Background
Globally, ischemic stroke is one of the leading causes of death and long-term disability [
1]. The number of patients with cardioembolic ischemic stroke resulting from nonvalvular atrial fibrillation (AF), the major cause of cardioembolic ischemic stroke, is predicted to double by 2030 [
2,
3]. Consequently, a growing burden of warfarin-associated hemorrhagic transformation (HT) after cerebral ischemia can be expected [
4‐
6].
Early HT can occur as a complication of cardioembolic ischemic stroke [
7]. Additionally, a higher rate of hematoma expansion and a worse clinical outcome have been reported in warfarin-associated HT patients [
8‐
10]. However, no effective treatment strategy is available for prevention of HT in clinical practice. Experimental studies of cerebral ischemia have established increase in the permeability of the blood-brain barrier (BBB) after ischemia/reperfusion injury as one of the major causes of HT [
11,
12].
The glucagon-like peptide-1 receptor (GLP-1R) agonist exendin-4 (Ex-4) is a long-acting analog of the endogenous insulinotropic peptide GLP-1. Both GLP-1 and Ex-4 have multiple physiologic functions, such as the induction of glucose-dependent insulin release, inhibition of glucagon secretion, stimulation of B cell replication, and antiapoptotic action [
13]. Owing to their small molecule size, both GLP-1 and Ex-4 can diffuse across the BBB in the central nervous system and provide neuroprotection in cerebral ischemia [
14,
15]. While it has been reported that Ex-4 can protect against oxidative products and neuronal cell death caused by ischemic brain damage, it is yet unknown whether Ex-4 is effective in preventing warfarin-associated HT after cerebral ischemia.
Previous studies have shown that after a hemorrhagic stroke, cytotoxic events activate the ubiquitously expressed glycogen synthase kinase-3β (GSK-3β), which increases the expression of β-catenin [
16,
17] and subsequently decreases the expressions of claudins [
18]. There is substantial evidence that GSK-3β inhibition (tyrosine-216 dephosphorylation) reduces neuronal apoptosis [
19‐
21] and attenuates neuroinflammation in neurodegenerative models [
22‐
24]. Pharmacological stimulation of GLP-1R activates the phosphatidylinositol 3-kinase (PI3K)-Akt signaling pathway, and a number of studies have linked GSK-3β with the PI3K/Akt pathway, thereby showing that phosphorylated Akt inactivates GSK-3β via tyrosine-216 dephosphorylation. Herein, we hypothesized that Ex-4 would stabilize the BBB and suppress neuroinflammation through PI3K-Akt-induced inhibition of GSK-3β after warfarin-associated HT post-cerebral ischemia in mice.
Methods
Animals
All experiments were conducted using male C57BL/6 mice (body weight 18–25 g) at a constant temperature and with a consistent light cycle (from 07:00 to 18:00) under normal diet. This study was carried out in accordance with the Guide for the National Science Council of the Republic of China. All animals were treated according to protocols approved by the Institutional Animal Care and Use Committee of Fudan University.
A 5-mg warfarin sodium tablet (Coumadin™, Sigma-Aldrich, St. Louis, MO, USA) was dissolved in 375 mL of water. The C57 BL/6 mice were fed for 0, 6, 12, and 24 hours with ad libitum access to the treated water. Assuming a mouse body weight of 20 g and a water consumption rate of 15 mL/100 g per 24 h, this dosage corresponds to a warfarin uptake of 0.04 mg (2 mg/kg) per mouse over a 24-h period. Similar doses of warfarin have been previously used [
25]. After 24 h, the warfarin was withdrawn and middle cerebral artery occlusion was performed (Additional file
1: Figure S1). For the international normalized ratio (INR) measurement, the mice were under deep anesthesia, a peritoneal midline incision was performed, and 0.6 mL blood was drawn from the inferior caval vein as previously described [
26]. Blood was transferred to glass tubes (BD Vacutainer
®) containing sodium citrate as the anticoagulant. Measurements of INR values and prothrombin time were performed in the Department of Central Laboratory, Jingan District Centre Hospital, Shanghai, China.
