*2.2. Experimental Animals*

We used adult male Sprague-Dawley rats (8 weeks old, 250–350 g, DBL Co., Eumseong, Korea). We accommodated the animals in a continuous temperature- and humidity-controlled environment (22 ± 2 ◦C, 55 ± 5% humidity, and a 12-h light:12-h dark cycle), and fed them with Purina diet (Purina, Gyeonggi, Korea) and water, both provided ad libitum. This study was written in accordance with the Animal Research: Reporting in Vivo Experiments (ARRIVE) guidelines.

#### *2.3. Animal Surgery and Severe Hypoglycemia Induction*

We induced hypoglycemia (HG) using human insulin [5,36]. Before inducing hypoglycemia, rats were fasted for 15–16 h. The next day, we performed intraperitoneal injection of 10 U/kg regular insulin (Novolin-R, Novo Nordisk, Bagsværd, Denmark). We conducted hypoglycemia surgeries under controlled ventilation with a small rodent respirator (Harvard Apparatus, South Natick, MA, USA). We inserted a catheter into the femoral artery to continuously monitor the arterial blood pressure and into the femoral vein for glucose infusion after hypoglycemia. To monitor electroencephalogram (EEG) signals, we pierced a small hole in the skull bilaterally and then inserted a reference needle in the neck muscle. We measured the blood pressure and EEG using the BIOPAC System (BIOPAC System Inc., Santa Barbara, CA, USA) in succession. We maintained the core temperature of animals at 36.5–37.5 ◦C and measured blood glucose levels at 30-min intervals to monitor induction of hypoglycemia.

When insulin reduces blood glucose levels to a level low enough to induce hypoglycemia, an isoelectric state is observed. Arterial blood pressure was sustained between 160 and 200 mmHg during the entire EEG isoelectric period. The severity of hypoglycemia-induced neuronal cell death is strongly related to the duration of the isoelectric (iso-EEG) period. We abruptly ended the hypoglycemic state at 30 min from initiation of iso-EEG [5,36] by suppling 25% glucose solution intravenously for 2 h (2.5 mL/h, i.v.) to recover normal blood glucose levels.

#### *2.4. Injection of DCA*

To ascertain the e ffect of DCA on hypoglycemia-induced neuronal death, the experimental group was divided into 4 groups: Sham (Vehicle, DCA) and Hypoglycemia (Vehicle, DCA). The control group was injected with 0.9% normal saline. Following hypoglycemia, animals were treated with DCA (100 mg/kg, i.v.) once a day for 2 days.

#### *2.5. Brain Sample Preparation*

To estimate the neuroprotective e ffects of DCA, we sacrificed animals 1 week after injury. Animals were deeply anesthetized with urethane injection (1.5 g/kg, i.p.). Brains were transcardially perfused with 500–600 mL of 0.9% saline and this was followed by 600 mL of 4% formaldehyde (FA). After perfusion, we removed the brain and then proceeded to post-fixation in the same fixative solution for roughly 1 h. After that, brains were immersed for cryoprotection in 30% sucrose for 2–3 days until the samples sank to the bottom. After cryoprotection, brain samples were frozen using freezing medium and were cut on a cryostat into 30-μm thicknesses and then kept in storage solution.

#### *2.6. Detection of PDH and PDK2*

To confirm PDH and PDK2 levels, we performed PDH and PDK2 staining with primary antibodies for each. Immunohistochemistry with PDH and PDK2 antibodies (Abcam, Cambridge, UK) was performed. Brain tissues were incubated in monoclonal rabbit antibody to rat PDH (diluted 1:100, ab168379) and a monoclonal rabbit antibody to rat PDK2 (diluted 1:100, ab68164) overnight in a 4 ◦C incubator. After washing, the sections were immersed in secondary antibody (PDH: Alexa Fluor 594-conjugated donkey anti-rabbit IgG secondary antibody, PDK2: Alexa Fluor 488-conjugated donkey anti-rabbit IgG secondary antibody, respectively, both diluted 1:250; Molecular Probes, Invitrogen, Carlsbad, CA, USA) for 2 h at room temperature (RT). The brain tissues were placed on gelatin-coated slides for observation under a microscope. To check the intensity of fluorescence of PDH and PDK2, we used ImageJ (version 1.47c; NIH, Bethesda, MD, USA) and the following steps were executed: select 100 cells in one brain tissue and then click the menu option Analyze → Measurement. (magnification = 80×).

