Neuroprotection of NSC Therapy is Superior to Glibenclamide in Cardiac Arrest Induced Brain Injury via Neuroinflammation Regulation

Abstract

Cardiac arrest (CA) is common and devastating, and neuroprotective therapies for brain injury after CA remain limited. Neuroinflammation has been a target for two promising but undeveloped post-CA therapies: neural stem cell (NSC) engrafting and glibenclamide (GBC). It is critical to understand whether one therapy has superior efficacy over the other and to further understand their immunomodulatory mechanisms. In this study, we aimed to evaluate and compare the therapeutic effects of NSC and GBC therapies post-CA. In in vitro studies, BV2 cells underwent oxygen-glucose deprivation (OGD) for three hours and were then treated with GBC or co-cultured with human NSCs (hNSCs). Microglial polarization phenotype and TLR4/NLRP3 inflammatory pathway proteins were detected by immunofluorescence staining. 24 Wistar rats were randomly assigned to three groups (Control, GBC, and hNSCs, N = 8/group). After 8 minutes of asphyxial CA, GBC was injected intraperitoneally or hNSCs were administered intranasally in the treatment groups. Neurological deficit scores (NDS) were assessed at 24, 48, and 72h after return of spontaneous circulation (ROSC). Immunofluorescence was used to track hNSCs and quantitatively evaluate microglial activation subtype and polarization. The expression of TLR4/NLRP3 pathway-related proteins was quantified via Western Blot. The in vitro studies showed the highest proportion of activated BV2 cells with increased expression of TLR4/NLRP3 signaling proteins were found in the OGD group compared to OGD + GBC and OGD + hNSCs groups. NDS showed significant improvement in hNSC and GBC groups compared to Controls, and hNSC treatment was superior to GBC treatment. The hNSC group had more inactive morphology and anti-inflammatory phenotype of microglia. The quantified expression of TLR4/NLRP3 pathway-related proteins was significantly suppressed by both treatments, and the suppression was more significant in the hNSC group compared to the GBC group. hNSC and GBC therapy regulate microglial activation and the neuroinflammatory response in the brain after CA through TLR4/NLRP3 signaling, and exert multiple neuroprotective effects including improved neurological function and shortened time of severe neurological deficit. In addition, hNSCs displayed superior inflammatory regulation over GBC.

Introduction

Cardiac arrest (CA) is a common but devastating disease. More than 350,000 CAs occur outside of hospitals in the US every year [1]. The prognosis remains dismal, as about 10.8% of patients survive hospital discharge [1]. Nearly half of survivors suffer from cognitive deficits [2], and about 20% live with reduced quality of life [3].

In recent years, neuroinflammation has proven to play an important role in hypoxic-ischemic brain injury [4]. This makes the regulation of neuroinflammation a promising therapeutic target, although there remain controversies regarding the method and timing of therapeutic targets [5]. Microglia, as the innate immune cells residing in the central nervous system, are rapidly activated and play a major role in neurological disease models, including CA-related brain injury [6–8]. However, the detailed contributions of microglia in CA-related brain injury are not fully understood.

Stem cell therapy is considered to be an important method in the regulation of the immune response after brain injury. Neural stem cells (NSCs) have been shown to exert neuroprotective functions after CA by regulating microglia-mediated inflammation [9, 10]. Glibenclamide (GBC) has been reported to protect the brain in disorders such as stroke and traumatic brain injury in animal models [11, 12]. GBC can cross the blood-brain barrier (BBB) and improve neurological outcomes in a rat CA model [13]. However, these two novel interventions remain underdeveloped in preparation for clinical translation, and a direct efficacy comparison of these two novel interventions has never been investigated. As preclinical trials move toward translation, it is critical to understand whether one therapy is superior to the other in order to focus translational efforts most effectively.

Toll-like receptor (TLR) 4 is a crucial pattern-recognition receptor expressed on the surface of microglia [14]. TLR4 activation triggers the downstream signaling pathway of the NLR pyrin domain containing 3 (NLRP3) inflammasome and leads to the downstream maturation of the key inflammatory cytokine IL-1β through activation of caspase-1. [15, 16]. Of note, alterations in structure, expression, and activation of the NLRP3 inflammasome result in the inflammatory response and can alter the progression of ischemic stroke [17–22]. A recent study also revealed that the cerebral NLRP3 inflammasome was activated in a swine model of CA [23]. The above evidence is still insufficient to fully reveal whether the TLR4/NLRP3 signaling pathway can be regulated by immunomodulatory therapy targeting microglia in the early injury phase after the global ischemic attack and ultimately achieve satisfactory neuroprotective effects.

Although the exact mechanism remains unclear, GBC has been shown to suppress NLRP3 inflammasome activation via suppression of caspase-1 activation [24]. NSCs exert their neuroprotective functions through mitigating microglia-mediated inflammation [9]. Inhibition of the NLRP3 inflammasome is involved in the neuroprotective mechanism of NSCs, which attenuated microglia-mediated toxicity in an in vitro model [25]. However, whether the above findings can be extended to in vivo research at a more complex and critical level of global brain injury is still uncertain.

In this study, we aimed to evaluate and compare the therapeutic effects of hNSC and GBC therapies post-CA. We hypothesized that hNSC and GBC therapies exert their neuroprotective functions after CA by regulating neuroinflammation via the TLR4/NLRP3 signaling pathway.

Methods

Cell culture and treatments

The BV2 mouse microglial cell line was obtained from American Type Culture Collection (ATCC, Manassas, VA) and all BV2 cells in the control group, oxygen-glucose deprivation (OGD) group, and two treatment groups were cultured in uniform media components, including: high-glucose Gibco Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS) and 1% penicillin-streptomycin. Cells were passaged with 0.25% trypsin. The hNSC line was obtained from Millipore (ReNcellVM Human Neural Progenitor Cell Line, Catalog Number SCC008). The hNSCs were expanded and maintained in ReNcell hNSC Maintenance Medium (serum-free media, Millipore Cat. No. SCM005) supplemented with 20 ng/mL basic fibroblast growth factor (bFGF, Thermo Fisher, Cat# 13-256-029) and 20 ng/mL epidermal growth factor (EGF, Thermo Fisher Cat# PHG0314). hNSCs were passaged using Accutase (Millipore). All of the cells were maintained at 37°C in an incubator with 95% air and 5% CO₂. All cell culture vessels for maintenance of hNSCs were coated with 20μg/mL Laminin (Sigma, L2020) overnight before use [26].

