Salutary effects of glibenclamide during the chronic phase of murine experimental autoimmune encephalomyelitis

Formalin-fixed brain tissues from MS patients were obtained from the Brain and Tissue Bank of the Rocky Mountain MS Center, Anschutz Medical Campus, University of Colorado Denver, Aurora, CO. Detailed pathological examinations of the human tissues were conducted at the University of Colorado Health Sciences Center, Department of Pathology. The tissues were further examined at the University of Maryland by a board certified neuropathologist (RIM). Tissue samples containing one or more demyelinating lesions were obtained from 9 patients, 6 females, aged 61 ± 2.6 years, and 3 males aged 60 ± 8.8 years. The post-mortem interval was 9.8 ± 1.4 h. Controls included 2 females and 4 males, aged 47 ± 9 years, who had died rapidly from non-neurological diseases (acute cardiovascular or respiratory disorders) [15].

Female C57BL/6J mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Abcc8−/− mice, obtained as described [13], exhibited neurological function and spinal cord histology indistinguishable from wild-type (WT) mice. Mice were housed under pathogen-free conditions in the animal facility of the University of Maryland, School of Medicine.

EAE was induced in female WT and Abcc8−/− mice at 8 weeks of age, as described [13]. EAE was induced with myelin oligodendrocyte glycoprotein 35–55 (MOG_35–55) peptide (Biomer Technology, Pleasanton, CA) in complete Freund’s adjuvant containing Mycobacterium tuberculosis (H37RA; Sigma-Aldrich, St. Louis, MO). Mice were immunized by subcutaneous injection in the flank regions (left and right sides) with 200 μL total of an emulsion of MOG_35–55 peptide (200 μg in 100 μL phosphate-buffered saline (PBS) plus 100 μL of complete Freund’s adjuvant containing 0.4 mg of heat-inactivated M. tuberculosis). Each mouse then received 400 ng of pertussis toxin (List Biological Laboratories, Campbell, CA) intraperitoneally (IP) on post-induction day (pid) 0 and pid-2.

Abcc8−/− mice with induced EAE were described previously [13], and so were used in the present study for only a single experiment that examined astrocyte expression of TNF, BAFF, CCL2, and NOS2 in EAE. WT mice with induced EAE were randomly assigned to receive no treatment or glibenclamide treatment beginning either on pid-12 or on pid-24 (these groups are referred to below as “EAE/glib-pid-12” and “EAE/glib-pid-24”). Treatment consisted of administering 10 μg glibenclamide IP daily to mice in the treatment groups until the end of the experiment (pid-40). A stock solution of glibenclamide was prepared by placing 25 mg glibenclamide (#G2539; meets USP testing; Sigma-Aldrich) into 10 mL dimethyl sulfoxide (DMSO). We diluted 200 μL of this solution into 9.8 mL PBS; mice received 200 μL of the final solution.

Scoring of disease severity was carried out as previously described [13]. From pid-1 onwards, mice were assessed daily for signs of paralysis by two independent observers blinded to treatment group. Mice were assigned a clinical score of increasing severity: 1, limp tail; 2, hind limb paresis; 3, complete hind limb paralysis; 4, hind limb paralysis and body/front limb paresis/paralysis; 5, moribund. End point evaluations included mean severity of disease over time and mean day of disease onset (first day of score > 0). Paralyzed mice (scores 3 and 4) were moved to individual cages where food and water were placed at cage floor level. The weight of EAE mice was measured every 2 days, and mice were euthanized if there was a loss of more than 20% in weight or if they become dehydrated.

On pid-40, mice were euthanized by IP injection of pentobarbital (> 100 mg/kg), followed by intracardial perfusion of 10% neutral-buffered formalin (15 mL; Sigma-Aldrich). The spinal cords were removed, immersion fixed for 24 h in formalin, and either cryoprotected 48 h in 30% sucrose prior to cryosectioning (7 μm) or they were paraffin embedded for sectioning.

