Multiple sphingolipid abnormalities following cerebral microendothelial hypoxia



Hypoxia has been previously shown to inhibit the dihydroceramide (DHC) desaturase, leading to the accumulation of DHC. In this study, we used metabolic labeling with [3H]-palmitate, HPLC/MS/MS analysis, and specific inhibitors to show numerous sphingolipid changes after oxygen deprivation in cerebral microendothelial cells. The increased DHC, particularly long-chain forms, was observed in both whole cells and detergent-resistant membranes. This was reversed by reoxygenation and blocked by the de novo sphingolipid synthesis inhibitor myriocin, but not by the neutral sphingomyelinase inhibitor GW-4869. Furthermore, oxygen deprivation of microendothelial cells increased levels of dihydro-sphingosine (DH-Sph), DH-sphingosine1-phosphate (DH-S1P), DH-sphingomyelin (DH-SM), DH-glucosylceramide (DH-GlcCer), and S1P levels. In vitro assays revealed no changes in the activity of sphingomyelinases or sphingomyelin synthase, but resulted in reduced S1P lyase activity and 40% increase in glucosylceramide synthase (GCS) activity, which was reversed by reoxygenation. Inhibition of the de novo sphingolipid pathway (myriocin) or GCS (EtPoD4) induced endothelial barrier dysfunction and increased caspase 3-mediated cell death in response to hypoxia. Our findings suggest that hypoxia induces synthesis of S1P and multiple dihydro-sphingolipids, including DHC, DH-SM, DH-GlcCer, DH-Sph and DH-S1P, which may be involved in ameliorating the effects of stroke .


Progressive hypoxia leads to the accumulation of several dihydrosphingolipids in cerebral microendothelial cells. Hypoxia also increases sphingosine-1-phosphate and the activity of glucosylceramide (Glc-Cer) synthase. These changes reverse by inhibiting the de novo sphingolipid synthesis, which worsens hypoxia-induced endothelial barrier dysfunction and apoptosis, suggesting that the identified sphingolipids may be vasculoprotective.

Abbreviations used

acid sphingomyelinases


blood–brain barrier














detergent-resistant membranes




glucosylceramide synthase inhibitor




human cerebral endothelial cells




high-performance thin-layer chromatography


liquid chromatography/mass spectrometry/mass spectrometry


3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide


neutral sphingomyelinases


phosphate-buffered saline








sphingomyelin synthase


sphingosine kinases



Stroke is a prevalent medical emergency that constitutes the leading cause of disability worldwide. Despite our growing understanding of the pathogenesis of stroke, intravenous thrombolysis remains the only treatment for this devastating condition. Because only 4–6% of all the stroke patients qualify for this treatment, new treatments are desperately needed (Go et al. 2013). Significant changes in the metabolism of sphingolipids have been described post-stroke. In ischemia–reperfusion models, for example, ceramide and other dihydrosphingolipids increase in the reperfusion phase (Nakane et al. 2000; Yu et al. 2007; Hankin et al. 2011; Novgorodov and Gudz 2011). In addition, we have previously observed the elevation of ceramides in the CSF from stroke victims, particularly in those with poor neurological outcome (Testai et al. 2012). The pathophysiological implications of these observations, however, have not been fully elucidated. At the cellular level, hypoxia has been shown to inhibit dihydroceramide (DHC) desaturase activity leading to the accumulation of long-chain DHC in tumor cells and in rat lung endothelium, but the role of altered sphingolipids in cerebral endothelial cells under hypoxia remains unknown (Devlin et al. 2011).

