Neuroprotective Effects of Endogenous Secretory Receptor for Advanced Glycation End-products in Brain Ischemia

The receptor for advanced glycation end-products (RAGE) is expressed on human brain endothelial cells (HBEC) and is implicated in neuronal cell death after ischemia. We report that endogenous secretory RAGE (esRAGE) is a splicing variant form of RAGE that functions as a decoy against ischemia-induced neuronal cell damage. This study demonstrated that esRAGE was associated with heparan sulphate proteoglycans on HBEC. The parabiotic experiments between human esRAGE overexpressing transgenic (Tg), RAGE knockout (KO), and wild-type (WT) mice revealed a significant neuronal cell damage in the CA1 region of the WT side of parabiotic WT→WT mice, but not of Tg→WT mice, 7 days after bilateral common carotid artery occlusion. Human esRAGE was detected around the CA1 neurons in the WT side of the parabiotic Tg→WT pair, but not in the KO side of the Tg→KO pair. To elucidate the dynamic transfer of esRAGE into the brain, we used the blood-brain barrier (BBB) system (PharmaCo-Cell) with or without RAGE knockdown in endothelial cells. A RAGE-dependent transfer of esRAGE was demonstrated from the vascular to the brain side. These findings suggested that esRAGE is associated with heparan sulphate proteoglycans and is transferred into the brain via BBB to exert its neuroprotective effects in ischemia.

There are many hypotheses about the molecular mechanisms underlying the delayed neuronal cell death in ischemia, including mitochondrial dysfunction, oxidative stress, nitrosative stress and the calcium-glutamine hypothesis [12][13][14][15].
The receptor for advanced glycation end-products (RAGE) is a pattern-recognition receptor that is involved in ischemic brain injury [16][17][18]. It is also well known that RAGE is implicated in the development of diabetic vascular complications, atherosclerosis, neurodegenerative diseases, and inflammatory diseases [19][20][21][22][23][24]. In the brain, RAGE expression is reported in microglial and endothelial cells [12,25]. Kamide et al. demonstrated that the expression of RAGE could rise in the cerebrovascular endothelium and hippocampal neurons after transient brain ischemia by bilateral common carotid artery occlusion (BCCAO) [16]. We have previously identified and reported endogenous soluble RAGE (esRAGE), which is a truncated variant of RAGE isoforms generated by alternative splicing of the pre-mRNA transcripts and detectable in human circulation [26]. Recombinant proteins of the soluble form of RAGE were generated with gene technology and used in animal experiments [27,28]. Soluble forms of RAGE proteins are known to be generated in vivo by cleavage of membrane-bound full-length RAGE as well as esRAGE by metalloproteinsases [22]. The soluble RAGE proteins including esRAGE and the cleaved forms of RAGE are known to act as a decoy receptor and to ameliorate pathologies in experimental animal models [29]. For example, the application of soluble RAGE significantly reduced the infarct size and showed neuroprotective effects on cortical neurons in mouse MCAO models [30,31]. In addition, the expression of inflammatory molecules such as tumor necrosis factor α or inducible nitric oxide synthase was attenuated at the mRNA level in the hippocampal CA1 region of esRAGE transgenic (Tg) mice and in RAGE knockout (KO) mice, when compared with wild-type (WT) mice [16]. The neuroprotective role of esRAGE was observed. Regarding the esRAGE Tg mice used in the report, the mice were created to produce human esRAGE in the liver using the albumin promoter [28]. However, human esRAGE was also detected in the brain of this mouse model. Therefore, to evaluate the strict effect of only bloodstream esRAGE, we employed the parabiotic strategy [32], which will demonstrate a stable level of blood esRAGE, as a pilot study of intravenous injection of esRAGE protein.
In the present study, we found that blood esRAGE significantly inhibited transient brain ischemia-induced neuronal cell damage and apoptosis in the hippocampal CA1 subfield in the mouse BCCAO model. Blood esRAGE was transferred into the brain through the bloodbrain barrier (BBB) by RAGE on endothelial cells, potentially contributing to the neuroprotective effects in ischemia. Thus, endothelial RAGE has two different roles in ischemia, namely as an inducer of vascular injury and neuronal damage and as a transporter of esRAGE, a neuroprotector, to the brain.

