Mito-TEMPO, a Mitochondria-Targeted Antioxidant, Improves Cognitive Dysfunction due to Hypoglycemia: an Association with Reduced Pericyte Loss and Blood-Brain Barrier Leakage

Hypoglycemia is associated with cognitive dysfunction, but the exact mechanisms have not been elucidated. Our previous study found that severe hypoglycemia could lead to cognitive dysfunction in a type 1 diabetes (T1D) mouse model. Thus, the aim of this study was to further investigate whether the mechanism of severe hypoglycemia leading to cognitive dysfunction is related to oxidative stress-mediated pericyte loss and blood-brain barrier (BBB) leakage. A streptozotocin T1D model (150 mg/kg, one-time intraperitoneal injection), using male C57BL/6J mice, was used to induce hypoglycemia. Brain tissue was extracted to examine for neuronal damage, permeability of BBB was investigated through Evans blue staining and electron microscopy, reactive oxygen species and adenosine triphosphate in brain tissue were assayed, and the functional changes of pericytes were determined. Cognitive function was tested using Morris water maze. Also, an in vitro glucose deprivation model was constructed. The results showed that BBB leakage after hypoglycemia is associated with excessive activation of oxidative stress and mitochondrial dysfunction due to glucose deprivation/reperfusion. Interventions using the mitochondria-targeted antioxidant Mito-TEMPO in both in vivo and in vitro models reduced mitochondrial oxidative stress, decreased pericyte loss and apoptosis, and attenuated BBB leakage and neuronal damage, ultimately leading to improved cognitive function.


Introduction
In type 1 diabetes mellitus (T1DM), in the context of insulin deficiency, hypoglycemia may trigger cognitive impairment [1]. A study that included 718 older patients with T1DM found significantly lower global cognitive scores, language, executive function, and memory for recent episodes of severe hypoglycemia, suggesting that the cognitive status of patients with T1DM is strongly associated with a history of severe hypoglycemia [2]. However, the mechanisms by which hypoglycemia causes cognitive dysfunction have not been fully elucidated.
Previous studies have shown that neuronal death is caused, to some extent, by hypoglycemia itself [3]. Clinically, the first treatment for patients with T1DM who develop severe hypoglycemia is a rapid increase in blood glucose levels, known as "glucose reperfusion," which is essential for saving lives. However, glucose reperfusion allows patients to recover rapidly from severe hypoglycemia, leading to secondary damage that can precipitate more severe neuronal death [4], which is referred to as "glucose reperfusion injury." The mechanisms by which this injury leads to neuronal cell death may include the production of reactive oxygen species (ROS), disruption of the blood-brain barrier (BBB), and activation of inflammatory factors [5]. Thus, diabetic neuropathy is probably caused by oxidative stress induced by glucose reperfusion injury.
Our previous study reported that glucose reperfusion after severe hypoglycemia in diabetes can increase BBB leakage and pericyte loss, leading to cognitive dysfunction [6]. Therefore, protecting pericyte function and reducing BBB leakage may be key targets for the treatment of cognitive dysfunction. Interventions to reduce vascular oxidative stress have been shown to restore microvascular function and cognitive performance [7], and more strategies, such as aerobic exercise [8] and resistance training [9], can successfully achieve cognitive improvement in chronic diseases and diabetes. One study demonstrated that the antioxidant edaravone protects microvascular pericytes from apoptosis, which is essential for the induction and maintenance of the BBB [10]. The study hints that it is possible to find a suitable antioxidant to rescue brain damage caused by glucose reperfusion.
Mito-TEMPO is a mitochondria-targeted antioxidant that acts in ischemic tissues [11]. Previous studies by our team have confirmed that Mito-TEMPO can attenuate severe hypoglycemia-induced cardiac damage [12]. In terms of brain damage, animal experiments have shown that Mito-TEMPO not only inhibited mitochondrial ROS production and rescued mitochondrial respiratory function but also effectively reduced the accumulation of tau oligomers in mouse cortical neurons and improved cognitive function [13]. In vitro tests have shown that Mito-TEMPO protects neurons from bupivacaine-induced toxic damage [14].
Therefore, this study is the first to investigate whether the action of glucose reperfusion after severe hypoglycemia on cognitive dysfunction is related to pericyte dysfunction and BBB leakage caused by oxidative stress at both animal and cellular levels. We also aimed to determine whether the mitochondria-targeted antioxidant Mito-TEMPO could reverse these disruptions. The results of this study will help elucidate the mechanisms underlying the relationship between severe hypoglycemia and cognitive dysfunction. Simultaneously, it will provide new ideas and methods for the prevention and treatment of diabetic brain injury and the development of novel drugs.

