Hyperglycemia primes NETosis, which exacerbates ischemic brain damage

BackgroundHyperglycemia is common and associated with poor outcomes in acute ischemic stroke patients. It is not well understood how hyperglycemia exacerbates brain damage in ischemic stroke. Neutrophil extracellular traps (NETs) have shown an emerging role in noninfectious diseases. We aimed to determine the role of NETs in acute ischemic stroke with hyperglycemia. Methods: NETs were immunostained using NET markers (citrullinated histone H3 (H3Cit)) and quantified in thrombi retrieved from ischemic stroke patients undergoing endovascular treatment. BKS-db/db and wild-type mice were used to establish the permanent middle cerebral artery occlusion (pMCAO) model. Wild-type mice were injected with glucose to simulate acute hyperglycemia after middle cerebral artery occlusion. NETs were detected in the peri-ischemic brain tissue. After inhibition of NET formation, infarction volume, neurological function and inflammatory factors in pMCAO mice were evaluated. Results: H3Cit, a marker of NETs, was observed in almost all thrombi. H3Cit was much more abundant in thrombi from diagnosed diabetic patients and acute hyperglycemic patients compared with those in normglycemic patients. In pMCAO mice, NETs were induced by chronic diabetes and acute hyperglycemia. Inhibition of NET formation with the peptidylarginine deiminase 4 (PAD4) inhibitor Cl-amidine decreased the infarction volume both in db/db and wild-type mice with hyperglycemia. Neurological function deficits were alleviated by blocking NET formation, as shown in the grip strength and rotarod tests. The levels of TNF-α and IL-1β but not IL-6 coincided with NET formation. Conclusions: Hyperglycemia may exacerbate brain damage in ischemic stroke

4 through NETs. The underlying mechanisms deserve to be further studied.

Background
Stroke is the leading cause of disability and mortality worldwide. Hyperglycemia arises in nearly 40% of people with acute ischemic stroke [1]. Most of these individuals have diagnosed or pre-existing but unrecognized diabetes mellitus, and some nondiabetic patients may have acute stress hyperglycemia [2]. Compared to the risks in stroke patients with normoglycemia, hyperglycemic stroke patients have a much higher risk of short-term mortality and poor functional recovery, especially in nondiabetic stroke patients [3]. Additionally, many studies have shown that hyperglycemia is associated with symptomatic hemorrhagic transformation, infarct volume growth and worse clinical outcomes among patients treated with intravenous recombinant human tissue plasminogen activator (rtPA) and mechanical thrombectomy [4][5][6][7][8][9]. These observational studies consistently found an association between hyperglycemia and worse outcomes in acute ischemic stroke, but it has not been determined whether this is a cause-and-effect relationship. Blood glucose lowering treatments have failed to benefit ischemic stroke patients until now [10].
Mechanisms by which hyperglycemia exacerbates ischemic brain damage need to be explored. Recently, the emerging role of neutrophil extracellular traps (NETs) has been observed in many kinds of noninfectious diseases [11], including atherosclerosis and atherothrombosis [12]. Early reports demonstrated that NET formation is strictly dependent on glucose and is increased in diabetes [13][14][15].
Hyperglycemia alone in the presence or absence of diabetes in vivo and high glucose concentration in vitro increase NETosis [16]. In addition, levels of the NET markers citrullinated H3 (citH3), cell-free DNA (cfDNA), and nucleosomes were elevated in the plasma of stroke patients [17]. NETs were also found in ischemic stroke thrombi retrieved from patient occluded arteries [18] and the plasma of ischemic stroke patients, which may constitute a useful prognostic marker in patients with acute stroke [17].
In this study, we aimed to offer solid experimental evidence of hyperglycemiamediated exacerbation of ischemic brain damage that is mediated by NETs.

Patient thrombi collection and processing
Thrombi were collected from acute ischemic stroke patients after mechanical thrombectomy that was performed at Shanghai Jiaotong University Affiliated Sixth People's Hospital. All patients gave written informed consent under the approval of the Ethics Committee. Clinical patient data were collected, such as random glucose at admission and fasting and postprandial blood glucose the day after admission.

Animals
Male BKS-db/db mice (on a C57BL/JN background), aged 4 months old and weighing 40 to 45 g, and wild-type (WT) C57BL/JN mice, weighing 20-22 g, were purchased from the Model Animal Research Center of Nanjing University in China. The mice were housed under environmentally controlled conditions with a 12-h light-dark cycle and a standard diet and water. All animals experimental procedures were approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong 6 University.

