Exacerbated VEGF up-regulation accompanies diabetes-aggravated hemorrhage in mice after experimental cerebral ischemia and delayed reperfusion.

Reperfusion therapy is the preferred treatment for ischemic stroke, but is hindered by its short treatment window, especially in patients with diabetes whose reperfusion after prolonged ischemia is often accompanied by exacerbated hemorrhage. The mechanisms underlying exacerbated hemorrhage are not fully understood. This study aimed to identify this mechanism by inducing prolonged 2-hour transient intraluminal middle cerebral artery occlusion in diabetic Ins2Akita/+ mice to mimic patients with diabetes undergoing delayed mechanical thrombectomy. The results showed that at as early as 2 hours after reperfusion, Ins2Akita/+ mice exhibited rapid development of neurological deficits, increased infarct and hemorrhagic transformation, together with exacerbated down-regulation of tight-junction protein ZO-1 and up-regulation of blood-brain barrier-disrupting matrix metallopeptidase 2 and matrix metallopeptidase 9 when compared with normoglycemic Ins2+/+ mice. This indicated that diabetes led to the rapid compromise of vessel integrity immediately after reperfusion, and consequently earlier death and further aggravation of hemorrhagic transformation 22 hours after reperfusion. This observation was associated with earlier and stronger up-regulation of pro-angiogenic vascular endothelial growth factor (VEGF) and its downstream phospho-Erk1/2 at 2 hours after reperfusion, which was suggestive of premature angiogenesis induced by early VEGF up-regulation, resulting in rapid vessel disintegration in diabetic stroke. Endoplasmic reticulum stress-related pro-apoptotic C/EBP homologous protein was overexpressed in challenged Ins2Akita/+ mice, which suggests that the exacerbated VEGF up-regulation may be caused by overwhelming endoplasmic reticulum stress under diabetic conditions. In conclusion, the results mimicked complications in patients with diabetes undergoing delayed mechanical thrombectomy, and diabetes-induced accelerated VEGF up-regulation is likely to underlie exacerbated hemorrhagic transformation. Thus, suppression of the VEGF pathway could be a potential approach to allow reperfusion therapy in patients with diabetic stroke beyond the current treatment window. Experiments were approved by the Committee on the Use of Live Animals in Teaching and Research of the University of Hong Kong [CULATR 3834-15 (approval date January 5, 2016); 3977-16 (approval date April 13, 2016); and 4666-18 (approval date March 29, 2018)].

Chinese Library Classification No. R453; R364; Q2

Introduction

Diabetes mellitus is a prominent disease worldwide (Saeedi et al., 2019). Type 1 and type 2 diabetes increases the risk of ischemic stroke by 6.3-fold and 2.3-fold, respectively (Janghorbani et al., 2007), and one study reported that 31% of patients with ischemic stroke had diabetes (Reeves et al., 2010). Diabetic stroke is associated with a poor prognosis and high mortality (Eriksson et al., 2012; Lau et al., 2019).

The current standard intervention of large vessel occlusion is thrombolysis combined with endovascular thrombectomy (Shi et al., 2018). However, such reperfusion therapies are hindered by their narrow time window (for thrombolysis (alteplase): 4.5 hours after symptom onset; for endovascular thrombectomy: within 24 hours of last known normal to patients meeting DAWN trial eligibility criteria. DAWN excludes patients with baseline blood glucose > 400 mg/dL (22.2 mM), and for these patients, endovascular thrombectomy must be initiated within 6 hours of symptom onset; Jovin et al., 2017; Powers et al., 2019). In combination with a concern of increased intracranial hemorrhage risk in patients with diabetes/hyperglycemia (Celik et al., 2004) following reperfusion therapies (Reeves et al., 2010; Jiang et al., 2015), the rate of patients with diabetes receiving these beneficial treatments is lower than patients without diabetes (Reeves et al., 2010; Nathaniel et al., 2019; Saber et al., 2020).

Diabetic/hyperglycemic-exacerbated hemorrhage is likely to be the result of an aggravated inflammatory response primed by diabetic/hyperglycemic conditions (Jiang et al., 2021). Elevated serum matrix metallopeptidase 9 (MMP-9) has been associated with aggravated outcomes in clinical studies (Abdelnaseer et al., 2015; Zhong et al., 2017), as well as with cerebral hemorrhagic transformation in ischemia-reperfused animals (Elgebaly et al., 2010; McBride et al., 2020) and patients (Montaner et al., 2003). Furthermore, MMP-9 levels upon ischemic stroke has been found to be aggravated under hyperglycemic conditions in human serum (Setyopranoto et al., 2018) and in the cerebrums of diabetic mice (Kumari et al., 2011). Another matrix metallopeptidase, MMP-2, has also been considered to participate in the early stage of ischemic pathology (Yang and Rosenberg, 2015) and has found to be up-regulated very soon after experimental ischemia in baboons (Heo et al., 1999; Chang et al., 2003). MMP-2 has also been shown to destroy vascular integrity (Liu et al., 2012) and induce hemorrhagic transformation upon ischemia in mice (Lu et al., 2013). However, the mechanism/s underlying MMP-2/9 aggravation in diabetes upon ischemic stroke have not been well studied.

Although ischemic stroke has been investigated using hyperglycemic/diabetic animal models, most existing mice model studies induced either a short ischemia length (0.5–1.5 hours) or permanent occlusion in favor of reduced animal mortality (Tsuchiya et al., 2003; Villalba et al., 2018). These experimental paradigms have left the mechanism of diabetes-exacerbated outcomes under prolonged ischemia and delayed reperfusion largely unstudied. Thus, there is still limited progress in extending the window of reperfusion therapies for patients with diabetic stroke. However, a fast and complete reperfusion is important for outcome improvement in patients with stroke (Shi et al., 2018).

