Astrocytic N-Myc Downstream-regulated Gene–2 Is Involved in Nuclear Transcription Factor κB–mediated Inflammation Induced by Global Cerebral Ischemia

GLOBAL cerebral ischemia (GCI), which selectively destroys CA1-region neurons in the hippocampus, occurs after cardiac arrest or systemic hypotension during surgery.¹  More than half of the patients who survive cardiac arrest suffer severe neurologic injury.^2,3  Emerging evidence suggests that neuroinflammation plays a critical role in the pathophysiology of postischemic brain injury and worsens neuronal injury.^4,5  The inhibition of inflammation has remarkable protective effects on the ischemic brain and has recently emerged as a promising therapeutic strategy.⁶ 

Astrocytes are critical for multiple functions of the central nervous system (CNS), including modulating synaptic plasticity; maintaining glucose, ion, and glutamate homeostasis; and maintaining the blood–brain barrier.^7–9  However, in response to cerebral ischemia, astrocytes become reactive, which results in the activation of immune mediators, the release of large quantities of proinflammatory cytokines, and neuronal dysfunction and death.^10,11  The heterodimeric transcription factor nuclear transcription factor κB (NF-κB) is a key regulator of inflammation in both the peripheral nervous system and the CNS.¹²  Recent studies have reported that in the CNS, astrocytes are the primary cell type under NF-κB regulation, whereas neurons exhibit negligible NF-κB activity.^13,14  In addition, the selective inhibition of astroglial NF-κB signaling markedly alleviates ischemic injury in GAP-IκBα-dn transgenic mice.¹⁵ 

Our previous study revealed that in Leydig cells treated with the toxicant ethane dimethanesulfonate, NF-κB modulated the expression of N-Myc downstream-regulated gene–2 (NDRG2),¹⁶  a newly identified differentiation- and stress-associated gene that is highly expressed in the heart, muscle, kidney, and brain.¹⁷  Interestingly, our previous study and another study have found that in the mouse brain, NDRG2 is specifically expressed in astrocytes but not in neurons.^18,19  Additionally, we observed a significant upregulation of NDRG2 expression in the ischemic penumbra after cerebral ischemia induced by middle cerebral artery occlusion.²⁰  However, the specific role of NDRG2 in cerebral ischemic injury, especially in GCI, remains elusive. In addition, there are no reports on whether NDRG2 is involved in astrocytic NF-κB–mediated inflammation after cerebral ischemia. In this study, we investigated the role of NDRG2 in global cerebral ischemia–induced inflammatory and neuronal injury, which may reveal a new molecular target for cerebral ischemia therapy.

All animal-related procedures were approved by the Ethics Committee for Animal Experimentation of the Fourth Military Medical University (Xi’an, China) and were performed in accordance with university guidelines for animal experimentation. Adult male C57BL/6 mice age 8 to 12 weeks and weighing 20 to 23 g, 1-day-old C57BL/6 pups, and pregnant C57BL/6 mice were provided by the Experimental Animal Center of the Fourth Military Medical University (Xi’an, China).

Briefly, meninges-free hippocampal tissue from 1-day-old C57BL/6 mice was collected, minced, and dissociated with 0.025% trypsin to form a cell suspension, which was then added to Dulbecco’s modified Eagle’s medium (Gibco, USA) containing 10% fetal bovine serum. The cells were cultured in 25 μg/ml poly-L-lysine–coated (Sigma-Aldrich, USA) T75 flasks and maintained at 37°C in a mixture of 95% atmospheric air and 5% CO₂. Half of the medium was replaced every 3 days. When the cells reached confluence after 3 days, the flasks were shaken at 190 rpm for 19 h to remove microglia and oligodendrocytes. After shaking, immunofluorescence staining for glial fibrillary acidic protein (GFAP) determined that more than 95% of the cultured cells were astrocytes, as shown in figure 1, Supplemental Digital Content, https://links.lww.com/ALN/B598. After purification, the cells were subcultured into different plates for various experiments.

