Glycine Exhibits Neuroprotective Effects in Ischemic Stroke in Rats through the Inhibition of M1 Microglial Polarization via the NF-κB p65/Hif-1α Signaling Pathway

Ischemic stroke is a disease that results in high disability and high mortality worldwide. Because of its increasing prevalence, it has attracted widespread attention. Ischemic stroke is caused by arterial embolism, microangiopathy, or macroangiopathy, resulting in oxygen and glucose deprivation in the brain and leading to brain damage and neurologic deficits (1). Oxidative stress, neuroinflammation, and excitotoxicity are thought to be the main causes of the neuronal death and brain damage caused by ischemia (2). Neuroprotection has been a related problem due to the complex environment after ischemic stroke (3). Although we have extensively described the pathogenesis of ischemic stroke and physiological changes after onset, treatment options are limited. The immune system is closely involved in the various stages of brain damage and tissue repair processes induced by ischemic stroke (4, 5). Immune system–mediated neuroinflammation is heavily involved in determining the fate of the brain after ischemic stroke (6).

Microglia are innate immune cells of the CNS and are the major mediators of neuroinflammation (7, 8). Microglia are called resident macrophages in the brain, and their functions and characteristics are very similar to those of macrophages (9, 10). Microglia can be activated to be polarized after being stimulated, exerting proinflammatory or anti-inflammatory functions by secreting different substances (11–13). Hif-1α is a major player in inflammation and controls the expression of genes related to promoting inflammation (14, 15). Hif-1α–regulated inflammation can be mediated through the nuclear translocation of PKM2 (16). However, NF-κB regulates the expression of Hif-1α, and the activation of NF-κB p65 is positively correlated with the expression of Hif-1α (17). AKT is upstream of p65, and the level of p65 can be negatively regulated by AKT activation (18). PTEN is a very important regulator of the PI3K/AKT signaling pathway, which negatively regulates AKT activation (19). Previous results have shown that the inhibition of PTEN not only results in the neuroprotection of neurons but also inhibits inflammation (20–22).

Glycine is a simple nonessential amino acid that is an important component of many proteins. Glycine is also crucial for the synthesis of many biomolecules, such as porphyrin, creatine, and purine nucleotides (23). In the CNS, glycine is a major inhibitory neurotransmitter that binds to glycine receptors to inhibit postsynaptic neurons, and it is also a coagonist of excitatory NMDA receptors (a calcium-permeable ion channel) (24, 25). Previous studies have demonstrated that glycine has neuroprotective effects in many disease models, including ischemic stroke injury, hypoxia, anoxia, and intracerebral hemorrhage (26, 27). Clinical trials show that glycine treatment can improve the prognosis of patients with ischemic stroke. Although the neuroprotective effect on glycine has been studied, it remains largely unknown.

In this study, we demonstrated that glycine exhibited neuroprotective effects on neurons and microglia after ischemic stroke injury. Glycine plays a direct role in protecting neurons; furthermore, its beneficial effects are also indirectly mediated by microglia. In terms of mechanism, glycine inhibits NF-κB p65 and Hif-1α by activating AKT and downregulating PTEN, which suppresses ischemia-induced M1 microglia polarization and inhibiting inflammation, resulting in the inhibition of neuronal death followed by ischemic stroke.

Adult male Sprague–Dawley rats were housed with three rats per cage on a 12-h light/dark cycle in a temperature-controlled room (23–25°C) with free access to water and food. Animals were allowed at least 3 d for acclimation before experimentation. We used a total of 214 male rats in our in vivo experiments, and 46 male rats were discarded during the operational procedures. Fifteen adult pregnant female rats and 65 embryos were used in our cortical neuronal cultures experiments. All animal use and experimental protocols were approved and carried out in compliance with the Institutional Animal Care and Use Committee guidelines and the Animal Care and Ethics Committee of Wuhan University School of Medicine. Assignment to the experimental groups and the collection and processing of samples were performed randomly. The experiments were performed by investigators blinded to the assigned group of each animal.