The cerebral endothelium is a key component of the neurovascular unit and operates in concert with neurons, vascular muscle cells, and pericytes, to maintain cerebral homeostasis. Endothelial dysfunction has been associated with neurovascular uncoupling, and endothelial death affects the integrity of the blood–brain barrier (BBB) and increases cellular permeability by loosening tight junctions in stroke (del Zoppo 2010). Similar changes have been described in association with traditional vascular risk factors such as hypertension and diabetes (Olmez and Ozyurt 2012). In this context, it has been proposed that strategies pertaining to prevent or reverse endothelial cell dysfunction and death could be critical for stroke prevention and survival. Immortalized endothelial cultures are increasingly used as in vitro BBB models. In addition, they have been used to study the effect of hypoxia in vitro and to investigate cell-mediated degenerative processes including auto-antibody attack, which can easily penetrate a damaged BBB/BNB barrier, leading to destruction of the CNS/PNS and propagating the disease progression (Dohgu et al. 2007; Yan et al. 2008; Yang et al. 2010; Dasgupta et al. 2011; Burek et al. 2012). Therefore, they are suitable for investigating biochemical processes that take place in cellular hypoxia, a commonly used in vitro model of stroke (Camos and Mallolas 2010). In this study, we investigated the effect of hypoxia on a human cerebral endothelial cell (HCEC) line, which was established in 1997 from the temporal lobe of a patient undergoing surgery for idiopathic epilepsy by transfecting endothelial cell colonies with a pSV3-neoplasmid encoding the SV40 large T antigen and selecting on the basis of the ability to internalize acetylated low-density lipoproteins and the expression of the transferrin receptor (Muruganandam et al. 1997; Dasgupta et al. 2011). These cells express other hallmarks of the BBB phenotype including γ-glutamyl transpeptidase, alkaline phosphatase, and tight junctions, while being negative for any trace of invasivity in soft agar (Muruganandam et al. 1997).

The aim of this study was to investigate the effect of hypoxia on sphingolipid metabolism in cerebral human endothelial cells with the goal to better understand the molecular mechanisms that occur in stroke and to identify new targets for vasculoprotection.

Materials and methods

Standards and reagents

Sphingolipids and fatty acid standards, including sphingosine (Sph), dihydro-Sph (DH-Sph), sphingosine-1-phosphate (S1P), dihydro-S1P (DH-S1P), DHC, and N-heptadecanoyl sphingosine (17:0-Cer), were obtained from Avanti Polar Lipids (Alabaster, AL, USA). C6-nitrobenzoxadiazole-ceramide (NBD-Cer) was from Matreya Inc. (Pleasant Gap, PA, USA). The non-phosphorylated lipid standards were dissolved in methanol, whereas the sphingoid base phosphates were dissolved in methanol containing a trace amount of concentrated HCl and were stored at −20°C. [3H] Palmitic acid (43 Ci/mmol) was purchased from New England Nuclear (Boston, MA, USA). Silica gel high-performance thin-layer chromatography (HPTLC) plates were obtained from Whatman (Clifton, NJ, USA) and the protein assay kit was obtained from Bio-Rad Laboratories (Hercules, CA, USA). Chloroform, methanol, and acetic acid used for HPTLC were of ACS grade and obtained from Fisher Scientific (Pittsburgh, PA, USA). Hexamethylumbelliferyl (HMU)-phosphorylcholine was purchased from Moscerdam Substrates (Amsterdam, The Netherlands). The caspase 3 substrate DEVD-AFC was from Sigma (St. Louis, MO, USA) and the mitochondrial membrane potential dye JC1 from Life Technologies (Grand Island, NY, USA). Myriocin and GW4869 were from Sigma and d-threo-ethylendioxyphenyl-2-palmitoylamino-3-pyrrolidinopropanol (EtDOP4) from Dr. Shayman from University of Michigan.

Cell culture

The HCEC line was kindly provided by Dr. Yu from University of Georgia (Dasgupta et al. 2011). HCEC cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and 1% gentamicin. Hypoxia treatments were performed in serum-free DMEM. Hypoxia was achieved by using a Gas-Pak oxygen scavenging system as described previously (BD, Franklin Lakes, NJ, USA; Qi et al. 1995).

