Interleukin-25 Expressed by Brain Capillary Endothelial Cells Maintains Blood-Brain Barrier Function in a Protein Kinase Cϵ-dependent Manner*

Interleukin (IL)-25, a member of the IL-17 family of cytokines, is expressed in the brains of normal mice. However, the cellular source of IL-25 and its function in the brain remain to be elucidated. Here, we show that IL-25 plays an important role in preventing infiltration of the inflammatory cells into the central nervous system. Brain capillary endothelial cells (BCECs) express IL-25. However, it is down-regulated by inflammatory cytokines, including tumor necrosis factor (TNF)-α, IL-17, interferon-γ, IL-1β, and IL-6 in vitro, and is also reduced in active multiple sclerosis (MS) lesions and in the inflamed spinal cord of experimental autoimmune encephalomyelitis, an animal model of MS. Furthermore, IL-25 restores the reduced expression of tight junction proteins, occludin, junction adhesion molecule, and claudin-5, induced by TNF-α in BCECs and consequently repairs TNF-α-induced blood-brain barrier (BBB) permeability. IL-25 induces protein kinase Cϵ (PKCϵ) phosphorylation, and up-regulation of claudin-5 is suppressed by PKCϵ inhibitor peptide in the IL-25-stimulated BCECs. These results suggest that IL-25 is produced by BCECs and protects against inflammatory cytokine-induced excessive BBB collapse through a PKCϵ-dependent pathway. These novel functions of IL-25 in maintaining BBB integrity may help us understand the pathophysiology of inflammatory brain diseases such as MS.

Blood-brain barrier (BBB) disintegrity is commonly associated with synaptic and neuronal dysfunction in various disorders, such as multiple sclerosis (MS), Alzheimer disease, Parkinson disease, and amyotrophic lateral sclerosis (9,10). The role of BBB disruption has been investigated in these neuronal disorders, especially in MS and its animal model, experimental autoimmune encephalomyelitis (EAE), in which BBB disruption is associated with an increased infiltration of autoreactive inflammatory immune cells, resulting in tissue destruction and neurological impairment (11). Disruption of tight junction (TJ) molecules such as occludin, junctional adhesion molecule (JAM), and claudin-5 by inflammatory cytokines, including tumor necrosis factor (TNF)-␣, interferon (IFN)-␥, IL-1␤, and IL-17, is thought to contribute to the transmigration of inflammatory lymphocytes (12)(13)(14)(15)(16)(17). For example, Fabry et al. (13) showed that TNF-␣, IFN-␥, and IL-1␤ significantly up-regulated leukocyte migration through the BBB by reducing occludin. Afonso et al. (12) indicated that TNF-␣ and IL-1␤ decreased occludin expression and increased endothelial cell permeability and transcellular migration of T lymphocytes. In addition, Förster et al. (14) demonstrated that TNF-␣ had an inhibitory effect on occludin and claudin-5 expression in human brain capillary endothelial cells (BCECs). Consistent with these findings, TNF-deficient mice have delayed EAE onset compared with wild-type mice, suggesting that TNF-␣ plays an important role in inflammatory T cell infiltration into the central nervous system (18). Recently, Th17 cytokines, IL-17 and IL-22, were reported to down-regulate zonula occludens-1 and occludin expression in human BCECs and to promote inflammatory T cell infiltration (15). Taken together, these results suggest that inflammatory cytokines play a key role in enhancing BBB permeability in MS and EAE.
Interestingly, IL-25 mRNA is also detected in the whole brain tissue of normal mice (19), despite the fact that the brain is a fundamentally immunoprivileged site and is not correlated with allergic inflammation. Moreover, it has been reported that IL-25-deficient mice are highly susceptible to EAE because of reduced IL-13 production (20). These findings suggest that IL-25 plays an important role in the maintenance of the central nervous system, especially in the development of MS or EAE. Although IL-25-deficient mice with EAE die (20), those without immunization survive (21) even though they have an increased number of Th17 in periphery. Thus, there may be some suppressive mechanisms that require IL-25 to induce inflammation in the central nervous system.