Transient focal cerebral ischemia was induced using the suture occlusion technique (28). The entire process is the same as our laboratory previously described (29). Male Sprague–Dawley rats weighing 250–300 g were anesthetized with 4% isoflurane in 70% N₂O and 30% O₂ by using a mask. The rectal temperature was maintained at 37.0 ± 0.5°C using a homoeothermic blanket. A midline incision was made in the neck, the right external carotid artery (ECA) was carefully exposed and dissected, and a 3–0 monofilament nylon suture was inserted from the ECA into the right internal carotid artery to occlude the origin of the right middle cerebral artery (∼22 mm). After 90 min of occlusion, the suture was removed to allow reperfusion, the ECA was ligated, and the wound was closed. Sham-operated rats underwent identical surgery and/or intracerebroventricular (i.c.v.) injections except that the suture was inserted and withdrawn immediately. At 24 h after middle cerebral artery occlusion (MCAO), rats (n = 24) were reperfused with ice-cold 0.9% saline after anesthetization with 4% isoflurane in 70% N₂O and 30% O₂, and brains were quickly removed for Western blot analysis, RT-PCR analysis, behavioral testing, and 2,3,5-triphenyltetrazolium chloride (TTC) staining.

The brain was placed in a cooled matrix, and 2-mm coronal sections were cut. Individual sections were placed in 10-cm petri dishes and incubated for 30 min in a solution of 2% TTC in PBS at 37°C. The slices were fixed in 4% paraformaldehyde at 4°C for 24 h. All image collection, processing, and analysis were performed in a blinded manner and under controlled environmental lighting. The scanned images were analyzed using image analysis software (Image-Pro Plus version 6.0). The infarct volume was calculated to correct for edema. The normal volume of the contralateral hemisphere and the normal volume of the ipsilateral hemisphere were measured, and the infarct percentage was calculated as the percentage of contralateral structure to avoid mismeasurement secondary to edema (30).

Rats were anesthetized with a mixture of 4% isoflurane in 30% O₂ and 70% N₂O in a sealed perspective box. When the rats were deeply anesthetized, we used ear bars and an upper incisor bar to secure the rat’s head in a stereotaxic frame, and the rats were continuously under anesthesia with 4% isoflurane using a mask. Next, a small sagittal incision was made, and the bregma was located as the anatomical reference point. The cerebral ventricle (from the bregma: lateral, 1.5 mm; anteroposterior, −0.8 mm; depth, 3.5 mm) was performed using a 23-gauge needle attached via polyethylene tubing to a Hamilton microsyringe and drug infusion at a rate of 1.0 μl/min. Proper needle placement was verified by withdrawing a few microliters of clear cerebrospinal fluid into the Hamilton microsyringe.

Cortical neuron cultures were prepared from female Sprague–Dawley rats at 17 d of gestation (31). The pregnant rats were killed by cervical dislocation after anesthetization with 4% isoflurane in 70% N₂O and 30% O₂. The embryos were sprayed with 70% ethanol that was removed after the rats. The embryos were quickly decapitated, and the cortices were placed in ice-cold plating medium (neurobasal medium, 0.5% FBS, 2% B-27 supplement, 25 mM glutamic acid, and 0.5 mM L-GlutaMAX) following the removal of the meninges. The cortical neurons were plated on petri dishes coated with poly-d-lysine and suspended in plating medium. Half of the plating medium was removed and replaced with maintenance medium (neurobasal medium, 0.5 mM l-glutamine, and 2% B-27 supplement) in the same manner every 3 d. The cultured neurons were used in the experiments after 12 d.

Cortical microglia cultures were prepared from female Sprague–Dawley rats at 17 d of gestation. Pregnant rats died of cervical dislocation after anesthesia with 4% isoflurane with 70% N₂O and 30% O₂. Rats were sprayed with 70% ethanol and removed from the uterus of pregnant rats. The brain tissue of embryos was quickly separated. Pelleted cells were resuspended in warmed DMEM culture medium completed with 10% heat-inactivated FBS, 1% antibiotic-antimycotic, and 5 ng/ml carrier-free recombinant mouse GM-CS. A total of 1.3 × 10⁶ cells were seeded into tissue culture–grade poly l-lysine–coated T75 cell culture–treated flasks and placed in a 37°C incubator with relative humidity. The culture supernatant was replaced twice weekly with 10 ml of fresh completed medium until confluency of cells was observed at approximately 3 wk.

