Acute glucose fluctuation impacts microglial activity, leading to inflammatory activation or self-degradation

Diabetes mellitus is associated with an increased risk of Alzheimer’s dementia and cognitive decline. The cause of neurodegeneration in chronic diabetic patients remains unclear. Changes in brain microglial activity due to glycemic fluctuations may be an etiological factor. Here, we examined the impact of acute ambient glucose fluctuations on BV-2 microglial activity. Biochemical parameters were assayed and showed that the shift from normal glucose (NG; 5.5 mM) to high glucose (HG; 25 mM) promoted cell growth and induced oxidative/inflammatory stress and microglial activation, as evidenced by increased MTT reduction, elevated pro-inflammatory factor secretion (i.e., TNF-α and oxygen free radicals), and upregulated expression of stress/inflammatory proteins (i.e., HSP70, HO-1, iNOS, and COX-2). Also, LPS-induced inflammation was enlarged by an NG-to-HG shift. In contrast, the HG-to-NG shift trapped microglia in a state of metabolic stress, which led to apoptosis and autophagy, as evidenced by decreased Bcl-2 and increased cleaved caspase-3, TUNEL staining, and LC3B-II expression. These stress episodes were primarily mediated through MAPKs, PI3K/Akt, and NF-κB cascades. Our study demonstrates that acute glucose fluctuation forms the stress that alters microglial activity (e.g., inflammatory activation or self-degradation), representing a novel pathogenic mechanism for the continued deterioration of neurological function in diabetic patients.


Primary microglial cultures
Microglial cells were cultured according to a previously established protocol 1 . In brief, after sacrifice, the brain tissues of newborn Sprague Dawley (SD) rats were mechanically removed from the crania. After aseptic removal of the meninges and blood vessels, the brains were dissociated mechanically with scissors. Next, cell aggregations were suspended in 0.25% Trypsin-EDTA solution (10 min, 37℃) and gentle trituration was performed using a pipette.
The digested cells were filtered through an 80-μm pore mesh, pelleted and suspended in 10% FBS/DMEM. The brain cells were cultured on poly-L-lysine pre-coated flasks, and the medium was frequently refreshed at an interval of 2 days until the cell composition contained < 5% neurons. Subsequently, the microglial cells were separated by gently shaking (180 rpm, 5 h, 37℃) the flasks after a confluent monolayer formed (1014 days) and were reseeded into multi-well plates at a density of 2  10 5 cells/ml. These microglia were separately cultured in NG or HG medium and then used for the experiments within 36-48 h, at which point the cells were well attached. The purity was 95-98% as determined by immunocytochemistry for the microglia-specific marker, OX-42.

Cell viability assay
Cell survival was evaluated by the VisionBlue™ Quick Cell Viability Fluorometric Assay Kit (BioVision, Mountain View, CA, USA) according to the manufacturer's instructions. This fluorescence-based single-step cell viability assay utilizes the redox dye (resazurin) which is not fluorescent, but upon reduction by metabolically active cells, it becomes highly fluorescent. Therefore, viable metabolically active cells can be measured. Initially, cells were seeded on 96-well culture plates at a density of 1×10 5 cells/ml in a volume of 200 µl of cell medium. After completing the desired treatments, fresh cell medium containing 1/10 volume of VisionBlue reagent was added to each well, followed by a 37°C incubation for 2 h. The fluorescence (excitation: 540 nm; emission: 590 nm) was measured using a fluorometric microplate reader (FLx800 Fluorescence Reader, BioTek Instruments, Inc.). All data were normalized to background values.

