Ninjurin1 Deficiency Attenuates Susceptibility of Experimental Autoimmune Encephalomyelitis in Mice*

Background: Effect of Ninjurin1 deletion in the experimental autoimmune encephalomyelitis (EAE) mice has not been examined. Results: Ninjurin1 knock-out (KO) mice are resistance to EAE due to a defect of leukocyte recruitment into lesion sites. Conclusion: Ninjurin1 is a potent target molecule for treating inflammatory diseases such as multiple sclerosis. Significance: Our study proved contribution of Ninjurin1 in EAE pathogenesis in vivo and supports the importance of its targeting strategies. Ninjurin1 is a homotypic adhesion molecule that contributes to leukocyte trafficking in experimental autoimmune encephalomyelitis (EAE), an animal model of multiple sclerosis. However, in vivo gene deficiency animal studies have not yet been done. Here, we constructed Ninjurin1 knock-out (KO) mice and investigated the role of Ninjurin1 on leukocyte trafficking under inflammation conditions such as EAE and endotoxin-induced uveitis. Ninjurin1 KO mice attenuated EAE susceptibility by reducing leukocyte recruitment into the injury regions of the spinal cord and showed less adhesion of leukocytes on inflamed retinal vessels in endotoxin-induced uveitis mice. Moreover, the administration of a custom-made antibody (Ab26–37) targeting the Ninjurin1 binding domain ameliorated the EAE symptoms, showing the contribution of its adhesion activity to leukocyte trafficking. In addition, we addressed the transendothelial migration (TEM) activity of bone marrow-derived macrophages and Raw264.7 cells according to the expression level of Ninjurin1. TEM activity was decreased in Ninjurin1 KO bone marrow-derived macrophages and siNinj1 Raw264.7 cells. Consistent with this, GFP-tagged mNinj1-overexpressing Raw264.7 cells increased their TEM activity. Taken together, we have clarified the contribution of Ninjurin1 to leukocyte trafficking in vivo and delineated its direct functions to TEM, emphasizing Ninjurin1 as a beneficial therapeutic target against inflammatory diseases such as multiple sclerosis.


Ninjurin1 is a homotypic adhesion molecule that contributes
Immune responses mediated by leukocytes from blood have a great influence on the pathogenesis of several inflammatory central nervous system (CNS) diseases including multiple sclerosis, stroke, Alzheimer disease, Parkinson disease, trauma, and epilepsy (1). Multiple sclerosis is caused by autoimmune T cells coordinating with monocytes, dendritic cells, and B cells recruited from the blood; it is one of the most common diseases with neuromyelitis and encephalomyelitis (2). In particular, the remarkable accumulation of leukocytes in the plaque of a damaged region is a hallmark of multiple sclerosis and its animal model, EAE. 4 Therefore, there have been active attempts to modulate leukocyte diapedesis in these CNS diseases (3).
Ninjurin1 was originally identified as an adhesion molecule, ϳ17 kDa by Araki et al. with two transmembrane domains (14). The 12 residues, from Pro 26 to Asn 37 , on the N-terminal ectodomain of Ninjurin1 are critical for its homophilic binding (15). Our group previously reported that Ninjurin1 is preferentially expressed in myeloid cells and in the inflamed endothelium in the EAE rat brain and that its overexpression promotes the adhesion of leukocytes onto endothelial cell monolayers in vitro (16,17). Consistent with our results, Ifergan et al. (18) demonstrated that Ninjurin1 is restricted to endothelial cells and myeloid cells, particularly to dendritic cells at lesions from human brains with multiple sclerosis. Furthermore, the functional blockage of Ninjurin1 decreases the transendothelial migration (TEM) of monocytes by blocking rolling and other adhesive steps on endothelial cells, whereas it attenuates the clinical symptoms of EAE mice by decreasing leukocyte infiltration (18). Recently, it has been reported that in highly migratory T cells activated in the lungs of EAE rats, Ninjurin1 is transiently up-regulated and participates in the intravascular crawling of T cells in the CNS vessels (19). These previous results suggest that Ninjurin1 is a beneficial candidate that targets the TEM of leukocytes including myeloid-lineage cells and T cells.