For the oxygen-glucose deprivation (OGD) challenge, cells were transferred to a glucose-free extracellular solution (116 mM NaCl, 0.8 mM MgSO₄, 5.4 mM KCl, 1.0 mM NaH₂PO₄, 26 mM NaHCO₃, and 1.8 mM CaCl₂) and deoxygenated environment, placed in a humidified chamber (Plas-Labs, Lansing, MI), and maintained at 37°C in 5% H₂/85% CO₂/10% N₂ for 60 min. Next, the cells were replaced with fresh maintenance medium containing an appropriate concentration of reagents for 24 h during the recovery period in a 5% CO₂/95% O₂ incubator. The control cultures were first transferred to another extracellular solution (5.4 mM KCl, 116 mM NaCl, 0.8 mM MgSO₄, 1.8 mM CaCl₂, 1.0 mM NaH₂PO₄, 26 mM NaHCO₃, and 33 mM glucose) and placed in the humidified chamber, which was maintained at 37°C in 95% O₂/5% CO₂ for 60 min (32). Finally, fresh maintenance medium was replaced for the whole period at 37°C in a 95% O₂/5% CO₂ incubator.

Western blotting was performed as previously described (33). Briefly, a polyvinylidene difluoride membrane from Millipore was used to incubate the samples with the primary Abs against NF-κB p65 (rabbit, 1:1000), Hif-1α (rabbit, 1:1000), AKT (mouse, 1:1000), phospho-AKT (Ser⁴⁷³) (rabbit, 1:2000), PTEN (rabbit, 1:1000), and actin (rabbit, 1:2000) from Cell Signaling Technology. The primary Abs were labeled with secondary Abs, and protein bands were imaged using SuperSignal West Femto Maximum Sensitivity Substrate (Pierce, Rockford, IL). The EC3 Imaging System (Uplant; UVP) was used to obtain blot images directly from the polyvinylidene difluoride membrane. The Western blot data were quantified using Image-Pro Plus version 6.0.

Rats were treated with an overdose of isoflurane and then intracardiac perfusion with 0.9% saline, placed in 4% paraformaldehyde at 4°C for 24 h, and transferred into 30% sucrose solution in 100 mol/ml phosphate buffer at 4°C for 72 h. Then, the brain tissue was kept in 4% paraformaldehyde solution at 4°C overnight. Brain tissues were cut into 16-μm coronal sections by a Leica VT1000 S Vibratome (Leica Microsystems AG, Nussloch, Germany). Fluoro Jade–C (FJC) labeling was performed by using the standard protocol (34). Brain sections were first immersed in 1% sodium hydroxide in 80% ethanol for 5 min, then rinsed in 70% ethanol for 2 min, then 2 min in distilled water, and then incubated in 0.06% potassium permanganate solution for 10 min. Following a 2-min distilled water rinse, the brain sections were transferred into 0.0001% solution of FJC (Sigma-Aldrich) 10 min, which was dissolved in 0.1% acetic acid. Brain sections were rinsed with water three times for 1 min, then air dried on a warmer at 50°C for at least 5 min, and finally immersed in xylene for at least 1 min. The brain sections were mounted with DPX Mountant media (Sigma-Aldrich). The brain sections were photographed by a blinded investigator using an Olympus fluorescence microscope (IX51; Olympus, Japan). A series of photomicrographs were taken from three regions of the ipsilateral cerebral cortex with a ×20 objective, and FJC-positive cells were counted by ImageJ software (ImageJ). The data were expressed as cells per mm².

The immunostained brain sections were prepared as FJC staining, and the immunofluorescence staining steps were based on the description of the prior execution (35). The brain sections were treated with the primary Ab mouse anti-NeuN (neuronal-specific nuclear protein) from Chemicon. The secondary Ab was goat anti-mouse 488 from Molecular Probes (Eugene, OR). The DAPI probe was from Life Technologies. The sections were photographed by a blinded investigator using an Olympus fluorescence microscope (IX51; Olympus), and images were analyzed by ImageJ software (ImageJ).

Double-stranded oligonucleotides corresponding to the target sequences were cloned into the lenti-CRISPR-V2 vector and cotransfected packaging plasmids into HEK293 cells. Two days after transfection, the viruses were harvested, ultrafiltrated (0.22-mm filter; Millipore), and used to infect BV-2 cells in the presence of polybrene (8 mg/ml). The infected cells were selected with puromycin (1 mg/ml) for at least 5 d.

BV-2 cells and U251 cells were purchased from the Chinese Academy of Sciences Cell Bank. U251 cells were sown in a six-well plate (8 × 10⁵ cells per well) in DMEM supplemented with 10% heat-inactivated FBS, penicillin G (100 U/ml), streptomycin (100 mg/ml), and l-glutamine (2.0 mM) and incubated at 37°C in a humidified atmosphere containing 5% CO₂ and 95% air.

