Prostacyclin Prevents Pericyte Loss and Demyelination Induced by Lysophosphatidylcholine in the Central Nervous System*

Background: Pericyte damage is closely associated with the progression of neurodegeneration and neuronal dysfunction in the central nervous system (CNS). Results: Prostacyclin attenuates pericyte damage following vascular barrier dysfunction in the adult CNS. Conclusion: Prostacyclin therapy diminishes demyelination and neuronal deficits in pathophysiological conditions. Significance: This study provides the first evidence that pericyte protection contributes to attenuate disease progression in adult CNS. Pericytes play pivotal roles in physiological and pathophysiological conditions in the central nervous system. As pericytes prevent vascular leakage, they can halt neuronal damage stemming from a compromised blood-brain barrier. Therefore, pericytes may be a good target for the treatment of neurodegenerative disorders, although evidence is lacking. In this study, we show that prostacyclin attenuates lysophosphatidylcholine (LPC)-mediated vascular dysfunction through pericyte protection in the adult mouse spinal cord. LPC decreased the number of pericytes in an in vitro blood-brain barrier model, and this decrease was prevented by iloprost treatment, a prostacyclin analog. Intrathecal administration of iloprost attenuated vascular barrier disruption after LPC injection in the mouse spinal cord. Furthermore, iloprost treatment diminished demyelination and motor function deficits in mice injected with LPC. These results support the notion that prostacyclin acts on pericytes to maintain vascular barrier integrity.

Pericytes play pivotal roles in physiological and pathophysiological conditions in the central nervous system. As pericytes prevent vascular leakage, they can halt neuronal damage stemming from a compromised blood-brain barrier. Therefore, pericytes may be a good target for the treatment of neurodegenerative disorders, although evidence is lacking. In this study, we show that prostacyclin attenuates lysophosphatidylcholine (LPC)-mediated vascular dysfunction through pericyte protection in the adult mouse spinal cord. LPC decreased the number of pericytes in an in vitro blood-brain barrier model, and this decrease was prevented by iloprost treatment, a prostacyclin analog. Intrathecal administration of iloprost attenuated vascular barrier disruption after LPC injection in the mouse spinal cord. Furthermore, iloprost treatment diminished demyelination and motor function deficits in mice injected with LPC. These results support the notion that prostacyclin acts on pericytes to maintain vascular barrier integrity.
Vascular barrier dysfunction is a major anatomical feature in central nervous system (CNS) diseases, including multiple sclerosis, stroke, epilepsy, familial amyotrophic lateral sclerosis (1), and brain tumors (2). Because of an increase in vascular permeability, the CNS receives an influx of neurotoxic hemoglobinderived products (3). Maintaining the integrity of the vascular barrier is therefore considered necessary for protection against expanding and exacerbating damage in CNS diseases (4).
Although interendothelial contact is the principal cellular element of the vascular barrier, the formation and regulation of vascular barrier integrity appear to require cell-cell interactions in the vasculature. Vascular mural cells, or pericytes, make specific focal contacts with the endothelium within the microvasculature and have been recently highlighted as potentially contributing to vascular barrier integrity (5). Studies using mice with genetically deficient pericytes demonstrate that they are the major cell type that forms and maintains the vascular barrier during development (6) and adulthood (7). Moreover, agerelated pericyte loss is closely associated with the progression of neurodegeneration and neuronal dysfunction (8). These observations raise an interesting hypothesis that pericytes protect the CNS from damage resulting from disruption of the vascular barrier. However, direct evidence supporting or disputing this is lacking.