To fill this gap between basic research and the clinical situation, this study used Ins2 mice and 2-hour transient intraluminal MCAO to mimic delayed mechanical thrombectomy in patients with diabetic stroke, and investigated the causal mechanism of exacerbated intracerebral hemorrhage after prolonged ischemia in diabetes. Ins2 mice are a widely accepted model of type 1 diabetes (Lai and Lo, 2013) and have been used to study diabetic complications in various organ systems. The present results could aid the development of interventions to extend the treatment window of mechanical thrombectomy in patients with diabetic stroke by mitigating the exacerbated hemorrhage through the identification of suppression targets.

Materials and Methods

Animals

Twelve-week-old male Ins2 (n = 74; 23.7 ± 0.2 g) and Ins2 (n = 69; 25.3±0.2 g) mice were generated from C57BL/6-Ins2 and C57BL/6J mice from Jackson Laboratory (Bar Harbor, ME, USA). They were kept under a 12-hour light/dark cycle and given ad libitum access to food and water. Two to five mice were housed per cage. Blood glucose level was measured using a glucometer (Ascensia Elite XL, Bayer Healthcare AG, Leverkusen, Germany).

The mice of both genotypes were randomly assigned to the sham or MCAO groups for each of the two time-points of assessment (2 hours ischemia/2 hours reperfusion or 2 hours ischemia/22 hours reperfusion), resulting in a total of eight groups ().

Experiments were conducted according to local and institute regulations and were approved by the Committee on the Use of Live Animals in Teaching and Research of the University of Hong Kong [CULATR 3834-15 (approval date January 5, 2016); 3977-16 (approval date April 13, 2016); and 4666-18 (approval date March 29, 2018)], as well as in compliance with the National Institutes of Health's Guide for the Care and Use of Laboratory Animals. This study was reported in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines (Percie du Sert et al., 2020).

Induction of ischemia and reperfusion

Transient focal cerebral ischemia was induced using the intraluminal method (Lo et al., 2005; Li et al., 2012; Yang et al., 2012). In brief, mice were randomly assigned and were subjected to isoflurane (IsoFlo; 50019100, Zoetis Inc., Kalamazoo, MI, USA) inhalation anesthesia (2% isoflurane in 70% N2O/30% O2 for induction; 1.25% isoflurane in 70% N2O/30% O2 for maintenance). A filament coated with vinylpolysiloxane material (ESPE 7302, 3M Dental Products, St. Paul, MN, USA) was inserted into the right internal carotid artery and further advanced until reaching the middle cerebral artery. After 2 hours of ischemia, the filament was removed to commence reperfusion. Animals in the sham groups received the same procedures except for filament insertion. Relative cerebral blood flow in the middle cerebral artery territory was monitored by a laser Doppler flowmeter (Periflux 5000, Perimed AB, Järfälla, Sweden), with the reading at 5 minutes before ischemia set as 100%. Body temperature was maintained at 37 ± 0.5°C throughout the surgery. Mice were placed in an intensive care unit set at 30 ± 0.5°C during ischemia and for 4 hours after reperfusion commenced. Mice were sacrificed at either 2 hours (2 hours ischemia/2 hours reperfusion) or 22 hours (2 hours ischemia/22 hours reperfusion) of reperfusion ().

Study design.

Ins2 mice and Ins2 mice were subjected to 2 hours of middle cerebral artery occlusion (MCAO) or sham surgery. Two hours later, the filament was removed to commence reperfusion in mice of the MCAO group, while mice in sham group received another sham surgery. At either 2 hours (2 h ischemia/2 h reperfusion; abbreviated as 2h I/2h R) or 22 hours (2 h ischemia/22 h reperfusion; abbreviated as 2h I/22h R) of reperfusion, mice were sacrificed for brain collection after neurological assessment.

Survival rate and neurological assessments

Experimental mice were closely monitored and survival rate was recorded. At the end of the reperfusion period (either 2 hours or 22 hours after reperfusion), neurological deficits were evaluated by an observer blinded to the genotype of the mouse (Lo et al., 2005; Li et al., 2012; Yang et al., 2012). The scoring system was as follows: 0 – no observable neurological deficits (normal); 1 – left wrist drop and walks straight (mild); 2 – left wrist drop and walks in a circular motion (moderate); 3 – loss of righting reflex (severe); and 4 – dead.

Tissue collection

Following decapitation under anesthesia with intraperitoneal injection of a mixture of ketamine (100 mg/kg, Alfasan International BV, Woerden, Netherlands) and xylazine (6 mg/kg, Alfasan International) at either 2 hours or 22 hours after reperfusion, mouse brains were cut into five 2 mm-thick coronal slices. For infarct and histological analyses, brain slices were stained with 2% 2,5-triphenyltetrazolium chloride and fixed in 10% formalin (see below). For molecular analyses, infarct and penumbra areas of brain slice number two to four were collected at 1 mm away from the midline, snap-frozen in liquid nitrogen and homogenized together in 10 mM phosphate buffered saline (PBS), pH 7.4, containing protease inhibitors (4693159001, Roche Applied Science) and phosphatase inhibitors (524628, Merck KGaA, Darmstadt, Germany), followed by protein or mRNA extraction (see below).