However, the in vivo role of Ninjurin1 in a gene-deficient animal model and its direct regulation via Ninjurin1 own expression pertaining to the processes of TEM should be explored. We herein clarified the in vivo relevance of Ninjurin1 using both KO mice and a blocking antibody generated by immunization with the homophilic binding domain as the specific antigen (Ab 26 -37 ). Ninjurin1 KO and Ab 26 -37 -administrated mice exhibited protective effects against EAE by reducing leukocyte infiltration in the lesion site. In addition to the well known homophilic binding activity of Ninjurin1, we found that Ninjurin1 directly enhances TEM activity in a dose-dependent manner, which is shown in the Ninjurin1 KO bone marrow-derived macrophage (BMDM)s and Ninjurin1 siRNA or stable overexpressing Raw264.7 cells in vitro. Altogether, we determined the role of Ninjurin1 in vivo and discovered that TEM is regulated based on the amount of Ninjurin1 expression during the trafficking of immune cells under inflammatory conditions.

EXPERIMENTAL PROCEDURES
Animals-Ninjurin1 KO mice (C57BL/6J background) were backcrossed with C57BL/6 for at least seven generations. The breeding colony was established and maintained under pathogen-free conditions in the animal housing facility of the College of Pharmacy, Seoul National University, for the duration of the experiments under the rule of the Committee for Care and Use of Laboratory Animals at Seoul National University (SNU-101011-1). The primer sequences for genotyping are as follows: wild type (forward), 5Ј-GAG ATA GAG GGA GCA CGA CG-3Ј; Neo (forward), 5Ј-ACG CGT CAC CTT AAT ATG  CG-3Ј; reverse primer, 5Ј-CGG GTT GTT GAG GTC ATA  CTT G-3Ј. EAE Induction and Clinical Scoring-Sex-and age (6 -10 weeks)-matched C57BL/6 mice were immunized subcutaneously with an emulsion containing 100 g of myelin oligodendrocyte glycoprotein-(35-55) (MOG  ; Peptron Inc.) in complete Freund's adjuvant (CFA; Mycobacterium tuberculosis H37Ra, 4 mg/ml). Each mouse was injected with 300 ng of pertussis toxin intraperitoneally at 0 and 2 days after immunization. The mice were weighed and observed daily for clinical signs of EAE. The progression of EAE was graded according to the following scale: 0, no symptoms; 1, floppy tail; 2, mild paralysis of the hind limbs; 3, complete paralysis of one hind limbs and partial paralysis of another one; 4, complete paralysis of both hind limbs; 5, a moribund state or death.
Endotoxin-induced Uveitis (EIU) and Quantification of Retinal Adherent Leukocytes-Each mouse (male, 6 -10 weeks) received a single intraperitoneal injection of lipopolysaccharide (LPS, Sigma) in PBS at a dose of 9 mg/kg body weight. Twentyfour hours after the LPS injection, the chest cavity was opened, and a 24-gauge syringe was introduced into the left ventricle under deep anesthesia. After injection of 5 ml PBS to remove the erythrocytes and non-adherent leukocytes, 5 ml of rhodamine-conjugated concanavalin A (5 mg/kg) (Vector Laboratories) was perfused. After the eyes were enucleated, the retinas were flat-mounted. The flat mounts were imaged with a fluorescence microscope, and the total number of concanavalin A-positive adherent leukocytes per retina was counted.
Antibodies-For a custom-made anti-moue Ninjurin1 antibody, rabbits were immunized with keyhole limpet hemocyanin-conjugated synthetic peptide-bearing mouse Ninjurin1 residues 26 -37 (Ab 26 -37 ) and 1-15 (Ab 1-15 ) following standard procedures (Peptron, Inc. and Abfrontier, Inc.); these anti-Ninjurin1 antibodies were purified in each case with antigenspecific affinity chromatography. The isotype rabbit IgG purified with a Protein A column (Upstate Biotechnology) from the serum of normal rabbits was used for the in vivo blocking experiment in the EAE model. Ab 1-15 was used for immunostaining and Western blotting.
