Diabetic Stroke Severity: Epigenetic Remodeling and Neuronal, Glial, and Vascular Dysfunction

Type 1 diabetes (T1D) is a major risk factor for ischemic cerebrovascular disease that increases morbidity and mortality worldwide (1,2). Stroke occurs fivefold more often in patients with T1D and results in intense consequences compared with those in patients without diabetes; however, the mechanisms underlying stroke severity in patients with diabetes are unclear (3). Also, whether stroke is different in people with and without diabetes is unclear. Therefore, understanding these molecular changes and functional relationships with altered regulatory pathways can help explain the conceptual basis behind the severity of stroke in patients with diabetes. In addition, exploring the regulatory pathways and molecular mechanisms may be helpful in the future for designing preventive and therapeutic strategies.

The role of epigenetics was recently broadly characterized and found to be involved in the pathophysiology of stroke. Large clinical trials, such as the Diabetes Control and Complications Trial (DCCT) and Epidemiology of Diabetic Interventions and Complications Trial (EDIC), along with animal studies, have confirmed that hyperglycemia predisposes individuals to develop complications of diabetes, also referred to as the legacy effect (4,5). These reports suggest that hyperglycemia exposure mediates long-lasting epigenetic modifications that intensify global gene expressions by epigenetics modifications. Hence, exploring epigenetic factors underlying diabetic stroke is the need of the hour to better understand the basis of severity in patients with diabetes during stroke.

Epigenetic modifications, for example, DNA methylation, regulate phenotype and gene expression patterns and occur through the aid of enzymatic activity of DNA methyltransferases (DNMTs). Methylation usually occurs at the fifth carbon atom of cytosine residues and forms 5-methylcytosine (5-mC), which may be further oxidized to 5-hydroxymethylcytosine (5-hmC) (6,7). Studying DNMTs and global 5-mCs and 5-hmCs for their biological roles is of great scientific interest that may help in determining gene expression in the diseased and nondiseased state. Whereas 5-mC predicts a compacting chromatin inaccessible to transcription, 5-hmC determines efficient chromatin transcription. However, global 5-mC and 5-hmC levels in diabetic stroke have not been studied.

Specialized endothelial cells line the cerebral vessels and are enwrapped with pericytes and astrocytes that in turn connect to neurons providing nursing and maintenance. A number of different factors, including the extracellular matrix, tight junctions (TJs), pericytes, and astrocyte end-feet, together with adherens junctions, form junctional complexes and play a central role in the control of blood-brain barrier (BBB) integrity (8,9). Disruption in BBB integrity causes BBB permeability, which is manifested by the activation of matrix metalloproteinase-9 (MMP-9) that chops off junction barriers (10). Dysfunction of the BBB advances to stroke-like pathologies by increasing microvascular permeability that in turn affects the neuronal, glial, and vascular system. However, the roles of BBB and neuro-glio-vascular dysfunction in diabetic stroke pathology have not been studied much. The regulatory functions of vascular and neuronal integrity are essentially regulated by nitric oxide (NO) synthase (NOS) enzymes. The NOS enzymes produce NO, which regulates vascular tone, insulin secretion, and neuronal development via different isozymes, for example, endothelial NOS (eNOS) and neuronal NOS (nNOS). NOS regulation has been studied in ischemic injury, but an explanation is needed of its significance in diabetic stroke.

Our goal in the current study was to address two aspects: first, to explore whether ischemia-reperfusion (IR) injury in T1D severely disturbs the neuro-glio-vascular unit by causing intense inflammation, global alteration in epigenetic markers, and high MMP-9 activation, and second, to compare IR injury in T1D with non-T1D to find out the basis of IR injury severity in diabetes.

