Ninjurin 1 dodecamer peptide containing the N-terminal adhesion motif (N-NAM) exerts proangiogenic effects in HUVECs and in the postischemic brain

Nerve injury-induced protein 1 (Ninjurin 1, Ninj1) is a cell adhesion molecule responsible for cell-to-cell interactions between immune cells and endothelial cells. In our previous paper, we have shown that Ninj1 plays an important role in the infiltration of neutrophils in the postischemic brain and that the dodecamer peptide harboring the Ninj1 N-terminal adhesion motif (N-NAM, Pro26-Asn37) inhibits infiltration of neutrophils in the postischemic brain and confers robust neuroprotective and anti-inflammatory effects. In the present study, we examinedt the pro-angiogenic effect of N-NAM using human umbilical vein endothelial cells (HUVECs) and rat MCAO (middle cerebral artery occlusion) model of stroke. We found that N-NAM promotes proliferation, migration, and tube formation of HUVECs and demonstrate that the suppression of endogenous Ninj1 is responsible for the N-NAM-mediated pro-angiogenic effects. Importantly, a pull-down assay revealed a direct binding between exogenously delivered N-NAM and endogenous Ninj1 and it is N-terminal adhesion motif dependent. In addition, N-NAM activated the Ang1-Tie2 and AKT signaling pathways in HUVECs, and blocking those signaling pathways with specific inhibitors suppressed N-NAM-induced tube formation, indicating critical roles of those signaling pathways in N-NAM-induced angiogenesis. Moreover, in a rat MCAO model, intranasal administration of N-NAM beginning 4 days post-MCAO (1.5 µg daily for 3 days) augmented angiogenesis in the penumbra of the ipsilateral hemisphere of the brain and significantly enhanced total vessel lengths, vessel densities, and pro-angiogenic marker expression. These results demonstrate that the 12-amino acid Ninj1 peptide, which contains the N-terminal adhesion motif of Ninj1, confers pro-angiogenic effects and suggest that those effects might contribute to its neuroprotective effects in the postischemic brain.


N-NAM induced the proliferation of HUVECs.
To determine whether N-NAM (Fig. 1a) would induce endothelial cell proliferation, we treated HUVECs with 10, 50, or 100 nM of N-NAM or with 50 nM of scN-NAM (Fig. 1a) for 24 h, and then we counted the number of Ki67-positive cells. The numbers of Ki67-positive cells were increased by 207.1 ± 15.8% (n = 6) in the 10 nM N-NAM-treated cells compared with the PBS-treated controls, and the peak increase was detected with 50 nM N-NAM (300.0 ± 9.0%, n = 6) (Fig. 1b,c). To determine whether endogenous Ninj1 participated in the N-NAM-mediated induction of HUVEC proliferation, proliferation levels were examined after treating the cells with Ninj1 siRNA. The Ninj1 level was reduced to 22.3 ± 2.5% (n = 3) in cells transfected with Ninj1 siRNA, but it was unaffected in siCon-transfected cells (Supplementary Figure 1). Interestingly, HUVEC proliferation increased to 131.3 ± 7.4% (n = 6) in siNinj1-transfected cells and N-NAM treatment had no significant effect on that increase (125.4 ± 4.0%, n = 6) (Fig. 1e,f). In contrast, in the siCon-treated HUVECs, cell proliferation was unchanged but it increased significantly when the cells were cotreated with N-NAM (130.4 ± 3.9%, n = 6) (Fig. 1e,f). N-NAM-mediated HUVEC proliferation was confirmed by the MTT assay, which showed that cell survival was greater for N-NAM-treated cells than for scN-NAMtreated cells or PBS-treated control cells and the level increased significantly in the absence of endogenous Ninj1 (Fig. 1d,g). Together these results indicate that N-NAM induced HUVEC proliferation and that this induction required endogenous Ninj1.

N-NAM induced HUVEC migration.
