A Novel Role of Lactosylceramide in the Regulation of Tumor Necrosis Factor α-mediated Proliferation of Rat Primary Astrocytes

The present study describes the role of glycosphingolipids in neuroinflammatory disease and investigates tumor necrosis factor α (TNFα)-induced astrogliosis following spinal cord injury. Astrogliosis is the hallmark of neuroinflammation and is characterized by proliferation of astrocytes and increased glial fibrillary acidic protein (GFAP) gene expression. In primary astrocytes, TNFα stimulation increased the intracellular levels of lactosylceramide (LacCer) and induced GFAP expression and astrocyte proliferation. d-Threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol·HCl (PDMP), a glucosylceramide synthase and LacCer synthase (GalT-2) inhibitor, inhibited astrocyte proliferation and GFAP expression, which were reversed by exogenous supplementation of LacCer but not by other glycosphingolipids. TNFα caused a rapid increase in the activity of GalT-2 and synthesis of LacCer. Silencing of GalT-2 gene using antisense oligonucleotides also attenuated the proliferation of astrocytes and GFAP expression. The PDMP and antisense-mediated inhibition of proliferation and GFAP expression was well correlated with decreased Ras/ERK1/2 pathway activation. Furthermore, TNFα-mediated astrocyte proliferation and GFAP expression was also inhibited by LY294002, a phosphatidylinositol 3-kinase inhibitor, which was reversed by exogenous LacCer. LY294002 also inhibited TNFα-induced GalT-2 activation and LacCer synthesis, suggesting a phosphatidylinositol 3-kinase-mediated regulation of GalT-2. In vivo, PDMP treatment attenuated chronic ERK1/2 activation and spinal cord injury (SCI)-induced astrocyte proliferation with improved functional recovery post-SCI. Therefore, the in vivo studies support the conclusions drawn from cell culture studies and provide evidence for the role of LacCer in TNFα-induced astrogliosis in a rat model of SCI. To our knowledge, this is the first report demonstrating the role of LacCer in the regulation of TNFα-induced proliferation and reactivity of primary astrocytes.

The present study describes the role of glycosphingolipids in neuroinflammatory disease and investigates tumor necrosis factor ␣ (TNF␣)-induced astrogliosis following spinal cord injury. Astrogliosis is the hallmark of neuroinflammation and is characterized by proliferation of astrocytes and increased glial fibrillary acidic protein (GFAP) gene expression. In primary astrocytes, TNF␣ stimulation increased the intracellular levels of lactosylceramide (LacCer) and induced GFAP expression and astrocyte proliferation. D-Threo-1-phenyl-2decanoylamino-3-morpholino-1-propanol⅐HCl (PDMP), a glucosylceramide synthase and LacCer synthase (GalT-2) inhibitor, inhibited astrocyte proliferation and GFAP expression, which were reversed by exogenous supplementation of LacCer but not by other glycosphingolipids. TNF␣ caused a rapid increase in the activity of GalT-2 and synthesis of LacCer. Silencing of GalT-2 gene using antisense oligonucleotides also attenuated the proliferation of astrocytes and GFAP expression. The PDMP and antisense-mediated inhibition of proliferation and GFAP expression was well correlated with decreased Ras/ERK1/2 pathway activation. Furthermore, TNF␣-mediated astrocyte proliferation and GFAP expression was also inhibited by LY294002, a phosphatidylinositol 3-kinase inhibitor, which was reversed by exogenous LacCer. LY294002 also inhibited TNF␣-induced GalT-2 activation and LacCer synthesis, suggesting a phosphatidylinositol 3-kinase-mediated regulation of GalT-2. In vivo, PDMP treatment attenuated chronic ERK1/2 activation and spinal cord injury (SCI)-induced astrocyte proliferation with improved functional recovery post-SCI. Therefore, the in vivo studies support the conclusions drawn from cell culture studies and provide evidence for the role of LacCer in TNF␣-induced astrogliosis in a rat model of SCI. To our knowledge, this is the first report demonstrating the role of LacCer in the regulation of TNF␣-induced proliferation and reactivity of primary astrocytes.
Traumatic injury to the adult central nervous system (CNS) 1 results in a rapid inflammatory response by the resident astro-cytes, characterized mainly by hypertrophy, proliferation, and increased glial fibrillary acidic protein (GFAP) expression, resulting in astrogliosis (1)(2)(3)(4). Tumor necrosis factor-␣ (TNF␣) has been identified as one of the first cytokines to appear following CNS injury and has been implicated in exacerbation of CNS injury. TNF␣ induces proliferation of both primary astrocytes (5,6) and human astroglioma cell lines (7,8) as well as GFAP overexpression (9). Although many reasons have been put forward to explain the obvious lack of CNS regeneration following injury/neurotrauma, the robust formation of the glial scar, as a result of astrogliosis, is also known to interfere with subsequent neural repair or axonal regeneration (2,10). Thus, considerable effort is being directed toward understanding the mechanisms involved in astrocyte proliferation and reactivity in order to design therapeutic approaches to modulate gliosis, which is an impediment to neuronal recovery and axonal regeneration.
Studies from our laboratory and others have shown the involvement of sphingolipids such as ceramide and psychosine in the potentiation of cytokine-mediated inflammatory disease (11)(12)(13)(14). In addition, we have recently reported the involvement of lactosylceramide in the regulation of inducible nitricoxide synthase gene expression and the efficacy of the glycosphingolipid biosynthesis inhibitor (PDMP) in attenuating spinal cord injury (SCI)-induced inflammatory disease, demonstrating significantly improved functional outcome post-SCI (15). The activation of sphingomyelinases and the resulting sphingomyelin-ceramide pathway has been closely linked with TNF␣-induced apoptosis in numerous cell types. However, TNF␣ is also known to cause sphingosine 1-phosphate and lactosylceramide generation through activation of sphingosine kinase and lactosylceramide synthase (GalT-2), respectively, which have been implicated in inducing cell proliferation, which is antagonistic to the effects observed with ceramide (16,17). Whereas sphingosine 1-phosphate is mitogenic for primary astrocytes (18), LacCer has been linked with hyperproliferation of aortic smooth muscle cells in atherosclerosis (17).