Short hairpin RNA (shRNA) targeting Hif-1α was inserted into the pGIPZ vector as previously reported, and the pGIPZ control was generated with the control oligonucleotide 5′-GCTTCTAACACCGGAGGTCTT-3′. PCR-amplified mouse Hif-1α was cloned into the pcDNA3.1/hypro (+) vector between BamH I and Not I. pCDNA3.1(+)-Hif-1α–WT (WT Hif-1α) was generated using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA).

When the confluence of BV-2 cells reached 60–70% on the treatment day, the cells were transfected with Hif-1α shRNA or mouse Hif-1α plasmid WT Hif-1α for 8 h. The medium was then replaced with normal growth medium for 24 h. On the following day, the cells were then collected for Western blot analysis.

Total RNA was isolated from control brains and ischemic brains using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions; the first strand of cDNA was synthesized using 5-μg Superscript First-Strand Synthesis System for RT-PCR (Invitrogen). PCR was performed on the Opticon 2 Real-Time PCR Detection System (Bio-Rad) using the corresponding primers (Table I) and SYBR Green PCR Master Mix (Invitrogen). The cycle time value was normalized to the GAPDH level of the same sample. The mRNA expression level was then reported as the fold change compared with the control group.

Lactate dehydrogenase (LDH) release was analyzed using a colorimetric CytoTox 96 Cytotoxicity kit (Promega). Cell viability in the neuronal cultures was evaluated by the ability to take up thiazolyl blue tetrazolium bromide (mingest thiazolyl tetrazolium) (PowerWave X; BioTek, Winooski, VT). The two methods were performed following the manufacturer’s instructions.

The rats were subjected to a modified neurologic severity score test as reported previously. These tests are a battery of reflex, sensory, motor, and balance tests, which are similar to the contralateral neglect tests in humans. Neurological function was graded on a scale of 0 to 18 (normal score, 0; maximal deficit score, 18) (36).

The beam-walk test measures the complex neuromotor function of animals. The animal was timed as it walked across a (100 × 2-cm) beam. A box for the animal to feel safe was placed at one end of the beam. A loud noise was created to stimulate the animal to walk toward and into the box. Scoring was based on the time it took the rat to go into the box. The higher the score was, the more severe the neurologic deficit (37).

A modified sticky-tape test was performed to evaluate forelimb function. A sleeve was created using a 3.0 × 1.0-cm piece of yellow paper tape and was subsequently wrapped around the forepaw so that the tape attached to itself and allowed the digits to protrude slightly from the sleeve. The typical response is for the rat to vigorously attempt to remove the sleeve by either pulling at the tape with its mouth or brushing the tape with its contralateral paw. The rat was placed in its cage and observed for 30 s. Two timers were started; the first ran without interruption and the second was turned on only while the animal attempted to remove the tape sleeve. The ratio of the left (affected)/right (unaffected) forelimb performance was recorded. The contralateral and ipsilateral limbs were tested separately. The test was repeated three times per test day, and the best two scores of the day were averaged. The lower the ratio, the more severe the neurologic deficit (38).

The experimental design and analysis of this study resulted in data and statistical analysis that comply with recommendations. All data are expressed as the mean ± SE. Student t test and variance analysis were used for differences among groups. p < 0.05 was considered statistically significant.

We generated an experimental ischemic stroke rat model induced by MCAO for 1.5 h followed by various periods of reperfusion (39) and showed the structure of coronal brain slices after MCAO injury in rats (Fig. 1A, Supplemental Fig. 1). TTC staining was performed to detect the infarct volume of the rat brain 24 h after ischemia-reperfusion (I/R), and the results showed that the ischemia-induced cerebral infarction accounts for 21% of the total brain volume (Fig. 1B, 1C). Additionally, the i.c.v. injection of glycine (100 μg/100 g) immediately after MCAO (40) significantly improved ischemic stroke–induced cerebral infarction volume (Fig. 1B, 1C). In the pathologic condition of ischemic stroke, the cortex of the peri-infarct area is mainly impacted. Therefore, we performed FJC staining to evaluate the degenerate neurons in the peri-infarct area 24 h after I/R. The results showed that the number of FJC-positive degenerating neurons was downregulated by treatment with glycine (Fig. 1D). In addition, we performed a sequence of neurobehavioral tests in rats and showed that glycine promoted functional recovery after ischemic stroke (Fig. 1E–G). Together, glycine treatment reduces brain damage and promotes functional recovery after cerebral ischemia injury.

