Male C57BL/6 mice at 10–12 weeks of age (25–30 g, purchased from the Sino-British SIPPR/BK Laboratory Animal Co., Ltd.) were rendered diabetic by intraperitoneal (i.p.) injection of 60 mg/kg STZ (Sigma-Aldrich, St Louis, MO, United States) in 0.05 M sodium citrate (pH 4.5) daily for five consecutive days during the first week of the study (Zauli et al., 2010; Wang et al., 2011). Previous studies reported that estrogens exert neuroprotective effects in an animal model of ischemia (Suzuki et al., 2009; Shi et al., 2024). To avoid the interference of estrogen on ischemic stroke, only male mice were selected in this study. Fasting blood glucose levels were measured from tail blood samples using a OneTouch glucose meter (LifeScan), starting from the initiation of streptozotocin (STZ) injection and monitored weekly thereafter. A 5-day low-dose STZ injection regimen was used to ensure sustained hyperglycemia (Wang et al., 2011). After 14 days, whole fasting blood glucose levels from tail blood were measured, and mice with a blood glucose level>11.1 mmol/L were considered diabetic (Ju et al., 2018). The diabetic mice were randomly divided into two groups, half of the mice were treated daily by oral gavage with AKG (4 g·kg⁻¹·d⁻¹; Sigma-Aldrich, St Louis, MO, United States), and the other half were treated daily with the same amount of vehicle (saline solution) via gavage for 4 consecutive weeks. To ensure consistency in AKG intake among mice, we conducted a preliminary experiment to measure water intake. Based on daily water consumption, AKG at 2% in drinking water corresponds to doses of 4 g·kg⁻¹·d⁻¹ (Yuan et al., 2022; Cheng et al., 2024; An et al., 2021). After 4 weeks of AKG (Sigma-Aldrich) treatment, mice were used for EPCs isolation and assessment, or were subjected to permanent focal cerebral ischemia. Experiments were carried out in a random and blinded fashion. Experimental procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (United States). All animal experiments complied with the ARRIVE guidelines. The animal protocol was planned in compliance with the animal protection, animal welfare, and ethical principles and was approved by the IACUC of the Shanghai Jiao Tong University.

Permanent focal cerebral ischemia was induced in mice according to previously described methods (Dong et al., 2017; Dong et al., 2015). Briefly, mice were anesthetized with 3.5% chloral hydrate at 0.1 mL/10 g body weight by intraperitoneal injection. A 0.5-cm skin incision was made between the left orbit and ear. Then, the left distal middle cerebral artery was exposed through a craniotomy and permanently cauterized above the rhinal fissure. The body temperature was maintained at 37 °C ± 0.5 °C throughout the surgical procedure using a heating lamp and pad. Behavioral assessment (including Body Asymmetry Test and Beam Test) was performed 3 days after middle cerebral artery occlusion, then mice were euthanized, and their brains were immediately removed. Infarct volume was determined by staining with 2, 3, 5-triphenyltetrazolium chloride (TTC, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) and was analyzed with ImageJ software (Dong et al., 2017; Zhu et al., 2008). The experiments were performed in a blind and random fashion.

Additionally, on day 3 after middle cerebral artery occlusion, the mice were anesthetized and the ischemic brains were flushed with PBS and perfused in 4% PFA, before being dehydrated and embedded in paraffin. Serial sections (6 µm thick) were cut in the coronal plane from paraffin-embedded blocks. Every 10th section was processed for immunohistochemical staining. For quantification of angiogenesis, the area of ischemic boundary was immunostained with CD31 and von Willebrand Factor (vWF) antibody (Dong et al., 2017; Zhang et al., 2002; Chen et al., 2005).

Body Asymmetry Test—To measure motor asymmetry, the mice were examined using the elevated body swing test as described previously (Dong et al., 2017; Dong et al., 2015). The mice were examined for head swings while being suspended by their tails. The direction of the swing (right or left) was recorded when the mouse turned its head sideways by approximately at a 10° angle to the body’s midline. After each swing, each mouse was allowed to move freely in a Plexiglas box for at least 30 s before undergoing the next test; the trials were repeated 20 times for each animal. The frequency of the head swings toward the contralateral side was counted and normalized as follows: (number of lateral swings in 20 tests- 10)/10 %× 100%.