Experimental Animals and Treatment
Male C57BL/6J mice (20-25 g) were purchased from Shanghai Slacker Laboratory Animal Co., Ltd (Shanghai, China) and randomly divided into normal (NC, n = 25), diabetic (DM, n = 25), severely (DH, n = 25), and severely + Mito-TEMPO-treated (DHT, n = 25) groups. The DHT group was treated with intraperitoneal (i.p.) injection of 0.7 mg/kg/day Mito-TEMPO (SML0737; Sigma-Aldrich; 3 or 10 days treatment, Figure S1) [12,15]. The DH group was injected with the same amount of normal saline. T1D was induced with a single i.p. injection of streptozotocin (150 mg/kg; STZ, S0130, Sigma-Aldrich, St Louis, MO, USA). On the 3rd day post-injection, mice had blood glucose levels of >16.7 mmol/L and presented with polyuria, polydipsia, polyphagia, and wasting, suggesting successful induction of T1D. After fasting overnight, DH and DHT mice were induced with a single i.p. injection regular insulin (15 mU/g; Wanbang, Jiangsu, China) to maintain glucose levels of < 2.0 mmol/L for 90 min [12]; blood glucose levels were assayed from the tail vein blood [16]. To terminate episodes, mice were allowed to inject glucose (1 mg/kg i.p.) after severe hypoglycemia to ensure that their blood glucose level was > 10 mmol/L. All experimental procedures were performed in accordance with the Chinese Society for Experimental Animal Care and Use guidelines and were approved by the Fujian Animal Research Ethics Committee (approval number: FJMU IACUC 2021-0029).

Morris Water Maze (MWM) Task
After 24-h severe hypoglycemia modeling, 60 mice from all the four groups (NC, DM, DH, and DHT; n = 15 per group) were sacrificed for histological testing. We tested cognitive function by performing a water maze test on the remaining mice (NC, DM, DH, and DHT; n = 10 per group) 1 week after induction of severe hypoglycemia ( Figure S1). Subsequent assessment of cognitive function allows for better measurement of clinical outcomes and assessment of neuroprotection, as it considers a complete and comprehensive assessment of ongoing brain damage and possible recovery from severe hypoglycemia [17,18]. In the hidden platform experiment, a transparent platform of approximately 7 cm in diameter is fixed 1.5 cm below the water surface. The mice's swimming trajectory was tracked by a video camera. In the hidden platform experiment, mice were trained to swim four times a day for 5 days. The time it took for the mice to find the hidden platform was the latency period of the hidden platform. On day 6, a memory retention experiment was performed. The camera tracking system recorded the time spent swimming in each quadrant, the number of times the mouse crossed the original platform, and the number of swimming paths.

Hematoxylin-Eosin Stain
After 24-h of severe hypoglycemia modeling, some of the mice were euthanized for histological testing. Mice were anesthetized using an intraperitoneal injection of 2% sodium pentobarbital (2 ml/kg), and the fresh brain tissue was removed and fixed using a fixative for 24h. The tissue was dehydrated and waxed, followed by embedding and sectioning. This tissue was subjected to hematoxylin and eosin staining, followed by dehydration sealing of the sections in sequence, microscopic examination, image acquisition, and analysis.

Nissl Staining
Nissl-stained specimens were embedded in paraffin and cut with a microtome (RM2016, Leica, Wetzlar, Germany) to obtain 4-mm thick sections. Tissue slices were treated with toluidine blue for 2-5 min, rinsed with tap water, and treated with 1% glacial acetic acid. The degree of differentiation was determined using a microscope. The samples were then washed with tap water, dried in an oven, cleared in xylene for 10 min, and sealed with neutral gum. Microscopy inspection, image acquisition, and analysis were observed.

Evaluation of BBB Permeability and Measurement of Brain Water Content
Evans Blue is a commonly used azo dye. When the BBB structure is disrupted, plasma proteins bound to Evans Blue can penetrate the barrier into the tissue interstitial space, and this assay is commonly used to test the structural integrity of the BBB [19]. Evans blue solution (2% concentration) (Sigma, E2129, 4 ml/kg) was injected into the tail vein and was allowed to circulate in mice for 6 h. After heart perfusion with normal saline, the brain tissue was taken out and photographed with a camera. Afterward, the mouse brain tissue was homogenized in 50% trichloroacetic acid solution (Sigma, T5159), and Evans Blue was quantified in the heads by measuring the fluorescence intensity (620 nm excitation wavelength, 680 nm emission wavelength).
The water content of brain tissue was used to assess the degree of brain edema and the BBB functional disruption [20]. Mice were sacrificed by anesthesia and weighed to obtain wet weights (W). They were then oven-dried (65°C, 72 h) and weighed to obtain a dry weight (D). The brain water content was calculated using the following formula: brain water content (%) = (W − D) / W × 100%.

Transmission Electron Microscopy
The CA1 region of the hippocampus was identified as the site of extraction and a culture dish with an electron microscope fixative was prepared in advance. The small tissue blocks were removed in vitro and immediately placed into the culture dish containing fixative and cut into 1 mm 3 piece using a scalpel. The small tissue blocks were then transferred to EP tubes containing a new electron microscope fixative for further fixation at 4°C. The samples were rinsed three times using 0.1 M phosphate buffer PB (PH 7.4), followed by fixation, room temperature dehydration, osmotic embedding, polymerization, ultra-thin sectioning, and staining. The samples were subjected to transmission electron microscopy, and the collected images were analyzed.

