Coordinate effects of P2X7 and extracellular acidification in microglial cells

Extracellular adenosine 5′-triphosphate (ATP) is a damage-associated molecular pattern and contributes to inflammation associated diseases including cancer. Extracellular acidosis is a novel danger signal in the inflammatory sites, where it can modulate inflammation, immunity and tumor growth. Extracellular acidification was shown to inhibit P2X7-mediated channel currents, while it remains unknown how acidification and P2X7 together affect cellular responses. Here, we treated BV-2 microglial cells with ATP in a short period (<15 min) or a sustained acidified condition. For short acidification we compared the actions of neutralized ATP and acidic ATP in a condition with pH buffering. For sustained acidification, we treated cells with neutralized ATP in acidic medium or acidic ATP in medium without pH buffering. In the short acidified condition, neutralized ATP induced higher responses than acidic ATP to increase intracellular calcium and reactive oxygen species, decrease intracellular potassium and induce cell death. In contrast, these cellular responses and mitochondrial fission caused by neutralized ATP were enhanced by pH 6.0 and pH 4.5 media. P2X7 activation can also rapidly block mitochondrial ATP turnover and respiration capacity, both of which were mimicked by nigericin and enhanced by acidity. Taken together P2X7-mediated ionic fluxes and reactive oxygen species production are attenuated under short acidification, while sustained acidification itself can induce mitochondrial toxicity which deteriorates the mitochondrial function under P2X7 activation.


INTRODUCTION
Microglia are the resident immune cells in the central nervous system (CNS) [1]. In response to brain injury or immunological stimuli, these cells become activated and migrated to the site of injury and secrete numerous chemokines, reactive oxygen species (ROS), and pro-inflammatory cytokines. Microglia also actively involve in the phagocytosis of apoptotic cells and microbes. Although microglial activation is essential to eliminate pathogenic agents and prevent neuronal damage, it is a double edge sword during chronic inflammatory conditions such as neurodegenerative disease and CNS infections [2,3].
P2X7 is a ligand-gated ion channel receptor and is ubiquitously expressed in all types of cells. Among P2 purinoreceptor subtypes, P2X7 is usually activated by high concentration of ATP (> 500 μM) in a free acidic ATP4form [4,5]. Thus extracellular ATP that can be confined locally at sites of damaged tissues is regarded as an endogenous damage associated molecular pattern (DAMP). Activation of P2X7 rapidly triggers Na + , K + and Ca 2+ ions across the plasma membrane, which is followed by increasing membrane permeability for larger organic cations and forming a large, non-selective pore allowing molecules with molecular weight up to 900 Da to enter the cells [5,6]. The large pore formation consequently initiates several cellular events, including activation Research Paper of the NLRP3 inflammasome, opening of pannexin 1 and connexin hemichannels, membrane blebbing, ROS production, loss of mitochondrial membrane potential, and eventually cellular death [7][8][9]. P2X7 is highly expressed in both macrophages and microglia [10], and is critical for protective innate immune responses during the early phases of microbial infection [11]. Nevertheless, pro-inflammatory and pro-apoptotic actions of P2X7 contribute to the chronic inflammation and pathogenesis of various diseases, including arthritis, inflammatory bowel disease, neurodegenerative diseases, chronic pain, mood disorders and cancers [12][13][14][15][16][17][18].
Acidic environment frequently exists at the site of inflammation, representing a novel endogenous danger signal to alert the innate immunity [14,15,19]. In tissues with chronic inflammation, the pH values within the carotid atherosclerotic plaques, synovial fluid in rheumatic joints, and exhaled breath condensate of patients with acute asthmatic attack may be decreased to 6.8~5.2 [19][20][21]. Lower pH in the inflammatory loci mainly results from the activation of recruited immune cells, which undergo aerobic glycolysis and secrete more lactic acid to meet the higher energy demand [22,23]. Studies have demonstrated that acidic extracellular environment can modulate apoptosis, phagocytosis and trans-differentiation of neutrophils [24,25], regulate gene expressions [26,27], control phenotypic change and voltage-gated K + channels in astrocytes [28], impair P2Y receptor-induced migration in microglia [29], amplify proatherogenesis [30] and nociceptive responses [31]. Notably, NLRP3 inflammasome can sense acidic condition and trigger pH-dependent caspase-1 activation and IL-1β secretion. In contrast, extracellular alkaline condition strongly inhibits the IL-1β response to several known NLRP3 activators [32].
It is well known that inflammation can be induced by pro-inflammatory mediators and DAMPs [13]. Given that ATP and acidic pH are independent DAMPs, which can be concomitantly induced at the inflammatory loci, their combined action is an interesting issue to be investigated. To date only one study showed that the maximal ionic current response of P2X7 is suppressed by extracellular acidification, and five amino acid residues in the extracellular domain of P2X7 contribute to the inhibitory effect of acidification [33]. In this study, we applied an in vitro cell model to investigate the interplay of P2X7 agonists (ATP and BzATP) and acidic pH in BV-2 microglial cells. To this end, we designed two acidification modes, i.e. short time and long lasting acidification. For short time acidification, we treated cells with neutralized ATP (nATP) or acidic ATP (aATP). The latter gives an acidified solution and return to pH 7.4 when cells were kept in CO 2 incubator within 15 min. For long lasting acidification, we kept cells in NaHCO 3 -free complete media with pH adjustment and maintained in an incubator without CO 2 input. We found that extracellular acidification exerts dual actions to coordinate ATP-induced intracellular calcium increase, ROS production, mitochondrial dysfunction and cell death. Short acidified condition is sufficient to inhibit P2X7-mediated cellular responses, while long lasting acidified condition exerts an additive effect with P2X7.

