Cardiac arrest leads to microglia activation and leukocytes infiltration in the brain
Rats undergoing10-min asphyxial cardiac arrest and cardiopulmonary resuscitation modeling (n = 259) or sham operation (n = 75) were randomly assigned to appropriate groups for different experimental purposes (Fig. 1). We first sought to examine the influence of cardiac arrest on the status of microglia. At 48 h post-surgery, single-cell suspension dissociated from cortical and hippocampal brain tissues were prepared to differentiate the populations of resident microglia and infiltrating leukocytes by flow cytometry labeling with CD45 and CD11b [23, 24]. As reported [25, 26], activated microglia possess a higher level of CD45 than resting microglia, while infiltrating leukocytes own the highest amounts of CD45. Consequently, CD11b-positive cells including infiltrating myeloid-lineage leukocytes and microglia could be classified as 3 populations: low CD45 population (CD45low, CD11b+), intermediate CD45 population (CD45int, CD11b+), and high CD45 population (CD45high, CD11b+), and they may represent resting microglia, activated microglia, and invading leukocytes, respectively (Fig. 2a, b). Cells in the sham-operated hippocampus expressed a low level of CD45 (CD45low, CD11b+), manifesting as tight cell clusters on the graph with very few cells in the CD45int and CD45high range. As a contrast, cells in the post-cardiac arrest hippocampus revealed much higher expression of both CD45 and CD11b, reflected as looser cell clusters on the graph with many more cells in the CD45int and CD45high range. Similar results were observed in cells from cortical brain tissues. It indicates a large number of microglia shifts from an immuno-surveilling state to an activated state after cardiac arrest, accompanied by massive exudation of myeloid-lineage leukocytes in the post-cardiac arrest brain tissues.
Cardiac arrest triggers microglial pyroptosis and an increase of caspase-1 activity in the activated microglia
We next investigated whether the activated microglia were correlated with pyroptosis and promoted the inflammatory response after cardiac arrest. Flow cytometry was conducted to measure the cell viability and caspase-1 activity in microglia, which were labeled with LIVE/DEAD Fixable Near-IR Dead Cell Stain fluorescent probe and FAM-FLICA fluorescent probe, respectively. The CD11b+ cells were at first gated into CD45low CD11b+, CD45int CD11b+ and CD45high CD11b+ by CD45 labeling, as described above. Then, all microglia including resting (CD45low, CD11b+) and activated microglia (CD45int, CD11b+) were plotted using the two kinds of fluorescent probes to establish a four-quadrant gate (Fig. 2c, d): Q1, necrotic microglia (FLICAlow, LIVE/DEADhigh); Q2, pyroptotic microglia (FLICAhigh, LIVE/DEADhigh); Q3, live microglia expressing caspase-1 (FLICAhigh, LIVE/DEADlow); Q4, live microglia without expression of caspase-1 (FLICAlow, LIVE/DEADlow ) .
As illustrated, the number of pyroptotic microglia and live microglia expressing caspase-1 in the post-cardiac arrest brain tissues (hippocampus and cortex) was significantly higher than that in the sham-operated brain tissues (Fig. 2c, d). Further analysis revealed that the number of cells undergoing pyroptosis and live cells expressing caspase-1 was predominantly increased in the CD45int CD11b+ population (Fig. 2e, f) but not in the CD45low CD11b+ population (Fig. 2g, h), indicating that the activated microglia population was the primary source of pyroptotic microglia and cells with elevated caspase-1 activity after cardiac arrest. These results imply that cardiac arrest triggers microglial pyroptosis and an increase of caspase-1 activity in the activated microglia, and the latter may also subsequently undergo pyroptosis.
