New Insights into Glycosphingolipid Functions—Storage, Lipid Rafts, and Translocators

Glycosphingolipids are key components of eukaryotic cellular membranes. Through their propensity to form lipid rafts, they are important in membrane transport and signaling. At the cell surface, they are required for caveolar‐mediated endocytosis, a process required for the action of many glycosphingolipid‐binding toxins. Glycosphingolipids also exist intracellularly, on both leaflets of organelle membranes. It is expected that dissecting the mechanisms of cell pathology seen in the glycosphingolipid storage diseases, where lysosomal glycosphingolipid degradation is defective, will reveal their functions. Disrupted cation gradients in Mucolipidosis type IV disease are interlinked with glycosphingolipid storage, defective rab 7 function, and the activation of autophagy. Relationships between drug translocators and glycosphingolipid synthesis are also discussed. Mass spectrometry of cell lines defective in drug transporters reveal clear differences in glycosphingolipid mass and fatty acid composition. The potential roles of glycosphingolipids in lipid raft formation, endocytosis, and cationic gradients are discussed.

Introduction

Eukaryotes and some bacteria contain a heterogeneous class of lipids called glycosphingolipids (GSLs). These lipids are composed of a ceramide backbone and a sugar headgroup (Fig. 1). The hydrophobic ceramide part, consisting of a sphingoid base and N‐linked fatty acid, inserts in a cellular membrane, whereas the hydrophilic sugar headgroup mostly faces the extracellular space. At the cellular level, GSLs organize protein function by forming glycosignaling domains, where clustering of specific proteins within bilayer‐spanning lipid rafts regulates their signal transduction (Hakomori and Igarashi, 1995). In addition, cells use the capacity of GSLs to form ordered domains to create selectivity in membrane sorting (Simons and van Meer, 1988).

GSLs exhibit a huge heterogeneity of structure. The sphingoid bases can vary in length, saturation, hydroxylation, and branching. The main sphingoid base in mammals is sphingosine (Karlsson et al., 1973). Sphinganine, which corresponds to sphingosine without the trans double bond, and phytosphingosine (hydroxyl‐sphinganine) are also common sphingoid bases. The fatty acid is amide‐linked to the amino group of the sphingoid base. The fatty acid species are cell‐type–dependent, can vary in length, saturation, and hydroxylation, but are mostly long (≥C16) and saturated. GSLs can be divided into two main classes based on the first sugar linked to the ceramide backbone, glucose or galactose.

The majority of glycosphingolipid synthesis in resting cells occurs from recycled breakdown products (e.g., sphingosine) derived from the lysosome (Gillard et al., 1998). It is unclear whether the lysosomal breakdown products, such as sphingosine, escape from lysosomes via a transporter or are able to passively slip across the lysosomal membrane. Transport to the endoplasmic reticulum (ER) may occur via discrete protein complexes in areas of close membrane contact between the late endosome and the ER (Holthuis 2005, Ko 2001). The accumulation of sphingosine in the lysosome of many GSL storage diseases (Rodriguez‐Lafrasse et al., 1994) suggests transport from the lysosome can be frustrated, but the consequences of this, if any, have yet to be determined. Once transported, sphingosine is phosphorylated at the ER, forming sphingosine‐1‐phosphate, an important intracellular messenger (Funato 2003, Ghosh 1994). Once dephosphorylated, sphingosine is then acylated to ceramide. In specialized cell types, luminal ceramide is then converted to GalCer. From the ER, ceramide can also follow the vesicular pathway to early Golgi compartments where it is converted to glucosylceramide (GlcCer). Alternatively, ceramide reaches the trans Golgi via CERT, a ceramide transfer protein that is able to extract ceramide from the ER and deliver it to the Golgi after docking to phosphatidylinositol‐4‐phosphate (Hanada et al., 2003). This pathway may also involve the membrane contacts between the ER and the trans Golgi (Ladinsky et al., 1994). Synthesis of the phosphosphingolipid sphingomyelin (SM) in the trans Golgi (Huitema et al., 2004) depends on this pathway for ceramide supply (Hanada et al., 2003). It has been reported that ER to Golgi transport of ceramide may be regulated via an ER sphingosine‐1‐phosphate phosphohydrolase. Anterograde transport of ceramide, probably via CERT, and at least some proteins are regulated by the degradation of ER sphingosine‐1‐phosphate (Giussani et al., 2006). Feedback regulation via the formation of sphingosine in the lysosome and lipid transport and synthesis is an interesting topic for future research.