In the present study, we sought to delineate the role of glycosphingolipids in TNF␣-induced astrocyte proliferation and GFAP expression. LacCer generated through TNF␣ stimulation was found to be the effector molecule that regulated TNF␣-induced proliferation of astrocytes and GFAP expression through the Ras/MEK/ERK pathway. LacCer generation in response to TNF␣ stimulation was found to be regulated in a PI3K-dependent manner, since 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002 or LY; a PI3K inhibitor) attenuated TNF␣-induced GalT-2 activation and LacCer production. In vivo, PDMP-treatment was efficacious in attenuating pathological ERK1/2 activation, astrocyte proliferation, and GFAP overexpression in a rat model of SCI. These studies underline the efficacy of modulating the GSL pathway in suppressing gliosis in SCI, which finds relevance in other CNS disorders as well. Cell Culture-Primary astrocyte-enriched cultures were prepared from the whole cortex of 1-day-old Sprague-Dawley rats as described earlier (12). Briefly, the cortex was rapidly dissected in ice-cold calcium/ magnesium-free Hanks' balanced salt solution (Invitrogen) at pH 7.4 as described previously (19). The tissue was then minced, incubated in Hanks' balanced salt solution containing trypsin (2 mg/ml) for 20 min, and washed twice in plating medium containing 10% fetal bovine serum and 10 g/ml gentamicin and then disrupted by triturating through a Pasteur pipette, following which cells were seeded in 75-cm 2 culture flasks (Falcon, Franklin, NJ). After incubation at 37°C in 5% CO 2 for 1 day, the medium was completely changed to the culture medium (Dulbecco's modified Eagle's medium containing 5% fetal bovine serum and 10 g/ml gentamicin). The cultures received half exchanges with fresh medium twice a week. After 14 -15 days, the cells were shaken for at least 24 h on an orbital shaker to remove the microglia and then seeded on multiwell tissue culture dishes. All the cultured cells were maintained at 37°C in 5% CO 2 . At 80% confluence, the cells were incubated with serum-free Dulbecco's modified Eagle's medium for 24 h prior to incubation with TNF␣ and other chemicals.

Reagents
BrdUrd Incorporation Assay-Proliferation of primary astrocytes was assayed by using the cell proliferation ELISA, BrdUrd colorimetric assay kit (Roche Applied Science) according to the manufacturer's protocol. Briefly, cells were seeded in 96-well plates in quadruplicate, and following overnight serum starvation they were stimulated with mitogenic stimulants. 2 h before termination of the proliferation assay, BrdUrd (10 M) was added to each well, following which cells were fixed, and levels of incorporated BrdUrd were assayed using a conjugated anti-BrdUrd enzyme. Colorimetric analysis was done by measuring absorbance at 370 nm using a spectramax MAX 190 (Molecular Devices) multiwell plate reading spectrophotometer.
Western Blot Analysis-The cells were washed with cold Tris-buffered saline (20 mM Trizma base and 137 mM NaCl, pH 7.5) and lysed in 1ϫ SDS sample loading buffer (62.5 mM Trizma base, 2% (w/v) SDS, 10% glycerol), and following sonication and centrifugation at 15,000 ϫ g for 5 min, the supernatant was used for the immunoblot assay. The protein concentration of samples was determined with the detergentcompatible protein assay reagent (Bio-Rad) using bovine serum albumin as the standard. Sample was boiled for 3 min with 0.1 volumes of 10% ␤-mercaptoethanol and 0.5% bromphenol blue mix. 50 g of total cellular protein was resolved by electrophoresis on 8 or 12% polyacrylamide gels, electrotransferred to polyvinylidene difluoride filter, and blocked with Tween 20-containing Tris-buffered saline (TBST; 10 mM Trizma base, pH 7.4, 1% Tween 20, and 150 mM NaCl) with 5% skim milk. After incubation with antiserum raised against rat GalT-2 (Abjent Inc.), phosphospecific ERK1/2 (Cell Signaling Tech Inc.), or GFAP (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), the membranes were then incubated with horseradish peroxidase-conjugated anti-rabbit or mouse IgG for 1 h. The membranes were detected by autoradiography using ECL-plus (Amersham Biosciences) after washing with TBST buffer.
Quantification of Ras Activation-After stimulation, primary astrocytes in 6-well plates were washed with ice-cold phosphate-buffered saline (PBS) and lysed in membrane lysis buffer (0.5 ml of 25 mM HEPES, pH 7.5, 150 mM NaCl, 1% Igepal CA-630, 0.25% sodium deoxycholate, 10% glycerol, 10 mM MgCl 2 , 1 mM EDTA, 25 mM NaF, 1 mM of sodium orthovanadate, and EDTA-free Complete™ protease inhibitor mixture). After centrifugation (5,000 ϫ g) at 4°C for 5 min, supernatant was used for the Ras activation assay. 100 g of supernatant was used for binding with agarose-conjugated Ras-binding domain (RBD) of Raf-1, which was expressed in BL21 (Invitrogen), Escherichia coli strain, transformed by pGEX-2T-GST-RBD in the presence of 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside as described previously (20). The binding reaction was performed at 4°C for 30 min in membrane lysis buffer. Following washing with membrane lysis buffer three times, Ras-RBD complexes were denatured by the addition of 2ϫ SDS sample buffer. Ras protein was identified by Western blot analysis with Ras antibodies from Upstate Biotechnology, Inc. (Lake Placid, NY). Densitometric analysis of the autoradiograph was carried out to determine Ras activation and represented as arbitrary density units.