Cardiac arrest induces NLRP3 inflammasome activation in microglia
Inflammasome assembling with NLRP3 and caspase-1 is considered to be the major signaling molecule that promotes the pyroptosis in microglia . Therefore, we tested the activation of NLRP3 inflammasome after cardiac arrest, aiming to provide potential targets for mediating microglial pyroptosis. Results from qRT-PCR showed that the mRNA level of NLRP3 was significantly upregulated after cardiac arrest, with a peak at the 12th hour (Fig. 3a). Consistently, the protein level of NLRP3 was elevated at 6 to 24 h after cardiac arrest (Fig. 3b, c). The mRNA level of caspase-1 and the protein level of pro-caspase-1 were not altered for the rats undergoing cardiac arrest compared to sham operation (Fig. 3a–c). However, the protein level of cleaved caspase-1, namely the active form of capase-1, was increased at 6 to 24 h in the post-cardiac arrest brain (Fig. 3b, c), indicating that the pro-caspase-1 underwent autocatalysis and activation after the cardiac arrest. In addition, the mRNA and protein levels of GSDMD were further evaluated via qRT-PCR and Western blotting to prove that the cleavage of caspase-1 here was involved in pyroptosis but not apoptosis . As expected, it was observed that the protein level of cleaved GSDMD was markedly increased in the post-cardiac arrest brain, with the elevated mRNA and protein levels of GSDMD (Fig. 3d-f), which implies that caspase-1 is cleaved and activated after cardiac arrest, leading to the downstream cleavage of GSDMD and pyroptosis. We also assessed the level of caspase-11, in order to exclude the effect of caspase-11-mediated non-canonical inflammasome activation on the cleavage of GSDMD and pyroptosis . It was found that there was no significant change in the protein levels of precursors of caspase-11 (pro-caspase-11) and cleaved caspase-11 of the rats in the cardiac arrest group compared to that in the sham group (Fig. 3d-f), indicating that it is the canonical NLRP3 inflammasome that mediates the microglia pyroptosis in the rats after cardiac arrest, and caspase-1 but not caspase-11 is involved in the process. The elevated expression of NLRP3 and cleaved caspase-1 were further confirmed by immunofluorescent staining at 12 h after cardiac arrest, along with ASC, another key molecule involved in the assembly of inflammasome (Fig. 3g, h).
To further validate whether the assembly of inflammasome principally occurred in microglia after cardiac arrest, immunofluorescent co-localization was performed. Using confocal microscopy, we observed that NLRP3 and caspase-1 were mainly co-localized in Iba-1-positive cells but not other types of cells such as neuron, astrocytes, endothelial cells, oligodendrocyte, or oligodendrocyte precursor cells to form the inflammasome (Fig. 4a–d). It should be emphasized that since Iba-1 is not a highly specific marker, the Iba-1-positive cells, in addition to identifying microglia, may also represent a small number of invading monocytic lineage cells like macrophages. The association of NLRP3, ASC, and caspase-1 represents the intracellular activation of the inflammasome . As shown by the co-immunoprecipitation (Fig. 4e-h), pairwise interactions among NLRP3, ASC, and caspase-1 were observed in both the hippocampus and cortex tissues at 12 h after cardiac arrest. Our quantitative analysis showed that the assembly of NLRP3 inflammasomes in microglia of post-surgery rats was significantly increased, featuring in the increment in the absolute amount. Interestingly, we found that there was a small amount of activated inflammasomes in sham group, which could be attributed to the effect of isoflurane anesthesia . Therefore, we speculate that the combination of isoflurane and cardiac arrest in our study may lead to additional inflammasome activation over that seen in cardiac arrest without anesthesia. Taken together, these findings imply that cardiac arrest activates the assembly of NLRP3 inflammasome in microglia, which leads to the self-cleavage of caspase-1 and triggers microglial pyroptosis.
Targeting NLRP3 with MCC950 prevents microglial pyroptosis and consequential neuroinflammation after cardiac arrest
To further test the role of NLRP3 inflammasome and correlated microglia pyroptosis in post-cardiac arrest brain injury, we targeted the NLRP3 with a highly selective inhibitor MCC950 . At 10 min after ROSC from cardiac arrest, the rats were randomized to receive daily once MCC950 or vehicle until euthanasia (Fig. 1). The body weight, the time from asphyxia to cardiac arrest, the time required for ROSC, and epinephrine usage were all comparable between the MCC950 and vehicle groups (Table 1). There were also no statistical differences in physiological variables, including MAP, heart rate, and rectal temperature at baseline or after ROSC between these two groups (Table 1).
