Glucosylceramide and Glucosylsphingosine Modulate Calcium Mobilization from Brain Microsomes via Different Mechanisms*

We recently demonstrated that elevation of intracellular glucosylceramide (GlcCer) levels results in increased functional Ca2+ stores in cultured neurons, and suggested that this may be due to modulation of ryanodine receptors (RyaRs) by GlcCer (Korkotian, E., Schwarz, A., Pelled, D., Schwarzmann, G., Segal, M. and Futerman, A. H. (1999) J. Biol. Chem. 274, 21673–21678). We now systematically examine the effects of exogenously added GlcCer, other glycosphingolipids (GSLs) and their lyso-derivatives on Ca2+ release from rat brain microsomes. GlcCer had no direct effect on Ca2+ release, but rather augmented agonist-stimulated Ca2+ release via RyaRs, through a mechanism that may involve the redox sensor of the RyaR, but had no effect on Ca2+ release via inositol 1,4,5-trisphosphate receptors. Other GSLs and sphingolipids, including galactosylceramide, lactosylceramide, ceramide, sphingomyelin, sphingosine 1-phosphate, sphinganine 1-phosphate, and sphingosylphosphorylcholine had no effect on Ca2+ mobilization from rat brain microsomes, but both galactosylsphingosine (psychosine) and glucosylsphingosine stimulated Ca2+ release, although only galactosylsphingosine mediated Ca2+ release via the RyaR. Finally, we demonstrated that GlcCer levels were ∼10-fold higher in microsomes prepared from the temporal lobe of a type 2 Gaucher disease patient compared with a control, and Ca2+ release via the RyaR was significantly elevated, which may be of relevance for explaining the pathophysiology of neuronopathic forms of Gaucher disease.

first and second messengers (1,2). Of the signaling pathways regulated by SLs, Ca 2ϩ homeostasis has received wide attention due largely to observations that sphingosine, sphingosine 1-phosphate, and sphingosylphosphorylcholine modulate Ca 2ϩhomeostasis via the edg receptors, a class of plasma membrane G-protein-coupled receptors (GPCRs) (3)(4)(5)(6). However, complex glyco-SLs (GSLs) and their lyso-derivatives ( Fig. 1) have also been implicated in regulating Ca 2ϩ homeostasis (7,8). Of the lyso-GSLs, galactosylsphingosine (GalSph, psychosine) (9 -13) has been shown to mobilize Ca 2ϩ from intracellular stores, possibly via activation of the ryanodine receptor (RyaR), the major Ca 2ϩ -release channel of the endoplasmic reticulum (ER), although this has never been unambiguously proven. Recent studies (reviewed in Ref. 14) have shown that complex GSLs, such as ganglioside GM1, can potentiate the activity of a nuclear envelope Na ϩ -Ca 2ϩ exchanger (15), and the sarco/endoplasmic reticulum Ca 2ϩ -ATPase (SERCA) can be modulated by gangliosides GM1 and GM3 (16,17). Since GSLs and lyso-GSLs accumulate in the sphingolipidoses, in which neuronal function is often severely impaired (18,19), and since altered Ca 2ϩ homeostasis has been implicated in a number of neurodegenerative diseases (20, 21), determination of the molecular mechanisms by which GSLs or lyso-GSLs modulate intracellular Ca 2ϩ signaling may be a prerequisite for determining the mechanism leading from GSL accumulation to neuronal cell dysfunction and/or death.
We recently demonstrated (22) that the simplest GSL, glucosylceramide (GlcCer), upon its accumulation in cultured neurons in a chemically induced model of type 2/3 Gaucher disease (the neuronopathic forms of Gaucher disease, Refs. 23 and 24), increases Ca 2ϩ mobilization from intracellular stores, presumably via the RyaR. As a result, neurons with elevated GlcCer levels showed enhanced sensitivity to agents that induce cell death via Ca 2ϩ mobilization (22,25). In the current study we systematically examine the effects of GlcCer, other GSLs and their lyso-derivatives on Ca 2ϩ release from isolated rat brain microsomes, and demonstrate that GlcCer and its lyso-derivative, glucosylsphingosine (GlcSph) (Fig. 1), both modulate Ca 2ϩ release but via different molecular mechanisms, and by a different mechanism to that of galactosylsphingosine (GalSph), none of which involve GPCRs. Moreover, Ca 2ϩ release was also enhanced in human brain microsomes obtained from a type 2 Gaucher disease patient, in which GlcCer levels are elevated ϳ10-fold.