Beam Test—Beam walking across a bridge was used to assess motor coordination and balance after stroke injury. The mice were trained for 5 days before the MCA occlusion to traverse a narrow round beam (5 mm in diameter, 900 mm in length) to reach an enclosed escape platform. They were placed on one end of the beam and the latency to traverse the central 80% of the beam toward the enclosed escape platform at the other end was recorded. Data are expressed as mean latency to cross the beam in three trials (Dong et al., 2017; Dong et al., 2015).

BM-EPCs were isolated and cultured according to the published methods (Xie et al., 2010; Peng et al., 2018). Bone marrow-derived mononuclear cells (BMMNCs) were isolated from the mice tibia and femur, seeded in six-well cell culture plates coated with rat vitronectin (1 mg/mL; Sigma-Aldrich, St Louis, MO, United States) at a density of 5 × 10⁶ cells/well, and cultured in endothelial growth medium-2 (EGM-2; Lonza, Walkersville, MD, United States). After 4 days of culture, nonadherent cells were removed and the adherent cells were further cultivated for 3 days. And then EPCs were either used for in vitro studies (including function assays and Western blot analysis).

The EPCs were characterized via flow cytometry for the co-expression of Sca-1 and Flk-1, and through Dil-acLDL/lectin/Hoechst triple staining, following our previously described methodologies (Xie et al., 2010; Peng et al., 2018). After a 7-day culture, the attached cells were labeled with Dil-acLDL (10 μg/mL; Invitrogen) at 37 °C for 4-h, washed three times in PBS, and immersed in 2% paraformaldehyde for 10 min. The cells were then incubated with FITC-labeled Ulexeuropaeus agglutinin (lectin, 10 μg/mL; Sigma–Aldrich) for 1-h. After nuclei staining with a Hoechst stain (5 μg/mL; Sigma–Aldrich), the cells were viewed under an inverted fluorescent microscope (Leica). Pictures were taken of high-power fields (magnification, ×200). Cells demonstrating triple-positive fluorescence of Dil-acLDL, lectin, and the Hoechst dye were identified as EPCs (Xie et al., 2010; Peng et al., 2018) (data not shown). The phenotype of the EPCs was further examined by flow cytometry. EPCs were placed in polypropylene tubes in 100 μL PBS with 1% albumin. They were then stained with FITC-conjugated Sca-1 antibodies (BD Biosciences) and PE-conjugated Flk-1 antibodies (BD Biosciences) for 1 h at 4 °C and washed three times in PBS with 1% albumin. Quantitations of Sca-1+/Flk-1+ cells were performed with a BD Flow cytometer (Xie et al., 2010; Peng et al., 2018) (data not shown).

1 × 10⁴ cells were plated per well of a 96-well plates precoated with vitronectin (1 μg/mL; Sigma-Aldrich, St Louis, MO, United States). After 4-h incubation time, nonadherent BM-EPCs were removed, and residual adherent BM-EPCs were stained with 10 μg/mL of Hoechst (Hoechst 33,258; Sigma-Aldrich, St Louis, MO, United States) and then fixed with 2% PFA. Adherent cells were counted at five random fields at magnification of ×100 per sample. Three wells were measured for each cell sample (Dong et al., 2017).

BM-EPCs were plated at a number of 5 × 10⁴ per well in the upper Boyden’s chamber with M199 (Hyclone, Shanghai, China), while M199 supplemented with 50 ng/mL of VEGF (R&D systems, Minneapolis, MN, United States) and 10% FBS (Bioind, Kibbutz Beit Haemek, Israel) was placed in the lower chamber. After 24-h incubation time, migrated BM-EPCs adhering to the lower membrane were fixed and then stained with 10 μg/mL of Hoechst 33,258 (SigmaAldrich). Then, the number of migrated BM-EPCs were determined by counting the BM-EPCs on the lower side of the membrane under a microscope at magnification ×100. In blinded analyses, for each sample, five images were taken from random fields (Dong et al., 2017).

A number of 5 × 10⁴ BM-EPCs were plated per well of a 96-well plate precoated with growth factor-reduced Matrigel-Matrix (BD Biosciences, San Jose, CA, United States). After 6-h of incubation, images of tube morphology were taken by inverted microscope (Leica), and tube numbers were counted at five random fields at magnification of ×100 per sample (Dong et al., 2017; Marrotte et al., 2010).