Brains were isolated from 14- to 16-day-old embryos from pregnant C57 mice and used to culture primary hippocampal neurons using the same procedure described for primary astrocyte cultures except that neurobasal medium containing 1% glutamate and 2% B27 was used to culture the primary neurons. The neurons were cultured in 6-well plates. On the first day of culture, 5 μM Ara-C was added to the neuronal cultures to reduce glial cell contamination. Then, Transwells (3 μm, Corning, USA) containing cultured astrocytes were placed on top of the mature neuron cultures. The cocultures were then incubated at 37°C in a mixture of 95% atmospheric air and 5% CO₂ in neurobasal media containing 1% glutamate, 2% B27, and 10% fetal bovine serum.

MA1800-57 cells were purchased from ScienCell (USA) and used as described in our previous study.²¹ 

The mice were randomly divided into the following three groups for in vivo studies: control, global cerebral ischemia/reperfusion (GCI/R), and pyrrolidinedithiocarbamate plus GCI/R. The mice in the control group underwent the same surgical procedure as the other mice but without bilateral common carotid artery ligation, whereas the mice in the GCI/R and pyrrolidinedithiocarbamate + GCI groups were subjected to GCI for 20 min followed by reperfusion for different durations. NF-κB, NDRG2, and proinflammatory cytokine (TNF-α, IL-6, and IL-1β) expression levels and locations were assessed at different time points (0 h, 2 h, 6 h, 12 h, 24 h, 48 h, 72 h, and 7 days) after reperfusion by Western blotting, enzyme-linked immunosorbent assay (ELISA), and immunofluorescence. In addition, neuronal injury and survival were analyzed at 72 h or 7 days after reperfusion by hematoxylin and eosin (HE), terminal deoxynucleotidyl transferase-mediated 2′-deoxyuridine 5′-triphosphate-biotin nick end labeling (TUNEL), and neuronal nuclei (NeuN) staining. Furthermore, hippocampal-dependent cognitive impairment was measured 7 days after reperfusion using the fear conditioning test.

Pure primary astrocyte cultures and cocultures were assigned to the control, oxygen-glucose deprivation and reoxygenation plus vehicle (OGD/R + vehicle) or OGD/R + pyrrolidinedithiocarbamate group. NF-κB, NDRG2, and proinflammatory cytokine (TNF-α, IL-6, and IL-1β) expression levels and locations were analyzed at different time points (0 h, 2 h, 6 h, 12 h, and 24 h) after OGD/R by Western blotting and ELISA. In addition, neuronal apoptosis was assessed 24 h after reoxygenation via cleaved caspase-3 expression and flow-cytometric analyses.

The astrocyte cell lines were transfected with the lentivirus and divided into five groups: cells without lentiviral transfection (control group), cells transfected with a control shRNA or shRNA-NDRG2 (shRNA-control group and shRNA-NDRG2 group, respectively), and cells transfected with a control lentivirus (LEN)-beta-galactosidase (LEN-LacZ) or LEN-NDRG2 lentivirus (LV-Con group or LV-NDRG2 group, respectively). In addition, the levels of proinflammatory cytokines (TNF-α, IL-6, and IL-1β) in the culture medium were detected using ELISA 24 h after OGD/R.

GCI was induced in the C57BL/6 mice as described in our previous publication.²²  A laser Doppler flow meter (PeriFlux System 5000, Perimed, Sweden) was used to monitor cortical cerebral blood flow. Only the mice whose mean cortical cerebral blood flow decreased to 10% of the preischemic value and recovered to 80% of the baseline after reperfusion were used for data analysis (fig. 2, Supplemental Digital Content, https://links.lww.com/ALN/B598). The mice whose mean cortical cerebral blood flow did not decrease to 10% (n = 11) and those that died during ischemia (n = 9) were excluded. Homeostatic variables, including temperature, mean arterial blood pressure, and blood gas and glucose levels, were monitored in each animal before, during, and after ischemia (table 1, Supplemental Digital Content, https://links.lww.com/ALN/B598).