All animal procedures were conducted at the University of Louisville Health Sciences Center, were in compliance with guidelines established by the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and were approved by the University of Louisville Institutional Animal Care and Use Committee. Male wild-type (WT; C57BL/6J) and Akita (genetic T1D mice, C57BL/6-Ins2^Akita/J) mice (8–10 weeks old) were obtained from The Jackson Laboratory (Bar Harbor, ME). The experimental mice groups were 1) sham, 2) IR, 3) sham^Akita, and 4) IR^Akita. Akita mice with a glucose level >400 mg/dL were used in the study. We assessed inflammatory (tumor necrosis factor-α [TNF-α], interleukin-6 [IL-6]) and anti-inflammatory (IL-10) cytokines and glial markers (glial fibrillary acidic protein [GFAP]), integrin-α M [CD11b]) using real-time quantitative (q)-PCR. Neuronal (neuronal nuclei [NeuN], neuronal-specific enolase [NSE], nNOS), vascular (zona occluden-1 [ZO-1], claudin-5, eNOS), and epigenetic (DNMT-1 and DNMT-3a) markers were determined using Western blot. MMP-9, vascular (vascular endothelial [VE]-cadherin, occludin), glial (connexin-43 [Cx-43], GFAP), and neuronal (NeuN) markers were quantified using immunohistochemistry (IHC) analysis.

All mice were allowed free movement in the cage (4–5 mice/cage) and free access to water and food. The mice were maintained at 12/12-h day-night cycle at ∼23°C room temperature. Mice were fasted overnight with free access to water just before the day of surgery. At the day of surgery, the mice were transferred to the surgery room, anesthetized with pentobarbital (50 mg/kg body weight), and operated on within 1 h. Anesthetized mice were orally intubated, mechanically ventilated, and the body temperature was maintained at 37 ± 1°C during surgery.

The common carotid artery was exposed through a midline neck incision and dissected free of the surrounding nerves. Lysine-coated monofilament (2-cm long; 5-0 or 6-0 [11,12]) was inserted into the left external carotid artery, advanced into the internal carotid artery, and wedged into the cerebral arterial circle to obstruct the origin of the middle cerebral artery. After 40 min of surgery, the filament was withdrawn to allow reperfusion for 1 day. The same anesthesia and surgical procedures, except the insertion of filament, were performed in sham groups of mice (13).

Mice were assessed with neurobehavioral tests after IR injury to determine ischemia severity. Neurological deficit scores were graded 0–12 (normal score = 0, maximum score = 12) after the mice performed several behavioral tests, including posture relax test, forelimb placing test, circling, and motor coordination tests. Only mice with a high-grade neurological deficit (≥9) were used for the study (14,15).

Microvascular permeability was measured as described earlier (10). Mice were anesthetized with pentobarbital sodium (50 mg/kg i.p.). A heating pad was used to maintain body temperature of the mice at 37 ± 1°C. A 14-mm hole was made in the skull using a high-speed microdrill (Fine Scientific, Foster City, CA). Fluorescein isothiocyanate (FITC)-conjugated BSA (BSA-FITC-albumin, 300 μg/mL) was infused through carotid artery cannulation. In vivo imaging with a BX61WI fluorescent microscope (Olympus, Tokyo, Japan) was used to examine the exposed area of the skull. Venules were identified by the topology of the pial circulation and blood flow direction. Selected third-order venular segments were recorded and used as the baseline. After the baseline reading was obtained, images of the venular segments were recorded. The lamp power and camera gain settings were held constant during the experiments. Data were interpreted with the software provided with the instrument and Image-Pro Plus 6.3 software (Media Cybernetics, Bethesda, MD). Leakage of FITC-BSA was assessed by changes in the ratio of fluorescence intensity in the interstitium to that inside the vessel. The results were averaged and presented.

Brain samples were collected for Western blot, q-PCR, and IHC analysis. The brain tissue samples were harvested from the experimental mice groups, washed with 50 mmol/L PBS (pH 7.4), and stored at −80°C until use.

Equal quantities of brain extracts (40 μg) were run on 10–15% polyacrylamide gel under reducing condition, and separated proteins were transferred to polyvinylidene fluoride membrane using an electrotransfer apparatus (Bio-Rad, Hercules, CA). After blocking with 5% nonfat dry milk for 1 h, the membranes were probed overnight with a primary antibody (eNOS, nNOS, claudin-5, ZO-1, DNMT-3a, DNMT-1, NSE, NeuN) at 4°C. The next day, the blots were probed with appropriate secondary antibody for 2 h at room temperature and were developed using the ChemiDoc XRS+ Molecular Imager (Bio-Rad). The images were recorded in the chemi-program of a gel documentation system (Bio-Rad). The membranes were stripped and reprobed with monoclonal anti-GAPDH antibody (Millipore, Billerica, MA) as a loading control. Each band density was normalized with a respective GAPDH density using Image Lab densitometry software (Bio-Rad).