Neutral ATP induced higher cellular responses than acidic ATP
Due to the rapid recovery (within 15 min) of acidic pH of aATP-containing medium to neutral condition in CO 2 -buffering incubator, we regarded the different responses of nATP and aATP as the effect of short time acidification on P2X7 response. In cell viability MTT assay, we found that both nATP and aATP can induce the concentration-dependent cell death, with nATP of higher potency and efficacy. Within 2 h of incubation, the IC 50 value of nATP was 0.92 ± 0.10 mM, and aATP at the highest concentration tested (3 mM) only induced 45 ± 3% cell death ( Figure 1A). The cell death induced by nATP and aATP (each at 3 mM) was 80% and 50%, respectively ( Figure 1A). Moreover, nATP treatment led to cell death with earlier onset than aATP ( Figure 1B). We also found that pre-treatment of P2X7 selective antagonist A438079 ( Figure 1A, 1B) and silencing P2X7 by siRNA approach (Figure 1C, 1D) can block the cell death induced by nATP and aATP, suggesting the involvement of P2X7 in extracellular ATP-induced cell death.
To further address the death mode, we conducted annexin V/PI staining. Results revealed that both forms of ATP at 3 mM dramatically and rapidly increased cell population with annexin V + /PI + staining, suggesting a mixed cell death mode (apoptosis and necrosis) is induced. Similar to the results of MTT assay, nATP at 3 mM induced higher degree of cell death than aATP ( Figure 1E, 1F). Likewise, A438079 and siP2X7 can abrogate the death caused by nATP and aATP. Accordingly, nATP induced higher extent of caspase 3 activation than aATP, and this event was blocked by siP2X7 ( Figure 1G).
Next we used Fluo-3 AM to monitor the [Ca 2+ ]i. We found that a rapid increase of [Ca 2+ ]i was evoked by nATP and aATP (each at 3 mM) within 20 min, and the response of nATP was significantly higher than that of aATP. Likewise, A438079 co-treatment abolished the increase of [Ca 2+ ]i ( Figure 2A). The weaker effects of aATP than nATP on cell viability and [Ca 2+ ]i increase suggest that short acidification can attenuate the ATP-induced P2X7 activation. We further compared the effects of nATP and aATP on intracellular ROS production and intracellular potassium. As shown in Figure 2B and 2C, both forms of ATP can increase cytosolic and mitochondrial ROS production in a time-dependent manner, and nATP triggered higher induction level within 15-60 min than aATP. Similarly, nATP induced a P2X7-dependent potassium efflux with faster onset and greater extents than aATP ( Figure 2D). The selectivity of A438079 on P2X7 was further tested with nigericin, which is a potassium ionophore known to partially mimic the cellular actions of exogenous ATP through P2X7. Our data revealed that nigericin-elicited potassium efflux was unaltered by A438079 ( Figure 2D).

