Mouse primary mixed glial cells were isolated and cultured as described previously (12). Briefly, the whole brain from pups on the first postnatal day was isolated and the meninges were removed in chilled Hanks' balanced salt solution. The brain was then dissociated in DMEM/F-12 medium containing trypsin-EDTA and incubated in 5% CO₂ for 12 min. The brain homogenate was centrifuged and filtered by using a cell strainer (100µm). Dissociated cells were washed and plated onto a 100-mm culture dish, and the medium was replaced every 3 days for 2~3 weeks. To isolate microglial cells, the above brain-mixed cultures were agitated for 8 h, and the liberated cell fraction was used for microglial cells. Mouse immortalized bone marrow-derived macrophages were prepared as described previously (17). To induce oxygen-glucose deprivation (OGD), culture medium was replaced with glucose-free DMEM, and the cells were placed in a humidified 37℃ incubator containing a mixture of 95% N₂ and 5% CO₂ for the indicated times.

Anti-human/mouse caspase-1 antibody was obtained from Santa Cruz (Santa Cruz, CA, USA) and kindly gifted from Dr. Emad Alnemri (Thomas Jefferson University). Anti-human/mouse IL-1β antibody was purchased from Cell Signaling Technology (Beverly, MA, USA) and R&D (Minneapolis, MN, USA). All the other antibodies were obtained from Cell Signaling Technology (PARP), Alexis (San Diego, CA, NLRP3), Abcam (Cambridge, MA, USA, HIF-1α) and eBioscience (San Diego, CA, CD11b-PE and F4/80-APC). LPS, CoCl₂, nigericin, ATP, and poly dA:dT were obtained from Sigma (St Louis, MO, USA), and z-VAD-fluoromethylketone was from Bachem (Torrance, CA, USA). All the culture media and supplements were purchased from Invitrogen (Grand Island, NY, USA).

Cells were lysed in a buffer containing 20 mM HEPES (pH 7.5), 0.5% NP-40, 50 mM KCl, 150 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, and protease inhibitors. Cell lysates were fractionated on SDS-PAGE, transferred onto PVDF membranes (BioRad, Hercules, CA, USA), and then Western-blotted using appropriate antibodies as indicated. Cell culture supernatants were precipitated by methanol/chloroform as described previously (17) and immunoblotted with appropriate antibodies.

To quantitate microglial cell population, cells were stained with anti-CD11b-PE and anti-F4/80-APC antibodies, and analyzed by flow cytometry. To detect mitochondrial ROS, cells were stained with MitoSox (Invitrogen) according to the manufacturer's protocol and analyzed by flow cytometry.

To test whether hypoxia or ischemia could trigger inflammasome signaling in brain microglial cells, we first observed caspase-1 processing in microglial BV-2 cells after oxygen-glucose deprivation (OGD), which induces ischemia-like injury. Extracellular release of processed caspase-1, a crucial readout of inflammasome activation, was not detected even by 12-h duration of OGD, while the secretion of processed IL-1β was observed in the cultural supernatant of BV-2 cells under OGD conditions longer than 6 h (Fig. 1A). This result suggests that OGD could induce IL-1β release in a caspase-1/inflammsome-independent manner.

Previous reports demonstrated that OGD causes brain cell death of not only neuron but also microglia (18,19), suggesting the possibility that the above OGD-mediated processing and secretion of IL-1β is due to cell death, but not to caspase-1/inflammasome. Indeed, longer duration of OGD than 8 h resulted in apoptotic cell death of BV-2 cells as determined by the cleavage of PARP (Fig. 1A, lower panel). We thus examined whether the OGD-mediated IL-1β release is caspase-dependent or caspase-independent using a pan-caspase inhibitor zVAD. As described in the above experiment, only processed IL-1β, but not caspase-1, was detected in the supernatants of BV-2 cells upon OGD treatment (Fig. 1B). However, the caspase inhibitor zVAD did not reduce the processing and release of IL-1β (Fig. 1B), indicating that IL-1β may be released by OGD in a caspase-independent manner.

The released IL-1β by OGD seems rather fragmented than processed into active IL-1β regardless of caspase activity (Fig. 1B). All these observations suggest that ischemia may not induce assembly of inflammasome and activation of caspase-1, but trigger nonspecific fragmentation and release of IL-1β, probably by cell death. Similar to the ischemia-mimicking OGD treatment, CoCl₂-induced hypoxia did not promote caspase-1 activation and IL-1β secretion in BV-2 cells, verifying that hypoxia is not implicated in the activation of inflammasome in microglial cells (Fig. 1C).