BV2 cells were seeded into 24-well plates at a density of 4×10⁴ cells/well. hNSCs were seeded into the Corning Transwell inserts (1×10⁴ cells/well). After OGD injury, BV2 cells were treated with GBC (10 μM) [27], co-cultured with hNSCs for 12 h during re-oxygenation, or received no treatment (Control) in high-glucose DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. A fourth, normal group of cells did not receive OGD injury or any treatment for 12 h.

Oxygen-glucose deprivation and re-oxygenation (OGD/R)

The BV2 cells were washed three times with PBS and subjected to glucose-free DMEM (A1443001, Gibco) pre-bubbled with 100% N₂ for 30 min. Cells were exposed to an atmosphere of 1% O2, 94% N2 and 5% CO2 for 3 h using one cell culture chamber of an Xvivo System (Biospherix Hypoxia Cytocentric Glove Chamber Incubation Cell Culture G300C, Biospherix, Ltd., Parish, NY), as previously published [28]. After OGD for 3 h, the medium was then changed to high-glucose DMEM containing 10% FBS and was returned to the normoxic incubator for culture.

Cell viability

BV2 cell viability was assessed with MTT assay according to the commercial kit (Trevigen, Gaithersburg, MD). MTT reagent (40 μL) was added to the culture medium in 24-well plates and incubated at 37°C for 2h. Then the detergent reagent was added to each well and incubated at 37°C for 2h. The absorbance of each sample was determined using a microplate reader at 570 nm.

Animal model

Twenty-four adult Wistar rats (Charles River, Wilmington, MA) were randomly divided into three groups (N=8): the hNSCs treatment group (hNSCs), the GBC treatment group (GBC), and the control group (Control). All protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Maryland, Baltimore. All rats were included in the study unless return of spontaneous circulation (ROSC) was not achieved within 2 minutes of cardiopulmonary resuscitation (CPR).

The asphyxial CA and CPR model has been well-described and validated in our previous studies [29, 30]. Rats were anesthetized and asphyxia was induced by neuromuscular blockade with vecuronium and was maintained for 8 minutes. CA was defined as the time at which electrocardiogram (ECG) was flat and mean arterial pressure (MAP) decreased <10mmHg. CPR was initiated, ventilator support was resumed, and resuscitation drugs (epinephrine 0.5mL/kg, sodium bicarbonate 1mL/kg) were administered. The time of the ROSC was defined as reaching a MAP >60mmHg. The body temperature was maintained at 36.5 ~ 37.5 °C by continuous rectal temperature monitoring.

GBC and hNSCs Treatment

A stock solution of drug formulation (#G2539; Sigma, St. Louis, MO) was prepared by dissolving 25mg GBC in 10mL dimethyl sulfoxide (DMSO) and diluted in saline before use, as described previously [12]. Diluted GBC was injected intraperitoneally with a loading dose (10 μg/kg) at 10 min after ROSC and three maintenance doses of (1.6 μg/kg) at 8, 16, and 24h after ROSC. These dose selections of GBC were previously shown to be effective without adverse effects of hypoglycemia [13]. The collected hNSCs (0.4×10⁶ hNSCs in 40 μl saline) were administered intranasally 3h after ROSC. The hNSCs were injected into the right and left nare twice, with an interval of 10 minutes between administrations. Each nare received a 10μl injection each time. Animals remained comatose after ROSC, and no anesthesia was needed during the intranasal delivery.

Electroencephalogram (EEG) record and analysis

Our prior preclinical study has demonstrated an association between EEG recovery and functional outcome after CA [29, 30]. As such, a four-channel EEG system (Tucker-Davis Technologies, USA) was used to record EEG signals. Baseline EEG data was collected for 5 min prior to CA, followed by 5 min washout recording without isoflurane to limit the confounding influence of isoflurane sedation. EEG data was transferred into the form of a quantitative EEG - information quantity (qEEG-IQ), which is a quantitative EEG indicator validated in our previous studies [29, 30]. After baseline correction, the qEEG-IQ at 1–3h post-ROSC was compared between groups.

Arterial blood gas (ABG) and lac assessments

ABG was recorded at baseline and 20 min after ROSC using the i-STAT1 system (Abaxis, Union City, CA), and ventilator adjustments were made based on ABG results. Blood lactic acid (lac) is a major biochemical parameter reflecting organ perfusion, and recovery of normal lac has been implicated as a concern in brain injury after CA [31]. Lac was collected as part of the ABG collection at baseline and 20 min after ROSC.

Neurology function evaluation

Neurological deficit scores (NDS, total score:80) were used to assess neurological function at 24h, 48h, and 72h post-CA [30, 32, 33]. In addition, we defined the duration of NDS<60 as the threshold of severe neurological deficit (SND) [30]. Finally, the survival rates were compared among the three groups.

Cresyl Violet Staining

Quantitative assessments of neuronal damage were performed in the hippocampus by cresyl violet staining, as we previously described [32]. Three traditional regions (CA1, CA2, CA3) in the hippocampus were used to determine the degree of ischemic injury. Coronal slides taken 3mm posterior to the bregma were randomly selected to evaluate the hippocampal region according to a standard rat’s brain atlas, and the pre-determined random field was set as the region of interest. All slices were observed in the pre-defined set of random bright-field (50/200 magnification) of the Leica DMi8 microscope (Leica Microsystems, Wetzlar, Germany). Neuronal damage in the areas of interest was quantified using histopathological damage scoring (HDS) [32].

Immunofluorescence staining

BV2 cells were fixed and incubated overnight with rabbit anti-NLRP3 (1:500, Abcam, San Francisco, CA), rabbit anti-TLR4 (1:500, Abcam, San Francisco, CA) and rabbit anti-Caspase-1 (1:500, Proteintech, Rosemont, IL) at 4°C. The number or proportion of positive cells with immunofluorescence signal was quantified by counting.