Temporary middle cerebral artery occlusion and drug treatment
Mice were anesthetized with ketamine/xylazine (65/6 mg/kg, i.p), and their body temperature was maintained at 37 °C by a heating pad and feedback control system (FHC, Bowdoin, ME, USA). A laser Doppler probe was fixed on the skull 5 mm lateral and 2 mm posterior to the bregma. A coated filament was placed on the right middle cerebral artery (MCA) with concurrent recording of laser Doppler cerebral blood flow to ensure that the cerebral blood flow decreased to below 25 % of the baseline. After 45 min, the filament was removed (Additional file
2: Figure S2). Either Ex-4 (10 mg/kg) or saline was injected through the tail vein immediately after reperfusion. In the Ex-4 + wortmannin group, we intravenously injected 15 μL/kg wortmannin (Sigma-Aldrich), a non-specific, covalent inhibitor of PI3K immediately after reperfusion.
Assessment of infarct volume, neurological deficits, and blood-brain barrier
All the mice were killed 72 h after temporary middle cerebral artery occlusion (MCAO), and brain tissues were incubated in 2,3,5-triphenyltetrazolium chloride (TTC) for 1 h. The infarct area in each slice was analyzed by a computerized image analysis system, and the infarct volume was calculated by multiplying the distance between sections [
27]. Neurological score was determined 72 h after MCAO, according to the graded scoring system described previously by Li et al. [
28]. Assessment of motor coordination deficits was performed on days 3 and 7 using the rota rod as previously described [
29]. Investigators who performed MCAO models, evaluation of infarct volumes, neurological scales, and rota rod were blinded to all the experimental protocols and drug treatments. To measure BBB permeability, Evans blue (Sigma-Aldrich) was dissolved in saline (2 %) and injected into the right jugular vein 72 h after MCAO. The animals were then killed, and the brain hemispheres were homogenized in 3 mL of
N,
N-dimethylformamide (Sigma-Aldrich); incubated for 18 h at 55 °C; and centrifuged. The supernatants were subjected to spectrophotometry at 620 nm.
Quantification of hemorrhagic transformation
The hemoglobin content in brain tissue was quantified by spectrophotometric assay. The hemispheric brain tissue was homogenized with phosphate-buffered saline (PBS) and centrifuged at 13,000×g for 30 min. The hemoglobin-containing supernatant was collected, 80 μL of Drabkin reagent (Sigma) was added to 20-μL supernatant aliquots, and the sample was kept standing for 15 min at room temperature. The optical density in each group was measured at 540 nm, and hemorrhage volume was expressed in equivalent units by comparison with a reference curve generated using homologous blood.
Western blotting
Striatal brain tissues from the MCA were lysed with radioimmunoprecipitation assay buffer (RIPA) containing protease inhibitors (Sigma-Aldrich, St. Louis, MO, USA). Proteins were separated by SDS-PAGE and then transferred onto a nitrocellulose membrane. The membranes were incubated overnight at 4 °C with the following primary antibodies: anti-p-GSK-3β (Tyr216, 1:1000, Abcam Inc., Cambridge, MA); anti-GSK-3β (1:1000, Abcam); anti-β-actin (1:5000, Sigma-Aldrich); anti-p-β-catenin (Ser33/37/Thr41, 1:2000, Cell Signaling Technology Inc., Danvers, MA); anti-β-catenin (1:1000, Abcam), anti-claudin-3 (1:2000, Santa Cruz, CA); anti-claudin-5 (1:2000, Santa Cruz); anti-p-Akt (Ser473, 1:2000, Cell Signaling); anti-Akt (1:2000, Cell Signaling); anti-ICAM-1 (1:1000, Abcam); anti-VCAM-1 (1:1000, Abcam); anti-IKK-β (1:2000, Santa Cruz); anti-NF-kB (1:2000, Santa Cruz); anti-HHE (1:1,000, Abcam); anti-Iba1 (1:1,000, Abcam); and anti-myeloperoxidase (MPO) (1:2000, Santa Cruz). Secondary antibodies conjugated with horseradish peroxidase were used, and immunoreactivity was visualized by chemiluminescence (SuperSignal Ultra, Pierce, Rockford, IL, USA). Bands of interest were analyzed and quantified using Scion Image.
The small interfering RNA (siRNA)-mediated GSK-3β gene knockdown was performed as previously described [
30]. Briefly, two pairs of GSK-3β siRNAs (21500 R12-1717, R12-1719; Cell Signaling) with a total volume of 4 μL (2 μL each) were stereotaxically injected to the right lateral ventricle following coordinates relative to the bregma: AP = −0.4 mm, L = −1.0 mm, and H = − 2.0 mm (from the brain surface) 48 h prior to MCAO.
