FTY720

Activation of Glucagon-Like Peptide-1 Receptor Promotes Neuroprotection in Experimental Autoimmune Encephalomyelitis by Reducing Neuroinflammatory Responses

Abstract The signaling axis of glucagon-like peptide-1 (GLP-1)/GLP-1 receptor (GLP-1R) has been an important component in overcoming diabetes, and recent reports have uncovered novel beneficial roles of this signaling axis in cen- tral nervous system (CNS) disorders, such as Alzheimer’s disease, Parkinson’s disease, and cerebral ischemia, accelerat- ing processes for exendin-4 repositioning. Here, we studied whether multiple sclerosis (MS) could be a complement to the CNS disorders that are associated with the GLP-1/GLP-1R signaling axis. Both components of the signaling axis, GLP-1 and GLP-1R proteins, are expressed in neurons, astro- cytes, and microglia in the spinal cord of normal mice. In particular, they are abundant in Iba1-positive microglia. Upon challenge by experimental autoimmune encephalomy- elitis (EAE), an animal model of MS, the mRNA expression of both GLP-1 and GLP-1R was markedly downregulated in EAE-symptomatic spinal cords, indicating attenuated activity of GLP-1/GLP-1R signaling in EAE. Such a downregulation obviously occurred in LPS-stimulated rat primary microglia, a main cell type to express both GLP-1 and GLP-1R, further indicating attenuated activity of GLP-1/GLP-1R signaling in activated microglia. To investigate whether increased activity of GLP-1R has a therapeutic benefit, exendin-4 (5 μg/kg, i.p.), a GLP-1R agonist, was administered daily to EAE- symptom- atic mice. Exendin-4 administration to symptomatic EAE mice significantly improved the clinical signs of the disease, along with the reversal of histopathological sequelae such as cell accumulation, demyelination, astrogliosis, microglial activation, and morphological transformation of activated microglia in the injured spinal cord. Such an im- provement by exendin-4 was comparable to that by FTY720 (3 mg/kg, i.p.), a drug for MS. The neuroprotective effects of exendin-4 against EAE were also associated with decreased mRNA expression of proinflammatory cytokines, such as in- terleukin (IL)-17, IL-1β, IL-6, and tumor necrosis factor (TNF)-α, all of which are usually upregulated in injured sites of the EAE spinal cord. Interestingly, exendin-4 exposure sim- ilarly reduced mRNA levels of IL-1β and TNF-α in LPS- stimulated microglia. Furthermore, exendin-4 administration significantly attenuated activation of NF-κB signaling in EAE spinal cord and LPS-stimulated microglia. Collectively, the current study demonstrates the therapeutic potential of exendin-4 for MS by reducing immune responses in the CNS, highlighting the importance of the GLP-1/GLP-1R signaling axis in the development of a novel therapeutic strategy for MS.

Keywords : Experimental autoimmune encephalomyelitis . Glucagon-like peptide 1 . Glucagon-like peptide 1 receptor . Exendin-4 . Microglia

Introduction

Glucagon-like peptide-1 (GLP-1) is an incretin hormone that exerts diverse biological functions via its specific G protein- coupled receptor, GLP-1 receptor (GLP-1R), such as in- creased β cell neogenesis, reduced β cell apoptosis, and im- proved insulin sensitivity. For this reason, GLP-1 and other GLP-1R agonists are thought to be key molecules in drug development for the treatment of diabetes mellitus [1]. Representatively, exendin-4, a GLP-1R agonist, is a novel drug for type II diabetes and has been on the market since 2005 (exenatide, Byetta®, Astrazeneca) [2].

Recent evidence has raised the possibility that GLP-1 and GLP-1R also play various biological roles in the central nervous system (CNS). GLP-1-immunoreactive fibers and GLP-1R appear to be present throughout the brain [3, 4]. Moreover, GLP-1 is synthesized by neurons in the solitary nucleus and can act as a neuropeptide [4]. Recently, valida- tion of the functional roles of the GLP-1/GLP-1R signaling axis has been evolving to include diverse CNS disorders [5]. Extensive gain- or loss-of-function studies have demon- strated that activation of this signaling axis leads to protec- tion against neuroinflammation, oxidative stress, neurotox- icity, memory impairment, and seizure activity [6–10]. In particular, diverse in vivo evidence has validated medically relevant benefits of exendin-4 in CNS disorders using asso- ciated rodent models: exendin-4 administration ameliorated brain
damage in Alzheimer’s disease (AD), Parkinson’s dis- ease (PD), amyotrophic lateral sclerosis (ALS), trauma, and cerebral ischemia [11–14]. In addition, these therapeutic benefits of exendin-4 are closely associated with reduced microglial activation, leading to reduced immune responses in the CNS [15, 16]. Therefore, more therapeutic benefits of exendin-4 may await to be uncovered in other CNS disor- ders, since its activity is associated with neuroinflammation, a common feature of diverse CNS disorders.