O₂ ^·- levels were estimated using the fluorescent probe dihydroethidium (DHE) (Sigma-Aldrich, St Louis, MO, United States), a commonly used cell-permeable dye which is sensitive to O₂ ^·- and may be oxidized to the red fluorescent molecule ethidium. BM-EPCs were washed, collected and resuspended in M199 medium, and then labeled with 1 µM DHE for 40-min at 37 °C in dark. After that, labeled BM-EPCs were washed three times with 5% BSA/PBS and then fixed in 2% PFA. Finally, the labeled BM-EPCs were analyzed by flow cytometry (Xie et al., 2010).

To further confirm whether the EPC-mediated ischemic angiogenesis might relate to the protection of AKG against cerebral ischemic injury in diabetic mice, a total of 1 × 10⁶ bone marrowderived diabetic EPCs and AKG-treated diabetic EPCs in 200 μL PBS were respectively systemically injected into diabetic mice via a tail vein just after middle cerebral artery occlusion, and equivalent volume of PBS was administered to control diabetic mice (Dong et al., 2015). On day 3 after stroke, behavioral tests were performed, and then the ischemic brains were serially cut and stained with 2% TTC for 5-min (55 °C) to determine the infarct area (Dong et al., 2017; Zhu et al., 2008).

Additionally, on day 3 after middle cerebral artery occlusion, the mice were anesthetized and the ischemic brains were flushed with PBS and perfused in 4% PFA, before being dehydrated and embedded in paraffin. Serial sections (6 µm thick) were cut in the coronal plane from paraffin-embedded blocks. Every 10th section was processed for immunohistochemical staining. For quantification of angiogenesis, the area of ischemic boundary was immunostained with CD31 and vWF antibody (Dong et al., 2017; Zhang et al., 2002; Chen et al., 2005).

To check whether the same results observed in EPCs could be obtained in human endothelial cells, we repeated these experiments on HUVECs. HUVECs were purchased from the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in RAPI 1640 medium (Gibco, America) supplemented with 10% foetal bovine serum (FBS) at 37 °C in a CO₂ incubator according to the manufacturer’s instructions. HUVECs were seeded in 6-well plates at a density of 1 × 10⁵ cells/mL to reach 80%–90% confluence. The cell number was manually counted in a hemocytometer. Cells were fasted overnight in serum-free 1,640 medium followed by incubation for 24-h with mannitol (33 mM) or high glucose (HG, 33 mM) in the absence or presence of α-Ketoglutanc acid disodium salt dihydrate (Sigma, 75,892). The glucose level (33 mM) used in the present work was determined according to previous studies (Thum et al., 2007; Eriksson et al., 2012; Xue et al., 2008).

To examine the cytotoxicity of AKG, the viability of cells after treatment with different doses of AKG were assessed via using Cell Counting Kit-8 (CCK-8, Beyotime, Shanghai, China). The cells were cultured in a 96-well plate at a density of 10,000 cells/well and incubated for 24-h. When the cells were to 90% confluence, AKG was added to different groups at final concentrations of 0.05, 0.1, 0.5, 1, 2, or 8 mM. After 24-h, the original medium was replaced with 100 μL medium containing 10% (v/v) CCK-8 and incubated with an additional 1-h. The optical density (OD) values at 450 nm were measured using a microplate reader (Yang et al., 2023).

1 × 10⁴ cells were plated per well of a 96-well plates precoated with vitronectin (1 μg/mL; Sigma-Aldrich, St Louis, MO, United States). After 4-h incubation time, nonadherent HUVECs were removed, were removed, and residual adherent HUVECs were stained with 10 μg/mL of Hoechst (Hoechst 33,258; Sigma-Aldrich, St Louis, MO, United States) and then fixed with 2% PFA. Adherent cells were counted at five random fields at magnification of ×100 per sample. Three wells were measured for each cell sample (Dong et al., 2017).

HUVECs were plated at a number of 5 × 10⁴ per well in the upper Boyden’s chamber with M199 (Hyclone, Shanghai, China), while M199 supplemented with 50 ng/mL of VEGF (R&D systems, Minneapolis, MN, United States) and 10% FBS (Bioind, Kibbutz Beit Haemek, Israel) was placed in the lower chamber. After 24-h incubation time, migrated HUVECs adhering to the lower membrane were fixed and then stained with 10 μg/mL of Hoechst 33,258 (SigmaAldrich). Then, the number of migrated HUVECs were determined by counting the HUVECs on the lower side of the membrane under a microscope at magnification ×100. In blinded analyses, for each sample, five images were taken from random fields (Dong et al., 2017).