Total RNA from the brain tissue was isolated using TRIzol reagent (Invitrogen, Grand Island, NY), following the manufacturer’s instructions. After the quantity and purity of total RNA was confirmed using NanoDrop 1000 (Thermo Fisher Scientific, Waltham, MA), the RNA was reverse transcribed to cDNA according to the manufacturer’s protocol (ImProm-II; Invitrogen). cDNA samples were amplified for the given genes using gene-specific primers (GFAP, CD11b, TNF-α, IL-6, IL-10; sequences given in Supplementary Table 1) using Stratagene Mx3000p (Agilent Technologies, Santa Clara, CA). CT values were determined after baseline and threshold adjustments, and the results were expressed in fold expression. The transcript levels of given genes were normalized with Rn18s.

Frozen brain blocks, prepared with optimal cutting temperature media (Triangle Biomedical Sciences, Durham, NC), were cut into 20-μm sections using a Leica CM cryostat (Leica Microsystems, Buffalo Grove, IL). After removing the mounting matrix and fixing with methanol for 10 min, tissues were blocked with blocking solution (0.1% Triton X-100 TBS, 0.5% BSA, and 10% normal donkey serum) for 1 h at room temperature. The sections were incubated with primary antibody (MMP-9, occludin, VE-cadherin, NeuN, GFAP, Cx-43) overnight at 4°C. After incubation with the appropriate fluorescence secondary antibodies for 60 min at room temperature, the sections were stained with DAPI (1:10,000) and mounted with antifade mounting media. Images were acquired using a FluoView 1000 laser scanning confocal microscope (Olympus, Allentown, PA), and the data were analyzed with Image-Pro Plus image analysis software.

Genomic DNA was isolated (Sigma-Aldrich, St. Louis, MO) from the brain sample, and total 5-mC and 5-hmC levels were determined using ELISA kits (Epigentek, Farmingdale, NY), according to the manufacturer’s instructions.

Fluoro-Jade C (FJC) stain (Sigma-Aldrich) was used to determine the presence of neuronal damage. Brain sections were processed according to the method described previously (16). Briefly, paraffin-embedded sections were deparaffinized, rehydrated, and transferred to 0.06% potassium permanganate solution for 10 min. Afterward the sections were incubated in a 0.0001% solution of FJC for 20 min and mounted with DPX. Images were captured, and the data were analyzed with Image-Pro Plus image analysis software.

Coronal sections (2 mm) were cut using a mouse brain slice matrix (Harvard Apparatus, Holliston, MA). The slices were stained for 20 min at 37°C with 2% 2,3,5-triphenyltetrazolium chloride (Sigma-Aldrich) and postfixed with 4% paraformaldehyde. Infarct area (pale white) of each brain section was determined with Image-Pro Plus image analysis software.

The cortical parts of the brain ipsilateral area were dried for 3 days in a drying oven at 100°C. The absolute water content was calculated as water content (%) = [(wet weight – dry weight)/wet weight] × 100.

All values are expressed as mean ± SEM. Statistical comparisons between two groups were performed by the Student t test. One-way ANOVA was used for more than two groups. A P value of <0.05 was considered to be statistically significant.

IR^Akita mice showed more infarct size (Fig. 1A and B), cerebral edema (Fig. 1C), proinflammatory cytokines (TNF-α and IL-6), and reduced anti-inflammatory cytokines (IL-10; Fig. 1D) compared with IR and sham^Akita groups. Cresyl violet staining demonstrated significant degradation of the cellular constituents as indicated by cell swelling (blubbing), decreased cell size (shrinkage), and extensively decreased cell number in the IR^Akita group compared with the sham, IR, and sham^Akita groups (Fig. 1E). These results indicate severe IR-injury in diabetic mice compared with nondiabetic mice.