Sustained acidification enhances nATP-induced calcium increase and ROS production
After we found the inhibition of P2X7-mediated responses by short time extracellular acidification, we were interested to understand the effect of sustained extracellular acidification on P2X7 activation, because sustained lower pH is commonly detected in the inflammatory microenvironment. Therefore, we treated cells in lower pH unbuffered DMEM (pH 6.0 or pH 4.5) together with nATP in 37° C incubator without CO 2 input to prevent the change of pH. In this condition, the pH of acidic media and aATP-containing DMEM was maintained for up to 2 h, the longest time period tested (data not shown). We found that acidification itself at either pH 6.0 or pH 4.5 rapidly increased [Ca 2+ ]i with similar extents at 10 and 30 min, while their responses were additive and synergistic to nATP (3 mM), respectively ( Figure 3A). Furthermore, we found that sustained extracellular acidification not only increased the cytosolic and mitochondrial ROS with earlier onset of pH 4.5 than pH 6.0, but also enhanced the responses of nATP ( Figure 3B). ATP-induced, but not low pH-induced changes in intracellular calcium (Figure 2A, Figure 3A) and ROS ( Figure 2B, 2C, Figure 3B) were blocked by P2X7 antagonist A438079. The additive or synergistic effect of nATP and low pH was also reduced by A438079 to the level of acidification itself ( Figure 3A, 3B).
With respect to cell viability, our data revealed that nATP-induced cell death was enhanced by acidification, both of pH 6.0 and pH 4.5 ( Figure 3C). In addition, the synergistic effect of pH 4.5 and ATP was inhibited by A438079, which failed to reduce the effect of low pH ( Figure 3C). To further verify the coordinated actions of P2X7 and acidification, we used the more specific P2X7 agonist BzATP and determined its effect on cell viability in different pH environment. We found that BzATP-induced cell death at 100 μM was enhanced by acidification, at either pH 6.0 or pH 4.5 ( Figure 3D). Moreover, higher cytotoxicity was induced by 300 μM BzATP and this action was also enhanced by pH 4.5 but blocked by A438079 ( Figure 3E). All these results suggest that in contrast to short period of acidification, sustained acidification can potentiate the cellular responses mediated by P2X7, implying an amplified outcome of both danger signals.

Coordinated effects of ATP and sustained acidification on mitochondrial membrane potential and mitochondrial morphology
As mitochondria are important in controlling cell viability that is instigated during P2X7 activation [34], we investigated the combinational effects of P2X7 agonist and sustained acidification on mitochondrial membrane potential (ΔΨm) and mitochondrial dynamics. Using JC-1 fluorescence probe we found that nATP (1 mM) can dramatically decrease ΔΨm and this effect was abrogated by A438079 ( Figure  4A). In acidification condition, the ΔΨm loss caused by nATP (1 mM) was enhanced ( Figure 4B). Meanwhile, the  calcium, cytosolic ROS, mitochondrial ROS and intracellular potassium, respectively. The fluorescence intensity of the control group was set as 100% to express the relative values. Results were the mean ± S.E.M. from 3 independent experiments. * p < 0.05, indicating the higher effects of nATP than aATP to increase intracellular calcium and ROS, but decrease intracellular potassium; # p < 0.05, indicating the significant antagonistic effect of A438079 on ATP responses. acidification-induced ΔΨm loss was unaltered by A438079 ( Figure 4C), ruling out the involvement of endogenous ATP-P2X7 pathway in the action of acidification.
In addition to ΔΨm measurement, image data of mitochondrial morphology revealed that nATP can dramatically change the microglial shape from the resting ramified form to the active spherical form ( Figure 5A). When incubating in acidic medium for 30 min, the cells swelled rapidly; and swelling phenotype at pH 6.0 condition was more obvious than pH 4.5 environment and was diminished by cotreatment with nATP ( Figure 5A). With Tom-20 staining our data demonstrated the abilities of nATP and acidification to induce mitochondrial fission as indexed by the changes of mitochondrial morphology from the elongation to fragmented appearance. When using Drp-1 staining as another index of mitochondrial fission, higher amounts of Drp-1 staining co-localized with Tom-20 signal were observed after nATP and acidic pH treatment alone. An even stronger signal was obtained upon nATP treatment in acidic medium ( Figure 5A). These results suggest a significant mitochondrial fission is induced by both danger signals.
After observing the mitochondrial fission induced by nATP and low pH, we further demonstrated the abilities of both stimuli to induce autophagy, as indexed by the accumulation of LC3II and downregulation of p62 ( Figure 5B). In addition, the less effects of pH 6.0 compared to pH 4.5 on these events were shown. As autophagy (or mitophagy) has been reported to exert either beneficial or suppressive role in cell viability [35], we used autophagosome inhibitor 3-methyladenine (3-MA) to determine the cell death role of autophagy under P2X7 activation. Our data indicated that BzATP-induced cell death can be reduced by 3-MA ( Figure 5C), suggesting the cell death effect of autophagy.