Measurement of infarct, swelling, and hemorrhagic transformation

Freshly cut brain slices were stained with 2% 2,5-triphenyltetrazolium chloride (TTC; T8877, Sigma-Aldrich, St. Louis, MO, USA) in PBS at 37°C for 7.5 minutes and then fixed overnight in 10% buffered formalin (Merck KGaA). The red live and white infarct areas on the posterior side of brain slices were measured using SigmaScan Pro (SPSS Inc., Chicago, IL, USA) by a researcher who was blinded to the genotypes of each mouse. Infarct areas and volume were estimated indirectly to minimize inaccuracy due to swelling (Swanson et al., 1990), and hemispheric swelling was estimated using the following equation: (ipsilateral volume − contralateral volume)/contralateral volume × 100%. Hemorrhagic transformations, identified as dark-brown areas on the posterior side of brain slice number three (at approximately bregma –0.34 mm), were presented as percentages of infarct and contralateral areas.

Histochemistry and immunohistochemistry

To further verify the presence of hemorrhage, the fixed brain slices were embedded in paraffin (T565, Fisher Chemical, Thermo Fisher Scientific, Waltham, MA, USA) and cut into sections of 7 μm thickness using a microtome (HM 315R, Microm International GmbH, Walldorf, Germany). Sections were mounted onto positive-charged glass slides and dried overnight in an oven set to 37°C, deparaffinized at 57°C for 30 minutes, followed by soaking in toluene twice for 5 minutes each, and rehydrated through a graded series of ethanol. After over-staining with Harris's hematoxylin for 5 minutes, the sections were differentiated in acid alcohol for a few seconds followed by bluing in Scott's tap and staining in 1% aqueous eosin for 3 minutes. Subsequently, the sections were dehydrated in ascending ethanol gradients followed by toluene for 3 times for 5 minutes each, mounted with Permount (SP15-500, Fisher Chemical) and covered with cover slips. Microscopic photos of infarct cores were taken using a light microscope (Eclipse 80i; Nikon, Tokyo, Japan) equipped with a digital camera (SPOT RT3 25.4 2 Mp Slider; Diagnostic Instruments, Inc., Sterling Heights, MI, USA), and patches of orange red indicated the presence of hemorrhage.

Western blot analysis

Lysates of infarct and penumbra areas of brain slices two to four were mixed with 2× ice-cold radioimmunoprecipitation assay lysis buffer in a 1:1 ratio. Supernatants were collected following centrifugation at 16,100 × g and 4°C for 30 minutes, mixed with Laemmli reducing loading dye at 95°C for 5 minutes, separated by SDS-PAGE at 30 mA (4% stacking gel and 10% separating gel, 10 μg protein per lane), and transferred onto polyvinylidene fluoride membranes (IPVH00010, Merck KGaA) at 300 mA for 2 hours on ice. The membranes were blocked with 5% skimmed milk (Nestlé S.A., Switzerland) in Tris-buffered saline with 0.1% Tween 20 detergent (0.1% TBST) and incubated with primary antibodies () diluted with 5% bovine serum albumin (USB Corporation, Affymetrix, Santa Clara, CA, USA) in 0.1% TBST at 4°C overnight followed by peroxide-conjugated anti-rabbit IgG (PI-1000, 1:2000, Vector Laboratories, Burlingame, CA, USA) or anti-mouse IgG (PI-2000, 1:5000, Vector Laboratories) secondary antibodies at room temperature for 1 hour. Immunoreactivities were detected using enhanced chemiluminescence reagents (RPN2106, GE Healthcare, Buckinghamshire, UK or K-12042-D10, Advansta, CA, USA) and light-sensitive films (47410 19291, Fujifilm, Tokyo, Japan), quantified using ImageJ (U.S. National Institutes of Health, Bethesda, MD, USA), and normalized using an endogenous protein (actin or α-tubulin), except for phospho (p)-Akt, p-Erk1/2, and p-p38 MAPK, which were normalized with their respective total expressions.

Real-time PCR analysis

After mixing lysates of infarct and penumbra areas of brain slices two to four with ice-cold RNAiso plus (9109, Takara Bio Inc., Japan) at a ratio of 1:4, total RNA was extracted by the phenol:chloroform extraction method, and cDNA was prepared from 2 μg of the extracted RNA (SuperScript VILO; 11754050, Life Technologies, USA), both according to the manufacturers’ instructions. Real-time PCR reactions were performed using the StepOnePlus system and SYBR Green technology (4385610, Life Technologies) with the primers listed in . The relative mRNA expression levels were shown as fold changes to the Ins2 sham group after normalizing with the endogenous gene β-actin, following the manufacturer's protocol.

Statistical analysis

No statistical methods were used to predetermine sample sizes; however, the sample sizes of this study (~20 animals in each group) were similar to those reported in our previous publication (Li et al., 2012). The total number of mice used in each group is shown in . Data are expressed as the mean ± standard error of mean (SEM) and statistical tests were performed using GraphPad Prism (v5.02; GraphPad Software, Inc., USA). Survival rate and the neurological score were analyzed using the log-rank test and Mann-Whitney U test, respectively. All other measurements were analyzed using a one-way analysis of variance followed by Tukey's HSD post hoc tests or unpaired Student's t-tests, except for the comparison of hemorrhagic transformation areas, in which case the Mann-Whitney U test was used due to non-normality of the data. A P-value of < 0.05 was considered statistically significant.

Results

Physiological parameters of Ins2 and Ins2 mice

Ins2 mice displayed hyperglycemia from 4 weeks of age and a decreased body weight from 13 weeks of age (). Ins2 mice had higher relative cerebral blood flow during ischemia, which was similar in both genotypes during reperfusion ().

.

Ins2 mice displayed hyperglycemia (A) and decreased body weight (B) from 4 and 13 weeks of age, respectively, when compared with Ins2 mice (n = 10–19; *P < 0.05, **P < 0.01, and ***P < 0.001, unpaired Student's t-test; mean ± SEM).