To determine whether N-NAM would induce HUVEC migration, we used a wound healing assay after treating cells with N-NAM (10, 50, or 100 nM) or scN-NAM (50 nM) for 12 h. Cell motility, determined by measuring wound widths, increased with 10 nM N-NAM, peaking after treatment with 50 nM N-NAM at 341.0 ± 12.0% (n = 5) of the PBS-treated control (Fig. 2a,b). However, cell migration was not increased in the scN-NAM-treated group (Fig. 2a,b). Importantly, cell migration increased significantly to 228.0 ± 14.0% (n = 3), following siRNA-mediated Ninj1 knockdown (KD), with co-treatment with N-NAM failed to induce an additional effect (Fig. 2c,d). As no change in cell migration was detected after treating HUVECs with siCon, but co-treatment with N-NAM and siCon significantly increased cell migration to 247.0 ± 4.0% (n = 5) compared with the control (Fig. 2c,d). These results indicate that N-NAM induced HUVEC migration in an endogenous Ninj1-dependent manner.

N-NAM induced HUVEC tube formation.
To confirm the proangiogenic effects of N-NAM, we examined tube formation by HUVECs. The induction of tube-like structures harboring branches, segments, and nodes was observed after culturing HUVECs on Matrigel for 12 h. The induction of tube formation was augmented dose-dependently by N-NAM, with 50 µMN-NAM increasing tube formation to 165.8 ± 5.9% (n = 10) versus the PBS-treated control cells, whereas scN-NAM had no such effect (Supplementary Figure 2). Similarly, the mean total tube length after 12 h of treatment was also increased significantly in the presence of N-NAM (Supplementary Figure 2). As expected, N-NAM significantly induced tube formation in siCon-transfected HUVECs (Fig. 3a,b). Tube formation was also enhanced to 166.3 ± 8.4% (n = 10) in Ninj1-KD HUVECs, but N-NAMmediated enhancement of tube formation was not detected in the absence of endogenous Ninj1 (Fig. 3a,b) www.nature.com/scientificreports/ lar results were obtained when total tube lengths were measured in Ninj1-KD HUVECs (Fig. 3a,c), indicating that N-NAM induced tube formation in HUVECs in an endogenous Ninj1-dependent manner.
Proangiogenic effects of N-NAM in a reconstituted tissue model of angiogenesis. Next, we examined N-NAM-mediated angiogenesis using a mouse Matrigel Plug assay. Matrigel plugs containing N-NAM (1 µM) or scN-NAM (1 µM) were embedded under the skin in the mid-ventral region of BALB/c mice and harvested 12 days later (Fig. 4a). Massive neovascularization was detected in the N-NAM-containing plugs but not in the scN-NAM-containing plugs (Fig. 4b). When we measured the hemoglobin content in the plugs, we found significantly more hemoglobin in the N-NAM-containing plugs (148.2 ± 4.8% (n = 3) versus PBS-treated controls) and a moderate level of induction in the sc-N-NAM-containing plugs (Fig. 4c). When we counted the number of endothelial cells were counted in paraffin sections of the Matrigel plugs after staining them with anti-CD31 antibody (an endothelial marker), N-NAM-containing plugs had a significantly higher number of CD31 + cells than the control, but the scN-NAM-containing plugs did not (Fig. 4d,e). These results further confirmed the pro-angiogenic effect of N-NAM.
Interaction between N-NAM and endogenous Ninj1 and induction of proangiogenic markers in HUVECs in a Ninj1 N-terminal-dependent manner. To investigate whether N-NAM binds endogenous Ninj1, a pull-down assay was performed using biotinylated-N-NAM (bt-N-NAM). The results demonstrate the interaction between bt-N-NAM and endogenous Ninj1 in bt-N-NAM-treated HUVECs; however, that inter-  www.nature.com/scientificreports/ action was not detected when bt-scN-NAM was used (Fig. 5a). Interestingly, the interaction was also undetectable when the HUVECs were preincubated with N-terminal-specific Ninj1 antibody (anti-Ninj1 2-81 antibody). Furthermore, the inhibitory effects were dose-dependent on the anti-Ninj1 2-81 antibody (Fig. 5b). Importantly, preincubation with Ninj1 C-terminal-specific antibody (anti-Ninj1 138-152 antibody) had no effect on the binding between bt-scN-NAM and endogenous Ninj1 (Fig. 5b). In addition, in the media of N-NAM-treated HUVECs, the VEGF and MMP9 protein levels (proangiogenic markers) were significantly higher than in PBS-treated controls, however, those enhancements were not observed in scN-NAM-treated cells (Fig. 5c,d). Moreover, the induction of VEGF and MMP9 did not occur when cells were preincubated with anti-Ninj1 2-81 antibody and then treated with N-NAM (Fig. 5c,d). These results indicate that the N-terminal region of Ninj1 is critical for the binding between exogenous N-NAM and endogenous Ninj1 and the proangiogenic effects of N-NAM.
