THE JOURNAL OF B~OLOC~CAL CHEMISTRY 0 1990 by The American Society for Biochemistry

and Molecular Biology. Inc.

Vol. 265, No. 21, Issue of July 25, pp. 12259-12266, 1990 Printed in U.S.A.

Progressive Hypoxia Inhibits the de Nouo Synthesis of Galactosylceramide in Cultured Oligodendrocytes*

(Received for publication, October 17,1989)

Ady

Kendler

and Glyn

Dawson
Research Center, Departments of Chicago, Chicago, Illinois of Pediatrics, 60637 Biochemistry, and Molecular

From the Joseph P. Kennedy, Jr. Mental Retardation Biology, and the Committee on Neurobiology, University

Neonatal rat oligodendrocyte (OLG) cultures exposed to 6 h of gradual, progressive hypoxia in a GasPak (BBL, Becton Dickinson) apparatus were not injured or metabolically impaired, but instead showed a specific inhibition of de nova synthesis (measured by [Hlpalmitic acid labeling) of the major myelin component galactosylceramide (GalCer). De novo synthesis of the Z-hydroxy fatty acid GalCer (HFA-GalCer) species, which requires OZ for its synthesis, was most severely inhibited (by 65%), while non-hydroxy GalCer species (NFA-GalCer) were less affected. The synthesis of membrane glycerophospholipids and sphingomyelin was unaffected by hypoxia. Treatment of OLG with 12 IIM oligomycin, an inhibitor of mitochondrial ATP synthesis, resulted in an inhibition (by 50-60%) of synthesis of all GalCer species. 13H]Palmitate labeling of NFA-ceramide, the ungalactosylated precursor of NFA-GalCer species, increased in both hypoxia and oligomycin treatments, suggesting that the conversion of newly synthesized ceramide to GalCer was blocked. Newly synthesized HFA-ceramide did not accumulate in OLG, but the small labeled HFA-ceramide pool present during hypoxia was not converted into HFA-GalCer. Pulse-chase studies indicated that NFAand HFA-ceramides labeled during these treatments were available for galactosylation and could be converted into GalCer upon reoxygenation. [3H]Galactose labeling of NFA-GalCer species was enhanced 2-fold in hypoxia, in contrast to the inhibition seen with [3H]palmitic acid labeling. Thus, while de nova GalCer synthesis was blocked in hypoxia, galactosylation of pre-existing ceramide pools was actually enhanced. Our evidence suggests that hypoxia results in a reversible inhibition of transport of newly synthesized ceramide from its site of synthesis to its site of galactosylation, but causes an increase in galactosylation of subcellular pools of pre-existing ceramide.

Myelination in the central nervous system is a perinatal event in which oligodendrocytes (OLG) elaborate membra* This investigation was supported by Grant HD06426 and Training Grant HD07009 of the National Institutes of Health and Grant PPO019 of the Multiple Sclerosis Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked aduertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: OLG, oligodendrocytes; HFA-GalCer, 2-hydroxy fatty acid form of galactosylceramide; NFA-GalCer, long chain non-hydroxy fatty acid form of GalCer; V.I. NFA-GalCer, very long chain non-hydroxy fatty acid form of GalCer; NFA-ceramide, non-hydroxy fatty acid form of ceramide; HFA-ceramide, 2-hydroxy fatty &id f&m oiceramide; SM, sphingomyelin; GlcCer, glucoiylce;amide; EMS, Matalons modified Eagles medium; CS, calf serum; C,

chloroform;

M, methanol; TLC, thin layer chromatography.

12259

nous sheaths which envelope many axons (1). Myelination is vital for the normal function of many central nervous tracts. OLG are highly active during myelination, synthesizing three to four times their weight in myelin membranes/day (2). The metabolic requirements of these cells are thought to exceed those of other brain cell types at this time (3), making them likely to be highly susceptible to injury caused by nutrient deprivation (starvation) and energy impairment (hypoxia, carbon monoxide poisoning). Clinically, normal myelin formation has been shown to be highly sensitive to nutrient and energy deprivation insults (4). The glycosphingolipid galactosylceramide (GalCer) constitutes about 30% of the lipid of myelin and is rapidly synthesized during myelination (5). More than half of the GalCer in myelin exists as a 2-hydroxy fatty acid form (hydroxy fatty acid GalCer, HFA-GalCer), and this modification is unique to myelin. GalCer is also unique in that it contains unusually long fatty acid moieties (6). Long (l&carbon) and very long (v.1.) (24-carbon) fatty acid moieties are found in non-hydroxy GalCer, hence the terminology NFA-GalCer (C-18) andv.1. NFA-GalCer (C-24). HFA-GalCer contains predominantly (C-24). It has been suggested that GalCer (especially HFA-GalCer) can form tightly packed bilayers by virtue of extensive intermolecular hydrogen bonding, and that this property is responsible for the close and regular compaction which is essential for the normal functioning of the myelin sheath (7). The accepted biosynthetic pathway for GalCer (depicted in Fig. 1) involves fatty acylation of sphingosine to form ceramide. Acylation with a non-hydroxy fatty acid will define the ceramide as NFAceramide; acylation involving a 2-hydroxy fatty acid will result in HFA-ceramide. UDP-Gal:ceramide:galactosyltransferase (EC 2.4.1.47) is the enzyme activity which catalyzes the galactosylation of ceramide to form GalCer (8). Galactosylation of NFA-ceramide is considered rate-limiting in NFAGalCer synthesis (9). However, it is 2-hydroxylation of free fatty acid, a reaction requiring molecular oxygen, that is thought to be the rate-limiting step in HFA-GalCer synthesis (10). In agreement with these theories of GalCer synthetic regulation is the observation that a large pool of NFA-ceramide is found in brain white matter, while HFA-ceramide is detectable in only trace amounts (11, 39). The locations in the cell of ceramide synthesis and galactosylation have been reported to be microsomal (12, 13). Studies of the cellular metabolism and trafficking of fluorescent derivatives of ceramide (14, 15) indicate that ceramide indeed exists as a precursor pool in a pre-Golgi area (possibly the endoplasmic reticulum), and that it is transported to the Golgi apparatus, where its conversion to more complex sphingolipids takes place (in cells other than OLG, ceramide is converted largely to glucosylceramide and to the phosphosphingolipid sphingomyelin, SM). Continued transport through the Golgi results in vesicular budding at the trans-Golgi and translocation to

12260
Pathway of

Effects of Hypoxia on Galactosylceramide

Galactosylceramide Biosynthesis

Synthesis in Oligodendrocytes

NFA-ceramide-----, Hydroxyletion 2-HydroxyFatty l cyl / UDP-palactose

NFA-GalCer

CoA \ -------->

HFA-ceramide----->

HFA-GaICer

1. Pathway of GalCer synthesis. Condensation of nonhydroxy-fatty acyl-CoA (NFA) or P-hydroxy-fatty acyl-CoA (HFA) with sphingosine yields NFA-ceramide or HFA-ceramide, respectively. Galactosylation of ceramide via a UDP-Gal intermediate results in formation of GalCer (NFA or HFA species). Note that the rate-limiting step in HFA-GalCer synthesis is thought to be the hydroxylation of the free fatty acyl-CoA, while the galactosylation of FIG.