Chromogen immunohistochemistry was performed on paraffin sections, as described in [13], using VECTASTAIN Elite ABC Kits (#PK-6100; Vector Laboratories) and Mouse on Mouse (M.O.M.) Elite Peroxidase Kit (#PK-2200; Vector Laboratories). Primary antibodies were directed against: CNPase (1:1000; MAB326; EMD Milipore, Billerica, MA), PDGFRα (1:500; sc-338; Santa Cruz Biotechnology, Santa Cruz, CA), SMI-312 (1:1000; SMI-312R; Covance Inc., Gaithersburg, MD), cluster of differentiation (CD) 45 (1:1500; ab10558; Abcam, Cambridge, MA); CD20 (1:100; sc-7735; Santa Cruz Biotechnology); CD3 (1:200; ab5690; Abcam); p65 (1:200; GTX102090; Genetex, Irvine CA); CD86 (1:500; bs-1035R; Bioss,Woburn, MA); CD163 (1:500; bs-2527R; Bioss); TNF (1:500; sc-1350; Santa Cruz Biotechnology); BAFF (1:400; ab16081; Abcam); CCL2 (1:400; ab8101; Abcam); NOS2 (1:500; 482,728; EMD Millipore). Sections were counterstained with hematoxylin and were examined using bright-field microscopy. Quantification was performed by blinded observers using ImageJ software (National Institutes of Health, USA). Cell infiltrates (H&E) in white matter were quantified by counting the number of positive quadrants with inflammatory infiltrates and are reported as the percentage of the total number of quadrants examined. Immunolabeling experiments for specific markers were quantified based on the number of positive cells/field, with each field being a 435 × 325 μm rectangle, and with 7–9 fields covering the entire white matter of one coronal section of the lumbar spinal cord of each mouse.

Fluorescence immunohistochemistry was performed on cryosections. Myelin was assessed by immunolabeling for myelin basic protein (MBP) (1:500; Ab40390; Abcam) and by staining with LFB. Reactive astrocytosis was assessed by immunolabeling for GFAP (1:500; C9205; Sigma-Aldrich). Unbiased measurements of signal intensity within regions of interest (ROIs) were obtained using NIS-Elements AR software (Nikon Instruments, Melville, NY, USA). The area that was evaluated was the dorsal half of the white matter of one lumbar spinal cord section from each mouse. Specific labeling for MBP or LFB or GFAP within the ROI was defined as pixels with signal intensity > 1.5–2.0× background. Percent demyelination in EAE groups without and with treatment was computed based on measurements in normal (non-EAE) mice.

Double immunolabeling was performed for GFAP (1:500; C9205; Sigma-Aldrich) plus TNF (1:500; sc-1350; Santa Cruz Biotechnology) or BAFF (1:400; ab16081; Abcam) or CCL2 (1:400; ab7202; Abcam) or NOS2 (1:500; 482,728; EMD Millipore). Double immunolabeling also was performed for SUR1 (1:800; custom rabbit antibody) plus CD31 (1:200; 550,274; BD Biosciences; San Jose, CA). For visualization, we used fluorescent-labeled species-appropriate secondary antibodies [1:500, Alexa Fluor 488 (green) and Alexa Fluor 555 (red); Invitrogen/Molecular Probes, Eugene, OR] at room temperature. The specificity of the immunolabeling for all proteins was tested in control sections by incubation with pre-immune serum, or after pre-adsorption of the antibody with the peptides used as immunogens.

Human tissues. For the analysis of human tissues, cryosections (10 μm) were stained with H&E or LFB to identify appropriate areas for further evaluation by immunohistochemistry. We used immunolabeling for CD68 (1:200; ab955; Abcam), CD45 (1:200; ab10558; Abcam) and CD3 (1:200; ab5690; Abcam) to categorize demyelinating white matter lesions as either active, chronic active or chronic inactive, per De Groot et al. [16], as well as to analyze subpial/cortical demyelinating lesions. Tissues from the 9 MS patients yielded 13 white matter lesions, including 5 active lesions, 4 chronic active lesions, and 4 chronic inactive lesions. From these 9 cases, we also identified 6 demyelinating subpial/cortical lesions, each from a different patient. Preactive white matter lesions with “normal appearing white matter” were not included in this study.