OGD/R was performed as described in our previous publication.²²  In brief, the medium was replaced by glucose-free Dulbecco’s modified Eagle’s medium, and the cells were placed in an anaerobic chamber, which was flushed for 15 min with 95% N₂ and 5% CO₂ and then sealed at 37°C for 4 h for astrocytes and 2 h for cocultures. After the 4-h or 2-h challenge, the cells were removed from the anaerobic chamber, and the medium was replaced with normal culture medium. Then, the cells were maintained in a cell incubator for different lengths of time to induce reoxygenation damage.

Pyrrolidinedithiocarbamate treatment was performed as described in previous studies.^23,24  For in vivo studies, pyrrolidinedithiocarbamate was dissolved in saline, and 200 mg/kg was administered by intraperitoneal injection 20 min before GCI. For in vitro studies, pyrrolidinedithiocarbamate was dissolved in dimethylsulfoxide and further diluted to a concentration of 50 μM, 100 μM, or 200 μM using Dulbecco’s modifed Eagle’s medium. Pyrrolidinedithiocarbamate was added 1 h before the onset of OGD exposure, and the final dimethylsulfoxide concentration was 0.003% (v/v).

The fear conditioning test, which is mainly a learning test, was conducted over the course of 2 days using training and context protocols. Each mouse was individually placed in a control chamber to acclimate for 880 s on the seventh day after reperfusion. Then, on the eighth day, for training, they were subjected to 5 unsignaled footshocks (1 s in duration, 0.8 mA) that were randomly given during the 880-s period and occurred during the last second of a 20-s sound stimulus. On the ninth day, hippocampal-dependent contextual fear memory was tested without auditory electrical stimuli for 880 s by exposing the animals to the same context; freezing time, defined as a lack of movement except for respiration, was determined by visually inspecting a video.

Proteins from the cultured cells and the hippocampal CA1 region were harvested at different time points and homogenized in RIPA lysis buffer (Beyotime, China) with 1 mM phenylmethylsulfonyl fluoride. Then, the collected proteins were separated using sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to Hybond ECL nitrocellulose membranes (Amersham Biosciences, United Kingdom). The following antibodies were used for immunoblotting: rabbit anti-NDRG2 (1:1,000), rabbit anti–NF-κB (1:500), rabbit anticleaved caspase-3 (1:1,000), rabbit anti–pro–caspase-3 (1:1,000) and mouse anti–β-actin (1:1,000). The membranes were then incubated with an horseradish peroxidase–conjugated secondary antibody for 2 h. Protein bands were visualized using a LI-COR Odyssey System (LI-COR Biotechnology, USA).

Twenty-four hours after the pure cultures were exposed to OGD, nuclear extracts were prepared in accordance with the protocol of the NE-PER Nuclear and Cytoplasmic Extraction Reagents Kit (Pierce, USA). Briefly, 200 μl of ice-cold cytoplasmic extraction reagent I was added to the washed astrocytes and incubated on ice for 10 min; then, 11 μl of ice-cold cytoplasmic extraction reagent II was added and incubated on ice for 1 min. The tube was then centrifuged in a microcentrifuge for 5 min at maximum speed (16,000 × g). The supernatant fraction (cytoplasmic extract) was immediately transferred to a clean prechilled tube, and 100 μl of ice-cold nuclear extraction reagent was added to the insoluble fraction by vortexing for 15 s every 10 min for a total of 40 min. The tube was then centrifuged for 10 min in a microcentrifuge at maximum speed. The nuclear extract fraction was then moved to a clean prechilled tube. All extracts were analyzed by Western blotting.

The tissue fluid of the hippocampal CA1 region and the cell culture medium were collected at different time points, and ELISA reagent was used to quantitatively determine the concentrations of proinflammatory cytokines (TNF-α, IL-6, and IL-1β). Absorbance was measured at 450 nm using a microplate reader (Infinite M2000, TECAN, Switzerland).