NFA-ceramide (see text).

is considered rate-limiting

in NFA-GalCer

synthesis

the plasma membrane, much as occurs in glycoprotein transport and processing (16-18). Primary cultures of neonatal OLG have been used by a number of researchers to study OLG development in vitro (19). OLG have been shown to synthesize increasing amounts of myelin components (including GalCer) with time in culture; myelin component synthesis appears to follow a time course similar to myelination in the intact animal (reviewed in Refs. 19, 20). Similarly, cultured OLG have been reported to elaborate myelin-like membrane processes (21, 22), suggesting that these cells are a useful model for in uivo OLG development and myelination. We found that exposure of cultured cells to gradual, progressive hypoxia in the GasPak is a slow injury process which is ideal for study of early events in cell injury, events which otherwise might occur too quickly to be analyzed biochemically. In cultured rat neonatal OLG, we found that gradual, progressive hypoxia selectively inhibits synthesis of GalCer, and that this inhibition is an early event which precedes cell injury. Investigation of the mechanisms of this early event suggested that hypoxia inhibited the transport of newly synthesized ceramide from its site of synthesis to its site of galactosylation, resulting in an accumulation of ceramide. Galactosylation of other, pre-existing pools of ceramides was normal or elevated. An early inhibition of GalCer synthesis by hypoxia may be a reason why myelination is so sensitive to hypoxia clinically.
MATERIALS AND METHODS

Primary Oligodendrocyte Culture-OLG were isolated from neonatal rat brain by a method similar to a modification (23) of the method of McCarthy and devellis (24). Cerebral hemispheres were dissected from decapitated 3-day-old Sprague-Dawley rats (SASCO, Inc., Omaha, NE), the meninges stripped by rolling on filter paper, and major blood vessels removed with a small spatula. Although the dissection was not done under sterile conditions, brains were stored in ice-cold sterile Matalons modified Eagles medium (EMS) (GIBCO) until mincing. All media used contained 60 units/ml penicillin G and 0.06 mg/ml streptomycin. The tissue was minced under sterile conditions and trypsinized in 10 ml of EMS with 0.25% trypsin and 100 rg/ml DNAse at 37 C. The tissue was then centrifuged at 150 x g and resuspended in EMS with 10% calf serum (CS) (GIBCO), dissociated through a narrow bore Pasteur pipette and a sterilized 105 pm polypropylene screen (Spectramesh, Medical Industries, CA) and plated onto poly-L-lysine-coated 75 cm* T-type tissue culture flasks (poly-L-lysine from Sigma, 100 rg/ml in 0.1 M borate buffer, pH 8.5). The plating density was approximately 0.8 brains/flask. With frequent feeding (10 ml of EMS, 10% CS) a bed layer of astrocytes developed over 2 weeks time. At 2 weeks, numerous round, dark cells appeared on the surface of the bed layer (OLG). After a 100-rpm preshake on a rotating platform (Orbit Shaker, Labline Instruments, Melrose Park, IL), the flasks were fed 7 ml of fresh

EMS, 10% CS and shaken overnight at 200 rpm at 37 C in order to dislodge the OLG. The flasks were then rapped sharply bv hand 50100 times to further dislodge OLG, taking-care not to-allow the bed layer to be stripped off. This process greatly increases the OLG yield. The dislodged cells were harvested, resuspended in EMS, 10% CS with a narrow bore Pasteur pipette, dissociated through a 30 pm filter (NITEX, TETKO, Inc.), and replated onto 60-mm Primaria (Falcon, Becton Dickinson) poly-L-lysine-coated culture dishes. Approximately lo-14 60-mm dishes were plated from the shake-off, yielding about 50 ~cg protein/dish. After several hours all cells had - of OLG attached, and the media could then be replaced with Bottensteins defined media with 3% CS (25) supplemented with cytosine arabinoside (Sigma, 10 pM) to inhibit astrocyte and fibroblast proliferation. Over the next 3-5 days, the OLG developed extensive networks of processes. The cells were positively identified as OLG by immunofluorescent staining with antibodies against the specific myelin components GalCer (rabbit anti-GalCer IgG was a gift from Drs. Charissa Dyer and Joyce Benjamins, Wayne State University) and myelin basic protein (anti-MBP antisera was provided by Dr. Anthony Campagnoni, UCLA School of Medicine). Staining and visualization of cultures was done as described in Ref. 23. Cultures stained intensely with these antibodies, with staining of cell bodies, processes, and occasional membrane sheets apparent. In contrast, cultures stained with anti-GFAP antisera (glial acidic fibrillary protein, an astrocyte marker, antisera from Sigma) revealed a small number of large cells clearly identifiable as astrocytes. OLG did not stain for GFAP; based on these observations cultures could be identified as 80-90% pure OLG-enriched cultures. Cultured OLG also synthesized GalCer detectable biochemically (described below), while bed layer cells did not. Cultures were used in experiments approximately 1 week after their plating onto 60.mm dishes. The bed layers were refed EMS, 10% CS and could be shaken off again after several days incubation. Usually bed layers were shaken off two times and then discarded. The age of the cultures was defined as the age of the pups at dissection plus days in culture, and so OLG were used at approximately 25 days of age. Hypoxia ond Oligomycin Treatments-OLG cultures were subjected to hypoxia in a GasPak Anaerobic Chamber (BBL, Becton Dickinson) at 37 C. This system chemically scavenges oxygen and results in an atmosphere of less than 0.4% 02. and 5% CO2 in 100 min (26). Alternately,OLG were treated with oligomycin (Sigma, stored in solution in ethanol), 12 nM, for 4 h in serum-free EMS. Measurement of ATP Content-ATP levels were determined by a modification of the method of Cole et al. (27). OLG were homogenized by sonicating on ice with a Heat Systems/Ultrasonics sonifier (Plainview, NY). Sonicates (corresponding to 30 pg of protein) were extracted with 0.4 ml of 5% perchloric acid on ice for 10 min. The extracts were neutralized with 0.4 ml of 1 N KOH. The resulting precipitates were centrifuged, and the supernatants transferred to other tubes. The pellets were re-extracted as above and the supernatants pooled and neutralized (as indicated by Litmus paper) with HCl/NaOH. The extracts were kept on ice to limit ATP hydrolysis. 0.1 ml of each extract was allowed to reach room temperature for exactly 1 min in a glass scintillation vial, then was mixed with 0.1 ml of Luciferin/Luciferase assay mixture (Sigma, Kit FLAA, diluted lofold) and immediately counted for 10 s in a scintillation counter. The cpm of duplicates were compared with an ATP standard curve for determination of ATP content, in picograms. Labeling of Oligodendrocvte Lipids-OLG (one plate, corresponding to 50 pg of protein and approximately 0.3 million cells) were labeled with 19.10-3Hlnalmitic acid (60 Ci/mmol. Du Pont-New Eneland Nuclear), D-[6-3H]galactose (31.5 Ci/mmol, Amersham C&p.), [l-Clpalmitic acid (55 mCi/mmol, Amersham Corp.), or [l-C] lignoceric acid (a gift of Dr. Inderjit Singh) in 3.0 ml of the Bottenstein

