salt. Thus, we devised a twostep purification procedure. First, CHAPSO-solubilized Golgi membranes were loaded onto the dye-agarose column in the presence of 1 mM UDP-Glc. GCS activity was then recovered (70% of total applied activity) in the unbound fractions with a 2.3-fold enrichment relative to the CHAPSO extract or a 474fold relative to the homogenate (Table I; Fig. 1A). After gel filtration to remove UDP-Glc, the unbound fractions from the first dye-agarose column were loaded onto a second dye-agarose column, which was equilibrated with buffer without UDP-Glc. Although most proteins passed through the second column, almost all the GCS activity was bound and was eluted successively with 0.15 M KCl, 20 mM UDP-Glc plus 20 mM NADH, and 1 M KCl (Fig. 1B). The 0.15 M KCl-eluted fraction was enriched ϳ5000-fold in GCS activity with a 1.15% recovery relative to homogenate (Table I). The amount of protein recovered in the UDP-Glc-eluted fraction was at the lower limits of detectibility (ϳ1 g/ml), resulting in a Ն10,000-fold enrichment of GCS with a recovery of Յ0.5% (Table I). The 1 M KCl-eluted fraction was also significantly enriched (3300-fold) in GCS activity (Table I). Furthermore, since CHAPSO partially inhibits GCS activity as noted above, the true enrichment of GCS in the purified fractions is greater than the values cited above (see Table I).
Polypeptide Composition of GCS Fractions-The polypeptide composition of GCS fractions during different steps of purification was assessed by SDS-PAGE (Fig. 2). Initial steps (Fig. 2,   lanes 1-4) were visualized by silver staining. To visualize the fractions eluted from the second dye-agarose column, which were very low in protein, aliquots were radioiodinated. When a radiolabeled aliquot of the 0.15 M KCl-eluted fraction was electrophoresed and visualized by autoradiography, the profile obtained (Fig. 2, lane 5) showed a similar pattern to that of the 3,300 f a 1 unit corresponds to 1 nmol of glucose transferred to C 6 -NBD-Cer per min under standard conditions. Units were calibrated using standard curves prepared with authentic C 6 -NBD-GlcCer.
b Recovery is expressed as a percent of total activity measured in the homogenate. c Measured in the absence of CHAPSO. When measured in the presence of 0.05% CHAPSO, activity was reduced by ϳ30%. same sample visualized by silver staining in lane 4. Three major bands (45, ϳ60, and 66 kDa) were visible in this fraction (Fig. 2, lane 5). The 45-kDa polypeptide was noticeably diminished in the UDP-Glc-eluted fraction, while the 60-and 66-kDa forms continued to predominate on the gel (Fig. 2, lane 6).
In an attempt to further purify GCS following dye-agarose chromatography, UDP-Glc-eluted GCS, including a radiolabeled aliquot, was centrifuged through a glycerol gradient. Preliminary experiments showed that only ϳ50% of the starting GCS activity was preserved after a 12-h ultracentrifugation in a glycerol gradient with 0.5 mM UDP-Glc present, preventing a quantitative evaluation of GCS enrichment. Nevertheless, the discrete peak of GCS activity in the glycerol gradient allowed us to assess the apparent relative size of the enzyme under nondenaturing conditions. Based on a standard curve of marker enzymes sedimented in the same tube, we estimated the sedimentation coefficient of GCS as ϳ4.2 s (Fig. 3A). The peak of GCS activity migrated to position immediately preceding malate dehydrogenase, which has a molecular mass of 70 kDa. Fig. 3B shows the glycerol gradient distribution of the radiolabeled 45-, 60-, and 66-kDa polypeptides present in dyeagarose purified GCS. The intensity of the 45-kDa polypeptide peaked earlier in the glycerol gradient than the peak of GCS activity (Fig. 3, B and C). Most of the radioactivity associated with the 66-kDa band was found to be rat albumin by immunoprecipitation; however, this technique did not completely remove the radioactivity in this region of the gel (data not shown). Furthermore, both the 60-and the 66-kDa bands peaked in intensity at fraction 11, which coincided with the peak of GCS activity (Fig. 3, B and C). Thus, although we cannot definitively rule out the possibility that another polypeptide is responsible for GCS activity, our data suggest that the ϳ60and/or ϳ66-kDa polypeptides are the GCS protein.

