Glycine triggers a non-ionotropic activity of GluN2A-containing NMDA receptors to confer neuroprotection

Ionotropic activation of NMDA receptors (NMDARs) requires agonist glutamate and co-agonist glycine. Here we show that glycine enhances the activation of cell survival-promoting kinase Akt in cultured cortical neurons in which both the channel activity of NMDARs and the glycine receptors are pre-inhibited. The effect of glycine is reduced by shRNA-mediated knockdown of GluN2A subunit-containing NMDARs (GluN2ARs), suggesting that a non-ionotropic activity of GluN2ARs mediates glycine-induced Akt activation. In support of this finding, glycine enhances Akt activation in HEK293 cells over-expressing GluN2ARs. The effect of glycine on Akt activation is sensitive to the antagonist of glycine-GluN1 binding site. As a functional consequence, glycine protects against excitotoxicity-induced neuronal death through the non-ionotropic activity of GluN2ARs and the neuroprotective effect is attenuated by Akt inhibition. Thus, this study reveals an unexpected role of glycine in eliciting a non-ionotropic activity of GluN2ARs to confer neuroprotection via Akt activation.


Glycine increases Akt phosphorylation independent of Ca 2+ influx through NMDAR channels.
To test the effect of glycine on the activation of cell survival-promoting kinase Akt (protein kinase B) in cultured mouse cortical neurons where no Ca 2+ pass through the NMDAR channels, we treated the neurons with glycine (100 μ M) for 30 min in a modified extracellular solution (5.0 mM EGTA, 137 mM NaCl, 5.4 mM KCl, 1.0 mM MgCl 2 , 25 mM HEPES, 33 mM Glucose, titrated to pH 7.4 with osmolarity of 300-320 mOsm) in which Ca 2+ was not included but with the addition of 5.0 mM EGTA to chelate the residual Ca 2+ . The activation of Akt was quantified by measuring Akt phosphorylation (p-Akt) on Ser473 in western blot assay 26,27 . The levels of p-Akt were quantified by calculating the ratio of p-Akt to total Akt (t-Akt). Our results showed that treatment of glycine (100 μ M) for 30 min increased Akt phosphorylation in the cortical neurons in which there were no Ca 2+ influx into the NMDAR channels (Fig. 1A).
To exclude the possibility that residual Ca 2+ in the extracellular solution might pass through NMDAR channels, we treated the neurons with the extracellular solution described above but with the addition of non-competitive NMDAR antagonist MK-801 28,29 . We named this specific Ca 2+ -free extracellular solution as ECS-1 (10 μ M MK-801, 5.0 mM EGTA, 137 mM NaCl, 5.4 mM KCl, 1.0 mM MgCl 2 , 25 mM HEPES, 33 mM Glucose, titrated to pH 7.4 with osmolarity of 300-320 mOsm). Since MK-801 is a use-dependent pharmacological agent, we first treated the neurons with NMDA (1.0 μ M) and glycine (1.0 μ M) for 1.0 min that opened the NMDAR channels and allowed MK-801 in the ECS-1 to fully block NMDARs [28][29][30] . The cultured neurons were then washed with ECS-1 for three times (10 min wash/each). As this procedure eliminates the possibility of Ca 2+ in activation of NMDAR channels, in this study we referred to this as NMDAR channel inactivation procedure (Fig. 1B). As shown in Fig. 1C, glycine (100 μ M) treatment increased Akt phosphorylation in the cortical neurons where the NMDAR channel activity was inhibited by the NMDAR channel inactivation procedure. In the same experimental conditions, the effect of glycine on Akt phosphorylation was found to be dose-dependent (Fig. 1D). These data indicate the possibility that glycine-induced Akt activation is independent of the channel activity of NMDARs.