Cell Culture-Raw264.7 and MBEC4 cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS, Invitrogen) and maintained in an incubator with a humidified atmosphere of 95% O 2 and 5% CO 2 at 37°C. For the BMDM culture, bone marrow cells were isolated from femurs and tibias and cultured in RPMI 1640 medium supplemented with 10% FBS, 1% penicillin, and streptomycin for 3 days and differentiated in RPMI 1640 containing 20 ng/ml macrophage-colony stimulating factor (Peprotech) for 3 days.
Construction of the Expression Vector and Stable Raw264.7 Cells-For the construction of stable Ninjurin1-overexpressed Raw264.7 cells, full-length cDNA corresponding to mouse Nin-jurin1 (NM_013610) was inserted into the pEGFP vector (GFP-mNinj1). The constructed GFP-mNinj1 plasmid was transfected in Raw264.7 cells (Nucleofector, Amaxa) and maintained in complete DMEM with G418 (500 g/ml). After several days the surviving colonies were selected. Purity and homogeneity (Ͼ90%) were determined by means of fluorescence microscope observations and Western blotting.
Cell Adhesion and TEM Assays-To measure the adhesion activity with Ab 26 -37 treatment, Raw264.7 cells activated by TNF␣ and IFN␥ (10 ng/ml, 12 h, Peprotech) were labeled with CFSE (5 M, 5 min) and preincubated with Ab 26 -37 for 1 h corresponding to each concentration. CFSE-labeled Raw264.7 cells were added to a MBEC4 monolayer activated with TNF␣ and IFN␥ (10 ng/ml, 12 h) and incubated for 20 min. After washing three times with PBS, the adherent cells were imaged in at least six positions by microscopy, and the percentages of CFSE-labeled cells were quantified in each picture.
For the measurement of TEM activity with Ab 26 -37 treatment, MBEC4 was seeded in the upper chamber of each transwell (6.5 mm diameter, 5 m pore size; Costar), and the cells were grown to form a confluent monolayer. CFSE-labeled Raw264.7 cells preincubated with Ab 26 -37 for 1 h corresponding to each concentration were added to the upper chamber, and 10 ng/ml TNF␣ and IFN␥ were used as a chemoattractant in the lower chamber. After incubation for 12 h, the trans-well was fixed with 4% paraformaldehyde, and its upper side was cleaned with cotton and then mounted. The migrated cells were imaged at six positions by means of microscopy and analyzed by counting the percentages of CFSE-labeled cells that had transmigrated. For TEM activity of BMDMs, siNinj1, or stable Raw264.7 cells, each cells were labeled with CFSE and added to MBEC4 monolayer containing with TNF␣ and IFN␥ (10 ng/ml, 12 h) in the lower chamber as chemoattractants and incubated for 12 h. The migrated cells were quantified using microscopy.
Accelerating Rotarod Test-Motor coordination and balance were assessed by an accelerating rotarod (Columbus Institute, Columbus, OH). The rotating rod was grooved to improve the grip. The experimental procedure consisted of two phases: constant-speed training and accelerating-speed testing. For the constant-speed training, mice were placed on the rod and allowed to adjust for 1 min at a fixed 2.0 rpm. If a mouse fell before the 1-min cutoff, the process was restarted. For the accelerating-speed test, the rotarod was accelerated (acceleration ϭ 17 rpm/min), and the latency to fall from the rod was recorded. This procedure was repeated three times, and the results were analyzed.
Data Analysis and Statistics-All data are presented as the means Ϯ S.E. and expressed as relative percentages and fundamental units. Statistical significance was calculated using an unpaired two-tailed Student t test for single comparisons and an analysis of variance test for multiple comparisons. *, p Ͻ 0.05 was considered statistically significant.

RESULTS
Construction of Ninjurin1 KO Mice-To determine the role of Ninjurin1 during CNS inflammation in vivo, Ninjurin1 KO mice were generated by removing exon 1 from among a total of 4 exons for Ninjurin1 located on chromosome 13 by homologous recombination (Fig. 1A) and were backcrossed with the C57BL/6 strain for at least seven generations. Ninjurin1 heterozygous (Hetero), homozygous (KO), and congenic wildtype (WT) mice were evaluated by genotyping via a genomic polymerase chain reaction (PCR) (Fig. 1B). C, one-dimensional SDS-PAGE of various tissues (brain, liver, kidney, and spinal cord) from WT and mutant mice was performed and blotted with Ab 1-15 antibody and anti-actin antibody. Actin was used as a loading control, and the arrowhead indicates nonspecific band.