Prostacyclin (prostaglandin I 2 ) is a metabolite of arachidonic acid or prostaglandin endoperoxides and is released from vascular endothelial cells (9). The I-type prostaglandin receptor (IP 3 receptor) is a G protein-coupled receptor that is primarily coupled to the activation of adenylate cyclase, which catalyzes the formation of 3Ј,5Ј cyclic adenosine monophosphate (cAMP) (9), an enhancer of vascular barrier integrity (10). Expression of the IP receptor is reported on vascular smooth muscle and platelets and is involved in both vasorelaxants on smooth muscle and inhibition of platelet aggregation. Some drugs that target prostacyclin signaling have been approved for treatment of peripheral vascular diseases, including pulmonary arterial hypertension and arterial thrombosis. Concerning the role of prostacyclin in vascular barrier integrity, there have been no reports that disrupting either prostacyclin or the IP receptor gene in mice leads to an abnormal vascular barrier. However, studies under pathological conditions reveal that after occlusion of the middle cerebral artery, a stable analog of prostacyclin inhibits brain edema that increases vascular permeability (11). One possible hypothesis for reducing edema formation with a prostacyclin analog is that prostacyclin signaling directly acts on endothelial cells and enhances endothelial barrier function (12). Moreover, it is known that vasodilatation is the result of relaxation of perivascular cells (e.g. vascular smooth muscle cells) wrapped around the endothelium. Therefore, perivascular cell activation by prostacyclin may also be involved in reducing edema formation. These observations prompted us to hypothesize that prostacyclin signaling in perivascular cells, such as pericytes, contributes to reducing capillary hydraulic permeability under pathological conditions in the adult CNS.
Lysophosphatidylcholine (LPC) is a bioactive proinflammatory lipid generated by hydrolysis of phosphatidylcholine via the action of phospholipase A 2 . The majority of LPC formed in tissue is normally metabolized or reacetylated; however, overstimulation of phospholipase A 2 results in phosphatidylcholine breakdown and subsequent accumulation of LPC in the damaged tissue (13). Thus, accumulation of LPC is a recognized hallmark of pathological conditions and exacerbates pathological features. In the pathologically affected CNS, LPC production has been revealed by imaging mass spectrometry of experimental ischemic brain injury (14). Experimental accumulation of LPC into the CNS is known to cause neurodegeneration, which is characterized by demyelination (15,16). Demyelination by LPC is understood to occur by phagocytosis (17) and direct changes of myelin-associated protein expression (18); however, other possible mechanisms have not been fully elucidated.
By studying the mouse spinal cord, we show that prostacyclin protects against LPC-mediated pericyte damage following vascular barrier dysfunction in the adult CNS. Intrathecal administration of a prostacyclin analog attenuates demyelination and motor deficits induced by LPC.

EXPERIMENTAL PROCEDURES
Animals-C57BL/6J mice and Wistar rats were obtained from Charles River Japan or Japan SLC. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Osaka University.
Primary Cultures of Pericytes and Vascular Endothelial Cells-Brain pericytes were prepared from cerebral cortices obtained from female rats 3 weeks postnatally as described previously with modifications (19). The cerebral cortices were digested in a mixture of 1 mg/ml collagenase type 2 (Worthington) and 6.7 g/ml DNase (Sigma) in Dulbecco's modified Eagle's medium (DMEM) for 90 min at 37°C. The cell pellet was separated by centrifugation in 20% (w/v) bovine serum albumin (BSA)/DMEM (1000 ϫ g, 10 min) and then incubated for another 45 min at 37°C with a mixture of 1 mg/ml collagenase-dispase (Roche Applied Science) and 6.7 g/ml DNase in DMEM at 37°C. Brain capillaries were centrifuged on a 33% continuous Percoll gradient (GE Healthcare) (1000 ϫ g, 10 min) and plated on culture dishes coated with 0.1 mg/ml collagen type IV and 0.1 mg/ml fibronectin. Cells were maintained in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS).
To purify vascular endothelial cells from a mixed cell population, cells were cultured in DMEM supplemented with 10% heat-inactivated FBS and 4 g/ml puromycin (Sigma). Attached cells were identified as vascular endothelial cells.
To purify pericytes from a mixed cell population, cells were detached with 0.25% trypsin solution and transferred to noncoated culture dishes. Attached cells were immunocytochemically identified as pericytes. The purity of the pericytes was calculated as the ratio of the number of platelet-derived growth factor receptor (PDGFR) ␤ and 4Ј,6-diamidino-2-phenylindole (DAPI, Sigma) double-positive cells to the total number of DAPI ϩ cells in culture (91.2 Ϯ 0.09%).