Both nATP and acidification reduce mitochondrial respiration
We also tested mitochondrial respiration function by measuring OCR. With the aid of several agents to interrupt mitochondrial respiration chain reaction including oligomycin (an inhibitor of ATP synthase), FCCP (an uncoupling agent of electron transport and oxidative phosphorylation), rotenone (an inhibitor of electron transfer from complex I to ubiquinone) and antimycin A (an inhibitor of complex III), we determined the resting OCR, ATP turnover, and uncoupled respiration of mitochondria. Our results showed that within 24 min incubation period, although nATP (1 and 3 mM) has no significant effects on resting OCR ( Figure 6A and 6B), it can attenuate ATP turnover and uncoupled respiration of mitochondria ( Figure 6B). All the inhibitory effects of nATP on respiration were reversed by A438079 ( Figure 6A, 6B). Notably, aATP at 3 mM led to an enormous and rapid decrease of mitochondrial respiration, ATP turnover and uncoupled respiration ( Figure 6A, 6B). These dramatic reductions caused by aATP (3 mM) were not blocked by A438079 nor due to cell death (data not shown). Because OCR assay is conducted in medium without pH buffering system, the finding of different effects of aATP and nATP, in particular at 3 mM of each, is suggested to be resulting from the acidity.
Next we determined whether impairment of mitochondrial respiration caused by nATP is resulting from potassium efflux. We found that nigericin (10 μM) alone did not alter resting OCR ( Figure 6C, 6D), while caused the same action as nATP to decrease the mitochondrial ATP turnover and uncoupled respiration ( Figure 6D). Moreover, acidification at pH 4.5 also significantly reduced the resting OCR ( Figure 6C, 6D), ATP turnover and uncoupled respiration ( Figure 6D). Moreover, high extracellular KCl itself led to a modest reduction of ATP turnover, and can attenuate the inhibitory effects of nigericin but not acidity on mitochondrial respiration ( Figure 6C, 6D). In acidified condition, a higher inhibition of mitochondrial oxidative phosphorylation function was detected upon nigericin treatment ( Figure 6C, 6D). These results suggest that increased intracellular acidity and loss of intracellular potassium may have a coordinative effect to impair mitochondrial oxidative phosphorylation.