In actively dividing cells “de novo” synthesis is more active and starts with condensation of serine and palmitoyl‐CoA to 3‐ketosphinganine (Fig. 2). Formation of sphinganine and subsequent acylation and desaturation on the cytosolic side of the ER produces ceramide without production of sphingosine as an intermediate. A variety of ceramide synthases have been identified with both sphingoid base (Venkataraman et al., 2002) and fatty acid preferences (LASS1–LASS6). LASS1 and LASS4 expression preferentially increase C18 ceramide, LASS2 mainly increases C16 and C24:1 ceramide, and LASS5 and LASS6 produce C16 ceramide (Mizutani 2005, Mizutani 2006, Riebeling 2003, Venkataraman 2002).

Ceramide is also the direct precursor for GlcCer. GlcCer is present in most eukaryotic cells and a few bacteria and serves as the major precursor for complex GSLs. It is synthesized by the UDP‐Glc:ceramide glucosyltransferase or GlcCer synthase CGlcT that occurs on the cytosolic side of Golgi membranes (Coste 1986, Futerman 1991, Jeckel 1992). Although the functions of GlcCer are unknown, in some yeast and bacteria, GlcCer may play an important role in the modulation of pH gradients; knocking out GlcCer synthase reduces growth in alkaline conditions (Rittershaus et al., 2006). Complex GSLs are made by the stepwise addition of individual sugars from their activated nucleotide precursors onto GlcCer. In mammals, the first reaction is the conversion of GlcCer to lactosylceramide by the LacCer synthase. This enzyme, like all further glycosyltransferases and sulfotransferases involved in the glycosylation of GSLs, acts in the Golgi lumen (Lannert et al., 1994). Sialoglycosphingolipids are also called “gangliosides,” and abbreviations have been assigned to them according to the number of sialic acids present and to their migration order in chromatography. Except for ganglioside GM4, which corresponds to NeuAcα2‐3GalCer, all gangliosides have LacCer as a precursor.

In contrast to sphinganine1‐phosphate, the synthesis of sphingosine‐ 1‐phosphate is dependent on the degradation of GSLs and sphingomyelin in late endosomes and lysosomes (Berdyshev et al., 2006). Once formed, sphingosine‐1‐phosphate is involved in signaling events such as the regulation of calcium gradients (Bagshaw 2005, Masgrau 2003). Termination of the signal proceeds via a sphingosine‐1‐phosphate phosphohydrolase (Le Stunff et al., 2002). The observation that exogenous addition of sphingosine‐1‐phosphate regulates cell proliferation (Zhang et al., 1991) is also consistent with the role of this second messenger in the control of proliferation. Overexpression of sphingosine kinase has been shown to activate autophagy, a process that occurs constitutively but is also activated in neurodegenerative diseases such as the secondary glycosphingolipid storage diseases (see Section IV.C). Whether there is any relationship between GSL storage and sphingosine‐1‐phosphate signaling has yet to be determined. Both sphingosine kinase 1 and sphingosine‐1‐phosphate phosphohydrolase modulate de novo sphingolipid synthesis (Berdyshev 2006, Le Stunff 2002). The interrelationships between sphingosine‐1‐phosphate signaling and the regulation of new membrane sphingolipid has yet to be determined.

Structural differences between GSLs and glycerolipids (Fig. 1) underlie the special behavior of GSLs in membranes. In GSLs, the region between the polar headgroup and the hydrophobic backbone contains chemical groups that can function both as hydrogen bond donor and hydrogen bond acceptor, in contrast to the glycerolipids, which only have hydrogen bond accepting properties in that part of the molecule (Pascher, 1976). Additional hydrogen bonding can occur between the sugar headgroups of GSLs. Another striking difference between GSLs and most glycerolipids is that the lipid chains are saturated over at least the first 15 carbons of both chains. In combination with the higher hydrogen bonding, the saturated nature results in denser packing. This is measured as an increased melting temperature, or T_m, which corresponds to the temperature above which a bilayer of a single lipid switches from a frozen state, the gel or solid‐ordered (s_o) phase, to a fluid state, the liquid‐crystalline or liquid‐disordered (l_d) phase. In addition, cholesterol, a rigid and flat cylindrical lipid, also interacts preferentially with sphingolipids via van der Waals interactions (Boggs, 1987). In model membranes containing mixtures of a high T_m lipid and cholesterol, a fluid‐fluid phase separation has been observed between the l_d and a liquid‐ordered l_o phase (Recktenwald and McConnell, 1981).