Immunohistochemistry and immunoFRET were performed on cryosections using custom goat or custom rabbit anti-SUR1, custom chicken anti-TRPM4, or goat anti-TRPM4 (G-20, sc-27,540, Santa Cruz Biotechnology) antibodies, as described [12, 15, 17]. Slides were incubated with a mixture of 2% donkey serum (D9663; Sigma-Aldrich) and 0.2% Triton X-100 for 1 h at room temperature prior to immunolabeling.

Single-label immunohistochemistry was performed using biotin-conjugated secondary antibody. Sections were incubated for 30 min in PBS with 0.3% H₂O₂ to block endogenous peroxidase activity, after which sections were placed in the above blocking solution for 1 h. Following overnight incubation with anti-SUR1 antibody (1:800; custom rabbit antibody [12]) at 4 °C, sections were incubated with biotinylated secondary antibody (1:500; BA-1000; Vector Laboratories, Burlingame, CA) for 1 h. After washing in PBS, sections were incubated in avidin biotin solution (Vector Laboratories) and the color was developed in diaminobenzidine chromogen solution (0.02% diaminobenzidine in 0.175 M sodium acetate) activated with 0.01% hydrogen peroxide. The sections were rinsed, mounted, dehydrated, and cover-slipped with DPX mounting medium (Electron Microscopy Services, Hatfield, PA). Omission of primary antibody was used as a negative control.

Double-label immunohistochemistry was performed using fluorescent secondary antibodies. To identify cell-specific expression, double immunolabeling was performed for SUR1 (1:800; custom rabbit antibody) plus S100B (1:400; ab868; Abcam) or GFAP (1:500; C9205; Sigma-Aldrich) or CD68 (1:200; ab955; Abcam) or CD3 (1:200; ab5690; Abcam). To show co-localization, double immunolabeling was performed for SUR1 (1:800; custom rabbit antibody) plus TRPM4 (1:800; custom chicken antibody [12], as well as for SUR1 (1:800; custom rabbit antibody) plus BAFF (1:400; ab16081; Abcam), or CCL2 (1:400; ab7202; Abcam), or NOS2 (1:500; 482,728; Calbiochem, San Diego, CA). For visualization, we used fluorescent-labeled species-appropriate secondary antibodies [1:500, Alexa Fluor 488 (green) and Alexa Fluor 555 (red); Invitrogen/Molecular Probes, Eugene, OR] or, in the case of chicken anti-TRPM4 antibody, a FITC-conjugated secondary antibody (The Jackson Laboratory), all at room temperature. Sections were cover-slipped with polar mounting medium containing anti-fade reagent and 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Immunolabeled sections were visualized using epifluorescence microscopy (Nikon Eclipse 90i; Nikon Instruments Inc., Melville, NY). Omission of primary antibody was used as a negative control. Specific labeling was defined as fluorescence intensity twice that of background.

Semi-quantitative analysis of SUR1 labeling intensity in various types of demyelinating lesions was performed as described previously for SUR1 [17]. Briefly, all sections had been immunolabeled as a single batch, and all images were collected using uniform parameters of magnification and exposure. Two areas encompassing demyelinating lesions were randomly selected for construction of a montage, with each montage composed of 4 images, each 300 × 400 μm, that were acquired at × 20 magnification. Images were evaluated independently by 2 observers for specific SUR1 immunoreactivity associated with each lesion type. Areas of maximum labeling were scored for each case using a semi-quantitative scale ranging from 0 to 4. The overall concordance between observers was more than 90%. In cases of disagreement, independent reevaluation was performed by a third observer to arrive at the final score. Co-localization of fluorescence signals in double immunolabeled sections (SUR1 plus BAFF or CCL2 or NOS2) was computed as Pearson’s correlation coefficient [18].

Antibody-based Förster resonance energy transfer (immunoFRET) for SUR1-TRPM4 was performed as described above [12, 17], except for the secondary antibodies (CY5-conjugated donkey anti-chicken and CY3-conjugated donkey anti-rabbit antibodies; The Jackson Laboratory), image acquisition using a Zeiss LSM710 confocal microscope, and controls that included omission of one or the other of the two primary antibodies.