Paraffin sections were deparaffinized in xylene and rehydrated. The sections were incubated with proteinase K (10 μg/ml) for 15 min at 37°C for antigen retrieval, and nonspecific antibody binding sites were blocked with 1% bovine serum albumin–phosphate buffer saline. The sections were incubated overnight at 4°C with rabbit anti-NDRG2 (1:100), mouse anti-NDRG2 (1:100), mouse anti-GFAP (1:500), rabbit anti–NF-κB (1:100), rabbit anti–TNF-α (1:100) or mouse anti-NeuN (MAB377, 1:100, Millipore, USA) antibodies in 1% bovine serum albumin–phosphate buffer saline. The sections were then washed in tris-buffered saline with Tween and incubated with an antimouse fluorescein isothiocyanate (FITC)–tagged secondary antibody (1:200) and an antirabbit CY3-tagged secondary antibody (1:200) for 2 h at room temperature. The nuclei were subsequently stained with 4′,6-diamidino-2-phenylindole (1 ng/μl), and the sections were mounted with 50% glycerol and examined under a fluorescence microscope.

The mice that were subjected to 20 min of GCI were euthanatized 3 days after reperfusion, and their brains were fixed via perfusion with a buffered 4% paraformaldehyde solution, paraffin-embedded, and sectioned at a thickness of 4 μm for HE and TUNEL staining. HE staining was performed as previously described²⁵  for morphologic observation. TUNEL staining was performed using an In Situ Cell Death Detection kit (Roche Diagnostics, Germany) according to the manufacturer’s instructions. The sections were treated with 0.3% (v/v) H₂O₂ for 20 min and then incubated in a TUNEL reaction mixture for 1 h at 37°C followed by converter-peroxidase for 30 min at 37°C. After washing 3 times in PBS, the sections were developed with 3,3′-diaminobenzidine for 5 min at room temperature. The number of TUNEL-positive neurons in the CA1 region was counted in 3 different fields for each section in a blinded manner using a light microscope at 400× magnification.

After OGD/R, neuronal apoptosis was assayed by flow cytometry. Briefly, the cells were washed with 1× annexin V–FITC binding buffer before staining with annexin V–FITC and propidium iodide (PI) for 15 min at room temperature in the dark. The stained cells were immediately analyzed by flow cytometry. Apoptotic and necrotic cells were quantitated based on annexin V binding and PI uptake. The annexin V–FITC⁺/PI⁻ and annexin V–FITC⁺/PI⁺ cell populations were considered to represent early and late apoptotic cells, respectively. In addition, the apoptosis index was calculated as the sum of the early and late apoptotic cell populations.

NDRG2 overexpression and interference in MA1800 cells were achieved via lentiviral transfection and were performed as described in our previous publication.²¹ 

The determination of the sample size was mainly based on our previous experience.^22,26  All values are expressed as the mean ± SD. All analyses were performed using the SPSS software (version 14.0, USA). Differences between two groups were analyzed using Student’s t tests. Differences among various groups were detected by two-way analysis of variance followed by the Tukey post hoc test. Variables that were measured repeatedly were compared using two-way repeated-measures analysis of variance. Two-tailed P values less than 0.05 were considered statistically significant.

First, we verified that NF-κB and NDRG2 colocalized in hippocampal astrocytes both in vivo and in vitro. Double immunofluorescence staining results revealed that NF-κB and NDRG2 primarily colocalized with the astrocyte marker GFAP in the hippocampal CA1 region in vivo (fig. 1A), demonstrating that NF-κB and NDRG2 were primarily expressed in astrocytes in the hippocampus. Additionally, we found that NF-κB colocalized with NDRG2 in the hippocampal CA1 region under normal conditions (fig. 1B, control group), and their expression significantly increased after GCI/R (fig. 1B). We also verified that NF-κB and NDRG2 colocalized in primary cultured hippocampal astrocytes in vitro (fig. 3, Supplemental Digital Content, https://links.lww.com/ALN/B598).