To provide further evidence that the effect of glycine on Akt phosphorylation was independent of extracellular Ca 2+ , we tested the effect of BAPTA, a Ca 2+ chelator that has faster calcium-binding kinetics than EGTA 31 . The experimental condition was same as that in Fig. 1C. But BAPTA (0.1, 1.0 or 5.0 mM) was included in the ECS-1 in the + BAP groups (Fig. 1E). BAPTA was included both during the NMDAR channel inactivation procedure and during treatment with glycine. Compared with the group without BAPTA treatment, BAPTA treatment did not interfere with glycine-induced elevation of Akt phosphorylation (Fig. 1E), suggesting that Ca 2+ -mediated channel activity of NMDARs does not contribute to the observed effect of glycine in cortical neurons. Elevation of Akt phosphorylation by glycine does not depend on the activation of glycine receptors. Glycine is the agonist for strychnine-sensitive glycine receptors. Glycine receptors are not significantly expressed in the mature but expressed in the developing cortex 32,33 . To exclude the possible effect of glycine receptors on the observed Akt activation by glycine, we used the same experimental design as that in Fig. 1C. But strychnine was added into the ECS-1 for all the treatment in both -Gly and + Gly group ( Fig. 2A). Our data showed that strychnine (10 μ M) failed to block the enhancement of Akt phosphorylation by glycine (100 μ M) in cortical neurons subjected to NMDAR channel inactivation procedure ( Fig. 2A). Thus, glycine-induced enhancement of Akt phosphorylation does not depend on the activation of strychnine-sensitive glycine receptors.
Glycine does not affect p38-MAPK activation in cortical neurons where NMDAR channel activities and glycine receptors are inhibited. The p38-MAPK is implicated in NMDAR-dependent LTD 34 , and was shown to be activated by non-ionotropic NMDAR signaling after chemical LTD induction 20,24 . The activation of p38-MAPK is also involved in excitotoxicity 35 . To determine whether glycine also altered p38-MAPK signaling independent of glycine receptors and the activation of NMDAR channels, we tested the effect of glycine on p38-MAPK phosphorylation in cortical neurons following the experimental procedure described in Fig. 2A. Our results showed that glycine (100 μ M) had no significant effect on p38-MAPK phosphorylation in our experimental conditions (Fig. 2B), suggesting that a specific activation of Akt but not p38-MAPK by glycine occurs in the condition in which NMDAR channel activities and glycine receptors were suppressed.
Glycine enhances Akt activation through a non-ionotropic activation of GluN2ARs. Our results thus far suggest a non-ionotropic activity of NMDAR to mediate the potentiation of Akt activation by glycine. To provide direct evidence for this possibility, we measured the effects of glycine on Akt phosphorylation in HEK293 cells transiently over-expressing NMDARs. The cDNAs of GluN1, GluN2A and/or GluN2B subunits were transfected in various combinations into the HEK293 cells 36 . The Ca 2+ -mediated channel activities of NMDARs expressed in the transfected cells were inhibited by the NMDAR channel inactivation procedure. Treatment of glycine (100 μ M) for 30 min had no effect on Akt phosphorylation in both non-transfected HEK293 cells and the cells transfected with cDNAs of green fluorescence protein (GFP) (Fig. 3A). However, glycine increased Akt phosphorylation in HEK293 cells transfected with cDNAs of GluN1 + GluN2A following the NMDAR channel inactivation procedure (Fig. 3B), but not in cells transfected with cDNAs of GluN1 + GluN2B (Fig. 3C). We also found that glycine did not increase Akt phosphorylation in HEK293 cells transfected with cDNAs of GluN1, GluN2A and GluN2B, respectively (Fig. 3D). Together, these results indicate that a non-ionotropic activity of GluN2ARs mediates the elevation of Akt phosphorylation by glycine.
To validate the role of a non-ionotropic activity of GluN2AR in mediating the enhancement of Akt activation by glycine in cortical neurons, we applied a GluN2A knockdown approach. The GluN2A protein expression was suppressed in the cultured cortical neurons transducted with GluN2A shRNA lentiviral particles (Fig. 4A). The same experimental design as that in Fig. 1C was applied to inhibit NMDARs. As shown in Fig. 4B, glycine (100 μ M) increased Akt phosphorylation in neurons transducted with lentiviral shRNA control, but the effect of glycine was significantly reduced in neurons transducted with lentiviral GluN2A shRNA. As another control of GluN2A shRNA against GluN2A, the GluN2B shRNA had no influence on the observed effect of glycine (Fig. 4C,D). These results lead us to conclude that glycine promotes Akt activation through a non-ionotropic activity of GluN2ARs in cortical neurons.