Animals
All animal experiments were conducted according to the Guidelines for Proper Conduct of Animal Experiments by the Science Council of Japan (1 June 2006) (www.scj.go.jp/ja/info/kohyo/pdf/kohyo-20-k16-2e.pdf). All animal procedures were performed in accordance with and approved by the Animal Care and Use Committee of Kanazawa University (AP-132929). The report was written in compliance with the ARRIVE guidelines (Animal Research: Reporting in Vivo Experiments) (www.nc3rs.org.uk/arrive-guidelines). The RAGE KO (Ager -/-) and WT (C57BL/6J) mice were produced by crossbreeding heterozygous mutant mice [20]. The esRAGE Tg (C57BL/6J) mice were generated as described previously [28]. Female mice of each cohort (age, 8-12 weeks; weight, 20-30 g) were used in the experiments. General anesthesia was induced with 2.0% sevoflurane and was maintained with 0.5% sevoflurane using a face mask [16]. To introduce parabiosis, mirror image incisions were made through the skin at the left and right flanks. The skin of the adjacent parabiont was sutured together. Before surgery and 14 d postoperatively, venous blood samples (100 µl) were collected from the tail vein. esRAGE levels in sera, culture media and tissue lysates were measured by human esRAGE enzyme-linked immunosorbent assay (ELISA) kit (B-Bridge) [28]. Cross-circulation was confirmed in a subset of parabiotic pairs by measuring the esRAGE blood concentration from one partner in the blood of the other partner [32]. Serum mouse soluble RAGE (sRAGE) was assayed with a mouse RAGE Quantikine ELISA kit (MRG00, R&D Systems, Inc., MN, USA). Protein concentrations of tissue lysate were measured using a BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA).

Surgical procedures (BCCAO)
Seven days after parabiosis, brain ischemia was induced by BCCAO for 20 min with the use of microvascular clips as described previously [16]. Laser-Doppler flowmetry was used to measure the cerebral cortical microperfusion (3 mm lateral to bregma). In our experimental model, mice that demonstrated < 15% of baseline control microperfusion during the first minute of occlusion were used for subsequent experiments. Rectal temperature was maintained at 36.5-37.5 °C by using a heat lamp and a blanket [16].

Neuronal cell quantification
Hippocampal cell death was assessed according to a standard protocol described previously [36,37]. Brain sections were stained with cresyl violet (Nissl) staining and hematoxylin-eosin staining. Cell counting was performed in five sections per hippocampus starting at 1.4 mm posterior to bregma encompassing the core of the lesioned area (rostrocaudal levels, −1.4, −1.5, −1.6, −1.7, and −1.8 mm from bregma). Injury to the CA1 subfield was evaluated in the medial and intermediate portion of the CA1 under a light microscope at a 100× magnification by manually counting [38]. We defined degenerating neurons as cells with pyknotic, shrunken nuclei and reduced size. Surviving neurons (Nissl-positive and no sign of neurodegeneration) were counted in five fields and expressed as the average number. For the detection of neuronal apoptosis, we used the Apop-Tag Fluorescein in situ Apoptosis Detection Kit (Chemicon International) according to the manufacturer's instructions [39]. Apoptotic neurons (caspase 3-positive neurons) were counted in five different sections [40]. Cell counting was done with blind assessments by two evaluators.

Blood-brain barrier assay
For this assay, we used a commercially available BBB kit (MBT-24H, PharmaCo-Cell Co., Nagasaki, Japan), which was composed of monkey brain capillary endothelial cells, rat brain pericytes, and rat brain astrocytes. The in vitro BBB models were established within 4 d of seeding the cells according to the manufacturer's instructions [41,42]. Trans-endothelial electrical resistance (TEER), which primarily reflects the flux of sodium ions through the cell layers in the culture conditions, was measured with an epithelial-volt-ohm meter and an Endohm-6 chamber electrode (World Precision Instruments, Sarasota, FL, USA) [42,43]. The TEERs of the coated, cell-free filters were subtracted from the measured TEER values of the models and shown as Ω cm 2 . The apparent permeability constants (Papp) between the luminal and abluminal chambers were calculated based on the distribution ratios [42,44]. The expression of endothelial RAGE was silenced using RAGE small hairpin RNA (shRNA) and the pSilencer 3.0-H1 vector (Ambion, Austin, TX) with Lipofectamine 2000 (Thermo Fisher Scientific) [42,45]. The RAGE-knockdown efficiency was >90% with our shRNA system (data not shown).

Statistical analysis
P values were calculated using two-tailed Student's t-test for pair wise comparisons, and one-way analysis of variance (ANOVA), followed by Bonferroni's test for multiple comparisons, unless otherwise stated. A P value of < 0.05 was considered statistically significant. These analyses were carried out with the use of Ekuseru-Toukei 2015 (Social Survey Research Information Co., Ltd., Japan).