Next, we raised an antibody (Ab 1-15 ) to detect Ninjurin1 expression, which was immunized with the peptide from Met 1 to Arg 15 of the mouse Ninjurin1 as an antigen (20). Western blotting with Ab 1-15 antibody revealed the complete disappearance of the expected band in the Ninjurin1 KO mice, clearly demonstrating the well controlled Ninjurin1 expression on the protein level as well as the specificity of the Ab 1-15 antibody to the mouse Ninjurin1 (Fig. 1C). Notably, the level of Ninjurin1 expression is heterogeneous in various tissues and is gene dosage-dependent when WT and mutant mice were compared (Fig. 1C).
By analyzing the phenotypical characteristics, some of the Ninjurin1 KO mice showed developmental retardation and dysfunctions such as ataxia and hydrocephalus (data not shown), whereas the others had normal phenotypes. In this study we used Ninjurin1 KO mice with a normal appearance and a hemogram similar to that of WT mice (Table 1).

Ninjurin1 Mutant Mice Attenuate the Susceptibility of MOG 35-55 -immunized EAE in a Gene Dosage-dependent
Manner-Initially, the susceptibilities to EAE were compared between WT and Ninjurin1 KO mice for a 30-day period after (MOG)  immunization. Interestingly, the Ninjurin1 KO mice showed significantly lower EAE clinical scores from 17 to 22 days after immunization compared with the WT mice ( Fig.  2A) with a difference in their peak scores and without a difference in the incidence and onset time of the disease (Fig. 2B). Furthermore, when we compare the EAE susceptibility between the Ninjurin1 Hetero and KO mice, the difference is likely to be distinguishable despite the loss of statistical significance in terms of the clinical score for EAE (Fig. 2C). Ninjurin1 KO mice had reduced mean and peak values, but a similar onset time for EAE compared with the Hetero group (Fig. 2D). Nevertheless, the difference in the EAE severity between the Hetero and KO mice is less than that between the WT and KO mice. In Fig. 1C, the amount of Ninjurin1 expression is dependent on the dosage of the gene. These results imply that EAE pathogenesis is prohibited by the deficiency of Ninjurin1 in mice, which is closely correlated with its expression level.
Because some Ninjurin1 KO mice have developmental retardation and ataxia, we performed the rotarod test to examine the relevance between the basal motor functions and the EAE limb paralysis. The latency to fall off the rod in the Ninjurin1 KO mice was slightly reduced but showed no statistical significance to their WT counterparts (data not shown). This result indicates that the weak limb paralysis of EAE in Ninjurin1 KO mice mainly stems from the leukocyte-promoting demyelination regardless of the basal motor function with normal appearance.   (15,16,18) and up-regulated in T lymphocytes (19) to promote their recruitment toward inflammatory sites. We, therefore, compared the number of leukocytes that infiltrated into the spinal cords of WT or Ninjurin1 KO mice with EAE. Consistent with previous results, a histological analysis of the spinal cords of mice revealed that Ninjurin1 KO mice had decreased accumulation of leukocytes with CD45, a pan-leukocyte marker, F4/80, a macrophage marker, and CD3, a lymphocyte marker (Fig. 3, A  and B) and, consequently, results with a smaller lesion area and less demyelination (Fig. 3C). These results imply that Ninjurin1 KO mice protected against EAE due to the reduced amounts of leukocyte adhesion and infiltration into the injury lesions.
A recent study showed that Ninjurin1 depletion promotes cell apoptosis and senescence in a p53-dependent manner (21). Although the hemogram between WT and Ninjurin1 KO mice did not differ under normal conditions (Table 1), the inflammatory EAE induction may alter the homeostatic functions of leukocytes in Ninjurin1 KO mice. Therefore, the hemogram was compared when the clinical score after EAE induction was significantly different between each group ( Table 2). As a result, we found no significant distinction in the number and propor-tion of leukocytes circulating inside blood vessels ( Table 2). This result suggests that inflammatory EAE stimulation has no effect on leukocytes homeostasis between WT and Ninjurin1 KO mice.