Neurons are structural and functional units of the CNS, whereas glial cells mainly play a supporting role; however, both cell types are very important (41). We cultured primary neurons and microglia from the rat cortex and challenged the neurons and microglia with OGD for 1.5 h to simulate ischemic stroke injury in vitro (42). The cell viability and LDH release results showed that treatment with glycine (100 μM) 1 h before OGD insult decreased OGD-induced neuronal death (Fig. 2A, 2B). Additionally, a similar result was observed in microglia (Fig. 2C, 2D). Because the function of glial cells is to support neurons in the CNS, we tested the indirect effects of microglial glycine treatment on neurons. Glycine treatment was administered to microglia 1 h before the OGD challenge. Then, we recovered the microglia culture medium (MCM) 22.5 h after OGD insult, mixed it with DMEM at a 1:1 ratio, and used this mixed medium to culture cortical neurons after OGD insult (Fig. 2E). We found that OGD-induced neuronal death was upregulated after treatment with MCM without glycine; however, glycine-treated MCM reduced OGD-induced neuronal death compared with the non-MCM group (Fig. 2F, 2G). In summary, treatment with glycine not only reduced OGD-induced neuronal death but also OGD-induced microglial death, and the treatment of microglia with glycine also has an indirect neuroprotective effect on neurons.

Microglia are innate immune cells of the CNS and are characterized by resident macrophages in the brain, which are major mediators of neuroinflammation (43). Microglia can be activated after ischemic stroke injury. Activated microglia have two phenotypes: classically activated (M1 microglia) and alternatively activated (M2 microglia), as described (13, 44). The representative markers of M1 microglia are IL-1β, TNF-α, and CD32. Arg-1, CD206, and YM-1 are markers of M2 microglia. To further investigate the mechanism by which glycine regulates microglia and indirectly exerts neuroprotection of neurons, RT-PCR analysis of mRNA levels of M1 and M2 microglial markers in rats 24 h after I/R was performed. The results showed that ischemic stroke injury promoted M1 microglial polarization, whereas glycine treatment inhibited ischemia-induced inflammation and promoted M2 microglial polarization (Fig. 3A, 3B, Table I). Previous studies have demonstrated that NF-κB and Hif-1α play a crucial role in inflammation (15, 45). We showed that p65 and Hif-1α were increased after cerebral ischemia, which supports the function of p65 and Hif-1α in promoting inflammation (Fig. 3C, 3D); however, glycine treatment inhibited these processes (Fig. 3C, 3D). Similar results were observed in primary cultured microglia in vitro (Fig. 3E–G). Together, glycine-induced anti-inflammatory effects may be mediated by NF-κB and Hif-1α.

To further investigate whether glycine-regulated microglial polarization is mediated by p65 and Hif-1α, we first determined the relationship between p65 and Hif-1α. Western blotting analysis of Hif-1α in rats 1 h after treatment with the NF-κB p65 inhibitor helenalin (200 μM, 2 μl) by i.c.v injection showed that Hif-1α was downregulated after the inhibition of p65 (Supplemental Fig. 2A). Similar results were observed in vitro (Supplemental Fig. 2B). Meanwhile, we treated cultured cortical microglia with helenalin (2 μM) 1 h after OGD insult, and the results showed that the level of Hif-1α was downregulated by the inhibition of p65 (Supplemental Fig. 2C). In addition, the transfection of p65 shRNA or the expression of WT p65 in BV-2 cells showed that the level of Hif-1α was downregulated or upregulated, respectively (Fig. 4A, 4B). To further confirm the relationship between glycine, NF-κB p65, and Hif-1α, we established a p65 knockout BV-2 cell line and showed that the regulation of Hif-1α by glycine after the knockout of p65 was no longer observed (Fig. 4C). These results suggest that glycine-regulated Hif-1α is mediated by p65. Furthermore, the transfection of Hif-1α shRNA or expression of WT Hif-1α in BV-2 cells did not result in a significant difference in p65 levels (Fig. 4D, 4E). NF-κB p65 is upstream of Hif-1α. Moreover, we pretreated rats with a p65 inhibitor 1 h before MCAO. Western blotting analysis of p65 and Hif-1α showed that glycine-regulated p65 and Hif-1α expression disappeared after I/R (Fig. 4F). RT-PCR analysis of M1 and M2 microglial markers in rats after I/R 24 h revealed that glycine-mediated microglial polarization was significantly inhibited (Figs. 4G, 4H, 3A, 3B). Similar results were also observed in cultured microglia in vitro (Supplemental Fig. 2D, 2E). Together, these results indicate that NF-κB p65/Hif-1α mediates glycine-regulated microglial polarization after ischemic stroke injury.