Blood esRAGE protects against neuronal cell damage
To elucidate the functional role of blood circulating esRAGE, we employed mouse parabiosis approaches by using human esRAGE Tg (human esRAGE overexpressed in blood), RAGE KO, and WT mice (Fig.  1A): (1) WT mice parabiosed with WT mice (WT→WT); (2) esRAGE Tg mice parabiosed with WT mice (Tg→WT); (3) esRAGE Tg mice parabiosed with RAGE-KO mice (Tg→KO). By using these three parabiosis pairs, brain ischemia was induced on either side of the pair (gray-colored right one) to investigate the neuronal cell damage and survival 7 d after BCCAO (Fig. 1B). Laser-Doppler flowmetry data showed no differences in ischemic conditions during and after BCCAO among the three groups (Fig. 1C). Severe ischemic brain damages were not observed after BCCAO (Fig. 1D). The transfer of human esRAGE protein from the esRAGE-Tg mice to the blood of WT mice in the Tg→WT pair was confirmed prior to the experiment. The results indicated that the level of human esRAGE in WT mice was the half of that of esRAGE Tg mice in the Tg→WT pair (Fig. 1E). Subsequently, we investigated the delayed neuronal cell death and cell survival by using a mouse BCCAO model. Both the hematoxylin-eosin and Nissl stainings demonstrated severe neurodegeneration in the CA1 neurons in the WT side of the WT→WT pair, but not in the WT side of the Tg-WT pair, 7 d after BCCAO (Fig.  2A). Quantitative analyses also showed a significant decrease in the number of delayed neuronal cell death in the CA1 region of the WT side of the Tg→WT pair, when compared with that of the WT→WT pair ( Fig. 2A). The terminal deoxynucleotidyl transferase dUTP nick end labeling staining was performed to evaluate apoptosis and demonstrated a significantly decreased number of apoptotic cells in the CA1 region of the WT side of the Tg→WT pair, in comparison with the WT→WT pair (Fig. 2B). These results suggested neuroprotective effects of esRAGE. However, we could not find any further synergistic or addictive neuroprotective effects of esRAGE under RAGE-deleted conditions; e.g. the KO side of the Tg→KO parabiosis pair ( Fig. 2A and B). We speculate that the neuroprotective effects observed in the KO side of the Tg→KO parabiosis pair could be caused by only RAGE-deletion.

Endothelial RAGE transfers blood circulating esRAGE to the brain
To address how endothelial esRAGE can protect against neuronal cell death in the ischemic brain, we first checked the in vivo presence of the human esRAGE protein by immunofluorescent examinations by using mouse parabiosis models. Therefore, we employed WT→WT, WT→Tg, Tg→WT, and Tg→KO parabiosis pairs and examined the underlined animals in each pair after 7 days of the parabiosis. Based on the results, human esRAGE signals were detected on the brain endothelial surface in the WT→Tg, Tg→WT, and Tg→KO pairs, but not in the WT→WT pair (Fig. 3A). Surprisingly, the human esRAGE protein was also observed in the brain parenchyma around the hippocampal CA1 neurons in the WT→Tg and Tg→WT parabiosis pairs (Fig. 3A). A costaining experiment also showed that the esRAGE signals were co-localized with NeuN, a neuronal marker, but not GFAP, an astrocytic maker (Fig. 3B). Based on these results, we suggested that the blood circulating esRAGE and endothelium-associated esRAGE proteins were transferred into the brain parenchyma through the BBB. No positive signals of human esRAGE protein were observed in the KO side of Tg→KO or WT→KO pair which was a negative control (Fig. 3A). In addition, a quantitative evaluation with ELISA demonstrated that human esRAGE was detectable and its concentration was significantly lower in the KO side of the brain parenchyma in Tg→KO when compared with the WT side of Tg→WT pair (Fig. 4A). These findings also support our hypothesis that esRAGE transfers into brain through the BBB. The transfer of esRAGE might depend on RAGE, even though some detectable level of human esRAGE was observed in the KO side of Tg→KO pair (Fig. 4A), which may be derived from the contamination of vessel esRAGE. This is based on the fact that serum levels of human esRAGE were 2.0-3.0 µg/ml in the Tg side of Tg→WT and Tg→KO pairs and ~1.0 µg/ml in the WT or KO side of Tg→WT or Tg→KO pairs (Fig. 4B). Mouse sRAGE levels were found to be ~8 ng/ml in WT mice, and at under detectable levels in the KO mice (Fig.  4C), indicating that mouse sRAGE is more than two orders of magnitude lower than human esRAGE levels in the serum of Tg parabiosed mice.