Sphingomyelinase and sphingosine-1-phosphate lyase assay

Acid sphingomyelinase (ASMase) activity was determined with the fluorogenic substrate (HMU)-phosphorylcholine as described previously (Testai et al. 2004a). Briefly, cells were harvested and washed with phosphate-buffered saline (PBS), the pellets were resuspended and lysed in 25 mM Tris–HCl, 150 mM NaCl, and 1% Triton X-100, pH 7.4. Approximately 50 μg of homogenates was mixed with the fluorogenic substrate (0.6 mM). The incubation was carried out in 150 mM sodium acetate buffer containing 1 mM EDTA, pH 4.5, to block any neutral sphingomyelinase (NSMase) activity. The HMU released was followed fluorometrically in a 96-well Synergy H1 Hybrid (BioTek Inc., Winooski, VT, USA) microplate reader using 370 nm excitation and 460 nm emission. The enzyme activity was calculated from the slope of the graph of intrinsic fluorescence plotted against time and standardized by mg of protein. For NSMase, the incubation was at 37°C in PBS containing 0.5 mM MgCl2, pH 7.4. The presence of phosphate was previously shown to inhibit any ASMase activity (Testai et al. 2004b). S1P-lyase was determined using the fluorogenic substrate 2S-ammonio-3R-hydroxy-5-[(2-oxo-2H-chromen-7-yl)oxy]pentyl hydrogen phosphate (Cayman Chemical, Seattle, WA, USA) as previously described (Bedia et al. 2009). Approximately, 300 μg of cell homogenate was incubated with 125 μM of the fluorogenic substrate in K3PO4, pH 7.5, containing 25 μM Na2VO4 and 0.25 mM pyridoxal phosphate. The reaction was followed in a 96-well microplate fluorescence reader for 6 h using 330 nm excitation and 460 nm emission. The enzyme activity was calculated from the slope of the graph of intrinsic fluorescence plotted against time and standardized by mg of protein.

Sphingomyelin synthase and glucosylceramide synthase assay

Cells (106) were grown in normoxia or hypoxia as described and in the presence or absence of the glucosylceramide synthase inhibitor (GCSi) EtDOP4 (1 mM final concentration). Cell were mechanically harvested, homogenized and sonicated in 0.2 mL of 10 mM Tris–HCl buffer pH 7.4 containing 1 mM EDTA and no Mg2+. Extracts (50 μg protein) were incubated at 37°C with saturating concentrations of NBD-Cer (1 μg) and exogenous phosphatidylcholine (20 μg) as the phosphorylcholine donor [for sphingomyelin synthase (SMS) activity] or UDP-glucose (500 μM) as the glucosyl donor (for glucosyceramide synthase) as described previously for up to 24 h (Yamaoka et al. 2004; Kilkus et al. 2008). The fluorescent NBD metabolites were extracted with chloroform : methanol : water (2 : 1 : 0.6 v/v). NBD-lipid in the lower phase was separated by HPTLC using chloroform : methanol : acetic acid : water (350 : 125 : 44 : 22.5 v/v) and quantified with a Bio-Rad Chemi-doc XRS scanner, using Quantity One software.

Lipid extraction and sample preparation for lipid quantification by LC/MS/MS

Cellular lipids were extracted by a modified Bligh and Dyer procedure with the use of 0.1N HCl for phase separation as previously described (Qin et al. 2009). For analysis of lipid detergent-resistant membranes (DRM), 0.5 mL of the DRM fraction was mixed with 3 mL of methanol : chloroform (2 : 1 v/v). C17-S1P (40 pmol), C17-Sph (30 pmol), and 17:0-Cer (30 pmol) were used as internal standards and were added during the initial step of lipid extraction. The extracted lipids were dissolved in methanol : chloroform (4 : 1 v/v), and aliquots were taken to determine total phospholipid content (Vaskovsky et al. 1975). Samples were concentrated under a stream of nitrogen, redissolved in methanol, transferred to autosampler vials, and subjected to consecutive LC/MS/MS analyses of sphingolipids (Qin et al. 2010).

Analysis of sphingoid bases, sphingoid base 1-phosphates and ceramides

Analyses of sphingolipids were performed by combined LC/MS/MS using an automated Agilent 1200 series liquid chromatograph and autosampler (Agilent Technologies, Wilmington, DE, USA) coupled to an AB Sciex 5500 QTRAP hybrid triple quadrupole linear ion trap mass spectrometer (Applied Biosystems, Foster City, CA, USA) equipped with a TurboIonSpray ionization source. Sphingolipids were ionized via electrospray ionization (ESI) with detection via multiple reactions monitoring (MRM). Analysis of sphingoid bases and the molecular species of ceramides used ESI in positive ions with MRM analysis. Sphingoid base-1-phosphates were analyzed as bis-acetylated derivatives via ESI in negative ions mode (Berdyshev et al. 2005). Standard curves for each of the sphingoid bases, sphingoid base 1-phosphates, and ceramides molecular species were constructed by adding increasing concentrations of the individual analyte to 30 or 40 pmol of the corresponding structural analogs used as the internal standard. Linearity and the correlation coefficients of the standard curves were obtained by a linear regression analysis. The standard curves were linear over the range of 0.0–300 pmol of each of the sphingolipid analytes with correlation coefficients (R2) > 0.98. Sphingolipid levels were expressed as fmol/nmol of total lipid P in whole cells and per mg of protein in lipid DRM.