Ins2 mice subjected to prolonged ischemia showed decreased survival rate and worse neurological outcomes

Ins2 mice had a significantly lower survival rate than Ins2 mice after reperfusion (Figure and ). The majority of lethality was observed in the first 4 hours after reperfusion in Ins2 mice, while death of Ins2 mice mostly occurred later. Moreover, Ins2 mice suffered from more severe neurological deficits at 2 hours after reperfusion (), whereas there was no between-group difference at 22 hours after reperfusion ().

More severe neurological deficits and lower survival rate in .

(A & B) Survival rate after 2 hours of ischemia recorded at various time points after reperfusion. The majority of deaths of Ins2 mice occurred before 4 hours of reperfusion. (C & D) Neurological deficits were graded from 0 to 4 at the end of the assigned period of reperfusion. A significantly higher neurological score was observed in Ins2 mice compared with Ins2 mice at 2 hours after reperfusion, but the difference diminished at 22 hours after reperfusion (2h I/2h R: Ins2 (n = 20), Ins2 (n = 20); 2h I/22h R: Ins2 (n = 24), Ins2 (n = 34); *P < 0.05, Log-rank (Mantel-Cox) test; #P < 0.05, Mann-Whitney U test; mean ± SEM).

Ins2 mice displayed accelerated infarct development and increased hemorrhage after prolonged ischemia

Figure and show posterior photographs of TTC-stained brain slices at 2 and 22 hours after reperfusion, respectively. At 2 hours after reperfusion, infarct areas were significantly larger in brain slices one to three of Ins2 brains compared with Ins2 brains (), with significantly larger infarct volume compared with Ins2 brains and a trend towards greater hemispheric swelling (Figure and ). However, these differences between Ins2 and Ins2 mice diminished at 22 hours after reperfusion (Figure , , and ).

Increased development rate of infarct area and infarct volume upon 2h I/2h R ischemic challenge in .

Representative TTC-stained brain slices of mice after 2h I/2h R (A) and 2h I/22h R (B). Calculated infarct area, infarct volume, and hemispheric swelling after 2h I/2h R are shown in C, E, and F; those after 2h I/22h R are shown in D, G, and H, respectively. 2h I/2h R: Ins2 middle cerebral artery occlusion (MCAO) (n = 12), Ins2 MCAO (n = 10), 2h I/22h R: Ins2 MCAO (n = 10), Ins2 MCAO (n = 10); *P < 0.05, **P < 0.01, unpaired Student's t-test; mean ± SEM. I: Ischemia; R: reperfusion; TTC: 2,5-triphenyltetrazolium chloride.

Hemorrhage (red-pink spreading) was observed in ipsilateral Ins2 brains as early as 2 hours after reperfusion (Figure and ), while eosin staining revealed its presence around blood vessels in both the infarct core (Figure ) and penumbra area (data not shown). Quantitative analysis revealed a significantly larger hemorrhagic area in brain slice number three of Ins2 mice at 2 hours after reperfusion (Figure and ), which was robustly exacerbated at 22 hours after reperfusion (Figure and ).

Increased hemorrhage in .

General view of the ipsilateral side of the brain of Ins2 (A) and Ins2 (B) mice after 2h I/2h R challenge. The black line outlining the reddish area indicates the presence of hemorrhage. Representative regions of the infarct core located in the cerebral cortex after 2 hours (C & D) and 22 hours (E & F) of reperfusion are shown. Sections were stained with hematoxylin and eosin, with which red blood cells were stained in an orange-red color. Scale bars: 50 μm. Quantification of the hemorrhagic area was presented as a ratio of the infarct core as well as a ratio of the contralateral side in brain slice number three of mice after 2 hours (G & H) and 22 hours (I & J) of reperfusion. *P < 0.05, **P < 0.01, Mann-Whitney U test; mean ± SEM). I: Ischemia; R: reperfusion.

Decreased tight junction protein ZO-1 level and increased blood-brain barrier-disrupting MMP expressions in Ins2 brains after prolonged ischemia

Infarct and penumbra areas of operated mice were subjected to Western blot analysis. At 2 hours after reperfusion, there was a reduced level of tight junction protein ZO-1 in operated Ins2 mice upon MCAO, which was further exaggerated in Ins2 mice (). Likewise, at 22 hours after reperfusion, only Ins2 mice showed a significant reduction in ZO-1 level ().

Vulnerable blood vessel integrity in .

Protein expressions of ZO-1, MMP-2, and MMP-9 for the two experimental groups, 2h I/2h R (A, C, & E) and 2h I/22h R (B, D & F), were semi-quantified using Western blot analysis. The corresponding fold changes of these proteins are shown in the bottom panel. WT: Ins2, AK: Ins2; A, C, & E: Ins2 Sham: n = 4, Ins2 Sham: n = 4, Ins2 2h R: n = 5, Ins2 2h R: n = 7; B, D, & F: Ins2 Sham: n = 4, Ins2 Sham: n = 4, Ins2 22h R: n = 5, Ins222h R: n = 5; *P < 0.05, **P < 0.01, and ***P < 0.001, vs. Ins2 Sham; #P < 0.05, ##P < 0.01, and ###P < 0.001, vs. Ins2 Sham; ++P < 0.01, vs. Ins2 2h R or Ins2 22h R; one-way analysis of variance followed by Tukey's HSD post hoc test; mean ± SEM. I: Ischemia; MCAO: middle cerebral artery occlusion; MMP: matrix metallopeptidase; R: reperfusion.