Primary Culture of Cortical Neurons-Cortical neurons were prepared from cerebral cortices of postnatal day 1 Wistar rats. The cerebral cortices were digested in a mixture of 0.25% trypsin (Gibco) and DNase in phosphate-buffered saline (PBS) for 15 min at 37°C followed by resuspension in DMEM containing 10% FBS. The cells were passed through 70-m nylon mesh and were collected by centrifugation (300 ϫ g, 5 min). The cells were resuspended with DMEM containing 10% FBS and were plated on poly-L-lysine (Sigma)-precoated Lab-Tek 4-well chamber slides (Nalge Nunc International) at a density of 1 ϫ 10 6 cells/well. Primary Cultures of Microglia and Oligodendrocytes-Microglia and oligodendrocytes were obtained from the brain of the postnatal day 1 rat or mouse. The tissues were digested in a mixture of 0.25% trypsin in PBS for 15 min at 37°C followed by resuspension in DMEM containing 10% FBS. The isolated cells were plated on poly-L-lysine-precoated culture dishes in DMEM containing 10% FBS. One week after culturing, the dishes were rotated 20 times, and microglia were detached from the cell layer following centrifugation (300 ϫ g, 5 min). Cell suspension was diluted with DMEM containing 10% FBS and plated on poly-L-lysine-precoated Lab-Tek 4-well chamber slides at a density of 5 ϫ 10 4 cells/well.
Primary Culture of Macrophages-Macrophages were prepared from bone marrow of adult mice (female, 8 weeks). Col-lected bone marrow cells were cultured in Roswell Park Memorial Institute 1640 medium containing 10% FBS and 50 ng/ml macrophage colony-stimulating factor (Sigma). After 1 day of cultivation, we used the cells for phagocytosis assay.
Measurement of Electrical Resistance-We used an in vitro BBB kit (PharmaCo-cell, catalog no. RBT-12), which is composed of rat brain vascular endothelial cells, pericytes, and astrocytes (20). To assess the vascular integrity, we measured the transendothelial electric resistance (TEER) across the upper and lower transwell chambers using ERS-2 (Millipore). Values are expressed as ϫ cm 2 and were calculated by the formula: average resistance of experimental wells Ϫ average resistance of blank wells ϫ 0.33 (area of the transwell membrane). Relative values were calculated relative to the TEER of the control at 0 h. To assess the effect of endothelial cells on the TEER value, we seeded vascular endothelial cells at a density of 1 ϫ 10 5 cells/ insert on the polyethylene terephthalate membrane Millicell insert (pore size 3 m). Cells were maintained in DMEM supplemented with 10% heat-inactivated FBS and 4 g/ml puromycin.
Cell Count-Cultures were fixed with 4% PFA in PBS for 30 min at room temperature. Nuclei were stained with a solution of DAPI at a final concentration of 1 g/ml in PBS for 30 min at room temperature. For observation, cells on the opposite surface of transwell filters were removed by wiping with cotton swabs. Images were captured with an upright microscope using a ϫ20 objective (Olympus) and analyzed using a DP-controller image system (Olympus). We determined the number of DAPI ϩ cells on the upper surface of transwell (endothelial cell), the lower surface of transwell (pericyte), or the bottom of the culture plates (astrocyte). We calculated the average cell number from six fields per sample.
Surgical Procedure-Anesthetized mice were subjected to laminectomy at thoracic level Th8 and injected with 2 l of 1% (w/v) LPC (Sigma) dissolved in PBS over a 1-min period (0.5 mm, depth at the midline) into the dorsal column of the thoracic spinal cord. Pharmacological treatment started immediately after LPC injection. For administration of pharmacological reagents, an osmotic pump (model no. 1007D; ALZET Corp) was filled with vehicle solution (saline) or iloprost (20 mg/kg/day, Cayman Chemical). The delivery tube connected to an osmotic pump was placed close to the lesion just after LPC injection. The pump was implanted subcutaneously in the dorsal skin.