Measurement of LacCer Synthesis-Cultured cells were incubated in growth medium containing [ 14 C]galactose (5 Ci/ml) for 24 h as described previously. The medium was removed, and the cell monolayer was washed with sterile PBS. After stimulation with TNF␣ (1 ng/ml) for various periods of time, cells were harvested and washed with ice-cold PBS and lysed by sonication. 200 g of protein was used for extraction of lipids using chloroform/methanol/HCl (100:100:1). The organic phase was dried under nitrogen. Glycosphingolipids were resolved by high performance thin layer chromatography using chloroform/methanol/0.25% KCl (70:30:4, v/v/v) as the developing solvent. The gel area corresponding to LacCer when compared with pure LacCer used as a standard (Matreya Inc.) was scraped, and radioactivity was measured employing Liquiscint (PerkinElmer Life Sciences) as the scintillating fluid.
GalT-2 Activity Assay-The activity of GalT-2 was measured using [ 3 H]UDP-galactose as the galactose donor and GlcCer as the acceptor as described previously (15). Briefly, following stimulation, cells were harvested in PBS, and cell pellets were suspended in Triton X-100 lysis buffer. Cell lysates were sonicated, and following protein quantification, 100 g of cell lysate was added to reaction mixture containing 20 M of cacodylate buffer (pH 6.8), 1 mM manganese/magnesium, 0.2 mg/ml Triton X-100 (1:2, v/v), 30 nmol of GluCer, and 0.1 mmol of UDP-[ 3 H]galactose in a total volume of 100 l. The reaction was terminated by adding 10 l of 0.25 M EDTA, 10 l of 0.5 M KCl, and 500 l of chloroform/methanol (2:1, v/v), and the products were separated by centrifugation. The lower phase was collected and dried under nitrogen. Following resolution on HPTLC plates, the gel was cut out, and radioactivity was measured in a scintillation counter. Assay without exogenous GluCer served as blank, and their radioactivity counts were subtracted from all respective data points.
Gal T-2 Antisense Oligonucleotide-A 20-mer antisense oligonucleotide of the following sequence (5Ј-CGC TTG AGC GCA GAC ATC TT-3Ј) targeted against rat GalT-2 was synthesized by Integrated DNA Technology. A scrambled oligonucleotide (5Ј-CTG ATA TCG TCG ATA TCG AT-3Ј) was also synthesized and used as control. Cells were counted and plated a day before transfection, and the following day they were treated with Oligofectamine (Invitrogen)-oligonucleotide complexes (200 nM oligonucleotide) under serum-free conditions. 48 h following transfection, the protein levels of GalT-2 were analyzed using polyclonal antibodies raised against rat GalT-2 (Abjent Inc.). 48 h following transfection, the cells were stimulated with TNF␣ (1 ng/ml), and Br-dUrd incorporation was assayed 18 h following stimulation.
Plasmids, Transient Transfection, and FACS Analysis-Dominant negative Ras expression vector (pCMVrasN17) was purchased from BD Biosciences. pEGFP expression plasmid was purchased from Clontech. p110*⌬kin, a kinase-deficient version of p110 (the catalytic subunit of PI3K) was obtained from Tanti et al. (21). 3 ϫ 10 5 cells/well were cultured in 6-well plates for 1 day before the transfection. Transfection was performed with plasmid concentration constant (2.5 g/transfection) and 8 l of Fugene transfection reagent (Roche Applied Science). 24 h following transfection, the cells were placed in serum-free media overnight. Following stimulation for 18 h, the cells were trypsinized and pelleted, and the cell pellets were washed with cold PBS and finally resuspended in 100 l of PBS. The cells were fixed in 70% ethanol at 4°C for 1 h. Following fixation, cells were pelleted and the cell pellets were washed with PBS three times. The DNA was stained with 7amino-actinomycin D (7-AAD). Cell cycle analysis was done. Events were acquired using a BD Biosciences FACSCalibur equipped with a 488-nm argon laser and CellQuest software. pEGFP was acquired using a 515-545-nm bandpass filter (FL1), and 7-AAD was acquired using a 670-nm longpass filter (FL3). DNA histograms were generated using Modfit LT software. The collected data were gated for doublet discrimination and pEGFP-positive events.
RNA Extraction and cDNA Synthesis-Following total RNA extraction using TRIzol (Invitrogen) per the manufacturer's protocol, singlestranded cDNA was synthesized from total RNA. 5 g of total RNA was treated with 2 units of DNase I (bovine pancreas; Sigma) for 15 min at room temperature in an 18-l volume containing 1ϫ PCR buffer and 2 mM MgCl 2 . It was then inactivated by incubation with 2 l of 25 mM EDTA at 65°C for 15 min. 2 l of random primers were added and annealed to the RNA according to the manufacturer's protocol. cDNA was synthesized in a 50-l reaction containing 5 g of total RNA and 50 -100 units of reverse transcriptase by incubating the tubes at 42°C for 60 min.
Real-time PCR-Total RNA isolation from rat spinal cord sections was performed using TRIzol (Invitrogen) according to the manufacturer's protocol. Real time PCR was conducted using Bio-Rad iCycler (iCycler iQ Multi-Color Real Time PCR Detection System; Bio-Rad). Single-stranded cDNA was synthesized from total RNA as described. The primer sets for use were designed (Oligoperfect™ designer, Invitrogen) and synthesized from Integrated DNA Technologies (Coralville, IA). The primer sequences for GFAP (forward, 5Ј-cca agc cag acc tca cag c-3Ј; reverse, 5Ј-ccg ata cca ctc ttc tgt ttc ttg-3Ј), glyceraldehyde-3phosphate dehydrogenase (forward, 5Ј-cct acc ccc aat gta tcc gtt gtg-3Ј; reverse-5Ј-gga gga atg gga gtt gct gtt gaa-3Ј). IQTM SYBR Green Supermix was purchased from Bio-Rad. Thermal cycling conditions were as follows: activation of DNA polymerase at 95°C for 10 min, followed by 40 cycles of amplification at 95°C for 30 s and 58.3°C for 30 s. The normalized expression of target gene with respect to glyceraldehyde-3-phosphate dehydrogenase was computed for all samples using Microsoft Excel data spreadsheet.