Multiple sclerosis (MS) is an autoimmune disease that oc- curs in the CNS and affects more than 2.1 million individuals globally, with a preponderance of those with younger age, and women [17]. In MS, the immune system attacks the myelin sheath, causing demyelination and neurodegeneration in the CNS, in which one of the key pathological features is neuro- inflammation modulated by activated microglia [18]. Activated microglia stimulate myelin loss and axonal damage, which is mediated by increasing oxidative stress and proin- flammatory cytokines such as interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF)-α [19, 20]. A number of reports, in particular, have provided direct evidence that experimental autoimmune encephalomyelitis (EAE) severity is reduced when microglia are inactivated by the administration of minocycline, a well-known inhibitor of microglial activation [21, 22]. In addition, FTY720 (fingolimod, Gilenya™, Novartis, 2010), a currently used drug for MS, has been shown to reduce the clinical symptoms of EAE and increase remyelination, along with decreasing microglial activation [23, 24]. In view of the link between microglia and the GLP-1/GLP-1R signaling axis, reports regarding the expres- sion of GLP-1R on microglia [25] and reduced GLP-1 secre- tion from activated microglia [26] further suggest a possible role for the microglial GLP-1/GLP-1R signaling axis in MS. In fact, there is a patent application that claims the importance of the GLP-1/GLP-1R signaling axis in MS by demonstrating prophylactic effects of exendin-4 via suppression of T cell functions in rodent models (WO 2011/024110A2, BGLP-1 receptor agonists for treating autoimmune disorders,^ Pfizer and Rinat). It is believed that this patent application indicates the GLP-1/GLP-1R signaling axis as a novel etiological factor for MS [5]; however, there have yet to be published reports. Furthermore, it is yet unclear whether exendin-4 has therapeu- tic potential for MS and whether its efficacy is associated with reduced pathological features in the CNS, particularly associ- ated with its efficacy in microglia.

In the present study, we report that microglia are the main loci for the GLP-1/GLP-1R signaling axis in the spinal cord and that the signaling components are downregulated in vivo upon EAE challenge and in vitro in activated microglia. Importantly, we report that the activation of the GLP-1/GLP- 1R signaling axis employing exendin-4 has therapeutic poten- tial against monophasic EAE, along with reduced neuroinflammatory responses.

Materials and Methods

EAE Induction and Drug Treatment in Mice

EAE was induced by recombinant myelin oligodendrocyte glycoprotein (MOG35–55, MEVGWYRSPFSRVVHLY RNGK, >95% purity, 2 mg/ml) immunization in 7-week-old female C57BL/6 mice, as previously reported [23]. Briefly, mice were immunized on day 1 by subcutaneous injection of an emulsion of MOG (200 μg) in complete Freund’s adjuvant (CFA), along with an injection of Bordetella pertussis toxin (400 ng/mouse, i.p. on days 1 and 3). Mice were weighed and monitored daily for EAE clinical symptoms as follows: 0, healthy mouse, no signs of EAE; 0.5, some lack of tone, however, some strength at the base of the tail; 1.0, total loss of tail tonicity and flaccid tail; 2.0, hind limb weakness; 2.5, incomplete paralysis of one or both hind limbs; 3.0, total pa- ralysis of one of both hind limbs; 4, hind and fore limb paral- ysis; and 5, death from disease.Exendin-4 (His-Gly-Glu-Gly-Thr-Phe-Thr-Ser- Asp-Leu- Ser-Lys-Gln-Met-Glu-Glu-Glu-Ala-Val-Arg-Leu-Phe-Ile- Glu-Trp-Leu-Lys-Asn-Gly-Gly-Pro-Ser-Ser-Gly-Ala-Pro- Pro-Pro-Ser-NH2) or FTY720 (a positive control; provided by Dr. Jerold Chun at The Scripps Research Institute (La Jolla,CA, USA)) were dissolved in normal saline (0.9% NaCl) and intraperitoneally injected every day for 13 days to symptom- atic EAE mice at a dose of 5 μg/kg or 3 mg/kg. The dose of exendin-4 (5 μg/kg) was chosen based on a literature compar- ison [27, 28].

Histological and Immunohistochemical Analysis of the Spinal Cord

Spinal cords were removed from mice of each group (naïve, vehicle-treated EAE, exendin-4-treated EAE, and FTY720- treated EAE groups) at the end of the experiment. Removed spinal cords were embedded in Tissue-Tek Optimal Cutting Temperature (OCT) compound, frozen on powdered dry ice, and cut into sections (20 μm) using a cryostat (J4800AMNZ, Thermo, Germany). Sections were fixed in 4% paraformaldehyde (PFA) and processed for staining or immunolabeling.

To determine an increased cell density, sections were stained with 0.5% cresyl violet in acetate buffer, dehydrated with a series of increasing ethanol baths and xylene, and mounted with Entellan medium. To determine demyelination, sections were stained with fluoromyelin solution (1:300 dilu- tion in phosphate-buffered saline), counterstained with 4′,6- diamidino-2-phenylindole (DAPI) solution (1:10,000 dilu- tion), and mounted with VECTASHIELD® mounting media. For immunohistochemical analysis, sections were blocked with 1% fetal bovine serum containing 0.5% Triton X-100, and labeled with primary antibodies against glial fibrillary acidic protein (GFAP) or ionized calcium-binding adapter molecule 1 (Iba1) for 16 h at 4 °C to determine astrogliosis or microglial activation and morphological transformation of activated microglia. Iba1-immunopositive cells were visual- ized by DAB staining system as described previously [29]. To determine NF-κB p65 expression, spinal cord sections of various EAE experimental groups underwent antigen-retrieval with 1× Tris-EDTA at 100 °C followed by blocking with 1% fetal bovine serum (FBS). Sections were then incubated with rabbit NF-κB p65 primary antibody (1:50) for 16 h at 4 °C, followed by incubating with a biotinylated secondary anti- body (1:200) for 2 h at room temperature. Sections were then incubated with ABC kit and NF-κB signals were visualized by DAB staining system. Alternatively, fixed tissue sections were processed for double immunofluorescent labeling. Sections were labeled with primary antibodies against GLP-1 or GLP-1R, along with appropriate antibodies against cell type- specific markers, including neuronal nuclei (NeuN, neurons), GFAP (astrocytes), and Iba1 (microglia) for 16 h at 4 °C. The sections were further incubated with secondary antibodies conjugated to Alexa-Fluor® 488 or Cy5, counterstained with DAPI, and mounted using VECTASHIELD® mounting media.