Frozen brain slices were incubated with the primary antibody goat anti-Iba1 (1:250, Abcam) to mark microglia, and the percentage of microglial cells in the activated or resting state was separately counted according to morphological characteristics [34]. Mouse anti-Ku86 antibody (1:500, Santa Cruz Biotechnology, Dallas, TX) was selected to track transplanted hNSCs. Rabbit anti-iNOS (1:50, Abcam) and mouse anti-CD206 antibodies (1:100, Abcam) were used separately to classify the pro-inflammatory (M1) and anti-inflammatory (M2) phenotypes of microglia [35]. These antibodies respectively reflect proinflammatory production of NO and citrulline by M1 microglia [36] and the function of pinocytosis and phagocytosis in M2 microglia [37, 38]. Rabbit anti-NLRP3 (1:100, Abcam) and rabbit anti-caspase-1 (1:100, Proteintech) were used to localize and semi-quantify the proteins related to the NLRP3-inflammasome signaling pathway. Goat anti-IL-1β (1:100, Invitrogen, Carlsbad, CA) was used to detect the expression of the inflammatory-related cytokine. We randomly included coronal slices containing the whole hippocampus, set the pre-determined random field as the region of interest, and used rat brain atlas to match groups to ensure that quantitative comparison was performed in the random areas originating from similar coronal planes. ImageJ or manual method was selected for cell counting, as we previously described [39]. The integrated density (IntDen) was applied in the quantification of positive signals (IL-1β). All immunofluorescence slices were observed in a Leica DMi8 microscope (Leica Microsystems, Wetzlar, Germany).

Western Blot

After removing the cerebral cortex and exposing the hippocampus, bilateral hippocampal tissue was collected for Western Blot analysis. The proteins of hippocampal tissue were transferred to the polyvinylidene fluoride membrane (PVDF; Millipore Sigma, USA) that was blocked with 5% milk for 2 h at room temperature and then incubated overnight at 4 °C with the following primary antibodies: anti-TLR4 (1:500, Millipore Sigma, USA), anti-NFκB (1:500, Millipore Sigma), anti-IL-1β (1:5000, Millipore Sigma). ImageJ software was used to measure the band densities by densitometry and quantify. β-actin (1:5000, Proteintech) was used as a reference and results were output with arbitrary units (A. U.).

Statistical Methods

Parametric data (such as qEEG-IQ, positive cell count, and proportion of positive cell number) were shown as mean ± S.E.M. and non-parametric data (such as NDS) were presented as median and 25–75^th interquartile range. Univariate analysis with post-hoc analysis was conducted to analyze parametric data such as immunofluorescence and qEEG-IQ. Repeated measurement was used to analyze aggregate NDS, and a nonparametric test (Mann–Whitney U) was used to compare NDS at each time point. Differences were considered significant at p < 0.05: ***P < 0.001, **P < 0.01, *P < 0.05. All statistical analyses were performed using SPSS (IBM SPSS Statistics, version 22).

Results

hNSCs decreased OGD/R induced BV2 loss

We first verified and compared the protective effects of GBC and hNSC on BV2 cells treated with OGD in vitro. After OGD/R injury, the cell viability in all three groups that underwent OGD decreased compared to the normal group (p < 0.001, Fig 1A). Administration of GBC had no significant effect on cell viability. Co-culture with hNSCs significantly increased cell viability compared to OGD (p < 0.05, Fig 1B) or OGD + GBC groups (p < 0.01, Fig 1B). Activated BV2 cells induced by OGD/R showed enlarged cellular bodies (Fig 1A, red arrow). A lower proportion of activated BV2 cells was found in the OGD + GBC group (p < 0.01, Fig 1C) and OGD + hNSCs group (p < 0.001, Fig 1C) compared to the OGD group. In addition, the OGD + hNSCs group had a lower proportion of activated BV2 cells when compared to the OGD + GBC group, (p < 0.05, Fig 1C).

Figure 1.

Proliferation and activation of BV2 cells after OGD. Cell viability decreased in three OGD-treated groups compared to the normal group (A). More activated BV2 cells with enlarged cellular bodies were found in the control group (A, red arrows). Further quantification showed that hNSCs treatment markedly protected the cells’ ability to proliferate (B) and inhibited the cells’ activation (C) after OGD. The TLR4 / NLRP3 signaling pathway was activated in BV2 cells after OGD. Immunofluorescence quantification of TLR4, NLRP3, and caspase-1 showed that TLR4/NLRP3-related inflammation was maximally up-regulated in the control group (D). hNSCs and GBC treatment downregulated the expression of inflammatory molecules, and the downregulatory effect of hNSCs treatment was more profound compared to GBC treatment (E, F, G). hNSCs or GBC treatment induced changes of phenotype related to the inflammatory process in BV2 cells. After OGD, either pro-inflammation (M1, marked with iNOS) or anti-inflammation (M2, marked with CD206) phenotype BV2 cells increased in the three OGD-treated groups (H). hNSCs and GBC- treated BV2 cells showed significant inhibition of M1 polarization and promotion of M2 polarization. The co-culture of hNSCs had a more significant effect than GBC on the above regulatory process (I, J). *: p < 0.05; **: p < 0.01; ***: p < 0.001. 200/400X, Scale bar = 75μm/50μm.

hNSCs inhibited OGD/R mediated inflammatory activation of BV2 cells

We further confirmed the changes of TLR4 / NLRP3 signaling pathway related proteins after GBC and hNSC treatment in vitro. The control BV2 microglia showed small soma with distal arborization. Activated BV2 cells induced by OGD/R showed enlarged cellular bodies with fewer branches that were shorter and/or resorbed into the cell body, and increased expression of TLR4/NLRP3 signaling-pathway-related proteins, including NLRP3, TLR4, and Caspase-1 (Fig 1D). The expression of these signaling pathway proteins was inhibited in both GBC and hNSC treated cells (p < 0.01, Fig 1E/F/G). Furthermore, hNSC treatment exhibited a stronger effect than GBC on the down-regulation of these proteins in BV2 cells (p < 0.05, Fig 1E/F/G).