Normal (no EAE) mice were administered fingolimod (FTY720, SML0700, Sigma-Aldrich) (3 mg/kg once daily × 3 days by gavage) or glibenclamide (#G2539; Sigma-Aldrich; 10 μg/mouse once daily × 3 days IP). Untreated normal mice served as controls.

To evaluate circulating leukocytes, whole blood was collected aseptically by cardiac puncture using sterile K3 EDTA VACUTAINER blood collection tubes. The fluorochrome-conjugated monoclonal antibodies, FITC rat anti-mouse CD45 (553080), PerCP rat anti-mouse CD4 (553052), and PE rat anti-mouse CD8a (553032), 2 μg each, all from BD Biosciences, were added to 100 μL of whole blood in a 12 × 75-mm-capped polystyrene test tubes (2058; BD Biosciences). The blood was vortexed gently and was incubated for 30 min in the dark at room temperature, after which 2 mL of × 1 FACS Lysing Solution (349,202; BD Biosciences; × 10 diluted to × 1 in distilled water before use) was added. The mixture was vortexed gently and was incubated for 10 min in the dark at room temperature. The suspension was centrifuged at 500 × g for 5 min, the supernatant was removed, 2–3 mL of wash buffer (PBS with 0.1% sodium azide) was added, and the suspension was centrifuged at 500 × g for 5 min. The supernatant was removed and 0.5 mL of 1% paraformaldehyde solution (prepared in PBS with 0.1% sodium azide) was added to fix the cells. Counts of total CD45+ leukocytes, and CD45 + CD4+ and CD45 + CD8+ T-lymphocytes in the three treatment groups were determined by FACS (BD LSR II Flow Cytometer; University of Maryland, Center for Innovative Biomedical Resources) [19].

To evaluate the splenocytes, spleens were removed immediately after euthanasia and were placed in PBS on ice. They were then sliced into small pieces (∼1 mm³) in a Petri dish using surgical blades, and a single cell suspension was prepared by gently forcing the pieces through a 100-μm cell strainer. The resulting cell suspension was centrifuged and resuspended in red blood corpuscle lysis buffer (× 1 FACS Lysing Solution, BD Biosciences), was shaken at room temperature for 10 min, and was washed 2× with PBS with 0.1% sodium azide by centrifugation at 500 × g and resuspension at 4 °C. Isolated splenocytes were counted to estimate total cell numbers. Prior to incubation with antibodies, cells were fixed for 20 min in 1% formaldehyde solution (prepared in PBS with 0.1% sodium azide). Surface labeling was performed using the fluorochrome-conjugated monoclonal antibodies: PE hamster anti-mouse CD3e (#561824), APC rat anti-mouse CD4 (#553051), and PE-CY5 rat anti-mouse CD8a (#553034), all from BD Biosciences and was carried out in FACS buffer (BD Biosciences) with 1-h incubation at room temperature. Counts of total CD3+, CD3 + CD4+, and CD3 + CD8+ T-lymphocytes in the three treatment groups were determined by FACS (BD FACSCanto II system; University of Maryland, Center for Innovative Biomedical Resources).

Mice with symptomatic EAE (n = 3) and control (non-EAE) mice (n = 3) were studied. Bodipy-glibenclamide (TRITC (red); Thermo Fisher Scientific, Waltham, MA), 30 μg, was dissolved in 30 μL DMSO, then was added to 300 μL normal saline (NS). This solution was infused into the jugular vein of the anesthetized mouse, was allowed to circulate 1 h, after which the mouse was euthanized and perfused with NS to remove any intravascular signal. Cryosections of the lumbar spinal cord were either stained with H&E or were immunolabeled for fluorescence immunohistochemistry, as above. Immunolabeling of adjacent sections was performed for MBP (1:500; ab40390; Abcam) and albumin (1:100; ab106582; Abcam), with species-appropriate Alexa Fluor 488-conjugated secondary antibody (green).