Western blot analysis of MMP-2, which is known to disrupt the blood-brain barrier following stroke, was significantly up-regulated only in Ins2 ipsilateral brains at 2 hours after reperfusion (). Interestingly, at 22 hours after reperfusion, the previously up-regulated MMP-2 expression in Ins2 mice returned to a normal level that was similar to that of sham-operated controls; however, MMP-2 expression in Ins2 mice was significantly up-regulated ().

The expression of MMP-9 was also significantly increased in Ins2 mouse brains at 2 hours after reperfusion when compared with that of the sham-operated Ins2 mice (), but not at 22 hours after reperfusion ().

Up-regulation of vascular endothelial growth factor and its downstream factors p-Erk1/2 and p-p38 MAPK, but not p-Akt in Ins2 brains after prolonged ischemia

At 2 hours after reperfusion, vascular endothelial growth factor (VEGF) and p-Erk1/2 were remarkably over-expressed in Ins2 ipsilateral brains when compared with Ins2 mice (Figure and ). The expression of p-p38 mitogen-activated protein kinase (p-p38 MAPK) was also significantly higher in Ins2 ipsilateral brains than that of sham-operated Ins2 controls (). In contrast, there were no significant changes in p-Akt level in operated mice of both genotypes when compared with levels of sham-operated mice ().

Up-regulation of VEGF and its endothelium destabilizing downstream factors in .

Protein expressions of VEGF, p-Erk1/2, p-p38, and p-Akt for the two experimental groups, 2h I/2h R (A, C, E, and G) and 2h I/22h R (B, D, F, and H), were semi-quantitated using Western blot analysis. The corresponding fold changes of these proteins are shown in the bottom panel. (WT: Ins2, AK: Ins2; A, C, E, and G: Ins2 Sham: n = 4, Ins2 Sham: n = 4, Ins2 2h R: n = 5, Ins2 2h R: n = 7; B, D, F, and H: Ins2 Sham: n = 4, Ins2 Sham: n = 4, Ins2 22h R: n = 5, Ins2 22h R: n = 5; *P < 0.05, **P < 0.01, and ***P < 0.001, vs. Ins2 Sham; ##P < 0.01, and ###P < 0.001, vs. Ins2 Sham; +P < 0.05, ++P < 0.01, and +++P < 0.001, vs. Ins2 2h R or Ins2 22h R; one-way analysis of variance followed by Tukey's HSD post hoc test; mean ± SEM). I: Ischemia; MCAO: middle cerebral artery occlusion; R: reperfusion; VEGF: vascular endothelial growth factor.

Similarly, at 22 hours after reperfusion, VEGF expression only remarkably increased in Ins2 mice (). While there was a significant increase of p-Erk1/2 expression in the ipsilateral brains of both genotypes, its level in Ins2 mouse brains was significantly higher than that of Ins2 mice (). The expression of p-p38 in both genotypes was also significantly higher than that of the sham-operated controls (). No significant differences in p-Akt level were found between any of the groups ().

Increased endoplasmic reticulum stress markers expression in Ins2 brains during the early stage of reperfusion

The expression of endoplasmic reticulum (ER) stress markers in ipsilateral brains was revealed by RT-PCR. CHOP was significantly up-regulated in both Ins2 and Ins2 brains at 2 hours after reperfusion, with a more pronounced increase in Ins2 mice (). In contrast, a significant increase in BiP expression was only found in Ins2 mice when compared with sham-operated Ins2 controls (). These increases vanished at 22 hours after reperfusion ().

.

mRNA expressions of ER stress-targets were analyzed after 2 hours of ischemic challenge followed by 2 hours (A) or 22 hours (B) of reperfusion. (A: Ins2 Sham: n = 4, Ins2 Sham: n = 4, Ins2 2h R: n = 5, Ins2 2h R: n = 7; B: Ins2 Sham: n = 4, Ins2 Sham: n = 4, Ins2 22h R: n = 5, Ins2 22h R: n = 5; *P < 0.05 and ***P < 0.001, vs. Ins2 Sham; ##P < 0.01 and ###P < 0.001, vs. Ins2 Sham; ++P < 0.01, vs. Ins2 2h R; one-way analysis of variance followed by Tukey's HSD post hoc test; mean ± SEM). ER: Endoplasmic reticulum; I: ischemia; MCAO: middle cerebral artery occlusion; R: reperfusion.

Discussion

In the current study, we demonstrated that induction of prolonged ischemia followed by reperfusion in diabetic Ins2 mice could mimic the exacerbated outcomes observed in patients with diabetes upon ischemic stroke but with delayed reperfusion. Multiple studies have reported more severe neurological deficits, larger infarcts, and hemorrhagic transformation in diabetic animals after experimental stroke (Toung et al., 2000; Kusaka et al., 2004; Elgebaly et al., 2010; Chen et al., 2011; Ye et al., 2011; Cui et al., 2012; Ning et al., 2012; Yan et al., 2012; Liao et al., 2014; Mishiro et al., 2014). After inducing 2-hour long ischemia, we found early mortality and hemorrhagic transformation in diabetic Ins2 mice at as early as 2 hours after reperfusion, with a robust increase in hemorrhagic transformation at 22 hours after reperfusion. This study revealed that the diabetes-exacerbated hemorrhagic transformation was due to a quicker and heavier compromise of blood vessel integrity, which was shown by the greater loss of tight junction protein ZO-1 from 2 hours of reperfusion in Ins2 mice. Furthermore, this rapid compromise of blood vessel integrity was associated with accelerated VEGF and p-Erk1/2 up-regulation in Ins2 mice, which suggests that VEGF is a potential target for attenuating aggravated hemorrhage following delayed reperfusion in diabetic stroke.