Induction of SCI in Rats-Sprague-Dawley female rats (225-250 g in weight) were purchased (Harlan Laboratories, Durham, NC) for induction of SCI. All rats were given water and food pellets ad libitum and maintained in accordance with the Guide for the Care and Use of Laboratory Animals of the United States Department of Health and Human Services (National Institutes of Health, Bethesda, MD). We have used a clinically relevant weight drop device for the induction of SCI in rats as described earlier (22). Briefly, rats were anesthetized by intraperitoneal administration of ketamine (80 mg/kg) plus xylazine (10 mg/kg) followed by laminectomy at T12. While the spine was immobilized with a stereotactic device, injury (30 g/cm force) was induced by dropping a weight of 5 g from a height of 6 cm onto an impounder gently placed on the spinal cord. Sham-operated animals underwent laminectomy only. However, no prophylactic antibiotics or analgesics were used in order to avoid their possible interactions with the experimental therapy of SCI.
Treatment of SCI-Within 30 min after induction of SCI, rats received the glycosphingolipid inhibitor, PDMP (Matreya). PDMP was dissolved in 5% Tween 80 in saline and diluted with sterile saline (0.85% NaCl) at the time of intraperitoneal administration to SCI rats. Animals (six per group) were randomly selected to form four different groups: vehicle (VHC) (5% Tween 80 in saline)-treated sham (laminectomy only) and SCI and PDMP (20 mg/kg in 5% Tween 80)-treated sham and SCI. A single dose of PDMP was administered every 24 h after the first dose until 72 h after injury. Animals were sacrificed under anesthesia 1 h, 4 h, 12 h, 24 h, 48 h, 72 h, and 1 week following treatment.
Preparation of Spinal Cord Sections-Rats were anesthetized and sacrificed by decapitation. Spinal cord sections with the site of injury as the epicenter (lesion epicenter) were carefully extracted from VHCtreated sham and SCI as well as PDMP-treated sham and SCI animals. Tissue targeted to be used for RNA and protein extraction was immediately homogenized in TRIzol (Invitrogen), snap frozen in liquid nitrogen, and stored at Ϫ80°C until further use. Total RNA was extracted per the manufacturer's protocol and used for cDNA synthesis as described earlier. Sections of spinal cord to be used for histological examination as well as immunohistochemistry were fixed in 10% neutral buffered formalin (Stephens Scientific, Riverdale, NJ). The tissues were embedded in paraffin and sectioned at 4-m thickness.
Immunohistochemical Analysis-Spinal cord sections were deparaffinized and sequentially rehydrated in graded alcohol. Slides were then boiled in antigen unmasking fluids (Vector Laboratories, Burlingame, CA) for 10 min, cooled in the same solution for another 20 min, and then washed three times for 5 min each in Tris-sodium buffer (0.1 M Tris-HCl, pH 7.4, 0.15 M NaCl) with 0.05% Tween 20 (TNT). Sections were treated with trypsin (0.1% for 10 min) and immersed for 10 min in 3% hydrogen peroxide to eliminate endogenous peroxidase activity. Sec-tions were blocked in Tris sodium buffer with 0.5% blocking reagent (TNB) (supplied with the TSA-Direct kit; PerkinElmer Life Sciences) for 30 min to reduce nonspecific staining. For immunofluorescent double labeling, sections were incubated overnight with anti-pERK1/2 antibody (1:100; mouse monoclonal; Cell Signaling) followed by antibodies against the astrocyte marker, GFAP (1:100; rabbit polyclonal; DAKO, Japan) for 1 h. Anti-GFAP was visualized using fluorescein isothiocyanate-conjugated anti-mouse IgG (1:100; Sigma), and pERK was visualized using TRITC-conjugated anti-rabbit IgG (1:100; Sigma). The sections were mounted in mounting media (EMS, Fort Washington, PA) and visualized by immunofluorescence microscopy (Olympus) using Adobe Photoshop software. Rabbit polyclonal IgG was used as control primary antibody. Sections were also incubated with conjugated fluorescein isothiocyanate anti-rabbit IgG (1:100; Sigma) or TRITC-conjugated IgG (1:100) without the primary antibody as negative control.
Statistical Analysis-All values shown in the figures are expressed as the means Ϯ S.E. of values obtained from at least three independent experiments. The results were examined by one-and two-way analysis of variance; then individual group means were compared with the Bonferroni test. p Ͻ 0.05 was considered significant.

TNF␣-induced Proliferation of Rat Primary Astrocytes Is
Mediated by GSL-TNF␣ stimulation of primary astrocytes, resulting in proliferation of astrocytes and their reactive transformation characterized by increased GFAP expression, is a complex multistep process. In the present study, we tested whether GSLs were somehow involved in astrocyte proliferation. Increasing concentrations of TNF␣ (0, 0.1, 1, and 5 ng/ml) induced proliferation of astrocytes, which was assayed by Br-dUrd incorporation (Fig. 1A). To address the involvement of GSL in TNF␣-mediated proliferation, primary astrocytes were pretreated for 0.5 h with several concentrations of the glycosphingolipid inhibitor PDMP (0, 10, 20, 30, and 50 M) or its corresponding inactive enatiomer, InPDMP (0, 10, 20, 30, and, 50 M), followed by stimulation with TNF␣ (1 ng/ml) for 18 h. PDMP dose-dependently inhibited cellular proliferation assayed by BrdUrd incorporation (Fig. 1B), whereas InPDMP has no effect (Fig. 1C). TNF␣ (at a concentration of 1 ng/ml) and PDMP (25 M) were used for subsequent studies. Furthermore, increasing doses of lactosylceramide (LacCer) induced proliferation of astrocytes; however, GluCer did not have a similar effect (Fig. 1D). Additionally, exogenously supplemented Lac-Cer but not GluCer was able to bypass PDMP-mediated inhibition of TNF␣-induced proliferation (Fig. 1E). A similar trend was observed with regard to GFAP gene expression. Pretreatment of astrocytes with PDMP inhibited TNF␣-induced GFAP mRNA and protein expression, which was reversed by exogenously supplemented LacCer (Fig. 1F). Furthermore, as shown in Fig. 2, exogenous supplementation of other GSL metabolites such as GalCer ( Fig. 2A) and gangliosides GM1 (Fig. 2B), GM3 (Fig. 2C), and GD3 (Fig. 2D) neither induced proliferation themselves nor reversed the PDMP-mediated inhibition of TNF␣-induced proliferation, thus indicating this to be a LacCer-specific effect. Therefore, a metabolite of the glycosphingolipid pathway, LacCer, may play a role in the regulation of TNF␣-mediated proliferation of astrocytes and GFAP expression, two processes that encompass astrogliosis.