Lipid metabolic studies

Cells were labeled with 1 μCi/mL [3H] palmitate and lipids were extracted as described previously (Kilkus et al. 2003). Typical labeling experiments were carried out in 100-mm Petri dishes containing 8 mL of serum-free medium. Cells (3 × 106/100 mm plate) were treated as indicated and then harvested and washed three times with PBS, lipids were extracted by chloroform : methanol : water (2 : 1 : 0.6 v/v) partition and samples were subjected to alkaline methanolysis to remove phosphoglycerides. Lipids were applied to HPTLC plates (10 × 10 cm; LHP-K TLC plates; Whatman Inc.). Ceramides were separated using chloroform : methanol : water (94 : 1 : 5 v/v) and SM were separated using chloroform : methanol : acetic acid : water (90 : 25 : 8.8 : 4.5 v/v). Lipids were visualized in iodine vapors then scraped off for quantification by liquid scintillation counting. Glycosphingolipid production was analyzed by 2D-HPTLC. Lipids from cell extracts were separated in the first dimension using chloroform : methanol : 30% ammonium hydroxide (65 : 20 : 4 v/v) and second dimension using chloroform : acetone : methanol : acetic acid : water (50 : 20 : 10 : 10 : 5 v/v). HPTLC plates were sprayed with EN3HANCE (Perkin-Elmer, Waltham, MA, USA) to facilitate autoradiography and bands were excised for radioactivity determination by liquid scintillation counting.

Isolation of lipid detergent-resistant membranes

Cell pellets were lysed in 1.5 mL of 25 mM MES, 50 mM NaCl, 1% Triton X-100, 1 mM Na3VO4, pH 6.5, supplemented with a protease inhibitor cocktail (leupeptin, phenylmethylsulfonyfluoride, and aprotinin) for 1 h at 4°C. After being homogenized 10 times in a loose-fit Dounce homogenizer, lysates were mixed with 1.5 mL of 80% sucrose in MBS (25 mM MES, 150 mM NaCl, pH 6.5) and overlayered with 3 mL 30% sucrose in MBS and then with 3 mL of 5% sucrose in MBS. After centrifugation for 18 h at 170 000 g in an SW40 rotor, 1 mL fractions were collected. The DRM fraction is flotillin-2 positive and was typically found between fractions 3 and 4 (Kilkus et al. 2003). The 1 mL fractions were analyzed for lipids by HPTLC and LC/MS/MS and for protein.

Measurement of transendothelial electrical resistance

Transendothelial electrical resistance (TER) was measured in an electrical cell-substrate impedance sensing system, Applied Biophysics Inc. (Troy, NY, USA) as previously described (Nagababu et al. 2009). Briefly, HCEC were cultured on gold microelectrodes (eight electrodes per plate) until reaching approximately 95% confluence. At this stage, tight cell–cell junctions are formed. Cells were rinsed with serum-free DMEM and electrodes were placed in an electrical cell-substrate impedance incubator for 1 h to stabilize basal electrical resistance. Once cells attained stable TER, drugs were added into electrodes for 1 h followed by exposure to hypoxia. The TER was monitored for a designated time. The resistance is expressed as normalized resistance in time course experiments.

Caspase 3 assay

Cells were treated with drugs at the concentrations and times indicated, harvested, and washed with PBS. Pellets were resuspended and lysed in 25 mM Tris–HCl, 150 mM NaCl and 1% Triton X-100, pH 7.4. Hydrolysis of the caspase 3 (DEVD-AFC) substrate was determined using 5–10 μg of cell extract in 100 μL of 25 mM HEPES buffer containing 2 mM dithiothreitol and 5 mM EDTA, pH 7.4. The reaction was followed up for 3 h in a Synergy H1 Hybrid reader (BioTek Inc.) microplate fluorescence reader at 37°C set at 400 nm excitation and 505 nm emission. Enzyme activity was calculated from the slope of intrinsic fluorescence plotted against time and standardized by μg of protein (Testai et al. 2004a).

Cell viability assays

Cells were plated in 24-well culture plates for 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, which was carried out as described previously (Wiesner et al. 1997).

Statistical analysis

Results are based on experiments run in triplicate (n = 3) at least twice. Statistical analyses were performed by Student's t-test and results were considered statistically significant when p < 0.05.