Brain Microsomes-Wistar rats, obtained from the Weizmann Institute Breeding Center, were sacrificed, their brains removed, separated into cerebral cortex and cerebellum, rapidly frozen in liquid N 2 , and stored at Ϫ80°C. Microsomes (from 10 -12 gm of tissue) were prepared essentially as described (26) with some modifications. Tissue was suspended at a ratio of 1:4 (w/v) in ice cold 0.32 M sucrose, 20 mM HEPES-KOH, pH 7.0, containing 0.4 mM phenylmethylsulfonyl fluoride, leupeptin (0.8 g/ml), and aprotinin (1.4 TIU) (buffer A), and homogenized at 4°C using 8 up and down strokes of a rotating Potter-Elvehjem homogenizer. After centrifugation (700 ϫ g av , 10 min), the resulting pellet (P1) was gently resuspended in one-fourth of the original volume of buffer A, centrifuged (700 ϫ g av , 10 min), and the two supernatants pooled (S1). Mitochondria were removed by centrifugation (8,000 ϫ g av , 45 min) of S1 and the resulting supernatant (S2) centrifuged (115,000 ϫ g av , 90 min) to obtain a microsomal pellet (P3), which was resuspended in 0.4 -0.8 ml of buffer A. Protein was determined (27), and the microsomes subsequently flash-frozen in liquid N 2 . Microsomes were stored at Ϫ80°C and used for up to six months after their preparation, during which time there was no change in their activity with respect to Ca 2ϩ release and uptake. Human brain microsomes were prepared exactly as described for rat brain microsomes. Microsomes were prepared from a control human brain of a young adult and from the brain of a type 2 Gaucher patient who died at 1 year of age.
Spectrophotometric Assay of Ca 2ϩ Uptake and Release-Ca 2ϩ uptake and release was measured by a spectrophotometric assay using the Ca 2ϩ -sensitive dye, antipyrylazo III (26,28,30), with some modifications. Brain microsomes (330 g in 8-15 l of buffer A) were added to 0.95 ml of 8 mM NaMOPS, pH 7.0, 40 mM KCl, 62.5 mM K 2 HPO 4 , and 250 M antipyrylazo III, in a plastic cuvette containing a magnetic stir bar, to which 1 mM MgATP, 40 g/ml creatine phosphokinase, and 5 mM phosphocreatine, pH 7.0, were added. Ca 2ϩ uptake and release were measured in a Cary spectrophotometer (Varian Australia Pty Ltd.) at 37°C by subtracting the absorbance at A 790 from A 710 at 2-s intervals.
The amount of Ca 2ϩ released from microsomes was expressed as a percent of total Ca 2ϩ in the microsomes, which was obtained by summing Ca 2ϩ taken up during the Ca 2ϩ -loading period together with endogenous Ca 2ϩ from the microsomal preparation (measured separately after addition of a Ca 2ϩ ionophore, A23187 (2 M), without Ca 2ϩ loading). The rate of Ca 2ϩ uptake into microsomes was calculated by measuring the linear portion of the slope after addition of Ca 2ϩ , agonist, or lipid. Occasionally, spontaneous quantal Ca 2ϩ release (calcium sparks) was observed. Spontaneous Ca 2ϩ release was considered to be a spark when A 710 Ϫ A 790 increased by Ͼ0.002 over the baseline, with the baseline defined as the A 710 Ϫ A 790 value measured immediately before the spark. GlcCer and GlcSph Analysis-Lipids were extracted (34) from the same human temporal lobe microsomes used for Ca 2ϩ analysis. GlcCer and GlcSph were eluted in one fraction by aminopropyl solid phase chromatography using a LC-NH 2 cartridge as described (34). GSLs and lyso-GSLs were separated by weak cation exchange solid phase extraction using a LC-WCX cartridge, and GSLs subsequently deacylated by alkaline hydrolysis (1 M KOH in methanol, 100°C, 24 h). The resulting lyso-GSLs were acetylated using 5 mM acetic anhydride containing [ 3 H]acetic anhydride (2 Ci) and NaOH (4 mM) in chloroform/methanol (1:1, v/v), as were the lyso-GSLs obtained in the initial fractionation. A detailed account of this method will appear elsewhere. 2

RESULTS
Ca 2ϩ mobilization from rat brain microsomes was analyzed using the Ca 2ϩ -sensitive dye, antipyrylazo III. This dye has been used to measure Ca 2ϩ release from muscle (28,35,36), which contains high levels of RyaRs (37), and from canine brain (30). By using a rigorous homogenization procedure and by optimizing the recovery of microsomal membranes with respect to Ca 2ϩ uptake, we were able to use this dye to measure Ca 2ϩ release in rat brain microsomes, from which significantly lower levels of RyaRs can be recovered. Upon its addition to the cuvette, Ca 2ϩ accumulated in microsomes and could be released by palmitoyl CoA (Fig. 2A), a RyaR agonist (38,39), to a similar extent to that previously reported in canine brain (30).