(25) media, 3% CS or serum-free media. Typically

cells were fed the

radioactive media at the start of the experiment and labeled continuously through the 6 h of hypoxic (or control) conditions. At the end of the labeling period, cells were harvested in buffered saline, sonicated as described above, and aliquots removed for ATP and protein (28) determinations. Lipids were then extracted from the sonicates in 3 ml of chloroform/methanol (C/M, l:l, v/v) with nonradioactive GalCer (5 fig/sample, SUPELCO, Bellefonte, PA) added. The extract was adiusted to C/M (2:l) with the addition of 1.5 ml of chloroform, and washed with.l.3 ml of water and then theoretical upper phase (C/M/W. 3:47:48) as described bv Folch, et al. (29). Part or all of the organic phase was then subjected to mild alkaline methanolysis (0.7 N NaOH in C/M (2:1), at room temperature for 60 min) to destroy glycerolipids. Sphingolipids were chromatographed on 20 X 20-cm

Effects of Hypoxia

on Galactosylceramide

Synthesis in Oligodendrocytes

12261

borate-impregnated silicic acid thin layer chromatography (TLC) plates (30) three consecutive times in the solvent system C/M/H,0 (144:25:2.8) (31). This procedure separated GalCer into several species (described in Fig. 2 and Results) and separated GalCer from labeled glucosylceramide. Bands were identified by autoradiography using Enhance spray (Du Pont-New England Nuclear) and XOmat XAR-2 film (Eastman Kodak Co.) and quantitated by scraping and liquid scintillation counting. In other experiments carrier GalCer (5 rg) was added to samples spotted onto TLC plates and the bands identified after development by orcinol staining (0.2% orcinol in 2.0 M sulfuric acid, developed at 120 C for 10 min). The bands could then be scraped and counted as described. Orcinol staining resulted in a loss of approximately 20% of the dpm present in a band due to a charring effect; since all samples in an experiment were treated the same; this did not result in any experimental artifacts. Labeled ceramides were separated by TLC in C/M/acetic acid (glacial) (94:1:5) (11) and quantitated as described above. Ceramide standards were obtained from Applied Science, Inc., PA. Glycerophospholipds were chromatographed in C/M/H,0 (144:25:2.8), identified by their comigration with standards; visualized by iodine staining, and quantitated as above. Incorporation of PHJPalmitic Acid into Different Moieties of GalCer-Palmitic acid, a 16-carbon saturated fatty acid, is incorporated into newly synthesized sphingosine in cells, and is elongated to form the long chain fatty acids utilized in sphingolipid synthesis (see Fig. 1). To confirm that the [3H]palmitic acid used to label OLG is really incorporated into these moieties and so actually reflects de nouo synthesis of ceramides and GalCer the following procedure (based on (32)) was performed: [3H]palmitic acid-labeled GalCer species were chromatographed as described above, visualized with iodine, scraped, and eluted from the silicic acid with C/M (1:l). Individual GalCer species were transferred to a glass tube, dried, and resuspended in 0.6 ml of 3 N HCl in water. The tubes were heated at 110 C for 3 h to allow GalCer acid-hydrolysis to proceed. This procedure liberates free fatty acid, sphingosine, and galactose from glycosphingolipids. After hydrolysis, the reaction mixture was extracted with 3 ml of C/M (2:l) (29). The upper phase (containing galactose) was collected and quantitated by scintillation counting. The lower phase was divided into halves, half was chromatographed in the ceramide system above (ll), the other on the GalCer system above (31). The ceramide system separated free fatty acid from any labeled ceramides, GalCer, and sphingosine; thus, the fatty acid band was conveniently scraped and quantitated. The GalCer system separated sphingosine from any labeled GalCer, fatty acid, or ceramide bands; and so the sphingosine could be easily quantitated. Results indicated that NFA-GalCer species were labeled 35% in the fatty acid moiety, 41% in sphingosine, and 24% in galactose. HFA-GalCer was labeled 24% fatty acid, 54% sphingosine, and 22% galactose. These proportions did not change in hypoxia. These results indicate that indeed labeled palmitate is incorporated mostly into the backbone of the GalCer moiety, and so reflects de nouo synthesis of GalCer. Only a small amount of palmitate is recycled into galactose and used to galactosylate ceramide. [3H]Galactose incorporation into GalCer species occurs exclusively via galactosylation of ceramide, however, as no [3H]galactose was incorporated to a detectable extent into ceramide. Thus, GalCer species labeled with VHlaalactose mav reflect galactosylation of either de nouo synthesized ceramide or pre:existing pools of ceramide in the cell. UDP-Galactose:Ceramide:Galactosyltransferase (EC 2.4.1.47) Assay-Activity was determined by a method similar to that of Brenkert and Radin (33). The following were added to a 1.5sml microcentrifuge tube and dried under nitrogen, 0.23 nmol UDP-Gal (Sigma, in ethanol), 250,000 dpm of UDP[1-3H]galactose (DuPont-New England Nuclear, 40 mCi/mmol), 15 pg of phosphatidylcholine (Sigma, in C/M), 6 fig each of HFAand NFA-ceramides. The final composition of the incubation mixture (.035 ml) was 100 mM Tris-HCl, pH 7.4, 8 mM MgCl,, 1 mM EDTA, and 20 erg of OLG protein or water (for controls). The mixture was sonicated by sonifier probe, with 5 half-second bursts (setting 3-4). and vortexed carefullv. The mixture was incubated at 37 C for 1 h, and then was extracted as described above, chromatographed on silicic acid plates, and the species of GalCer quantitated by scraping and counting. Controls were always done to verify that no GalCer synthesis occurred in the absence of cell sonicate. UDP-Gal:ceramide:galactosyltransferase activity toward NFA-ceramide was always approximately 20% of activity toward HFA-ceramide. Low activity toward NFA-ceramide has been observed by a number of authors (33-36), and the reasons for this are unknown.

Incorporation of rH]Galactose into OLG Glycoproteins-OLG were labeled with 2 &i/ml of ?Hlealactose in Bottensteins media. 3% CS as described above, under- hypoxic or control conditions. The following procedure is based on Ref. 37: 50 rg of OLG protein, determined by the method of Lowry (28), was precipitated with icecold trichloroacetic acid (lo%), along with 200 rg of carrier albumin. After 2 h at 4 C, the precipitates were filtered through Whatman GF/C glass fiber filters that had previously been washed with 1 mM galactose in 10% trichloroacetic acid (to block nonspecific binding of the label). The filters were washed extensively with cold trichloroacetic acid, cold absolute ethanol, and cold diethyl ether (to remove lipids). This process leaves only proteins on the filters, which were then dried in air. The filters were then extracted in 1 N NaOH overnight, neutralized, and quantitated by liquid scintillation counting. Results were expressed in dpm/50 pg protein, after subtraction of blank (filter w/o protein) values. RESULTS