Effect of hypoxia on sphingolipids in human cerebral microendothelial cells

Detergent-resistant membranes are sphingolipid-enriched microdomains that compartmentalize different cellular processes. The dynamics of sphingolipid metabolism was assessed by [3H] palmitate labeling and compared to their steady-state levels as determined by HPLC/MS/MS. HCEC cells, labeled with [3H] palmitate for 1 h and then incubated for 6 h under increasing hypoxic conditions showed a small but statistically significant increase in Cer production with no change in SM. Two new fast-moving bands not normally observed under standard culture conditions were also observed after hypoxia (Fig. 1). As de novo sphingolipid synthesis is dependent upon an oxygen-requiring DHC desaturase, we tentatively identified these new bands as DHC and DH-SM generated under hypoxic conditions.

Figure 1.

Sphingomyelin (SM) and ceramide (Cer) production in human cerebral endothelial cells (HCEC) in cells and detergent-resistant membranes (DRM) following hypoxia. Cells were incubated for 1 h with [3H]-palmitate and then grown in normoxia or hypoxia for 6 h. DRM were isolated and lipids extracted and resolved by high-performance thin-layer chromatography (HPTLC). Bands corresponding to Cer, SM, phosphatidylcholine (PC) and cholesterol (Chol) were scrapped and the radioactivity measured by scintillation counting. The labeling of SM was, on average, 3–4 times higher than Cer. SM and Cer were internally standardized by PC and Chol, respectively. Hypoxia resulted in the appearance of previously undetectable fast-moving bands (a) tentatively identified as dihydroceramides (DHC) and DH-SM. Hypoxia also increased the turnover of Cer, but not SM in both whole cells (b) and DRM (c). The radioactivity distribution in the sucrose-density gradient was not affected by hypoxia (d). Means ± SD. *p < 0.05.

We determined absolute levels of Cer and DHC by HPLC/MS/MS which confirmed a 5- to 10-fold elevation in DHC (99 ± 2 vs. 917 ± 203 fmol/nmol total lipid P) (Fig. 2) after hypoxia. The main contributors to total DHC were long-chain fatty acids C22:0, C24:0 and C24:1; however, both medium and long-chain fatty acids increased in response to hypoxia. The increase in DHC reversed after 1 h of reoxygenation and persisted after prolonged hypoxia (12 h) (data not shown). Similar changes in Cer and DHC were observed in lipid DRMs. Myriocin (serine-palmitoyltransferase inhibitor) blocked DHC increase in response to hypoxia, whereas GW4869 (NSMase inhibitor) and EtDOP4 (GCSi) had little effect on it (Fig. 3). The inhibitory effect of GW4869 and EtDOP4 on their respective enzymes was confirmed in vitro (data not shown).

Figure 2.

Ceramide (Cer) and dihydroceramide (DHC) changes in human cerebral endothelial cells (HCEC) and detergent-resistant membranes (DRM) following hypoxia. Cells were exposed to normal conditions or hypoxia for 6 h and DRM isolated. Cer and DHC were measured by HPLC/MS/MS in whole cells (a–c) or DRM (d–f). (a) and (d) show total Cer and DHC levels, (b) and (e) Cer subtypes, and c and f DCH subtypes. Cells were grown in normal conditions (open bars), hypoxia (black bars) or hypoxia followed up for 1 h reoxygenation (gray bar). Means ± SD. *p < 0.05.

Figure 3.

Species of dihydroceramide (DHC) generated in human cerebral endothelial cells (HCEC) treated with drugs that regulate the metabolism of sphingolipids and exposed to hypoxia. Cells were incubated in the presence of myriocin (serine-palmitoyltransferase inhibitor), GW4869 [neutral sphingomyelinase (NSMase) inhibitor], and d-threo-ethylendioxyphenyl-2-palmitoylamino-3-pyrrolidinopropanol (EtDOP4) (glucosylceramide synthase inhibitor). DHC levels were determined by HPLC/MS/MS. Results are expressed as the fold increase in DHC species in control versus cells exposed to 6 h of hypoxia. Myriocin, but not GW4869 or EtDOP4, blocked the effect of hypoxia on DHC. Mean ± SD. *p < 0.05.