Intracellular superoxide (O₂ ^·-) level was determined using DHE, a membrane-permeable dye which is oxidized to ethidium bromide in the presence of O₂ ^·- (Marrotte et al., 2010; Guerby et al., 2019). After treatment with or without 0.5 mM AKG for 24-h in RPMI 1640 with high glucose, HUVECs were trypsinized and rinsed with RPMI 1640, and then incubated with DHE (10–6 mol/L) for 30-min at 37 °C in dark. After incubation, cells were washed with 5% BSA (w/v) in PBS and re-suspended in 300 μL 2% paraformaldehyde. The DHE fluorescence intensity in cells was determined by flow cytometry.

Western blot analysis was performed as previously described (Zhang et al., 2002). Briefly, BM-EPCs were collected by trypsinization and washed with PBS. Total protein lysates were prepared in lysis buffer containing protease inhibitors (KC-440, KangChen), and centrifuged at 12,000 g for 10-min at 4 °C. Protein concentrations were determined using the Pierce BCA assay kit (Thermo Fisher), and samples containing equal amounts of protein were subjected to 10% (v/v) SDS/PAGE. Proteins were transferred to nitrocellulose membranes, blocked with 5% (w/v) non-fat dry milk (Beyotime) and incubated overnight at 4 °C with the following primary antibodies: MnSOD (1:1,000), CuZnSOD (1:500), TLR4 (1:1,000), NF-κB (1:600), IL-6 (1:1,000), IL-1β (1:1,000) and β-actin (1:5,000). Secondary antibodies were IRDye^® 800CW Goat anti-Rabbit Antibody (1:10,000) and Goat anti-Mouse Antibody (1:5,000). Bands were visualized using Odyssey Imager with Odyssey 1.1 software (Li-Cor) and quantified using NIH image J software.

HUVECs were collected by trypsinization and washed with PBS after treatment with or without 0.5 mM AKG for 24-h in RPMI 1640 with high glucose. Total protein lysates were prepared in lysis buffer containing protease inhibitors (KC-440, KangChen), and centrifuged at 12,000 g for 10 min at 4 °C. Protein concentrations were determined using the Pierce BCA assay kit (Thermo Fisher), and samples containing equal amounts of protein were subjected to 10% (v/v) SDS/PAGE. Proteins were transferred to nitrocellulose membranes, blocked with 5% (w/v) non-fat dry milk (Beyotime) and incubated overnight at 4 °C with the following primary antibodies: MnSOD (1:1,000), CuZnSOD (1:500), TLR4 (1:1,000), p-NF-κB (1:600), NF-κB (1:600), TNF-α (1:500), IL-6 (1:1,000), IL-1β (1:1,000) and β-actin (1:5,000). Secondary antibodies were IRDye^® 800CW Goat anti-Rabbit Antibody (1:10,000) and Goat anti-Mouse Antibody (1:5,000). Bands were visualized using Odyssey Imager with Odyssey 1.1 software (Li-Cor) and quantified using NIH image J software.

CLI-095 is a selective TLR4 antagonist that blocks the TLR4 signaling pathway in mice (Zhang et al., 2022). MD2-TLR4-IN-1 inhibits TLR4 activation by binding to its co-receptor MD2, thereby preventing ligand-induced (e.g., LPS) TLR4 signaling (Liu et al., 2019). JSH-23 selectively inhibits NF-κB by preventing nuclear translocation of the p65 subunit, thus suppressing transcription of NF-κB target genes (Shin et al., 2004; Kumar et al., 2011). To further examine whether AKG exerted anti-inflammatory roles by inhibiting the TLR4/NF-κB pathways, antibodies against TLR4 intracellular binding domain (CLI-095), extracellular binding domain (MD2-TLR4-IN-1) and a selective NF-κB inhibitor (JSH-23) were added at final concentrations of 10 μmol/L. Briefly, HUVECs were pretreated with or without CLI-095, MD2-TLR-4-IN-1 or JSH-23 for 1-h before incubation with or without 0.5 mM AKG for 24-h in RPMI 1640 with high glucose for 24-h (Janovec et al., 2021; Oh et al., 2022). Then, the cells were harvested by trypsin digestion and centrifugation for Western blotting analysis.