Hypoxia-An important principle that has emerged from the study of molecular mechanisms of cell injury is that injury is a gradual process, involving, in its early stages, specific biochemical events which represent the cells response to the insult. This response greatly precedes nonspecific terminal events in injury, such as mitochondrial death and membrane hydrolysis (38). While terminal injury events may be common to all cell types, the early responses may be highly individualized. Thus, it was important to determine the course of events in oligodendrocyte (OLG) hypoxia, focusing especially on unique, differentiated characteristics of OLG. Gradual, progressive, hypoxia in the GasPak for 6 to 10 h had no detectable effect on cell morphology; at approximately 12 h the cells began to lose their processes and greater than 80% appeared swollen (data not shown). After 6 h of hypoxia, cells in medium supplemented with 3% calf serum contained 81% of their normal ATP level (Table I). Interestingly, OLG in serum-free medium showed a more dramatic loss of ATP over time, with levels down to 50% of normal at 6 h and 44% at 10 h, suggesting that these cells were more sensitive to rapid energy impairment than cells in the presence of serum. OLG lipid metaboTime Course of Oligodendrocyte lism after 6 h of hypoxia is depicted in Table II. OLG that

had been made hypoxic in media with 3% serum for 6 h or control OLG were fed fresh, oxygenated, medium containing [Hlpalmitic acid. The cells were allowed to label for 2 h, and the incorporation of [Hlpalmitate into a variety of lipids determined. A number of membrane glycerophospholipids were labeled to the same extent as, or greater than, control cells labeled for 2 h. Labeling of the sphingolipid SM was also relatively unaffected by hypoxia (decreased by 15% in one experiment, but increased 18% in another). Incorporation into three species of GalCer, however, was dramatically reduced (to 55%, or less, of controls), despite the reoxygenated enviTABLE I OLG A TP levels during hypoxia Control or hypoxic cells were harvested, and 30 rg of sonicated protein extracted with perchloric acid and KOH as described under Materials and Methods. The neutralized extracts were then assayed for ATP content using a modified Luciferin/Luciferase procedure. ATP content is expressed below as a percent of control. In each experiment duplicates were within 5% of each other, and values below represent the average of two experiments except where indicated by a (*) (one experiment). Typical control ATP values are 3-5 pg of ATP/pg protein.
Time Condition 0 4 h 6 10 of hypoxia

Serum-free 3% serum

media

100% 100%

89%

50% 81%

44%

12262
Reoxygenated

Effects of Hypoxia on Galactosylceramide

OLG: TABLE II lipid metabolism

Synthesis in Oligodendrocytes

after hypoxia Control OLG or OLG that had been subjected to 6 h of hypoxia in Bottensteins media, 3% CS were fed oxygenated media containing 2 &i/ml [3H]palmitic acid. The cells were allowed to label for 2 h, and then harvested. Lipids were extracted and identified by TLC as described under Materials and Methods. Bands corresponding to specific lipids were scraped and quantitated by liquid scintillation counting. Data from a typical experiment is presented (in dpm/50 rg OLG protein) and is also given as percent labeling in hypoxic cells relative to that of controls. Percent values from a repeat experiment are given in the last column. PE, phosphatidylethanolamine; PA, phosphatidic acid; PC, phosphatidylcholine; PI, phosphatidylinositol; PS, phosphatidylserine.
Lipid species Experiment Control 1 Experiment Hypoxic 1 Experiment 1 Experiment 2 96 of control % of control

Gd\
GalCar v.I.NFA NFA HFA _ 3 OLG H-Palm Gal Cer Std OLG 14 C.Lign

FIG. by

2. Species

of galactosylceramide

(GalCer)

synthesized

PE+PA PC PI + PS SM V.I. NFA GalCer NFA GalCer HFA GalCer ronment.

160,837 13,668 13,652 1,633 3,417 3,260 2,377

201,499 14,357 15,679 1,384 1,202 1,782 1,027

125 105 115 85 35 55 43

129 140 185 118 22 38 36

These studies, along with ATP data and morphologic examination, suggest that the cells were not injured, nor even metabolically impaired by 6 h of hypoxia, but rather appeared to display a specific defect in the synthesis of the myelin lipid GalCer. Synthesis of other lipids, including another sphingolipid SM, was not inhibited by hypoxia. The rest of our studies focused on the inhibitory effect of hypoxia on GalCer synthesis. Synthesis of GalCer in Cultured OLG-Palmitic acid (16carbon, saturated) is considered a basic building block of sphingolipids. Palmitic acid is incorporated into newly synthesized sphingosine in the cell and is also elongated to the long chain fatty acids which are utilized by the GalCer synthetic pathway (Fig. 1) (8). The incorporation of [3H]palmitic acid into ceramide and GalCer in OLG, therefore, represents synthesis of new molecules (de nouo synthesis), and we have confirmed this by analyzing the distribution of label in the different moieties of [3H]palmitate-labeled GalCer (see Materials and Methods). Fig. 2 depicts OLG sphingolipid species separated by thin layer chromatography, as described under Materials and Methods. Based on their comigration with bovine brain GalCer standards (as in (8)), [3H]palmitatelabeled bands were identified as v.1. NFA-GalCer, NFAGalCer, and HFA-GalCer. [4C]Lignoceric acid, a 24-carbon saturated fatty acid, was selectively incorporated into the v.1. NFA but not the NFA-GalCer band (see Fig. 2), and Wlabeled stearic acid (18-carbons, saturated) selectively labeled the NFA-GalCer species (not shown), thus supporting the above designations. Of the HFA-GalCer species, a single band was labeled by both [3H]palmitic acid and [Cllignoceric acid, suggesting a 24-carbon (and possibly also 18-carbon) fatty acid composition. In this study, cultured rat OLG, aged 25 days, incorporated radiolabeled [3H] and [i4C]palmitic acid into GalCer in 6 h such that the NFA/HFA ratio was between 2:l and 4:l (typically 15,000 dpm in NFA, 5,000 dpm in HFA). Effect of Hypoxia on de Nova Galcer Synthesis-Fig. 3A shows that 6 h of hypoxia in the GasPak apparatus caused a decrease in labeling with [3H]palmitic acid of HFA-GalCer by 65%, while NFA-GalCer forms were less affected (NFA decreased 39%, v.1. NFA by 27%). Although the decrease in labeling of the NFA-GalCer forms was variable, from as little as lo-25% to as much as 30-50%, the effect on HFA-GalCer was always more dramatic; the reduction in its labeling being approximately twice the reduction of the NFA species in that

OLG. This is a composite picture of OLG sphingolipid species separated on borate-impregnated thin layer chromatography plates (30) three times in chloroform/methanol/water (144:25:2.8) as described in Ref. 31 and under Materials and Methods. The left-most lane depicts an autoradiograph of [3H]palmitic acid-labeled OLG GalCer species, so designated due to their comigration with bovine brain GalCer standards (middle lane), according to Ref. 31. The rightmost lane depicts an autoradiograph of OLG GalCer species labeled with [C]lignoceric acid (24-carbon, saturated); this pattern confirms our designation of the top-most NFA-GalCer band as very long chain non-hydroxy-fatty acid-GalCer. While labeled lignoceric acid is not incorporated into the long chain non-hydroxy-fatty acid-GalCer (NFA-GalCer) band, [Clstearic acid (18-carbon, saturated) is incorporated into NFA-GalCer (not shown), confirming the designation of this band. A single 2-hydroxy-fatty acid-GalCer (HFA-GalCer) band is labeled both with nalmitic acid and lignoceric acid by OLG, suggesting a 24- (and possibly 18-) carbon fatty acid composition. Note that this system separates glucosylceramide (GlcCer) from GalCer; [3H]palmitate labels both GlcCer bands very weakly (10% of GalCer labeling), but [Cllignoceric acid labels the upper, very long chain species quite well.