Immunohistochemical Analysis-One day after operation, mice were transcardially perfused with 4% PFA in PBS, and the spinal cords were postfixed for 24 h at 4°C, followed by immersion in 30% sucrose/PBS for 24 h at 4°C. Because the pericyte coverage varies depending on the level of the spinal cord (21), we carefully dissected the spinal cord from one segment rostral to one segment caudal to the lesion center (Th7-9). Spinal cord tissue was embedded in Tissue-Tek OCT compound (Sakura Finetek) and stored at Ϫ80°C. We prepared cryosections of 30-m in thickness and mounted them on adhesive silanecoated slides (Matsunami Glass). Sections were dried in air at room temperature for 1 h. After washing in PBS, sections were incubated with PBS containing 0.1% Triton X-100 and 10% swine serum (Vector Laboratories) for 1 h at room temperature. The sections were incubated with primary antibodies overnight at 4°C and then incubated with fluorescent secondary antibodies for 2 h at room temperature. Th following antibodies were used: goat anti-mouse CD13 antibody (1:100; R&D Systems, catalog no. AF2335); rabbit anti-mouse CD31 (vascular endothelial cell marker, 1:100, Abcam, catalog no. ab28364); Alexa Fluor 488 and Alexa Fluor 568 conjugated to isotype-specific antibodies (1:500, Invitrogen); and streptavidin-conjugated Alexa Fluor 488 (1:500, Invitrogen).
In a separate set of studies, cryosections of 30 m in thickness were blocked and permeabilized with 10% swine serum containing 0.1% Triton X-100. Sections were incubated with antibodies against rabbit anti-mouse cleaved caspase-3 and biotinylated anti-mouse PDGDR␤ overnight at 4°C and then incubated with Alexa Fluor 568 conjugated to anti-rabbit IgG (1:500, Invitrogen) and streptavidin-conjugated Alexa Fluor 488 (1:500, Invitrogen).
To visualize vascular endothelial cells using lectin, sections were incubated with DyLight 594-labeled Lycopersicon esculentum (tomato) lectin (vascular endothelial cell marker, 1:100; Vector Laboratories, catalog no. DL-1177) according to the manufacturer's instructions. The length of lectin ϩ vasculature in the dorsal column of the spinal cord was measured using ImageJ software. The average length was calculated from three sections per sample.
To evaluate the pericyte coverage of endothelial cells, the CD13-or PDGFR␤-positive area in the dorsal column of spinal cord was determined using the ImageJ area measurement tool and divided by the lectin-positive capillary area (20). The average ratio was calculated from three sections per sample.
To evaluate the pericyte damage, the number of cleaved caspase-3/PDGFR␤ double-positive cells was counted and divided by the number of PDGFR␤-positive cells. We calculated the average length from three sections per sample.
Western Blot Analysis-The tissues or cells were homogenized in lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS) containing protease inhibitors (Roche Applied Science). The lysates were clarified by centrifugation at 13,000 ϫ g at 4°C for 20 min, and the supernatants were collected and normalized for protein concentration. Proteins were separated by 8% SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Immobilon-P, Millipore). After blocking with PBS containing 5% skim milk and 0.05% Tween 20, the membranes were incubated with primary antibodies.
Measurement of Evans Blue Leakage-Evans blue dye (2% in saline, 4 ml/kg; Sigma) was injected intravenously into the tail vein in adult mice under anesthesia. One hour after the injection, mice were transcardially perfused with saline to remove the intravascular dye. The spinal cord tissues (Th7-9) was removed and weighed for quantitative measurement of Evans blue extravasation. Evans blue dye was extracted by incubating tissues in 1 ml of formamide (Sigma) for 48 h at room temperature. The level of Evans blue was quantified using a light absorbance at a wavelength of 620 nm.