hNSCs regulated the anti-inflammation phenotypes of BV2 cells

The in vitro results based on microglial polarization phenotypes showed that both GBC and hNSC significantly inhibited the polarization of microglia to M1 and promoted the polarization to M2, and this therapeutic effect showed a significant difference between the two groups. BV2 cells emerged as M1 (pro-inflammation) and M2 (anti-inflammation) phenotypes based on mutually exclusive processes involved in the inflammatory response [40]. The cell number proportion of M1 phenotype microglia marked with iNOS in GBC and hNSCs groups were significantly lower than that in the control group (p < 0.001, Fig 1H/I). At the same time, the comparison between treatment groups showed that the hNSCs group had the lowest cell number proportion of M1 phenotype cells (p < 0.01, Fig 1H/I). In terms of the anti-inflammatory phenotype, the hNSCs group had a significantly higher proportion of M2 BV2 cells marked with CD206 (hNSCs VS. Control: p < 0.001, hNSCs VS. GBC: p < 0.001, Fig 1H/J). GBC treatment also promoted M2 phenotype polarization after OGD compared to the control group (GBC VS. control: p < 0.001, Fig 1H/J).

hNSCs improved neurological function and shortened the duration of severe neurological deficit

In vivo study showed that the treatments of GBC and hNSCs improved NDS and shortened the duration of SND. hNSCs showed a greater neuroprotective effect compared to GBC. The aggregate analysis of surviving NDS showed a significant improvement in the hNSCs and GBC groups compared to the Control group (hNSCs VS. Control: p < 0.01, GBC VS. Control: p < 0.05). The aggregate comparison between the two treatment groups showed that the hNSCs group had better functional recovery than the GBC group (hNSCs VS. GBC: p < 0.01). We also found that the hNSCs group had markedly better recovery at all evaluation time points (24, 48, and 72h) compared to the Control group (24h: p=0.004; 48h: p=0.019; 72h: p=0.032, Fig 2A). The hNSC group had better NDS recovery at 24h compared to the GBC group (hNSCs VS. GBC: p = 0.049, Fig 2A).

Figure 2.

Quantitative analysis of neurological function and outcome between groups. The aggregate analysis of surviving NDS shows a significant functional improvement in hNSCs and GBC groups compared to the control group (NSCs VS. Control: p < 0.01, GBC VS. Control: p < 0.05). Further comparison between the two treatment groups showed that the hNSCs group had better functional improvement than the GBC group on aggregate surviving NDS (hNSCs VS. GBC: p < 0.01). Of note, hNSCs therapy displayed better functional recovery than the GBC group at all assessment points (A). Both therapeutic interventions significantly shortened the length of time under severe neurological deficit (SND) (B). The cumulative survival analysis exhibited a trend of improved survival after hNSCs or GBC therapy (C). *: p < 0.05; **: p < 0.01.

The duration of SND in GBC (40.5 ± 11.3h) and hNSCs (26.25 ± 11.8h) groups was shorter than in the Control group (69.0 ± 3.0h), indicating that both hNSCs treatment and GBC treatment consistently and significantly reduce the duration of severely impaired state after resuscitation (hNSCs VS. Control: p = 0.005; GBC VS. Control: p = 0.048, Fig 2B). Kaplan-Meier analysis showed a trend of improvement of the survival rate in both treatment groups compared to the Control group (4/8, 50%), and the survival rate was similar at 72hr in the GBC and hNSCs groups (both 6/8, 75%), (Fig 2C).

qEEG-IQ recovery

Quantitative analysis of qEEG-IQ showed that there was no significant difference between the three groups when compared separately in these time periods (Supplementary Table 1, Supplementary Fig 1).

Negative Correlations between blood lac and NDS in biochemistry analysis

The correlation analysis between lac and neurological outcomes showed that the lac level 20 minutes after ROSC was negatively correlated with the recovery of neurological function. The primary ABG baseline parameters of the three groups are shown in Supplementary Table 2. Bivariate analysis was performed to evaluate the correlation between blood lac and neurological function (NDS) at different time points. The blood lac was negatively correlated with the NDS of rats at all time points. The correlations were significant at 24h (Correlation coefficient = −0.462, p = 0.039), 48h (Correlation coefficient = −0.550, p = 0.017), and 72h (Correlation coefficient = −0.626, p = 0.008) after ROSC (Supplementary Fig 2).

Therapeutic intervention alleviated histopathological damage

Histological staining by cresyl violet demonstrated that therapeutic intervention alleviated histopathological damage. Histopathological damage scoring (HDS) revealed variable tolerance to ischemic insult in different regions of the hippocampus. Region CA1 was the most ischemic-sensitive area, with the highest HDS, and CA3 was the least affected hippocampal region, with the lowest average HDS (Fig 3A). Compared to the Control group, the hNSCs group had less neuronal damage in CA1 (hNSCs VS. Control: 38.4% ± 7.4% VS. 78.3% ± 5.9%; p = 0.001), CA 2 (hNSCs VS. Control: 10.1% ± 2.8% VS. 49.8% ± 8.2%; p < 0.001) and CA3 (hNSCs VS. Control: 2.3% ±0.6% VS. 28.7% ± 8.8%; p = 0.031), (Fig 3B). The comparison between the two treatments indicated that the protective effect of hNSCs therapy was significantly stronger than GBC (60.2% ± 6.6%) in the ischemia-sensitive area of CA1 (p = 0.039, Fig 3B).

Figure 3.

Ischemic neuron assessment by cresyl violet staining. The damaged neurons in the hippocampus were evaluated by region (CA1, CA2, CA3). Neurons in the CA1 region underwent the most serious injury after ischemia, as this region had the highest proportion of damaged neurons (A). Quantitative analysis (HDS, %) among the three groups showed that the percentage of injured neurons in the hippocampus of rats transplanted with hNSCs was significantly lower than that of the control group in CA1, CA2, and CA3 (B). In the ischemia-sensitive region, CA1, hNSCs therapy exhibited a stronger protective effect than GBC (B). *: p < 0.05, **: p < 0.01; 50/200X, Scale bar = 500 /100μm.