Permanent middle cerebral artery occlusion (pMCAO)
The mice were anesthetized with intraperitoneal injection of sodium pentobarbital (75 mg/kg) (Sigma). A skin incision was made between the right eye and the right ear, and the temporal muscle was removed to perform the craniotomy. After resecting the dura mater, the distal portion of the middle cerebral artery (MCA) was exposed and permanently electrocoagulated. The sham operation only exposed the MCA, but the electrocoagulation was not performed. Acute stress hyperglycemia was induced with intraperitoneal injection of 0.2 ml 50% glucose immediately after electrocoagulation, followed by repeated injections every four hours until sacrifice 24 hours later. Mice were intraperitoneally injected with Cl-Amidine (Cayman Chemical, Michigan) half an hour before electrocoagulation of the MCA. Blood glucose was detected in the tail vein using the analyzer (Performa, Basel, Switzerland) before the operation, immediately after the MCA electrocoagulation and at the indicated times after the operation.

Measurement of infarction volume
The mice were deeply anesthetized and sacrificed at the indicated time points. The brains were rapidly frozen at -20°C for 10 min and then sectioned into 2 mm coronary slices and stained with 2,3,5-triphenyltetrazolium chloride. Images were taken, and the infarction area was calculated by using ImageJ (version 1.47t; National Institutes of Health, Bethesda, MD; http://imagej.nih.gov/ij/). The percentage of infarction volume was calculated by the following equation: Infarction volume (%)=Infarction volume/Total brain volume*100%.

Immunofluorescence and immunohistochemical staining
The mice were perfused transcardially with pre-cooled 4% paraformaldehyde (pH 7.4; Sigma) under deep anesthesia. The brains of the mice and the thrombi of stroke patients were embedded in paraffin and cut into 5 µm sections. For immunofluorescence staining, the sections were immersed in 0.2% Triton X-100 for permeabilization and in 10% goat serum for blocking and were then incubated with primary antibodies against Ly6G (1:400, BD Biosciences) and H3Cit (1:400, Abcam) overnight at 4°C and subsequently two secondary antibodies (anti-rat IgG coupled to FITC 488,1:1000; anti-rabbit IgG coupled to DyLight 594, 1:1000 (Jackson Immuno Research)) for 1 hour at room temperature. Nuclei were stained with DAPI. The slices were then mounted with coverslips using antifade reagent (Molecular Probes, Invitrogen) and sealed with nail polish. Images were acquired with a laser confocal microscope (SP8; Leica Microsystems, Wetzlar, Germany). For immunohistochemical staining, the sections were rinsed and the endogenous peroxidase activity was blocked with 3% H 2 O 2 . Then, the sections were blocked with 5% goat serum and incubated with diluted primary antibodies against H3Cit (1:200, Abcam) overnight at 4°C. A biotinylated secondary anti-rabbit IgG antibody (1:1000, Roche) was added for 30 min, followed by the addition of Sav-HRP conjugates and incubation at room temperature for another 30 min. Diaminobenzidine (DAB) was applied to reveal the color of antibody staining. The stained slides were imaged using a light microscope (SP8;Leica Microsystems) and CaseViewer software. ImageJ software was used to analyze the H3Cit content. Five 20× fields of view per section stained for H3Cit were chosen at random and used to calculate the average amount of H3Cit in each thrombus. Classical H&E staining was conducted based on a routine procedure.

Western blotting
Peri-infarction brain tissues were lysed in the ice-cold radio immunoprecipitation assay buffer (Thermo Fisher Scientific) with protease and phosphatase inhibitors.
8 Protein extracts were separated by SDS-PAGE. The proteins were transferred to PVDF membranes. After blocking with 5% skimmed milk, the membranes were incubated with the diluted primary antibodies against H3Cit (Abcam). GAPDH was used as the internal standard. After washing with TBST buffer, the membranes were incubated with secondary antibodies (IRDye-700 nm anti rabbit and IRDye-800 nm anti mouse) (Roche) and photographed using an Odyssey imaging system (LI-COR Bioscience). The measurement of band immunofluorescence intensity was determined by Quantity One (Bio-Rad).

Evaluation of neurological deficits
A rotarod test was performed (Columbus Instruments, America) for four consecutive days before pMCAO for learning and at 24 hours after pMCAO. The measurement of the last day before pMCAO was defined as the preoperation test. A grip strength test (Bio-ACTIVW-M, Bioseb, France) was used to evaluate the motor function on the day before and at 24 hours after pMCAO.