Here, we show that IL-25 controls BBB disruption induced by TNF-␣ in the BCEC line, MBEC4, and primary BCECs. A permeability assay and transmigration assay confirm that IL-25 suppresses the TNF-␣-induced increase in permeability and T cell migration across the BBB. IL-25 increases TJ proteins, occludin, JAM, and claudin-5, and restores the reduced expression of them, induced by TNF-␣ through a protein kinase C⑀ (PKC⑀)-mediated pathway. Additionally, RT-PCR and immunohistochemistry indicate that the IL-25-expressing cells are endothelial cells in the brain. Collectively, these results suggest that IL-25 produced by BCECs protects from TNF-␣-induced excessive collapse of the BBB and inflammatory CD4 ϩ T cell infiltration across the BBB in MS or EAE.
Cell Cultures-All animal experiment protocols were approved by the Animal Experiment Committee of Nagoya University. All primary cell cultures were prepared from C57BL/6J mice (Japan SLC).
Primary microglia were isolated from mixed glial cell cultures prepared from newborn mice on day 14 using the "shaking off" method as described previously (22,23). The purity of the cultures was almost 100%, as determined by immunostaining with anti-CD11b antibodies. Astrocytes were prepared as described previously (24). Briefly, mixed glial cell cultures were trypsinized after the microglia were collected and replated in Petri dishes. Cultures that had undergone two passages were used as astrocytes. The purity of the cultures was greater than 95% as determined by immunostaining with anti-glial fibrillary acidic protein antibodies.
Primary neurons were prepared from the neocortices of embryonic day 17 embryos as described previously (25,26). The purity of the cultures was more than 95% as determined by NeuN-specific immunostaining. Primary meningeal fibroblasts were prepared by mincing meninges of newborn mice after removing the blood vessels and then cultured for 10 days before each assay.
Primary BCECs were purified from adult mice as reported by Ge et al. (27) and Ohtsuki et al. (28). In brief, the cerebrums were homogenized and separated by a 16% dextran solution. This homogenate was centrifuged, and the resulting pellet was filtered through a 40-m cell strainer. The microvessels retained in the cell strainer were then digested in PBS containing 0.5 mg/ml Liberase Blendzyme Type I (Roche Applied Science) and 120 units/ml DNase (Sigma) at 37°C for 30 min. CD31positive cells were then sorted using the magnetic-activated cell sorting (MACS) system. The purified cells were washed and resuspended in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 10% horse serum, 50 g/ml heparin, and endothelial cell growth supplement (Sigma) and then plated onto tissue culture dishes coated with collagen (Roche Applied Science). The purity was Ͼ98%, as determined by DiIlabeled acetylated low density lipoprotein uptake.
The BCEC line, MBEC4 (a kind gift from Dr. T. Tsuruo) (29), was maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and used as an established BBB model.
RNA Isolation and Semi-quantitative RT-PCR-Total RNA was extracted from all primary cells and SV40-transduced human BCEC lines (established by T. Kanda, Yamaguchi University, Japan) using the RNeasy mini kit (Qiagen). After 0.2 g of total RNA was denatured for 5 min at 65°C, the RT reaction was performed as described previously (22).
EAE-Myelin oligodendrocyte glycoprotein (MOG)-induced EAE was generated as described previously (30,31). In brief, mice were injected subcutaneously with 0.2 ml of emulsion containing 200 g of MOG-  in PBS combined with an equal volume of complete Freund's adjuvant (CFA) containing 300 g of heat-killed Mycobacterium tuberculosis H37Ra. Mice were injected with pertussis toxin intraperitoneally on the day of immunization and 2 days after immunization (200 ng/mouse). EAE mice showed disease onset 14 days after immunization, which peaked 18 days post-immunization as shown in supplemental Fig. 1.