Discussion
Glucose is the major energy source for the brain. The glucose concentration in the brain is approximately 20% of that of plasma glucose concentration 2 . Five main glucose transporters (GLUT 1-5) mediate glucose uptake in the brain and various peripheral tissues 3 . Microglia can express various GLUTs under different circumstances, and previous studies have indicated different GLUTs to be expressed in microglia to ensure sufficient glucose influx to meet the energy demand. Given that GLUT1 and GLUT3 (a main neuronal GLUT) are expressed in many tissues at variable levels and are thought to regulate basal glucose uptake 4 , and that GLUT4 is only expressed in discrete brain areas 5 , the protein levels of GLUT1 were examined to ensure a constant and stable influx of glucose in both NG-and HG-cultured BV-2 cell lines even if glucose levels are low, and to further explore whether primitive GLUT1 expression is disturbed by glucose fluctuations in this study. Regarding the plasma membrane glucosensors, neurons contain the Kir6.2 subunit of the ATP-dependent potassium channel (K-ATPKir6.2), SGLTs 6 , and GLUT2 7-9 ; additionally, glial cells express glucosensors, including the GLUT2 and Kir6.2 subunit of the K-ATP channel 7,8 . Although these brain cells also express other glucose transporters, they are currently not regarded as sensors.
Interestingly, defects in hypothalamic and brainstem membrane glucosensors, in particular, GLUT2 and sweet taste receptors, have been implicated in some diseases involving disorders in brain glucose metabolism such as obesity, diabetes mellitus, multiple sclerosis, Alzheimer's, Parkinson's, and Huntington's diseases 10 . Therefore, in this study, GLUT2 was the first priority to be examined.
Interestingly, microglia appear to be the only cells in the CNS expressing GLUT5 11 . The exact function of GLUT5 in microglia in relation to glucose metabolism is not known.
GLUT5 has a low affinity for glucose, and its affinity for fructose is much higher. It has been proposed that microglial GLUT5 expression is not correlated with prolonged hyperglycemia and that GLUT5 expressed in the endothelial cells and microglia of the brain is responsible for the transport of fructose. Fructose feeding in both young and older adult rats increased the expression of both GLUT5 mRNA and protein in the brain 12 . It is plausible that increased brain expression of GLUT5 allows fructose to play a role as an alternative energy source, at least in the short term. Furthermore, fructose-fed rats have activated microglia and primary rat microglia cultured under high fructose conditions showed upregulation of inflammatory pathways 13 . Therefore, microglia may play a key role in fructose-induced metabolic disorders.
Consistent with previous studies, although GLUT5 is regarded as a fructose transporter, we still examined it and indeed observed that the GLUT5 protein levels were not markedly affected in cultured BV-2 cells exposed to glucose fluctuations ( Supplementary Fig. 2).
Recently, a study proposed by Zhang et al. indicated that an HG (35 mM) condition can augment LPS-induced rat microglial activation and inflammatory cytokine levels 14 . Despite some consistency with our results, an obvious dissimilarity to the present study is that Zhang et al. did not remark on the viewpoint related to any GLUTs. Additionally, the brain glucose concentration is thought to be 5-fold lower than that of plasma (2030 mM in diabetic mice) 2 ; yet, microglia were illogically cultured in vitro under 2550 mM glucose conditions, a concentration range rarely observed in the brain during physiological or pathophysiological conditions. Therefore, compared to the study by Zhang et al., we think that using a glucose fluctuation between 5.5 mM (NG) and 25 mM (HG) in the present study may be more appropriate to offer new insight into the pathophysiology of diabetic neurodegeneration.  NG-cultured cells were treated with different concentrations of BAY (015 M) for 30 min before media being replaced from NG to HG. Cultures were continuously incubated in HG medium containing BAY at indicated concentrations for another 72 h. Control group was the NG-cultured cells treated with vehicle under a constant NG condition. After harvest, western blotting was used to detect the iNOS and COX-2 levels. GAPDH served as a protein loading control. Figure 8. High mannitol-induced hyperosmolarity exerts no effect on cell proliferation, inflammatory activation and inactive self-degradation in cultured BV-2 microglia. (a) Two BV-2 cell lines were separately cultured in NG and HG media (i.e., constant NG or HG). As indicated, some cells were exposed to a glucose shift (i.e., NG-to-HG or HG-to-NG) or treated with the high mannitol (i.e., NG-to-HM or HG-to-HM) for 072 h. After harvest, cells were subjected to a cell proliferation assay with the assistance of a Cellometer Auto T4 instrument. Each point represents the mean ± SEM from at least three independent experiments performed in triplicate. *p < 0.05 versus constant NG group (left panel) or constant HG group (right panel) at the same time point. (b) As indicated, the media of both NG-and HG-cultured BV-2 cells were simply renewed or replaced by HM medium.

Supplementary
After incubation for the indicated time points, western blotting was used to detect GLUT2, iNOS, and COX-2 levels. GAPDH served as a protein loading control. The representative blots from one of four independent experiments are presented. (c) The media of both NG-and HG-cultured BV-2 cells were simply renewed or replaced by NG, HG, and HM media, as indicated. After 48 h-incubation, an ELISA was applied to measure TNF- releasing level. ARA-C (2 µg/ml) was added to cultures to eliminate the interfere of different proliferation rates. Data are represented as mean ± SEM from at least three individual experiments. *p < 0.05 versus constant NG group; # p < 0.05 versus constant HG group. (d) Similar to the treatments in panel b, western blotting was used to detect Bcl-2, cleaved caspase-3, and LC3B levels. GAPDH served as a protein loading control. The representative blots from one of four individual experiments are presented.
iNOS, and COX-2 expression. GAPDH served as a protein loading control. The representative blots from one of four independent experiments are presented. (b-e) As indicated, both NG-and HG-cultured microglia were simply treated with or without LPS (1 µg/ml, 24 h); also, an NG-to-HG shift with or without LPS treatment was performed in NG-cultured microglia. After harvest, cultures were subjected to various biochemical assays for determining iNOS and COX-2 levels (b), TNF- release (c), and productions of peroxides (d) and ROS (e). In panel b, GAPDH served as a protein loading control and the representative blots from one of four independent experiments are presented. Data are represented as mean ± SEM from four independent experiments performed in triplicate. *p < 0.05 versus constant NG group; # p < 0.05 versus constant NG group treated with LPS. (f) Similar to the treatments in panel a, western blotting was executed to detect Bcl-2, cleaved caspase-3, and LC3B levels at the indicated time points. GAPDH served as a protein loading control. The representative blots from one of four individual experiments are presented. (g) A TUNEL assay was performed 24 h after the HG-to-NG shift or simple medium renewal. The representative photos of bright-field and matching TUNEL fluorescence are shown.