particular experiment. [3H]Palmitic acid labeling of glycerolipids (including membrane phospholipids and di- and triglycerides) during 6 h of hypoxia was unaffected (data not shown); similarly labeling of sphingomyelin during hypoxia (Fig. 3A) was not different from controls. This data agrees with the results seen in Table II (the reoxygenation study), in that hypoxia appeared to inhibit GalCer synthesis without an inhibition of other lipids. Fig. 3A also suggests that there was a differential effect on the three GalCer species during the course of hypoxia, with the inhibition of HFA-GalCer more dramatic than the inhibition of NFA-GalCer species. Since hypoxia caused a reduction in OLG ATP content (by 19%, see Table I) as well as a reduction in oxygen available to the cells, it was of interest to determine whether an ATP decrease alone was capable of mimicking the effect of hypoxia on GalCer synthesis. The mitochondrial respiratory inhibitor oligomycin was chosen to address this question because oligomycin acts directly on the mitochondrial F,-ATPase enzyme and does not interfere with electron transport processes in the cell. OLG were treated with 12 nM oligomycin for 4 h in serum-free media; otherwise labeling conditions and lipid analysis procedures were exactly the same as in the hypoxia experiments. Oligomycin (12 nM) was used because this dose had a small effect on OLG ATP content (approximately a 10% decrease in ATP), while still actively inhibiting GalCer metabolism. Higher doses of oligomycin caused greater ATP depletion (36 nM caused a 50% depletion in cells) but inhibited GalCer synthesis completely (not shown). Fig. 3B shows that oligomycin (12 nM) caused a global inhibition of GalCer synthesis relative to controls, with v.1. NFA-GalCer reduced by 70 f 6%, NFA-GalCer by 58 f 9%, and HFA-GalCer by 58 + 6%. Oligomycin also inhibited SM synthesis by 46%. These results suggested that 12 nM oligomycin did not mimic the selective effect of hypoxia on de nouo HFA-GalCer synthesis. All oligomycin experiments henceforth were performed

Effects of Hypoxia

on Galactosylceramide

Synthesis in Oligodendrocytes
TABLE III

12263

20

40

60

SO

100

[3H]-palmitate

incorporated

(%

of

control)

Effect of hypoxia and oligomycin on OLG ceramide synthesis Cultured OLG were labeled as described under Materials and Methods with [3H] or [Wlpalmitate during 6 h of hypoxia or 4 h of 12 nM oligomycin at 37 C (HFA-ceramide could be detected only with [4C]palmitate labeling, as described in the text). The cells were extracted and the ceramides analyzed by TLC as described. Bands corresponding to HFA or NFA-ceramides were scraped and quantitated by liquid scintillation counting. The results were normalized to cell protein and expressed as % of control labeling. All values are averages of measurements made in two or three independent experiments. Typical values for control cells were NFA-ceramide, 8000 dpm/50 yg protein, with both isotopes (4000 in oligomycin experiments), and HFA-ceramide, 500 dpm/50 pg (14C only). Condition Ceramide
(HFA or NFA) Incorporation

% of control
Hypoxia HFA NFA HFA NFA 108 f 5% 134 f 11% 103 + 0%
210 + 70%

Oligomycin

10

20

30

40

50

60

70

so

90

100

[3H]-palmitate

incorporation

(%

of

control)

FIG. 3. Effect of hypoxia and oligomycin on de now OLG GalCer synthesis. A, 25-day-old rat OLG cultures were labeled with [3H]palmitic acid (2 &i/ml) in Bottensteins defined media, 3% calf serum continuously during 6 h of gradual hypoxia. Cells were harvested and extracted as described under Materials and Methods, the glycolipids chromatographed, autoradiographed, scraped, and quantitated by liquid scintillation. The results, once normalized to 50 Kg of protein, are represented as % label incorporation (relative to controls) per species of GalCer and sphingomyelin (SM). Standard errors are included. Students t test indicated that the labeling during hypoxia of HFA-GalCer was significantly different than the NFA GalCer labeling, with p < 0.01. Typical values for control incorporation are 10,275 dpm/50 rg protein for v.1. NFA-GalCer, 5000 dpm/50 rg for NFA-GalCer, and 5100 dpm/50 fig for HFA-GalCer species. B, OLG were labeled with [3H]palmitate (2 &i/ml) in serum-free media in the presence of 12 nM oligomycin for 4 h. Cells were then harvested and their lipids analyzed as in A. Results are expressed as % of control + range. Typical control values are 2100 dpm/50 fig for v.1. NFAGalCer, 1800 dpm/50 pg for NFA, and 2100 dpm/50 fig for HFAGalCer. in serum-free

labeled with [Wlpalmitic acid, as it labeled so weakly (at approximately 5% of NFA-ceramide labeling). Previous studies have detected only trace amounts of HFA-ceramide in white matter (11, 39). Table III shows that labeling of this small HFA-ceramide pool did not appear to be affected by hypoxia. However, since the HFA-ceramide pool did not accumulate the dpm lost from HFA-GalCer (up to 2000 dpm/ 50 pg protein) this suggested that HFA-ceramide synthesis did not proceed at its normal rate during the inhibition of HFA-GalCer synthesis. Oligomycin (12 nM) similarly resulted in no detectable change in HFA-ceramide labeling (Table III).

Pulse-Chase Studies on /Cl or pH]Palmitic

Acid-labeled

EMS

and all hypoxic

treatments

in EMS,

3%

serum.

Cells-The results in Table III suggested that the increased dpm in the labeled NFA-ceramide pool during hypoxia or oligomycin treatment represented molecules of ceramide that were not being converted into NFA-GalCer. To confirm that this labeled ceramide pool was actually available for conversion into GalCer, pulse-chase experiments were performed as follows: OLG were treated with 12 nM oligomycin or hypoxia for 4 or 6 h, respectively, in the presence of [Cl or [3H] palmitate. The cells were then washed twice with media, fed fresh (oxygenated) medium and incubated for 18 h. Fig. 4A shows the results of a typical experiment with oligomycin (12 nM). Labeling of the NFA-ceramide pool increased dramatically during treatment (from 3580 to 10,420 dpm/50 Kg protein), far beyond a value corresponding to dpm lost from

Effect of Hypoxia on de Nouo Synthesis of Ceramide-Table III shows the effects of both hypoxia and oligomycin on the incorporation of [3H]palmitic acid or [WJpalmitic acid into ceramide, the ungalactosylated precursor of GalCer (Fig. 1). Hypoxia resulted in an increase in labeled NFA-ceramide in OLG of 34 + 10% over controls. The labeled NFA-ceramide pool in OLG is very large, containing about as many dpm of [Hlpalmitic acid as v.1. NFA-GalCer and NFA-GalCer combined (approximately 8000 dpm/50 pg protein/6 h labeling), in agreement with previous observations that a large NFAceramide pool exists in white matter (11). The actual dpm lost from NFA-GalCer species in hypoxia were more than accounted for by the increase in dpm in the NFA-ceramide pool, suggesting that NFA-ceramide was labeled with [3H] palmitic acid normally or at an enhanced rate during hypoxia,
but that its metabolism to GalCer was somehow blocked.