Microglia’s recruitment among groups in the hippocampus

We further analyzed microglial recruitment in the hippocampus. The quantitative results from Iba-1 fluorescence staining showed that microglial recruitment was inhibited to different degrees in both treatment groups, and the inhibition was more significant in the hNSC group. Transplanted hNSCs (Ku86, red) were located in the entire hippocampus in the hNSCs group 3 days after treatment. Staining using Iba-1 (green) showed that more microglia with the larger cell bodies were activated in the Control group (Fig 4A). Further quantification analysis indicated that the increases of microglia in the GBC (372.4 ± 79.6/mm², p = 0.006) and hNSCs groups (222.1±54.1/mm², p = 0.002) were significantly lower than that in the Control group (739.3 ±95.9/mm², Fig 4B).

Figure 4.

The recruitment of microglia in the hippocampus and tracking of transplanted hNSCs after CA. Immunofluorescence-labeled microglia showing differences in microglial recruitment and activation between groups after ischemia-reperfusion injury of CA (A). Quantitative results showed that the GBC and hNSCs therapy suppressed the recruitment and activation of microglia (B). **: p < 0.01; 50X, Scale bar = 50/250μm.

Treatment suppressed the activation subtype of microglia

Comparison of the cell-morphologies of microglia showed that the proportions of cells in the rest and activated states were markedly changed by the two treatments. The functional states of microglia involved in the process of inflammation were defined and classified based on validated methods [34, 41, 42]. In brief, microglia were classified as resting (Type 1), initiating microglial activation (Type 2), activated but non‐phagocytic (Type 3), and phagocytic (Type 4) according to morphology [41].

In the hippocampus, type 1 microglia were the most common type in the hNSCs group, with a higher proportion compared to the GBC group (p = 0.017) or Control group (p = 0.002). In contrast, the proportion of type 3 activated microglia was markedly lower in the hNSCs group compared to the GBC (p = 0.041) or Control groups (p = 0.029), (Fig 5A/B). A similar pattern of microglial subtypes was present in the cortex. The proportion of type 1 microglia in the hNSCs and GBC groups was prominently higher than in the Control group (hNSCs VS. Control: p = 0.003; GBC VS. Control: p = 0.009). Similarly, the highest proportion of tyle 3 and type 4 activated microglia was found in the Control group, and this proportion was significantly higher than the proportion of these cells in the hNSCs group (type3: p = 0.022; type4: p = 0.036), (Fig 5C/D). Microglia were classified according to morphology (Fig 5E). More detailed data on microglia activation subtypes are shown in Supplementary Table 3.

Figure 5.

The definition and comparison of microglial resting-state vs. activated state based on morphology in the cortex and hippocampus. In the hippocampus, the hNSCs group retained the maximum resting-state microglia (type 1) and minimum activated, non-phagocytic microglia (type 3) compared to GBC or control groups (A, B). In the cortex, the hNSCs group and GBC group had significantly higher proportions of resting-state microglia (type 1) compared to the control group. The hNSCs treatment group had the smallest proportion of activated microglial cells (type 3 and type 4) in the cortex (C, D). Microglia classification according to morphology (E). *: p < 0.05, **: p < 0.01; 400X, Scale bar = 25μm.

hNSCs therapy changed the dominant polarization phenotype

The analysis of microglial polarization phenotype with iNOS (M1) and CD206 (M2) indicated that both hNSC and GBC treatments had a similar effect on decreasing the average percentage of M1 microglia and increasing the average percentage of M2 microglia. In the M1 phenotype comparison, the percentages of iNOS positive cells in all Iba-1 positive microglia in the hNSCs and GBC groups were significantly lower than that in the Control group (hNSCs VS. GBC VS. Control: 5.77% ± 2.14% VS. 15.26% ± 3.48% VS. 40.33% ± 3.07%. hNSCs VS. Control: p < 0.001, GBC VS. Control: p = 0.001, Fig 6A/B). In addition, the hNSCs (43.0% ± 7.1%) and GBC (38.9% ± 8.9%) groups had significantly increased percentages of CD206/Iba-1 positive cells compared to the Control group (15.8% ± 3.6%). (hNSCs VS. Control: p = 0.012; GBC VS. Control: p = 0.03, Fig 6C/D).

Figure 6.

hNSCs inhibit the polarization of pro-inflammatory phenotype of microglia (M1) in the hippocampus. (A) In the control group, Iba-1 positive cells were dense, and more iNOS positive cells were found compared to the treatment groups. (B) Quantitative analysis showed that the proportion of M1 polarization microglia in the control group was significantly higher than that in the hNSCs group and GBC group. (C) hNSCs treatment promoted polarization of the anti-inflammatory phenotype of microglia (M2). Immunofluorescence staining revealed more CD206 positive cells (M2) were co-located with Iba-1 positive cells. (D) The effects of GBC and hNSCs were similar in promoting M2 phenotype polarization in microglia compared to the control group. **: p < 0.01; ***: p < 0.001; 200/400X, Scale bar = 100/50μm.

TLR4/NLRP3 signaling pathway activated in microglia

TLR4 expression in microglia was down-regulated by hNSCs or GBC

Our results indicated that TLR4 was expressed in a portion of microglia (Fig 7A). Quantified positive cells identified by double labeling with TLR4 and Iba-1 showed that the proportions of double-positive cells in the GBC group (20.2% ± 1.4%) and hNSCs group (12.9% ± 2.4%) were significantly lower than that in the Control group (37.5% ± 5.6%) (hNSCs VS. Control: p<0.001; GBC VS. Control: p <0.01, Fig 7B).

Figure 7.