MMP-2, which can destroy vascular integrity (Liu et al., 2012) and induce hemorrhagic transformation upon ischemia in mice (Lu et al., 2013), has been found to be up-regulated very soon after MCAO in baboons (Heo et al., 1999; Chang et al., 2003). Furthermore, MMP-9 has been associated with hemorrhagic transformation following ischemia in multiple animal and clinical studies (Montaner et al., 2003; del Zoppo, 2010; Elgebaly et al., 2010; Khatri et al., 2012; McBride et al., 2020). In the current study, we observed a premature increase of MMP-2 in Ins2 mice at 2 hours after reperfusion compared with Ins2 mice, in which the increase was only evident at 22 hours after reperfusion. We also observed a potentially higher MMP-9 expression in Ins2 mice at 2 hours after reperfusion, yet this difference was not significant, which may be due to the limited sample size. Together with a greater decline in ZO-1 levels starting at 2 hours after reperfusion, these results indicate that early increased expressions of MMPs in diabetic animals after experimental stroke result in a quicker and more severe compromise of blood vessel integrity and therefore hemorrhagic transformation, which may account for the worse outcomes and earlier mortality in Ins2 mice.

Up-regulation of VEGF has been reported in murine models of transient MCAO starting from 1 hour after reperfusion and persisting for at least 1 day (Hayashi et al., 1997; Shimamura et al., 2006; Yao et al., 2010; Wu et al., 2014), but its role in exacerbating diabetic ischemic stroke has not been well studied. We observed a remarkable and persistent VEGF up-regulation during reperfusion in Ins2 mice, which suggests that diabetic conditions play a role in magnifying VEGF over-expression after prolonged ischemia. The deleterious role of VEGF reported in rodents after stroke includes increasing infarct volume, blood-brain barrier breakdown, and hemorrhagic transformation (Zhang et al., 2000; Kaya et al., 2005). Meanwhile, antagonism of VEGF signaling has been found to significantly reduce brain swelling in both normal adult (van Bruggen et al., 1999) and diabetic mice (Kim et al., 2018) at 1 day after reperfusion. In parallel with worse neurological deficits, earlier mortality, increased hemorrhagic transformation, and blood-brain barrier breakdown, as demonstrated by ZO-1 down-regulation in Ins2 mice, the current study further substantiated the deleterious role of VEGF in exacerbating diabetic ischemic stroke during reperfusion.

Erk1/2 and p38 MAPK are two downstream factors of the VEGF pathway that are responsible for the proliferation and migration of endothelial cells during angiogenesis, respectively, while Akt is another downstream factor promoting endothelial cell survival and vasodilation (Kowanetz and Ferrara, 2006; Melincovici et al., 2018). Inhibiting the MEK-Erk1/2 pathway reduced MMP-9 up-regulation in the middle cerebral artery upon MCAO in rats (Maddahi et al., 2009), while inhibiting p38 activation attenuated hypoxia-induced brain vascular hyperpermeability in mice (Clauss et al., 2001; Issbrucker et al., 2003). At 2 hours after reperfusion, while there was no significant change in the level of p-Akt, we detected significant increases in p-Erk1/2 and p-p38 MAPK expressions in diabetic Ins2 mice only. These persisted at high levels at 22 hours after reperfusion, at which point the expressions in Ins2 mice also increased, but to a lesser extent. These results suggest that the rapidly exaggerated VEGF expression found in diabetic Ins2 mice triggered an earlier and stronger activation of its downstream factors Erk1/2 and p38; this likely caused premature angiogenesis that started immediately after reperfusion instead of 12–24 hours following ischemic insults in normoglycemic rodents and 3–4 days in humans (Beck and Plate, 2009). Therefore, diabetes-aggravated VEGF expression is likely to be responsible for the exaggerated hemorrhagic transformation observed in patients with diabetes during reperfusion after ischemia.

Other than prompting angiogenesis, VEGF, Erk1/2, and p38 MAPK are pro-inflammatory factors that might also aggravate I/R injury via triggering inflammation (del Zoppo, 2009; Hong et al., 2019). VEGF can induce activation and proliferation of microglia (Mosher et al., 2012) and promote recruitment of neutrophils and monocytes/macrophages (Cursiefen et al., 2004; Sinnathamby et al., 2015). Its downstream players, Erk1/2 and p38 MAPK, also trigger inflammation cascades leading to the production of various cytokines and inflammatory responses (Hommes et al., 2003; Kaminska, 2005; Arthur and Ley, 2013). Importantly, Erk1/2- and p38 MAPK-triggered inflammatory cytokines have been linked to aggravation of cerebral I/R injury (Barone et al., 2001; Wang et al., 2004; Maddahi and Edvinsson, 2010). During the acute phase of stroke, uncontrolled and excess inflammation is a major contributor to tissue damage (del Zoppo, 2009; Hong et al., 2019) and is exacerbated under diabetic conditions (Tureyen et al., 2011). Therefore, the early increase in VEGF, p-Erk1/2, and p-p38 MAPK found in Ins2 mice is also suggestive of a diabetes-triggered robust inflammation during reperfusion in response to prolonged ischemia, which, in addition to premature angiogenesis discussed above, could explain the early deteriorating outcomes in diabetes. This should be investigated in future studies.