TNF␣ Stimulation Results in Altered Levels of LacCer-To understand the mechanism of TNF␣-induced astrocyte proliferation mediated by LacCer, in situ levels of lactosylceramide were quantified. [ 14 C]LacCer was resolved and characterized by Rf value using commercially available standard LacCer by HPTLC as described under "Materials and Methods." As shown in Fig. 3A, a sharp increase in LacCer levels was observed within 2-5 min following stimulation with TNF␣. Upon TNF␣ stimulation, LacCer levels increased to ϳ2.5-fold of those observed in unstimulated cells. Correspondingly, a rapid increase in GalT-2 enzyme activity was also observed upon TNF␣ stimulation (Fig. 3B). The role of GalT-2 and its product LacCer in cell proliferation was further confirmed by silencing GalT-2 gene using antisense (AS) DNA oligomers against rat GalT-2 mRNA and a sequence-scrambled oligomer as a control. As shown in Fig. 3C, diminished protein levels of GalT-2 by AS oligonucleotides correlated with diminished synthesis of [ 14 C]LacCer upon TNF␣ stimulation. Silencing of GalT-2 with AS oligomers decreased the TNF␣-induced astrocyte proliferation (Fig. 3D), whereas supplementing LacCer exogenously bypassed the inhibition, presumably because the signaling events downstream of LacCer can be triggered upon the addi- tion of LacCer. Correlating with decreased astrocyte proliferation, diminished GFAP mRNA (Fig. 3E) and protein levels (Fig.  3F) were observed upon GalT-2 silencing using AS oligomers. However, in the presence of exogenous LacCer, AS-mediated inhibition of GFAP expression was blunted, thus further establishing the involvement of LacCer in astrogliosis.

Activation of Small GTPase Ras and ERK1/2 Is Involved in LacCer-mediated Regulation of TNF␣-induced Proliferation-
Since we have previously established the redox-dependent regulation of small GTPase Ras by LacCer (15), the possible involvement of Ras in LacCer-mediated regulation of TNF␣induced astrocyte proliferation was investigated. Primary astrocytes were transiently co-transfected with dominant negative Ras, DN-Ras (H-Ras N17 mutant), and pEGFP as a transfection marker followed by cell cycle analysis of the GFPgated cells by FACS. Upon TNF␣ and LacCer stimulation, the percentage of cells in S-phase was significantly increased in the mock-transfected group. However, the DN-Ras-transfected group showed a significantly decreased percentage of cells in S-phase (Fig. 4A). TNF␣-and LacCer-induced GFAP mRNA and protein expression was also significantly attenuated in DN-Ras-transfected cells and GFAP expression (Fig. 4B). These results show that Ras is involved and necessary for cellular proliferation as well as for GFAP expression. The inability of exogenous LacCer to bypass the inhibition by DN-Ras demonstrated that Ras is necessary for LacCer-mediated proliferation and GFAP gene expression and suggests that Ras is downstream of LacCer in the signaling cascade that induces astrogliosis. The role of Ras was further confirmed by assaying Ras activity using the glutathione S-transferase-conjugated Raf-1 RBD (Ras binding domain). As expected, TNF␣ stimulation enhanced the activation of Ras, which was attenuated upon PDMP pretreatment. PDMP-mediated inhibition of TNF␣-induced Ras activation was fully reversed by the addition of LacCer (Fig. 4C). Furthermore, TNF␣-induced Ras activation was also attenuated upon silencing of GalT-2 expression by AS oligonucleotides, thus further confirming LacCermediated regulation of Ras activation (Fig. 4D). To further examine the signaling events downstream of Ras that mediate proliferation and GFAP expression, we investigated the involvement of two well established downstream effectors of Ras, ERK1/2 (Fig. 4) and PI3K (Fig. 5). In correlation with the effect on proliferation and regulation of GFAP expression reported earlier (9), inhibition of the ERK1/2 pathway by PD98059 (a MEK1/2 inhibitor) pretreatment attenuated TNF␣-mediated astrocyte proliferation (Fig. 4E) and GFAP (Fig. 4F) expression, and this effect could not be reversed even by exogenous supplementation of LacCer, indicative of MEK-ERK1/2 being downstream of LacCer in the signaling cascade. Further, the effect of MEK1/2 inhibitor observed on cell proliferation was also confirmed by directly assaying ERK1/2 activation using antibodies specific for the phosphorylated (activated) form of ERK1/2. TNF␣-induced phosphorylation of ERK1/2 was inhibited both by PDMP and MEK1/2 inhibitor PD98059. However, exogenous LacCer supplementation could only reverse PDMPmediated inhibition of ERK1/2 activity and not PD98059-mediated inhibition (Fig. 4, G and H, respectively). In addition, AS-mediated silencing of GalT-2 expression attenuated TNF␣induced ERK1/2 activation (Fig. 4I). These results establish LacCer-mediated regulation of astrocyte proliferation and GFAP expression to be through the small GTPase Ras/ERK1/2 pathway.