TLR4 expression in microglia was downregulated by hNSCs and GBC treatment. Representative double immunofluorescence staining with TLR4 and Iba-1 showed that TLR4 was expressed in a variety of central nervous system cells, including microglia (A). Both TLR4 and Iba-1 expression had an increasing trend in the control group (A). The quantitative results showed that after treatment, the percentages of TLR4+ / Iba-1+ cells in microglia were significantly decreased in the GBC and hNSCs groups (B). Therapeutic intervention inhibited the expression of the NLRP3 inflammasome. Using co-staining with Iba-1, microglia with an expression of NLRP3 were double marked in 10X and 40X microscopic fields. All groups had the expression of NLRP3 at different levels (C). Further quantitative analysis based on the proportion of NLRP3 positive cells showed that hNSCs therapy had a significant and stronger regulatory effect on the expression of this key protein compared to GBC (D). Evaluation of the expression of Caspase-1 in microglia showed the proportion of Caspase-1 positive microglia (marked as both Caspase-1 and Iba-1 positive) in all microglia (marked as Iba-1 positive) decreased after treatments (E). There was a significant decrease in this proportion in the two treatment groups compared to the Control group. In addition, the comparison between the two treatment groups suggested that the hNSCs group had a stronger inhibition of Caspase-1 expression in microglia (F). Localization and quantification of IL-1β expressed in the three groups showed IL-1β was expressed in a small amount in the extracellular area (G). The levels of IL-1β in GBC and hNSCs treated groups were similar and significantly lower than in the control group (H). *: p < 0.05, **: p < 0.01, ***: p < 0.001; 50 100/200/400X, Scale bar = 200/100/50μm.

NLRP3 positive cells in microglia reduced after treatment

We found the expression of NLRP3 immunofluorescence signal in some microglial cells (Fig 7C). Quantified staining of microglia showed that the proportion of NLRP3 positive cells in all Iba-1 positive cells was highest in the Control group (27.2% ± 5.8%), intermediate in the GBC group (15.7% ± 1.8%), and lowest in hNSCs group (4.4% ± 0.8%) (Fig 7D) (hNSCs VS. Control: p<0.001; GBC VS. Control: p = 0.034; hNSCs VS. GBC: p = 0.038, Fig 7D).

Inhibition of the inflammasome-related protein in microglia

Formation of the NLRP3-inflammasome activates Caspase-1, and Caspase-1 initiates the process of proteolysis and converts pro-IL-1β into bioactive IL-1β, which is subsequently secreted by the cell [43]. We used cell number proportion of Caspase-1+ microglia (marked as both Caspase-1+/Iba-1+) to all microglia (marked as Iba-1 positive) to evaluate the caspase-1 expression (Fig 7E). We found that there was a significant decrease of this proportion in the hNSCs group (28.9% ± 2.9%, hNSCs VS. Control: p < 0.01) or GBC group (48.8% ± 6.2%, GBC VS. Control: p = 0.034) compared to the Control group (67.1% ± 6.4%), (Fig 7F). The proportion of caspase-1 expression in the hNSCs group was significantly lower than in the GBC group (P=0.023), (Fig 7F). This inhibitory effect was verified by evaluating the integrated density (IntDen) of IL-1β expression (Fig 7G). Compared to the Control group (6030.2 ± 1124.5), hNSCs transplantation (2798.8±350.2, p = 0.009) or GBC treatment (2991.7 ± 619.5, p = 0.014) markedly downregulated IL-1β expression (Fig 7H).

TLR4/NLRP3 signaling proteins inhibited by therapies

TLR4 (receptor expression), NFκB (transcription factor), NLRP3 (assembly of inflammasome), and IL-1β (inflammatory cytokine expression) are four important nodes in the activation of the TLR4/NLRP3 signaling pathway [44]. We evaluated the changes in expression of these steps after therapeutic intervention (Fig 8A). Western blot results demonstrated that the expression of TLR4, NFκB, NLRP3, and IL-1β was significantly downregulated by hNSCs and by GBC therapy (TLR4: hNSCs VS. Control, p < 0.001, GBC VS. Control: p = 0.007. NFκB: hNSCs VS. Control, p < 0.001, GBC VS. Control: p = 0.005. NLRP3: hNSCs VS. Control, p < 0.001, GBC VS. Control: p < 0.001. IL-1β: hNSCs VS. Control, p < 0.001, GBC VS. Control: p < 0.001, Fig 8B/C/D/E). It is also notable that there are significant differences between the two therapies in regulating the activation of the signaling pathway (TLR4, hNSCs VS. GBC: p = 0.029. NFκB, hNSCs VS. GBC: p = 0.02. NLRP3, hNSCs VS. GBC: p = 0.018. IL-1β: hNSCs VS. GBC: p = 0.042. Fig 8B/C/D/E). Detailed data on the expression of these proteins are also shown in Supplementary Table 4.

Figure 8.

The neuroinflammatory regulation of hNSCs therapy is driven by the TLR4 / NLRP3 pathway. Three rats in each group were included for western blot (WB) quantitative analysis. (A) Representative Western Blots for four key proteins involved in receptor expression, signal transduction, assembly of the inflammasome, and expression of inflammatory factors in the TLR4 / NLRP3 pathway. (B-E) The quantitative analysis between the three groups showed that GBC treatment and hNSCs treatment significantly reduced the expression of these proteins compared to the control group (***: p < 0.001) (arbitrary units, A. U.). The comparison between the two treatment groups showed that the downregulatory effect of hNSCs treatment on TLR4/NLRP3 pathway-related proteins was greater than the GBC group. (*: p < 0.05, **: p < 0.01).

DISCUSSION

This study confirmed the therapeutic effects of GBC and hNSC therapies in ischemic brain injury after CA via immunoregulation and is the first to demonstrate the shared molecular mechanisms of immunoregulation therapy with chemical drugs (GBC) and cell therapy (hNSC). Our study has provided scientific support that these two treatments provide therapeutic benefit in neurological prognosis of rats after cardiac arrest at three levels: animal, cell, and molecule. Intranasal delivery of hNSCs is feasible, achieves a satisfactory therapeutic effect, and holds stronger efficacy in improving functional outcomes after CA than GBC therapy in the early phase (first 72 hours) after CA. Further, we have unveiled the immune mechanisms related to these two therapies, namely the downregulation of the TLR4/NLRP3 inflammasome pathway (Fig 9).

Figure 9.