Next, we studied the effects of GCI/R on the expression of NF-κB and NDRG2 and on the inflammatory response. Compared to the control group, in the mice that experienced ischemia, the protein expression levels of both NF-κB and NDRG2 were significantly increased immediately after reperfusion (0 h) and were maintained at high levels up to 72 h after reperfusion (fig. 1C). In addition, the increase in NDRG2 expression was consistent with but slightly less than the increase in NF-κB expression. Proinflammatory cytokine (TNF-α, IL-6, and IL-1β) levels in the hippocampal CA1 region after GCI/R were studied using ELISA. As shown in figure 1D(a), compared to those in the control group, the expression levels of TNF-α and IL-6 markedly increased from 0 h to 72 h after reperfusion, but the levels of IL-1β expression were not significantly different between the control group and the GCI/R groups. Importantly, using double immunofluorescence staining, we found that TNF-α colocalized with NDRG2 in the hippocampal CA1 region after GCI/R (fig. 1D[b]).

Then, pure cultured hippocampal astrocytes were subjected to an OGD/R treatment to mimic in vivo GCI/R. Western blotting analysis revealed that the NF-κB and NDRG2 protein expression levels in astrocytes were markedly increased from 0 h to 24 h after OGD/R (fig. 2A). We also used ELISA to detect the levels of proinflammatory cytokines (TNF-α, IL-6, and IL-1β) that were secreted by astrocytes into the culture medium at different time points after OGD/R. The results showed that the secretion of TNF-α and IL-6 increased immediately at 0 h and 2 h and reached a maximum 24 h after OGD/R (fig. 2B). However, the secretion of IL-1β was only significantly increased 24 h after OGD/R injury (fig. 2B; *P = 0.0495 vs. the control group).

We used pyrrolidinedithiocarbamate, a specific inhibitor of NF-κB activity, to explore the role of NDRG2 in NF-κB–mediated inflammation after ischemia. As shown in figure 3A, NDRG2 expression was markedly increased 1, 2, and 3 days after GCI/R, whereas treatment with pyrrolidinedithiocarbamate 20 min before GCI surgery dramatically reduced the increase in NDRG2 expression. In addition, pyrrolidinedithiocarbamate treatment significantly reduced the increases in the proinflammatory cytokines TNF-α and IL-6 in the hippocampal CA1 region after GCI/R injury (fig. 3, B and C).

In vitro experiments showed that NF-κB expression significantly increased both in the cytoplasm and in the nucleus after OGD/R injury and demonstrated the nuclear translocation of NF-κB after OGD/R injury (fig. 4A). However, pyrrolidinedithiocarbamate treatment reduced the increase in NF-κB expression in the nucleus without influencing its expression in the cytoplasm (fig. 4A). In addition, we analyzed the effects of different doses of pyrrolidinedithiocarbamate (50 μM, 100 μM, and 200 μM) on NDRG2 expression after OGD/R (fig. 4, Supplemental Digital Content, https://links.lww.com/ALN/B598) and found that treatment with 100 μM pyrrolidinedithiocarbamate significantly reduced the increase in NDRG2 expression induced by OGD/R (fig. 4B). Moreover, pyrrolidinedithiocarbamate treatment dramatically reduced the increase in the expression of the proinflammatory cytokines TNF-α and IL-6 after OGD/R injury (fig. 4, C and D).

In in vivo experiments, GCI/R-injured neurons in the hippocampal CA1 region were detected using HE, TUNEL, and immunofluorescence staining. As revealed by HE staining 3 days after GCI surgery (fig. 5A), the neurons remained intact in the control group, and abundant nuclei centered in the cytoplasm were clearly stained. In contrast, neurons were severely damaged and lacked normal morphology after GCI/R, but pyrrolidinedithiocarbamate treatment reduced the neuronal damage. TUNEL staining 3 days after GCI surgery revealed that the percentage of TUNEL-positive cells in the CA1 region of the hippocampus was 65 ± 4% in the GCI/R group and that treatment with pyrrolidinedithiocarbamate significantly reduced the percentage of TUNEL-positive cells to 46 ± 4% (fig. 5B). We also evaluated the survival of neurons via NeuN staining 7 days after GCI/R. As shown in figure 5B, the percentage of NeuN-positive cells in the hippocampal CA1 region was 44 ± 4% in the GCI/R group, and treatment with pyrrolidinedithiocarbamate significantly increased the percentage of NeuN-positive cells to 69 ± 4% (**P = 0.0042 vs. the control group; ##P = 0.0058 vs. the GCI/R group).