For passive transfer, splenocytes were harvested from C57BL/6 mice 12 days after immunization with MOG and cultured in RPMI 1640 medium supplemented with 10% fetal calf serum, L-glutamine, and sodium pyruvate in the presence of MOG-(35-55) for 4 days. Then CD4 ϩ T cells were sorted using CD4 ϩ microbeads (Miltenyi Biotec) with Ͼ95%. Dead cells were removed by Ficoll (GE Healthcare). Cells were then washed and injected intravenously into naive C57BL/6 mice (5 ϫ 10 6 cells/mouse). Animals received 200 ng/mouse pertussis toxin on day 0 and day 2 after transfer. IL-25 was administered at 500 ng/mouse/day from day Ϫ2 to day 14 every other day.
BBB Permeability and CD4 ϩ T Cell Transmigration Assay-The permeability of MBEC4 and primary BCEC monolayers was evaluated using the permeability marker, Evan's blue albumin (EBA), as described previously (32) and by measuring transendothelial electrical resistance (TER). Confluent monolayers of MBEC4 cells on the Transwell inserts (3-m pore size) were incubated with or without 50 ng/ml TNF-␣ and 50 ng/ml IL-25 for 24 h. The monolayers were then washed with assay buffer (118 mM NaCl, 4.7 mM KCl, 1.3 mM CaCl 2 , 1.2 mM MgCl 2 , 1.0 mM NaH 2 PO 4 , 25 mM NaHCO 3 , and 11 mM D-glucose (pH 7.4)). This buffer (1 ml) was added to the outside of the insert (the abluminal side). Assay buffer containing 4% bovine serum albumin (Sigma) mixed with 0.67 mg/ml Evan's blue dye (Sigma) was loaded on the luminal side of the insert for 1 h. The concentration of EBA in the abluminal chamber was determined by measuring the absorbance of aliquots at 630 nm with a microplate reader. Clearance was calculated as follows: where V A is the volume of the abluminal side, [C] A is the concentration of EBA on the abluminal side, and [C] L is the concentration of EBA on the luminal side.
TER was measured using a Millicell-ERS (Millipore). Resistances of blank filters were subtracted from those of filters with cells before final resistances (⍀/cm 2 ) were calculated.
To assess CD4 ϩ T cell transmigration, we sorted CD4 ϩ T cells from the spleens of EAE mice using the MACS system. After 500 l of RPMI 1640 medium containing 10% fetal bovine serum, 2 mM L-glutamine, and 1 mM sodium pyruvate was added to the abluminal side, sorted CD4 ϩ T cells were then loaded onto the luminal side and incubated for 24 h at 37°C, 5% CO 2 . Following migration, 50 l of 0.5 M EDTA was added to the abluminal side, and the plates were placed on a shaker for 15 min to mobilize cells. Cells in the abluminal side were then harvested and counted by flow cytometry (Cytomics FC500, Beckman Coulter).
Western Blotting-Cells were lysed in TNES buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Nonidet P-40, 2 mM EDTA, and 0.1% SDS) with a protease inhibitor mixture (Roche Applied Science) and a phosphatase inhibitor mixture (Sigma). Thirty g of protein from the total lysate was examined by Western blotting as described previously (31).
Immunohistochemistry-Frozen brain and spinal cord sections from normal mice were fixed with 4% paraformaldehyde and stained for IL-17E and vWF followed by Alexa 488-conjugated anti-goat IgG and Alexa 568-conjugated anti-rabbit IgG as described previously (25,26). The stained sections were examined with a deconvolution fluorescence microscope system (BZ-8000, Keyence) and analyzed by Dynamic Cell Count image analysis program (Keyence).