Brain Frozen Section ROS Assay
Frozen sections of mouse brain were rewarmed at room temperature and dried. These frozen sections were incubated with ROS staining solution containing DHE probe (Sigma, D7008, 1:500) at 37°C for 30 min in a dark place. The nuclei were stained with DAPI and sealed. The sections were then observed under a fluorescence microscope. The nuclei of the DAPI-stained cells were blue under ultraviolet excitation (excitation wavelength 330 nm, emission wavelength 470 nm) and red for ROS-positive expression (excitation wavelength 510 nm, emission wavelength, 590 nm). The fluorescent images were observed under a conventional fluorescence microscope (Dmi8; Leica) and collected. The Image Pro Plus 6.0 software was used to calculate red fluorescence intensity.

Enzyme-Linked Immunosorbent Assay (ELISA)
The levels of ROS, malondialdehyde (MDA), and total superoxide dismutase (SOD) in the mouse brain tissue were measured using ELISA. Brain tissue was weighed and added to saline at a ratio of weight (g):volume (ml) = 1:9. The tissue was cut up and homogenized in an ice water bath and centrifuged at 2500-3000 rpm for 10 min; 10% of the supernatant of the homogenate was collected. Repeated measurements were made using ROS, MDA, and SOD kits (Mlbio, Shanghai, China).

Adenosine Triphosphate (ATP) Assay
Adenosine triphosphate (ATP) levels were measured in the mice's brain tissue using the Enhanced ATP Assay Kit (Sigma, MAK190). The ATP levels were determined by adding 20 mg of tissue to 100 μL of lysate, homogenizing, and centrifuging at 12,000 g for 5 min at 4°C. An ATP standard curve was generated, and the ATP levels were measured using a microplate reader (SpectraMax i3x; Molecular Devices, CA, USA).

Cell Culture and Intervention
Human brain microvascular perivascular (HBVP) cells were purchased from ScienCell (Cat. 1200; Carlsbad, CA, USA). When cell density reached 80-90%, cells were digested down using 0.25 % trypsin (Cat. 25200072; Gibco, Waltham, MA, USA) using different pretreatments and divided into 5.5 mM normal glucose group (NC), 35 mM high glucose group (HG), 0 mM glucose deprivation group (GD), and glucose deprivation + Mito-TEMPO intervention group (Mito-T). Cells in the GD group were treated with high glucose for 12 h, followed by glucose deprivation for 24 h, and then retreated with high glucose at 35 mM for 4 h ( Figure S2). The Mito-T group was supplemented with 100 μM Mito-TEMPO based on the premise of the GD group.

Cell Viability
The activity of the HBVP cells was assayed using the CCK-8 Cell Activity Kit (Cat. BS350B; Biosharp, Hefei, China). After each group of cell samples was subjected to the corresponding intervention, 10 μl of CCK-8 solution was added to the 96-well plate, and the cells were incubated for 1 h, while protected from light, and the absorbance was measured at 450 nm using a spectrophotometer (MultiskanGO; Thermo Fisher Scientific, WA, USA).

Fluorescence Detection of Apoptosis
At the end of treatment, the Hoechst staining solution (Cat. 33342; Beyotime), Annexin V-FITC staining solution, and propidium iodide (PI) were added to the medium, and the cells were incubated at 37°C for 15 min, while protected from light. After staining, the cells were carefully washed and observed under a conventional fluorescence microscope (Dmi8; Leica). Apoptotic cells exhibited dense or fragmented nuclei under green fluorescence.

Detection of Mitochondrial ROS in HBVP Cells
Mitochondrial ROS production was detected using the Mito-SOX stain (Cat. 40778ES50; YEASON). After treatment with different concentrations of glucose, the cells were incubated for 30 min at 37°C in the dark using MitoSOX (5 μM ). After washing the cells three times with PBS, the cellular mitochondrial ROS were observed under a fluorescence microscope. The intensity of the red fluorescence was calculated using the Image Pro Plus 6.0 software.

Detection of Mitochondrial Morphology
Mitochondrial morphology was observed using the fluorescent probe MitoTracker Red (Cat. C1048B; Beyotime). After different interventions, the mitochondrial morphology was analyzed using a confocal laser-scanning microscope (Leica SP8, Germany) after incubation for 30 min in the dark using 250 nM of MitoTracker Red.

Mitochondrial Membrane Potential Assay
The mitochondrial membrane potential (MMP) was measured using the MMP Assay Kit for JC-1 (Cat. C2006; Beyotime). Cell culture medium and JC-1 staining solution were mixed in a 1:1 ratio, added to the cells, and the cells incubated at 37°C for 20 min. The supernatant was removed, and an appropriate amount of cell culture medium was added. The cells were then observed under a fluorescence microscope (DMi8; Leica). The formation of J-aggregates by JC-1 indicated a high mitochondrial membrane potential, whereas if JC-1 remained in its monomeric form, it indicated a low mitochondrial membrane potential.