Seven days after GCI surgery, we performed a fear conditioning test, which is a hippocampal-dependent cognitive test, to evaluate hippocampal neuronal injury. The movement track of the mice (fig. 5A, Supplemental Digital Content, https://links.lww.com/ALN/B598) and freezing time before the footshocks (fig. 5B, Supplemental Digital Content, https://links.lww.com/ALN/B598) showed that there were no differences among the control, control + pyrrolidinedithiocarbamate, GCI/R, and pyrrolidinedithiocarbamate + GCI/R groups. As shown in figure 5C, the freezing times of the mice in the control group and the control + pyrrolidinedithiocarbamate group were 39 ± 3% and 38 ± 7%, respectively. GCI/R significantly reduced the freezing time to 23 ± 2% that of the control, and pretreatment with pyrrolidinedithiocarbamate significantly increased the freezing time to 31 ± 2% that of the control.

In in vitro experiments, OGD/R was induced in astrocyte and neuron cocultures, and flow cytometry and Western blotting for cleaved caspase-3 were used to explore the effects of pyrrolidinedithiocarbamate on neuronal apoptosis. As shown in figure 6A, 24 h after reoxygenation, the apoptotic index of the neurons in the OGD group was 45.6 ± 0.2%. In addition, pretreating astrocytes with pyrrolidinedithiocarbamate (100 μM) dramatically attenuated OGD/R-induced neuronal apoptosis to 22.0 ± 4.0% (fig. 6A). As shown in figure 6B, 24 h after reoxygenation, cleaved caspase-3 expression in the neurons had significantly increased in the OGD group, and pretreating astrocytes with pyrrolidinedithiocarbamate markedly decreased cleaved caspase-3 expression in neurons.

To detect the direct relationship between NDRG2 and proinflammatory cytokine secretion after OGD/R, we used lentiviruses to transfect MA1800 astrocytes. First, we successfully established MA1800 cell lines in which NDRG2 expression was stably up- or down-regulated, as verified by Western blotting (fig. 7A). Then, we detected proinflammatory cytokine secretion from the MA1800 cells after OGD and 24 h of reoxygenation. As shown in figure 7B, in the shRNA-control group, the levels of TNF-α, IL-6, and IL-1β were 2,204 ± 472, 491 ± 76 and 222 ± 42 pg/mg protein, respectively. In addition, the downregulation of NDRG2 expression by shRNA-NDRG2 decreased the levels of TNF-α, IL-6, and IL-1β to 1,282 ± 322 (*P = 0.0491), 344 ± 50 (*P = 0.0495), and 144 ± 21 pg/mg protein (*P = 0.0448), respectively; these levels were significantly lower than those in the shRNA-control group. In the LV-control group, the levels of TNF-α, IL-6, and IL-1β were 2,471 ± 532, 521 ± 66, and 247 ± 50 pg/mg protein, respectively. NDRG2 overexpression via LV-NDRG2 markedly increased the levels of TNF-α, IL-6, and IL-1β to 3,598 ± 419 (#P = 0.0449), 712 ± 89 (#P = 0.0401), and 262 ± 48 pg/mg protein (#P = 0.7327), respectively; these levels were higher than those in the LV-control group.