DISCUSSION
P2X7 activation and functions were investigated in details over the past decade and its roles in the danger signaling with respect to intracellular potassium efflux and NLRP3 inflammasome activation have considerably given an attention very recently [6,7]. Numerous studies have mentioned this receptor's functions in regulation of various molecular events and participation in inflammation associated diseases like arthritis and Alzheimer [36,37]. Moreover, the role of P2X7 in microglial activation that contributes to neuronal inflammation has been demonstrated [38,39]. Extracellular acidification is also a major cause of neuronal diseases, as the decreased extracellular pH in the brain microenvironment is common at the conditions of brain trauma, epilepsy [40] and early stage of brain tumors [41]. Moreover, acid increases inflammatory pain has been demonstrated [42]. Here we are keen to investigate the coordinate actions of both danger signals, P2X7 and extracellular acidification, in microglia. We found that acidification  can exert opposite effects on regulating P2X7 action depending on the acidification periods. Acidification can inhibit P2X7-mediated intracellular stress responses at the upstream P2X7 receptor-coupled signaling level including channel currents as previously reported [33], but sustained acidification itself can exert similar effects as P2X7 to induce intracellular stress responses, which display the coordinate enhancement with P2X7 action for mitochondrial toxicity.
It is well known that ATP is a cytolytic stimulus and sustained activation of P2X7 by ATP leads to cell death in macrophages and microglia [6,9,[43][44][45][46][47]. Nevertheless, the cytotoxicity induced by P2X7 is cell type dependent. Compared to previous studies in EOC13 microglia [9] and RAW264.7 macrophages [44], our data indicate that P2X7-mediated cytotoxicity in BV-2 is much remarkable. In mouse primary microglia, high concentration of ATP (5 mM) also can rapidly and significantly induce cell death as we observed in BV-2 cells [46]. In N9 microglial cells, maximal release of LDH within 6 h incubation occurred at a concentration of about 0.5 mM ATP [47]. All these findings suggest that the different cytotoxicity of exogenous ATP in microglia might depend on the expression level of P2X7 and cellular context. Besides cytotoxicity, we also observed a weak inhibition (20%) of cell proliferation after BzATP (100 μM) treatment for 6 h in BV-2 cells (data not shown). This action of P2X7 is in agreement with previous study observed in murine luteal cells [48].
In this study we demonstrated the higher extents of intracellular calcium increase, cytosolic and mitochondrial ROS production and potassium loss induced by nATP than aATP at the same concentration in BV-2 microglial cells. Consistently a higher degree of cell death accompanied by an earlier onset was observed in response to nATP than aATP. The extensive ionic fluxes and ROS production have been shown to contribute to the cell death in response to P2X7 activation in microglia [9,10]. Previous study showed that the maximal ionic current response of P2X7 occurring within several milliseconds is suppressed by the extracellular acidified condition, and five amino acid residues in the extracellular domain of P2X7 are involved in this functional inhibition by acidic pH [33]. Because low pH of aATP-containing medium is rapidly neutralized to pH 7.4 within 15 min in CO 2 incubator, the different extent of response caused by nATP and aATP is ascribed to the effect of short acidification on P2X7. Therefore, our current data are in agreement with previous notion indicating that the signaling of P2X7 in ionic current fluxes is negatively regulated by an acidic extracellular pH.
In this study we found that sustained acidic pH itself is a stimulus to activate microglia by altering cell shape from ramified outlook to spheroid, and to induce intracellular stress response as P2X7 activation. Extracellular acidosis can cause intracellular acidification [49,50], calcium release from intracellular calcium stores and calcium influx [29,51]. Notably, pH 6.0 can enlarge microglial size more apparently than pH 4.5 and this effect is reversed by P2X7 activation. Currently we do not have sufficient data to answer this finding, but cytoskeleton rearrangement by retraction of lamellipodia upon acidification of extracellular milieu as reported might be the cause [50].
Mitochondrial oxidative phosphorylation plays a crucial role in maintaining cell metabolism and viability [52,53]. As reported, P2X7 activation causes a large increase of mitochondrial calcium and fragmentation [54]. Our findings demonstrate that P2X7 activation and intracellular acidification can individually induce mitochondrial dysfunction and display the potentiation action of each other. This notion is based on the findings of nATP and acidification to decrease mitochondrial membrane potential, induce mitochondrial fission and impair mitochondrial respiratory function. These findings can explain why 3 mM aATP can suppress mitochondrial respiration with more pronounced effect than nATP.
Recently a few studies demonstrate the relationship between acidification and mitochondrial activity. Mitochondria have outer membrane and inner membrane, and H + flowing across the inner membrane into intermembrane space constitutes the electrochemical polarization of the mitochondrial membrane potential and favors the flow of H + back into the matrix and couple to the ATP production. Extracellular acidification creates H + on the cytosol and blocks the proton flux across the inner membrane into the intermembrane space. Moreover, extracellular acidification (lower than pH 5.5) itself can change intracellular calcium homeostasis via different processes, such as induction of calcium influx through acid sensing ion channels [55], depletion of intracellular Ca 2+ stores [51], and alternation of the Ca 2+ exchange function of mitochondria [56].
Because ATP and acidity are effectors present in the microenvironments of inflammatory loci, their actions associated with cancer biology have been documented. As reported, extracellular acidification is a severe pathological feature of solid tumors [57,58]. Acidity in inflammatory tumor microenvironment is due to lactate secretion caused by anaerobic glycolysis and CO 2 production from the pentose phosphate pathway under the hypoxia [58]. Acidic extracellular pH can modulate cancer cell metastasis through an intracellular signaling cascade that is different from hypoxia to induce the expression of certain genes of prometastatic factors [58,59]. Moreover, acidic tumor milieu can modulate cancer cell metabolism and susceptibility to therapeutic efficacy [59,60]. On the other hands, P2X7 activation is associated with a complexity of modulatory actions on tumor growth, invasiveness and release of proinflammatory cytokines [16,17,[61][62][63]. To date, the potential interplay between P2X7 and low extracellular pH that contributes to cancer progression has not been investigated and is an interesting issue for further study. www.impactjournals.com/oncotarget Taken together we demonstrate the coordinate actions mediated by two endogenous danger signals in microglia. Extracellular acidification exerts different actions on P2X7 depending on the acidity periods. Short period of acidification upon P2X7 activation can target P2X7 and block the ionic responses mediated by P2X7 and the subsequent actions on mitochondrial functions. On the other hand, sustained acidification itself is an effective danger stimulus to induce intracellular stress response. The action of acidification overcomes its inhibitory action on P2X7 molecule and leads to a net potentiation effect with P2X7 activation. This study for the first time provides a detailed understanding on the coordinate actions of P2X7 and extracellular acidification in microglia, and highlights the mechanistic crosstalk of both DAMPs in regulation of inflammatory responses.