Data were expressed as mean ± SEM. One-way ANOVA followed by Newman-Keuls post hoc analysis was used for comparison of variables in more than two groups. A value of P < 0.05 was considered statistically significant.

A 5-day low-dose STZ injection regimen was used to ensure sustained hyperglycemia (Liu et al., 2018). Five consecutive intraperitoneal daily injections of STZ (60 mg·kg⁻¹·d⁻¹) were administered to male C57BL/6 mice, after 14 days, whole blood glucose levels were measured with a OneTouch meter (LifeScan) and mice with a blood glucose level >11.1 mmol/L were considered diabetic. Half of the diabetic mice were treated daily by oral gavage with AKG (4 g·kg⁻¹·d⁻¹) for four consecutive weeks, and the other half were treated daily by oral gavage with vehicle (saline solution) for four consecutive weeks. As shown in Figure 1 significant increase in fasting blood glucose levels and a considerable reduction in body weight were observed in the STZ-treated mice. However, 4 weeks treatment with AKG had no obvious effect on fasting blood glucose levels and body weight in diabetic mice.

After 4-week of treatment, mice were subjected to cerebral ischemia (Figure 2A). Compared to control mice, diabetic mice exhibited more severe cerebral ischemic injury (+28.3%, p < 0.05) and higher neurological deficits at 3 days after cerebral ischemia. However, the infarct volume was significantly reduced (−41.4%, p < 0.01) and the neurological deficits were improved in AKG treated diabetic mice compared with diabetic mice (Figures 2B–E). These results indicate that AKG treatment could protect diabetic mice against cerebral ischemic injury.

Angiogenesis is the fundamental processes during formation of new blood vessels after vascular injury, following ischemic stroke (Liman and Endres, 2012). Therefore, local angiogenesis was assessed by immunostaining at 3 days after cerebral ischemia using an antibody against the angiogenesis marker, i.e., CD31 (Figure 2F). It was found that the number of CD31⁺ microvessels in the boundary area of ischemic brains was reduced in diabetic mice compared with control (−37.2%, p < 0.05), which was significantly increased in 4-week AKG treated diabetic mice (+38.3%, p < 0.05 vs. Diabetes) (Figure 2F). ECs specifically produce vWF, a marker that stains newly formed vessels (Yu et al., 2019). To confirm whether angiogenesis is activated, we evaluated another angiogenic marker vWF to validate this study. It was found that the number of vWF microvessels in the boundary area of ischemic brains was reduced in diabetic mice compared with control (−46.5%, p < 0.01), which was significantly increased in 4-week AKG treated diabetic mice (+42.9%, p < 0.01 vs. Diabetes) (Figure 2G).

Diabetic cardiovascular complications are frequently associated with EPCs dysfunction and reduced EPC-mediated angiogenesis in response to ischemia (Loomans et al., 2004; Tepper et al., 2002). Therefore, after 4 weeks of treatment, the BM-EPC functions were determined in diabetic mice (Figure 3A). Results showed that the adhesion, migration and tube formation functions were significantly impaired in diabetic BM-EPCs, which were rescued by AKG treatment (p < 0.01, Figures 3B,D; p < 0.05; Figure 3C).

To investigate the potential mechanisms underlying AKG protecting BM-EPCs functions, the expression of MnSOD, CuZnSOD and the levels of intracellular O₂ ^·- were examined in diabetic mice after 4-week AKG treatment (Figure 4A). Compared to control, both MnSOD and CuZnSOD were significantly decreased (−19.2%, p < 0.01 and −28.1%, p < 0.05 respectively) in BM-EPCs from diabetic mice. BM-EPCs from AKG treatment diabetic mice showed higher levels of both MnSOD and CuZnSOD compared to diabetic mice (+23.0%, p < 0.01 and +24.7%, p < 0.05) (Figure 4B). As shown in Figure 4C, intracellular O₂ ^·- level of BM-EPCs from diabetic mice was increased ∼50% compared to control group (+48.4%, p < 0.01). However, it was significantly attenuated in BM-EPCs from AKG treated diabetic mice when compared with that of the diabetic mice (−45.3%, p < 0.01).