Measurement of cytokine concentration
Striatal brain tissues from the MCA were homogenized and collected by centrifugation at 15,000×
g for 30 min at 4 °C and then stored at −70 °C until the assay was performed. The supernatant was assayed for tumor necrosis factor-α (TNF-α) and interleukin-1 beta (IL-1β) using enzyme-linked immunosorbent assays (ELISA; R&D Biosystems) as described previously [
31].
Concentration of 8-hydroxy-2′-deoxyguanosine (8-OHdG) in brain DNA was measured by Piao et al.’s method [
32], with slight modifications. Briefly, 200 mg of brain tissue was homogenized in 0.25 M sucrose solution. DNA was extracted from the homogenate under anaerobic conditions. The 8-OHdG content in the brain was measured by using an HPLC-ECD as previously described [
33]. Each brain sample was examined in duplicate.
Immunohistochemistry
Seventy-two hours after MCAO, the mice were anesthetized and first perfused with saline followed by fixation with buffered paraformaldehyde (4 %). The brains were removed and post-fixed in 4 % paraformaldehyde; the paraformaldehyde was then removed and replaced with 30 % sucrose solution overnight. Then 15-μm coronal sections were obtained on a cryostat. The slices were blocked with PBS containing 5 % bovine serum albumin (BSA), 10 % goat serum, and 0.3 % Triton-X 100. Next, the slices were incubated with the primary antibodies anti-Iba1 (1:250, Abcam) and anti-TNF-α (1:100, Santa Cruz) overnight at 4 °C. Then Alexa Fluor 488 or 595 labeled secondary antibody (Molecular Probes Inc., Eugene, OR, USA) for 2 h at room temperature. The tissue sections were washed twice in PBS and then immersed in DAPI (Molecular Probes) solution (1:1000 dilution) for 10 min. The sections were finally rinsed in distilled water and fixed with a coverslip with anti-fade mounting medium.
Assessment of microglia activation
First, microglia activation were counted and morphologically characterized based on the following criteria. Cells with an oval cell body containing a small volume of cytoplasm and long, thin, delicate, and radially branched processes were classified as ramified microglia [
34]. Activated microglia were defined as having an enlarged soma (width greater or equal to 30 μm) and a broad-flattened appearance with the common presence of several lamellapodia [
35]. This morphological classification was then confirmed by using a methodology of semi-automatic image analysis to analyze the cell body to cell size ratio in Iba1-stained brain sections as described before [
36] by ImageJ software.
Statistical analysis
All values are expressed as mean ± standard deviation (SD). Differences between means were analyzed using either one-way or two-way ANOVA followed by Newman–Keuls post hoc testing for pair-wise comparison using SigmaStat v 3.5®. A P value <0.05 was considered statistically significant.
Discussion
Atrial fibrillation is a severe independent risk factor of stroke, its attributable risk increasing with age up to more than 20 % [
47]. INR-driven oral anticoagulation with vitamin K antagonists to an INR of 2–3 reduces the risk of an ischemic stroke by over 60 % and has been the standard of stroke prevention in patients with AF [
48]. However, anticoagulation therapy is closely related to HT after ischemia. In addition, cardioembolic stroke also carries with it an increased risk of HT [
49]. The chief mechanism of HT is considered to be blood leakage due to disruption of the BBB. Our results showed that pretreatment with warfarin could significantly increase the INR level in a time-dependent manner and dramatically enhance Evans blue leakage provoked by MCAO. Although the infarct volume and neurological deficits were not significantly different between the groups with or without warfarin treatment, warfarin significantly promoted the HT after cerebral ischemia, which is consistent with the permeability measurement results.