The Ang1-Tie2 and AKT signaling pathways were involved in the proangiogenic effects of N-NAM in HUVECs. Expression of angiopoietin-1, angiopoietin-2, and tie receptors has been reported to play important role in angiogenesis in the ischemic brain 23 and in particular, angiopoietin-1 inhibits endothelial cell apoptosis via the AKT/survivin pathway 24,25 . Western blot analyses showed that the Ang1 level increased significantly, and the Ang2 level decreased gradually and significantly in N-NAM (50 nM)-treated HUVECs ( Fig. 6a,b). In addition, phospho-AKT (Ser473) and phospho-eNOS (Ser1177) levels were also significantly increased after N-NAM (50 nM) treatment (Fig. 6c,d). To further confirm the importance of the above mentioned signaling in the proangiogenic effect of N-NAM, tube formation was examined after pretreating HUVECs with sTie2-Fc, the soluble extracellular domain of the Tie2 receptor, or with wortmannin. N-NAM-mediated tube formation was suppressed in a dose-dependent manner (Supplementary Figure 3) and decreased to almost basal levels by 0.5 µg/ml sTie2-Fc and 10 nM wortmannin, a PI3K inhibitor (Fig. 6e,f). Taken together, these results indicate that the Ang1-Tie2 and AKT signaling pathways play important roles in N-NAM-mediated tube formation in HUVECs.
Proangiogenic effects of N-NAM in the postischemic brain. Next, we investigated proangiogenic effects of N-NAM in the postischemic brain using a rat model of cerebral ischemia. In our previous report, we  . 7b). When blood vessel densities were measured in the cortical penumbra of the ipsilateral hemispheres (Fig. 7b, asterisk) using an anti-RECA-1 antibody (a marker of endothelial cells), RECA-1-positive vessels were detected in the MCAO + PBS controls 7 days after MCAO (Fig. 7c). Importantly, vessel densities was significantly enhanced in the MCAO + NAM animals (259.1 ± 24.4% (n = 5) versus MCAO + PBS controls) but not in the MCAO + scN-NAM animals (Fig. 7c,d).
Similarly, total vessel length was also markedly greater in the MCAO + N-NAM group (324.7 ± 42.4% (n = 5) of the MCAO + PBS controls) but not in the MCAO + scN-NAM animals (Fig. 7c,e). Together, these results demonstrate a robust proangiogenic effect of N-NAM in the postischemic brain.

Attenuated vessel leakage and upregulation of angiogenesis marker induction by N-NAM in postischemic brains.