Statistical analysis
The results are presented as the mean ± standard deviation (SD). Data were analyzed using GraphPad Prism 5 (GraphPad Software, San Diego, CA) and and Bonferroni's multiple comparison tests were used to confirm significant differences. A p < 0.05 was considered significant.

NETs were induced by diabetes and hyperglycemia in the thrombi of stroke patients
We collected thrombi from 41 ischemic stroke patients after the thrombectomy procedure, of which 9 were confirmed as having diabetes, 7 were confirmed as having acute hyperglycemic non-diabetes and 25 were normoglycemic patients. H&E staining showed an abundance of granulocytes in the thrombi ( Figure 1A, B, C).
Nuclear material that appeared to be extracellular was found in the magnified H&E pictures ( Figure 1D, E, F). Staining for H3Cit, a defining marker of NETs, confirmed the formation of NETs in these thrombi ( Figure 1G, H, I). The content of H3Cit was greatly increased in the area that had abundant granulocytes, as shown in representative images from the three different groups of patients ( Figure 1J, K, L).
To further confirm the presence of NETs in the thrombi, immunofluorescent staining was performed combining H3Cit with DNA dye (DAPI). From the representative patient images, we clearly showed that H3Cit accompanied DNA that was released from the nucleus (Figure 2).

Blood glucose measurements of the pMCAO mice
Blood glucose measurements from the tail vein are shown in figure 3. Animals had free access to food and water. The basal level of blood glucose in the db/db mice exceeded the upper limit (35 mmol/L) of the analyzer and was 9.68±0.54 mmol/L in WT mice before the operation. Immediately after the operation, there was a transient but significant increase in blood glucose in the WT (20.10±3.57 mmol/L) and WT+HG mice (18.62±1.80 mmol/L). Two hours after the operation, significantly higher levels of blood glucose in WT+HG mice were maintained until the sacrifice compared to those of the WT mice (p<0.001). Although the blood glucose in db/db mice decreased spontaneously, the levels of blood glucose were much higher in the db/db mice than in the WT mice until 20 hours after the operation (p<0.05) ( Figure   3).

Neutrophils and NETs were increased in diabetic and acute hyperglycemic pMCAO mice
To detect neutrophil infiltration in ischemic brain tissue, the peri-infarction brain tissue was immunofluorescently stained for the neutrophil marker Ly6G. Compared to the levels in WT mice at 24 h after MCAO, the number of infiltrated neutrophils was much higher in the brains of db/db and WT+HG mice ( Figure 4A). Next, we evaluated H3Cit in the peri-infarction brain tissue. In the normoglycemic WT MCAO mice, some star dot signals of H3Cit appeared intracellularly, indicative of reduced and early NETosis. These results were completely different from the strong H3Cit string signals that appeared extracellularly in db/db MCAO and WT+HG MCAO mice, which are indicative of increased and complete NETosis ( Figure 4B). To further confirm the NETs in peri-infarction brain tissue, we used the PAD4 inhibitor Cl-Amidine to inhibit NETosis in MCAO mice. In db/db mice, compared to expression in the sham MCAO mice, H3Cit expression increased significantly (0.57±0.11 versus 0.29±0.08, p<0.05), which was inhibited when Cl-Amidine was administered (0.36±0.08 versus 0.57±0.11, p<0.05) ( Figure 4C). In the WT MCAO mice with normoglycemia, there was a significant increase in H3Cit protein expression compared to that of the sham MCAO mice (0.25±0.10 versus 0.09±0.01, p<0.05).
Moreover, the WT MCAO mice with hyperglycemia contained double the amount of H3Cit (0.54±0.12) compared to that of the WT MCAO mice with normoglycemia (0.25±0.10, p<0.05) ( Figure 4D). H3Cit expression was inhibited after injection of Cl-Amidine in both normoglycemic and hyperglycemic MCAO mice.

Inhibition of NETosis decreased the infarction in diabetic and hyperglycemic stroke animals
In db/db MCAO mice, the infarction volume was much larger than in the WT MCAO mice (30.64%±3.62% versus 22.34% ±2.39%, p<0.01). This difference in infarction volume between db/db MCAO mice and WT MCAO mice disappeared when Cl-amidine was injected intraperitoneally at a low dose of 30 mg/kg or a high dose of 60 mg/kg before the operation. Compared to the infarction volume in the db/db MCAO mice injected with vehicle, Cl-amidine at a dose of 60 mg/kg significantly decreased the infarction volume (30.64%±3.62% versus 21.76%±3.61%, p<0.05) ( Figure 5A). In WT MCAO mice, the increase in infarction volume was also shown in the WT mice with hyperglycemia induced by glucose injection compared to that of the WT mice with normoglycemia (36.38%±3.77% versus 22.34%±2.39%, p<0.01). This difference in infarction volume between WT MCAO mice and WT+HG MCAO mice diminished but remained significant when Cl-amidine was injected intraperitoneally.