The glycine-GluN1 binding site mediates the non-ionotropic activation of GluN2ARs. To determine how glycine exerts its effect through the non-ionotropic activation of GluN2ARs, we tested the effect of glycine-GluN1 binding site antagonist L-689560 on glycine-induced Akt activation after Ca 2+ -mediated channel activities of NMDARs were inhibited [40][41][42][43] . The L-689560 was included in the ECS-1 in the wash step of the NMDAR channel inactivation procedure and in the step of glycine treatment (Fig. 1B). The cultures were then treated with ECS-1 containing glycine (100 μ M) and L-689560 for 30 min. We showed that after the channel activities of NMDARs were inhibited, L-689560 (50 μ M) blocked glycine-induced Akt phosphorylation in the cultured neurons and in the HEK293 cells transfected with GluN1 + GluN2A (Fig. 5A,B). These data suggest that the glycine-GluN1 binding is required for the non-ionotropic activation of GluN2ARs.
As a control study, we also tested the effect GluN2B antagonist Ro 25-6981 (5.0 μ M) 42,43 . Our data showed that Ro 25-6981 had no significant effect on glycine-induced Akt activation in the neurons and the HEK293 cells transfected with GluN1 + GluN2A + GluN2B after Ca 2+ -mediated NMDAR channel activity was inhibited (Fig. 5C,D). D-serine is the endogenous agonist of glycine-GluN1 binding site 44 . As a further support for the role of glycine-GluN1 binding in mediating the effect of non-ionotropic GluN2ARs, we tested the role of D-serine in Akt phosphorylation in the cortical neurons and the HEK293 cells transfected with GluN1 + GluN2A or GluN1 + GluN2B following NMDAR channel inactivation procedure. D-serine increased Akt phosphorylation in both cortical neurons and HEK293 cells transfected with cDNAs of GluN1 + GluN2A but not with those of GluN1 + GluN2B (Fig. 5E-G).
Glycine prevents glutamate neurotoxicity-induced neuronal death through non-ionotropic activation of GluN2ARs. As Akt is a survival-promoting kinase that plays a crucial role in preventing neuronal death 26 47 . The levels of lactate dehydrogenase (LDH) released from injured neurons was also measured to quantify the neuronal damage 27 . Our data showed that after glutamate neurotoxicity insult, glycine (100 μ M) treatment protected against the death of cortical neurons in which Ca 2+ -mediated NMDAR channel activity was inhibited (Fig. 6B-D).
Our results further demonstrated that the neuroprotective effect of glycine (100 μ M) was reduced in injured cortical neurons in which the GluN2A expression was suppressed by lentiviral GluN2A shRNA (Fig. 6E,F). The neurons in both shRNA control and GluN2A shRNA groups were subjected to the same experimental procedures described in Fig. 6B-D. We conclude that the neuroprotective effect of glycine is at least in part mediated through non-ionotropic activation of GluN2ARs in glutamate neurotoxicity-induced neuronal injury.
To determine the roles of Akt activation and glycine-GluN1 binding in glycine-induced neuroprotection, we tested the effect of Akt inhibitor IV and glycine-GluN1 binding antagonist L-689560 in our experimental model. The experimental condition was the same as that described in Fig. 6A-D. The Akt inhibitor IV and L-689560 were included in both wash and treatment steps. We found that both Akt inhibitor IV (1.0 μ M) and L-689560 (50 μ M) significantly reduced glycine-induced neuroprotective effect in neurons where Ca 2+ -mediated channel activity of NMDARs were inhibited (Fig. 6G,H). Thus, Akt activation and glycine-GluN1 binding mediate glycine-induced neuroprotection. Together, these results provide functional evidence for the role of non-ionotropic activity of GluN2ARs in mediating the neuroprotective effect of glycine.