Statistical Analyses
All data were statistically analyzed using SPSS 25.0 and are expressed as mean ± SD. Normality and homogeneity of data were assessed by Shapiro-Wilk and Levene's tests, respectively. Data from the MWM test were analyzed using repeated-measure two-way analysis of variance (ANOVA). One-way ANOVA and LSD (Fisher's least significant difference) tests were used for normally distributed data, while Kruskal-Wallis test was used for non-normally distributed variables. Differences were indicated as statistically significant at P < 0.05.

Mito-TEMPO Improves Cognitive Impairment due to Severe Hypoglycemia
Changes in general indicators in mice are shown in Fig. 1A-C. Compared with the NC group, the mice in the DM, DH, and DHT groups showed a significant decrease in body weight after STZ injection (Fig. 1A), whereas blood glucose increased significantly on day 3 (Fig. 1B).
To verify the effect of severe hypoglycemia on cognitive dysfunction in T1D mice and the ameliorative effect of Mito-TEMPO on cognitive function, the MWM test was administered in each group of mice. No significant differences were found in the swimming speeds of the four groups during MWM test (P > 0.05, Fig. 1D). The escape latency (time spent searching for a hidden platform) for each group of mice is shown in (Fig. 1E). As the number of training days increased, the escape latency increased in the DM group compared to that in the NC group, but no significant difference was observed (P > 0.05), while the escape latency was significantly longer in the DH group (P < 0.01). Escape latency was shorter in the DHT group than in the DH group (P < 0.05). In the spatial exploration experiment with the platform removed ( Fig. 1F-G), mice in the DH group passed through the original platform location significantly less often than those in the NC and DM groups (P < 0.01). The number of times the platform was crossed was higher in the DHT group than in the DH group (P < 0.05). This indicates that glucose reperfusion after severe hypoglycemia in the diabetic state can significantly impair cognitive function in mice, which can be reversed by Mito-TEMPO.

Mito-TEMPO Attenuates Neuronal Damage in Cortical and Hippocampal CA1 Regions Caused by Severe Hypoglycemia
We used hematoxylin and eosin staining to assess the histological alterations of neurons in the cortex and the CA1 region of hippocampus ( Fig. 2A). In the NC group, the structure of the neurons in the cortical and CA1 regions was regular, and 4-5 layers of cells were neatly and closely arranged, surrounded by red cytoplasm. Large round nuclei were visible inside the neurons, with 1-2 nucleoli clearly visible inside the nuclei. Some degree of damage and a decrease in the number of neurons were visible in the DM group, but the overall condition was better than that in the Neuronal damage was observed in each group of mice by Nissl staining. Nissl staining is specific for neurons. As shown in Fig. 2b-d, the number of neurons in the hippocampus and cortex was significantly reduced in the DH group compared with that in the NC group (NC vs. DH, P < 0.01; DM vs. DH, P < 0.05), while the staining of Nisinia vesicles was reduced, light, and blurred, suggesting neuronal damage. After Mito-TEMPO treatment, the number of neurons in the hippocampus and cortex of mice in the DHT group was significantly increased (P < 0.05), while the staining of nisin microsomes was darker and clearer, suggesting a protective effect of Mito-TEMPO on neurons.

Mito-TEMPO Reduces Blood-Brain Barrier Leakage and Brain Edema due to Severe Hypoglycemia
As shown in Fig. 3a, the NC and DM groups showed almost no staining of the mice brain with Evans Blue, whereas the DH group showed significant penetration of Evans Blue dye. Compared with the DH group, the DHT group showed reduced Evans blue leakage. Further quantification of Evans Blue in the brain (Fig. 3B) showed that severe hypoglycemia could lead to BBB leakage in diabetic mice and that Mito-TEMPO could reverse this damage. The water content of the brain tissue of each group of mice was calculated by weighing the weight of the brain tissue before and after drying. The results (Fig. 3C) showed that the water content of the brain in the DH group was significantly higher than that in the NC and DM groups (NC vs. DH, P < 0.01; DM vs. DH, P < 0.05). There was a decrease in the brain water content in the DHT group compared to that in the DH group (P < 0.05).
As shown in Fig. 3D, the BBB was normal in NC mice. Compared to the NC group, the DH group showed severe damage to the BBB, with marked dilatation of the rough endoplasmic reticulum and local degranulation. The capillary lumen was markedly atrophied and collapsed. The basement membrane structure was blurred. The endothelial cells were heavily edematous, with loss of intercellular tight junctions (TJ) and narrowing of the intercellular spaces. Astrocyte footplate was markedly edematous with sparse stroma. At further magnification, the pericytes showed intracellular low electron density edematous areas. The nucleus was irregularly shaped, with an intact nuclear membrane, a slightly widened perinuclear gap, and a heterochromatin border set. The mitochondria were heavily swollen, with lysis of the intra-membrane matrix and reduction and loss of cristae. In contrast, the DM group showed mild damage to these indicators and the BBB and pericytes. After Mito-TEMPO treatment, compared with the DH group, the mice in the DHT group showed significantly reduced BBB damage, an increased number of TJ, reduced pericyte edema, and near-normal mitochondrial morphology.