Oligomycin treatment (12 nM) resulted in a larger and more variable increase in labeling of the large NFA-ceramide pool (by 110 f 70%). HFA-ceramide, the precursor to HFAGalCer, could only be detected upon autoradiography when

NFA-Galcer (1595 dpm/50 pg), suggesting that an enhanced incorporation of label into NFA-ceramide was occurring in addition to an accumulation of ungalactosylated NFA-ceramide. Chasing for 18 h reduced the NFA-ceramide pool, however, along with a simultaneous increase in NFA-GalCer by approximately the same amount as chased from ceramide (approximately 5500 dpm/50 pg). Thus, it appeared that the accumulated labeled NFA-ceramide was readily converted into GalCer upon restoring the cells to aerated conditions. The same pulse-chase results were obtained with hypoxia, although the increase in NFA-ceramide labeling in hypoxia was not as dramatic (see Table III). The increase in labeling of NFA-ceramide in hypoxia or oligomycin treatment beyond a value corresponding to dpm lost from GalCer raised the possibility that the apparent inhibition of the conversion of NFA-ceramide to GalCer may have been an artifact due to a change in the specific activity of the NFA-ceramide pool. To investigate this possibility, an experiment was done to compare the rate of chasing of NFA-ceramide into GalCer in

12264

Effects of Hypoxia on Galactosylceramide moo IA I

Synthesis in Oligodendrocytes

control

HypoxlOlipom.18

Chase

n
E

NFA-wr NFA-GalCer

0
[3H]-galactose
FIG.

100

200

incorporation

(%of

control)

a z P E + 2 0 E E ,

1200, R 1000 800 6ofJ 400 200 0 control HypoxlOligom. 6 h Chase H HFA-cer HFA-Ga!cer

5. [3H]Galactose labeling of GalCer during hypoxia. OLG were labeled continuously with [3H]galactose (2 &X/ml) in Bottensteins media, 3% serum during hypoxia. Cells were harvested and GalCer species analyzed as in Fig. 3A and under Materials and Methods. Dpm incorporated into GalCer species were normalized to 50 pg of protein, and depicted as a percent of control values k range (the data is an average of two experiments, done in duplicates). Typical values for controls were 900 dpm/50 pg for v.1. NFA-GalCer, 1200 dpm/50 Kg for NFA, and 500 dpm/50 rg for HFA-GalCer.

FIG. 4. Pulse-chase studies on [3H]/[4C]Palmitate-labeled OLG. OLG cultures were treated with hypoxia (6 h) or 12 nM oligomycin (4 h) as described in Fig. 3, in the presence of radiolabeled palmitate (2 &i/ml). The cultures were then washed twice, fed fresh (aerated) media, and allowed to incubate a designated time before harvesting. Ceramide and GalCer analysis was carried out as described under Materials and Methods, in Table III and Fig. 3. The experiments below depict typical data for 12 nM oligomycin; similar results were obtained with hypoxia, though the increase in NFAceramide was not as dramatic as with oligomycin (Table III). A, chasing of NFA-ceramide into NFA-GalCer. This experiment is representative of four experiments done. 13H]Palmitate is the label; each bar depicts dpm normalized to 50 pg of cell protein and is the average of duplicates that were within 5-10% of each other. B, chasing of HFA-ceramide into HFA-GalCer. This experiment is representative of two experiments done. [C]Palmitate is the label required to detect HFA-ceramide (note difference in scale compared with Fig. 4~). Each bar depicts dpm and is an average of duplicates that were within 510% of each other.

control and hypoxic cells. The results of this study showed that control OLG chased labeled NFA-ceramide into NFAGalCer at a rate of 8745 dpm/50 pg protein/6 h chase, and hypoxic OLG at a rate of 7792 dpm/50 pg protein/6 h chase. These numbers, which are 11% apart, suggest that that the specific activities of labeled NFA-ceramide pools in hypoxic cells were virtually the same as in controls. NFA-ceramide was therefore as available for galactosylation in hypoxic or oligomycin-treated OLG as it was in controls, but its conversion to GalCer was blocked. Fig. 4B indicates that the small pool of labeled HFAceramide (about 500 dpm; note difference in scales in Fig. 4) was not affected by either hypoxia or oligomycin, despite the decrease in labeling of HFA-GalCer. However, upon reoxygenation, this pool is readily chased into HFA-GalCer in only 6 h of chasing. Thus, it seemed that the limited amount of HFA-ceramide synthesized during hypoxic or oligomycin treatments could not be converted into GalCer, as was seen with the NFA-ceramide pool above. Galactosytation of Ceramide Pools: fH]Galactose Labeling of GalCer-The inhibition of conversion of both HFA- and NFAceramides into GalCer in hypoxic or oligomycin-treated OLG prompted an investigation of the possibility that these conditions inhibited the galactosylation of ceramide, and so blocked GalCer synthesis. A possible mechanism for such an inhibition would be a decrease in cytosolic levels of UDP-Gal,

whose synthesis depend on ATP. [3H]Galactose was incorporated by OLG into the three GalCer species described in Fig. 2, at a ratio of approximately 51 (total NFA/HFA, corresponding to 1000 dpm/50 pg protein in NFA and 200 dpm/50 pg in HFA). Because galactosylation is the terminal event in GalCer synthesis, the incorporation of [3H]galactose into GalCer may reflect galactosylation (via a UDP-[3H]Gal intermediate) of preexisting pools of ceramide in the cell as well as ceramide made de nouo during the labeling period. [3H]Galactose was not incorporated to any significant extent into ceramides (data not shown), and so is a reflection only of galactosylation of ceramide, and not de nouo GalCer synthesis. Fig. 5 shows the results of labeling OLG with [3H] galactose during 6 h of hypoxia. The incorporation of label into v.1. NFA and NFA-GalCer species was enhanced more than 2-fold (223% of control for v.1. NFA-GalCer and 251% for NFA-GalCer), suggesting increased galactosylation of some NFA-ceramide pool. The labeling of HFA-GalCer was not increased by hypoxia, however, and so suggested that the apparent galactosylation observed was not an artifact due to a change in the specific activity of UDP-[3H]Ga1 in the cytosol during hypoxia. This was confirmed by studies on the the incorporation of [3H]galactose into trichloroacetic acid-precipitable material (glycoproteins) in hypoxic cells (described under Materials and Methods). Results indicated that galactosylation of proteins in hypoxic cells was not different than in control cells (2000 dpm/50 pg protein). Thus, it appeared that the inhibition of galactosylation of de nouo synthesized ceramide in hypoxic cells was not due to a depletion of UDP-Gal. To test the possibility that the observed galactosylation of pre-existing ceramide pools was due to an enhanced UDP-Gal:ceramide:galactosyltransferase activity in the cells, in vitro activities of homogenates of control and hypoxic cells were determined as described under Materials and Methods. Table IV indicates that no change in activity could be detected by this assay.
DISCUSSION

We present evidence that the synthesis of the major myelin glycolipid galactosylceramide (GalCer) by developing OLG is compromised by hypoxia. Recent work in cell injury has shown that while the terminal events of cell injury and death are common to many cell types (Ca influx and membrane lipid peroxidation, for example) (38, 40), different cell types respond to injurious stimuli in unique, specific ways that occur well before injury. These responses may represent de-

Effects of Hypoxia on Galactosylceramide

TABLE IV The effect of 6 h of hypoxia on UDPGa~:ceramide:galactosy~transferase activity in OLG OLG were harvested after 6 h in either hypoxic or control conditions, sonicated, and 20 pg of protein was incubated with NFA and HFA ceramide substrates and UDP-f3H]Gal in buffer for 1 h at 37 C (as under Materials and Methods):The mixture was extracted with chloroform/methanol and the linids senarated by TLC as described. Typical results are expressed as dpm of [3H]GaiCer produced by 20 rg of OLG protein (an average of duplicates), and also as enzyme activity (pmol (~10~~) of UDP-Gal incorporated into GalCer/20 lg protein/h). Chntd
GalCer species dmn Activity dam Activitv Hypoxia