Neurons have been the focus of numerous therapeutic strategies for cerebral ischemia, all of which have failed to yield favorable patient outcomes. However, in the last decade, a growing body of evidence has indicated that astrocytes may be a viable target for cerebral ischemia therapy during both the acute phase of ischemia and the recovery period.²⁷  Under transient focal cerebral ischemia, astrocytes can distribute functional mitochondria to adjacent neurons, which promotes neuronal viability and recovery.²⁸  In addition, the upregulation of GLT-1 expression in CA1 astrocytes—but not DG or CA3 astrocytes—significantly reduced CA1 neuronal death after transient forebrain ischemia.¹  Furthermore, astrocytes promote axon regeneration by releasing multiple axon growth–supporting molecules after severe spinal cord injury.²⁹  However, within minutes after ischemic injury, injured neurons and glial cells in the core and ischemic penumbra produce proinflammatory mediators, cytokines, and reactive oxygen species, which activate astrocytes.³⁰  Activated astrocytes can produce high levels of proinflammatory cytokines, including IL-6, TNF-α, and IL-1β,³⁰  which hinder ischemic recovery by directly inducing neuronal apoptosis³¹  and/or increasing toxic nitric oxide levels¹  and inhibiting neurogenesis.³²  In addition, inhibiting astrocyte-induced inflammation could alleviate ischemic injury,¹⁵  which may provide a new therapeutic strategy for cerebral ischemia.

As mentioned, NF-κB is a key regulator of inflammation in the CNS, and astrocytes are the primary cell type under NF-κB regulation.^12–14  NF-κB is a transcription factor that is normally bound by its inhibitory protein, IκBα, and is sequestered in the cytoplasm. Ischemia triggers the degradation of IκBα by the ubiquitin proteasome system, resulting in the translocation of NF-κB to the nucleus, where it regulates gene transcription.³³  In this study, we found that after GCI/R or OGD/R, astrocytic NF-κB expression was markedly increased, and proinflammatory cytokine expression also increased, which indicates that astrocytic NF-κB is involved in ischemia-induced inflammatory injury. However, the question of whether inhibiting astrocytic NF-κB activity can alleviate ischemic injury remains controversial. In a previous study, transgenic mice were generated that expressed the IκBα super-repressor under the transcriptional control of the GFAP promoter; this study reported that the selective inhibition of astrocytic NF-κB did not reduce infarct size or cell death 48 h after permanent middle cerebral artery occlusion.³⁴  However, a recent study using GAP-IκBα-dn transgenic mice found that the selective inhibition of astroglial NF-κB significantly reduced the retinal ischemia–induced loss of retinal neurons.¹⁵  Pyrrolidinedithiocarbamate is an inhibitor of NF-κB that stops the degradation of IκBα and the translocation of NF-κB into the nucleus,³⁵  and pyrrolidinedithiocarbamate treatment can reduce infarct volumes and improve neurologic behavior that has been impaired by focal cerebral ischemic injury.³⁶  However, the effects of pyrrolidinedithiocarbamate on GCI have rarely been reported. GCI selectively destroys the hippocampal CA1 region and causes severe hippocampal-dependent cognitive impairment,³⁷  and the current study revealed that pyrrolidinedithiocarbamate significantly reduced the effects of GCI/R, including the increased levels of proinflammatory cytokines (TNF-α and IL-6) and neuronal injury, as well as hippocampal-dependent cognitive impairment. More importantly, in vitro experiments showed that pyrrolidinedithiocarbamate treatment markedly reduced the translocation of astrocytic NF-κB into the nucleus and dramatically decreased the levels of proinflammatory cytokines and neuronal apoptosis induced by OGD/R, thus demonstrating the neuroprotective effects of the inhibition of astrocytic NF-κB signaling. However, the detailed molecular mechanisms have not been well defined.