Cell culture and extracellular pH adjustment
BV-2 microglial cells were kept growing by culturing in complete DMEM supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 3.7 g/l NaHCO 3 , 100 units/ml penicillin and 100 μg/ml streptomycin. In acidification experiments, the final pH of culture medium was adjusted from pH 7.4 to 6.0 or 4.5, and cells were maintained in NaHCO 3 -free complete medium (i.e. in unbuffered DMEM) and placed in an incubator at 37° C without CO 2 after treatment with tested agents.

Preparation of nATP and aATP
To distinguish the effects of acidification on ATP action are coming from rapid inhibition of early signaling of P2X7 as previously demonstrated in the evoked cationic channel currents [33] and/or the late-occurring cellular actions by acidification itself, we designed two acidification modes, i.e. short time and long lasting acidification. For short time acidification, we need to readjust pH from acidic condition under P2X7 activation back to pH 7.4. Nevertheless, it is not easy to change the medium pH during the experimental course without causing some unexpected interference. Because ATP disodium salt hydrate dissolving in water or NaHCO 3 -containing DMEM gives a mild acidified solution (pH 6.0 and 4.5 for 1 mM and 3 mM ATP, respectively), and ATP disodium salt hydrate solution in NaHCO 3 -containing DMEM can return to pH 7.4 when it was kept in CO 2 incubator within 15 min, this feature allowed us to compare the cellular responses of aATP and nATP by directly treating them in BV-2 cells without subsequent pH adjustment procedure. To prepare nATP we neutralized stock aATP solution to pH 7.4 by NaOH.