Colored or fluorescent images were obtained with a micro- scope equipped with a DP72 camera (Olympus). Double im- munofluorescent images were collected using laser scanning confocal microscopy (Eclipse A1 Plus, Nikon, Tokyo, Japan). The degree of colocalization of GLP-1 and GLP-1R with different CNS cell markers (NeuN, GFAP and Iba1) were quantified using Nikon confocal microscopic analysis tool, whereas fluorescence intensity of fluoromyelin and GFAP staining was quantified using the ImageJ software (National Institute of Mental Health, Bethesda, MD). Quantification of CV staining, Iba 1 immunolabeling, and NF- κ B immunolabeling was performed by counting the number of labeled cells in the white matter region of the spinal cord.

Rat Primary Microglia Culture

Primary microglia cells were cultured from the cerebral corti- ces of neonatal rats (2 days old, Orient Bio). Briefly, rat cor- tices were triturated into single cells in DMEM/F12 contain- ing 10% fetal bovine serum and 1% penicillin/streptomycin and plated onto a poly-L-lysine-coated 75 cm2 T-flask for 2 weeks. Microglia were detached by mild shaking and ap- plied to a nylon mesh to remove cell clumps. Microglia (5 × 105 cells/well) were seeded on poly-L-lysine-coated 6- well plates. After 24 h, cells were treated with lipopolysaccha- ride (LPS, 1 μg/ml) and/or exendin-4 and further incubated for the indicated time points.

Quantitative Real-Time Polymerase Chain Reaction

Spinal cords were removed from mice of each group (naïve, vehicle-treated EAE, and exendin-4-treated EAE groups) at the end of the experiment and used for RNA isolation. In case of cells, rat primary microglia were stimulated with LPS in the presence or absence of exendin-4 for the indicated time points and processed for RNA isolation. Total RNA (1 μg) isolated from spinal cord tissues or rat primary microglia using RNAiso Plus was reverse transcribed and synthesized cDNA was proc- essed for quantitative real-time polymerase chain reaction (qPCR) using target gene-specific primer pairs (Table S1). Target genes including GLP-1, GLP-1R, IL-1β, IL-6, TNF-α, and IL-17 were amplified using Power SYBR Green PCR Master Mix in a StepOnePlus™ Real-Time PCR system (Appl, Biosys). The expression levels of tar- get genes were calculated as a fold change relative to the control, following normalization to mouse β-actin or rat GAPDH.

Western Blot Analysis

Protein samples from vehicle-, LPS- stimulated, and exendin-4 exposed rat primary microglia were prepared using cell lysis buffer. Proteins (10 μg) of each group were separated by 10% sodium dodecyl sulfate polyacrylamide agarose gel electrophoresis (SDS-PAGE) and transferred to the polyvinylidene fluoride membrane. The membrane was then blocked with 5% skim milk, washed with Tris-buffered saline containing 0.2% Tween 20 (TBST) and incubated with rabbit IκBα (1:1000) and mouse β-actin (1:5000) overnight at 4 °C. Membranes were then washed with TBST and exposed to respective secondary antibodies (goat anti-rabbit or goat anti-mouse, 1:10,000) for 2 h at room temperature, washed, and visualized using the ECL western blotting substrate (Thermo Scientific). The band intensity of each protein was calculated using the ImageJ software.

Materials

A recombinant MOG35–55 was obtained from Peptron (Daejeon, Republic of Korea). CFA containing Mycobacterium tuberculosis H-37 RA (5 mg/mL) was pur- chased from Chondrex (Redmond, WA, USA). Bordetella pertussis toxin or exendin-4 was supplied by List Biological Laboratories (Campbell, CA, USA) or Tocris Bioscience (Bristol, UK). Primary antibodies were obtained from Abcam (Cambridge, UK; antibodies against GLP-1, GLP- 1R, and Iba1), Sigma-Aldrich (St. Louis, MO, USA; GFAP), Santa Cruz Biotechnology (Santa Cruz, USA; NF-κB p65), Cell Signaling (Hitchin, UK; IκBα and β-actin), or Millipore (MA, USA; NeuN). Alexa-Fluor® 488- or Cy5-conjugated secondary antibody was obtained from Thermo Fisher Scientific (MA, USA) or Jackson Immuno Research (PA, USA). A kit for reverse transcription, RNAiso Plus, or Power SYBR Green PCR Master Mix was purchased from Agilent Technologies (Santa Clara, CA, USA), Takara (Japan), or Thermo Fisher Scientific (MA, USA). Entellan medium, VECTASHIELD® mounting media, or Tissue- Tek® OCT compound was supplied by Merck (Whitehouse Station, NJ, USA), Vector Laboratories, Inc. (Burlingame, CA, USA), or Surgipath (Richmond, IL, USA). All other re- agents that were not specified were also from Sigma-Aldrich.