Given the direct bioactivity of MCC950 on NLRP3, we first examined the influence of MCC950 on the assembly of NLRP3 inflammasome, the cleavage of caspase-1 and GSDMD, and the maturation of IL-1β and IL-18. The results from qRT-PCR and Western blotting showed that MCC950 significantly suppressed the levels of NLRP3, ASC, IL-1β, and IL-18 at 12 h after cardiac arrest, in both hippocampus and cortex (Fig. 5a–c). Also, MCC950 markedly inhibited the activity of cleaved caspase-1 but not the level of pro-caspase-1, suggesting that MCC950 suppresses the activation but not the generation of caspase-1. Besides, we found the similar result that MCC950 could significantly inhibit the activation of GSDMD and reverse the increased level of cleaved GSDMD after surgery, with no effect on the level of full-length GSDMD (Fig. 5a–c). Consistently, less NLRP3-positive, ASC-positive, and cleaved caspase-1-positive cells were observed in the MCC950-treated group (Fig. 5d, e).
We then examined whether MCC950 prevents microglial pyroptosis and the increased caspase-1 activity in activated microglia caused by cardiac arrest. In consistent with the above results, a reduced number of live cells expressing caspase-1 and pyroptotic cells were observed in the MCC950 but not the vehicle groups at 48 h after cardiac arrest (Fig. 6a, b). MCC950 treatment significantly reduced the number of pyroptotic cells and live cells expressing caspase-1 compared to the vehicle group in the CD45int CD11b+ population (Fig. 6a, b). However, there was no difference in the number of pyroptotic and caspase-1-positive cells between the MCC950-treated and the vehicle-treated groups in the CD45low CD11b+ population (Fig. 6a, b). Moreover, the delivery of MCC950 led to a significant reduction in the number of CD45high CD11b+ cells (Fig. 6c, d) accompanied by the inhibition of IL-1β and IL-18, suggesting that the intervention of NLRP3 by MCC950 prevents the post-cardiac arrest inflammatory response. Strikingly, MCC950 inhibited not only the mature IL-1β and IL-18 but also their precursors, which could be attributed to the overall suppression of inflammatory response and thus reduced the transcription of these two cytokines by nf-κb [31,32,33]. Therefore, our findings provide evidence that targeting NLRP3 by MCC950 suppresses the occurrence of microglial pyroptosis and consequential inflammatory response after cardiac arrest.
Targeting NLRP3 with MCC950 improves survival and neurologic outcome after cardiac arrest
We then explored whether the suppression of microglia pyroptosis and consequential inflammatory response improved outcomes by targeting NLRP3 with MCC950 in cardiac arrest modeling rats. We found that the 7-day survival rate in the MCC950 group (72.7%, 16 of 22) was significantly higher than that in the vehicle group (40.9%, 9 of 22) (Fig. 7a). In addition, we used the NDS to assess the neurologic function after cardiac arrest and found that rats in the vehicle group presented lower NDSs at 24, 48, 72 h and on day 7 after ROSC than those in the MCC950 group (Fig. 7b, c), indicating that the neurologic deficit caused by cardiac arrest was alleviated by MCC950 treatment.
We also performed the Morris water maze test to evaluate the effect of MCC950 on short-term spatial learning and memory ability after cardiac arrest. As a comparison, sham-operated rats were used in this part of the experiment. During the hidden-platform training on day 9 to day 12 after surgery, rats in the sham, vehicle, and MCC950 groups all showed a gradual decline in latency to find the hidden platform over time (F = 60.086, P = 0.000) (Fig. 7d). In addition, analysis of the training data by repeated-measures ANOVA showed that escape latency differed significantly among the groups (F = 13.662, P = 0.000), with no significant interaction between groups and time points (F = 2.455, P = 0.113). Compared with sham operation, cardiac arrest led to extended latencies on day 10 to day 12 in the vehicle group (all P < 0.05 vs. sham), whereas the extended latencies were partly prevented by MCC950 treatment (all P < 0.05 vs. vehicle) using the Tukey’s post hoc test. The poor performance of vehicle animals to climb up the hidden platform should not be attributed to slower swimming speed because the mean swimming speed was comparable among the three groups (data not shown). On day 13 after cardiac arrest, all rats received the probe test to assess their short-term memory (Fig. 7e, f). Results showed that the frequency of crossing the platform area was lower in the vehicle group than the sham group, whereas this figure was increased after MCC950 treatment. Besides, rats administrated with MCC950 spent markedly more time searching in the platform quadrant (Q3) in comparison with vehicle rats, while vehicle-operated rats spent increased time in the other three non-platform quadrants (Fig. 7e, f). These findings from Morris water maze test indicate that MCC950 treatment rescued the spatial learning and memory deficiency caused by cardiac arrest.
