N-azidoacetylmannosamine-mediated chemical tagging of gangliosides.

Peracetylated N-α-azidoacetylmannosamine (Ac4ManNAz) is metabolized by cells to CMP-azidosialic acid. It has been demonstrated previously that in this way azidosialic acid-containing glycoproteins are formed that can be labeled on the cell surface by a modified Staudinger ligation. Here, we first demonstrate that the same procedure also results in the formation of azidosialic acid-containing gangliosides. Deoxymannojirimycin, an inhibitor of N-glycan processing in proteins, decreases the total cell surface labeling in Jurkat cells by ∼25%. Inhibition of ganglioside biosynthesis with N-[5-(adamantan-1-yl-methoxy)-pentyl]1-deoxynojirimycin reduces cell surface labeling by ∼75%. In conclusion, exposure of cells to Ac4ManNAz allows in vivo chemical tagging of gangliosides.

Glycoconjugate-metabolizing glycosyltransferases and glycosidases have become important drug targets in recent years. Nature provides numerous lead compounds, primarily polyhydroxylated alkaloids (denominated iminosugars), that exert important biological activities by inhibiting specific glycosidases (1,2). Effective therapeutic strategies based on interfering with glycoprocessing enzymes have been described. Two iminosugar-based drugs are now used in the clinic. Miglitol (N-hydroxyethyldeoxynojirimycin), inhibiting the intestinal glycosidases sucrase and maltase, is used for the treatment of diabetes mellitus type II (3). Miglustat (N-butyldeoxynojirimycin), inhibiting the glycosyltransferase glucosylceramide synthase, is in use for the treatment of Gaucher disease (4)(5)(6).
Glucosylceramide synthase, the transferase responsible for the assembly of glucosylceramide from UDP-glucose and ceramide, is a key enzyme in the biosynthesis of neutral glycosphingolipids and sialic acid-containing gangliosides. More recently, the reduction of ganglioside levels was identified as a therapeutic approach for diabetes mellitus type II (7)(8)(9). Partial inhibition of glucosylceramide synthase, therefore, also appears as an attractive therapeutic target for the prevention and treatment of diabetes mellitus type II.
As part of our efforts to obtain effective glucosylceramide synthase inhibitors, we searched for a means to monitor the inhibitory effect of selected iminosugars on ganglioside biosynthesis in living cells. Bertozzi and coworkers (10-13) earlier developed a strategy for in vivo labeling of cell surface glycoproteins. Their approach is based on the finding that N-a-azidoacetylmannosamine is accepted by the CMP-sialic acid biosynthesis machinery. The resulting CMP-azidosialic acid in turn is recognized by sialic acid transferases, leading to the biosynthesis and cell surface expression of azidosialic acid containing N-linked glycoproteins. The azide can next be chemoselectively tagged by a modified Staudinger ligation. To promote the formation of azidosialic acid, Saxon and Bertozzi (10) established that peracetylated N-a-azidoacetylmannosamine (Ac 4 ManNAz) is vividly taken up by cells and efficiently metabolized to N-a-azidoacetylmannosamine by cytosolic esterases. The realization that gangliosides, like GM3, also contain a sialic acid residue at the nonreducing end led us to explore whether the strategy developed by Saxon and Bertozzi (10) would also have merit in cell surface labeling of gangliosides. Here, we demonstrate the validity of this with the finding that Jurkat cells cultured in the presence of Ac 4 ManNAz express azidosialic acidcontaining glycosphingolipids at levels at least as high as those of azidosialic acid-containing N-linked glycoproteins. We further show that cell surface labeling of azidosialic acid-containing glycoproteins and gangliosides can be suppressed independently by the proper selection of iminosugars: those that inhibit N-linked glycan processing mannosidases and those that inhibit glucosylceramide synthase, respectively.

Cell culture conditions
Jurkat cells were grown and maintained in RPMI 1640 medium with 2 mM L-glutamine (Biowhittaker, Baltimore, MD) containing 10% fetal calf serum (Gibco, Carlsbad, CA) at 5% CO 2 . Cells were seeded at ?1.5 3 10 5 ml in 5 ml flasks for flow cytometry or 50 ml flasks for lipid analysis. Cell viability was assessed during maintenance and before and after the various stages of labeling using trypan blue dye exclusion.
Both peracetylated N-azidoacetylmannosamine and N-[5-(adamantan-1-yl-methoxy)-pentyl]1-deoxynojirimycin were added to cell cultures from stock solutions in DMSO. The final concentration of DMSO in the culture medium was 1% (v/v). Deoxymannojirimycin (DMM; Sigma-Aldrich, Nieuwegein, The Netherlands) was added from stock solution in ethanol. As negative controls, equal volumes of the appropriate solvent were added to the cell culture.