(A) Proposed two-way model of TLR4/NLRP3 signal pathways. TLR4 senses multiple compounds (signal 1) including lipopolysaccharide (LPS) and induces activation of the transcription factor NF-κB. This causes the expression of NLRP3 and proIL-1β. NLRP3 senses different pathogen-associated molecular patterns (PAMPs) and danger-associated molecular patterns (DAMPs) (signal 2), which drives the assembly of the NLRP3 inflammasome (NLRP3, ASC, and pro-caspase-1). Caspase-1 is activated by pro-caspase-1 and cleaves pro-IL-1β (pro-inflammatory cytokine). Finally, the mature cytokines induce inflammation. (B) Possible mechanism of TLR4 / NLRP3 signaling pathway affecting microglia polarization. ASC: apoptosis-associated speck-like protein containing a CARD (Caspase activation and recruitment domain). DAMPs: danger-associated molecular patterns. IL: interleukin. NFκB: nuclear factor κ-light-chain-enhancer of activated B-cells. PAMPs: pathogen-associated molecular patterns.

Neuroprotective therapies for brain injury after CA remain a persistent clinical challenge, despite modest improvements in outcomes after the institution of targeted temperature management (TTM). Preclinical studies have suggested that hNSC transplantation may be a promising therapeutic strategy, with both cell replacement and immunomodulatory mechanisms [45, 46]. In the more widely explored field of regional ischemic brain injury, different opinions exist regarding the detrimental versus beneficial impact of microglia in the acute phase of ischemic brain injury, as it is understood that neuroinflammation plays roles in both promoting neurological repair and exacerbating neuronal injury [4]. This suggests that neuroinflammation may play a variable and time-dependent role in the process of pathological damage and repair after ischemic brain injury [47]. Unlike the regional brain injury of ischemic stroke, global ischemia after CA and its induced neuroinflammation may more widely affect brain tissue fate, which brings greater uncertainty to the progress and repair of this type of injury. This study indicates that stem cell therapy targeting inflammatory pathways plays an earlier role than the benefit through cell replacement mechanisms, which may have a role in a later stage after resuscitation. This opens the door to an early treatment window for critically ill patients.

Our comparison of hNSCs and GBC indicates that animals may have a stronger benefit from NSC therapy than from GBC. In an attempt to explain this difference in therapeutic effect for future clinical translation, we performed tracking and localization of NSCs and microglia. Three days after NSC transplantation, immunofluorescence signals targeting the human nucleus (Ku86+ cells) were still found in the hippocampus, indicating that grafted cells had a retention period of at least several days. When combined with the cells’ proliferation abilities, this maintains the stability of this intervention well beyond the day of administration. In contrast, GBC therapy was inevitably affected by drug pharmacokinetics, with an average half-life of 3.3 +/− 1.5 hours [48], which leads to an unstable concentration and makes it difficult for the drug to distribute to the injured brain. Although this study adopted the most conventional administration regimen for GBC based on published studies [11, 49–51], future studies with a better temporal pharmacokinetic profile may potentiate improvement by the treatment. Our in vitro experiment showed lack of protection with GBC compared to hNSCs in an OGD environment, which may be attributed to the hypoglycemic function of GBC, potentially leading to a more rigorous glucose deprivation that overcame the protective action of the drug. Further understanding requires future investigations.

Comatose post-CA patients are admitted to intensive care units (ICUs) due to high mortality, leading to higher costs of care [52]. Thus, ICU length of stay (LOS) is of clinical significance. We analyzed the treatment effect of GBC and hNSCs on the duration of SND to correspond to ICU LOS, and we demonstrated that both therapies shorten the duration of SND after CA, with hNSCs having a greater therapeutic effect. Unlike prior studies, in which cell transplantation took place 24h to 7 days after injury [45], this study innovatively provides an early (<24h) therapeutic strategy, contributing to earlier recovery from SND and mitigating the final severity of brain injury for subjects suffering CA.

Our two treatment groups achieved the same trend of improved survival rate, although no significant difference was demonstrated when compared to the control group. The death cases were limited in each group (2 deaths in each treatment group and 4 deaths in the Control group), and limited sample size can make it difficult to distinguish survival differences between groups. Another potential explanation for this survival finding is that this result may be related to the changes in the brain microenvironment after hypoxic-ischemic injury [53, 54]. It can be inferred that individuals with the most severe brain injury after ischemia would have a worse brain microenvironment, which would inevitably have a negative influence on the survival of NSCs, and ultimately, would weaken the therapeutic effect of NSC transplantation on rats that may die in the end. It should be noted that in clinical translation, functional outcome is a more important measure than simple survival. This study demonstrated improved functional outcomes in both treatment groups, with the hNSC group having even better functional recovery than the GBC group.

Until recently, cell replacement was considered the primary therapeutic mechanism of NSCs transplantation [55]. However, in our study, we demonstrate that the residential neurons in the hippocampus had better fate at 72-hours with NSCs treatment, indicating that there is a neuroprotective mechanism independent of cellular replacement that occurs after NSCs transplantation. We observed microglial activation and involvement in the neuroinflammatory process from three sequential perspectives. In the early post-ischemic phase, microglia migrate to the injured area and proliferate [5, 56]. Our results indicated that the control group rats had a more pronounced increase of Iba-1 positive cells in the hippocampus, indicating that treatment with NSCs or GBC produced a marked effect on immunoregulation, even in the very early stage of microglial behavior. In the next stage of activation, microglia undergo obvious morphological changes. Specifically, the morphology of the microglia changes from the resting state, composed of a small soma with multiple thin branches, to the activated state, characterized by a large soma and a few thick branches [34]. Our results at this stage showed that there were more resting state microglia in the hippocampus and cortex of rats treated with NSCs or GBC, whereas the control group was composed of more activated microglia. At the final stage, activated microglia will polarize into two phenotypes with opposite functions, M1 vs. M2, with the M1 phenotype considered to be pro-inflammatory and the M2 phenotype considered to be anti-inflammatory [57]. At this final stage of polarization, we again found that NSC and GBC treatment resulted in the recruitment of more anti-inflammatory M2 microglia compared to the control group, which displayed more M1, inflammatory microglia. These results constitute complete regulation of the microglial inflammatory activation process and indicate the important involvement of neuroinflammation as an important target of these specific neuroprotective therapies.