Previous studies have demonstrated that AKT activation inhibits M1 macrophage polarization and that AKT is upstream of p65 (18). In this study, we confirmed that glycine-regulated microglial polarization after ischemic stroke is dependent on p65 and Hif-1α. We hypothesized that AKT activation mediates glycine-regulated microglial polarization. Western blotting analysis of S473 p-AKT showed that the level of p-AKT was downregulated after I/R in vivo and in vitro. Meanwhile, treatment with glycine increased the level of p-AKT (Fig. 5A, Supplemental Fig. 3A). Next, Western blotting analysis of p65 and Hif-1α in rats 1 h after treatment with the AKT inhibitor IV (IV) (100 μM, 2 μl) by i.c.v injection showed that p65 and Hif-1α were upregulated after the inhibition of AKT (Fig. 5B, 5C), and similar results were observed in vitro (Supplemental Fig. 3B, 3C). Furthermore, Western blotting analysis of p65 and Hif-1α after pretreatment with IV 1 h before I/R showed that treatment with glycine did not result in significant changes in the levels of p65 and Hif-1α after the inhibition of AKT in vivo and in vitro (Fig. 5D, Supplemental Fig. 3D). In addition, RT-PCR analysis of M1 and M2 microglial markers revealed that the inhibition of AKT significantly suppressed glycine-induced M2 microglial polarization (Fig. 5E, 5F, Supplemental Fig. 3E, 3F). Together, glycine regulates the AKT activation that reduces the levels of NF-κB p65 and Hif-1α, subsequently promoting anti-inflammatory activity following ischemic stroke in vitro.

PTEN is a tumor suppressor, and the inhibition of PTEN activates AKT through the PI3K signaling pathway (46, 47). The inhibition of PTEN reduces ischemia-induced neuronal death (48). Western blotting analysis of PTEN in rats 1 h after glycine treatment demonstrated that PTEN was downregulated by glycine (Fig. 6A). To confirm whether glycine-induced AKT activation is mediated by PTEN, we used a PTEN-deficient U251 cell line (Fig. 6B). Western blotting analysis of p-AKT in U251 cells 1 h after treatment with glycine showed that the level of p-AKT did not significantly change (Fig. 6C). These results indicate that glycine-mediated AKT activation is dependent on PTEN. Together, glycine inhibits PTEN regulation of the AKT/NF-κB/Hif-1α signaling pathway and subsequently suppresses M1 microglia polarization.

In Fig. 2F, 2G, we demonstrated that glycine-treated microglia can indirectly reduce OGD-induced neuronal death. We also confirmed that glycine-mediated microglial polarization is dependent on the PTEN/AKT/NF-κB/Hif-1α signaling pathway. In this study, we pretreated cultured cortical microglia with the NF-κB inhibitor helenalin (2 μM) or IV (1 μM) 1 h before glycine treatment and then used the MCM to culture neurons (Fig. 7A). Cell viability and LDH release showed that the inhibition of AKT or NF-κB inhibited the indirect neuroprotection of neurons by glycine (Fig. 7B, 7C). Meanwhile, we pretreated cultured cortical neurons with the NF-κB inhibitor helenalin or IV and demonstrated that the OGD-induced neuronal death was directly decreased or increased, respectively (Fig. 7D, 7E). Furthermore, TTC staining, FJC staining, and neurologic function scores showed that the inhibition of AKT or NF-κB blocked the glycine-induced decrease in infarct volume and reduction in FJC-positive degenerative neurons and improved functional recovery (Fig. 8A–F).

Ischemic stroke is a global disease that has drawn increasing attention because of its high disability and mortality rates; however, the therapeutic options for it are limited. Glycine confers protection against neuronal death under in vitro and in vivo experimental conditions (49, 50). Clinical trials show that glycine treatment can improve the prognosis of patients with ischemic stroke (51). However, the mechanism of the neuroprotective role of glycine remains unclear. In this study, we demonstrate that glycine not only inhibits neuronal death directly after I/R but indirectly through microglia. Glycine inhibits ischemia-induced M1 microglial polarization and promotes anti-inflammatory effects by inhibiting PTEN and activating AKT and then inhibiting NF-κB p65 and Hif-1α, resulting in the indirect inhibition of ischemia-induced neuronal death.