Labeling of cell surface azides
After 3 days of incubation in the presence of 50 mM peracetylated N-azidoacetylmannosamine, cells were collected by centrifugation at 1,500 rpm for 10 min, washed three times in cold PBS, and resuspended in PBS containing 2% (v/v) fetal calf serum. The cells were distributed on a six-well plate in 1 ml, after which an equal amount of 0.5 mM biotin-phosphine in PBS was added. After incubation at room temperature for 3 h under mild shaking, the cells were collected by centrifugation and washed three times in cold PBS. Either the cells were labeled with streptavidin-FITC for the purpose of flow cytometry or total lipids were isolated as described below. FITC labeling was accomplished by incubation of cells with 1 ml of 1:1,000 streptavidin-FITC (Gibco) in PBS for 1 h in the dark at 4jC, after which the cells were washed three times in cold PBS. Flow cytometry was performed using a FACSscan (Becton Dickinson, Palo Alto, CA) with settings optimized for FITC fluorescence.

Isolation of gangliosides and ganglioside ligation product
Lipids were extracted with chloroform-methanol (1:1, v/v), and phase separation was performed according to Bligh and Dyer (16). The aqueous phase was evaporated to dryness under N 2 . The samples were desalted on a SPE C18 column (Bakerbond, Deventer, The Netherlands). In short, the dried fractions were dissolved in 1 ml of water containing 0.1 M NaCl (pH 4.5). The solution was applied to the column, which had been preequili-brated with 2 ml of the same watery solution. Subsequently, the column was desalted with 30 ml of water, after which the lipids were eluted with 20 ml of a mixture containing equal volumes of chloroform and methanol. The eluent was evaporated to dryness under N 2 .

In vitro Staudinger ligation of metabolically labeled gangliosides
Desalted gangliosides were dissolved in 500 ml of a mixture of equal volumes of chloroform and methanol containing phosphine-biotin in a final concentration of 1 mM. Next, 100 ml of water was added, and the reaction was allowed to proceed overnight at room temperature under mild stirring, after which the organic solvents and the water were evaporated to dryness under N 2 .

Ganglioside detection
Gangliosides were detected by analysis of the acidic glycolipid fraction obtained after Folch extraction using chloroformmethanol-water (65:25:4) as a solvent (17). Gangliosides were quantified after the release of oligosaccharides from glycosphingolipids by ceramide glycanase detection (18). The oligosaccharides were labeled at their reducing end with the fluorescent compound anthranilic acid (2-aminobenzoic acid) before analysis using normal-phase high-performance liquid chromatography.

Cell surface labeling of azidosialosides
Jurkat cells were cultured for 3 days in medium containing 50 mM Ac 4 ManNAz. Labeling of cell surface azidosialosides was performed as described in Materials and Methods. Cells were harvested and washed in labeling buffer and labeled for 3 h with phosphine-biotin. Next, cells were incubated with streptavidin-FITC, washed, and resuspended for flow cytometry analysis. Intense labeling of the cell surface was obtained by this procedure. The presence of Ac 4 ManNAz did not influence the rate of cell proliferation. Cell viability, as assessed by trypan blue exclusion, was not affected by the procedure. Very similar results were obtained with murine B16 melanoma cells cultured in DMEM containing 10% fetal calf serum, 100 U/ml penicillin, and 0.1 mg/ml streptomycin at 10% CO 2 (data not shown).