To address potential markers associated with epigenetics, mice brains were evaluated for levels of DNMTs and global %5-mC and %5-hmC levels. Western blots exemplified the highest protein expression of DNMT-1 and -3a in the sham^Akita group. The IR group exhibited a significant increase in DNMT-1 and -3a protein levels compared with the sham group, and the IR^Akita group showed a considerable decrease in DNMT-1 and -3a compared with the sham^Akita group (Fig. 2A and B). Furthermore, the IR group showed an increase in global 5-mC levels and a decrease in global 5-hmC levels compared with the sham group, and the IR^Akita group showed a decrease in global 5-mC and 5-hmC levels compared with the sham^Akita group (Fig. 2C and D). Hence, DNMTs and 5-mC levels were increased in the IR group and decreased in IR^Akita group compared with the sham and sham^Akita groups, respectively. These results represented differential epigenetic remodeling after IR injury in diabetic versus nondiabetic mice.

Intracarotid FITC-BSA infusion exemplified the highest macromolecular pial venular permeability in the IR^Akita group. The IR and sham^Akita groups also exhibited considerable high venular permeability compared with the sham group (Fig. 3A and B). We further evaluated endothelial junction proteins because these actively regulate selective barrier functions across the vessel walls. IHC analysis using endothelial junction proteins (occludin and VE-cadherin) determined drastically reduced expression in the cortical vessels of the IR and IR^Akita groups compared with their respective sham groups (Fig. 3C and D). Similarly, Western blot analysis using TJs (ZO-1 and claudin-5) confirmed remarkably decreased expressions in the IR^Akita group compared with the sham, IR, and sham^Akita groups (Fig. 3E and F). Protein expressions of ZO-1 and claudin-5 were also considerably reduced in the IR and sham^Akita groups compared with the sham group (Fig. 3E and F). Because the vascular impairment is exacerbated with MMP-9 activation, we further determined vascular MMP-9 expression using IHC analysis. The highest elevation in vascular MMP-9 expression among the different groups was observed in the IR^Akita group. However, significant enhancement in vascular MMP-9 was also observed in the IR and sham^Akita groups compared with the sham group (Fig. 3G and H).

To address glial markers, we determined transcript levels of GFAP and CD11b using q-PCR analysis. Whereas the IR group showed significantly increased GFAP and CD11b mRNA levels, the IR^Akita group showed a noticeable decrease in GFAP and CD11b mRNA compared with their respective shams. GFAP and CD11b were also significantly high in the sham^Akita mice compared with the sham mice (Fig. 4A). To address the vascular connection to glia, we evaluated Cx-43, an astral-gap junction marker, through IHC analysis of cerebral vessels. Extremely reduced immunoreactivity of Cx-43 was observed in the cortical vessels of the IR^Akita mice compared with the sham^Akita mice. Conversely, increased Cx-43 immunoreactivity was observed in the IR group compared with the sham group (Fig. 4B and C). Furthermore, IHC analysis confirmed reduced GFAP immunoreactivity in the hippocampus region of IR-injured Akita brains compared with the other brains, and significant amplification in the GFAP immunoreactivity was observed in sham^Akita and IR brains compared with sham brains (Fig. 4D and E). These results showed differential regulation of glial activation after IR injury in diabetic versus nondiabetic conditions.

NeuN staining confirmed notable loss of neurons in the hippocampus region after ischemic injury in IR and IR^Akita brains compared with their respective shams. Diabetic Akita brains also exhibited decreased NeuN expression compared with sham brains (Fig. 4D and E). Furthermore, Western blot analysis confirmed the lowest expression of NeuN in IR^Akita brains (Fig. 5A and B). Western blot analysis illustrated reduced expression of NSE, another neuronal-related marker, in the IR^Akita group compared with the sham^Akita group. Interestingly, increased expression of NSE was also observed in the IR group compared with the sham group (Fig. 5A and B). We performed FJC staining to address neuronal loss, and the highest degenerating neurons were observed in IR^Akita brains. FJC staining confirmed neurodegeneration in IR and sham^Akita brains compared with sham brains (Fig. 5C and D). These results suggested intense neuronal damage during IR injury in diabetes.