#### *2.7. Detection of Neuronal Death*

To investigate neuronal death, brain samples were cut into 30 μm thicknesses on gelatin-coated slides for staining (Fisher Scientific, Pittsburgh, PA, USA). Fluoro-Jade B (FJB) staining was performed as descried by Hopkins and Schmued [40]. First, the slides were immersed in a basic alcohol solution (100 →70%) and then washed with distilled water. After that, the slides were submerged in 0.06% potassium permanganate for 15 min. Secondly, the slides were dipped in 0.001% Fluoro-Jade B (Histo-Chem Inc., Je fferson AR, USA) for 30 min and washed 3 times for 10 min each in distilled water. After drying, we mounted a cover glass on each slide with a mixture of distyrene, a plasticizer, and xylene (DPX) solution and then observed the signals with a fluorescence microscope using blue (450–490 nm) excitation light. To quantify the result, we selected five coronal brain sections that were gathered from each animal starting 4.0 mm posterior to Bregma and collecting 5 sections at 75 μm intervals. A blind researcher counted the number of degenerating neurons in the same scope (magnification = 10×) of the hippocampal subiculum (900 × 1200 μm), CA1 (900 × 400 μm), and dentate gyrus (900 × 1200 μm) from both hemispheres. The total average number of degenerating neurons from each region was used for statistical analysis.

#### *2.8. Detection of Live Neurons*

To determine the number of live neurons present, we performed NeuN staining with a monoclonal anti-NeuN, clone A60 antibody (diluted 1:500, EMD Millipore, Billerica, MA, USA). We used it as the primary antibody in PBS with 30% Triton X-100 overnight in a 4 ◦C incubator. After washing three times for 10 min with PBS, and then incubating in anti-mouse IgG secondary antibody (diluted 1:250, Vector Laboraorise, Burlingame, CA, USA) for 2 h at RT, then, the samples were washed again. Next, the brain samples were put into the ABC solution (Burlingame, vector, CA, USA) for 2 h at RT on the shaker. After that, the samples were washed repeatedly three time for 10 min. To activate immune responses, we put samples in 0.01 M PBS bu ffer (100 mL) with 3,30-diaminobenzidine (0.06% DAB agar, Sigma-Aldrich Co., St. Louis, MO, USA) and 30% H2O2 (50 μL) for 1 min. After mounting the sample on the slides, we then used an Axioscope microscope to analyze the immunoreaction. A blinded experimenter counted the number of live neurons in a constant area (magnification = 10×) of the hippocampal subiculum (900 × 1200 μm), CA1 (900 × 400 μm), and dentate gyrus (900 × 1200 μm) from both hemispheres. The total average number of live neurons from each region was used for statistical analysis.

#### *2.9. Detection of Oxidative Injury*

To estimate oxidative injury after hypoglycemia, we conducted immunofluorescence staining with 4-hydroxy-2-nonenal (4HNE) antibodies (diluted 1:500, Alpha Diagnostic Intl. Inc., San Antonio, TX, USA) as described previously [4]. Brain sections were incubated in a mixture of polyclonal rabbit anti-HNE anti-serum and phosphate bu ffered saline (PBS) containing 0.3% TritonX-100 overnight in a 4 ◦C incubator. After washing three times for 10 min each with PBS, samples were also immersed in a mixture of Alexa Fluor 594-conjugated goat-anti rabbit IgG secondary antibody (diluted 1:250, Invitrogen, Grand Island, NY, USA) for two hours at room temperature. Finally, the sections were placed on gelatin-coated slides. To measure the oxidative stress signals, we used ImageJ using the following order of commands: click the menu option Image → Adjust → Color threshold and dark background. Then, we modulated brightness according to the intensity of oxidative stress. The image was converted to 8-bit images (magnification = 10×). The fluorescence intensity was expressed as mean gray value.