Palmitoyl CoA-induced Ca 2ϩ release was enhanced upon preincubation with C8-GlcCer (10 M) by ϳ3-fold (Fig. 2B), and could be blocked by preincubation with 350 M ryanodine (Fig.  2C), a concentration similar to or lower than that used previously to block RyaR-mediated Ca 2ϩ release measured using antipyrylazo III (26,28,40). In contrast, C8-GlcCer did not induce Ca 2ϩ release by itself (Fig. 2D), demonstrating that GlcCer is not an agonist of the RyaR, but rather modulates its activity. LC-GlcCer enhanced Ca 2ϩ release to a similar extent to that of C8-GlcCer, using either palmitoyl CoA or GalSph as RyaR agonists (Table I and Fig. 3). C8-GalCer, over a range of concentrations (Fig. 3), and long-acyl chain GalCer (Table I), were completely ineffective in modulating agonist-induced Ca 2ϩ release, as were a variety of other sphingolipids (Table I), demonstrating a highly specific mode of sensitization of the RyaR by GlcCer.
Although GlcCer did not induce Ca 2ϩ release by itself (Fig.  2D), a significant increase in spontaneous quantal Ca 2ϩ release (calcium sparks) (41,42) was observed in the presence of C8-GlcCer. In untreated microsomes, ϳ2 sparks per hour were observed over the time-course of a typical experiment, which increased to ϳ6 sparks per hour in the presence of C8-GlcCer, both of which could be completely blocked by ryanodine (Table  II). The amount of Ca 2ϩ released per spark was higher in the presence of C8-and LC-GlcCer than in controls, although it was not statistically significant (Table II). Likewise, LC-GlcCer caused a significant increase in spark frequency, but no increase was observed with other GSLs. These data strengthen the idea that GlcCer sensitizes the RyaR.
We next examined the ability of GlcCer to mobilize Ca 2ϩ via other mechanisms. C8-GlcCer did not affect InsP 3 -induced Ca 2ϩ release from cerebellar microsomes, a rich source of the InsP 3 R (43), which could be blocked by the InsP 3 R antagonist, heparin (44) (Fig. 4A). Neither C8-GlcCer, C8-GalCer (Fig. 4B) Table II). Data are representative traces showing absorbance change (A 710 ϪA 790 ) of antipyrylazo III versus time, with an increase in absorbance demonstrating an increase in free Ca 2ϩ in the cuvette, and a decrease in absorbance demonstrating a decrease in free Ca 2ϩ due to microsomal Ca 2ϩ uptake. effect on the rate of Ca 2ϩ influx into microsomes via the Ca 2ϩ -ATPase, SERCA. Thus, we conclude that GlcCer specifically modulates Ca 2ϩ mobilization via the RyaR and not via the InsP 3 R or SERCA. Since neither pretreatment with C8-GlcCer, LC-GlcCer, or LC-GalCer had any effect on the B max or K D of Recent studies have demonstrated that RyaR activity can be enhanced by its redox state (45)(46)(47). Preincubation with the reducing agent, DTT, completely abolished the ability of C8-GlcCer (Fig. 5) to enhance agonist-induced Ca 2ϩ release, and blocked calcium sparks (Table II), suggesting that GlcCer may modulate the redox state of the RyaR via its redox sensor (48,49).