Distinction of azide-containing N-linked glycoproteins and gangliosides
To distinguish between the presence of azide moieties in N-linked glycoproteins and gangliosides, cells were cultured in the presence or absence of 1 mM DMM and 10 mM N-[5-(adamantan-1-yl-methoxy)-pentyl]1-deoxynojirimycin (AMP-DNM). DMM specifically inhibits the processing of high-mannose-type N-linked glycan to sialic acid-containing complex-type structures. We demonstrated previously that 1 mM DMM prevents the formation of complex-type glycan in glycoproteins in various cell types (21). Using radioactive motioning labeling, we observed that the presence of 1 mM DMM in the culture medium of Jurkat cells also completely blocked the conversion of EndoH-sensitive glycan to resistant structures in newly formed glucocerebrosidase molecules, indicating effective inhibition of glycan processing (data not shown). The presence of 1 mM DMM in the culture medium did not reduce the cell surface concentration of the ganglioside GM3, as detected by flow cytometry using antibody directed to the ganglioside.
AMP-DNM specifically inhibits the first step in glycosphingolipid biosynthesis catalyzed by glucosylceramide synthase (IC 50 ? 150 nM). Incubation of Jurkat cells with 10 mM AMP-DNM also completely inhibits in these cells the conversion of C6-NBD-ceramide to C6-NBD-glucosylceramide and subsequent glycosphingolipids (Fig. 1). We also observed by flow cytometry analysis that after 3 days of culture of Jurkat cells in the presence of 10 mM AMP-DNM, cell surface ganglioside GM3 is reduced by 80% (data not shown). AMP-DNM is known not to interfere with glycoprotein biosynthesis and processing (22).
Culturing of Jurkat cells for 3 days in the presence of 1 mM DMM and 0.05 mM Ac 4 ManNAz resulted in a 24.0% reduction of fluorescent cell surface labeling (Fig. 2). In a second independent experiment, the reduction of fluorescence by DMM was 27%. The presence of 10 mM AMP-DNM led to an 86.5% reduction of fluorescence (Fig. 2). In the second independent experiment, the reduction of fluorescence by AMP-DNM was 68%. The combined presence of DMM and AMP-DNM led to an almost complete (95%) loss of fluorescence in all experiments. Very similar observations were made with murine melanoma cells (data not shown). Our findings suggest that a very large proportion of the cell surface azidosialosides in Jurkat cells are found in sialic acid-containing glycosphingolipids.

Demonstration of the formation of azidoGM3
To further substantiate that azidogangliosides are indeed formed in Jurkat cells exposed to Ac 4 ManNAz, we isolated gangliosides and analyzed their oligosaccharides released by ceramide glycanase treatment and fluorescent labeling with anthranilic acid (AA). In Jurkat cells, the gangliosides almost exclusively consist of GM3 molecules. Figure 3 shows that two trisaccharides derived from GM3 were detected. The retention time of the first peak coincides with the normal sialic acid-galactose-glucose-AA generated from GM3. Incubation with phosphine reagent did not influence its chromatographic behavior. The second peak shifted in its retention time after incubation with the phosphine reagent, indicating that it indeed contains an azide moiety.

DISCUSSION
Our study reveals that exposure of Jurkat cells to Ac 4 ManNAz results in the formation of chemically tagged ganglioside, in particular the most abundant ganglioside, GM3. The proportion of azidoGM3 is ?60% of total GM3, suggesting a very efficient incorporation of azidosialic acid in gangliosides. This is not entirely surprising, because different forms of sialic acid occur in nature, species that are either acylated or glycolated at the N atom. Apparently, the azide group modification in sialic acid is equally well tolerated by the ganglioside biosynthetic machinery. It is of interest that the proportion of cell surface-tagged glycoproteins is actually lower than that of gangliosides. Again, this is not entirely surprising if one considers the estimated ratio of ganglioside to glycoprotein molecules at the cell surface. Gangliosides are largely located at the cell surface. Because nearly all ganglioside is GM3 in Jurkat cells, these cells contain per gram wet weight ?200 nmol of sialic acid associated with glycolipid. Assuming that ?1% of all cellular protein is cell membrane glycoprotein with an average mass of 50 kDa, Jurkat cells would contain ?20 nmol of glycoprotein per gram wet weight. To explain the observed ratio (3:1) of FITC-labeled ganglioside to glycoprotein, it would require on average at least three sialic acids per membrane glycoprotein.
The pioneering work of Bertozzi and colleagues (10-13) has led to a convenient procedure to chemically tag sialic acid-containing glycoconjugates that can be subsequently labeled at the cell surface. Our study reveals that this also includes gangliosides besides glycoproteins. This approach may have interesting applications. First, it offers a novel tool to screen synthetic and natural compounds that interfere in ganglioside biosynthesis and/or trafficking to the cell surface. Second, given the postulated role for GM3 in modulating insulin receptor-mediated signaling, it may be of interest to analyze the impact of various synthetic tags on this process, which is impaired in type 2 diabetes mellitus. Fig. 3. Analysis of oligosaccharides derived from GM3. Demonstration of azidoGM3 formation in cells cultured with Ac 4 ManNAz. Jurkat cells were labeled with Ac 4 ManNAz, and glycosphingolipids were isolated as described in Materials and Methods. Oligosaccharides were removed from glycosphingolipids by ceramide glycanase digestion, labeled with anthranilic acid, and separated by HPLC as described in Materials and Methods. Oligosaccharides were reacted with phosphine-biotin or incubated identically without the agent. The dotted line, a double peak of two closely related molecular species (a 1 and a 2 ), displays the oligosaccharides not exposed to the Staudinger reagent. The solid line represents the chromatogram of oligosaccharides exposed to the Staudinger reagent, revealing a selective shift of a 2 to b.