Adherence of Leukocytes Is Decreased in the Ninjurin1 KO Mice with EIU-Attachment of leukocytes on the luminal side of the vessel is the initial event for TEM under inflammation (5). Retinal vasculature is a beneficial in vivo method to investigate

TABLE 2 Hemogram of WT or Ninjurin1 KO mice after EAE induction
EAE was induced in 12-week-old mice (WT; n ϭ 8, KO; n ϭ 8). Peripheral blood was collected from the abdominal vein at 14 days after immunization when clinical scores were significantly different. Cell counts were performed using an automated cell counter with veterinary parameters and reagents. Significant differences of each value between the WT and Ninjurin1 KO mice were noted when p Ͻ 0.05. WBC, white blood cells. leukocyte adhesion on the vessel network because of its twodimensional property and easy visualization with a flatmounted technique (22). Furthermore, inflammation in the retina can be easily generated by EIU, which is accompanied by acute and intense leukocyte infiltration into the vitreous in the retina within 24 h after an intraperitoneal injection of LPS at a sublethal dose (23). Using the retinal model coupled with EIU, we compared the number of leukocytes attached on the retinal vessel between Ninjurin1 KO mice and its WT counterpart. Ninjurin1 KO mice had less attachment of leukocytes on the vessel wall of the retina compared with the WT mice (Fig. 4, A  and B FEBRUARY 7, 2014 • VOLUME 289 • NUMBER 6 inflammation. To examine the blockage effects of the Ninjurin1 homophilic binding domain, a custom-made antibody, Ab 26 -37 , was generated using the peptide from Pro 26 to Asn 37 of the N-terminal ectodomain of Ninjurin1 as an antigen (Fig. 5A). The treatment with the Ab 26 -37 antibody decreased the adhesion (Fig. 5B) and TEM (Fig. 5C) of the Raw264.7 cells (murine peritoneal macrophage cells) on or across MBEC4 cell (mouse brain microvascular endothelial cells) monolayers in vitro (Fig.  5, B and C). Moreover, the intraperitoneal administration of the Ab 26 -37 antibody at 4, 6, 8, and 10 days after EAE induction significantly attenuated its clinical symptoms from 14 to 17 days after EAE immunization (Fig. 5D) with a significant difference in peak score and no difference in the onset time (Fig. 5E). Staining with anti-CD45 or anti-CD3 antibody revealed that the accumulation of leukocytes decreased in the Ab 26 -37 -in-jected KO mice compared with an IgG control group (Fig. 5,  F and G). These results mean that selective blockage corresponding to its homophilic binding domain inhibits the adhesion and TEM activity and, in turn, the accumulation of leukocytes, finally leading to protective effects against EAE progression.

Protective Role of Ninjurin1 KO Mice against EAE
Ninjurin1 Is Up-regulated in the Myeloid Cells and Inflamed Endothelial Cells in the Spinal Cord of EAE Mice-Next, Western blotting with the Ab 1-15 antibody showed Ninjurin1 upregulation in the spinal cord with EAE according to clinical symptoms (Fig. 6, A and B).
To examine the Ninjurin1-expressing cell-types in the EAE spinal cord, we first tested our Ab 1-15 antibody, which is specific to the mouse Ninjurin1 sequence. Strong and abundant Ninjuirn1-positive signals were detected in the lesion sites in which leukocytes were accumulated (Fig. 6C). These signals merged with the anti-CD45 (pan-leukocytes) antibody and the anti-F4/80 (macrophages/monocytes) antibody with a strong intensity level and high percentages and also merged with the anti-CD3 (lymphocytes) antibody with weak intensity and low proportions (Fig. 6C). Furthermore, the Ninjurin1-positive signals merged with the anti-CD31 (endothelial cells) antibody in the inflamed regions but not in non-inflamed regions (Fig. 6E). The signals of the Ab 1-15 staining largely disappeared in the Ninjurin1 KO mice tissues, although some background signals remained (Fig. 6, D and F). Thus, these results show that Nin-jurin1 is expressed in leukocytes and in inflamed endothelial cells (strong/abundant expression in myeloid-lineage cells and weak/partial expression in the lymphoid-lineage cells). These results are consistent with previous findings (16,18,19).