Association of esRAGE with heparan sulfate proteoglycan (HSPG) of vascular endothelial cells
We further investigated whether there was an association between esRAGE and the cell surface of human HBEC in cultures by using fluorescence microscopy. Faint signals of esRAGE were detected in the endothelial cells, which became much more enhanced with the addition of recombinant esRAGE into the cell culture media (Fig.  5A). On the other hand, treatment with heparin (0.1 IU/ml) or heparitinase (1 mU/ml) prevented the accumulation of esRAGE with or without the addition of esRAGE protein in HBEC (Fig. 5A). We obtained the same data in human HMVEC (data not shown). Colocalization of esRAGE and heparin sulfate signals were observed (Fig. 5B). Subsequently, we examined the release of esRAGE protein into the cell culture media of human HBEC with or without treatment with heparin and heparitinase. Based on the results, treatment with heparin (0.1 IU/ml) significantly increased the concentration of esRAGE protein in the cell culture media of HBEC (Fig.  5C). Pretreatment of the HBEC culture media with recombinant esRAGE (1 µg/ml) for 1 h increased the basal level of esRAGE. Similarly, heparin (0.1 IU/ml) or heparitinase (1 mU/ml) also significantly upregulated the esRAGE concentration (Fig. 5C). Furthermore, intravenous heparin injection significantly increased the serum concentration of human esRAGE in vivo in the esRAGE Tg mice (Fig. 5D). These results indicate that esRAGE is associated with HSPG of the endothelial cells.
In order to confirm our hypothesis of the esRAGE transfer through the BBB, we used an in vitro BBB model composed of brain endothelial cells, pericytes, and astrocytes, whose barrier function was intact with higher TEER levels (Fig. 6A). When 20 μg/ml of human esRAGE protein was added to the vessel side, the esRAGE concentrations of the brain side increased significantly in a time-dependent manner without endothelial RAGE knockdown (control) (Fig. 6B). In contrast, no such increase was observed in the RAGE knockdown model (Fig. 6B), indicating that the transfer of human esRAGE protein occurred through the endothelial RAGE.

DISCUSSION
Previous reports demonstrated a pivotal role of neuronal and microglia RAGE in ischemia-induced neuronal death and inflammation in mouse models of distal permanent middle cerebral artery occlusion (MCAO), which cause focal ischemic stroke [46,47]. On the other hand, in the BCCAO model, which is known to induce global ischemia and subsequent delayed neuronal cell death, the induction of RAGE expression in the hippocampal vascular endothelial cells preceded that neurons and glia, indicating that endothelial RAGE could contribute to delayed neuronal death by enhancing vascular injuries, and potentially leading to microcirculatory disturbances [16]. For the therapeutic strategies against stroke and neuronal cell death, both esRAGE and enzymatically cleaved soluble RAGE were applied to ischemic brain injury models of MCAO and BCCAO. Administration of soluble RAGE has been reported to reduce the infarct size of the brain and microglia/macrophage infiltrations in a mouse MCAO model [48]. Using esRAGE-Tg mice, delayed neuronal cell death in the hippocampus was attenuated in the BCCAO model [16]. However, the precise mechanisms of the decoy-type receptor of soluble RAGE and esRAGE in the protection against ischemiainduced neural damage are still unclear. In this study, we first observed that esRAGE was associated with HSPG of the vascular endothelial cell surface in which it accumulated both in vitro and in vivo (Fig. 1), which is consistent with previous findings [49]. We also found that excessive circulating esRAGE was protective against delayed neuronal damage in ischemia in our parabiosis models (Fig. 2). Therefore, this study demonstrated for the first time that esRAGE was transferred into the brain from blood through the BBB and localized around the hippocampal CA1 neurons (Fig. 3). These results suggested a functional role of the decoy type receptor against RAGE ligands including damageassociated molecular patterns released in brain ischemia. The human esRAGE is approximately 50 kDa (of a large molecular size) and is usually considered to be untransportable through the BBB. However, using Tg→WT and Tg→KO parabiosis mice in addition to the in vitro BBB model with or without RAGE knockdown, we found that endothelial RAGE could function as a transporter of esRAGE protein from blood to brain (Figs. [3][4][5][6]. We speculated that RAGE may form an oligomer complex with esRAGE on endothelial cells, which has been previously described whereby RAGE itself forms oligomer with its extracellular domain and transcytoses esRAGE from the luminal to the abluminal side [50]. These findings suggested that both circulating and endothelium-associated and accumulated esRAGE could protect against vascular damages in ischemia; similarly, the esRAGE transferred into the brain could also function as a neuroprotective factor in ischemic brain injuries. Further studies are required to explore how esRAGE can prevent neuronal damage in brain ischemia. It is conceivable that endothelial RAGE is a double-edged sword in brain ischemia; intracellular RAGE signal transduction can induce vascular inflammation and result in the derangement of microvascular circulation, while RAGE can transport the neuroprotective esRAGE into the brain and prevent neuronal cell death. Thus, a condition enriched with esRAGE could be neuroprotective in ischemia.
In conclusion, we confirmed that esRAGE could prevent neuronal cell death in the BCCAO ischemic mouse model and could be transferred into the brain through the BBB, where it accumulates around the neurons and functions as a neuroprotective factor. Accordingly, we suggest that a supplementation of esRAGE in ischemic brains will be a potential therapeutic tool for the prevention of neuronal cell death.