Astrocyte primary cultures were prepared from post-natal day 3–5 C57B6 mice or Wistar rats using a modified version of the method described [20]. Animals were euthanized with a barbiturate overdose, whereupon brains were dissected from the skull and were placed in ice-cold Dulbecco’s modified Eagle’s medium containing 4.5 g/L glucose (Invitrogen), supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin. The meninges were completely removed, and the cortices were isolated. Cortices were passed through 70- and 40-μm nylon meshes, and cell suspensions were plated in the above media. Astrocyte cultures were maintained to confluency for ~12 days, with media changes every 3 days. Confluent cultures were rotary shaken at 250 rpm for 18 h to detach contaminating cells, and were washed × 3 with the above media. GFAP immunolabeling demonstrated that cultures were > 95% GFAP+. Astrocyte cultures were maintained for experiments < 2 months.

To model the astrocyte response in MS/EAE, astrocyte cultures were exposed to TNF (20 ng/mL; 8902SC, Cell Signaling, Danvers, MA) plus IFNγ (20 ng/mL; 8901SC, Cell Signaling) overnight [1]. After activation, cells were studied by patch clamp electrophysiology to detect functional SUR1-TRPM4 channels, or they were processed to measure mRNA, as described below.

Data are shown as mean ± standard error of the mean (SE). Non-parametric data (murine clinical scores and SUR1-immunolabeling scores) were analyzed using the Mann-Whitney or the Kruskal-Wallis test with Dunn’s post hoc comparisons, as appropriate. Parametric data were analyzed using the t test or the ANOVA with Fisher’s post hoc comparisons, as appropriate. Statistical analyses were performed using OriginPro version 8 (OriginLab Corp, Northampton, MA). For all experiments, P < 0.05 was considered to be statistically significant.

Clinical scoring was assessed daily on pid-1–40. Untreated EAE mice showed a typical course of EAE, marked by worsening paralysis scores beginning on pid-9–12, peaking several days later, then persisting unchanged through to pid-40 (Fig. 1). The mean clinical score during the late disease period (pid-40) was 2.14 ± 0.10. The day of disease onset and the mean clinical scores at different times during the disease course are shown in Table 2.

In murine EAE, pharmacological inhibition of SUR1 by glibenclamide beginning on pid-10 was found to yield improved clinical scores compared to untreated mice [13]. Here, we examined the effects of glibenclamide administered starting during the chronic phase of EAE. Mice were treated once daily with glibenclamide, 10 μg IP, beginning at pid-24 or, for comparison, at pid-12 and continuing through to pid-40. The clinical course of EAE/glib-pid-12 mice resembled that reported for mice beginning treatment on pid-10: progression of clinical symptoms was halted and clinical scores thereafter declined slightly then stabilized (Fig. 1). In the EAE/glib-pid-24 mice, clinical scores before treatment were as severe as in untreated mice (Fig. 1; Table 2). However, clinical scores improved noticeably within 2 days of beginning treatment and, by 1 week, scores had improved to levels similar to those in the EAE/glib-pid-12 group. The decrease in clinical severity with glibenclamide in the EAE/glib-pid-24 group persisted through the remainder of the experiment. In the EAE/glib-pid-24 group, the mean clinical score for the late disease period (pid-40) was 1.17 ± 0.24, which was significantly different from that in untreated EAE mice (2.14 ± 0.10), but not statistically different from the scores in the EAE/glib-pid-12 group (0.78 ± 0.28) (Fig. 1; Table 2).

At pid-40, spinal cord myelin was evaluated using LFB staining and immunolabeling for MBP [23] (Fig. 2). Commensurate with the clinical findings, white matter tracts in untreated EAE mice stained weakly with LFB and immunolabeled weakly for MBP. Compared to normal controls (no EAE), the spinal cords of untreated EAE mice showed 37.9 ± 5.0% demyelination assessed by LFB, and 26.3 ± 3.4% by MBP. By comparison, the spinal cords of EAE/glib-pid-24 mice showed 23.5 ± 4.2% demyelination assessed by LFB, and 16.9 ± 1.9% by MBP.