Animal Welfare and Ethical Statement

C57BL/6J mice were housed with food and water ad libitum under the controlled conditions of constant temperature (23 ± 1 °C), relative humidity (60 ± 10%), and a 12:12-h light/dark cycle. All animal experimental procedures used in this work were approved by the Institutional Animal Care and Use Committee (IACUC) of Lee Gil Ya Cancer and Diabetes Institute (LCDI), Gachon University, Republic of Korea (# of an approved protocol: LCDI-2013-0055).

Data Analyses

Statistical analyses were performed using the GraphPad Prism Version 5.02 (GraphPad, La Jolla, CA, USA). All data are presented as the mean ± SEM. P values were determined using a Student’s t test between two groups or one-way ANOVA followed by a Newman-Keuls post hoc test for mul- tiple comparisons. P < 0.05 was considered to be statistically significant.

Results

GLP-1 and GLP-1R Are Ubiquitously Expressed in Microglia of the Spinal Cord and Are Downregulated in EAE Mice

We determined the cellular localization and expression of GLP-1 and GLP-1R in the spinal cord of naïve mice using double immunofluorescence labeling for cell type-specific markers including NeuN (mature neurons), GFAP (astro- cytes), and Iba1 (microglia). GLP-1 was expressed in neurons, astrocytes, and microglia (Fig. 1). GLP-1R was also expressed in these three cell types (Fig. 2). Interestingly, both GLP-1 and GLP-1R were highly expressed in microglia, which was ob- served across the white matter of the spinal cord. These results suggest that microglia are the main loci for the GLP-1/GLP- 1R signaling axis in the spinal cord.

We next determined whether EAE influences mRNA ex- pression levels of the signaling components, GLP-1 and GLP- 1R, in the spinal cord using qPCR analysis. In the spinal cords of EAE-symptomatic mice, mRNA expression of both GLP-1 and GLP-1R was markedly downregulated compared with naïve mice, being approximately 50% lower (Fig. 3). These results indicate that the activity of the GLP-1/GLP-1R signaling axis may be attenuated in EAE.

GLP-1 and GLP-1R Are Downregulated in Cultured Microglia That Have Been Activated by LPS

Our results reveal that both GLP-1 and GLP-1R are abundant in microglia and that the mRNA expression of these mole- cules is markedly downregulated in symptomatic spinal cords. Furthermore, microglia are known to be activated in the spinal cord of EAE mice, suggesting that the observed downregula- tion of GLP-1 and GLP-1R in EAE may occur in activated microglia. Therefore, we examined whether GLP-1 and GLP- 1R are downregulated in activated microglia. Temporal changes in GLP-1 and GLP-1R gene expression were deter- mined by qPCR in cultured rat primary microglia that had been activated with LPS. In LPS-stimulated microglia, mRNA expression of GLP-1 (Fig. 4a) and GLP-1R (Fig. 4b) was downregulated compared with control cells, which was remarkable 2 days after LPS exposure. These results indicate that GLP-1 and GLP-1R expression is downregulated in acti- vated microglia, raising the possibility that the observed downregulation of both components of GLP-1 signaling in spinal cords of EAE mice may be due to their downregulation in activated microglia.

Activation of GLP-1R Improves Clinical Symptoms and Reduces Spinal Cord Damage in EAE

Activation of the GLP-1/GLP-1R signaling axis may have relevance to therapeutic potential for EAE, in view of the observed attenuation of the signaling axis, which was ad- dressed here using exendin-4, a GLP-1R agonist, under the therapeutic regimen. EAE-symptomatic mice were given exendin-4 (5 μg/kg, i.p.) daily for 13 days. Exendin-4 admin- istration significantly improved the clinical symptoms of EAE mice between days 34 and 43, with the exception of day 36, when compared with the vehicle-treated group (Fig. 5a). Moreover, the effectiveness of exendin-4 was obvious when analyzed based on the cumulative clinical score between days 30 and 43 or the responsiveness calculated from the clinical score on the first day of administration for each mouse. Exendin-4 administration markedly reduced the cumulative score (Fig. 5b) and improved the responsiveness (Fig. 5c) compared with the vehicle-administered group. This neuro- protective effect of exendin-4 was comparable to FTY720 (3 mg/kg, i.p., Fig. 5), a currently used drug for MS. These data demonstrate that exendin-4 is therapeutically effective for EAE, indicating an enhancement of GLP-1 signaling activity as a possible strategy to treat MS.

Fig. 2 GLP-1R is also abundantly expressed in microglia of the naïve spinal cord. (a–c) Confocal images showing double immunofluorescence labeling of GLP-1R (green) and markers for specific neural cell types (red), such as NeuN (neurons, a), GFAP (astrocytes, b), and Iba1 (microglia, c) in the spinal cord of the naïve mouse. Scale bars, 5 μm. Quantitative colocalization analysis of GLP- 1R with each specific cell marker (d)

Fig. 3 a, b Expression levels of GLP-1 and GLP-1R mRNA are downregulated in the injured spinal cords of EAE mice. Mice were challenged with EAE, and spinal cord samples were used to determine changes in GLP-1 and GLP-1R mRNA expression levels. qPCR analysis showing that GLP-1 and GLP-1R mRNA levels were reduced in the lumbar spinal cord of EAE-symptomatic mice (control gene: β-actin). n = 4 per group. *P < 0.05 vs. naïve (t test)