To determine whether N-NAM induced functional blood vessel formation, we tested the permeability of the vessels using staining after administrating IgG or fluorescein isothiocyanate (FITC)dextran. IgG or FITC-dextran was administered 1 h prior to sacrifice at 7 days post-MCAO (Fig. 7a). When we administered IgG and measured the intraparenchymal leakage area in the cortical penumbra of the ipsilateral hemisphere (asterisk in Fig. 8a), it was reduced in MCAO + N-NAM group to 75.4 ± 4.2% (n = 7) of the total area in MCAO + PBS control animals (Fig. 8b). In addition, after injecting FITC-dextran, the percent-leakagearea in the MCAO + N-NAM group was significantly decreased, 48.0 ± 8.5% (n = 12) of that in the MCAO + PBS controls (Fig. 8c), indicating that leakage from the new vessel was attenuated by N-NAM. In addition, the levels of Ang1 and Ang2 were significantly up-or down-regulated, respectively, in the cortical penumbra of the ipsilateral hemispheres in the MCAO + N-NAM group 7 days post-MCAO, but not in the MCAO + scN-NAM group (Fig. 8d,e). These results confirmed the proangiogenic potency of N-NAM in the postischemic brain. www.nature.com/scientificreports/

Discussion
In a previous study, we reported that Ninj1 was up-regulated during the acute phase after MCAO 22 and that intranasally administered N-NAM had robust neuroprotective and anti-inflammatory effects in the same rat MCAO model 26 . In the present study, we have shown that N-NAM has proangiogenic effects on HUVECs and in the postischemic brain. HUVEC proliferation, migration, and tube formation were promoted by N-NAM, and those effects were accompanied by the up-or down-regulation of Ang1 and Ang2, respectively, and the induction of proangiogenic markers such as VEGF and MMP9 (Fig. 8f). Moreover, the intranasal administration of N-NAM as late as 4 days post-MCAO conferred a robust proangiogenic effect, as evidenced by increased vessel formation and the expression of proangiogenic markers. Although a few previous papers have reported the importance of the N-terminal Ninj1 peptide under various pathological conditions, this is the first report to show that it plays a critical role in angiogenesis. Furthermore, we demonstrated that this effect is due to inhibition of the suppressive effect that endogenous Ninj1 has on angiogenesis. The proangiogenic effect found in our siRNA-mediated Ninj1 KD experiments indicates that endogenous Ninj1 has a negative effect on angiogenesis, which is in keeping with a previous report that angiogenesis was enhanced after functionally blocking Ninj1 using a neutralizing antibody or siRNA in diabetic mice 15 . In addition, increase in the function and survival of endothelial cells following treatment with a Ninj1 antibody was reported in a similar diabetic animal model 20 . In the present study, we found that endogenous Ninj1 KD enhanced angiogenesis and that in that situation, N-NAM exerted no angiogenic effect, which suggested that the proangiogenic effects of N-NAM resulted from the suppression of endogenous Ninj1. In the postischemic brain, penumbral endothelial cells begin to proliferate after 12-24 h and active angiogenesis is observed 3-4 days Figure 6. The Ang1-Tie2 and PI3K/AKT signaling pathways were involved in the N-NAM-mediated proangiogenic effect. (a-d) HUVECs were incubated with N-NAM (50 nM) for 15, 30, 60, or 120 min, and Ang1 or Ang2 levels (a,b) and total and phosphorylated AKT and eNOS levels (c,d) were assessed by immunoblotting. Representative images are presented (b,d) and results are presented as the means ± SEMs (n = 3 for Ang1, Ang2, and eNOS and n = 5 for AKT) (a,c). (e-f) HUVECs were treated with N-NAM (50 nM) or scN-NAM (50 nM) for 12 h with or without sTie2-Fc (0.5 µg/ml) or wortmannin (10 nM) pretreatment and tube formation was examined. Representative images are presented and the number of tubes and total tube lengths are presented as the means ± SEMs (n = 10). Scale bar in e, 500 µm, **p < 0.01, *p < 0.05 versus the PBS-treated controls, # p < 0.05 versus 50 nM N-NAM-treated cells. www.nature.com/scientificreports/ post-stroke 4,25 . It is interesting to note here that in our previous reports, we showed a dual surge of Ninj1 expression in the postischemic brain, one at ~ 12 h and another 4-6 days post-MCAO 22,26 . The neuroprotective effect of N-NAM administered immediately after MCAO, shown in our previous study 26 was attributed to blocking Ninj1 from ~ 12 h, which resulted in the inhibition of neutrophil infiltration during the acute