Expression of the mature oligodendrocyte marker, 2′,3′-cyclic-nucleotide 3′-phosphodiesterase (CNPase), was used to verify the data on LFB and MBP (Fig. 2). Compared to normal controls, the spinal cords of untreated EAE mice showed 60.8 ± 4.6% demyelination, compared to 8.2 ± 2.1% in EAE/glib-pid-24 mice.

We performed counts of cells positive for platelet-derived growth factor receptor-α (PDGFR-α), a marker of oligodendrocyte precursor cells (OPC) (Fig. 2). Cells with PDGFR-α were significantly increased in both untreated EAE and EAE/glib-pid-24 mice, consistent with increased proliferation of OPC.

To assess for axonal sparing, spinal cords were examined for the pan-axonal neurofilament marker, SMI 312 (Fig. 2). Quantification revealed axonal loss of 54.9 ± 4.7% in untreated EAE mice, compared to normal controls, versus losses of 12.3 ± 1.6% in EAE/glib-pid-24 mice.

Lumbar spinal cords from normal controls (no EAE), untreated EAE mice, and EAE mice treated with glibenclamide beginning on pid-24 were examined by H&E staining (Fig. 3). Normal controls showed no inflammatory infiltrates, whereas meningeal, perivascular, and parenchymal inflammatory infiltrates were observed in the white matter of all untreated EAE spinal cords. Foci of inflammation were significantly reduced in the spinal cords of EAE/glib-pid-24 mice: 41.5 ± 1.5% of the quadrants in the untreated EAE mice were positive for inflammatory infiltrates, compared with 21.2 ± 2.4% of the quadrants in the EAE/glib-pid-24 group (Fig. 3).

To characterize the inflammatory cells present, sections of lumbar spinal cords were immunolabeled for markers of leukocytes (CD45), B cells (CD20), and T cells (CD3). Inflammatory lesions in untreated EAE lumbar spinal cords contained significantly increased numbers of CD45+, CD20+, and CD3+ cells, compared to normal controls (Fig. 3). The number of these cells was significantly reduced in EAE/glib-pid-24 mice (Fig. 3).

To further characterize the effect of delayed glibenclamide administration on immune modulation in vivo, we counted cells positive for the nuclear factor-κB (NF-κB) subunit, p65. Cells positive for p65 were significantly increased in untreated EAE mice compared to normal controls (Fig. 4), whereas this increase was significantly blunted in EAE/glib-pid-24 mice (Fig. 4).

Similarly, reactive astrocytosis evaluated using GFAP immunoreactivity was markedly increased in untreated EAE mice compared to normal controls (Fig. 4), and this increase was significantly blunted in EAE/glib-pid-24 mice (Fig. 4).

We examined macrophage/microglial polarization using the M1 marker, CD86 [24], and the M2 marker, CD163 [24] (Fig. 4). Cells with prominent CD86 were significantly increased in untreated EAE mice compared to normal controls, with a similar increase in EAE/glib-pid-24 mice (Fig. 4). CD163 also was increased in untreated EAE mice, but was increased to a greater extent in EAE/glib-pid-24 mice (Fig. 4).

One possible explanation for the reduced inflammatory burden with glibenclamide could be that the drug acts to reduce the number of circulating peripheral immune cells, as is found with fingolimod, which acts in part by preventing lymphocyte egress from lymph nodes [25, 26]. We tested this hypothesis by treating normal (non-EAE) mice with either fingolimod or glibenclamide for 3 days, following which we used FACS to examine peripheral blood for two major classes of effector T cells, CD45 + CD4+ (helper), and CD45 + CD8+ (cytotoxic) cells. As expected, total CD45+ leukocytes as well as CD45 + CD4+ and CD45 + CD8+ T-lymphocytes were sharply reduced with fingolimod [19, 27] (Fig. 5a, b). By contrast, no changes in these cell counts were observed with glibenclamide treatment (Fig. 5a, b).