To investigate the potential mechanisms underlying AKG protecting BM-EPCs functions, the expression of TLR4, NF-κB, IL-6 and IL-1β were examined in diabetic mice after 4-week AKG treatment (Figure 4D). Compared to control, the expression of TLR4, NF-κB, IL-6 and IL-1β were significantly increased (+35.4%, p < 0.01; +35.9%, p < 0.05; +40.1%, p < 0.05 and +31.1%, p < 0.05) in BM-EPCs from diabetic mice. BM-EPCs from AKG treatment diabetic mice showed lower levels of TLR4, NF-κB, IL-6 and IL-1β compared to diabetic mice (−32.6%, p < 0.01; −59.7%, p < 0.01; −52.5%, p < 0.01 and −30.9%, p < 0.05).

To determine whether AKG can enhance the therapeutic effect of diabetic BM-EPCs on cerebral ischemic injury reduction and whether the EPC-mediated ischemic angiogenesis may contribute to the protection of AKG against cerebral ischemic injury in diabetic mice, 1 × 10⁶ bone marrow-derived EPCs from normal mice, diabetic mice and diabetes mice with AKG treated were injected via tail vein into normal mice immediately after middle cerebral artery occlusion, and control mice received equal volume of vehicle (PBS) (Figure 5A).

As shown in Figures 5B–E, normal BM-EPCs, diabetic BM-EPCs and AKG treated diabetic BM-EPCs transplantation attenuated infarct volume in mice (−44.3%, p < 0.01; −21.6%, p < 0.01 or −50.7%, p < 0.01) and the corresponding neurobehavioral outcomes were significantly improved in the three groups of EPC-treated mice compared with controls. What is more, AKG-treated diabetic BM-EPCs showed a stronger therapeutic efficacy against cerebral ischemic injury than diabetic BM-EPCs. Furthermore, the angiogenesis in ischemic brain was assessed on day 3 after middle cerebral artery occlusion. It was found that the number of CD31⁺ microvessels in the boundary area of ischemic brains was significantly increased in the three groups of EPC-treated diabetic mice compared with controls (+54.9%, p < 0.01; +20.6%, p < 0.05 or +62.6%, p < 0.01). However, AKG-treated diabetic BM-EPCs exerted a markedly stronger effect on local angiogenesis promotion compared to the diabetic BM-EPCs (+42.0%, p < 0.05) (Figure 5F).

These findings collectively demonstrated that the transplanted EPCs could promote local neovascularization capacity and protect against the cerebral ischemic injury in diabetic mice, and the AKG increased the therapeutic effect of EPCs from diabetic mice on angiogenesis promotion and cerebral ischemic injury reduction.

To check whether the same results observed in EPCs could be obtained in human endothelial cells, we repeated these experiments on HUVECs.

To evaluate the effects of different concentrations of AKG on HUVECs viability, a CCK-8 assay was performed. It was found that cell viability was not affected after incubating with 0.05, 0.1, 0.5, 1, 2 and 8 mM AKG for 24 h. Based on this, 0.05, 0.5 and 1 mM AKG were selected in the functional studies.

To determine whether AKG exerted direct beneficial effects in protecting HUVECs functions, we further performed cell functions, HUVECs were treated with vehicle (M199), high glucose (33 mM) or high glucose with additional AKG (0.05, 0.5 and 1 mM) respectively for 1 day. Equimolar mannitol addition controlled for any potential effect of increased osmolarity under high-glucose conditions. It was found the adhesion function was significantly impaired in high-glucose-treated HUVECs, which was rescued by AKG treatment (0.5 mM for 1 day, p < 0.01, 1 mM for 1 day, p < 0.05), but was not affected after incubating with 0.05 mM AKG (Figure 6B). As a control for potential osmotic effects of added high glucose, equimolar mannitol produced no impairment of adhesion function in HUVECs (Figure 6B). Hence, we next selected the concentration of 0.5 mM for the subsequent in vitro experiments. Transwell migration assay showed that the migration number was also significantly increased after treating with 0.5 mM AKG compared with HG (0.5 mM for 1 day, p < 0.01) (Figure 6C).