To confirm the specificity of the Ab 1-15 antibody, liver tissues, which are known to show high Ninjurin1 expression levels, were tested (Fig. 1C) (20). Strong Ninjurin1 signals were observed in WT tissues, although these were mostly absent without any nonspecific background in the Ninjurin1 KO samples (data not shown), supporting the specificity and availability of our Ab 1-15 antibody for tissue staining.
An IFN␥ Treatment Increases Ninjurin1 Expression in Raw264.7 or MBEC4 Cells-Because some inflammatory cytokines regulate leukocyte recruitment toward legion sites in EAE pathogenesis (24), we examined the change in Ninjurin1 expression in macrophage and endothelial cells treated with TNF␣ or/and INF␥ in vitro. In the Raw264.7 cells, INF␥ treatment increased Ninjurin1 expression in a dose-dependent manner, whereas that of TNF␣ had no effect (Fig. 7, A and B). Ninjurin1 expression was also up-regulated by INF␥ treatment in BMDMs (Fig. 7C), the purity (Ͼ90%) of which was evaluated by immunostaining with the anti-F4/80 antibody, a macrophage-specific marker (data not shown). In MBEC4 cells, INF␥ treatment increased Ninjurin1 expression (Fig. 7D). These results indicate that inflammatory stimulation with INF␥ upregulates Ninjurin1 expression in both macrophage and endothelial cells and consequently might enhance the TEM. However, its regulation might vary according to the type of cytokines that require further investigation.
Ninjurin1 Expression in the BMDMs and Raw264.7 Cells Is Directly Involved in Their TEM Activity across an MBEC4 Monolayer-Through previous and our current studies with blocking antibody or peptide, it has already been established that the Ninjurin1 homotypic adhesion property is crucial for the trafficking of leukocytes (14,18,19). However, direct evidence of whether the Ninjurin1 expression level might control the TEM activity has been lacking. Therefore, with an in vitro modified Boyden chamber assay, we compared the TEM activity levels penetrating a monolayer of MBEC4 cells with Ninjurin1 KO BMDMs and Raw264.7 cells suppressed with Ninjurin1 RNAi (siNinj1) or overexpressing the GFP-tagged mouse Ninjurin1 plasmid (GFP-mNinj1). The Ninjurin1 KO BMDMs (Fig. 8, A and B) and siNinj1 Raw264.7 cells (Fig. 8,  C and D) showed less TEM activity than the WT BMDMs and the siCont Raw264.7 cells, respectively. In contrast, stable GFP-mNinj1 Raw264.7 cells showed increased TEM activity compared with the GFP Raw264.7 cells (Fig. 8, E and  F). These results suggest that the Ninjurin1 expression level directly regulates the TEM of immune cells across the endothelial monolayer.

DISCUSSION
Herein we clarified the contributions of Ninjurin1 using gene-deficient mice (Fig. 1) and antibody-neutralized mice coupled with the EAE and EIU animal model in vivo. Ablation of Ninjurin1 in mice alleviated EAE susceptibility (Fig. 2) with reduced leukocyte infiltration (Fig. 3), and decreased adherence of leukocytes on retinal vessels was shown in EIU mice (Fig. 4). In addition, a custom-made antibody, Ab 26 -37 , specific to the homophilic binding domain of the Ninjurin1 N-terminal ectodomain, effectively attenuated EAE susceptibility (Fig. 5). Furthermore, we discovered that direct control via Ninjurin1 expression is crucial for the TEM of immune cells (Fig. 8).
For a gene depletion study, we used Ninjurin1 KO mice with a normal phenotype that exhibits similar hemograms (Table 1 and Table 2) as well as BMDMs F4/80 ϩ proportional to the WT counterpart (data not shown). However, a recent paper indicated that Ninjurin1 KO mice died prematurely (21). Consistent with this result, we also observed developmental defects in some of the Ninjurin1 KO mice. Therefore, this novel mechanism of Ninjurin1 on p53 activity and cell survival could explain these features of Ninjurin1 KO mice with severe developmental defects.