We determined eNOS and nNOS levels that impart main roles in regulating vascular tone and glia and neuronal integrity. Western blot analysis confirmed decreased protein expression of eNOS and nNOS in the IR^Akita group compared with the sham^Akita group, while increased eNOS and nNOS expression was observed in the IR group compared with the sham group (Fig. 6A and B). The sham^Akita group also represented a remarkable increase in nNOS with respect to the sham group (Fig. 6A and B). These data further suggest differential IR injury outcomes in diabetic versus nondiabetic brains.

Our results suggest that IR injury in T1D is severe in infarct volume, intense inflammation, cell death, remodeling of global epigenetic markers, and intense vascular MMP-9 activation. Interestingly, the severity of IR injury in T1D is exacerbated by differential regulations of global epigenetic, vascular, neuronal, and glial functions compared with IR injury in non-T1D.

Although T1D is much less common than type 2 diabetes, the symptoms and injuries are equally abrupt and sometimes more severe. To address the basis of severity in T1D, we performed our studies in the well-established genetic T1D Akita mouse model. Akita mice closely mimic human T1D because they have a genetic defect in the insulin 2 (Ins2^+/−) gene and therefore induce hyperglycemia naturally. Besides that, these mice show higher diabetic traits (blood glucose, 27.3 ± 5.3 mmol/L for males and 13.6 ± 3.8 mmol/L for females), with decreased reactive immunologically detectable insulin (20.7 to 9.1% in males and 45.9 to 49.6% in females) (17). Akita mice have been used for studying the effects of diabetes on cerebral vasculature (18), sexual dimorphism during diabetes (17,19,20), myelinated fiber loss (21), peripheral neuropathy, and memory performance (22). Akita mice also showed adverse inflammatory and epigenetic remodeling in the heart (23) and impaired vascular density in the brain (18). However, there are no reports of these mice being used to study the effects and mechanisms of ischemic injury. After creating IR injury in Akita hyperglycemic (>400 mg/dL) mice, we found larger cerebral infarcts, more edema, increased cell death, intense inflammation, and reduced anti-inflammation compared with IR-injured nondiabetic mice. We also observed high functional neurological deficits in IR^Akita mice compared with IR mice. A study of a T1D BioBreeding rat model described the association of diabetic-IR injury with increased infarct size even at tight blood glucose control (24). Similarly, a clinical study that used X-ray computed tomography in 104 patients also confirmed the correlation between hyperglycemia and cerebral infarct size in patients with stroke (25).

The cumulative results of large clinical trials (DCCT, EDIC) and animal studies confirmed the abnormality of the cells under hyperglycemia and invokes phenomenon of altered epigenetics in perpetuating diabetic complications (4,5). Understanding epigenetic modifications may help in exploring novel epigenetic mechanisms that could be targeted for early standpoint or therapeutic aspects against intense ischemic injury during diabetes. We therefore addressed potential epigenetic markers (global %5-mC, %5-hmC, and methylation enzymes). DNMT-1 (maintenance methylation) and DNMT-3a (de novo methylation) were found to be highest in the sham^Akita group. Our results further showed that DNMT-1, DNMT-3a, and global 5-mC levels were decreased in diabetic brains and increased in nondiabetic brains after ischemic insult. Stroke in the hyperglycemic state adversely affects potential epigenetic markers that increase the stroke severity in diabetic mice. These results certainly suggest that differential epigenetic alterations can be associated with intense IR injury outcomes in diabetic mice compared with nondiabetic mice. Similar to global %5-mC levels, global %5-hmC levels were also decreased in IR-injured diabetic brains, suggesting sheared and inactive chromatin status that can be associated with intense cellular damage in diabetic brains after ischemia. In agreement with our findings, previous studies have reported an increase in methyltransferases levels in middle cerebral artery occlusion and Akita mice models (23,26,27), and a study of the hippocampus region of 10 patients with Alzheimer’s disease found decreased levels of 5-mC and 5-hmC (28). Hence, the decrease of 5-mC and 5-hmC in the study of patients with Alzheimer’s disease and in our study suggests that these epigenetic modifications are involved in the pathogenesis of disease. Liyanage et al. (29) reported that there are dynamic changes in DNA methylation that are associated with ischemic injury. Accordingly, it could be suggested that the observed epigenetic changes induce gene expression alterations that ultimately accumulate and develop to disease pathology.