RNA interference
Mouse siP2X7 (Cat no. SC-42576) and scramble nonspecific siRNA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). BV-2 cells were transfected with 100 nM siRNA by DharmaFECT Transfection Reagents (Dharmacon Research, Lafayette, CO, USA) at 50% confluence following the manufacturer's instruction. After 48 h of siRNA transfection, cells were treated with indicated drugs and then harvested for analysis.

Western blotting
After cells were treated with specific ligands in a mentioned time points, cells were washed twice with cold PBS and then lysed using RIPA lysis buffer. After cells were harvested and sonicated, total cell lysates were centrifuged at 18,000 g at 4° C for 30 min. Protein concentrations were determined with the Bio-Rad Protein Assay. Aliquots (60 µg) of soluble protein were resolved by SDS-PAGE, and samples were then transferred onto Immobilon-P membranes (Millipore). Nonspecific binding was blocked by incubating the blots with TBST containing 5% nonfat milk for 1 h at room temperature. After incubation with the appropriate primary antibody, the membranes were washed three times with TBST and then incubated with the appropriate HRP-conjugated secondary antibody for 1 h. The membranes were washed three times with TBST, and the protein bands were detected with ECL detection reagent. To assure equal loading of samples analyzed by Western blotting, we used β-actin as an internal control.

Measurement of cytosolic and mitochondrial ROS production
The intracellular levels of hydrogen peroxide (H 2 O 2 ) and mitochondrion-specific superoxide (O 2 -) were measured with DCFDA and MitoSOX Red, respectively. BV-2 cells were seeded in 96-well plates for overnight followed by the indicated treatments. At the endpoint, cells were treated with DCFDA (5 μM) or MitoSOX Red (5 μM) for 30 min. The cells were then washed with PBS and collected as single cell suspensions. Fluorescence signal was detected using fluorescence spectrophotometer (Varioskan TM Flash Multimode Reader, MA, USA) under the excitation and emission collected at 488 nm/520 nm and 510 nm/580 nm for DCFDA and MitoSOX Red, respectively.

MTT assay
BV-2 cells (1 × 10 4 per well) were plated in 96well plates and incubated overnight at 37° C followed by the time courses as indicated above. MTT (5 mg/ml) was incubated for 1 h at 37° C, the supernatants were aspirated, and the formazan granules generated by the live cells were dissolved in DMSO. The OD values at 550 and 630 nm were measured by use of a microplate reader. The net absorbance (OD550-OD630) indicating the enzymatic activity of the mitochondria and implying cell viability was represented as 100% of the individual control.

Annexin V/PI staining
BV-2 cells were seeded (1 × 10 7 cells/well) and incubated overnight at 37° C followed by the treatment mentioned. Following the treatment cells were suspended in an annexin V binding buffer and stained with both annexin V-FITC and propidium iodide (PI) at room temperature for 15-30 min and loaded on a flow cytometer (BD FACSCalibur, Franklin Lakes, NJ, USA) available for FL1 (annexin V) and FL2 (PI) bivariate analyses. Analysis by calculating percentage of the cells in the respective quadrants was done by using CellQuest PRO software.

Intracellular calcium and potassium measurements
After treatment with indicated agents, the cells were loaded with l μM Fluo-3 AM (for calcium measurement) or 5 μM Asante Potassium Green-2 for 30 min at 37° C under a dark condition, followed by washing twice with PBS. Then the cells were re-suspended in 500 μl PBS, and the intracellular calcium and potassium levels were analyzed by flow cytometry.