Detection of Alexa Fluor 555-conjugated Cadaverine-Mice were intravenously injected with 0.1 ml of Alexa Fluor 555conjugated cadaverine at a concentration of 0.5 mg/ml (Life Technologies, Inc.) under anesthesia. Two hours after the injection, mice were transcardially perfused with saline to remove the intravascular dye. Mice were subsequently perfused with 4% PFA in PBS, and the spinal cords were postfixed for 24 h at 4°C. Low magnification images of cadaverine leakage were obtained using fluorescent microscopy (SZX-16, Olympus). To observe cadaverin leakages under high magnification, tissues were immersed in 30% sucrose/PBS for 24 h at 4°C and embedded in Tissue-Tek OCT compound. We prepared cryosections with a thickness of 30 m and performed immunohis-tochemical analysis using a rabbit anti-mouse CD31 antibody (1:100, Abcam, catalog no. ab28364).
Kluver-Barrera Staining-One day after the operation, mice were perfused with Bouin fixture containing 9% PFA and 7% picric acid. The spinal cord tissues were removed and postfixed with Bouin solution for 2 h. The samples were washed in 70% ethanol, followed by decolorization with 70% ethanol containing lithium carbonate. Subsequently, the samples were dehydrated in an ascending ethanol series and then embedded in paraffin (tissue preparation, T580, FALMA) through Clear Plus (FALMA). The paraffin-embedded tissues were cut into 7-mthick sections using a microtome (RM2145, Leica), and sections were mounted on glass slides. After deparaffinization using Clear Plus, the slides were stained with 0.1% luxol fast blue (LFB) in distilled water at 60°C overnight. The slides were decolorized using both 0.05% lithium carbonate in distilled water and 70% ethanol, followed by Nissl staining with 0.1% LFB in distilled water. After dehydration, coverslips were placed on slides and sealed with Entellan (Merck). All samples were analyzed using a light microscope (BX61; Olympus). The demyelinated area was calculated by subtracting the blue area from the total area in the dorsal horn of the spinal cord using ImageJ 1.44p software. The average demyelinated area was calculated from five samples per groups.
Basso Mouse Scale Score-Recovery of hindlimb motor function was scored by the Basso Mouse Scale open-field locomotor rating scale, which was developed specifically for mice, and it results in scores ranging from 0 (complete paralysis) to 9 (normal mobility) (22). The number of errors at each footstep was measured during 100 steps. Mice were observed individually for 2 min in an open field.
Behavioral Data Recording-Mice were allowed to freely explore the open field. In total, 50 steps were video recorded for each hindlimb (23). Scores were given to every step according to the flowing scheme: 0, normal step; 1, minor error (slight insecurity of foot placement); 2, major error (foot slipped completely from beam surface). The behavioral score is defined as the sum of the score of 50 steps.
Phagocytosis Assay-Phagocytosis activity was evaluated using a phagocytosis assay kit (IgG FITC) (Cayman Chemical) according to the manufacturer's instructions. Phagocytosis activity was measured by a plate reader (SpectraMax M2, Molecular Device).
Data Analysis and Statistics-The data are presented as the means Ϯ S.E. The statistical analyses were performed with Student's unpaired t tests, one-way analysis of variance with post hoc Tukey's tests, Scheffe's tests, or two-way repeated measures analysis of variance. A value of p Ͻ 0.05 was considered as statistically significant.

LPC Perturbs Vascular Barrier Integrity and Induces Pericyte
Loss in Vitro-We first investigated the effect of LPC accumulation on the vascular barrier in vitro; the in vitro BBB consisted of cultured purified astrocytes, brain capillary endothelial cells, and pericytes. To determine whether LPC compromises the integrity of the BBB, we treated these cells with various concentrations of LPC for 24 h and assessed the TEER. LPC treatment significantly reduced the electrical resistance of the BBB (Fig.  1A). We next examined which cell types in the BBB are reactive to LPC. We cultured the cells in the presence of 50 M LPC for 24 h and counted the number of endothelial cells, pericytes, and astrocytes. The number of pericytes was significantly reduced after LPC treatment for 24 h (Fig. 1, B and C). However, LPC did not change the number of endothelial cells or astrocytes compared with control groups (Fig. 1, B, D, and E). We confirmed that LPC treatment does not decrease the number of neurons and microglia (Fig. 1, F and G), whereas LPC treatment reduces the number of MBP ϩ oligodendrocytes (Fig. 1H). These results suggest that pericyte damage is involved in LPC-induced disruption of vascular barrier integrity.