Mito-TEMPO Attenuates Oxidative Stress and Impaired Mitochondrial Energy Metabolism in the Brain of T1D Mice Caused by Severe Hypoglycemia
Brain tissue ROS immunofluorescence staining images are shown in Fig. 4A, and the ROS fluorescence intensity was significantly increased in the DM and DH groups compared to the NC group (P < 0.01). The brain ROS fluorescence intensity was further increased in mice after severe hypoglycemia (DH group) compared with that in the DM group (P < 0.01). Brain ROS fluorescence intensity was significantly reduced in mice in the DHT group compared to that in the DH group (P < 0.01). The brain ROS, MDA, and SOD levels of mice in each group were measured using ELISA. The results showed that the DH group had the highest ROS and MDA levels, which were statistically significant compared with the NC and DM groups (P < 0.01; Fig. 4B-C), and Mito-TEMPO treatment improved these indices (P < 0.01). In contrast, SOD levels were significantly lower in the DH group than in the NC and DM groups (P < 0.01; Fig. 4D). SOD levels were higher in the DHT group than in the DH group (P < 0.01; Fig. 4D). Evaluation of mitochondrial ATP content in the brain tissue showed that severe hypoglycemia significantly reduced brain mitochondrial ATP production in diabetic mice (NC vs. DH, P < 0.01; DM vs. DH, P < 0.05; Fig. 4E), whereas Mito-TEMPO ameliorated this impairment (P < 0.01).

Mito-TEMPO Improves Pericyte Dysfunction and Loss of TJ Proteins due to Severe Hypoglycemia
As shown in Fig. 5A-C, the expression of mouse pericytespecific proteins PDGFR-β and α-SMA was significantly decreased in the DH group compared to the NC and DM groups (NC vs. DH, P < 0.01; DM vs. DH, P < 0.05), suggesting that severe hypoglycemia could cause a decrease in the number of pericytes. This damage was reversed by Mito-TEMPO (DH vs. DHT, P < 0.05). Similarly, the expression of the tissue TJ proteins occludin and claudin-5 was significantly decreased after severe hypoglycemia compared to that in the NC and DM groups (NC vs. DH, P < 0.01; DM vs. DH, P < 0.05; Fig. 5A, Fig. 5D-E). A greater increase in the expression of occludin and claudin-5 was observed in the DHT group than in the DH group (P < 0.05). The expression of MMP-9, which responds to pericyte inflammatory status, Fig. 3 Mito-TEMPO reduces BBB leakage and brain edema due to severe hypoglycemia. A Naked-eye view of Evans blue exocytosis in mouse brain. B Quantification of Evans Blue extravasation. C Mouse brain water content. D Representative images of hippocampal BBB and pericyte morphology under transmission electron microscopy are shown. Scale bars: 2 μm (left), 1 μm (right). Ast, astrocyte; P, pericyte; N, nucleus; Nu, nucleolus; RER, rough endoplasmic reticulum; Cap, capillary lumen visible as a myelin-like structure; BM, basement membrane; EC, endothelial cell; TJ, tight intercellular junction; M, mitochondria; BBB, blood-brain barrier; SD, standard deviation; NC, normal control (normal glucose); DH, severe hypoglycemia; DM, type 1 diabetes mellitus. N = 3 mice per group. Data are presented as means ± SD. *P < 0.05 vs. NC, **P < 0.01 vs. NC, & P < 0.05 vs. DM, && P < 0.01 vs. DM, # P < 0.05 vs DH, ## P < 0.01 vs DH was higher in the DH group than in the NC and DM groups (NC vs. DH, P < 0.01; DM vs. DH, P < 0.05; Fig. 5A,  Fig. 5F). The expression of the inflammatory factor MMP-9 decreased after treatment with Mito-TEMPO (P < 0.05).