For neuropathologic examination of MS patients, sections from paraffin-embedded blocks of the cerebrum were stained with hematoxylin and eosin and subjected to IL-25 immunostaining with the labeled StreptAvidin-Biotin method. In brief, sections from paraffin-embedded blocks of the cerebrum from three MS patients (46 years old, female with a 3-year history; 54 years old, female with a 24-year history; and 71 years old, female with a 1-month history) were stained with hematoxylin and eosin and subjected to IL-25 immunostaining. Eight-micrometer-thick sections were placed on silanized slides (Dako, Glostrup, Denmark). Immunohistochemistry for IL-25 was performed with monoclonal antibody against mouse IL-17E (dilution, 1:100; Santa Cruz Biotechnology) and the labeled StreptAvidin-Biotin method. After deparaffinization, nonspecific endogenous peroxidase activity was blocked by pretreatment with 3% H 2 O 2 for 5 min. For antigen retrieval, the sections were treated at 100°C for 30 min in beakers filled with 10 mM citrate buffer. After the temperature had decreased to 60°C slowly over 1 h, the sections were removed. After a wash in PBS, Fc receptors were blocked with 10% normal nonimmune goat serum (Dako) for 30 min at room temperature prior to incubation with IL-25 antibody. Sections were treated with anti-IL-25 monoclonal antibody for 2 h at room temperature and washed in PBS. Sections were then incubated at room temperature with biotin-conjugated anti-goat IgG for 20 min and washed in PBS, followed by incubation at room temperature with streptavidinhorseradish peroxidase for 20 min. Immunoreactivity was visualized with 3,3Ј-diaminobenzidine/peroxidase (Wako). IL-25positive endothelial cells were enumerated in both active MS lesions identified as demyelinated area with cellular infiltration and without these findings (intact area) in hematoxylin and eosin.
Statistical Analysis-Differences between the means of experimental groups were analyzed using the two-tailed Student's t test with Welch's correction.

JOURNAL OF BIOLOGICAL CHEMISTRY 31837
These cells were markedly reduced after the development of EAE (18 days after immunization), but IL-25 expression significantly increased in the EAE recovery phase (25 days after immunization) (Fig. 1, C and  D). IL-25 administration ameliorated EAE induced by passive transfer of MOG-reactive CD4 ϩ T cells (supplemental Fig. 3).
To further explore whether inflammatory cytokines down-regulated IL-25 expression in BCECs, we assessed IL-25 mRNA expression by real time RT-PCR in the BCEC line, MBEC4. Although TNF-␣, IL-17, IFN-␥, IL-1␤, and IL-6 reduced IL-25 mRNA expression in MBEC4 cells, TGF-␤ and IL-10 increased it (Fig. 1E). In addition, a decrease of IL-25 expression in BCECs was also observed in the active MS lesion (Fig. 2, A-F). The number of IL-25-expressing endothelial cells in the severe MS lesion was significantly less than in the normal appearing white matter (Fig.  2G). Thus, these results suggest that IL-25 is expressed by BCECs and may play a novel role in maintaining BBB function to block inflammatory cell infiltration into the central nervous system.

IL-25 Prevents TNF-␣-induced
Permeability of the BBB-To confirm whether IL-25 rescues TNF-␣disrupted BBB functions, we performed a permeability assay and CD4 ϩ T cell transmigration assay by measuring the levels of EBA, TER, and the number of transmigrated CD4 ϩ T cells through confluent layers of MBEC4 cells. TNF-␣ increased the permeability of EBA and the transmigration of CD4 ϩ T cells and decreased TER. Adding IL-25 to TNF-␣-stimulated BCECs significantly reduced EBA levels on the abluminal side of the well (the outside of the insert) and up-regulated the electrical resistance of BCECs compared with cells treated with TNF-␣ alone (Fig. 3, A and B). Moreover, the addition of IL-25 significantly decreased TNF-␣-induced CD4 ϩ T cell transmigration across the BBB (Fig. 3C). Although IL-25 alone did not significantly change the levels of EBA and TER (Fig. 3, A and B), it significantly suppressed the number of transmigrated CD4 ϩ T cells on the abluminal side compared with the unstimulated control (Fig. 3C). In addition, BCECs expressed the IL-25 receptor, and TNF-␣ did not change the expression levels (supplemental Fig.  4). These results further support the notion that IL-25 prevents the disturbance of BBB permeability. (12)(13)(14). To elucidate whether IL-25 affects this reduction in TJ protein expression by TNF-␣, we assessed the effect of IL-25 on the expression of occludin, JAM, and claudin-5. IL-25 significantly increased their expression levels in a dose-dependent manner (Fig. 4, A and B). Among these proteins, claudin-5 expression was most strongly enhanced by IL-25. Furthermore, IL-25 dose-dependently reversed the TNF-␣-induced down-regulation of occludin, JAM, and claudin-5 protein levels (Fig. 4, C and D). IL-25 restored the TNF-␣-induced decrease of occludin expression more effectively than any other molecules that have been reported as stabilizers of BBB such as IFN-␤, TGF-␤, dexamethasone, angiotensin II, and simvastatin (supplemental Fig. 5). IL-25 more effectively restored the TNF-␣-induced reduction of JAM and claudin-5 proteins as the restoration of occludin (data not shown).