Iloprost Attenuates LPC-mediated Pericyte Loss and Vascular Barrier Dysfunction-We next wished to assess whether prostacyclin signaling can mitigate LPC-induced pericyte dam- age. We first prepared pericyte cultures obtained from rat brain and examined IP receptor expression on pericytes. Immunocytochemical analysis revealed IP receptor expression in PDGFR␤ ϩ or CD13 ϩ pericytes ( Fig. 2A). To determine whether IP receptor activation modulates LPC-mediated pericyte damage, we pretreated pericytes with or without iloprost, an analog of prostacyclin, and then stimulated the cells with LPC. Immunocytochemical analysis revealed that LPC treatment increases the ratio of PDGFR␤ ϩ caspase-3 ϩ cells compared with the control and that iloprost treatment inhibits the LPC-mediated increase in the number of caspase-3 ϩ pericytes (Fig. 2B). As a control, we found that  iloprost treatment alone was similar to control conditions (data not shown). We next asked whether prostacyclin signaling could prevent LPC-mediated vascular barrier disruption in the BBB model. We pretreated cultures with iloprost for 15 min, and then added LPC to the cultures. Treatment with iloprost prevented LPC-mediated reduction of the TEER (Fig. 2C). There was no significant difference in TEER value between the control and iloprost groups (Fig. 2C). We evaluated whether these changes of TEER value are mediated by vascular endothelial cells. Single culture of endothelial cells resulted in a lower TEER value than that observed in the BBB model, as reported previously (24), and pharmacological treatment did not induce significant changes in the TEER value (Fig. 2D), suggesting that the involvement of endothelial cells prevented the change of TEER in the BBB model. We next examined whether prostacyclin affects LPC-mediated oligodendrocyte loss, but we did not detect a significant difference in the MBP ϩ cell number in the presence or absence of iloprost (Fig. 2E). We also confirmed that MBP expression in oligodendrocytes is not affected by treatment with LPC and/or iloprost (Fig. 2F). These data indicate that prostacyclin signaling rescues LPC-mediated pericyte damage and consequently protects LPC-mediated vascular barrier dysfunction in vitro.

Prostacyclin Signaling Prevents LPC-mediated Vascular Barrier Disruption and Degradation of Tight Junction Proteins in
Vivo-As our in vitro evidence positions prostacyclin signaling as an important player for vascular barrier integrity, we next studied in vivo effects of LPC and prostacyclin signaling. Specifically, we examined whether LPC disrupts vascular barrier integrity in the spinal cord of adult mice in vivo. We intravenously injected Alexa Fluor 555-conjugated cadaverine to observe the leakage of blood component to the CNS. Mice with LPC injection showed a fluorescent cadaverine signal in the spinal cord (Fig. 3A), suggesting that LPC injection leads to disruption of the vascular barrier in the spinal cord. To examine the role of prostacyclin in vascular barrier disruption, we intrathecally administered iloprost or saline to mice immediately after LPC injection. Treatment with iloprost inhibited LPC-mediated cadaverine leakage into the spinal cord. Obser- vation at high magnification showed that the leaked cadaverine was distributed in the cell around the injection site (Fig. 3A), and this distribution is not detected in sections from the mice treated with a combination of LPC and iloprost (Fig. 3A). We next investigated endogenous protein leakage through the vasculature by LPC injection. Immunohistochemical analysis showed that LPC causes the leakage of serum IgG through the vasculature in the spinal cord (Fig. 3A) and that iloprost treatment inhibits LPC-mediated IgG leakage into the spinal cord (Fig. 3A).