Our recent studies have shown that the neuroprotective effects of GBC [58] may be associated with down-regulation of NLRP3 inflammasome signaling [59]. Based on this prior evidence, the present study investigated the NLRP3 inflammasome as a key signaling protein. We made a further extension to include TLR4, given that the classic TLR4 / NLRP3 signaling cascade consists of signal 1 (TLR4) and signal 2 (NLRP3) [44], as shown in the diagrammatic sketch. In our in vitro experiment, activated BV2 cells induced by OGD/R showed higher expression of NLRP3, IL-1β, TLR-4, and Caspase-1, which could be decreased by GBC and strongly suppressed by hNSCs co-culture. In our in vivo experiments, TLR4/NLRP3-related signaling proteins, such as TLR4, NFκB, NLRP3, caspase-1, and IL-1β, were uniformly downregulated in the two treatment groups, with more pronounced downregulation in the hNSCs group. This evidence provides a connection linking early inflammatory cell activation, namely microglial activation, with inflammatory signal transduction through the NLRP3 inflammasome and demonstrates the ability to regulate this cascade via GBC and hNSC therapy, with the suggestion that hNSCs may provide a stronger down-regulation.

The existing studies on NSCs vary in delivery time from 24 hours to 30 days [60–62]. The current study suggests that intranasal administration of hNSCs in the very early phase after CA (3h) is feasible and leads to improved functional outcomes and shorter duration of severe neurological deficit, with a duration of action beyond initial administration (at least 3 days). This may suggest that a single dose can provide a lasting benefit, presenting a feasible, realistic therapy for critically ill patients. In contrast, while GBC therapy also provided improved functional outcomes, shortened severe neurological deficit, and inflammatory downregulation, the strength of these effects was not as robust as hNSC treatment. This may suggest that additional repeated doses or a better temporal pharmacokinetic profile of GBC drug therapy may be needed to achieve similar outcomes. Risks of hypoglycemia with GBC may limit its use in critically ill patients. Implications in clinical practice would be dependent on therapy availability and ease of administration. At present, GBC is the more widely available intervention, but, if available, a single-dose treatment such as hNSCs could present an easier option for compliance and delivery. Long term benefit and outcomes or potential side effects of these two treatments requires further investigation, but this study suggests that efforts focused on the advancement and translation of hNSC therapy after CA may be of greater benefit.

This study is limited in that control investigations did not include sham vehicles in the in vitro studies. In the in vivo studies, the GBC control did not receive a sham injection of saline, and the hNSC control did not receive a sham intranasal administration. The amounts of vehicle used in the experimental groups were felt so small that they would not affect the results. This study draws an association of neurological injury/function after CA with the TLR4/NLRP3 inflammatory pathway but does not provide clear-cut proof of a cause and effect relationship. Using more advanced molecular biology methods, such as gain/loss of function models and gene transfer models, might further validate our immunostaining findings. The use of pathway agonists and antagonists may aid in developing a more cause and effect association in the future. This study provides a launching pad for future studies to continue aspiring toward a better understanding of these interventions to improve functional outcomes after CA. At present, no clinical trial has evaluated either treatment, GBC or hNSC, in post-cardiac arrest patients. These individual studies likely require evaluation before a randomized trial comparing the two therapies could be considered. Ultimately, this would provide the clearest understanding of the translational, clinical significance of our preclinical findings.

Conclusions

hNSC and GBC therapy regulate microglial activation and the neuroinflammatory response in the brain after CA via the TLR4 / NLRP3 signaling pathway, leading to neuroprotective effects, improved neurological function, and shorter time of severe neurological deficit. In most comparisons of functional outcomes, inflammatory signaling pathways, and microglial activation, hNSCs showed a more robust therapeutic effect in immunomodulation and functional outcome when compared to GBC, owing to the characteristic differences in cell therapy vs. drug therapy.

Supplementary Material

Supplementary Fig 1

Quantitative comparisons show qEEG-IQs were similar among the three groups at baseline, CA periods, 1–2h post-ROSC, and 2–3h post-ROSC. The results showed that qEEG-IQ recovered over time. The degree of brain injury in the earliest stage (0–3h) after CA was similar among the three groups.

Supplementary Fig 2

Correlation analysis of blood biochemical indices (blood lactate) and neurological function. Kendalls’ tau-b correlation analysis revealed that there were significant negative correlation between blood lac tested at 20min after ROSC and NDS at 24h (CC = −0.462, P = 0.039), (A); 48h (CC = −0.550, P = 0.017), (B); 72h (CC = −0.626, P = 0.008), (C) after ROSC. CC: Correlation coefficient.

Abbreviations

NSC

neural stem cell

CA

cardiac arrest

GBC

glibenclamide

OGD

oxygen-glucose deprivation

hNSCs

human NSCs

TLR4

Toll-like receptor 4

NLRP3

NLR pyrin domain containing 3

NDS

neurological deficit scores

ROSC

return of spontaneous circulation

IntDen

integrated density

OGD/R

oxygen-glucose deprivation and re-oxygenation

AD

Alzheimer’s disease

DMEM

Dulbecco’s Modified Eagle Medium

bFGF

basic fibroblast growth factor

EGF

epidermal growth factor

FBS

Fetal Bovine Serum

CPR

cardiopulmonary resuscitation

MAP

mean arterial pressure

ABG

arterial blood gas

ECG

electrocardiogram

EEG

Electroencephalogram

DMSO

dimethyl sulfoxide

qEEG-IQ

quantitative EEG - information quantity

lac

Blood lactic acid

SND

severe neurological deficit

PFA

paraformaldehyde

HDS

histopathological damage scoring

BSA

bovine serum albumin

RIPA

radioimmunoprecipitation assay

PMSF

phenylmethanesulfonyl fluoride

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

PVDF

polyvinylidene fluoride membrane

ECL

enhanced chemiluminescence

A. U.

arbitrary units

TTM

targeted temperature management

ICUs

intensive care units

LOS

length of stay

ROS

reactive oxygen species

RNS

reactive nitrogen species

Data Availability

All data supporting the conclusions of this manuscript are provided in the text and figures. Please contact the author for data requests.