Effect of hypoxia on S1P and other sphingoid bases

Cells exposed to 6 h of hypoxia showed increasing levels of sphingosine (from 203 ± 26 to 252 ± 20 fmol/nmol of total lipid P), S1P (from 19 ± 4 to 51 ± 2 fmol/nmol of total lipid P), DH-S1P (from 10.2 ± 4.7 to 22.9 ± 6.2 fmol/nmol of total lipid P) and DH-Sph (from 35 ± 1 to 55 ± 5 fmol/nmol of total lipid P) and a 3-fold increase in the S1P/Cer ratio (Fig. 4).

Figure 4.

Hypoxia increases sphingosine-1-phosphate (S1P), dihydrosphingosine and dihydro-S1P (DH-S1P). Cells were grown for 6 h under normal conditions (open bars) or hypoxia (black bars) and sphingolipids determined by HPLC/MS/MS. (a) Levels of S1P, dihydrosphingosine (DH-Sph), and DH-S1P in human cerebral endothelial cells (HCEC). (b) S1P/ceramide (Cer) ratio in HCEC. Means ± SD. *p < 0.05.

Activity of enzymes that regulate the metabolism of sphingolipids in hypoxia

The activities of key enzymes that participate in the metabolism of sphingolipids, including ASMase, NSMase, GCS, and SMS were determined in vitro. Cells exposed to hypoxia showed an increase in GCS activity by 40% which was reversed by reoxygenation (Fig. 5). Hypoxia was associated with a 2.5-fold increase in the activity of S1P-lyase (hypoxia: 5.7 ± 0.6 IF/h.μg; normoxia: 2.1 ± 0.4 IF/h.μg; p = 0.01), but did not affect ASMase, NSMase, and SMS (data not shown).

Figure 5.

Effect of hypoxia on the conversion of ceramide (Cer) into complex lipids in human cerebral endothelial cells (HCEC) cells. (a) Activity of glucosylceramide (GlcCer) synthase and sphingomyelin (SM) synthase. Hypoxia (6 h) activates GlcCer synthase in HCEC cells, but not SM synthase activity and this is reversed by reoxygenation. (b) Effect of a glucosylceramide synthase inhibitor (GCSi) on Cer and SM formation. Cells were grown under normal conditions or 6 h of hypoxia in the presence or absence of the GCSi [d-threo-ethylendioxyphenyl-2-palmitoylamino-3-pyrrolidinopropanol (EtDOP4), 1 mM]. Cer and SM turnover were determined as indicated in methods. The ratio of newly formed Cer/SM increased under hypoxic conditions, suggesting that newly formed Cer is not routed toward the synthesis of SM, but to other metabolic pathways. The additional increase in Cer/SM noted in the presence of the GCSi suggests that under hypoxia, the toxic Cer is utilized for the synthesis of glucosphingolipids. *p < 0.05 compared to control.

Effect of hypoxia on glucosylceramides

Cells were incubated for 1 h with [3H] palmitic acid and [3H]-glycosphingolipids measured by HPTLC as indicated in methods. Cells exposed to hypoxia showed a fast-moving band in close proximity to glucosylceramide (GlcCer), which we tentatively identified as dihydro-GlcCer (DH-GlcCer). The activity of GCS was determined by measuring the conversion of NBD-(C6)Cer to NBD-GluCer. In normoxic conditions, increasing amounts of (C6)-DHC inhibited the production NBD-GluCer, indicating that GCS can utilize either Cer of DHC as substrates (Fig. 6). The activity of GCS increased during hypoxia, whereas SMS activity was unaffected. The ratio of Cer/SM increased in response to hypoxia and this was further accentuated by the GCSi EtDOP4 (Fig. 6). Collectively, our results indicate that part of the Cer generated during oxygen deprivation is not routed to the production of SM, but is channeled by GCS into GlcCer, perhaps as a protective mechanism to lower ceramide, as has been observed in tumor cells.

Figure 6.

(a) Effect of hypoxia on the synthesis of glucosylceramide (GlcCer). Human cerebral endothelial cells (HCEC) were labeled with [3H] palmitate and grown in normal conditions or hypoxia (6 h). Lipids were analyzed by 2-D high-performance thin-layer chromatography (HPTLC) as indicated in methods. After hypoxia, a doublet was observed in the mobility of GlcCer, which was tentatively identified as dihydro-GlcCer. (b) Effect of dihydroceramide (DHC) on glucosylceramide synthase (GCS) activity. HCEC extracts were incubated with increasing amounts of C6-DHC and the activity of GCS determined in vitro by using C6-NBD-Cer. Lipids were resolved by HPTLC and the intensities of the bands corresponding to NBD-Cer and NBD-GlcCer were quantified by densitometry. Results are expressed as arbitrary units of density (*intensity/mm2). Extracts incubated with NBD-Cer and without UDP-Glu were used as negative control.