Discussion
Using a Ca 2+ -free ECS-based procedure to inactivate the channel activity of NMDARs in cultured cortical neurons and HEK293 cells expressing GluN2ARs, we tested the effect of glycine on Akt phosphorylation, a cellular process playing important role in neuronal survival. We provided the first evidence that glycine induced a potentiation of Akt phosphorylation independent of the channel activity of NMDARs. We confirmed that glycine-induced non-ionotropic activation of GluN2ARs, but not GluN2BRs, mediated the enhancement of Akt activation. Thus, our study identified a non-ionotropic function of GluN2ARs. To ensure no Ca 2+ -mediated channel activities of NMDARs contributing to glycine-induced effect in our study, we established a NMDAR channel inactivation procedure in which Ca 2+ was not included in the ECS but with the addition of Ca 2+ chelator EGTA and the use-dependent open channel blocker MK-801. To allow MK-801 to fully block the channels  of NMDARs, we pretreated the cells with NMDA and glycine to open the NMDAR channels. Thus, Ca 2+ influx through NMDARs would not likely occur in our experimental conditions.
Increasing evidence supports the non-ionotropic function of NMDARs [19][20][21][22][23][24][25] . It has been recently shown that a non-ionotropic activation of NMDAR was insensitive to the glycine-GluN1 site antagonist 20,21,24 . However, our study shows that the glycine-GluN1 binding is required to activate the non-ionotropic activity of GluN2ARs. These findings suggest that glycine triggers a non-ionotropic activity of GluN2ARs through the glycine-GluN1 binding site. By testing the effects of glycine in HEK293 transfected with different combinations of NMDARs, we were able to obtain direct evidence to reveal that a non-ionotropic activation of GluN2ARs but not GluN2BRs mediates the enhancement of Akt activation.
The novel non-ionotropic NMDAR may have functional significance. NMDAR-mediated neurotoxicity induces neuronal death and neurodegeneration in various CNS disorders including ischemic stroke, traumatic brain injury and neurodegenerative diseases [48][49][50] . However, the use of NMDAR antagonists as neuroprotective agents was disappointing in clinical trials [51][52][53] . A simple possibility is that these antagonists, while suppressing NMDAR-mediated neurotoxicity, block the biological and/or neural survival-promoting effects of NMDARs 1,16-18 . Thus, identification of molecular mechanisms by which specific NMDAR subtype selectively exerts its effect on neuronal survival or death would provide a critical basis for the development of potent therapy for CNS injuries and neurodegenerative diseases.
GluN2ARs and GluN2BRs play different role in neuronal survival or death 14,16 . But the underlying molecular mechanism remains unclear. It is recently reported that a non-ionotropic function of NMDARs was required for β -amyloid-induced synaptic depression and synaptic loss [21][22][23] , providing new evidence for the involvement of GluN2BRs in neurotoxicity. Our observation for a non-ionotropic activation of GluN2ARs by glycine explains in part why GluN2AR plays a different role than GluN2BR in neuronal survival.
The molecular mechanism underlying glycine-induced non-ionotropic NMDAR activation is unclear. It is possible that glycine binds to GluN1 and induces conformational changes in GluN2A. If this is the case, why it cannot change the conformation of GluN2B? The different structure between GluN1-GluN2A coupling and GluN1-GLuN2B coupling may be the possible reason. Future study is required to reveal the mechanisms.
How non-ionotropic activity of GluN2ARs enhances Akt activation is unclear. It is likely that the C-terminal domain of GluN2A may mediate the effect of glycine on Akt activation. Akt deactivation is believed to be a causal mediator of cell death 26,45,46 . Enhancement of Akt activity exerts pro-survival effect in neuronal injury and neurodegenerative diseases 26,45,46 . In this study, we identify Akt as a downstream neuroprotective signal of glycine that activates non-ionotropic GluN2ARs. We provide evidence that non-ionotropic activation of GluN2ARs by glycine reduces glutamate neurotoxicity-induced Akt deactivation and thus prevents cortical neuronal death. Akt is known to influence neuronal survival through activation or inhibition of substrates 26,45,46 . For example, activated Akt promotes survival through phosphorylation of transcription factors forkhead/FOXO, NF-κ B and mdm2 or through phosphorylation of Bcl-2 family members Bad and Bim 26,45,46 . Further study is needed to determine which Akt-dependent signal pathway mediates the activation of non-ionotropic GluN2ARs by glycine.
It would be important to determine where is glycine coming from considering the brain does not normally have glycinergic transmissions. Glial cells would be the source of glycine release, especially during the ischemia injury process. In this study, we show that D-serine has similar effect to glycine, suggesting that D-serine is also the endogenous agonist for this non-ionotropic NMDA receptor activity.