Discussion
In this study, we showed that (1) hyperglycemia primed the formation of NETs in thrombi of stroke patients and peri-infarction brain tissue of pMCAO mice; (2) NETs play a role in thrombosis in diabetic ischemic stroke; (3) NETs exacerbate ischemic brain damage in diabetes; and (4) the inhibition of NETosis could eliminate the 14 harmful effect of hyperglycemia on ischemic brain injury.
Thrombus constitution provides hints on thrombus formation. Therefore, we analyzed the composition of thrombi in stroke patients. Previous studies showed that NETs contributed to coagulation and platelet aggregation, thus promoting thrombosis [19], such as deep vein thrombosis [20] and thrombi in acute myocardial infarction [21]. Although the NETs were verified in thrombi retrieved from large vessel occlusion stroke patients and were associated with a cardioembolic origin, the broad range of NET amounts among the different thrombi suggested that other factors may affect NETosis [18]. Early reports showed elevated levels of circulating NETs in type 2 diabetes [15]. Here, we confirmed an abundance of NET formation in thrombi of stroke patients that was associated with high blood glucose, either in diabetes or acute stress hyperglycemia. These results indicate that hyperglycemia primes NET formation in the thrombi of stroke patients. NETs play an important role in thrombosis in ischemic stroke.
Neutrophil granulocytes have been key players in cerebral ischemia [22]. The release of proteases associated with decondensed DNA triggers neurotoxicity [23].
Histones are released into the extracellular space, activating the immune system and causing further cytotoxicity [24,25]. Therefore, we hypothesize that NET formation accelerates and exacerbates injury, especially in hyperglycemia.
Hyperglycemia upon admission is attributed to existing chronic diabetes and acute stressful hyperglycemia [26]. To simulate the clinical scenario, we used db/db mice, WT mice and WT mice injected with glucose to establish pMCAO models for simulating the three kinds of ischemic stroke patients based on blood glucose. The results showed that induction of hyperglycemia was successful in db/db and WT+HG mice after pMCAO. There was a significant but transient rise in blood glucose immediately after the operation, which may be attributed to the stress of the operation. Blood glucose dropped gradually in WT mice and more obviously in db/db mice after the operation. Loss of appetite after the operation may explain this effect. In the pMCAO mice, we verified that NETs increased in ischemic brain tissue, which is in line with a recent report [27]. Moreover, the NET content in ischemic brain tissue was more highly induced in hyperglycemia. Then, as expected, we found that infarct volume was much larger in db/db and hyperglycemic pMCAO mice compared to those of the normoglycemic mice. This may be explained by the contribution of NETs to thrombus stabilization and growth [21] or the direct cytotoxicity of histones [24].
Since neutrophils and NETs orchestrate the initiation and modulation of inflammation [28,29], we evaluated the proinflammatory cytokines produced in pMCAO mice. TNF-α and IL-1β increased in db/db and hyperglycemic mice, which indicates a proinflammatory role of hyperglycemia. The downregulation of TNF-α and IL-1β after NET inhibition demonstrates that NETs drive inflammatory responses in pMCAO mice, as shown in other reports [30,31].
In this study, we employed a pMCAO model other than the tMCAO mouse model. The infarction area was more uniform in individuals with pMCAO than in those with tMCAO. Although the most effective treatment for ischemic stroke is rapid recanalization, the proportion of patients who receive thrombolysis or mechanical thrombectomy is still very low. Most patients suffer permanent ischemia.
Neutrophils could be transported to the brain by reperfusion and produce more NETs in tMCAO. This is not contradictory to the results of the present study.

Conclusions 16
In conclusion, our results showed that hyperglycemia activates neutrophils to overproduce NETs and identified NETs as a key factor that exacerbates ischemic brain damage in diabetes. The upstream signaling pathways controlling NETosis in hyperglycemia or diabetes need further exploration.  Inhibition of NETosis decreased the infarction in diabetic and hyperglycemic stroke animals. Inflammation associated with NETosis in diabetic and hyperglycemic stroke animals. The leve