Characterization of GCS
Purified fractions obtained after the second dye-agarose column were used to examine the enzymatic characteristics and substrate specificity of GCS. GCS activity was previously reported to be stimulated by phospholipids (41,42). Thus, we first investigated effects of phospholipids on GCS activity. Purified GCS showed an almost absolute requirement for phospholipids for activity (Table II, Fig. 4). In contrast, GCS activity in the Golgi and crude CHAPSO extract was not stimulated significantly by phospholipids (data not shown). The optimal phospholipid concentration for stimulating GCS activity in purified fractions was 0.1 mM, with higher concentrations causing a relative suppression of activity (Fig. 4). Of the phospholipids tested, phosphatidylcholines showed the greatest ability to stimulate GCS activity, although phosphatidylethanolamine was almost as effective (Table II). Among phosphatidylcholines with different fatty acid moieties, GCS activity was maximally stimulated by those containing unsaturated fatty acids, with C 16:1 and C 18:1 fatty acids being the most effective (Table II).
Next we examined the specificity of GCS for various acceptor substrates. Among the ceramide analogs tested, the D-erythroand L-erythro-isomers of C 6 -NBD-Cer were the best substrates (Table III). Glucosylation of C 6 -NBD-Cer from a commercial source, which contains a mixture of both isomers, gave similar results to the pure stereoisomers (Table III). Neither D-nor L-threo-NBD-Cer were used as substrates by the enzyme, showing that the enzyme is stereospecific. D-erythro-C 6 -NBD-dihydroceramide was glucosylated to about 25% of control values obtained with C 6 -NBD-Cer (Table III). Among the different C n -NBD-Cer tested, the hexanoyl-and octanoyl-analogs were the best GCS substrates, while little or no measurable activity was found toward shorter (propanoyl-and pentanoyl-) and longer (tetradecanoyl-) ceramide analogs (Table III). Similarly, GCS was ϳ4.5 times more active toward short chain [ 14 C]C 6 -Cer than toward a longer chain [ 14 C]C 10 -Cer (data not shown). 2 However, the glucosylation yield of the radioactive C 6 -Cer was only ϳ25% of that of C 6 -NBD-Cer (data not shown). By contrast, C 5 -DMB-Cer containing a different fluorophore was a poor substrate with 21% glucosylation compared to C 6 -NBD-Cer (Table III). We also tested the effects of D-threo-PDMP, a known specific inhibitor of GCS (43). PDMP inhibited the ac-

various phospholipids on GCS activity
Dye-agarose-purified GCS, eluted in the absence of exogenous phospholipid, was incubated with 5 nmol of C 6 -NBD-Cer, 2.5 mol of UDP-Glc, 0.05% CHAPSO, and various phospholipids for 30 min at 37°C as described under "Experimental Procedures." Lipids were then extracted and separated by thin layer chromatography, and the amount of C 6 -NBD-GlcCer formed was quantified by image analysis. The results are means of Ն2 replicates for each lipid tested and are expressed as a percent of the amount of GCS activity measured in the presence of 0.1 mM DOPC. GCS 4. Effect of phospholipid concentration on GCS activity. Dye-agarose-purified GCS, eluted in the absence of exogenous phospholipids, was incubated in the presence of various concentrations of synthetic (di-C 18:1 , di-C 16:1 and C 16:0 , C 18:1 ) phosphatidylcholines as described in Table II.

TABLE III
Acceptor substrate specificity of GCS Dye-agarose-purified GCS was incubated with 5 nmol of various fluorescent acceptor substrates, 2.5 mol of UDP-Glc, 0.05% CHAPSO, and 0.1 mM DOPC for 15 min at 37°C. GCS activity was quantified as described in Table II  tivity of purified GCS ϳ45% at 1 M and 85% at 10 M (Fig. 5).
Finally, we evaluated the ability of GCS to use various donor substrates to glycosylate C 6 -NBD-Cer. Of the UDP-hexoses, GCS was able to utilize UDP-Glc efficiently but had little or no activity using UDP-glucuronic acid, UDP-galactose, UDP-Nacetylglucosamine, UDP-mannose, or UDP-xylose as hexose donors (Table IV). UDP-Glc was the best glucose donor among diphosphoglucose nucleotides, but TDP-glucose and CDP-glucose also were efficient glucose donors (Table IV). Surprisingly, ADP-glucose was also used as a substrate by GCS, leading to about 6% glucosylation of C 6 -NBD-Cer compared to UDP-Glc. DISCUSSION We report here for the first time a method for the purification of GCS. Several features of the method were critical for the successful purification of this enzyme. First, the modifications that we introduced in the homogenization and Golgi fractionation procedure (see "Experimental Procedures") significantly improved the enrichment and recovery of GCS activity relative to our previous work (16). Second, we found that inclusion of UDP-Glc and DOPC as protective agents improved the stability of GCS. Similarly, UDP and phospholipids were reported to stabilize UDP-Glc:dolichyl-phosphate glucosyltransferase (44); however, UDP had no protective effect on GCS activity (data not shown). Finally, we found that green dye-agarose could be used in a two-step procedure to purify GCS based on the selective binding of GCS to the dye-agarose in the absence, but not in the presence, of UDP-Glc. This procedure led to a ϳ20-fold enrichment of GCS in the UDP-Glc-eluted fractions and was the only chromatographic procedure that we tried that produced any enrichment in the enzyme. Dye-agaroses have been used previously in the purification of numerous proteins such as dehydrogenases, kinases, and serum proteins (for review, see Ref. 45) as well as UDP-Glc:dolichyl-phosphate glucosyltransferase (46). The observation that the binding of GCS to dye-agarose is inhibited by UDP-Glc suggests a competition between UDP-Glc and dye molecules for the active site of the enzyme. Similar results were described for another glycosyltransferase purified on dye-agarose (47).