ER stress has been found to participate in the pathology of I/R injury (Nakka et al., 2010; Xin et al., 2014; Rastogi and Srivastava, 2019; Thiebaut et al., 2019), and inhibiting ER stress could attenuate I/R injury in diabetic rats (Srinivasan and Sharma, 2011a, 2012; Bai et al., 2021). The current study investigated BiP (GRP78) and CHOP, two unfolded protein response (UPR) markers known to alter mRNA levels during ischemia/reperfusion of various tissues (Bilecová-Rabajdová et al., 2010; Li et al., 2014; Noh et al., 2015). BiP is a pro-survival UPR chaperone, while CHOP plays a major role in ER stress-mediated apoptosis, is a major cause of I/R-induced neuronal cell death, and has been linked to more severe ischemic stroke outcomes (Xin et al., 2014; Rastogi and Srivastava, 2019). At 2 hours of reperfusion after prolonged ischemia, while there was similar non-significant increase in mRNA levels of BiP in both genotypes, CHOP expression was higher in diabetic Ins2 mice when compared with normoglycemic Ins2 mice. This indicates that diabetic conditions aggravated ER stress, as observed in the diabetic rat brain (Srinivasan and Sharma, 2011b), heart (Li et al., 2020), and kidney (Zhang et al., 2020) upon I/R injury, and rapidly shifted the BiP-driven pro-survival UPR to CHOP-driven pro-apoptotic UPR (Xin et al., 2014; Nakka et al., 2016; Rastogi and Srivastava, 2019) at 2 hours after reperfusion. Alternatively, the chronic inflammation and higher oxidative stress environment (Liu et al., 2015; Levard et al., 2021) in diabetes may deposit an excess CHOP up-regulation upon I/R injury, although the possible effect of inflammation on ER stress and UTR warrants further investigations (Hotamisligil, 2010; Hasnain et al., 2012).

Given the close link between hypoxia, ER stress, and inflammation (Hotamisligil, 2010; Hasnain et al., 2012; Ramakrishnan et al., 2014; Levard et al., 2021), it is likely that ER stress and CHOP are involved in inducing VEGF expression. Excess ER stress causes an accumulation of reactive oxygen species (ROS) (Malhotra and Kaufman, 2007; Zeeshan et al., 2016), which induces VEGF expression (Kim and Byzova, 2014). Furthermore, CHOP has been found to escalate ROS production during ER stress by both inducing ERO1α expression and resuming protein synthesis (Malhotra and Kaufman, 2007; Nakka et al., 2016). Alternatively, CHOP has been suggested to induce VEGF expression by potentiating the HIF-1α-VEGF pathway as ATF4-siRNA (Pereira et al., 2014) (ATF4 is a transcription regulator of CHOP) and CHOP-siRNA has been found to suppress VEGF transcription in vitro (Hu et al., 2017). Therefore, it is likely that the aggravated CHOP expression is responsible for the rapidly increased VEGF expression found in diabetic Ins2 mice, which resulted in extensive hemorrhagic transformation and probably exacerbated other outcomes upon prolonged ischemia and reperfusion.

Limitations

Although this study postulates the relationship between diabetes-exacerbated hemorrhagic transformation and aggravated VEGF up-regulation during delayed reperfusion, the potential of suppressing the VEGF pathway in attenuating diabetes-exacerbated stroke outcomes was not explored in the present study due to the high mortality of mice following prolonged ischemia. In addition, neurological score was used as a measurement of neurological deficit in this study. Although carried out in a blinded manner, the subjective nature of the assessment could intrinsically bring inaccuracy. More objective methods for assessing neurological deficit, such as the adhesive removal test, should be included in future research.

Conclusion

We demonstrated that diabetic conditions might play a key role in the rapid exacerbation of hemorrhage transformation at the reperfusion phase following prolonged ischemia. We postulate that this rapid exacerbation seen under diabetic, but not normoglycemic conditions, is at least partially underpinned by diabetes-aggravated premature VEGF up-regulation shortly after reperfusion. This has deleterious effects via rapid triggering of vascular disintegration and, hence, hemorrhagic transformation, and this premature VEGF aggravation is associated with diabetes-enhanced CHOP expression.

Exacerbated hemorrhage is a major concern in patients with diabetes/hyperglycemia (Celik et al., 2004) following reperfusion therapies (Reeves et al., 2010; Jiang et al., 2015). This adverse effect contributes to the contraindication of reperfusion therapy in patients with diabetes beyond the therapeutic window. By considering the relationship between aggravated VEGF up-regulation and hemorrhage, this study suggests a potential strategy of using VEGF antagonists in mitigating exacerbated hemorrhage in patients with diabetic stroke following delayed reperfusion therapy. The administration of VEGF antagonists, such as bevacizumab and sunitinib, before reperfusion therapy could attenuate VEGF-induced premature vascular disintegration, thus limiting hemorrhagic transformation and extending the therapeutic window of reperfusion therapy in patients with diabetic stroke for a better clinical outcome. In future research, the effect of VEGF antagonism during delayed reperfusion in diabetic animals should be explored by the use of VEGF antagonists like bevacizumab and sunitinib.

Additional files:

.

.

.

Original blot images for .

Original blot images for (A), (B), (C), (D), (E) and (F). (L#: Lane number; Gp: Group; S#: Sample ID; WT: Ins2+/+; AK: Ins2Akita/+; Sham: Sham control; MCAO: 2h I / 2h R or 2h I / 22h R; Mouse #211 died during reperfusion and was excluded from analysis). I: ischemia; MCAO: middle cerebral artery occlusion; R: reperfusion.

.

Original blot images for .

Original blot images for (A), (B), (C), (E), (F), (G) and (H). (L#: Lane number; Gp: Group; S#: Sample ID; WT: Ins2+/+; AK: Ins2Akita/+; Sham: Sham control; MCAO: 2h I / 2h R or 2h I / 22h R; Mouse #211 died during reperfusion and was excluded from analysis). I: ischemia; MCAO: middle cerebral artery occlusion; R: reperfusion.