Then, we examined the effect of CoCl₂ on inflammasome signaling in the primary mixed glial cultures. As observed in BV-2 cells, CoCl₂ did not trigger caspase-1 activation in the LPS-primed mixed glial cells, whereas nigericin, a well-known NLRP3 stimulator, clearly induced the processing of caspase-1 (Fig. 2A). CoCl₂ treatment remarkably increased the monomeric or oligomeric level of HIF-1α, indicating that CoCl₂ induces hypoxic condition in brain mixed glial cultures. In contrast to the OGD treatment, no IL-1β was released by CoCl₂ treatment in LPS-primed mixed glial cells (Fig. 2A), demonstrating that hypoxia is not likely to induce inflammasome activation. To verify these observations in other cell types than brain mixed glial cells, we treated LPS-primed mouse bone marrow macrophages and PMA-differentiated THP-1 macrophages with CoCl₂. As observed in mixed glial cultures, CoCl₂-induced hypoxic conditions failed to activate caspase-1 in both macrophages, but NLRP3 stimulators, such as nigericin and LPS, triggered a robust activation of caspase-1, (Fig. 2B and C). All the above results indicate that hypoxia may not be a critical stimulus to the activation of inflammasome signaling.

Next, we examined whether CoCl₂-induced hypoxic conditions facilitate or potentiate inflammasome signaling in primary mixed glial cells, as hypoxia has been implicated in the development of inflammation (10). Interestingly, CoCl₂ did not potentiate, but rather inhibit ATP- or nigericin-induced NLRP3 inflammasome activation in LPS-primed brain mixed glial cells, as extracellular release of caspase-1 and IL-1β in their mature forms was attenuated in the presence of CoCl₂ (Fig. 3A). On the contrary, CoCl₂ had no effect on AIM2 inflammasome activity in response to synthetic double-stranded DNA, poly dA:dT (17) (Fig. 3B). This observation demonstrates that CoCl₂ may unexpectedly inhibit NLRP3-mediated caspase-1 activation in brain mixed glial cells.

To lineate the regulative role of CoCl₂ on NLRP3 inflammasome in bone marrow-derived macrophages, we determined active caspase-1 and IL-1β in the cultural supernatants of LPS-primed bone marrow macrophages after stimulation with ATP or nigericin. Unlike mixed glial cultures, NLRP3 inflammasome activity was not affected by CoCl₂ treatment in bone marrow macrophages (Fig. 3C), indicating that CoCl₂-induced hypoxia may play a unique inhibitory role on NLRP3 inflammasome of brain mixed glial cells.

Then, we determined the effect of CoCl₂ at a various treatment time before ATP stimulation to provide a molecular insight into the regulation of NLRP3 inflammasome by CoCl₂. As shown in Fig. 4A, longer treatment of CoCl₂ than 4 h reduced ATP-mediated caspase-1 activation and IL-1β release, but 1-h treatment of CoCl₂ was not sufficient to attenuate NLRP3 inflammasome activation. This result suggests that CoCl₂ might not directly interfere with the interaction of NLRP3 and ASC or with active assembly of NLRP3 inflammasome, because NLRP3 inflammasome is activated only after ATP stimulation. We can infer that a new protein synthesis might be required for the negative regulation of NLRP3 inflammasome by CoCl₂. The effect of CoCl₂ was also examined with various LPS priming times. CoCl₂ effectively attenuated NLRP3-dependent caspase-1 activation and IL-1β release in mixed glial cultures regardless of LPS-priming time (Fig. 4B).

In the central nervous system, microglia is the primary cell type responsible for recognizing abnormal molecular pattern, such as ATP, from tissue injury and triggering inflammatory responses (20). According to the previous report, microglia represents only 10% of the mixed glial cultures from the neonatal mouse brain (12). To exclude the possibility that the above finding was resulted from other cell types than microglia, we isolated microglia from mixed glial cultures. Similar to the previous report, the population of microglia was 6.7% among the mixed glial cultures, as determined by the cell populations expressing both CD11b and F4/80 (Fig. 5A). In contrast, microglial cells were enriched by 81.6% after isolation by the shaking method (Fig. 5A).

Next, we examined NLRP3 inflammasome activity in both mixed glial culture and isolated microglial cell. ATP-mediated caspase-1 activation and IL-1β release were similarly decreased by CoCl₂ cotreatment not only in mixed glial cultures but also in microglial cells (Fig. 5B). This data indicates that CoCl₂-induced hypoxia may negatively regulate NLRP3 inflammasome activity in brain microglial cells.

Although it is still unclear how NLRP3 inflammasome is activated upon a variety of stimuli including microbial toxins and crystalline substances, mitochondrial reactive oxygen species (ROS) has recently been suggested as a critical mediator facilitating the assembly of NLRP3 inflammasome (21). We thus examined whether mitochondrial ROS is important for ATP-induced NLRP3 inflammasome activation in brain mixed glial cells and whether CoCl₂-mediated inhibition of NLRP3 signaling is dependent on mitochondrial ROS levels. In bone marrow-derived macrophages, ATP treatment caused a significant production of mitochondrial ROS, albeit to a lesser extent than rotenone which specifically damages the electron transport chain complex I leading to the generation of mitochondrial ROS (Fig. 6A). Contrary to bone marrow-derived macrophages, ATP failed to produce mitochondrial ROS in LPS-primed brain mixed glial cells (Fig. 6B), suggesting that mitochondrial ROS may be dispensable for ATP-mediated caspase-1 activation in mixed glial cells. Furthermore, CoCl₂ had no effect on the generation of mitochondrial ROS (Fig. 6B). Taken together, CoCl₂-mediated inhibition of NLRP3 inflammasome activity in brain mixed glial cells seems independent of mitochondrial ROS production.