Using the solubilized and purified GCS, we also obtained new information about the enzymatic characteristics of this protein. First, we found that purified GCS had almost no activity in the absence of exogenous phospholipid. Activity was restored by the addition of phospholipids, with the highest enhancement of activity observed with low concentrations (0.1 mM) of unsaturated, long chain (C 16:1 or C 18:1 ) phosphatidylcholine (Table II, Fig. 4). In contrast, phospholipids had little effect on GCS activity in the Golgi or CHAPSO-solubilized Golgi membranes. Presumably, endogenous phospholipids present in Golgi membranes were sufficient to stimulate maximal activity in the Golgi fractions. These endogenous phospholipids may have been depleted during purification of the enzyme by dye-agarose chromatography, causing an almost complete loss of GCS activity, which could be restored by exogenous phospholipids. A stimulating effect of exogenous phospholipids on glycosyltransferase activities was documented previously (42, 48 -50). More recently, stimulation of solubilized GCS activity by phosphatidylcholine was reported (51); however, the concentrations reported for optimal activity were ϳ100-fold higher (8 -10 mM) than the value that we report here (0.1 mM).
Our results on the specificity of purified GCS toward ceramide analogs are in good agreement with previously published studies concerned with GCS activity in cultured cells or membrane preparations. First, we found that GCS is stereospecific, utilizing erythro-but not threo-C 6 -NBD-Cer as a substrate, similar to results reported using cultured fibroblasts (18). We also found that D-erythro-C 6 -NBD-dihydroceramide was glucosylated to about 25% of control values obtained with C 6 -NBD-Cer (Table III). This result supports recent observations on the metabolism of C 6 -Cer analogs in Chinese hamster ovary cells (52). Second, we found that C 6 -and C 8 -ceramides were better substrates than ceramides with longer or shorter N-acyl chains, confirming observations reported for GCS activity in microsomal preparations (41). Finally, we found that C 6 -NBD-Cer is a better substrate for GCS than is [ 14 C]C 6 -Cer. By contrast, C 5 -DMB-Cer, containing a different fluorophore, was poorly glucosylated relative to C 6 -NBD-Cer (Table III), consistent with previous findings in cultured cells (53). We also examined the nucleotide specificity of GCS and found that, surprisingly, GCS is able to efficiently utilize CDP-Glc and TDP-Glc as glucose donors (see Table IV). While CDP-Glc and TDP-Glc are not naturally occurring glucose donors, earlier studies have also described UDP-Glc:glucosyltransferases that are able to utilize CDP-Glc and TDP-Glc (54,55).
In summary, we have presented for the first time a method for the purification of GCS, applied this method to isolate a ϳ10,000-fold enriched GCS fraction from rat liver, identified two polypeptides (60 -70 kDa) as likely candidates for the GCS protein, and further characterized the purified enzyme. This FIG. 5. Effect of D-threo-PDMP on GCS activity. Dye-agarosepurified GCS was preincubated with 2.5 mol of UDP-Glc, 0.05% CHAPSO, and 0.1 mM DOPC for 5 min at 37°C in the presence of D-threo-PDMP. The reaction was initiated by addition of 5 nmol of C 6 -NBD-Cer, and the results were quantitated as described in Table II and under "Experimental Procedures." The data are means of triplicate determinations and are expressed as a percent of control values.

TABLE IV
Donor substrate specificity of GCS Dye-agarose-purified GCS (after removal of UDP-Glc by gel filtration) was incubated with 2.5 mol of various NDP hexoses, 5 nmol of C 6 -NBD-Cer, 0.05% CHAPSO, and 0.1 mM DOPC for 15 min at 37°C. GCS activity was quantified as described in Table II and