Assessment of well-known histopathological features in EAE further support the therapeutic efficacy of exendin-4, involving demyelination, cell accumulation in the white mat- ter, astrogliosis, and microglial activation [23, 30]. In vehicle- treated EAE mice, demyelination (Fig. 6a, d), increased cell density (Fig. 6b, d), astrogliosis (Fig. 6c, d), and microglial activation (Fig. 7a, b) in the white matter of the lumbar spinal cord were identified by decreased myelin staining (fluoromyelin, Fig. 6a, d), an increased number of stained cells (cresyl violet, Fig. 6b, d), or increased immunoreactivity of GFAP (astrogliosis, Fig. 6c, d) or Iba1 (activated microglia, Fig. 7a, b). In addition, the morphological transformation (i.e., from ramified to amoeboid morphology) of Iba1- immunopositive microglia was dramatically increased in the injured spinal cord (Fig. 7). All these histopathological sequel- ae of EAE were markedly reduced by exendin-4 administra- tion (Figs. 6 and 7).

Exendin-4 Reduces the Expression Level of Proinflammatory Cytokines in the Injured Spinal Cords of EAE and LPS-Stimulated Cultured Microglia

Several proinflammatory cytokines are considered to be major pathogenic components in EAE, including IL-1β, IL-6, IL-17, and TNF-α [20, 31, 32], and GLP-1 has been suggested to be a potent modulator of the expression of proinflammatory cy- tokines in the CNS [7]. Expression of these cytokines at the mRNA level was determined by qPCR analysis. As reported, mRNA expression of all these cytokines was upregulated in the injured spinal cords of EAE mice (Fig. 8), which was reduced by exendin-4 administration (Fig. 8). These data in- dicate that activation of GLP-1R by exendin-4 can regulate proinflammatory cytokine expression at the transcriptional level during EAE.

Fig. 4 a, b Expression levels of GLP-1 and GLP-1R mRNA are downregulated in LPS-stimulated rat primary microglia. Cells were stimulated with LPS (1 μg/ml) for different lengths of time (1, 2, and 3 days) and used to determine changes in GLP-1 and GLP-1R mRNA expression levels. qPCR analysis showing that GLP-1 and GLP-1R mRNA levels were reduced in LPS-stimulated microglia (control gene: GAPDH). n = 3 per group.

Fig. 5 Exendin-4 administration reduces clinical symptoms in EAE mice. Exendin-4 (Ex4, 5 μg/kg), a positive control drug (FTY720: FTY, 3 mg/kg), or vehicle (veh) was intraperitoneally injected daily into EAE- symptomatic mice, from days 29 to 42 (for 13 days), and the experiment was terminated on day 43. a The mean clinical score of mice exposed daily to exendin-4 or FTY720 compared with the vehicle group. The box indicates the period of daily drug exposure. The clinical scores of each group were statistically analyzed by the Mann-Whitney test. b The cu- mulative score of each mouse exposed daily to exendin-4 or FTY720 compared with the vehicle. The cumulative scores were calculated as the sum of all clinical scores of each mouse during the drug exposure time (starting the day after exposure, days 30–43). c Responsiveness of each mouse exposed daily to exendin-4 or FTY720 compared with the vehicle. Responsiveness was calculated as the percentage decrease in clinical score at the end of the experiment (day 43) versus the score at the 1st exposure of each mouse (day 29). n = 10–12 per group. *P < 0.05, **P < 0.01, and ***P < 0.001 vs. EAE exposed to vehicle (EAE + veh). The Mann-Whitney test (a) and Newman-Keuls multiple range test (b, c).

In microglia, shown here to express GLP-1R abundant- ly, proinflammatory cytokines such as IL-1β, IL-6, and TNF-α are produced upon their activation [19, 20], resulting in immune responses in the CNS. To determine whether exendin-4 exposure also reduces the expression of these cytokines in activated microglia, the mRNA level of IL-1β, IL-6, and TNF-α was analyzed by qPCR in LPS-stimulated rat primary microglia treated with different concentrations of exendin-4 (10 or 100 nM). Exendin-4 exposure significantly reduced the expression of IL-1β (Fig. 9a) and TNF-α (Fig. 9c), but not IL-6 (Fig. 9b) at the mRNA level in LPS-stimulated microglia. These data indicate that activation of GLP-1R by exendin-4 can regulate the mRNA expression of proinflam- matory cytokines in activated microglia, suggesting a possible link between EAE and microglia, in view of the role of GLP-1 receptor signaling in proinflammatory cytokine expression.

Exendin-4 Suppresses Activation of NF-κB Signaling in EAE and LPS-Stimulated Rat Primary Microglia

The transcription factor NF-κB activation is considered as the major source of inflammatory cascades in several neuroinflammatory disorders that exacerbate the disease path- ogenesis in various CNS diseases including autoimmune en- cephalomyelitis [33]. In this study, exendin-4 administration significantly attenuated the number of NF-κB p65- immunopositive cells in EAE spinal cord (Fig. 10a, b) indi- cating the anti-inflammatory effects of exendin-4 via sup- pressing NF-κB activation. NF-κB activation is known to oc- cur in glial cells to initiate inflammatory reactions in EAE [reviewed in [33]]. Therefore, we determined in which cell types NF-κB activation occurs between activated microglia or reactive astrocytes. Our results revealed that NF-κB p65 was profusely expressed in activated microglia rather than reactive astrocytes in EAE spinal cord, as most of NF-κB p65-immunopositive cells were also Iba1-immunopositive cells whereas very few NF-κB p65 signals were colocalized with GFAP+ cells (Fig. S2).