In Vitro, Mito-TEMPO Ameliorates Glucose Deprivation-Induced Decrease in HBVP Cells Viability and Apoptosis
To verify the cytotoxicity of Mito-TEMPO, we tested the effects of different concentration gradients of Mito-TEMPO on the activity of HBVP cells in the NC group. We found that Mito-TEMPO with a concentration of 200 μM could significantly reduce the activity of HBVP cells (P < 0.01, Fig. 6A). It is suggested that the appropriate therapeutic dose of Mito-TEMPO should be less than 200 μM. As shown in the figure (Fig. 6B), the activity of HBVP cells decreased significantly after 24 h of glucose deprivation compared to high glucose, and further decreased after 36 h. Compared with glucose deprivation (Fig. 6C), the cell viability of HBVP cells increased after treatment with 75 μM and 100 μM Mito-TEMPO (P < 0.05, P < 0.01, respectively), but the effect of 100 μM Mito-TEMPO was better than that of 75 μM (P < 0.01). As shown in the figure (Fig. 6D), compared with the NC group (normal sugar group), HBVP cells viability was decreased in the HG group (P < 0.05), the most significant decrease in HBVP cells viability was observed in the GD group (P < 0.01), and Mito-TEMPO improved cell viability (P < 0.01). Hoechst staining showed the effect of different concentrations of glucose in culture conditions Fig. 4 Mito-TEMPO attenuates oxidative stress and impaired mitochondrial energy metabolism in mouse brain caused by severe hypoglycemia. A Immunofluorescent images of brain tissues, including nuclear staining with DAPI and ROS staining with DHE. Scale bars: 1000 μm. B Determination of ROS content in brain tissues using an ROS ELISA kit. C Determination of MDA content in brain tissues using an MDA ELISA kit. D Determination of SOD content in brain tissues using an SOD ELISA kit. E Determination of ATP content in brain tissues using an ATP colorimetric/fluorometric assay kit. ATP, Adenosine triphosphate; ELISA, enzyme-linked immunosorbent assay; ROS, reactive oxygen species; MDA, malondialdehyde; SOD, total superoxide dismutase; NC, normal control (normal glucose); DH, severe hypoglycemia; DM, type 1 diabetes mellitus; SD, standard deviation. N = 3 mice per group. Data are presented as means ± SD. *P < 0.05 vs. NC, **P < 0.01 vs. NC, & P < 0.05 vs. DM, && P < 0.01 vs. DM, # P < 0.05 vs DH, ## P < 0.01 vs DH on the apoptosis of HBVP cells. As shown in the figure (Fig. 6E), the NC group had large nuclei and darker fluorescence with a dark blue color. The number of apoptotic cells in the HG group was between those in the NC and GD groups. After treatment with Mito-TEMPO, apoptosis was reduced in the Mito-T group.

Glucose Deprivation Can Cause Increased ROS Production, Mitochondrial Disruption, and Decreased Mitochondrial Membrane Potential (MMP) in HBVP Cells, Which Can Be Reversed by Mito-TEMPO
As shown in Fig. 7A, red fluorescence was significantly increased in the HG and GD groups compared to the NC group (NC vs. HG, P < 0.05; NC vs. GD, P < 0.01; Fig. 7A-B), but more significantly in the GD group (HG vs. GD, P < 0.05; Fig. 7A-B), suggesting that glucose deprivation caused ROS accumulation in HBVP cells. Red fluorescence intensity was lower in the Mito-T group than in the GD group (P < 0.01; Fig. 7A-B). The effects of different glucose concentrations on the mitochondrial morphology of HBVP cells were compared. As shown in Fig. 7C, the mitochondria in the NC group were round tubes or thin filaments and interlocked to form a complex meshwork. After glucose deprivation, the mitochondria in the GD group were heavily fragmented, with short rods, small circles, and broken dots in a dispersed distribution. Mito-TEMPO intervention reversed mitochondrial breakage in HBVP cells.
JC-1 red fluorescence represents normal membrane potential, green fluorescence represents reduced membrane potential, and a decrease in the ratio of red to green fluorescence represents a decrease in MMP. As shown in Fig. 7D-E, the intracellular red fluorescence intensity of HBVP cells in the GD group was significantly reduced, the green fluorescence intensity was increased, and the aggregate/monomer ratio (i.e., red/green fluorescence ratio) was reduced compared to that in the NC group (P < 0.01). The above indices in the HG group were between the two groups, with statistically significant differences between the groups (NC vs. HG,