Synthesis in Oligodendrocytes

12265

NFA HFA

318.0 1605.0

3.0 16.0

274.0 1596.5

2.7 15.9

and ATP conservation maneuvers in hypoxic skeletal muscle (42), or unfortunate vulnerabilities, such as excitatory amino acid release by ischemic hippocampal neurons (43), or prostaglandin release by complement-injured OLG (44). GalCer, in particular HFA-GalCer, is considered an important ingredient for proper compaction of myelin (7). The early and specific inhibition of GalCer synthesis by hypoxia presented here may be the reason why myelination in the developing brain is so sensitive to energy impairment, such as occurs in carbon monoxide poisoning (4). The work presented here provides evidence that 6 h of progressive hypoxia in the GasPak apparatus did not injure or metabolically impair developing rat OLG, as shown by morphologic examination, minimal effects on ATP content (in the presence of serum), and lipid metabolism studies (Table II). OLG did not appear to be injured until 12 h of hypoxia, and this places them well within the injury time scale reported for other cultured brain cells, with neurons at 8 h (45) and astrocytes at 18 h (46). It is unclear why cultured cells are so resistant to hypoxia, although it is likely that the relatively uncrowded nature of the cell monolayer allows for an excess of nutrients and oxygen/cell. Such reserves, coupled to the gradual increase in hypoxia over time, stretch out the time course of cell injury. This is very useful, as it allows study of the cells early and specific responses to hypoxia. 6 h of hypoxia did not inhibit the de nouo synthesis ([3H] palmitic acid incorporation) of glycerophospholipids or the phosphosphingolipid SM but did dramatically inhibit the de nova synthesis of HFA-GalCer (by 65%) and less so the synthesis of NFA-GalCer (v.1. NFA by 27%, NFA by 39%) (Fig. 3). The depletion of molecular 0, by hypoxia could explain the selective inhibition of HFA-GalCer relative to NFA-GalCer species, since the 2-hydroxylation of fatty acids has been shown to be dependent on molecular oxygen (10). We then determined whether the small (19%) decrease in ATP during hypoxia could be responsible for the observed inhibition of NFA-GalCer synthesis. The mitochondrial inhibitor oligomycin acts directly on the F,-ATPase in mitochondria to decrease ATP synthesis without interfering with electron transport processes. A low (12 nM) dose of oligomycin decreased cell ATP levels by 10% and was found to inhibit synthesis of all species of GalCer by 60-70%, and SM by 50%. Balch et al. (47) have previously shown that the transport of vesicular stomatitis protein protein from the endoplasmic reticulum of cells to the Golgi apparatus for processing is completely and reversibly blocked by an equally small (15%) lowering of cellular ATP content. To test the hypothesis that the lo-20% decrease in ATP caused by both hypoxia and 12

fense mechanisms, expression (41)

such

as

neuronal

heat

shock

protein

nM oligomycin was preventing the transport of NFA-ceramide from its site of synthesis to its site of galactosylation, the labeling of NFA-ceramide (normally present as a large precursor pool in OLG) in hypoxia and oligomycin treatment was examined. [3H]Palmitate labeling of NFA-ceramide was indeed increased by 34% in hypoxic cells, and this increase was more than large enough to account for the decrease in dpm from the v.1. NFA and NFA-GalCer species combined (Table 3). A similar accumulation of ceramide has been reported in BHK-21 cells treated with 2-deoxyglucose, a competitive inhibitor of glycosylation (48). An increase in NFAceramide labeling was also seen following oligomycin treatment. Since the increase in labeling of the NFA-ceramide was often considerably larger than the dpm lost from the GalCer species (Fig. 4A), it appeared that both hypoxia and oligomycin were causing an enhanced incorporation of [3H]palmitic acid into NFA-ceramide, beyond a value corresponding to an accumulation of unconverted ceramide. Pulse-chase studies showed that this accumulated pool of labeled NFAceramide was certainly auailczble for galactosylation, as it was converted into GalCer upon chasing with aerated media (Fig. 4A). Similarly, the rate of chasing of NFA-ceramide into GalCer was virtually the same in previously hypoxic cells as in controls, indicating that the apparent inhibition of conversion could not be explained by a change in the specific activity of the NFA-ceramide pool available for galactosylation. Studies on the incorporation of [Hlgalactose into GalCer species (Fig. 5) and into OLG glycoproteins indicated that the inhibition of conversion of NFA-ceramide to NFA-GalCer was not due to a decrease in available UDP-Gal. Further, it was determined that VDP-Gal: ceramide: galactosyltransferase activity in hypoxic OLG homogenates (Table IV) was not decreased relative to controls. It is hypothesized that the likely site of accumulation of NFA-ceramide in hypoxic OLG is the endoplasmic reticulum. Studies on the cellular distribution and metabolism of a fluorescent ceramide analogue, CG-NBD-ceramide (14, 15) have shown that CG-NBD-ceramide accumulates in perinuclear structures when fed to cells at 0 C. With warming of the cells to 20 or 37 C, fluorescence moves to the Golgi apparatus. This step is associated with the conversion of ceramide to its natural anabolic products, which in fibroblasts are glucosylceramide and SM. We propose that while synthesis of ceramide in OLG is localized to the endoplasmic reticulum (corresponding to the perinuclear or pre-Golgi compartment observed in Refs. 14 and 15), the galactosylation of ceramide occurs in the Golgi apparatus, and that transport of ceramide to the Golgi is required for galactosylation. It is this ATP-dependent transport that we suggest is inhibited by hypoxia and oligomycin treatment, resulting in the accumulation of NFA-ceramide. The observation that ATP depletion alone (by oligomycin treatment) did not mimic the selective effect of hypoxia on HFA-GalCer synthesis (Fig. 3) suggests that depletion of O2 in hypoxia was dramatically inhibiting the 2-hydroxylation of fatty acids, a rate-limiting step in HFA-GalCer synthesis (Fig. 1). The data presented in Table 3 support this hypothesis: it indicates that synthesis of HFA-ceramide was inhibited by hypoxia, since it did not accumulate during the inhibition of HFA-GalCer synthesis. Oligomycin treatment also appeared to inhibit HFA-ceramide synthesis, however, and so indicates that an ATP depletion alone may block fatty acid 2-hydroxylation, although not as effectively as 0, depletion. It may be that ATP depletion in oligomycin treatment results in cellular conservation of 0, for respiration, a process which may involve down-regulation of fatty acid hydroxylation. Results in