Vascular junctions play a very crucial role in regulating BBB integrity, and their loss is associated with pathogenesis of diseases (10,16). We reported an ∼80% decrease in ZO-1 in IR^Akita versus sham^Akita mice and an ∼57% decrease in ZO-1 in IR versus sham mice. Similarly in claudin-5, there is an ∼60% decrease in IR^Akita versus sham^Akita mice and an ∼40% decrease in IR versus sham. The percentage of the decrease is more in IR^Akita because the TJs were already disrupted in the diabetic condition and became worse after IR injury. Earlier reports also suggest that the TJs are disrupted in hyperglycemia, corresponding to almost 40% (30).

We further reported functional BBB integrity loss associated with neuro-glia-vascular dysfunction that manifests with intense inflammation and global epigenetic remodeling after IR injury in diabetic mice. Extensive BBB disintegration in IR^Akita mice increases the chances of blood proinflammatory molecules entering the brain and worsening the IR injury outcomes during diabetes. MMP-9, which is the hallmark for cellular junctions, is activated by inflammation and by creating IR injuries, and we have shown that MMP-9 activation chops off vascular junction proteins and raises BBB permeability (10). In the current study, we observed more fold change of MMP-9 in IR mice compared with IR^Akita mice, although the basal level of MMP-9 was high in sham^Akita compared with WT mice. In agreement, we have reported from our laboratory that the basal expression level of MMP-9 is high in diabetic mice compared with the control (31,32). We observed the highest vascular MMP-9 expression and cerebrovascular permeability in IR-injured diabetic mice, suggesting that persistent hyperglycemia might accelerate brain vasculature and BBB damage. However, controversy remains for subjects with diabetes: on the one hand, sustained BBB integrity was reported (33), whereas on the other hand, increased BBB damage was revealed during diabetes (34). In our study, we found increased BBB permeability in diabetic Akita mice. Previous studies have demonstrated that the ischemic brain in db/db mice (type 2 diabetes) and streptozotocin-induced mice and rats (T1D) also exhibits increased cerebrovascular dysfunction (35,36).

A cascade of events is mediated after ischemic brain injury, yielding Ca²⁺-dependent activation of the NOS isoforms nNOS and eNOS (37). The role of NOS isozymes in cerebral ischemia damage was described in the study performed in transgenic mice lacking expression of nNOS or eNOS and in in vitro and in vivo models of cerebral ischemia. The study suggested that nNOS plays key roles in neurodegeneration, whereas eNOS has a prominent role in maintaining cerebral blood flow and preventing neuronal injury (38). By looking at eNOS expression, which serves as a major weapon against different vascular diseases, differential regulation was observed during IR injury in diabetic versus nondiabetic mice, indicating its involvement in enhancement of IR pathology in diabetes.

Intensive vascular injury affects vascular-glial interactions during IR injury in diabetes. Astrocytes, the major glial subtype, establish glial network and communicate through gap junctions. A number of studies reported that astrogliosis, a process of glial activation, increases after an ischemic injury to provide support to the neurons. However, we observed decreased GFAP immunoreactivity against IR injury in diabetic conditions. In agreement with our findings, previous reports suggest that diabetic hyperglycemia inhibited ischemia-induced activation of astrocytes and, therefore, caused damage to astrocytes (39,40), probably due to oxidative damage of DNA (41). An earlier study showed that hyperglycemic Akita mice have decreased gap junction communication in oocytes, as demonstrated by lower expression of Cx-43 (42). Hypoglycemia has been found to induce microglial (CD11b) activation (43), whereas in our study, we found that hyperglycemia induced a decrease of CD11b in IR-injured diabetic mice. The differential regulation of glial, astrocyte, and astrocytic gap junction in IR-injured diabetic versus nondiabetic mice further indicated their probable involvement in IR severity during diabetes. Loss of glial activation in IR^Akita mice indicated loss of neurons, which was confirmed by FJC and NeuN expressions. FJC and NeuN have previously been used as specific markers for neurons after ischemic stroke (44) and to confirm the evolving phase of infarction after middle cerebral artery occlusion (45). Furthermore, increased levels of neuronal NSE and nNOS in IR mice, whereas decreased levels in IR^Akita, also indicated differential regulation of the neuronal microenvironment after an IR insult in diabetic versus nondiabetic mice. The increase of the NSE level in IR mice is in agreement with previous findings (46,47), but the decrease of NSE in IR^Akita is suggestive of neuronal inability to transcribe NSE due to persistent hyperglycemia. In support, another study showed an increase in the NSE mRNA level in patients with diabetes, but a decrease occurred in subjects with diabetic neuropathy (48).