The distribution of lymphocyte subsets in the spleen is different from that in peripheral blood, with more CD8+ cytotoxic T cells found in the spleen [28]. We examined the effect of fingolimod and glibenclamide on the two major classes of T cells in the spleen, CD3 + CD4+, and CD3 + CD8+ cells. As with peripheral blood, reductions in cell numbers were observed with fingolimod [25], but not with glibenclamide, although the ratios of CD3 + CD4+ to CD3 + CD8+ cells [29] remained similar in the three groups (Fig. 5c).

Since SUR1-TRPM4 is expressed predominantly by astrocytes during the chronic phase of EAE [13], another possible explanation for the reduced inflammatory burden with glibenclamide could be that the drug acts on reactive astrocytes to reduce their expression of pro-inflammatory cytokines, chemokines, and neurotoxic factors. Tissues from lumbar spinal cords were double immunolabeled for GFAP and the cytokine, TNF [1, 30, 31], the B-cell immunostimulant, BAFF [9, 32‐34], the leukocyte chemokine, CCL2 [35‐37], or for a principal source of nitric oxide, NOS2 [8, 38, 39], all of which have been implicated in astrocytes in EAE or MS. In each case, double labeling of tissues from untreated EAE mice showed prominent expression of the four pro-inflammatory agents in astrocytes (Fig. 6a–e). By comparison, tissues from EAE/glib-pid-24 mice showed significantly less expression of the four pro-inflammatory agents by astrocytes (Fig. 6a–e).

We also examined the expression of TNF, BAFF, CCL2, and NOS2 in sections from the lumbar spinal cords of Abcc8−/− mice subjected to EAE; the favorable clinical course of these animals was reported previously [13]. Similar to findings with glibenclamide, gene suppression of Abcc8 also was associated with significantly less TNF, BAFF, CCL2, and NOS2, compared to untreated WT EAE mice (Fig. 6e). These finding are consistent with glibenclamide acting via SUR1, not via an off-target mechanism.

The foregoing findings suggested that glibenclamide was acting centrally. Normally, however, glibenclamide exhibits poor CNS penetration [40]. In EAE lesions, blood-brain and blood-spinal cord barriers are known to be compromised, as evidenced by extravasation of serum proteins such as albumin [41, 42]. Since glibenclamide in the circulation is highly protein bound (> 99%) [43, 44], we reasoned that extravasated serum proteins would transport glibenclamide into EAE lesions. To test this, we administered the fluorescent derivative of glibenclamide, bodipy-glibenclamide, to mice with clinically symptomatic EAE, and we assessed lumbar spinal cord lesions for albumin and for the fluorescent bodipy label.

As expected [41, 42], immunolabeling showed abundant albumin extravasation within demyelinating lesions, compared to non-lesion tissues (Fig. 7a, b). In addition, demyelinating lesions showed prominent fluorescent labeling due to bodipy-glibenclamide, which was absent from non-lesion tissues in EAE mice as well as from control normal mice (Fig. 7c–f). In addition, non-demyelinating lesions, which were identified by immune cell infiltrates, also showed prominent fluorescent labeling due to bodipy-glibenclamide (Fig. 7g). These data indicate that glibenclamide can reach central targets when blood-brain and blood-spinal cord barriers are compromised by inflammation.

In CNS ischemia, SUR1 is highly expressed in microvascular endothelium, where it accounts for protein extravasation (vasogenic edema) [45]. This prompted us to question whether SUR1 expression in microvessels could similarly be linked to protein extravasation in EAE lesions. However, in contrast to ischemia, in EAE, SUR1 expression was found to be minimal or absent in microvascular endothelium, whereas it was very prominent in perivascular structures consistent with astrocytes and astrocytic foot processes (Fig. 7h).

The foregoing experiments suggested that SUR1-TRPM4 channels in astrocytes might be the target of glibenclamide and might be involved in the secretion of pro-inflammatory mediators by astrocytes. To explore this, we modeled the astrocyte response in MS/EAE by exposing primary astrocyte cultures to TNF plus IFNγ overnight [1]. Cell lysates analyzed by qPCR showed upregulation of mRNA for Abcc8 and Trpm4 (Fig. 8a), which encode the SUR1 and TRPM4 subunits, respectively.

