**3. Results**

#### *3.1. DCA Inhibits PDK2 after Hypoglycemia*

In previous studies, PDKs were shown to be key regulators in glucose metabolism and to inhibit PDH by phosphorylating the enzyme during brain injury [26,45]. In the present study, we hypothesized that PDKs activation blocks entry of pyruvate into the citrate acid cycle in mitochondria, leading to reduction of ATP formation, and thus, causes neuronal cell death in hypoglycemia. As a result of PDK2 immunostaining, we found that the level of PDK2 significantly increased in the hypoglycemia-induced group compared to the sham group. However, DCA, the inhibitor of PDK2, reduced the level of PDK2, and consequently reduced neuronal cell death (Figure 1A,C).

**Figure 1.** Effects of sodium dichloroacetate (DCA) on hypoglycemia-induced pyruvate dehydrogenase kinase 2 (PDK2) activation and pyruvate dehydrogenase (PDH) reduction. Fluorescent images show the effect of DCA treatment on PDK2 level after hypoglycemia. (**A**) Difference in PDK2 intensity between vehicle- and DCA-treated groups in the vulnerable CA1 after hypoglycemia. Scale bar = 10 μm. (**B**) Difference in PDH intensity between vehicle- and DCA-treated groups in the vulnerable CA1 after hypoglycemia. In normal state, neuronal cells maintain an adequate amount of active PDH, while hypoglycemia causes a significant reduction of active PDH. However, DCA administration recovers PDH activity. Scale bar = 10 μm. Bar graph shows quantification of (**C**) PDK2 or (**D**) PDH intensity in CA1 region. Data are mean ± S.E.M., n = 3 from each group. \* Significantly different from vehicle treated group, *p* < 0.05.

#### *3.2. DCA Increases PDH after Hypoglycemia*

To continually maintain life, cells must use glucose as a fuel. This process is mainly controlled by the enzyme PDH, which regulates the entry of glycolytic products into the citric acid cycle by converting pyruvate into acetyl-CoA in the mitochondria [46]. PDH is usually suppressed by PDHinduced phosphorylation [47]. According to previous studies, pyruvate dehydrogenase activity is reduced in neurodegenerative brain diseases such as Huntington and Alzheimer's [48,49]. Based on these previous results, we conducted PDH staining to investigate if active PDH is similarly inhibited after hypoglycemia and to determine if this results in the loss of neuronal cells. We discovered that the level of PDH significantly decreased in the hippocampal CA1 region in the hypoglycemia-induced group compared with the sham group. In the present study we found that the administration of DCA increased the level of PDH and reduced hypoglycemia-induced neuronal death (Figure 1B,D).

#### *3.3. DCA Decreases Neuronal Death after Hypoglycemia*

Severe neuronal death is caused by hypoglycemia and subsequent glucose reperfusion when estimated at seven days after injury [4]. After hypoglycemia, we performed NeuN staining in order to confirm the number of surviving neurons, and also Fluoro-Jade B (FJB) staining in order to detect degenerating neurons in the hippocampal subiculum (sub), CA1 and dentate gyrus (DG). First, Fluoro-Jade B staining, a selective marker of degenerating neurons, exposed broad hippocampal

neuronal cell death in the subiculum (sub), CA1, and dentate gyrus (DG) after insult. Rats treated with DCA (100 mg/kg, i.v., two days) displayed a significant reduction in hippocampal neuronal death after hypoglycemia (Figure 2A). As demonstrated in Figure 2B, rats given DCA showed reduced FJB (+) neurons in the subiculum, CA1, and DG by 53%, 76%, and 76%, respectively, compared with rats given only saline plus glucose. Moreover, sham-operated groups showed live neurons in the hippocampal subiculum, CA1 and dentate gyrus via NeuN staining. There were no significant differences in the NeuN (+) cell numbers between vehicle- and DCA- treated group. Compared to the sham group, the number of surviving neurons was significantly decreased at 1 week after hypoglycemia. However, the number of surviving neurons in the DCA-treated group was significantly higher than in the vehicle-treated group (Figure 2C). As shown in Figure 2D, rats given DCA showed increased NeuN (+) neurons in the subiculum, CA1, and DG by 35%, 51%, and 35%, respectively, compared with rats given only saline plus glucose.