According to the 2014 National Diabetes Statistics Report, ∼29.1 million people, or 9.3% of the U.S. population, have diabetes. In this population, 21.0 million people have been diagnosed with diabetes, and 8.1 million people (∼27.8%) with diabetes are undiagnosed. Compared with diagnosed people who receive some treatment, the undiagnosed people face more risk of stroke because they do not receive any treatment. Our study can be helpful in that direction. To show the close resemblance of our mouse model with the population with diabetes receiving some treatment, we treated Akita mice with insulin and observed less stroke severity after creating the IR injury compared with untreated mice (Supplementary Figs. 1 and 2).

Although some reports suggest that DNA methylation levels have been found altered in T1D patients, the information about the effect of antidiabetic therapy on epigenetics is scarce. In concordance with our study, altered epigenetic changes have been observed in the kidney of db/db diabetic mice in a tissue-specific manner. The authors further reported aberrant DNA methylation, changes in histone modifications, and mRNA expression in the diabetic kidney and that these changes were resistant to pioglitazone (49). In another study by Ishikawa et al. (50), there was increase in the DNA methylation of the Ins1 (insulin 1) promoter in isolated islets from Zucker diabetic fatty rats. However, use of metformin treatment suppressed DNA methylation and upregulated insulin gene expression. Evans-Molina et al. (51) observed 15- and 7-fold increases in expression of Ins1/2 and GLUT2 in cultured islets harvested from mice after 6 weeks of oral pioglitazone therapy. These islets also showed increased altered epigenetic modifications, mRNA, and protein levels. Although there are few reports on the effect of antidiabetic therapy on epigenetic markers, literature suggests that with the progression of diabetes, the epigenetic changes are also progressive, although the changes are slowed with the use of antidiabetic therapy. The study data clearly indicate that the stroke effect in the Akita mice was significantly enhanced compared with the stroke effect in the WT mice. The results show strong evidence that epigenetics and neuro-glio-vascular markers are drastically changed in diabetic IR^Akita mice and contribute to the disease pathology. However, additional studies in this direction are warranted. On the basis of the results obtained, we have proposed a hypothesis that explains the severity in diabetic stroke (Fig. 7).

In summary, our study provides a novel insight in understanding the mechanistic basis of IR injury severity in T1D by comparing with nondiabetic injury. Through the current report we observed differential regulations of epigenetic markers and neuro-glio-vascular components that explain the reasons of IR severity in subjects with diabetes. The involvement of intense inflammation, potential markers associated with epigenetic alterations, and high vascular MMP-9 activation in diabetes profoundly intensify the injury outcomes. These findings pave the way for further studies that might be helpful in developing better preventive and therapeutic approaches for diabetic stroke.

Acknowledgments. The authors are indebted to research volunteer Guoxin Chu, MD, Jilin Province People's Hospital, Changchun, China, for his support and help in the surgical experiments.

Funding. This work was supported by a National Institutes of Health grant from the National Heart, Lung, and Blood Institute (HL-107640) to N.T.

Duality of Interest. No potential conflicts of interests relevant to this article were reported.

Author Contributions. A.K. designed and performed the studies and wrote the manuscript. P.K.K. performed the experiments and participated in the experimental analysis. N.T. obtained the funding, helped in the experiments and with the data analysis, and participated in the writing of the manuscript. All authors approved the final form of the manuscript. N.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.