GLP-1 and long-acting Ex-4 induce numerous biological actions through the G protein-coupled GLP-1 receptor (GLP-1R). GLP-1R is reportedly expressed in a wide range of tissues, including the brain. Moreover, GLP-1R stimulation has shown neuroprotective actions in previous findings, thereby establishing that GLP-1R stimulation protects hippocampal neurons from amyloid-β peptide and glutamate-induced toxicity [
50,
51]. As the GLP-1R agonist Ex-4 is permeable to the BBB with a relatively long half time, it has possible clinical applications. Several studies have shown that Ex-4 can protect against oxidative products and neuronal cell death caused by ischemic brain damage [
15]. However, to the best of our knowledge, whether GLP-1R stimulation is associated with warfarin-associated HT has not yet been studied. Herein, we reported that Ex-4 prevented the exacerbation of HT caused by warfarin without affecting the infarct volume. The mechanism whereby Ex-4 prevented the exacerbation of HT might involve maintenance of the expression of tight junction proteins and suppress the neuroinflammation associated with warfarin treatment. The pathways that strengthen the antiapoptotic and neuroprotective effects of Ex-4 after cerebral ischemia mostly converge on activation of the transcription factor cAMP response element-binding protein (CREB) by phosphorylation. In the present study, the PI3K/Akt-GSK-3β signaling pathway appeared to contribute to the protection afforded by Ex-4 in the warfarin-associated HT model.
PI3K/Akt plays a crucial role in the cell death/survival pathway through several different downstream targets including GSK-3β [
52]. A temporal increase in phospho-Akt after cerebral ischemia has been reported, and GSK-3β dephosphorylation at tyrosine-216 is accelerated as a downstream target of Akt [
53]. The inactivation of GSK-3β via tyrosine-216 dephosphorylation mediates neuronal survival after cerebral ischemia [
43]. In addition, the inactivation of GSK-3β results in stabilization of β-catenin, a protein that plays a role in cell adhesion. As a result, free β-catenin is allowed to accumulate and be translocated to the nucleus, binding to the transcription factors to alter target gene expressions [
54], such as those of tight junction proteins claudin-3 and claudin-5 [
18,
39]. Furthermore, GSK-3β inactivation may also decrease NF-kB expression, thereby reducing neuroinflammation.
In this study, Akt phosphorylation at Ser473 and GSK-3β dephosphorylation at tyr216 were increased in warfarin-associated HT after cerebral ischemia. Administration of Ex-4 substantially decreased HT and maintained the stability of BBB. The reduced dye extravasation and brain hemoglobin level were similar to that achieved by inhibition of GSK-3β. Evidence supporting enhanced BBB stabilization by Ex-4 including decreased adherens (VCAM-1 and ICAM-1) and increased tight junction (claudin-3 and claudin-5) proteins could be totally abolished by wortmannin, a specific PI3K inhibitor. These results suggest that warfarin-associated HT reduced the expression of tight junction proteins. This effect was prevented by treatment with Ex-4 through the PI3K/Akt-GSK-3β pathway. Furthermore, Ex-4 reduced the warfarin-induced hemorrhage volume via a protective effect on vascular endothelial cells.
Inflammation has been recognized as a key contributor to the pathophysiology of cerebral ischemia [
55]. Inflammation includes a series of cellular events such as infiltration of neutrophil cells and activation of microglia/macrophages and astrocytes [
56]. We found that warfarin-associated HT significantly upregulated Iba1-positive cells. Microglia/macrophage activation, together with elevated expression of pro-inflammatory cytokines such as IKK-β, NF-kB, TNF-α, and IL-1β, demonstrated that the warfarin-associated HT induced a neuroinflammation after cerebral ischemia. It has also been reported that activated microglia/macrophages are major sources of metalloproteinase generation, which is closely associated with ischemia-induced cerebral hemorrhage and edema. NF-kB is a central mediator of these inflammatory processes. Recent evidence has shown that the PI3K/Akt signaling pathway may be an endogenous negative feedback regulator of NF-kB-mediated pro-inflammatory responses [
57,
58]. Several pro-inflammatory NF-kB target genes including TNF-α and IL-1β could mediate the deleterious effects on neurons under ischemic conditions. In the present study, we showed that warfarin-induced HT markedly induced the activation of microglia/macrophages and consequently increased the production of pro-inflammatory cytokines and Ex-4 significantly inhibited the neuroinflammation induced by warfarin through the PI3K/Akt-GSK-3β pathway. Moreover, suppression of oxidative damage is also a key factor in neuroprotection. Using 8-OHdG and HHE as markers of oxidative stress, our study showed that Ex-4 reduced the warfarin-induced accumulation of oxidative DNA damage and lipid peroxidation after cerebral ischemia.
Acknowledgements
We are grateful to Baoguo Xiao for his technical support and Min Guo for assisting in preparing this manuscript.