Discussion
This study showed that glucose reperfusion after severe hypoglycemia in a T1D mouse model can lead to BBB leakage, pericyte dysfunction, and neuronal damage, causing the onset of cognitive dysfunction, which is consistent with the results of our previous study [6]. Further findings in this study revealed that the above mechanism of injury might be related to the excessive activation of oxidative stress and mitochondrial dysfunction due to glucose reperfusion after hypoglycemia. Interventions using the mitochondria-targeted antioxidant Mito-TEMPO in both in vivo and in vitro models have revealed that it may improve cognitive function in mice by resisting mitochondrial oxidative stress, reducing pericyte loss and apoptosis, and attenuating BBB leakage and neuronal damage.
Severe hypoglycemia is a common and serious complication of insulin therapy in patients with T1DM [21]. The proportion of patients with T1DM with cognitive dysfunction is obviously increased, and one of the reasons may be related to recurrent episodes of hypoglycemia during insulin therapy [1,2]. Studies in rodents have shown that severe hypoglycemia leads to significant brain damage through various mechanisms, including ROS production, mitochondrial permeability shifts, and oxidative DNA damage [22]. Among these, glycemic recovery after hypoglycemia (i.e., glucose reperfusion injury) is thought to be one of the direct causes of exacerbated brain damage in patients with diabetes [5]. In this study, we constructed a T1D mouse model of severe hypoglycemia to observe the effects of severe hypoglycemia on cognitive function in mice experiencing a rapid rise in blood glucose. Our results revealed that glucose reperfusion after hypoglycemia could cause increased levels of oxidative stress and mitochondrial dysfunction in the brain. We found significant damage to hippocampal and cortical neurons in histology, and impaired cognitive function was observed in mice. Intervention with the mitochondria-targeted antioxidant Mito-TEMPO significantly reduced neuronal death and improved cognitive function in mice while reducing mitochondrial ROS production and improving mitochondrial morphology and function. Therefore, the role of oxidative stress in cognitive dysfunction due to hypoglycemia remains unclear.
Oxidative stress is a severe imbalance between ROS and reactive nitrogen species production and antioxidant defenses [23]. Mitochondria are a major source of intracellular ROS [24] and are considered one of the major targets of  [25]. Hypoglycemia, a source of oxidative stress, may lead to neuronal damage in the CA1 region of the hippocampus and accelerate cognitive decline by exacerbating hyperglycemia-induced oxidative stress and inflammation [26]. Among these, glucose reperfusion following severe hypoglycemia is considered a period of marked ROS production and oxidative stress. The scavenging capacity of SOD, an effective free radical scavenger, plays a key role in maintaining the dynamic balance between ROS production and mitochondrial integrity [27]. MDA is a product of the reaction of lipids with oxygen-free radicals, and its content represents the degree of lipid peroxidation. SOD and MDA are important indicators for evaluating oxidative stress in terms of antioxidant and oxidative capacity, respectively [28], and are often used together in the field of research related to oxidative stress. In this study, increased brain tissue ROS and MDA levels were found in the DH group of mice, accompanied by a decrease in SOD activity, confirming that glucose reperfusion after severe hypoglycemia can enhance oxidative stress in the brains of T1D mice.
Several studies have shown the beneficial effects of natural mitochondrial antioxidants, such as Okamoto maple [29] and melatonin [30], in neurodegenerative diseases. During induced acute hypoglycemia, vitamin C infusion (as an antioxidant) may reduce oxidative stress and inflammation in patients with T1DM [31]. Mito-TEMPO is a specific scavenger of the mitochondrial superoxide [15]. Mito-TEMPO was reported to cross the blood-brain barrier and prevent nicotine-induced ischemic brain injury [32], and also Data are presented as means ± SD. *P < 0.05 vs. NC, **P < 0.01 vs. NC, & P < 0.05 vs. HG, && P < 0.01 vs. HG, # P < 0.05 vs GD, ## P < 0.01 vs GD. HBVP, human brain microvascular perivascular; ROS, reactive oxygen species; SD, standard deviation; NC, normal control (normal glucose); HG, high glucose; GD, glucose deprivation improved cognitive function by reducing the accumulation of tau oligomers in the cortical neurons of mice [13]. Therefore, Mito-TEMPO may have therapeutic potential in hypoglycemia-induced brain injury. In this study, intervention with Mito-TEMPO, a mitochondria-targeted antioxidant, reduced ROS and MDA expression, increased SOD activity in mouse brain tissue, and improved neuronal death and cognitive dysfunction in the cortical and hippocampal CA1 regions of mice caused by glucose reperfusion after severe hypoglycemia, a mechanism of action that may be mediated by the targeted scavenging of mitochondrial ROS and attenuation of oxidative stress. Therefore, we suggest that by enhancing antioxidant defenses, for example, through the administration of antioxidants, it may be possible to reduce oxidative stress-induced damage and improve synaptic dysfunction and neuronal damage caused by hypoglycemia and glucose reperfusion.
Brain endothelial cells interact with astrocyte endfeet, pericytes, and basement membranes to form neurovascular units, which are essential components of the BBB [33]. The BBB controls the composition of the neuronal internal environment and is essential for normal neuronal and synaptic functions [34]. Degeneration of the BBB and dysregulation of blood vessels can be detected in the early stages of patients with Alzheimer's disease (AD) [35]. Furthermore, damage to the BBB is now considered one of the key mechanisms leading to diabetic encephalopathy [36]. Hypoxia/ glucose deprivation can lead to mitochondrial dysfunction and a subsequent reduction in ATP production, triggering destruction of the BBB and exacerbating brain damage [37]. Our results show that glucose reperfusion after severe hypoglycemia could reduce ATP content in the brain of mice, and treatment with Mito-TEMPO resulted in a significant increase in ATP content, suggesting an improvement in mitochondrial function. A previous study by our team [6] found