Patch clamp electrophysiology was used to determine whether functional SUR1-TRPM4 channels were being upregulated by TNF + IFNγ-mediated astrocyte activation. For these experiments, intracellular and extracellular solutions contained Cs⁺ as the principal monovalent cation, because Cs⁺ blocks all K⁺ channels, including K_ATP channels, but it permeates non-selective cation channels such as SUR1-TRPM4. Channel currents were induced using diazoxide, which activates SUR1-regulated channels exclusively. Control non-activated astrocytes showed no response to diazoxide, indicating that quiescent astrocytes do not express SUR1-regulated cationic channels (Fig. 8b, c). By contrast, in activated astrocytes, diazoxide elicited large inward currents at physiological potentials that were blocked by the SUR1 or TRPM4 inhibitors, glibenclamide or 9-phenanthrol, respectively (Fig. 8b, d), consistent with the expression of functional SUR1-TRPM4 channels [12].

Exposure of astrocytes to TNF + IFNγ induced not only the expression of SUR1-TRPM4 channels but also upregulated mRNA for Tnf, Baff, Ccl2, and Nos2, with mRNA abundance being many-fold greater than in control astrocytes not activated by TNF + IFNγ [32] (Fig. 8e). Incubation with TNF + IFNγ in the presence of glibenclamide resulted in significant attenuation of mRNAs for Tnf, Baff, Ccl2, and Nos2 (Fig. 8e), replicating in vitro some of our in vivo observations.

To explore the potential relevance of the aforementioned findings to MS, post-mortem tissues from the cerebral hemispheres of patients with MS and controls were assessed for SUR1-TRPM4 expression. Tissues were analyzed by histopathologic methods to identify areas of inflammation and demyelination. Demyelinating white matter lesions were categorized using CD68, CD45, and CD3 immunolabeling as per De Groot et al. [16], as active, chronic active, or chronic inactive. We also analyzed subpial/cortical areas of demyelination [46, 47].

We examined the expression of SUR1 in these various tissues. As illustrated in Fig. 9a, control white matter showed no specific labeling for SUR1, active white matter lesions showed distinct SUR1 labeling of stellate, astrocyte-like cells (Fig. 9b), and chronic active white matter lesions exhibited intense, widespread labeling of cells and processes. SUR1 immunolabeling also was prominent in subpial/cortical lesions, whereas control cortex showed only faint labeling in scattered large neuron-like cells (Fig. 9c). Semi-quantitative evaluation showed that different lesion types exhibited different levels of SUR1 expression, with chronic active lesions having the most intense labeling (Fig. 9d).

Double immunolabeling was carried out to identify the cell type (s) involved. Co-labeling for S100B showed that, in white matter lesions, SUR1 was expressed predominantly by astrocytes (Fig. 10a). Double labeling for SUR1 and CD68/ED1 (microglia/macrophage) or CD3 (T cells) showed minimal or no overlap with SUR1 expression (not shown). Stellate cells expressed SUR1 not only in the cell body but also in processes—in some chronic active white matter lesions, perivascular astrocytes could be seen to extend processes ending in perivascular end-feet that immunolabeled strongly for SUR1 (Fig. 10b, c). In subpial/cortical lesions as well, SUR1 expression was found predominantly in GFAP+ astrocytes, including the glia limitans (Fig. 10d, e).

Double immunolabeling showed co-expression of SUR1 and TRPM4 in stellate cells (Fig. 11a), as well as in perivascular end-feet (Fig. 11b). ImmunoFRET showed that SUR1 and TRPM4 physically co-assembled to form heterodimers (Fig. 11c), consistent with assembly of functional SUR1-TRPM4 channels [12].

In MS white matter lesions, astrocytes were previously reported to upregulate BAFF [32], CCL2 [35], and NOS2 [38, 39]. Co-immunolabeling for SUR1 and BAFF, CCL2 or NOS2 confirmed that astrocytes in MS lesions that express SUR1-TRPM4 channels also express these pathogenic molecules (Fig. 12). High values of Pearson’s correlation coefficient for each one (0.58–0.84) reflect a high degree of co-localization with SUR1.