The weaker inflammatory response from the Ninjurin1 KO mice with EAE is thought to stem from the impairment of leukocyte-endothelium adhesions (Figs. 2 and 3). However, Nin-jurin1 depletion in our KO mice is not restricted to myeloid cells or endothelial cells; moreover, MOG 35-55 -induced EAE mice show chronic inflammation that lasts for a long time (25). These limitations left us to debate whether the anti-inflammatory effect seen in the Ninjurin1 KO mice was from specific events related to the leukocyte-endothelium interactions. We resolved this argument somewhat using the EIU animal model (Fig. 4), which shows acute inflammation properties that induce rapid and intense leukocyte adhesion on the vessel wall of the retina within 24 h after LPS administration (23). This short time response in EIU mice might exclude the contributions of other effectors, except for leukocyte-endothelium interactions. Therefore, our study with the EIU model supports the role of Ninjurin1 on leukocyte-endothelium adhesions under inflammation (Fig. 4).
With the neutralizing study using the Ab 26 -37 antibody (Fig.  5), we confirmed that the homotypic binding property of Nin-jurin1 is crucial for leukocyte trafficking and its targeting is beneficial for treating EAE. Indeed, Ifergan et al. (18) previously tried to block the functions of Ninjurin1 in EAE with a commercial antibody produced from immunization with the N-terminal regions of the mouse Ninjurin1. On the other hand, our custom-made Ab 26 -37 antibody was generated by immunization with a peptide corresponding to the homophilic binding domain of Ninjurin1 (Fig. 5). Thus, our antibody is thought to be an improvement for Ninjurin1 targeting strategies against its , and actin is used as a loading control. B, D, and F, the TEM activity was analyzed by a modified Boyden chamber assay. Each cell, labeled with 5 M CFSE for 5 min, was added onto the MBEC4 monolayer containing 10 ng/ml TNF␣ and INF␥ in lower chamber as chemoattractants. After 12 h, the trans-well fixed at 4% paraformaldehyde, and its upper side was cleaned using cotton, after which it was mounted. The migrated cells were visualized using microscopy. Representative images (left) are shown and a transmigration index (right, n ϭ 3) is shown relative to the WT BMDMs, siCont, or GFP Raw264.7 cells normalized to 100% in each case: **, p Ͻ 0.01; ***, p Ͻ 0.001. Scale bar ϭ 50 m.
homophilic domain with a more specific interaction in EAE compared with the previous trial.
Because Ninjurin1 is known as an adhesion molecule (15), its homotypic binding properties can aptly explain how Ninjurin1 mediates leukocyte trafficking. However, it is likely that Ninju-rin1 fulfils additional functional and regulatory complexities in leukocyte trafficking beyond its well established adhesion properties. Indeed, both leukocytes and endothelial cells undergo diverse cellular events during diapedesis (4,5). For example, dynamic morphological alternations like "podosomes" (26) or "filopodia" (27) in leukocytes and like "migratory cup" in endothelial cells (28) are found at the docking portions facing the penetrating leukocytes on the endothelial cell monolayer. Furthermore, the activity of numerous effectors including chemokines and their cognate receptors (24), cytoskeletal and vesicular regulations (29,30), extracellular matrix components (31), and intercellular signaling molecules (32) should be well orchestrated for the completion of TEM in a stage-restrictive or non-restrictive manner. It is, therefore, notable that Ninjurin1 might possibly play a role in the processes of each of the steps during diapedesis described above. In the future, whether Nin-jurin1 might have other stage-specific roles and, if it does, how it interacts with the homotypic binding activity in the leukocyte-endothelium interaction would be interesting subjects to further investigate.
Finally, because Ninjurin1 was also detected in various other cell types, including neurons, oligodendrocytes, and astrocytes in the CNS (14,16), an investigation of the coordinated interactions between Ninjurin1-expressing cells under inflammation will be helpful when seeking to understand CNS diseases. The growing evidence suggests that many diseases share inflammation as a convergent downstream signal after injury despite the divergent initial triggers (33). Therefore, our results are applicable for the treatments of inflammatory CNS diseases, including atherosclerosis (34), stroke (35), and trauma (36) as well as other diseases such as rheumatoid arthritis and various wounds.