Table 1

Number of experiments performed and body weight of mice used

	Ins2+/+ Sham	Ins2Aktia/+ Sham	Ins2+/+ MCAO	Ins2Aktia/+ MCAO	Ins2+/+ Sham	Ins2Aktia/+ Sham	Ins2+/+ MCAO	Ins2Aktia/+ MCAO
	---------------------------------------2h I / 2h R--------------------------				----------------------------2h I / 22h R-------------------------
Number of mice with experiment performed	10	9	24	21	10	10	25	34
Excluded due to incomplete occlusion/reperfusion	–	–	4	1	–	–	1	0
Included	10	9	20	20	10	10	24	34
Body weight before experiment (g)	25.3±0.5	24.1±0.4	26.0±0.4#	24.0±0.4++	24.6±0.5	23.8±0.4	25.1±0.3	23.4±0.2++

This table shows the number of mice used for each group. Mice were randomly drawn for the experiments. In mice with MCAO performed, relative cerebral blood flows were measured using a laser Doppler at 5 minutes before and during ischemia as well as 5 minutes after reperfusion, and animals were excluded from analysis if the measurements indicated any incomplete occlusion or reperfusion. In mice subjected to MCAO, the body weight of Ins2 mice were significantly lower than that of Ins2 mice. #P < 0.05, vs. Ins2 Sham; ++P < 0.01, vs. Ins2 MCAO; one-way analysis of variance followed by Tukey’s HSD post hoc test; mean ± SEM. I: Ischemia; MCAO: middle cerebral artery occlusion; R: reperfusion.

Additional Table 1

Antibodies used in Western blot analysis

Target	Dilution	Cat#	Supplier
α-Tubulin	1:20000	sc-5286	Santa Cruz Biotechnology, Santa Cruz, CA
Actin	1:5000	MAB1501	Merck KGaA, Darmstadt, Germany
MMP-2	1:1000	4022	Cell Signaling Technologies, MA
MMP-9	1:1000	2270	Cell Signaling Technologies, MA
ZO-1	1:1000	40-2300	Zymed Laboratories, South San Francisco, CA
VEGF	1:1000	sc-7269	Santa Cruz Biotechnology, Santa Cruz, CA
TotalAkt	1:1000	9272	Cell Signaling Technologies, MA
p-Akt	1:1500	9277	Cell Signaling Technologies, MA
Total Erk1/2	1:2000	9107	Cell Signaling Technologies, MA
p-Erk1/2	1:2000	9106	Cell Signaling Technologies, MA
Total p38 MAPK	1:1000	9212	Cell Signaling Technologies, MA
p-p38 MAPK	1:1000	9211	Cell Signaling Technologies, MA

MAPK: Mitogen-activated protein kinase; MMP: matrix metallopeptidase; VEGF: vascular endothelial growth factor.

Additional Table 2

Primer sequences used in real-time PCR analysis

Target	Primer sequence (5’ to 3’)	Reference
β-actin
	Forward: GACGGCCAGGTCATCACTATTG	Binet et al.,
	Reverse: CCACAGGATTCCATACCCAAGA	2013
BiP
	Forward: AAGGTGAACGACCCCTAACAAA	Binet et al.,
	Reverse: GTCACTCGGAGAATACCATTAACATCT	2013
CHOP
	Forward: GTTGAAGATGAGCGGGTGGCAGC	Wang et al.,
	Reverse: GCACGTGGACCAGGTTCTGCTT	2012

Binet F, Mawambo G, Sitaras N, Tetreault N, Lapalme E, Favret S, Cerani A, Leboeuf D, Tremblay S, Rezende F, Juan AM, Stahl A, Joyal JS, Milot E, Kaufman RJ, Guimond M, Kennedy TE, Sapieha P (2013) Neuronal ER stress impedes myeloid-cell-induced vascular regeneration through IRE1a degradation of netrin-1. Cell Metab 17:353-371.

Wang Z, Wang H, Xu ZM, Ji YL, Chen YH, Zhang ZH, Zhang C, Meng XH, Zhao M, Xu DX (2012) Cadmium-induced teratogenicity: association with ROS-mediated endoplasmic reticulum stress in placenta. Toxicol Appl Pharmacol 259:236-247.

Table 2

Physiological conditions of Ins2 mice and their wildtype littermates before and after ischemia

	Ins2+/+ MCAO	Ins2Aktia/+ MCAO	Ins2+/+ MCAO	Ins2Aktia/+ MCAO
	-----------2h	I / 2h R-------	--------2h	I / 22h R--------
N number	20	20	24	34
Relative cerebral blood flow (%)
5 min before ischemia	100	100	100	100
During ischemia	11.0±1.3	14.4±1.4	11.7±1.1	15.0±1.1*
5 min after reperfusion	102.2±10.6	118.4±8.8	120.7±11.0 108.5±8.9
Body temperature (°C)
At ischemia	36.9±0.0	36.9±0.0	36.8±0.0	36.8±0.0
At reperfusion	36.9±0.1	37.0±0.0	36.9±0.0	36.9±0.0

Relative cerebral blood flows were measured using a laser Doppler at 5 minutes before and during ischemia as well as 5 minutes after reperfusion. The reading of relative blood flow at 5 minutes before ischemia was set as 100%, and the sequential changes of blood flow were calculated as a reference to this. Relative cerebral blood flows were similar in both Ins2 and Ins2 mice after reperfusion, but they were faster in Ins2 mice during ischemia. *P < 0.05, vs. Ins2 MCAO at 22 h R (22 hours after reperfusion), unpaired Student’s t-test. Data are expressed as the mean ± SEM. I: Ischemia; MCAO: middle cerebral artery occlusion; R: reperfusion.