**Figure 2.** Effects of DCA on hypoglycemia-induced neuronal death. Hypoglycemia resulted in neuronal death in the subiculum (sub), CA1, and dentate gyrus (DG) of hippocampus one week after hypoglycemia. (**A**) Fluoro-Jade B (FJB) positive neuronal cells are observed in the CA1, subiculum, and DG one week after hypoglycemia. Scale bar = 50 μm. (**B**) The number of FJB (+) cells were significantly lower in the DCA-treated group than that in the control group after insult. Data are mean ± S.E.M., n = 8 from each hypoglycemia group. (**C**) NeuN (+) cells show live neurons in the hippocampal subiculum, CA1 and DG. Scale bar = 100 μm. (**D**) The number of NeuN (+) cells were considerably more increased in the DCA-treated group than in the control group after insult. Data are mean ± S.E.M., n = 3 for each sham group and hypoglycemia group. \* Significantly different from vehicle treated group, *p* < 0.05.

#### *3.4. DCA Reduces Hypoglycemia-Induced Oxidative Injury*

Mitochondria are the main sources of ROS production in many disorders such as ischemia, traumatic brain injury, or seizure [50]. Mitochondria taken from brain during hypoglycemia exhibit increased capacity to produce ROS [51]. Our previous study suggested that ROS can be produced by NADPH oxidase activation after hypoglycemia, so the production of ROS causes oxidative stress-induced neuronal cell death in the hippocampus [4]. We conducted 4-hydroxynonenal (4HNE) staining to estimate the degree of oxidative damage after hypoglycemia. Rat brains were immunohistochemically stained with a 4HNE antibody one week after injury to reveal whether oxidative stress had occurred in the hippocampal neurons and whether DCA can reduce this negative effect on neurons. In the sham-operated group, both saline controls and DCA-treated rats demonstrated no di fference in the fluorescence intensity of the 4HNE staining in the subiculum, CA1, and DG. In the hypoglycemia-operated group, 4HNE fluorescence intensity increased in the hippocampus region of the saline only group and but decreased in the DCA-treated group after hypoglycemia (Figure 3A). As demonstrated in Figure 3B, group treated with DCA and then 25% glucose showed an approximately 53% reduction in the intensity of 4HNE in the subiculum, 60% in the CA1, and 56% in the DG compared with the group provided only glucose (Figure 3B). Therefore, this result indicates that DCA can reduce oxidative injury, and thus spare hippocampal neurons from hypoglycemic insult.

**Figure 3.** Administration of DCA decreases oxidative injury after hypoglycemia. Neuronal oxidative injury was determined by 4-hydroxy-2-nonenal (4HNE) (red color) staining in the hippocampal regions seven days after hypoglycemia. ( **A**) Sham-operated groups rarely displayed the 4HNE signal in the hippocampus. DCA decreased the intensity of 4HNE fluorescence in the hippocampus compared with the saline-treated group after hypoglycemia. Scale bar = 100 μm. (**B**) 4HNE fluorescence intensity in the hippocampus. The fluorescence intensity indicates a significant gap between saline- and DCA-treated groups. Data are mean ± S.E.M., n = 5 for each sham group. n = 8 for each hypoglycemia group. \* Significantly di fferent from vehicle treated group, *p* < 0.05.

#### *3.5. DCA Reduces Hypoglycemia-Induced Microglia and Astrocyte Activation*

Another study showed that hypoglycemia induces microglia and astrocyte activation in the cerebral cortex and the hippocampus [52]. Severe hypoglycemia was induced in rats and then brains were harvested one week after insult. We therefore evaluated the degree of microglia and astrocyte activation between four groups: Sham (Vehicle, DCA) and Hypoglycemia (Vehicle, DCA). As a result of the staining, sham-operated groups showed resting microglia and small astrocytes. Also, activated astrocytes have the potential to be harmful because they can express NOS and produce neurotoxic NO [53,54]. Compared with the sham-operated groups, microglia and astrocyte activation in the CA1 increased by approximately 66% and 60%, respectively, in the hypoglycemia-operated group. However, after insult, microglia and astrocyte activation decreased around 56% and 46% in the DCA-treated group compared to the saline-treated group. Therefore, administration of DCA can decrease microglia and astrocyte activation associated with inflammation (Figure 4A–D).