Since SOD is an important antioxidant enzyme which can clear the O₂ ^·- and protect cells from oxidative damage, the expression levels of MnSOD and CuZnSOD were assessed. Compared to control, both MnSOD and CuZnSOD were significantly decreased in HUVECs in HG (−28.0%, p < 0.05; −26.5%, p < 0.01), while AKG increase the expression levels of MnSOD and CuZnSOD from HG (+26.2%, p < 0.05; +19.0%, p < 0.05). These results suggest that the increase of MnSOD and CuZnSOD may contribute to the therapy mitigated the adhesion incapacity of HUVECs in high glucose by AKG (Figure 7A). As shown in Figure 7B, intracellular O₂ ^·- level of HUVECs from HG was increased ∼40% compared to control group (+38.6%, p < 0.01). However, it was significantly attenuated in HUVECs from AKG treated when compared with HG (−41.6%, p < 0.01).

To more thoroughly assess whether AKG suppressed inflammatory responses through TLR4, TLR4 signaling inhibitors (CLI095 and MD2-TLR-4-IN-1) were incubated with HUVECs. Cell adhesion assay showed that compare with HG, CLI-095 can increase the number of adherent cells (+38.6%, p < 0.01). AKG-induced adhesion function improvement vanished after pretreatment with the TLR4 signaling inhibitor CLI-095 in high glucose-incubated HUVECs (Figure 8A). MD2-TLR-4-IN-1 does not affect the number of adherent cells. AKG-induced adhesion function improvement also disappeared after pretreatment with the TLR4 signaling inhibitor MD2-TLR-4-IN-1 in high glucose-incubated HUVECs (Figure 8A). The expression levels of TLR4, NF-κB, p-NF-κB, IL-6, TNF-α and IL-1β, were examined through Western blotting. Comparing HG to the Con (+29.9%, p < 0.05; +54.7%, p < 0.05; +52.3%, p < 0.01; +38.4%, p < 0.05; +50.7%, p < 0.05 and +41.9%, p < 0.05 respectively), there was a substantial rise in the expression of TLR4, NF-κB, p-NF-κB, IL-6, TNF-α and IL-1β intracellular levels, while AKG remarkably attenuated these effects (−35.2%, p < 0.01; −72.2%, p < 0.01; −34.5%, p < 0.01; −35.6%, p < 0.05; −53.9%, p < 0.01 and −65.7%, p < 0.01 respectively) (Figure 8B). These studies showed that, in HG-stimulated HUVECs, AKG reduced the expression of TLR4 and pro-inflammatory cytokines. AKG’s effect on the inflammatory cytokine also disappeared after pretreatment with the TLR4 signaling inhibitor MD2-TLR-4-IN-1 in high glucose-incubated HUVECs. These findings imply that TLR4 may contribute to AKG’s ability to reduce HG-induced inflammation in HUVECs.

To further investigate whether AKG inhibited pro-inflammatory cytokine production via NF-κB signaling pathway, NF-κB signaling inhibitors (JSH-23) were incubated with HUVECs. AKG-induced adhesion functional improvement vanished after pretreatment with the NF-κB signaling inhibitor JSH-23 in high glucose-incubated HUVECs (p < 0.05; Figure 9A). The expression levels of p-NF-κB, IL-1β, IL-6 and TNF-α, were examined through Western blotting. Comparing HG to the Con, there was a substantial rise in the expression of p-NF-κB, IL-1β, IL-6 and TNF-α intracellular levels (+41.3%, p < 0.01; +44.2%, p < 0.01; +48.2%, p < 0.05 and +53.1%, p < 0.01, respectively), while AKG remarkably attenuated these effects (−35.7%, p < 0.05; −33.5%, p < 0.05; −58.7%, p < 0.01 and −38.2%, p < 0.05) (Figure 9B). These studies showed that, in HG-stimulated HUVECs, AKG reduced the expression of p-NF-κB, IL-1β, IL-6 and TNF-α. Treatment with inhibitors of NF-κB signal transduction (JSH-23) and AKG all suppressed HG-induced overexpression of p-NF-κB, IL-1β, IL-6 and TNF-α in HUVECs (p < 0.05 and p < 0.01). Additionally, this protective effect disappeared after NF-κB signaling inhibitor JSH-23 addition. These findings imply that NF-κB may contribute to AKG’s ability to reduce HG-induced inflammation in HUVECs.