The Role of PI3K in TNF␣-mediated Regulation of Astrocyte Proliferation-The involvement of the second effector of Ras, PI3K, in astrocyte proliferation was also examined. PI3K has been reported to be involved in cell survival pathways and pro- Primary astrocytes were treated with [ 14 C]galactose overnight. Following TNF␣ stimulation (1 ng/ml), cells were harvested at the time points indicated, and [ 14 C]LacCer was analyzed by HPTLC as described under "Materials and Methods." Also shown is a representative image of the TLC carried out using GSL extracted from cells stimulated with TNF␣ for 5 min (A). The enzyme activity of GalT-2 was assayed as described under "Materials and Methods" using cell lysates derived from cells stimulated with TNF␣ for various durations as shown (B). For the silencing of the GalT-2 gene, the cells were transfected with either GalT-2 antisense DNA oligomer (AS) or its sequence-scrambled DNA oligomer (Scr) as described under "Materials and Methods." 48 h after transfection, the protein levels of GalT-2 as well as [ 14 C]LacCer synthesis were analyzed as described earlier (15)  liferation in various cells types including primary astrocytes (18). Pretreatment with LY (30 M), a PI3K inhibitor, significantly attenuated TNF␣-induced proliferation of primary astrocytes (Fig. 5A). Transient transfection with p110*⌬kin, a kinase-deficient version of p110 (the catalytic subunit of PI3K) (21), significantly reduced the percentage of cells in S-phase upon TNF␣ stimulation (Fig. 5B). However, in the presence of LacCer, the LY-and p110*⌬kin-mediated inhibition of astrocyte proliferation was effectively blunted. The reversal of LY-and p110*⌬kininduced inhibition by exogenous LacCer shows a differential location of the ERK1/2 kinases and PI3K in the signaling cascade triggered by LacCer, resulting in astrocyte proliferation. Reversal of the effect of PI3K inhibition by LacCer suggested PI3K to be upstream of LacCer, whereas the nonreversal of MEK1/2inhibition by LacCer indicated ERK1/2 to be downstream of LacCer. To further understand the role of PI3K in the mechanism of TNF␣-mediated regulation of proliferation, we examined the possibility that PI3K might be involved in the regulation of LacCer generation in response to TNF␣ stimulation. Pretreat-ment with LY inhibited TNF␣-induced LacCer synthesis (Fig.  5C), which correlated with inhibition of TNF␣-induced GalT-2 activation as well (Fig. 5D). These results suggest two things: first that PI3K is involved in TNF␣-mediated astrocyte proliferation and, second, that PI3K is involved in regulation of GalT-2 activity and LacCer synthesis. Since not much is presently known about the post-translational modifications of GalT-2 that might regulate its activity, the involvement of PI3K offers some clues about the mechanism, which remains to be investigated in more detail and was beyond the scope of this study. Furthermore, pretreatment with LY inhibited TNF␣-mediated Ras (Fig. 5E) and ERK1/2 activation (Fig. 5F) that was bypassed by exogenously supplied LacCer. LY also inhibited TNF␣-mediated GFAP mRNA (Fig. 5G) and protein expression (Fig. 5H), and this inhibition could be effectively bypassed by exogenously supplied LacCer. These results taken together demonstrated the involvement of PI3K in regulation of GalT-2 activation and LacCer biosynthesis in response to TNF␣ stimulation. Through the regulation of LacCer synthesis, PI3K regulates the downstream FIG. 4

. The involvement of small GTPase Ras and ERK1/2 in LacCermediated regulation of TNF␣-induced proliferation and GFAP gene expression in primary astrocytes.
Dominant negative Ras (DN-Ras) was transiently co-transfected with pEGFP (transfection marker) in primary astrocytes followed by stimulation with TNF␣ and/or LacCer. The cell cycle status of the GFP-gated cell population was assayed by FACS analysis (A), and GFAP mRNA and protein expression (B) was assayed in DN-Ras-and mock-transfected primary astrocytes. The effect of inhibition of Lac-Cer synthesis via PDMP (C) or through AS-mediated silencing of GalT-2 (D) was examined on Ras activation using a glutathione S-transferase-tagged Raf-1 Ras binding domain as described under "Materials and Methods." Pretreatment with LacCer and/or PDMP (25 M) for 0.5 h or transfection with AS oligomers for 48 h was followed by TNF␣ stimulation for 5 min, cell lysates were used to assay levels of activated Ras, which is represented as a graph following densitometric analysis of the autoradiograph, and the data are expressed as arbitrary density units (A.D.U.). To examine MEK/ERK pathway involvement in proliferation and GFAP expression, upon pretreatment for 0.5 h with PD98059 (30 M), a MEK1/2 inhibitor, and/or LacCer (10 M), followed by stimulation with TNF␣ for 18 h, cell proliferation (E) and GFAP expression was assayed (F). ERK1/2 activation was assayed upon LacCer depletion using PDMP (G) or GalT-2 silencing (I) or in response to the MEK inhibitor PD98059 (H). PDMP or PD98059 was pretreated for 0.5 h followed by stimulation with TNF␣ for 20 min and immunoblot using phosphorylated ERK1/2 as described under "Materials and Methods." In the case of AS-mediated silencing of GalT-2, cells were transfected with the oligomers. 48 h following transfection, the cells were stimulated with TNF␣ for 20 min, following which cells lysates were prepared and assayed for ERK1/2 activation. ***, p Ͻ 0.01 in A as compared with unstimulated pcDNA-transfected cells. #, p Ͻ 0.001 as compared with mock-transfected TNF␣stimulated cells in A.
signaling events such as activation of the Ras/ERK1/2 signaling cascade, which regulates proliferation and GFAP expression.