IL-25 Blocks the Impaired Expression of TJ Proteins Caused by TNF-␣ in BCECs-Many reports indicate that TNF-␣ downregulates TJ protein expression in human and mouse BCECs
To further clarify whether IL-25 has the same effect on primary BCECs, these cells were purified   NOVEMBER 13, 2009 • VOLUME 284 • NUMBER 46

IL-25 Repairs BBB Disruption
using MACS system. The final purity of the primary BCECs was approximately Ͼ90%, as determined by morphology, vWF staining, and low density lipoprotein uptake (supplemental Fig.  6, A-C). As found in MBEC4 cells, IL-25 increased the expression of occludin, JAM, and claudin-5 in purified primary BCECs and reversed the TNF-␣-induced decrease in their expression levels (supplemental Fig. 6, D and E). Collectively, these results suggest that IL-25 blocks TNF-␣-induced disruption of BBB function by up-regulating the expression of occludin, JAM, and claudin-5 in MBEC4 cells and primary BCECs.

DISCUSSION
IL-25 is known to induce Th2 responses and to be produced by cells related to allergic responses such as Th2 cells, mast cells, and lung macrophages stimulated with anti-CD3/CD28, IgE, and allergens (1,3,4). These findings suggest that IL-25 is an inflammatory mediator during Th2-mediated inflammation. On the other hand, a previous study indicated that IL-25 mRNA was expressed in the whole brain tissue of normal mice (19), indicating that IL-25 also plays a distinct role in the central nervous system. Consistent with this hypothesis, in this study we revealed that IL-25 protected excessive disruption of the BBB, resulting in decreased inflammatory T cell infiltration.
Kleinschek et al. (20) showed that IL-25-deficient mice immunized with MOG peptide/CFA had increased inflammatory T cell infiltration and a high mortality rate. These findings support our results that IL-25 reverses excessive BBB disruption and inflammatory T cell infiltration. Although Kleinschek et al. (20) also showed that IL-25 was produced by microglia, we found that unstimulated microglia did not express IL-25 mRNA. Additionally, we found that microglia did not express IL-25 mRNA even when stimulated with LPS, IFN-␥, IL-17, TNF-␣, IL-6, IL-4, or IFN-␤ in vitro (supplemental Fig. 2). We also previously reported that microglia did not produce the Th2  cytokine, IL-4 (35). Microglia produced IL-5 at low levels, just above the detection limit (ϭ 5 pg/ml) (36). Thus, microglia probably do not favor Th2 cytokine production, especially under normal conditions. Our results do not exclude the possibility that other factors induce IL-25 in microglia. However, at least in normal mouse brain, IL-25 mRNA is expressed in BCECs. Accordingly, BCECs are able to regulate their own permeability via producing IL-25. Additionally, Kleinschek et al. (20) showed that IL-25 suppressed EAE through inducing production of IL-13 but not IL-4 (20). However, they also showed that IL-13-deficient mice immunized with MOG peptide/CFA survived, although all IL-25-deficient mice immunized died. This implies that IL-25 suppresses EAE not only through an IL-13-mediated pathway but also via another mechanism. It is possible that IL-25 suppresses EAE by ameliorating the disruption of BBB permeability.