To quantitatively assess the disruption of vascular barrier integrity, we measured the volume of Evans blue dye in the spinal cord after LPC injection. In particular, we detected a large amount of Evans blue dye in the spinal cord tissue after LPC injection (Fig. 3B) and observed that Evans blue leakage in the spinal cord of mice was decreased by treatment with iloprost (Fig. 3B). Because the vascular barrier is maintained by the interendothelial tight junctions that form impermeable seals between the cells (4), we further analyzed whether LPC injection degenerates tight junction proteins in the mouse spinal cord tissue. Western blot analysis revealed that LPC injection decreases the protein expression of claudin-5, occludin, and ZO-1 in the spinal cord tissue around the lesion site (Fig. 3, C  and D). These findings support the result that LPC injection disrupts vascular barrier integrity. To investigate whether prostacyclin also prevents LPC-mediated vascular barrier disruption and decreased expression of tight junction-related proteins, we analyzed the expression of proteins related to tight junction formation in the spinal cord and found that iloprost treatment significantly prevented LPC-induced decreases of protein expression 1 day after the operation (Fig. 3, C and D). These data indicate a protective role of prostacyclin-mediated signaling in LPC-dependent disruption of the vascular barrier in the CNS.
Prostacyclin Signaling Prevents LPC-mediated Loss of Pericyte Coverage in the Spinal Cord-We next assessed whether prostacyclin inhibits pericyte loss in vivo. We investigated the change in pericyte coverage after LPC injection. Immunohistological analysis revealed that LPC injection decreases the degree of pericyte coverage, based on staining for CD13 or PDGFR␤ on vascular endothelial cells (Fig. 4, A-C). Intrathecal administration of iloprost abolished the LPC-mediated reduction of pericyte coverage (Fig. 4, A-C), suggesting that iloprost treatment prevents LPC-mediated pericyte loss in the CNS. We examined the total length of lectin ϩ neovessels in the sections and found that there were no significant differences between the groups (Fig. 3D). These results indicate that the change in pericyte coverage does not result from the loss of vascular endothelial cells.
Because LPC induces pericyte apoptosis in vitro, we examined whether LPC injection increases the number of caspase-3 ϩ pericytes in vivo. Mice that received LPC injection presented increased numbers of caspase-3-positive pericytes (Fig. 4, E and F). Intrathecal administration of iloprost inhibited the LPC-mediated increase in the number of caspase-3-positive pericytes. These results demonstrate that prostacyclin signaling reduces LPC-mediated pericyte damages.

Prostacyclin Signaling Prevents Demyelination and Motor Impairments after LPC Injection into the Spinal Cord-To
determine whether the protective effect of prostacyclin on pericytes attenuates pathological events caused by vascular barrier disruption, we prepared spinal cord sections from mice that were injected with LPC with or without intrathecal administration of iloprost. We estimated the demyelination area by Kluver-Barrera's staining (Fig. 5A). Iloprost treatment effectively reduced demyelination in the spinal cord after LPC treatment (Fig. 5, A and B), which is presumably caused by rescuing vascular barrier integrity (Fig. 5, A and B).
Some myelinated fibers in the dorsal column of the spinal cord regulate motor behavior, and therefore, LPC injection into the dorsal spinal cord injures these fibers and causes motor dysfunction. To investigate whether iloprost treatment ameliorates motor function, we performed two behavioral tests to assess lower limb function. Treatment with iloprost significantly improved motor function of the lower limbs as assessed by the Basso Mouse Scale score and behavioral data recording in the acute phase. The therapeutic effect of prostacyclin could be observed throughout the observation periods (Fig. 5, C and  D). Thus, by presumably protecting vascular barrier function, prostacyclin signaling ameliorates motor deficits induced by LPC injection into the spinal cord.
Protection against Demyelination by Prostacyclin Is Not Associated with Altered Phagocytosis Activation-It is known that LPC induces the recruitment of activated macrophages in the CNS, which associates with rapid clearance of myelin debris by phagocytosis (17). Therefore, we determined whether the effect of prostacyclin and LPC on demyelination results from changes in phagocytosis activity. We prepared brain microglia and bone marrow-derived macrophages from mice and treated the cells with LPC and iloprost. We found that LPC did not directly increase the phagocytosis activity in microglia and macrophages, respectively. In addition, we detected no change of phagocytosis activity in the presence or absence of iloprost (Fig.  5, E and F), suggesting that prostacyclin-mediated protection against demyelination does not result from modulation of phagocytosis. It is known that several cytokines lead to rapid phagocytosis (25); we conclude that LPC and iloprost do not change cytokine expression in macrophage and microglia.