Effect of sphingolipids on endothelial cell death

We investigated the activity of caspase 3 (DEVDase) in response to hypoxia in the presence and absence of drugs that regulate the metabolism of sphingolipids. The addition of myriocin (inhibitor of de novo ceramides synthesis) or the GCSi EtDOP4 had an additive effect to hypoxia-induced caspase 3 activation. In agreement with these results, both myriocin and EtDOP4 increased cell death in response to hypoxia as measured by the MTT method (Fig. 7).

Figure 7.

Effect on drugs that regulate the metabolism of sphingolipids on hypoxia-mediated cell death. Cells were grown in normal conditions (open bars) or hypoxia (black bars) as indicated in the text in the presence or absence of drugs that regulate the metabolism of sphingolipids. Both the inhibitor of the de novo ceramide synthesis, myriocin (a), and the inhibitor of glucosylceramide transferase d-threo-ethylendioxyphenyl-2-palmitoylamino-3-pyrrolidinopropanol (EtDOP4) (b) increased caspase 3 (DEVDase) activity in response to hypoxia. In addition, both myriocin and EtDOP4 increased hypoxia-mediated cell death as measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) method (c). Means ± SD. *< 0.05 compared to control hypoxia.

Effect of hypoxia on transendothelial electrical resistance

The physiological effect of the sphingolipid metabolic changes in HCEC was investigated by measuring TER, which constitutes a highly sensitive measure of endothelial cell barrier function. A decrease in TER suggests weakened cell–cell junctions and increased permeability. Compared with normal conditions, hypoxia reduced the TER by approximately 30%. Myriocin and EtDOP4 had an additive effect to hypoxia and reduced the normalized resistance by approximately 60% (Fig. 8). These results highlight the importance of both the de novo synthesis and glycosphingolipids in preserving cerebral microendothelial function.

Figure 8.

Transendothelial electrical resistance (TER). Human cerebral endothelial cells (HCEC) were grown in normal conditions or hypoxia. Compared with normal conditions, the TER of HCEC exposed to hypoxia was reduced by 30%. The inhibitor of the de novo ceramide synthesis myriocin and the inhibitor of glucosylceramide transferase d-threo-ethylendioxyphenyl-2-palmitoylamino-3-pyrrolidinopropanol (EtDOP4) independently potentiated the effect of hypoxia on endothelial cell permeability. n = 3 independent experiments per condition. Means ± SD. *p < 0.05 compared to control hypoxia.


Substantial changes in the metabolism of sphingolipids have been observed following a stroke. In animal models in vivo, ceramides and their dihydro-precursors have been shown to increase in ischemia–reperfusion (Novgorodov and Gudz 2009, 2011). Injury and experimental hypoxia inhibit the activity of DHC desaturase leading to accumulation of DHC in both tumoral cells and lung models (Devlin et al. 2011). However, the effect of hypoxia on the cerebral endothelium and BBB and on other bioactive sphingolipids has not previously been investigated. We now report that in addition to DHC, dihydrosphingolipids such as DH-Sph, DH-S1P, DH-SM and DH-GlcCer increase in HCEC cells (an in vitro BBB model) following exposure to hypoxia. The current study supports previous data obtained from DHC desaturase knockout mice and further shows that the increase in DHC and DH-SM occurs in both isolated DRM and whole HCEC cells. In addition, we tentatively identify DH-GlcCer as an abnormal glycolipid following hypoxia (Kok et al. 1997).