To reaffirm these in vivo findings, we determined the ex- tent IκB degradation in LPS-stimulated rat primary microglia in the presence or absence of exendin-4. The activation of NF-κB signaling involves IκBα degradation [34]. Exendin-4 treatment significantly attenuated LPS-induced IκBα degra- dation (Fig. 10c, d), indicating that exendin-4 suppress activa- tion of NF-κB pathway in activated microglia.

Discussion

GLP-1-triggered activation of GLP-1R exerts pharmacolog- ical potential for hyperglycemia by regulating insulin release and sensitivity. Through these biological features, a few drugs that activate GLP-1R, including exendin-4, have been successfully marketed to treat type II diabetes. In addition to diabetes, the GLP-1/GLP-1R signaling axis is considered to be a key component in overcoming CNS disorders, due to the presence of both signaling components throughout the brain [3, 35] and the importance of diabetic or hyperglyce- mic conditions as risk factors for CNS disorders [36]. As a result, we know that several types of CNS disorders can be therapeutically assessed by the activation of GLP-1R, in- cluding AD, PD, cerebral ischemia, and peripheral neuropa- thy [5, 9, 37, 38]. The current study identified the same benefit of GLP-1R activation, employing the GLP-1R ago- nist exendin-4, in MS using a monophasic EAE model, along with the downregulation of the signaling components in the injured spinal cords of EAE. Moreover, the present study identified microglia as the main loci for the presence of both GLP-1 and GLP-1R in the spinal cord and demon- strated their activation as a factor for the downregulation of both signaling components, supporting a possible role of microglial activities in GLP-1 signaling in MS. Activation of GLP-1 signaling triggered by exendin-4 administration reduced the immune response in the injured spinal cords of EAE, which was further supported by in vitro data demon- strating a reduced immune response in activated microglia upon activation of GLP-1 signaling. These collective data suggest that enhancement of GLP-1 signaling activity may have medically relevant potential in MS.

Fig. 7 Exendin-4 administration reduces microglial activation and morphological transformation into amoeboid cells in the injured spinal cord of EAE-symptomatic mice. Spinal cord samples from naïve, EAE, and EAE mice exposed to enendin-4 (Ex4) were histologically assessed for the expres- sion of microglial cells through Iba1 immunohistochemistry. a Representative images of Iba1-immunopositive cells in the injured spinal cord. Scale bar, 20 μm. Satellite images of each panel are arrowhead-indicated representative cell shapes. b Quantification of Iba1-immunopositive cells in the white matter region of injured spinal cord. c Quantification of morphologically transformed (i.e., amoeboid) Iba1-positive cells in the injured spinal cord of EAE- symptomatic mice. n = 4 per group. ***P < 0.001 vs. naïve and ##P < 0.01, ###P < 0.001 vs. vehicle-treated EAE (EAE + veh), respectively. Newman-Keuls multiple range test.

Fig. 8 Exendin-4 administration reduces the mRNA expression level of proinflammatory cytokines in the spinal cord of EAE-symptomatic mice. Spinal cord samples from naïve and EAE mice exposed to drugs were used to determine changes in the expression level of proinflammatory cytokines based on qPCR analysis, such as IL-1β (a), IL-6 (b), IL-17 (c), and TNF-α (d). n = 4 per group. **P < 0.01 and ***P < 0.001 vs. naïve and ##P < 0.01 and ###P < 0.001 vs. vehicle-treated EAE (EAE + veh), respectively. Newman-Keuls multiple range test.

The regional or cellular locations of GLP-1 and GLP-1R in the CNS have been independently reported. Immunohisto- chemical analysis has demonstrated that GLP-1-positive neu- ral cells are present in the nucleus of the solitary tract and the dorsal and ventral parts of the medullary reticular nucleus [3, 4]. Moreover, GLP-1 immunoreactive nerve fibers have been seen throughout the forebrain, including the hypothalamic, thalamic, and cortical areas [3]. In addition to GLP-1, its bind- ing receptor, GLP-1R has also been reported to be expressed in the spinal cord [25]. Similarly, here, we observed that both GLP-1 and GLP-1R are present in the spinal cord. More spe- cifically, these are ubiquitously expressed in microglia, and relatively less in neurons and astrocytes, implicating microglia as the main cell type in which GLP-1 signaling is activated. This notion is further supported by previous reports. Microglia express the proglucagon gene and secrete GLP-1 in a cAMP-dependent manner [26]. In addition to the ligand, GLP-1R is also specifically expressed on microglia in the spinal cord [25] or brain [39]. There are also several reports to explain GLP-1R localization in different CNS cell types: these reports demonstrated neurons and endothelial cells as main loci for GLP-1R expression in normal or diabetic brain as well, but they did not determine GLP-1R expression in glial cells [9, 16, 39–44]. In view of the important roles of microg- lia in MS, the ubiquitous expression of GLP-1 and GLP-1R in microglia shown in this study may serve to control the activ- ities of GLP-1 signaling across the spinal cord.

Fig. 9 Exendin-4 reduces the m RNA expression level of proinflammatory cytokines in LPS-stimulated rat primary microglia. Rat primary microglia were treated with LPS (1 μg/ml) and/or exendin- 4 (Ex4; 0, 10, and 100 nM) for 6 h and used to determine changes in the expression level of proinflammatory cytokines based on qPCR analysis, such as IL-1β (a), IL-6 (b), and TNF-α (c). n =4 per group. ***P < 0.001 and #P < 0.05 vs. untreated control and LPS only, respectively. Newman- Keuls multiple range test.