phase. On the other hand, proangiogenic effect of delayed N-NAM administration, beginning 4 days post-MCAO in the present study, might be derived from functional blocking of the 2nd surge of Ninj1 in the postischemic brain, which occurs from 4-6 days post-MCAO. Because that 2nd surge of Ninj1 decreased gradually but was maintained until 10 days post-MCAO in both the cortical and striatal penumbrae 22 , N-NAM might exert other effects, in addition to proangiogenic effects, related to the repair and reconstruction of the postischemic brain. Further studies are required on this topic. In our previous reports, we showed that the neuroprotective and anti-inflammatory effects of N-NAM in the postischemic brain were due to the inhibition of interactions between neutrophils and endothelial cells 26 . We observed a direct interaction between N-NAM and endogenous Ninj1 in this study, so we suggested that functional blocking of the homophilic binding of Ninj1 via competitive binding underlies the proangiogenic effects of N-NAM. However, we cannot exclude the possibility that functional blocking of the heterophilic interactions between Ninj1 and another molecule (not yet identified) also play a role or that N-NAM could act as a novel ligand for another receptor. Furthermore, because the angiogenesis-related effects of Ninj1 have been reported in cells other than endothelial cells, for example in pericytes 27,28 and macrophages 11 , it is possible that N-NAM also affects those cells. Although we failed to find any functional motif or proteins harboring sequences homologous to N-NAM in a BLAST search, additional studies are required to investigate the heterophilic interactions of Ninj1 by N-NAM and possible novel roles of N-NAM.
In this study, we have demonstrated that the Ang1-Tie2 and AKT signaling pathways are critically involved in the proangiogenic effects of N-NAM. Ninj1-KD-mediated Ang1-Tie2 activation and its proangiogenic effects in endothelial cells were previously reported in a diabetic mouse model 15 . Knockdown of the Ninj1 with siRNA increased Ang1 expression and decreased the endogenous Ang1 antagonist, Ang2, in cavernous endothelial cells in diabetic mouse which resulted in increased endothelial cell proliferation and decreased apoptosis 15 . Interestingly, similar modulation of Ang1-Tie2 signaling by Ninj1 has also been documented in pericytes 11 . During early ocular development, Ninj1-overexpressing macrophages decreased Ang1 and increased Ang2 in pericytes, resulting in hyoid vessel endothelial cell apoptosis 11 . Furthermore, in hind limb ischemia animal model, Ninj1 was induced in pericytes and modulated Ang1/Ang2 expression 27,28 . Pericytes play a critical role during the www.nature.com/scientificreports/ initial endothelial cell sprouting, guiding newly formed vessel 29 . Later, during the vascular maturation, pericytes are strongly associated with immature vessels and induces vascular stabilization and maturation 30,31 . Ninj1 in pericyte is involved in the vessel maturation through the association between pericytes and endothelial cells, leading to blood flow recovery from ischemia 28 . In the present study, we reported that N-NAM up-and downregulates Ang1 and Ang2, respectively, in HUVECs and promotes vascular permeability in the postishcemic brain. Therefore, the contribution and modulation of Ang1-Tie2 signaling pathway in the presence of both endothelial cells and pericytes might provide more information and needs further study. However, although we observed direct binding between N-NAM and endogenous Ninj1, the link between such binding and the activation of AKT signaling was not established. Given that the cytoplasmic domain of Ninj1 is too small to directly activate downstream signaling molecules, it has been speculated that Ninj1 might be involved in the clustering Figure 8. Functional blood vessel formation and proangiogenic marker induction by N-NAM in the postischemic brain. N-NAM (5 µg) or scN-NAM (5 µg) was administered intranasally three times daily at 4, 5 and 6 days after 60 min of MCAO, and IgG or fluorescein isothiocyanate (FITC)-dextran (60 mg/kg) was injected intravenously (i.v.) 1 h prior to sacrifice 7 days post-MCAO (Fig. 7a). Coronal brain sections were prepared and stained using biotynylated rat anti-IgG antibody for IgG staining (b) or FITC-dextran images acquired with confocal microscopy (c). IgG-positive area or FITC intensity were measured using Scion www.nature.com/scientificreports/ of membrane proteins such as integrins