Although in this study we showed that IL-25 augments BBB properties, other molecules have also been reported to increase TJ protein expression. TGF-␤1 enhances BBB functions in MBEC4 cells (37). Human brain microvascular endothelial cells pretreated with IFN-␤, an approved drug for the treatment of MS, decrease Th1 cell migration and suppress increased paracellular permeability of small molecules such as [ 3 H]inulin and [ 14 C]sucrose (38). Angiotensin II, which controls cerebral blood flow, memory retention, and neuronal regeneration, restricts the passage of molecular tracers across the BBB and is required for BBB maintenance (39). Glucocorticoid up-regulates occludin and claudin-5 expression and increases barrier properties (40). Statins, simvastatin and lovastatin, which are 3-hydroxy-3-methylglutaryl-CoA reductase inhibitors and have an anti-inflammatory effect, reduce human BBB permeability (41). All of these molecules were reported previously to have a suppressive effect against MS or EAE or were established as effective drugs for treating these diseases. As IL-25 reportedly has a suppressive effect on EAE (20), and BBB disruption is critical in the development of the disease, these also support our findings that IL-25 restores disrupted BBB permeability. Moreover, IL-25 blocks the TNF-␣-induced permeability more effectively than any other molecules by restoring the TNF-␣-induced decrease of TJ protein expression (supplemental Fig. 5).
PKC signaling has been shown to affect endothelial barriers. PKC includes 12 isozymes that are classified into three subtypes by differences in their mechanisms of action: conventional (␣, ␤I, ␤II, and ␥), novel (␦, ⑀, , , and ), and atypical (, , and ). Rat primary BCECs reportedly express PKC␣, -␤, -␦, -⑀, andisozymes (42). Some of these PKC isozymes appear to be associated with TJ assembly or openings in the BBB (33). Hypoxia/ aglycemia-mediated permeability augmentation was inhibited by a conventional PKC inhibitor, Gö6976, suggesting that PKC␣ or -␤ signaling may enhance TJ opening (43). Alterations in permeability induced by chemokine monocyte chemoattractant protein-1 (MCP-1)/CCL2 were also mediated by PKC␣ (44). Furthermore, inhibiting PKC␦ protects against hypertensive encephalopathy by preventing a breakdown of the BBB in rats (45), suggesting that PKC␦ signaling may lead to BBB disruption. Moreover, it is reported that PKC is up-regulated following hypoxia, which increases BBB paracellular permeability, suggesting that PKC may also be involved in BBB dis-ruption (46). Collectively, many of the PKC isozymes are reported to be associated with increased BBB permeability. However, to our knowledge, the role of PKC⑀ in BBB maintenance has not been previously reported. In this study, we demonstrated for the first time that the activation of PKC⑀ up-regulated claudin-5 and protected the BBB from TNF-␣-induced disruption. In Caco-2 cells, a human colorectal carcinoma, epidermal growth factor-mediated activation of PKC⑀ reportedly protects TJs from acetaldehyde (47), which support our findings in this study.
The content of the TJ proteins in the tissue or isolated BBB endothelial cells reportedly correlates well with barrier properties (48 -51). The high TJ protein levels may account for the maintenance and protection of the BBB integrity. In this study, we showed that IL-25 increased the expression levels of occludin, JAM, and claudin-5 through PKC⑀-dependent signaling. Several pathways are reported as downstream of PKC⑀-dependent signaling. For example, Weiler et al. (52) demonstrated that PKC activation leads to the increased expression of occludin through MEK1/MEK2 signaling. On the other hand, phosphorylation and redistribution of tight junction proteins are reportedly required for the barrier function. Suzuki et al. (47) showed that PKC⑀ enhanced epidermal growth factor-mediated redistribution and induction of occludin in Caco-2 cells. In addition, Yoo et al. (53) revealed that PKC⑀-dependent signaling pathway led to phosphorylation of occludin in T84 epithelia. These suggest that IL-25 protects BBB function via up-regulation and probably direct phosphorylation of tight junction proteins through PKC⑀-mediated pathway. In conclusion, our results suggest that IL-25 produced by BCECs protects from TNF-␣-induced excessive collapse of the BBB and inflammatory T cell infiltration across the BBB through a PKC⑀-dependent pathway in MS or EAE.