Measurement of mitochondrial oxygen consumption rate
the oxygen consumption rate (OCR) was measured by the Extracellular flux analyzer XF24 (Seahorse Bioscience, Houston, TX, USA). BV-2 cells were plated at 4 × 10 5 cells/well in a Seahorse 24-well V7 microplate (Seahorse Bioscience) and cultured in normal high glucose DMEM growth medium (with NaHCO 3 , glutamine, penicillin/ streptomycin and 10% FBS, pH 7.4) for 24 h in a 5% CO 2 incubator at 37° C. Then, the medium was removed and cells were incubated in XF assay medium in the absence of NaHCO 3 and FBS for 1 h at 37° C in measuring chamber without CO 2 input according to the routine assay procedure of manufacturer's protocol. ATP, nigericin, ionomycin and the mitochondrial complex inhibitors [oligomycin, carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP), rotenone and antimycin A] were freshly prepared in XF24 assay media. When determining the effect of acidification, high KCl and A438079 on OCR, medium of pH 4.5, pH 6.0, containing KCl (130 mM) or A438079 (10 μM) was added into wells prior to inserting the plate into the Seahorse XF24 Extracellular Flux analyzer. After 26 min to measure the basal respiration, nATP, aATP or nigericin was injected from port A; oligomycin (2.5 μM) was injected into each well at 50 min from port B, by FCCP (1 μM) at 74 min from port C and rotenone (2.5 μM) plus antimycin A (2.5 μM) at 98 min from port D. Because oligomycin is an inhibitor of ATP synthase by blocking the flow of protons through the F 0 subunit, the decreased OCR after oligomycin treatment indicates the status of ATP turnover. FCCP is an uncoupling agent which can disrupt ATP synthesis by transporting hydrogen ions across the mitochondrial membrane instead of the proton channel of ATP synthase (Complex V). This leads to a rapid consumption of energy and oxygen without the generation of ATP. Rotenone and antimycin A co-treatment can completely interfere with the electron transport chain in mitochondria. Rotenone inhibits the transfer of electrons in complex I to ubiquinone, while antimycin A blocks the electrons from complex III. OCR was recorded as pMoles per minute, and in some cases it was calculated as percentage of the OCR value before the treatment of tested agents as we previously described [64]. All the measurements and calculations were obtained by using seahorse instrument software (seahorse bioscience) according to the manufacturer's protocol. Averages of three wells were taken per data point.

Measurement of mitochondrial membrane potential (ΔΨm)
To investigate the mitochondrial membrane potential, BV-2 cells were seeded in 6-well plates a day before the experiment. After treatment for indicated times, the microglial cells (1 × 10 6 /ml) were incubated at 37° C for 20 min with 5 μg/ml JC-1, then washed twice before subjecting to flow cytometry. The cyanin dye JC-1 facilitates discrimination of energized and deenergized mitochondria because the normally green fluorescent dye forms red fluorescent aggregates when concentrated in energized mitochondria in response to their higher membrane potential [65]. Therefore, cell number with red fluorescence was determined as an index of mitochondrial membrane potential.

Mitochondrial imaging
BV-2 cells were fixed with 4% paraformaldehyde at 37° C for 20 min and then permeabilized with 0.2% Triton X-100 for 15 min. After blocking with 5% BSA with normal IgG (1:300) for 1 h, immunostaining was performed with primary antibody against Tom-20 (Santa Cruz Biotechnology, CA) or Drp-1 (Abcam, Cambridge, UK) in 1% BSA overnight at 4° C. Tom-20 is a subunit of the translocase of the mitochondrial outer membrane (TOM) complex, and becomes a common marker of mitochondria. Upon induction of external stress or apoptosis, Drp-1 (Dynamin related protein-1) translocates from the cytosol to mitochondria to initiate outer mitochondrial membrane fission. After washing with PBS, cells were incubated with secondary antibody in 1% BSA (in PBS) for 1 h at room temperature and then mounted with DAPI Fluoromount-G (SouthernBiotech, Birmingham, AL). Images were acquired using a 100 X Plan-Neofluar oil objective of LSM 880 with Airyscan SR microscopy (Carl Zeiss Micro Imaging GmbH, Jena, Germany).

Statistical analysis
Data were presented as the mean ± S.E.M. Students t test was used to access the statistical significance of the differences between samples, and p < 0.05 was considered statistically significant.

Author contributions
Ponarulselvam Sekar designed and performed the experiments, analyzed the data, prepared and reviewed the manuscript. Dr. Duen-Yi Huang conducted the confocal microscopy experiment for mitochondrial dynamics. Dr. Shwu Fen Chang involved in the discussion, experimental guidance and reviewed the manuscript. Dr. Wan-wan Lin designed the experiments and prepared and reviewed the manuscript.