**Figure 4.** Administration of DCA decreases microglia and astrocyte activation after hypoglycemia. To detect microglia activation, we conducted CD11b staining. Hypoglycemia-induced microglia activation triggers an immune response. CA1 is vulnerable to damage by activated microglia. (**A**) The green fluorescence (CD11b staining) shows activated microglia in the hippocampal CA1 area. Sham-operated group shows almost no microglial activation. However, after hypoglycemia, microglia cell number, intensity, and morphology were greatly enhanced in the vehicle-treated group compared to the DCA-treated group. Scale bar = 20 μm. Data are mean ± S.E.M., n = 5 for each sham group, n = 8 for each hypoglycemia group. (**B**) The grade of microglia activation in the CA1. ( **C**) The red fluorescence (GFAP staining) shows activated astrocyte in the hippocampal CA1 area. The sham-operated group shows almost no astrocyte activation. However, after insult, DCA administration prevented astrocyte activation in the CA1. Scale bar = 20 μm. ( **D**) The graph represents the intensity of activated astrocytes in the CA1 region. Data are mean ± S.E.M., n = 3 for each sham group and hypoglycemia group. \* Significantly di fferent from vehicle treated group, *p* < 0.05.

#### *3.6. DCA Prevents Hypoglycemia-Induced Blood-Brain Barrier (BBB) Disruption*

Many other studies reported that blood-brain barrier (BBB) disruption occurs under brain injury conditions such as ischemia, traumatic brain injury, and seizure [55–57]. In severe hypoglycemia, BBB is destroyed as early critical events, because glucose deprivation destroys endothelial cells that make up BBB [58–60]. To determine disruption of the BBB, we conducted IgG staining by detecting the degree of extravasation of serum immunoglobulin G (IgG) with a previously described method [43] and myeloperoxidase (MPO) staining to look for an influx of neutrophils after hypoglycemia. Firstly, IgG produced coronas with a concentration gradient around the blood vessels in the hippocampus area, and IgG staining was observed in broad areas in the brain due to hypoglycemia. Generally, in the sham-operated group, little leakage of IgG occurred, but the group experiencing hypoglycemic status experienced IgG leakage from damaged vessels. Figure 5 shows that a significant di fference was found in the concentration gradient of IgG leakage between the sham and hypoglycemia-induced groups. IgG leakage via BBB disruption was apparent in the vehicle-treated group following hypoglycemia. However, administration of DCA significantly decreased hypoglycemia-induced IgG leakage through BBB disruption in the hippocampus (Figure 5A,B). In additionally, we conducted MPO staining to check BBB permeability increased by the destruction of BBB after hypoglycemia. The increase of BBB permeability causes an influx of neutrophils, and neutrophils are identified through MPO staining. As a result of the MPO staining, sham-operated groups did not show MPO (+) cells. After hypoglycemia, compared with the sham-operated groups, MPO (+) cells were found to be significantly increased, but DCA administration reduced the infiltration of neutrophils in the hippocampus (Figure 5C,D).

**Figure 5.** Administration of DCA decreases blood-brain barrier (BBB) breakdown and an influx of neutrophils after hypoglycemia. ( **A**) Whereas sham-operated group had little leakage of IgG, the hypoglycemia group had a large quantity of IgG leakage when the BBB broke down. However, the DCA–treated group had significantly reduced IgG leakage. Scale bar = 200 μm (4×). (**B**) The quantification of IgG leakage in the whole hippocampus. Data are mean ± S.E.M., n = 5 for each sham group, n = 8 for each hypoglycemia group. Data are mean ± S.E.M., n = 5 for each sham group, n = 8 for each hypoglycemia group. ( **C**) Representative immunofluorescence images show expression of the neutrophil marker MPO in the cortex and hippocampus. DCA prevent Scale bar = 20 μm. ( **D**) The bar graph indicates the number of MPO (+) cells in the cortex and hippocampus. Data are mean ± S.E.M., n = 3 for each sham group and hypoglycemia group. \* Significantly di fferent from vehicle treated group, *p* < 0.05.