In the current study, we further identified that gene expres- sion of both GLP-1 and GLP-1R was downregulated in the spinal cord of EAE-symptomatic mice. Both were also down- regulated at the mRNA level in cultured microglia, the main cell type to express these signaling components, upon their activation. Consistent with the current observation, microglial GLP-1 secretion has been reported to be decreased in response to LPS stimulation [26]. However, there also exists a contrast- ing report to ours on the temporal changes of the GLP-1R mRNA level in injured spinal cords. In neuropathic pain mod- el by nerve ligation, GLP-1R was identified to be upregulated in injured spinal cords through immunohistochemical analysis [25]. Besides this controversial finding in the spinal cord, there are additional controversial, but complicated, findings in the brain [40, 45, 46]. Upon LPS exposure, GLP-1R ex- pression was upregulated in the brain of 3 weeks old mice, but not in the brain of aged (3 months old) mice [46]. In trauma- challenged brain, GLP-1R expression was also upregulated in astrocytes [39, 40]. Similarly, GLP-1R expression was upreg- ulated in astrocytes and GABAergic interneurons after ische- mic challenge [45]. Unlikely the temporal changes of GLP-1R expression in cell levels, its expression in the CA1 region of ischemic brain was not changed 2 days after ischemic chal- lenge, but downregulated 4 days after the challenge [45]. The latter is consistent with our findings. These dynamic changes could occur in MS. Ordinarily, MS patients experience peri- odic events such as relapse and remission [17]. In this context, GLP-1R expression levels could be altered depending on the disease phases. In this study, we used a monophasic EAE model that expresses a single relapse phase. Therefore, it might to be interesting to determine GLP-1R expression by the different disease phases. Despite these uncertain possibil- ities, it is true that GLP-1R mRNA was downregulated in the injured spinal cord of EAE mice and activated microglia, im- plicating a link between the downregulation of GLP-1 signal- ing components and inflammatory disease progression. As observed in the present study, exendin-4-triggered GLP-1 sig- naling reduced damage in EAE mice where GLP-1 and GLP-1R were downregulated. This notion is also supported by studies showing that GLP-1R is downregulated in non-alcoholic steatohepatitis, an inflammatory disorder of the liver, and in mouse hepatocytes following a high-fat diet, an inducer of inflammation [47]. Moreover, activation of GLP-1 signaling by exendin-4 administration reduces steatohepatitis in fatty liver [47] and ischemic brain [45]. Despite the discrepancy in the expression of GLP-1R in the spinal cords of EAE and neuropathic pain models, we clearly demonstrate that activation of GLP-1R reduced spinal cord damage in EAE mice. This finding, along with the downreg- ulation of the ligand and its receptor in GLP-1 signaling, im- plies that this signaling is a crucial factor in the regulation of EAE symptoms, further indicating that GLP-1R activation or upregulation of GLP-1/GLP-1R expression could alleviate the clinical symptoms of MS. Interestingly, the latter is supported, in part, by our findings that the mRNA expression level of GLP-1 and GLP-1R was restored by exendin-4 administration in EAE mice (Fig. S1). Similarly, GLP-1R expression is up- regulated in the hepatocyte cell lines, HepG2 and Huh7 [47], and in the CA1 region of ischemic brain [45] in response to exendin-4 treatment.

Fig. 10 Exendin-4 administration reduces NF-κB p65 expression in the injured spinal cord of EAE-symptomatic mice and attenuates IκBα degradation in LPS-stimulated rat primary microglia. a, b Spinal cord samples from naïve, EAE, and EAE mice exposed to enendin-4 (Ex4) were immunohistochemically assessed for the expression of NF-κB p65- immunopositive cells. a Representative images of NF-κB- immunopositive cells in the injured spinal cord. Scale bar, 20 μm. b Quantification of NF-κB p65-immunopositive cells in the white matter region of the injured spinal cord. c, d Rat primary microglia were treated with LPS (1 μg/ml) and/or exendin-4 (Ex4; 100 nM) for 2 h and used to determine changes in the expression level of IκBα based on western blot analysis. c Representative western blot images of IκBα expression in LPS-stimulated rat primary microglia cells. d Quantification of IκBα expression relative to the β-actin. n = 4 per group. ***P < 0.001 vs. naïve and ##P < 0.01 vs. vehicle-treated EAE (EAE + veh) in (b) and vs. vehicle and LPS-treated primary microglial cells in (d), respectively. Newman-Keuls multiple range test.

Even with the significant neuroprotection by exendin-4 in EAE, the reduced GLP-1R expression can affect the efficacy of exendin-4 like the diminished drug efficacy, which was not determined in the current study. Interestingly, the similar pat- tern was reported in case of FTY720, a drug for MS. Upon EAE challenge, S1P1, a known target of FTY720, was down- regulated in the injured spinal cord of EAE mice [48], but FTY720 administration after EAE onset remarkably improved disease symptoms [23]. Moreover, exendin-4 administration reversed gene expression levels of GLP-1R (the current study) and FTY720 administration did so for S1P1 [48]. Similarly, GLP-1R expression was reduced in the CA1 region of ische- mic brain 4 days after ischemic challenge, which was reversed with exendin-4 administration [45].