Experimental Procedures
General methods. Randomization was used to assign samples to the experimental groups, and to collect and process data. All animal experiments were approved and carried out in compliance with the IACUC guidelines of University of Nevada and Wuhan University School of Medicine. All experimental protocols were approved by the Animal Care and Ethics Committee of University of Nevada and Wuhan University School of Medicine.
Neuronal culture. The cortical neuronal cultures were prepared from female C57BL/6 mice at gestation day 17 as described 27,54 . Briefly, dissociated neurons were suspended in plating medium (Neurobasal medium, 2% B-27 supplement, 0.5% FBS, 0.5 μ M L-glutamine, and 25 μ M glutamic acid) and plated on poly-D-lysine coated Petri dishes. After 1 day in culture, half of the plating medium was removed and replaced with maintenance medium (Neurobasal medium, 2% B-27 supplement, and 0.5 μ M L-glutamine). Thereafter, maintenance medium was changed in the same manner every 3 days. The cultured neurons were used for experiments at 12 days after plating.
Scientific RepoRts | 6:34459 | DOI: 10.1038/srep34459 transfected in cultured HEK293 cells 36 . Transfections were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) as described in our previous studies 15,27 . DNA-Lipofectamine complexes were made in serum-free medium Opti-MEM. To prevent NMDAR-induced cell death, the transfected HEK293 cells were treated with 1.0 mM DAPV to prevent NMDAR-induced excitotoxicity 55 . The transduction of GluN2A and GluN2B shRNA lentiviral particles were performed in cultured cortical neurons based on the manufacturer's instructions.
Western blotting. Western blotting assay was performed as described previously 15,27 . For the detection of phospho-Akt, the samples prepared in the same day were used. The polyvinylidene difluoride membrane (Millipore, Bedford, MA, USA) was incubated with primary antibody against phospho-Akt (Ser473) (Cell Signaling Technology, Beverly, MA), Akt (Cell Signaling Technology, Beverly, MA), phospho-p38-MAPK (Cell Signaling Technology, Beverly, MA), p38-MAPK (Cell Signaling Technology, Beverly, MA), β -actin (Santa Cruz Biotech), GluN2A (Santa Cruz Biotech), or GluN2B (Santa Cruz Biotech). Primary antibodies were labeled with horseradish peroxidase-conjugated secondary antibody, and protein bands were imaged using SuperSignal West Femto Maximum Sensitivity Substrate (Pierce, Rockford, IL, USA). The EC3 Imaging System (UVP, LLC, Upland, CA) was used to obtained blot images directly from the polyvinylidene difluoride membrane. For the detection of total Akt, the same polyvinylidene difluoride membrane was stripped and then re-incubated with primary antibody against total Akt (Cell Signaling Technology). The quantification of Western blot data was performed using ImageJ software.
Neuronal viability assays. Double staining of propidium iodide (PI) and fluorescein diacetate (FDA) was performed to detect neuronal viability using a modified procedure 47 . Briefly, cultures were rinsed with extracellular solution and incubated with FDA (5 μ M) and PI (2 μ M) for 30 min. The cultures were washed with extracellular solution and then viewed on an Olympus fluorescent microscope (IX51, Olympus). Neuronal viability was determined by calculating the number of PI-labeled cells over FDA-labeled cells. The investigator for the cell count was blinded to the experimental treatment.
The lactate dehydrogenase (LDH) is a cytoplasmic enzyme retained by viable cells with intact plasma membranes and released from cells with damaged membranes. The LDH release was measured using CytoTox 96 Cytotoxicity kit based on the manufacturer's instructions (Promega, Madison, WI) 27 . The levels of maximal LDH release were measured by treating the cultures with 10× lysis solution (provided by the manufacturer) to yield complete lysis of the cells. Absorbance data were obtained using a 96-well plate reader (Molecular Devices, Palo Alto, CA) at 490 nm. According to the manufacturer's instructions, the LDH release (%) was calculated by calculating the ratio of experimental LDH release to maximal LDH release.
Statistics. Student's t test or ANOVA test was used where appropriate to examine the statistical significance of the differences between groups of data. Newman-Keuls tests were used for post-hoc comparisons when appropriate. All results are presented as mean ± SE. Significance was placed at p < 0.05.