NDRG2 is specifically expressed in astrocytes in the brain^17,18  and is involved in astrocyte activation and proliferation.³⁸  Additionally, NDRG2 is associated with the pathogenesis of neurodegenerative disorders, including Alzheimer disease,³⁹  depression,⁴⁰  and frontotemporal lobar degeneration.⁴¹  We recently reported the spatial-temporal expression of NDRG2 in rat and mouse brains after middle cerebral artery occlusion injury;^20,21  our report indicated that astrocytic NDRG2 played an important role in cerebral ischemia. However, the specific role of NDRG2 in GCI injury and the underlying mechanisms have not been reported. In this study, we demonstrated that astrocytic NDRG2 expression markedly increased after GCI surgery or OGD injury, which was in accordance with the findings of our previous studies.^20,21  Interestingly, we found that the increase in NDRG2 expression was consistent with the increase in NF-κB and proinflammatory cytokine expression both in vivo and in vitro and that NDRG2 clearly colocalized with TNF-α in the hippocampal CA1 region, which indicated that NDRG2 may be involved in NF-κB–mediated inflammation after cerebral ischemia. In addition, our previous study revealed that NF-κB increased the expression of NDRG2 by binding to its promoter region in Leydig cells.¹⁶  In this study, we found that pyrrolidinedithiocarbamate treatment significantly reduced the increase in NDRG2 expression induced by GCI/R or OGD/R. Based on these results, we hypothesize that ischemic stimuli activate NF-κB signaling, inducing its translocation from the cytoplasm to the nucleus, where it binds to the ndrg2 gene promoter region to promote NDRG2 expression.

Although pyrrolidinedithiocarbamate reduced the expression of NDRG2 and proinflammatory cytokines, this finding is not definitive proof that there is direct regulation between NDRG2 and proinflammatory cytokine expression. Using lentiviruses, we found that silencing NDRG2 expression markedly reduced the secretion of proinflammatory cytokines in the medium after OGD/R, whereas the overexpression of NDRG2 increased the secretion of TNF-α and IL-6, which indicated that NDRG2 could directly regulate proinflammatory cytokine expression after cerebral ischemia. A recent study on a mouse model of cortical stab injury reported that deleting NDRG2 attenuated reactive astrogliosis and inflammatory responses through the regulation of IL-6/STAT3 signaling.⁴²  In addition, our previous study revealed that NDRG2 translocated to the nucleus after cerebral ischemia, which suggests that NDRG2 may be a novel transcription factor.²¹  Therefore, we hypothesized that these proinflammatory cytokines may be NDRG2 target genes; this hypothesis warrants further exploration.

In addition, we previously found that electroacupuncture or sevoflurane pretreatment alleviated ischemic injury by inhibiting the increase in NDRG2 expression and NDRG2 nuclear translocation.^26,43  Therefore, we proposed that astrocytic NDRG2 may be a novel molecular target in cerebral ischemia therapy;²¹  however, the additional roles of astrocytic NDRG2 in cerebral ischemia remain elusive. In this study, we found that NDRG2 was involved in NF-κB–mediated inflammation induced by cerebral ischemia, which provides a novel direction for the study of NDRG2 in cerebral ischemia.

There are some limitations to the current study. A substantial and growing body of evidence suggests that inflammatory microglia also contribute to the pathophysiology of neuronal death after GCI/R,^44,45  but we were not able to determine whether the proinflammatory cytokines were secreted by astrocytes or microglia in our in vivo experiments. However, we analyzed the colocalization of astrocytic NDRG2 and proinflammatory cytokines after GCI/R, and in vitro experiments further demonstrated the direct relationship between NDRG2 and proinflammatory cytokine expression in astrocytes. Therefore, our results suggest that astrocytic NDRG2 is involved in the inflammatory responses induced by cerebral ischemia, and its underlying molecular mechanisms will be the focus of our future studies.

In conclusion, the current study provides the first evidence that astrocytic NDRG2 is significantly up-regulated after GCI/R and is involved in NF-κB–mediated inflammation. Furthermore, we found that the NF-κB activation inhibitor pyrrolidinedithiocarbamate could alleviate ischemia-induced neuronal injury by inhibiting the increase in NDRG2 expression and NDRG2-mediated inflammation. These results reveal a new mechanism for this disorder and provide evidence that may be used to develop novel stroke therapies that target astrocytic NDRG2.

The authors thank American Journal Experts for assisting in the preparation of this manuscript.

This study was supported by the Natural Science Foundation of China, Beijing, China (81771411; 81571279; 81371446), the Defense-related Science and Technology Foundation of China, China (3607026), and the Key Project of the Shaanxi Major Basic Research and Development Program, Xi’an, Shaanxi, China (2017ZDXM-SF-059).

The authors declare no competing interests.