The Efficacy of PDMP in Attenuation of Astrogliosis in SCI-To test the physiological relevance of our observations and further investigate the role of LacCer in astrogliosis in vivo, we examined the effect of PDMP in the rat SCI model. Rapid and chronic activation of ERK1/2 has been proposed to be a mechanism that operates in astroglial activation following acute CNS injury (23,24). Furthermore, astrogliosis triggered in response to secondary inflammatory disease has been widely reported to be detrimental for axonal regeneration and recovery in SCI (25)(26)(27). As shown in Fig. 6A, a robust activation of ERK1/2 is observed within 1 h post-SCI. Activated ERK1/2 levels steadily rise until 48 h and remain substantially elevated even until 1 week post-SCI. However, PDMP treatment 0.5 h post-SCI effectively attenuates chronic ERK1/2 activation (Fig.  6A). Additionally, PDMP treatment effectively attenuated GFAP mRNA (Fig. 6B) and protein expression (Fig. 6C), which was highly up-regulated in VHC-treated SCI. Furthermore, double immunofluorescence analysis of spinal cord sections from the lesion epicenter of VHC-treated SCI rats showed a significant increase in GFAP (Fig. 7D) and activated ERK1/2 (Fig. 7E) levels and their co-localization (Fig. 7F) 24 h following injury, whereas PDMP-treated SCI rats showed significantly attenuated GFAP expression (Fig. 7J) and ERK1/2 activation (Fig. 7K), as well as their co-localization (Fig. 7L), thus demonstrating the efficacy of PDMP in vivo in controlling chronic ERK1/2 activation and GFAP expression. To determine whether administration of PDMP suppressed astrogliosis after SCI, we counted the number of GFAP and BrdUrd double FIG. 5. The involvement of PI3K in TNF␣-induced cell proliferation and GFAP gene expression in primary astrocytes. Pretreatment with LY294002, a specific PI3K inhibitor, for 0.5 h was followed by stimulation with TNF␣, and cell proliferation was assayed (A). A kinase-deficient PI3K catalytic subunit, p110⌬kin, was transiently co-transfected with pEGFP (transfection marker) in primary astrocytes followed by stimulation with TNF␣ and/or LacCer. Cell cycle status of the GFP-gated cell population was assayed by FACS analysis (B). Following pretreatment with LY294002 and TNF␣ simulation, [ 14 C]LacCer production (C) and GalT-2 enzyme activity (D) were assayed at different time points as described under "Materials and Methods." Ras activation was examined using glutathione S-transferase-tagged Raf-1 Ras binding domain as described under "Materials and Methods." Upon pretreatment with LacCer and/or LY294002 (30 M) for 0.5 h followed by TNF␣ stimulation, cell lysates were used to assay levels of activated Ras, which are represented as a graph following densitometric analysis of the autoradiograph (arbitrary density units; A.D.U.) E, to examine MEK/ERK pathway involvement, upon pretreatment for 0.5 h with LY294002 (30 M) and or LacCer, ERK1/2 activation was assayed by immunoblot analysis using phosphorylated ERK1/2 antibodies as described under "Materials and Methods" (F). To examine PI3K involvement in GFAP gene expression, upon pretreatment for 0.5 h with LY294002, followed by stimulation with TNF␣, GFAP mRNA (G), and protein levels were assayed (H). **, p Ͻ 0.01 as compared with TNF␣-stimulated; ***, p Ͻ 0.001 as compared with LY-treated in A. ***, p Ͻ 0.001 as compared with unstimulated mock-transfected control; #, p Ͻ 0.001 compared with mock-transfected TNF␣-stimulated cells; @, p Ͻ 0.001 as compared with p110⌬kin-transfected TNF␣-stimulated cells in B.
positive cells in tissue sections. Double immunostaining with GFAP and BrdUrd revealed numerous proliferating astrocytes along the margins of the lesion in the VHC-treated SCI (Fig.  8B) as compared with VHC-treated sham (Fig. 8A), PDMPtreated sham (Fig. 8C), or PDMP-treated SCI (Fig. 8D). Correspondingly, PDMP-treated rats had a significantly better neurological outcome with a BBB score of 14 as compared with the VHC-treated rats with a score of 7, which could be due to attenuation of inflammatory events such as iNOS expression and proinflammatory cytokine expression as reported earlier (15) as well as astrogliosis as shown in this report (Fig. 8F). Thus, these studies indicate the involvement of glycosphingolipids in astrogliosis at the site of lesion in an in vivo model of SCI. These observations find critical relevance in other neuroinflammatory diseases as well, since astrogliosis and its detrimental effects are common to a number of CNS disorders. DISCUSSION Astrogliosis is a prominent and ubiquitous reaction characterized by proliferation of astrocytes with up-regulated expression of GFAP. Although the functional role of astrogliosis is not clearly defined, numerous studies have documented its pathological interference with the function of residing neuronal circuits, thus preventing axonal remyelination and inhibiting axonal regeneration (28,29). We have previously reported the involvement of LacCer in inducible nitric-oxide synthase gene expression in primary astrocytes and the anti-inflammatory efficacy of PDMP treatment in protecting against white matter vacuolization, demyelination, and neuronal apoptosis resulting in a profoundly improved neurological outcome in a rat model of SCI (15). Since PDMP treatment profoundly attenuated the inflammatory disease process post-SCI, including GFAP expression, which is a characteristic feature of astrogliosis, in this study we sought to investigate the involvement of GSL in proliferation of astrocytes and GFAP expression, the two processes that culminate in astrogliosis. This study demonstrates a novel pathway of LacCer-mediated regulation of TNF␣-induced astrocyte proliferation and GFAP expression through signaling events involving PI3K and the Ras/ERK1/2 pathway in primary astrocytes. These conclusions are based on the following findings (1). TNF␣-stimulation induced the activity of GalT-2 and increased the production of LacCer (2). The inhibition of GSL synthesis by PDMP or antisense oligonucleotides to GalT-2 inhibited astrocyte proliferation and GFAP expression, which was reversed by LacCer but not by other GSLs (GluCer, GalCer, GM1, GM3, and GD3) (3). Inhibition of LacCer synthesis also inhibited the activation of the Ras/ERK1/2 pathway (4). PI3K through an as yet unknown mechanism regulated GalT-2 enzyme activity and LacCer production (6). PDMP treatment effectively attenuated chronic ERK1/2 activation and GFAP expression in a rat model of SCI. Fig. 9 shows a schematic representation of the possible mechanism of regulation of