that severe hypoglycemia can cause TJ protein deficiency and increased BBB leakage in diabetic mice; however, the exact mechanism has not been clarified. TJ can be disrupted by oxidative stress, and changes in TJ protein levels and/or in the cellular localization/transport of TJ proteins are among the factors contributing to BBB disruption [38]. In this study, mice experiencing severe hypoglycemia and glucose reperfusion were found to have reduced TJ protein expression and increased blood-brain barrier leakage, and electron microscopic findings also showed disruption of the BBB and loss of TJ. This damage was reversed by the Mito-TEMPO intervention. Based on the above, we hypothesized that BBB leakage and TJ loss are related to the oxidative stress caused by glucose reperfusion after severe hypoglycemia. Therefore, two main questions must be addressed: At which target does oxidative stress generated by hypoglycemia primarily cause BBB leakage? What cellular functions need to be focused on?
There is evidence that BBB disruption in patients with AD is associated with pericyte dysfunction and that pericyte loss occurs early in the AD disease process, at the stage of mild cognitive impairment [39]. Pericytes are highly sensitive to oxidative stress and in many diseases, such as diabetes and AD, pericytes are found to be the first NVU (neurovascular unit) cells to die [40]. BBB disruption can be prevented by reducing oxidative stress and protecting pericyte function [41]. MMPs are protein-degrading enzymes that degrade extracellular matrix proteins and are common culprits of oxidative stress-induced BBB damage. MMP-9 is the major MMP that is most closely associated with barrier permeability following oxidative injury [42]. It has been reported that oxidative stress can contribute to the secretion and activation of MMP-9 by pericytes [43], which in turn leads to TJ disruption and increased BBB leakage. Based on the importance of pericytes in the maintenance of BBB structure and function, we focused on pericytes as the cause of BBB leakage due to glucose reperfusion after severe hypoglycemia. In our previous study, we found that hypoglycemia could induce cell loss and increase MMP-9 expression in the perivascular brain of diabetic mice [6]. The brain lacks a glycogen reserve and is highly dependent on glucose [44]; the stored glucose can only last for 30 min. Therefore, in this animal model, under the condition of severe hypoglycemia (< 2.0 mmol/L) for 90 min, the glucose content in the brain may be infinitely close to 0 mmol/L. To further verify whether pericyte injury is induced by glucose reperfusion after severe hypoglycemia, we established a severe hypoglycemia model (complete glucose deprivation) in vitro and found that glucose deprivation followed by re-hyperglucose (simulating glucose reperfusion after severe hypoglycemia in the diabetic state of T1DM) on top of high glucose cultures could cause increased mitochondrial ROS production, mitochondrial disruption, and decreased mitochondrial membrane potential, ultimately leading to apoptosis of HBVP cells. Furthermore, an increase in pericyte numbers and improvement in pericyte function after intervention with Mito-TEMPO were observed in both in vivo and in vitro trials. Therefore, we suggest that mitochondrial oxidative stress induced by hypoglycemia and/or glucose reperfusion leads to pericyte dysfunction and reduced numbers, causing BBB leakage and, ultimately, cognitive impairment.
Chronic hyperglycemia plays an important role in BBB function and cognitive dysfunction in diabetes [45] and may contribute to neuronal damage by increasing polyol pathway fluxes, late glycosylation end-product formation, and oxidative stress [46]. In this study, increased brain ROS production was observed in the T1D group. In vitro, high glucose decreased HBVP cells viability, increased cellular ROS production, caused mitochondrial disruption, and decreased mitochondrial membrane potential. However, no further effects of hyperglycemia on reduced pericyte numbers, BBB leakage, neuronal damage, or cognitive dysfunction were observed. We hypothesize that this is due to the short duration of hyperglycemia in the mouse model used in this study (3 days of hyperglycemia followed by execution for histological testing and 10 days of hyperglycemia followed by cognitive behavioral testing). The acute short duration of hyperglycemia was insufficient to cause damage to the BBB and neurons in mice, leading to significant cognitive impairment, which is not contradictory to the conclusion of the current study that hyperglycemia is an important risk factor for cognitive impairment. Our studies have demonstrated that acute short-term hyperglycemia can cause a rapid increase in oxidative stress (e.g., 3 days of hyperglycemia in mouse experiments resulted in a significant increase in ROS in the mouse brain, and in cellular assays, 40 h of hyperglycemia incubation resulted in a significant increase in oxidative stress indicators in HBVP cells).
This study has some limitations. The model of insulininduced severe hypoglycemia used in this study is less accurate than the use of the glucose clamp in terms of simulating the duration and degree of hypoglycemia. However, the insulin-induced hypoglycemia model has the advantages of being simple, easy to perform, and equally efficient and is now widely used as a test method, second only to the glucose clamp, in studies related to acute and chronic complications associated with hypoglycemia. Another limitation is the absence of pharmacological or genetic approaches to verify the above results and identify the key targets of Mito-TEMPO. Our team will aim to explore these issues in our next study.

Conclusions
In summary, this innovative study investigated the mechanism of the effect of glucose reperfusion after severe hypoglycemia on cognitive dysfunction at both the animal and cellular levels and found that hypoglycemia leads to increased mitochondrial oxidative stress through blood glucose reperfusion, causing pericyte dysfunction and BBB leakage, ultimately leading to neuronal death and cognitive dysfunction. More importantly, the use of the mitochondriatargeted antioxidant Mito-TEMPO reversed these pathophysiological changes and improved cognitive dysfunction caused by glucose reperfusion after severe hypoglycemia by scavenging the overproduction of mitochondrial ROS and improving pericyte mitochondrial function.