Unfortunately, most evidence to date that domains enriched in GSLs exist in biological membranes is indirect (Harder, 2003). However, GSLs are clustered on erythrocytes (Thompson and Tillack, 1985), gangliosides GM1 and GM3 are concentrated in domains (Parton, 1994), and GM1 can be enriched in different domains than GM3 on the same cell (Gomez‐Mouton et al., 2001). Most indications for the lipid domain association have been obtained by using the detergent‐resistance criterion. Although this technique has had great prospective value for how membrane signaling may work (Harder, 2003), physical studies have suggested there is no straightforward physical basis for why extraction at 4 °C would provide information concerning the situation at 37 °C (Heerklotz, 2002).

To avoid detergent extraction, pulse electron paramagnetic resonance (EPR) spin‐labeling methods and single‐molecule optical techniques have been applied (Dietrich 2002, Kenworthy 2000, Subczynski 2003) to monitor the entry and exit of probe molecules in domains in model membranes or in living cells. Results obtained using single molecule microscopy of saturated fluorescent analogues of phosphatidylethanolamine (Schutz et al., 2000) suggest that lipid analogues can be used to detect small (50 nm), stable (0.5–2 min) cell surface microdomains that may be related to the well‐established annular lipids that surround membrane proteins. Overall results have provided evidence for small/unstable rafts in unstimulated cells and for larger stabilized rafts induced by oligomerization of GPI‐anchored proteins or ligand binding (Harder, 2003), in which some proteins preferentially partition. The size of the confining domain for a GPI‐anchored protein is reduced when cells are treated with inhibitors of GSL synthesis; other studies have shown increases in phospholipase C susceptibility as well as changes in expression, suggesting GSLs contribute to the physical properties of microdomains. Using a fluorescent analogue of lactosylceramide, which is taken up by caveolar‐dependent mechanism, has shown that this specific form of endocytosis is dependent on the level of endogenous GSL, suggesting that some level of cell surface GSL is necessary for this mechanism of endocytosis, which may be lipid raft‐mediated (Cheng et al., 2006c). These processes are discussed in Section II.B and C. In conclusion, many studies suggest GSL can be heterogeneously organized on living cells, and this has functional consequences, although there is no strong consensus yet on the size, shape, and dynamics of lipid rafts.

One of the major setbacks in glycosphingolipid research has been a lack of tools with which to analyze them. Glycosphingolipids show increasing hydrophilicity with increasing size of the sugar headgroup, this leads to an increasing ability to form monomers in aqueous solution, decreasing solubility in organic solvents. Hence, difficulty arises with complete extraction from tissues and cells. Once extracted, utility of a phase split to purify the lipid components; higher GSLs escape into the aqueous phase which then leads to unacceptable losses. Higher GSLs are also more prone to adhere to the side of plastic tubes. The following techniques overcome many of these problems.