TNF␣-induced astrocyte proliferation and GFAP expression by LacCer. Based on data presented in this report, TNF␣ through the activation of PI3K results in the activation of GalT-2, leading to LacCer biosynthesis. Lac-Cer generation recruits and activates the small GTPase (Ras) that activates the downstream ERK1/2 pathway, thus resulting in astrocyte proliferation and GFAP expression and triggering astrogliosis. Studies from our laboratory and others have reported the mechanism for the LacCer-mediated regulation of Ras to be reactive oxygen species-dependent in primary astrocytes (15) and other cell types (17). TNF␣-induced activation of the small GTPase Ras could be through the direct activation of Src kinases associated with the LacCer-enriched glycosphingolipid signaling domains (GSDs) present on the cell surface. A number of studies have shown that several transducer molecules such as Src kinase associate with these GSDs and form functional units known as lipid rafts, which mediate signal transduction and cellular functions (30). In particular, LacCer has been shown to play an important role in the stabilization of GSDs (31). The addition of exogenous LacCer to the plasma membrane also initiates a similar downstream signaling cascade following incorporation into the plasma membrane. The amphipathic properties of GSLs that make them capable of being incorporated into cellular membrane have been extensively exploited to identify the functions of specific GSLs (32,33). LacCer addition to the outer leaflet of the plasma membrane bilayer could induce redistribution and reorganization of the lipids in both leaflets of the plasma membrane, possibly resulting in the sorting of LacCer into GSDs. Coupling of the inner and outer leaflets of the lipid bilayer has been documented in artificial membranes (34), and this phenomenon is also thought to occur at the plasma membrane of living cells, based on the localization of various lipid-anchored proteins at the plasma membrane (35,36). If such a coupling occurs, changes in the inner leaflet protein localization could result in the activation of Src kinases that have been shown to be associated with GSDs (37). In particular, an Src kinase, Lyn, has been found to directly associate with LacCer in GSDs (33). ROS generation and Src kinase activation may be followed by Grb/SOS-mediated Ras activation, which triggers the downstream MEK1/2-ERK1/2 pathway. The data presented in this study identify a glycosphingolipid, LacCer, as a bioactive signaling molecule regulating astrogliosis by mediating astrocyte proliferation and GFAP expression. In addition, the blockade of trauma-mediated ERK activation and GFAP expression in the SCI model (as reported in this study) and the inflammatory process and neuronal apoptosis (15) by PDMP further establish LacCer, generated through GalT-2 stimulation, to be a potent signaling lipid molecule that triggers inflammation and astrogliosis and show that it may have relevance to various neuroinflammatory diseases.
Glial cells can secrete TNF␣, which, in turn, can act on these cells in an autocrine manner. TNF␣ can induce the proliferation of astrocytes (5, 6) and overexpression of GFAP (9), both components of astrogliosis. Astrogliosis is a prominent and ubiquitous reaction of astrocytes to many forms of CNS injury, often implicated in the poor regenerative capacity of the adult mammalian CNS (38). As in other CNS disorders, SCI initiates reactive gliosis as part of a response to restore homeostasis at the site of primary injury. However, with this comes the unfortunate burden of massive deposition of molecules that inhibit axonal growth and recovery (25). TNF␣, a potent pleiotropic proinflammatory cytokine, is generated during the inflammatory response in SCI. Within 15 min the mRNA levels of TNF␣ are increased in most cellular components of the CNS (39 -42). Although the levels of other proinflammatory cytokines are barely detectable after 24 h, the protein levels of TNF␣ continue to increase during the first week following SCI (43). This increase in TNF␣ protein levels is probably attributable to leukocyte infiltration, leading to secretion of other proinflammatory cytokines as well at the site of primary injury (44,45). Although the functional role of astrogliosis is not clearly defined, numerous studies have documented its pathological interference with the function of residing neuronal circuits, thus preventing axonal remyelination and inhibiting axonal regeneration (10). A number of strategies have been tested for modulation of astrogliosis following neurotrauma such as ablation of astrocytes (46,47) and alteration of the extracellular matrix associated with the astroglial scar (48) but with mixed results (49).
Chronic ERK activation has been reported in human reactive astrocytes in subacute and chronic lesions including infarct, mechanical damage, chronic epilepsy, and progressive multifocal leukocephalopathy (23,50). Since neurons, oligodendrocytes, and most inflammatory cells showed little or no detectable activation, activation of the ERK pathway has been deemed obligatory for the triggering and persistence of reactive astrocytes (24). We have previously demonstrated the efficacy of PDMP in attenuation of ERK activation in primary astrocytes, which was effective in inhibition of post-SCI inflammatory disease (15). PDMP treatment post-SCI showed a profoundly improved neurological outcome post-SCI as compared with the untreated rats. Since TNF␣ has long been known to trigger the ceramide pathway, which can be further converted to bioactive intermediates such as sphingosine or complex GSL, predicting the specific actions of these intermediates and the enzymes regulating their levels is rather complex.
In conclusion, we have dissected the signaling pathway involved in TNF␣-induced astrocyte proliferation and GFAP expression and established the role of LacCer in mediating these processes. The results presented further establish LacCer as a significant bioactive lipid molecule capable of mediating inflammatory disease processes in SCI as opposed to the earlier perception of LacCer as simply being a precursor for complex gangliosides. At present, the ongoing challenge for research focused on spinal cord regeneration is to modulate the response of astrocytes to injury so as to gain from the potential neurotrophic effects while at the same time tempering their scarring effect. This report proposes GSL modulation as a potential target to attenuate astrogliosis and the inflammatory disease processes in neuroinflammatory diseases.