12266

Effects of Hypoxia

on Galactosylceramide

Synthesis in Oligodendrocytes

Fig. 4 suggest that the small pool of HFA-ceramide synthesized in hypoxic OLG accumulates in a pre-Golgi compartment, as was seen for NFA-ceramide. This small HFA-ceramide pool was also readily chased into HFA-GalCer in 6 h after reoxygenation (Fig. 4B). Evidence in the literature suggests that compartments of ceramide galactosylation other than microsomal do in fact exist in OLG. A number of researchers have presented evidence that galactosylation may occur in myelin itself (49%51), and since myelin is an extension of OLG processes, this suggests galactosylation may occur in OLG plasma membrane. Similarly, both plasma membrane and Golgi apparatus preparations have been shown to be capable of synthesizing SM (14, 15, 52). Our apparently paradoxical data indicating that the incorporation of [3H]galactose into NFA-GalCer actually increased 2-fold in hypoxia, while de nouo synthesis of NFA-GalCer ( [3H]palmitate incorporation) was inhibited, could be explained by the existence of two such subcellular pools of NFA-ceramide. Galactosylation of plasma membrane (pre-existing) pools of ceramide could be occurring at a normal or enhanced rate during hypoxia, while newly synthesized ceramide would be trapped in a pre-Golgi compartment, and prevented from being galactosylated in either the Golgi or the plasma membrane.
Acknowledgments-We and Dr. Robert Farrer would like to thank for valuable discussions. REFERENCES 1. Peters, A., and Vaughn, J. E. (1970) in Myelination (Davison, A. N., and Peters, A., eds), pp. 3-79, Charles C. Thomas, Springfield IL 2. Norton, W. (1981) in Basic Neurochemistry (Siegel, G. J., Albers, R. W., Agranoff, B. W., and Katzman, R., eds) pp. 63-92, Little, Brown, Boston 3. Kennedy, C., Grave, G. D., Jehle, J. W., and Sokoloff, L. (1970) Neurology 20,613-618 4. Leviton, A., Gilles, F. H., and Dooling, E. C. (1983) in The Developing Human Brain: Growth and Epidemiologic Neuropatho1og.y (Leviton, A., Gilles, F. H., and Dooling, E. C., eds) WrightyPSG, Boston 5. Norton, W., and Poduslo, S. E. (1973) J. Neurochem. 21, 759773 6. Radin, N. S., and Akahori, Y. (1961) J. Lipid Res. 2,335-341 7. Abrahammson, S.. Dahlen, B.. Lofgren, H., Pascher, I.. and S. Sundell (1976) Nobel Symposium 34, l-21, Goteberg, Sweden 8. Kishimoto, Y. (1983) in The Enzymes (Boyer, P. D., ed) pp. 358385, Academic Press, Inc., New York 9. Radin, N. S. (1972) in Sphingolipids, Sphingolipidoses, and Allied Disorders (Volk. B. W.. and Aronson. S. M.. eds) DD. 475-486. Plenum Publishing Co.: New York _H., and Y. Kishimoto (1987) Int. J. Biochem. 19, 10. Shigematsu, 41-46 P., Dulaney, J. T., and Moser, H. W. (1973) 11. Sugica, M., Connoly, Lipids 8, 401-406 12. Shah, S. N. (1971) J. Neurochem. 18,395402 13. Sato, C., Black, J. A., and Yu, R. K. (1988) J. Neurochem 50, 1887-1893 R. E. (1989) Methods Cell Biol. (part A) 75-85 14. Pagano, 15. van Meer. G.. Steltzer. E. H. K.. Wiinaendts-van-Resandt, R. W., andSimons, K. (1987) J. Cell Sk. 105,1623-1635 A. (1988) J. Biol. Chem. 263, l-4 16. Kornberg, Dr. Larry Hancock

17. Beckers, C. J., Keller, D. S., and Balch, W. E. (1987) Cell 50, 523-534 18. Bergmann, J. E., and Singer, S. J. (1983) J. Cell Biol. 97, 17771787 19. Pfleffer, S. (1984) Neurochem. 5, 233-298 20. Pleasure, D., Kim, S. U., and Silberberg, D. H. (1984) Neurochem. 5, 175-197 21. Rome, L. H., Bullock, P. N., Chiappelli, F., Cardwell, M., Adinolfi, A. M., and Swanson, D. (1986) J. Neurosci. Res. 15, 49-65 22. Szuchet, S., Stefansson, K., Wollman, R. L., Dawson, G., and Arnason, B. G. W. (1980) Brain Res. 200, 151-159 23. Dyer, C., and Benjamins, J. A. (1988) J. Neurosci. 8, 4307-4318 24. McCarthy, K. D., and de Vellis, J. (1980) J. Cell Biol. 85, 890902 25. Bottenstein, J. E. (1986) Proc. Natl. Acad. Sci. U. S. A. 83, 19551959 26. Seip, W. F., and Evans, G. L. (1980) J. Clin. Microbial. 11, 226233 27. Cole, H. A., Wimpenny, J. W. T., and Hughes, D. E. (1967) Biochem. Biophys. Acta 143,445-453 28. Lowry, 0. H., Rosebroueh. N. J.. Farr, A. L., and Randall, R. J. (19ii) J Biol. Chem. i&3,26&275' 29. Folch. J.. Lees. M.. and Sloane Stanlev. I G. H. (1957) J. Biol. Chkm. i26,497-509 30. Igisu, H., Takahashi, H., Suzuki, K., and Suzuki, K. (1983) Biochem. Biophys. Res. Commun. 110,940-944 31. Morell, P., and Radin, N. (1969) Biochemistry 8, 506-512 32. Rosenberg, A., and E. Chargaff (1958) J. Biol. Chem. 233,13231326 33. Brenkert, A., and Radin, N. S. (1972) Brain Res. 36, 183-193 34. Koul, O., Chou, K. H., and F. B. Jungalwala (1980) Biochem. J.

186,959-69
35. Morell, P., Constantino-Ceccarini, E., and Radin, N. S. (1970) Arch. Biochem. Biophys. 141, 738-748 36. Basu, S., Schultz, A. M., Basu, M., and Roseman, S. (1971) J. Biol. Chem. 246,4272-4279 37. Rapaport, R. N., and J. A. Benjamins (1981) J. Neurochem. 37, 164-71 38. Jennings, R. (1981) Am. J. Pathol. 102, 239-291 39. Kishimoto, Y.. and Mitrv, M. T. (1974) Arch. Biochem. Biophys. _ _

161,426-34

-.

B.. and Gutteridae. J. M. C. (1985) Trends Neurosci. 40. Halliwell. 8,22-26 41. Subjeck, J. R., and Shyy, T. T. (1986) Am. J. Physiol. 250, part 1, Cl-C6 42. Harris, K., Walker, P. M., Mickle, D. A., Harding, R., Gatley, R., Wilson, G. J., Kuzon, B., McKee, N., and Romaschin, A. D. (1986) Am. J. Physiol. 250, H213-H220 43. Goldbere. M. P.. Weiss, J.. Pham. P. C.. and Choi. D. (1987) J. Pharmhcol. Eip. The;. 243, 784-791 Y., Imagawa, D. K., and Shin, M. L. (1987) J. Neurochem. 44. Shirazi, 48, #1,271-278 45. Rothman, S. (1984) J. Neurosci. 4, 1884-1891 46. Yu, A. C. H., Gregory, G. A., and Chan, P. H. (1989) J. Cereb. Blood Flow Metab. 9, 20-28 47. Balch, W. E., Elliot, M. M., and Keller, D. S. (1986) J. Biol. Chem. 261,31,14681-14689 48. Suzuki, Y., Ecker, C. P., and Blough, H. A. (1984) Eur. J. Biochem. 143,447-453 E., and Suzuki, K. (1975) Brain Res. 93, 49. Constantini-Ceccarini, 358-362 50. Koul, O., Singh, I., and Jungalwala, F. B. (1988) J. Neurochem. 50, #2, 580-588 L. E., Benjamins, J. A., and Skoff, R. P. (1984) J. 51. Townsend, Neurochem. 43,139-145 D. R., and Kennedy, E. P. (1982) Biochemistry 21, 52. Voelker,
I

2753-2759