These results demonstrate that long-term treatment with MCC950 after cardiac arrest provides persistent neuroprotection to improve 7-day survival and neurologic outcome.
Targeting NLRP3 with MCC950 ameliorates histological injury after cardiac arrest
To assess the histological damage caused by cardiac arrest and explore the profound effect of MCC950, all the rats involved in the Morris water maze test were euthanized on day 14 after cardiac arrest to conduct Nissl and immunohistochemical staining (Fig. 7g, h). The results from Nissl staining demonstrated that cardiac arrest induced neuron loss in the hippocampal CA1 region, a vulnerable region to global ischemia, whereas the neuron loss was partially restored by MCC950 (Fig. 7g). Consistently, less NeuN+ cells (neuron) were noticed in the vehicle group compared to the sham group, while a significantly increased number of NeuN+ cells was found in the MCC950 group. We also examined the injury to dendrite by immunostaining for MAP-2, a protein that was enriched in neuronal dendrites and acted as a stabilizing molecule for the dendritic cytoskeletal integrity (Fig. 7h). Results showed that cardiac arrest caused a dramatic dendritic loss in the vehicle group, as compared with sham controls. However, the dendritic loss was markedly reversed after MCC950 treatment. These findings, therefore, indicate that MCC950 substantially prevents the neuron loss and dendritic injury induced by cardiac arrest.
Neuron loss is usually accompanied by glial activation to clean up the cell debris. As illustrated in Fig. 7h, microglia and astrocytes were dramatically activated in the post-cardiac arrest hippocampal CA1 region, as evidenced by increased immunoreactivities of Iba-1 for microglia and GFAP for astrocytes. The number of Iba-1-positive microglia and GFAP-positive astrocytes were both markedly reduced by MCC950 intervention compared to the vehicle group, indicating that the activation of microglia and astrocytes were inhibited by MCC950. Based on the results of flow cytometry, the number of activated microglia was elevated after cardiac arrest modeling, which was perceived as an important link to the pyroptosis implicated in the vigorous post-cardiac arrest inflammation and subsequent brain damage. We thus stained the brain sections with CD68 antibody, a marker for activated microglia, to further explore the inhibitory effect of MCC950 on the activated microglia. Unlike the findings observed at 48 h after cardiac arrest, there was a significant decrease in the intensity of CD68 staining in the presence of MCC950 treatment after cardiac arrest (Fig. 7h). This might be due to the inhibition of early inflammatory response by MCC950 after cardiac arrest, thereby reducing the further activation of microglia. Taken together, these results suggest that MCC950 significantly prevents histological injuries, which might be via interfering NLRP3 inflammasome activation and suppressing microglia pyroptosis in the brain after cardiac arrest.
Targeting caspase-1 by Ac-YVAD-cmk prevents microglial pyroptosis and consequential neuroinflammation after cardiac arrest
To further verify the role of NLRP3 inflammasome in mediating post-cardiac arrest microglial pyroptosis and its consequential brain injury, we used a selective inhibitor Ac-YVAD-cmk to target caspase-1, the canonical executor of pyroptosis . In this part, rats were randomized to receive one dose of Ac-YVAD-cmk (400 ng in 4 μL, i.c.v.) or vehicle before the induction of asphyxial cardiac arrest (Fig. 1). There was no significant difference found in the ratio of achieving ROSC between the Ac-YVAD-cmk (35/42, 83.3%) and vehicle (35/44, 79.5%) groups. Moreover, the body weight, the time from asphyxia to cardiac arrest, the time required for ROSC, epinephrine usage, and physiological variables were all comparable between these two groups (Table 2).
We found that the mRNA levels coding for IL-1β and IL-18 in hippocampus and cortex of post-cardiac arrest rats were downregulated after treatment with Ac-YVAD-cmk. However, there was no significant change for the mRNA levels of NLRP3, ASC, caspase-1, and GSDMD with Ac-YVAD-cmk treatment (Fig. 8a). Ac-YVAD-cmk treatment also significantly inhibited the activity of cleaved caspase-1 and cleaved GSDMD, and decreased the protein level of pro-IL-1β, pro-IL-18, IL-1β, and IL-18 in both hippocampus and cortex, while the levels of NLRP3 and ASC were not changed after Ac-YVAD-cmk treatment (Fig. 8b, c). These results were further confirmed by immunofluorescence staining. Inhibition of caspase-1 by Ac-YVAD-cmk dramatically reduced cleaved caspase-1 but not NLRP3 or ASC, as compared with those vehicle controls (Fig. 8d, e). These findings suggest that Ac-YVAD-cmk restrains the cleavage of caspase-1, and thereby prevents the maturation of IL-1β and IL-18 in the brain after cardiac arrest.