It has been proposed that the DHC-desaturase operates as a cellular oxygen sensor that regulates the balance between ceramides and DHC (Devlin et al. 2011). Our findings are in agreement with this hypothesis and we extend these observations to human cerebral endothelium and other more complex sphingolipids. DH-Sph is both a precursor of DHC and a catabolite, whereas DH-S1P is a catabolite. We have observed that the addition of exogenous (C6)-DHC to either neurons or oligodendrocytes leads to increased DH-Sph and DH-S1P, but not Sph or S1P, whereas exogenous (C6)-Cer increases Sph and S1P, suggesting that the pathways are separate (data not shown). S1P synthesis requires the concerted action of ceramidase and sphingosine kinases and, once formed, S1P is either metabolized to hexadecenal and ethanolamine phosphate by the S1P-lyase or recycled to Sph and ceramides. In this study, the activity of S1P-lyase, as measured with a commercially available fluorogenic substrate, was decreased by hypoxia, suggesting that the elevated levels of S1P (and possibly dihydro-S1P) are driven, at least partially, by a hypocatabolic state. The activity change of S1P-lyase in our study, however, should be interpreted with caution as it was low. Nonetheless, these data support the results obtained in adipocytes which respond to hypoxia by down-regulating S1P-lyase expression (Ito et al. 2013).

DRMs are membrane microdomains that contain enzymes that participate in endothelial vessel tone regulation, such as endothelial nitric oxide synthase, and cell signaling (Sprenger et al. 2006; Allen et al. 2007; Parton and Simons 2007). These structures are enriched in cholesterol and sphingolipids, and changes in the lipid composition are known to interfere with protein trafficking to the cell surface and cell adhesion (Marquez and Sterin-Speziale 2008; Maalouf et al. 2010). In addition, significant remodeling of DRM has been observed in response to hypoxia (Botto et al. 2008). Thus, we investigated the effect of oxygen deprivation in whole cells and DRM. The most dramatic changes we observed in hypoxia were the increase in DHC, particularly C24:0 and C24:1, in both whole cells and lipid DRM isolated from hypoxic HCEC cells. It has been described that the activity of DHC desaturase decreases with increasing length of the acyl chain (Michel et al. 1997; Mikami et al. 1998). Thus, it is reasonable to hypothesize that this differential affinity, which may be further exacerbated in hypoxia, is responsible for the disproportionate accumulation of long-chain DHC. The 4,5-trans-double bond that differentiates sphingosines from sphinganines (dihydrosphingolipids) was until recently thought to be required for biological activity such as the initiation of cell death, but Zheng et al. (2006) and others have recently suggested that DHC may regulate autophagy and exhibit protective effects (Siddique et al. 2013). In addition, DHC has been shown to inhibit ceramide-induced channel formation in mitochondria, and the pharmacological or genetic ablation of DHC-desaturase was found to be protective against apoptosis (Stiban et al. 2006; Breen et al. 2013; Siddique et al. 2013). Here, we demonstrate that the inhibitor of de novo sphingolipid synthesis, myriocin, blocks the increase in DHC, affects TER, and decreases cell viability (as measured by both caspase 3 activation and the MTT method) in response to hypoxia, supporting the notion that dihydrosphingolipids have an active biological action.

An important conclusion that can be drawn from metabolic studies is that under hypoxic conditions, the newly formed Cer is not used in the synthesis of SM but is routed to the production of GlcCer. As GlcCer synthase is activated by hypoxia and the GCSi EdDOP4 increases ceramide levels, affects TER, and potentiates endothelial cell death, we infer a protective role for GlcCer synthase in stroke as a way to decrease cellular ceramide levels, which could otherwise initiate apoptosis. This observation fits well with previous studies showing that the activity of GlcCer synthase increases during cerebral ischemia and that glycosphingolipids are protective against stroke (Yamagishi et al. 2003; Liu et al. 2005; Hisaki et al. 2008). Certainly, GlcCer synthase must confer some resistance against hypoxia by removing pro-apoptotic ceramides, a common phenomenon observed in drug-resistant tumors but, as with other cellular responses to stress (such as increased protein synthesis during ER stress), this action is most likely brief (Liu et al. 2011).

In summary, the results of this study expand our understanding of changes in the metabolism of sphingolipids that occur in cerebral endothelial cells in response to hypoxia, and identify both de novo sphingolipid synthesis and more complex glycolipid synthesis as likely endogenous protective mechanisms against oxygen deprivation. More importantly, we provide evidence supporting the idea that drugs that decrease ceramides and downstream metabolites should have vasculoprotective properties and therefore may be clinically useful in stroke management.

Acknowledgements and conflict of interest disclosure

Funded by minority recruitment supplement at UIC to FDT, USPHS grants NS36866 and HD09402 to GD, and PO1-HL 98050 to VN and EB.

All experiments were conducted in compliance with the ARRIVE guidelines. The authors have no conflict of interest to declare.