#### *2.10. Detection of Microglia and Astrocyte Activation*

To semi-quantitatively measure inflammation after injury, we performed CD11b and GFAP staining, a well-described indicator of the inflammatory response. Immunohistochemical staining was conducted with a mouse antibody to rat CD11b (diluted 1:500; AbD Serotec, Raleigh, NC, USA) and of goa<sup>t</sup> antibody to rat GFAP (diluted 1:1000; Abcam, Cambridge, UK) in PBS containing 0.3% TritonX-100 overnight in a 4 ◦C incubator. After rinsing, the sections were immersed in Alexa Fluor 488-conjugated donkey anti-mouse IgG secondary antibody and Alexa Fluor 594-conjugated donkey anti-goat IgG secondary antibody (both diluted 1:250; Molecular Probes, Invitrogen) for two hours at RT. A randomly blinded experimenter then measured the intensity of astrocytes in the CA1 region. In the case of microglia, we used five sections from each animal to estimate scoring within the same CA1 region (magnification = 20×). To quantify microglial activation, a randomly blinded observer performed the measurements. Microglial activation criteria based on the number and intensity of CD11b positive cells and on their morphology are as follows [41,42]. The scores for each criterion are as follows: for *CD11b-positive cells number*, a score of 0 denotes no cells with continuously stained processes are present, score 1 denotes 1–9 cells with continuous processes per 100 μm2, score of 2 denotes 10–20 cells with continuous processes per 100 μm2, and score of 3 denotes >20 cells with continuous processes per 100 μm2; for *Intensity*, a score of 0 denotes no expression, 1 denotes weak expression, 2 denotes average expression, and a score of 3 denotes intense expression. When ImageJ is used, based on the max value 225, a score of 1 has intensity to mean value 1–50, a score of 2 has intensity to 51–100 and the last score of 3 has intensity to 100–max value 225.; and for *Morphology*, a score of 0 denotes 0% have activated morphology (amoeboid morphology with enlarged soma and thickened processes), 1 denotes 1–45% of microglia have the activated morphology, 2 denotes 45–90% of microglia, and a score of 3 denotes >90% of microglia have the activated morphology. Therefore, the total score includes the combination of three individual scores (CD11b-immunoreactive cell number, morphology and intensity), ranging from 0 to 9 (magnification = 20×).

#### *2.11. Detection of BBB Disruption and Neutrophils*

To check the degree of BBB disruption, we evaluated leakage of endogenous serum IgG after hypoglycemia [43]. We conducted IgG staining with anti-rat IgG, which detects serum IgG released when the BBB is disrupted. Brain sections were incubated in a mixture of anti-rat IgG serum (diluted 1:250, Burlingame, Vector, CA, USA) in PBS containing 0.3% TritonX-100 on a shaking shelf for 2 h at RT. After rinsing three times for 10 min each with PBS, we treated ABC solution for two hours at RT on a shaker. The ABC immunoperoxidase was used to reveal IgG-like immunoreactivity [44]. After that, the sections were exposed to 3,3-diaminobenzidine (DAB ager, Sigma-Aldrich Co., St. Louis, MO, USA) in 0.1 M PBS bu ffer to trigger immune responses. After mounting, we observed the immune responses using an Axioscope microscope (Carl Zeiss, Munich, Germany) and quantified extravasation of endogenous serum IgG in the whole brain using the Image J program. The procedure for measuring IgG is as follows: Click the menu option Image →Type →8 bits and then, edit →invert. To measure the area, the menu option Analyze →Measure was selected. Mean gray values (magnification = 4×) were used. In addition, to measure the degree of BBB disruption, we evaluated any influx of neutrophils after hypoglycemia. We conducted myeloperoxidase (MPO) staining with rabbit anti-MPO (diluted 1:100, Invitrogen, Grand Island, NY, USA). The brain sections were soaked in the primary antibody solution with PBS containing 0.3% TritonX-100 overnight in a 4 ◦C incubator. After washing, samples were immersed in a mixture of Alexa Fluor 594-conjugated goat-anti rabbit IgG secondary antibody (diluted 1:250, Invitrogen, Grand Island, NY, USA) for two hours at room temperature. We put the sample on the slides and then, the blinded tester counted the number of neutrophils in the hippocampus from both hemispheres under fluorescent microscope. The total average number of neutrophils from the hippocampus region was used for statistical analysis.