We next detected whether the inhibition of caspase-1 by Ac-YVAD-cmk suppressed microglia pyroptosis and the caspase-1 activity in activated microglia caused by cardiac arrest. As expected, targeting caspase-1 by Ac-YVAD-cmk substantially reduced the number of pyroptotic cells and live cells expressing caspase-1 in all microglia population (CD45low CD11b+ and CD45int CD11b+) as well as in the CD45int CD11b+ population after cardiac arrest (Fig. 9a, b). Again, the suppression of microglial pyroptosis by Ac-YVAD-cmk led to a notably decreased number of CD45high CD11b+ population compared to the vehicle group (Fig. 9c, d). Thus, targeting caspase-1 by Ac-YVAD-cmk prevents microglial pyroptosis and consequential neuroinflammation after cardiac arrest.
Targeting caspase-1 by Ac-YVAD-cmk ameliorates neurological injury after cardiac arrest
We then assessed whether the suppression of microglia pyroptosis by targeting caspase-1 with Ac-YVAD-cmk ameliorated neurological injury in a rat model of cardiac arrest. Post-cardiac arrest rats in the Ac-YVAD-cmk group exhibited statistically higher NDSs at 24, 48, 72 h, and 7 days than those in the vehicle group (Fig. 10a, b), suggesting that Ac-YVAD-cmk intervention reduced neurologic damage. Furthermore, during the training period of the Morris water maze test, the escape latency became progressively shorter in all groups over time (F = 71.673, P = 0.000), and repeated-measures ANOVA revealed that escape latency differed significantly among the groups (F = 13.768, P = 0.000), with no significant interaction between groups and time points (F = 2.040, P = 0.075). Besides, Tukey’s post hoc test showed that Ac-YVAD-cmk substantially shorted the latency in finding the hidden platform on day 10, 11, and 12 after cardiac arrest modeling groups (Fig. 10c), while the mean swimming speed was comparable between the Ac-YVAD-cmk and the vehicle groups (data not shown). In the probe trial of the Morris water maze test, the frequency of crossing the platform area was increased in the Ac-YVAD-cmk-treated group compared with the vehicle-treated groups (Fig. 10d, e). Moreover, rats with an intracerebroventricular injection of Ac-YVAD-cmk had a trend toward spending more time searching in the platform quadrant (Q3) in comparison with vehicle rats, while vehicle-operated rats seemed to spend increased time in the other three non-platform quadrants (Fig. 10d, e).
After the Morris water maze test, all rats in this part of the experiment were euthanized, and brain sections were prepared for histological injury evaluation. Our results of Nissl staining revealed that there were significantly more viable neurons in the Ac-YVAD-cmk-treated group compared to the vehicle group (Fig. 10f). Ac-YVAD-cmk-treated rats exhibited more NeuN+ neurons in hippocampal CA1 region than vehicle-treated rats (Fig. 10g). Besides, post-cardiac arrest rats in the vehicle group exhibited extensive loss of MAP2-immunoreactive dendrites in the hippocampus CA1 region, which was partially recovered by Ac-YVAD-cmk treatment (Fig. 10g). We next carried out the Iba-1, CD68, and GFAP immunohistochemical staining to verify the activation of microglia and astrocytes. As expected, extensive Iba-1-positive, GFAP-positive, and CD68-positive cells appeared in the hippocampal CA1 region in the post-cardiac arrest rats treated with vehicle solution compared to the sham controls. However, Ac-YVAD-cmk treatment significantly reduced Iba-1-positive, GFAP-positive, and CD68-positive cells in the hippocampal CA1 region of post-cardiac arrest rats. These results suggest that pharmacological blockade of caspase-1 by Ac-YVAD-cmk considerably suppressed cardiac arrest-induced activation of microglia and astrocytes. Overall, the findings from this part suggest that pre-treatment with Ac-YVAD-cmk provides profound effects in preventing neurologic deficit and alleviating neuropathological injury induced by cardiac arrest, which could be attributed to the suppression of microglial pyroptosis and ensuing inflammatory responses.