DISCUSSION
LPC damages the peripheral vasculature by causing cytoskeletal changes in vascular endothelial cells and by inducing cell death in vascular smooth muscle cells (26). Although LPC injection into the adult CNS also disrupts the vasculature, this effect was attributed to immunocyte activation (17). However, our experiment using an in vitro BBB model suggests that LPC directly disrupts the vascular barrier independent of immune cells. More specifically, we found that LPC leads to pericyte apoptosis in vitro and also causes pericyte loss without disruption of vascular endothelial cells in vivo. These observations suggest that LPC at least partly acts directly on the vasculature, especially pericytes, and disrupts the vascular barrier. The signal transduction mechanism of LPC-mediated disruption of the vascular barrier and pericyte loss remains an open question. Ovarian cancer G protein-coupled receptor 1 (also known as G protein-coupled receptor-68), GPR4, and G2A have been identified as receptors for LPC. Although GPR4 is critically involved in the impairment of endothelial barrier function by LPC (27), it is still unknown which receptor subtypes are associated with the LPC-mediated cell death. Further studies are required to identify the receptor and the downstream signal transduction of LPC-mediated pericyte damage.
Under pathological conditions in the CNS, intraventricular treatment of the stable analog of prostacyclin has neuroprotective effects by increasing blood flow, reducing platelet aggregation, and directly protecting neurons (28,29). In addition, we previously revealed that prostacyclin treatment directly enhanced axon regeneration and remyelination in the injured adult CNS (30,31). Designing therapies that enhance prostacyclin signaling is thus considered beneficial for CNS disorders, and clarifying the effect of prostacyclin on the homeostasis of CNS vasculature is required to ensure the therapeutic potential of prostacyclin. This study describes a previously unidentified role of prostacyclin that can potentially preserve the vascular integrity so often undermined in the pathological CNS. We should note that endogenous prostacyclin expression is considered abundant in the vascular endothelium, and this context provides the possibility that constitutive expression of prostacyclin in endothelial cells supports the homeostasis of the vasculature. This hypothesis is consistent with our previous findings that prostacyclin secreted by neovessels acts on IP receptors on both injured axons (30) and oligodendrocyte precursor cells (31) by facilitating axon regeneration and remyelination in the adult spinal cord. The novel aspects of the present findings are as follows: prostacyclin protects against pericyte  damage but prostacyclin does not have play a protective role in oligodendrocytes at the start of demyelination. Recent studies have shown multiple functions of pericytes, including BBB formation and maintenance, vascular stability and angio-architecture, regulation of capillary blood flow, and clearance of toxic cellular by-products necessary for normal functioning of the CNS (3)(4)(5)(6)(7)(8). Pericyte degeneration is identified in neurodegenerative disorders such as sporadic amyotrophic lateral sclerosis (32) and Alzheimer disease (33). Although evidence for the protective role of pericytes on CNS homeostasis in humans is incomplete, a study using an Alzheimer disease model mouse demonstrated that pericyte loss leads to progression of an Alzheimer disease-like neurodegeneration pathogenic cascade (34), suggesting a direct link between pericyte loss and the progression of neurodegeneration. With regard to other pericyte functions that are not limited to vascular-related events, it has been demonstrated that pericytes can sense inflammation and promote leukocyte transmigration from the vessel and behave as a stem cell population capable of commitment to neuronal cells (35).
Another role of pericytes under pathological states is a tissue repair response to CNS injury, the scar tissue that forms at injuries seals lesions in the CNS. Blocking pericyte generation failed to seal the injured tissues, showing that pericytes are involved in the formation of spinal cord scar tissue (36). Thus, pericytes contribute to tissue connection and repair of the injured spinal cord. Continued search for the molecular cues that prevent pericyte loss in CNS diseases may ultimately lead to the discovery of new therapies to control pericyte loss and the progression of neurodegeneration.