We next examined the extent of Ca 2ϩ release from brain tissue obtained from a Gaucher disease type 2 patient to determine the physiological significance of these findings. Previous studies have suggested that GlcCer accumulates in Gaucher brain tissue (50,51), but the extent of accumulation was highly variable, ranging from 5-80-fold compared with normal brains. Using a new method for separation of GSLs and lyso-GSLs (34), and a new method to quantify these lipids in which the free NH 2 group of lyso-GSLs is derivatized with [ 3 H]acetic anhydride, 2 an ϳ13-fold higher level of GlcCer was detected in microsomes prepared from the temporal lobe of a type 2 Gaucher patient (i.e. a neuronopathic patient) compared with a control brain, 3 and GlcSph was also detected in the Gaucher brain, although at levels ϳ5-fold less than GlcCer, with no detectable GlcSph in control brain microsomes (Table III). Intriguingly, palmitoyl CoA-induced Ca 2ϩ release from Gaucher brain microsomes was significantly higher than from human control brain microsomes, and could be reduced by DTT to control levels (Fig. 6). Thus, these data demonstrate a physiological and pathophysiological link between GlcCer accumulation in Gaucher brains and enhanced levels of Ca 2ϩ release via the RyaR.
In contrast to GlcCer, GlcSph and GalSph directly stimulated Ca 2ϩ release from rat cortical microsomes, albeit at concentrations of ϳ5-10-fold higher than GlcCer. Unexpectedly, and in contrast to the ability of ryanodine to block GlcCerenhanced, agonist-induced Ca 2ϩ release (Fig. 2C), ryanodine was unable to block GlcSph-mediated Ca 2ϩ release (Fig. 7A), and GlcSph-induced Ca 2ϩ release was not enhanced by GlcCer (not shown), suggesting that GlcSph mediates Ca 2ϩ release via a mechanism independent of the RyaR. However, ryanodine did inhibit GalSph-mediated Ca 2ϩ release (Fig. 7B), demonstrating that GalSph is a RyaR agonist. Neither GlcSph-nor GalSph-induced Ca 2ϩ release could be blocked (not shown) by preincubation with GDP␤S (100 M), an inhibitor of G-protein activation, or by pertussis toxin (1 g/ml), an inhibitor of the heterotrimeric G proteins, G i and G o , implying that GPCRs are not involved in mediating the action of GalSph and GlcSph on Ca 2ϩ release; this is further supported by the inability of sphingosine-1-phosphate, a GPCR agonist, and of sphinganine-1phosphate and sphingosylphosphorylcholine (all at 100 M), to induce Ca 2ϩ release from rat brain microsomes (not shown). DISCUSSION The major finding of the current study is that GlcCer mobilizes Ca 2ϩ from microsomes via a mechanism involving modulation of the activity of a major Ca 2ϩ channel of the ER, the RyaR. This finding could be of relevance for understanding the pathophysiology of neuronal forms of Gaucher disease, in which GlcCer accumulates. This contention is strongly supported by our observation that GlcCer accumulates in microsomes prepared from a type 2 Gaucher disease brain, in which Ca 2ϩ release is also enhanced. Remarkably, the molar concentration of endogenous GlcCer in the human brain microsomes was similar to that added exogenously to rat brain microsomes in order to enhance agonist-induced Ca 2ϩ release 4 via the RyaR.