and that those complexes might mediate signaling pathways in the cytoplasm 32,33 . Further studies are required to elucidate the downstream signaling of Ninj1 by which N-NAM regulates angiogenesis. Although ischemic stroke is a major cause of death and disability, few therapeutic options are available, and only a small proportion of stroke patients receive acute reperfusion therapies. Nevertheless, as the survival rate has increased, long-term treatments have become necessary to address the sequelae of stroke. Therefore, researchers have begun to focus on recovery after stroke, including a search for drugs and treatments that promote functional recovery. Given the importance of angiogenesis during recovery, N-NAM might be a useful starting point for the development of therapeutics that target neuro-recovery from postischemic damage. Attenuation of delayed inflammation by N-NAM may also contribute to promote functional recovery following cerebral ischemia since numerous reports, including ours, have demonstrated the anti-inflammatory effects of this peptide [17][18][19][20]22 . It is important to note here that since in the process of reparative angiogenesis, the interplay between the immune cells and growing blood vessels is important, pro-angiogenic effect is associated with antiinflammatory effect if the postischemic brain. Moreover, because it has been shown that Ninj1 is a multifunctional protein involved not only in vessel formation but also in nerve regeneration and cell-cell interactions under various pathological conditions 10,[34][35][36][37] , N-NAM might affect repair process in the postischemic brain, such as neurite outgrowth or synaptogenesis.

Conclusion
The present study showed the 12-amino acid Ninj1 peptide, which contains the N-terminal adhesion motif of Ninj1, confers pro-angiogenic effects and suggest that those effects might contribute to its neuroprotective effects in the postischemic brain. Thus inhibition of Ninj1 using N-NAM might be a promising approach to ameliorate delayed damage and enhance reparative responses. Additional studies are required to access the effect of N-NAM modulating the cross-talk among those multifunction and to clarify the N-NAM-induced signaling pathways in vivo wherein the interactions among endothelial, pericytes, and immune cells are allowed. . MCAO was carried out as described previously 26 . Briefly, 8-week-old male Sprague-Dawley rats (250-300 g) were anesthetized with 5% isoflurane in 30% oxygen/70% nitrous oxide and maintained using 0.5% isoflurane in the same gas mixture during surgery. Occlusion of the right middle carotid artery was induced for 1 h by advancing a nylon suture (4-0; AILEE, Busan, Korea) with a heat-induced bulb at its tip (~ 0.3 mm in diameter) along the internal carotid artery for 20-22 mm from its bifurcation with the external carotid artery. This procedure was followed by reperfusion for up to 7 days. A thermoregulated heating pad and a heating lamp were used to maintain a rectal temperature of 37 ± 0.5 °C during surgery. Animals were randomly allocated to the sham, MCAO + PBS (phosphate-buffered saline), MCAO + N-NAM, or MCAO + scN-NAM group. Animals in the sham group underwent an identical procedure, but the MCA was not occluded.
Wound healing assay. HUVECs were seeded in 24-well plates and grown to a 90% confluency. The wounds were made across the center of wells using a yellow tip. After washing with M199, cells were treated with or without peptides in the same medium (M199 containing 2% FBS). Cell migration was assayed using live cell imaging microscope (JuLi Stage; NanoEnTek, Seoul, South Korea) for 12 h. Cell migration into wounds were analyzed using the ImageJ software (https ://image j.nih.gov/ij/downl oad.html) MRI Wound Healing Tool (National Institute of Health (NIH), Bethesda, MD), and the rate of cell migrations were calculated using the following equation: ((area at 0 h − area at 12 h)/area at 0 h) × 100.
HUVECs (5 × 10 4 ) were seeded on Matrigel with or without peptides and with or without co-treating antibodies or sTie2 for 12 h. Tube formation was photographed and calculated by measuring tube numbers and lengths in four different fields per well using a real-time cell history recorder (JuLi stage; NanoEnTek, Seoul, South Korea). Data analysis was performed using the Angiogenesis Tools in ImageJ (https ://image j.nih.gov/ij/downl oad.html) (NIH, Bethesda, MD).
Matrigel plug assay. Matrigel plug assay was performed as previously described 39 . In brief, Matrigel