The well-known biological feature of activated GLP-1 sig- naling is an anti-inflammatory response in the nervous system, which has been evidenced through extensive in vivo studies employing specific disease models, including AD [49], ALS [12], cerebral ischemic stroke [44, 45], and neuropathic pain [25, 50]. In MS, an excessive immune response is also con- sidered to be a feature for disease progression, in which the production of diverse proinflammatory cytokines is a well- known pathogenic event [31, 32]. Likely, in an animal model of MS (EAE), several cytokines are involved in disease pro- gression, particularly IL-17, IL-1β, IL-6, and TNF-α [20, 23]. In the present study, exendin-4 administration decreased the mRNA expression level of these proinflammatory cytokines in the spinal cords of EAE mice. Similarly, in LPS-stimulated microglia, we observed that exendin-4 exposure reduced the mRNA expression of IL-1β and TNF-α, but not IL-6. Therefore, the neuroprotective effects of GLP-1R activation in EAE may be associated with a reduced inflammatory re- sponse. This link is supported by several findings. Exposure to an active fragment of GLP-1 suppressed the production of IL- 1β in LPS-stimulated rat primary astrocytes [7] and improved synaptic and learning impairments in the inflamed brain [51]. In particular, there have been several reports dealing with the link between microglial activation and GLP-1 signaling acti- vation. Liraglutide, a long-lasting GLP-1 analog, has been shown to reduce microglial activation in a mouse model of AD [49] and in the inflamed brain following exposure to x-ray, along with reduced the levels of proinflammatory mediators such as IL-6, IL-12q70, IL-1β, and nitrite [52]. Moreover, the neuroprotective effects of exendin-4 were accompanied by reduced microglial activation in cerebral ischemia [16, 53] and PD [15]. In the latter, exendin-4 administration suppressed TNF-α and IL-1β expression in the brain of a MPTP-induced PD model [15].

It is of note that activation of GLP-1 signaling by exendin-4 administration affects morphological transformation of acti- vated microglia in EAE. The number of amoeboid transfor- mation of Iba1-positive microglia was dramatically increased in EAE spinal cord, whereas the morphology was significant- ly reversed into ramified cells by exendin-4-administeration. In addition, morphological transformation of activated microglia into amoeboid cells is closely associated with microglial polarization, specifically proinflammatory M1 phe- notype [54]. Therefore, exendin-4 affects microglial polariza- tion by reducing M1 phenotype in EAE. This notion can be further supported by our findings on exendin-4’s anti- inflammatory effects in EAE and LPS-stimulated microglia. All determined cytokines are characteristic feature of M1 phenotype [55, 56]. These cytokines were upregulated in EAE or LPS-stimulated microglia, whereas exendin-4 treat- ment reduced this upregulation. Moreover, activation of NF-κB signaling is known as a marker for M1 polarization of activated microglia [56]. We also showed that exendin-4 treatment dramatically attenuated NF-κB p65 expression in EAE and blocked IκBα degradation in LPS-stimulated prima- ry microglia, further demonstrating that exendin-4 can sup- press microglial polarization into M1 phenotype. In addition to morphological transformation, GLP-1 signaling is assumed to be associated with microglial proliferation in EAE. It was demonstrated that microglial activation and proliferation oc- curs during EAE [57]. In our study, the number of activated microglia was increased in EAE and gene expression levels of GLP-1 signaling components were suppressed. Even though there is no direct evidence, downregulation of GLP-1 signal- ing components might enhance microglial activation and pro- liferation in EAE.

Several mechanisms on how GLP-1 signaling regulates production of inflammatory mediators have been reported, including NF-κB, PI3K, and MAPK [58, 59]. Among these, it has been evidenced that GLP-1 exerts anti-inflammatory effects by inhibiting NF-κB pathway, leading to attenuation of the production of diverse inflammatory mediators [reviewed in [59]]. Exendin-4, a GLP-1 receptor agonist, di- rectly inhibited the NF-κB signaling in LPS-stimulated mac- rophages, which, in turn, attenuated secretion of proinflamma- tory cytokines, such as IL-1β, TNF-α, and IL-6 [58]. In line with these findings, we also observed that exendin-4 treatment significantly attenuated NF-κB p65 expression in EAE spinal cord. Moreover, NF-κB p65 was expressed abundantly in Iba1-immunopositive cells in EAE spinal cords, demonstrat- ing that NF-κB signaling is activated in activated microglia during EAE. Therefore, the attenuation of NF-κB activity in EAE by exendin-4 administration may occur in activated mi- croglia. This notion was, in part, supported by our in vitro findings that exendin-4 suppressed NF-κB activity through inhibiting IκBα degradation in LPS-stimulated primary mi- croglia. Collectively, NF-κB pathway appears to play a critical role in GLP-1 signaling-triggered anti-inflammation. Additionally, in view of the pathogenic role of NF-κB path- way in EAE [60, 61], the protective effects of exendin-4 in EAE may be associated with suppression of NF-κB activity.

We provide evidence that GLP-1 and its receptor, GLP-1R, are downregulated in the spinal cords of EAE mice, assumedly in activated microglia, despite the fact that a clear mechanistic role of local microglial regulation of this signal- ing awaits further identification. We also provide evidence that activation of GLP-1R is neuroprotective against MS using a mouse EAE model, in which exendin-4 administration is therapeutically potent with anti-inflammatory activities such as reduced microglial activation and the expression of proin- flammatory cytokines. Given the growing evidence regarding an important role of GLP-1 signaling in diverse CNS disor- ders, the present study adds MS to the disease types in which this signaling is beneficial, further suggesting the regulation of GLP-1 signaling as a therapeutic strategy for MS treatment.