The impetus for the current study was our earlier observation that upon its accumulation in cultured hippocampal neurons, GlcCer caused changes in neuronal functionality, inasmuch as a large increase in Ca 2ϩ release from intracellular stores was observed in response to glutamate or caffeine stimulation. Moreover, neurons were more sensitive to glutamateinduced neuronal toxicity and to toxicity induced via various other cytotoxic agents, which could be blocked by preincubation with antagonistic concentrations of ryanodine (22,25). Our current finding that GlcCer modulates agonist-induced Ca 2ϩ release from brain microsomes via the RyaR is consistent with these earlier observations, and the lack of effect of GalCer and other related GSLs and SLs might even imply that GlcCer plays a physiological role in regulation of the RyaR. This is further supported by the increase in the frequency of spontaneous quantal Ca 2ϩ release (sparks) upon incubation of microsomes with GlcCer. Ca 2ϩ sparks, sudden localized increases in intracellular Ca 2ϩ (52), have been observed in muscle and in brain (41,(53)(54)(55), and have been suggested to be of key importance in Ca 2ϩ signaling in the nervous system (56). The specificity of C8-GlcCer and LC-GlcCer to increase spark frequency strongly supports a central role for GlcCer in the regulation of Ca 2ϩ homeostasis via its sensitization of the RyaR. However, it is difficult to directly extrapolate the findings reported herein concerning sparks in brain microsomes to live neurons. Although we do not know, as yet, the precise molecular mechanism by which GlcCer sensitizes the RyaR, the ability of DTT to abolish GlcCer modulation of both Ca 2ϩ sparks and of agoniststimulated Ca 2ϩ release suggests that GlcCer may modulate the redox state of the RyaR via its redox sensor. A number of ion channels have been predicted to have an oxidoreductase domain (48). In the case of the RyaR, various cysteines within the channel have been proposed to regulate oxidative and nitrosative responses (46,57), and GlcCer may interact with the putative redox sensor (45).
If GlcCer acts as a physiological or pathophysiological modulator of the RyaR, which is located in the ER, GlcCer must presumably be present in this organelle, even though it is mainly synthesized distal to the ER, in the Golgi apparatus (58). However, our understanding of the intracellular distribution of GSLs may need to be re-evaluated in light of recent findings demonstrating that GlcCer affects a number of activities associated with the ER and other intracellular organelles (59,60). Indeed, a recent study demonstrated that sphingolipid-specific glycosyltransferases are found in a mitochondrialassociated ER subcompartment of rat liver (61) which could not be ascribed to contaminating Golgi apparatus membranes. Interestingly, the ceramide glucosyltransferase showed specificity for ceramide bearing phytosphingosine as sphingoid base, suggesting that different pools of GlcCer may be synthesized at different subcellular locations. In support of this are our current findings that microsomes obtained from human Gaucher brain tissue contains significant levels of GlcCer, although no  assessment of the purity of these microsomes was attempted due to limited availability of the human brain tissue.
GlcCer is the only GSL used in this study that sensitizes the RyaR, whereas GalSph and GlcSph both acted as agonists to stimulate Ca 2ϩ release from microsomes. Recently, GalSph was shown to induce Ca 2ϩ release from cultured cells (62) via a cell surface GPCR, T-cell death-associated gene 8 (TDAG8). However, TDAG8 is expressed at very low levels in brain and has a narrow tissue distribution (63). The lack of effect of GDP␤S and pertussis toxin on GalSph and GlcSph-induced Ca 2ϩ release from brain microsomes suggests that they do not act via GPCRs in microsomes, although the possibility that GalSph also binds to a cell surface orphan GPCR in neurons, as has been suggested in HL-60 cells (11), cannot be excluded. The lack of effect in microsomes of sphingosine-1-phosphate and sphingosylphosphorylcholine, which bind to cell surface GPCRs (6), together with the antagonistic effect of ryanodine on Gal-Sph-induced Ca 2ϩ release, support the notion that GalSph (9 -13) acts as an agonist of the RyaR, whereas GlcSph mediates Ca 2ϩ release via a mechanism independent of the RyaR.
Whether any or all of the effects of GalSph and GlcSph on Ca 2ϩ mobilization are of physiological relevance in normal cells is a matter of debate since lyso-GSL concentrations in cells are normally in the subnanomolar range (see Ref. 34). 5 However, lyso-GSLs accumulate at much higher levels in GSL storage diseases, such as Gaucher and Krabbe's disease, particularly in the brain, and the lyso-GSLs, rather than the GSLs, have been implicated in the mechanisms underlying disease pathology, especially neuropathology (34,64). Irrespective of the physiological relevance of lyso-GSLs in mobilizing Ca 2ϩ from microsomes, the specificity of GlcCer compared with both other GSLs and lyso-GSLs on Ca 2ϩ mobilization reported both in this study in microsomes and in our previous study in cultured neurons (22), and our recent study on activation of CTP:phosphocholine cytidylyltransferase by GlcCer (60), suggest that GlcCer is an important intracellular messenger that plays keys roles in both the regulation of phospholipid synthesis and in intracellular Ca 2ϩ homeostasis.