Glycine is a simple nonessential amino acid that is an important part of many proteins and is also a major inhibitory neurotransmitter that binds to glycine receptors to inhibit postsynaptic neurons (52). Treatment with glycine reduced the ischemic stroke–induced brain damage and neuronal death. We found that glycine can directly reduce OGD-induced neuronal death. However, glycine can inhibit OGD-induced microglial death and indirectly reduce neuronal death. Previous studies have demonstrated that the activation of GluN2A-containing NMDA receptors, but not GluN2B-containing NMDA receptors, contributes to glycine-induced neuroprotection in ischemic stroke injury (40). These studies show that the neuroprotection of neurons by glycine can also be achieved by microglia. Microglia are innate immune cells of the CNS and major mediators of neuroinflammation (7). Microglia are characterized as the resident macrophages in the brain and can be activated during pathological conditions (10). Activated microglia have two phenotypes: M1 microglia and M2 microglia (11). M1 microglia are proinflammatory and secrete oxidative metabolites and cytokines such as IL-1β, TNF-α, CD32, and iNOS, which promote brain damage under pathological conditions (53). In contrast, M2 microglia express anti-inflammatory mediators, such as Arg-1, CD206, and YM-1, which prevent inflammation and contribute to recovery after brain injury (54). Microglia are activated to polarize into the M1 phenotype after 24 h of ischemic stroke. Treatment with glycine resulted in the suppression of ischemia-mediated M1 microglia polarization and promotion of M2 microglia polarization. M2 microglia polarization promotes anti-inflammatory effects that indirectly reduce OGD-induced neuronal death.

Previous studies have shown that NF-κB and Hif-1α play a crucial role in inflammation (15, 45). The activation of the NF-κB pathway leads to the expression of various proinflammation-associated genes, including cytokines, chemokines, and adhesion molecules (55). In macrophages, the upregulation of Hif-1α promotes M1 macrophage polarization and inflammation (56). Several studies have shown that controlling the HIF-1α gene by NF-κB provides important additional and parallel regulatory levels for the HIF-1α pathway. The levels of NF-κB p65 and Hif-1α increased after ischemic stroke, and treatment with glycine suppressed the upregulation of NF-κB p65 and the subsequent inhibition of Hif-1α. We confirmed that NF-κB p65/Hif-1α mediates glycine-regulated microglial polarization. AKT is upstream of NF-κB. The activation of AKT can negatively regulate NF-κB p65. We further indicated that glycine-regulated NF-κB p65 and Hif-1α are mediated by AKT activation. Moreover, PTEN mediates glycine-regulated AKT/NF-κB/Hif-1α signaling pathways. PTEN is downregulated by glycine in microglia. PTEN is a tumor suppressor, and the inhibition of PTEN activates AKT through the PI3K signaling pathway (46, 47). The downregulation of PTEN has a neuroprotective effect on ischemic stroke injury. We demonstrated that the inhibition of PTEN lipid phosphatase activity activates AKT and that the inhibition of PTEN protein phosphatase activity inhibits synaptic GluN2B-expressing NMDA receptors, which leads to protection against ischemic neuronal death (22). We have also shown that the inhibition of PTEN by enhancing GABA_A receptor expression and function prevents neuronal death in OGD, in vitro OGD, and ischemic stroke models in vivo (57). These studies show that glycine modulates microglial polarization through regulation of the PTEN/AKT/NF-κB/Hif-1α signaling pathway and subsequently indirectly reduces ischemia-induced neuronal death, brain damage, and functional recovery.

The results of the current study reveal that neuroinflammation after ischemic stroke can be mediated by glycine. In the pathologic condition of ischemic stroke, glycine inhibits PTEN, which activates AKT and subsequently inhibits NF-κB p65 and Hif-1α, thereby inhibiting ischemia-induced M1 microglia polarization and promoting anti-inflammatory effects, which indirectly result in reduced ischemia-induced neuronal death. These findings provide a new understanding of the function of glycine in ischemic stroke and provide a basis for the extending the use of glycine in clinical trials.

We thank American Journal Experts for linguistic assistance.

The authors have no financial conflicts of interest.