For tissue homogenates, GSLs can be extracted by the addition of 3.2 vol of CHCl₃/MeOH (1:2.2) for 10 min at room temperature, and the phases are split by the addition of 1 vol of CHCl₃ and 1 vol of H₂O (Bligh and Dyer, 1959). This quick extraction procedure gives similar recoveries of GSLs to other commonly used methods of lipid extraction (Miller Podraza 1992, van Echten 1990). Upper phase GSLs can then be recovered by SePak C₁₈ columns (Sillence et al., 2000). Briefly, C₁₈ SepPak columns (Waters, Milford, MA) are washed with 1 mL MeOH and 1 mL water. The CHCl₃ phase was dried down and loaded in 50‐μL of CHCl₃/MeOH/H₂O (1:2.2:1). The corresponding aqueous phase was loaded onto the columns and washed with 5 × 1 mL water. GSLs are then eluted with 5 × 1 mL CHCl₃:MeOH (1:3) and 1 mL of MeOH, and the samples are dried under nitrogen. At this point, the cholesterol and GSLs can be quantified as the following describes. Sometimes further purification is necessary (e.g., for mass spectrometry), in which case glycerophospholipids are removed from samples by saponification by the addition of 1 mL of chloroform and 1 mL of 0.2 M NaOH in methanol and incubated overnight at 37 °C and further purified by silicic acid chromatography (Vance and Sweeley, 1967). Columns are washed with 5 mL of CHCl₃ to remove fatty acids, nonalkali labile sterols, and alkylglycerols. Neutral GSLs are eluted with 6 mL acetone/MeOH (9:1). Gangliosides are eluted with 6 mL CHCl₃:MeOH (1:3) and the eluates dried under nitrogen. The samples are resuspended with 1 × 50 μL and then 1 × 100 μL of CHCl₃:MeOH (1:2.2) and transferred to a 1.5‐mL tube. For mass spectrometry, the samples are dried down in a SpeedVac and resuspended in 20 μL MeOH.

Matrix‐assisted laser desorption/ionization (MALDI) mass spectrometry can be performed as described previously (Hunnam et al., 2001). Briefly, a mixture of the sample (1 μL, 100 pmol) and matrix (1 μL of a saturated solution of 2,5‐dihydroxybenzoic acid in acetonitrile) are crystallized on the MALDI target. Positive ion reflectron MALDI spectra can then be obtained using a time‐of‐flight (TOF) mass spectrometer calibrated externally with hydrolyzed dextran sugars.

To dried lipid extracts, 10 μL incubation buffer (1 mg/mL sodium cholate in 50 mmol sodium acetate pH 5.0) was added. After vigorous vortexing and spinning in a benchtop picofuge, 10 μL of 50 mU/10 μL of ceramide glycanase (E.C. 3.2.1.123, Calbiochem, La Jolla, CA) in incubation buffer is added to cleave the glycans. Glucosylceramide is only partially digested due to the specificity of the glycanase (Wing et al., 2001). After this period, 10 μL of water is added to each enzyme digest followed by 80 μL of anthranilic acid and sodium cyanoborohydride to each digest and incubated in the 80 °C oven for 1 h. Derivatized oligosaccharides can be purified on DPA‐6S (Supelco, Ballafonte, PA) columns preequilibrated with 2 × 1 mL CH₃CN. One milliliter 97:3 CH₃CN:H₂O is added to each sample and vortexed prior to loading the samples onto the columns. Columns are washed with 4 × 1 mL 99:1 CH₃CN:H₂O and then with 0.5‐mL 97:3 CH₃CN:H₂O and the derivatized oligosaccharides eluted with 2 × 0.6 mL water into screw‐cap Eppendorfs and stored at 4 °C in the dark until ready for NP‐HPLC.

Because of the specificity of the glycanase, GlcCer cannot be quantitated using the procedure applied for higher GSL assay. Fortunately GlcCer can be quantitated using glucocerebrosidase instead of the ceramide glycanase. Due to the preponderance of free glucose in most samples, GlcCer is first purified on silicic acid columns. After GSL extraction and sample loading (see earlier), columns are washed with 2 mL of CHCl₃ and 2 mL of CHCl₃:MeOH 98:2. GlcCer is then eluted with 2 mL of CHCl₃:MeOH 97:3, 2 mL of CHCl₃:MeOH 96:4, 2 mL of CHCl₃:MeOH 95:5, and 2 mL of CHCl₃:MeOH 94:6. Samples are dried down and resuspended in 15 μL of 50 mmol sodium acetate buffer (pH 5) with 0.1% triton and 0.25% taurocholate; 0.4 U of purified glucocerebrosidase (Cerezyme or Ceredase) is added and incubated at 37 °C for 18 h. Samples of glucocerebrosidase can have significant oligosaccharide contamination; therefore, the enzyme has to be purified using a spin column (MW cut off 10 kDa). Samples are then derivatized using the same method as for the higher GSLs.