The Subtype of GluN2 C-terminal Domain Determines the Response to Excitotoxic Insults

Summary It is currently unclear whether the GluN2 subtype influences NMDA receptor (NMDAR) excitotoxicity. We report that the toxicity of NMDAR-mediated Ca2+ influx is differentially controlled by the cytoplasmic C-terminal domains of GluN2B (CTD2B) and GluN2A (CTD2A). Studying the effects of acute expression of GluN2A/2B-based chimeric subunits with reciprocal exchanges of their CTDs revealed that CTD2B enhances NMDAR toxicity, compared to CTD2A. Furthermore, the vulnerability of forebrain neurons in vitro and in vivo to NMDAR-dependent Ca2+ influx is lowered by replacing the CTD of GluN2B with that of GluN2A by targeted exon exchange in a mouse knockin model. Mechanistically, CTD2B exhibits stronger physical/functional coupling to the PSD-95-nNOS pathway, which suppresses protective CREB activation. Dependence of NMDAR excitotoxicity on the GluN2 CTD subtype can be overcome by inducing high levels of NMDAR activity. Thus, the identity (2A versus 2B) of the GluN2 CTD controls the toxicity dose-response to episodes of NMDAR activity.

In answering questions relating to subunit-specific function (including excitotoxicity), it is becoming clear that pharmacological approaches are of limited use, given the tools currently available (Neyton and Paoletti, 2006). Although GluN2B-specific antagonists are highly selective and have demonstrated a role for GluN2B-containing NMDARs in excitotoxicity (Liu et al., 2007), attempts to study the role of GluN2A (Liu et al., 2007) employed a mildly selective GluN2A-preferring antagonist (NVP-AAM007) at a concentration shown by others to antagonize GluN2B-containing NMDARs (Berberich et al., 2005;Frizelle et al., 2006;Martel et al., 2009;Neyton and Paoletti, 2006;Weitlauf et al., 2005), rendering some of the findings hard to interpret. Moreover, the less-controllable conditions in an intact brain render a weakly selective competitive antagonist, such as NVP-AAM007, of limited value for in vivo studies. Another important issue is that receptors can exist in a triheteromeric form that contains both a GluN2A and a GluN2B subunit (Hatton and Paoletti, 2005;Rauner and Kö hr, 2011), where the role of each subunit cannot be established using currently available pharmacological tools.
Additional problems in relating function to GluN2 subunit composition include their different spatiotemporal expression profiles. For example, in younger neurons, GluN2B is predominant and as such may mediate excitotoxicity simply because Neuron 74, 543-556, May 10, 2012 ª2012 Elsevier Inc. 543 Open access under CC BY license.
most NMDARs are GluN2B-containing. Moreover, GluN2B-and GluN2A-containing NMDARs may be enriched at extrasynaptic and synaptic sites, respectively (Groc et al., 2006;Martel et al., 2009;Tovar and Westbrook, 1999, but see Harris and Pettit, 2007;Thomas et al., 2006). Since receptor location may be a determinant of excitotoxicity irrespective of subunit composition (Hardingham and Bading, 2010), a location-dependent effect may be misinterpreted as a subunit-specific effect.
We have eschewed pharmacocentric approaches in favor of molecular genetics to determine whether equivalent levels of Ca 2+ influx through GluN2A-and GluN2B-containing NMDARs differentially affect neuronal viability. We hypothesized that any differences would be due to their large CTDs because this is the primary area of sequence divergence, as well as being the part of GluN2 known to bind intracellular signaling/scaffolding proteins (Ryan et al., 2008). By studying signaling from wildtype and chimeric GluN2A/2B subunits, using both acutely expressed subunits as well as a mouse knockin model, we find that the presence of the CTD 2B in an NMDAR renders Ca 2+ influx through this receptor more toxic than the presence of CTD 2A . This difference is observed in vivo as well as in vitro and is attributable in part to enhanced physical/functional coupling of CTD 2B to the PSD-95/nNOS signaling cassette, which suppresses prosurvival CREB-mediated gene expression, rendering neurons vulnerable to excitotoxic cell death.

RESULTS
The CTDs of GluN2B and GluN2A Differentially Influence Excitotoxicity Independent of the Identity of the Rest of the Subunit We wanted to investigate whether the subtype of GluN2 CTD influences the excitotoxicity of a given amount of NMDAR-mediated ion flux. We created constructs encoding chimeric receptors based on GluN2B and GluN2A but with their respective CTDs replaced (denoted as CTR) with each other's (GluN2-B 2A(CTR) and GluN2A 2B(CTR) , respectively, Figure 1A). In rat hippocampal neurons, we first expressed either wild-type GluN2B WT or GluN2B 2A(CTR) , at a developmental stage where endogenous NMDARs are overwhelmingly GluN2B-containing (Martel et al., 2009). Expression of GluN2B WT or GluN2B 2A(CTR) both enhanced whole-cell currents to a similar level ( Figure 1B) and did not differentially affect the proportion of extrasynaptic NMDARs ( Figure 1C), as assessed by the ''quantal block'' method of irreversibly blocking synaptically located NMDARs (Papadia et al., 2008). Thus, any differential CTD-specific effects on excitotoxicity can be studied without the complicating factor of altered NMDAR location, which itself influences survival/death signaling via mechanisms that are likely to be independent of GluN2 subtype (Hardingham and Bading, 2010;Martel et al., 2009;Papadia et al., 2008).
We next studied whether expression of GluN2B WT or GluN2B 2A(CTR) had different effects on vulnerability to excitotoxicity. NMDA (20 mM) was applied for 1 hr to neurons transfected with vectors encoding either GluN2B WT , GluN2B 2A(CTR) or control vector, and neuronal death was assessed 24 hr later. GluN2B WT strongly increased the level of cell death compared to the control, consistent with NMDAR currents being higher ( Figures  1D and 1E). However, expression of GluN2B 2A(CTR) caused a significantly lower enhancement of cell death than GluN2B WT ( Figures 1D and 1E), despite NMDAR currents being equal (Figure 1B), suggesting that CTD 2B promotes cell death better than CTD 2A . The same result was found when the experiment was repeated in DIV18 neurons (see Figure S1A available online), indicating that the differential effect of CTD 2B versus CTD 2A on cell death also holds true in more mature neurons.
To further investigate the differential CTD subtype effects on excitotoxicity, we compared NMDAR-dependent cell death in neurons expressing GluN2A WT and GluN2A 2B(CTR) . Expression of GluN2A WT and GluN2A 2B(CTR) did not differentially affect the proportion of extrasynaptic NMDARs ( Figure 1C) and caused similar increases in NMDAR currents ( Figure 1F); although, because of the lower affinity of GluN2A for NMDA, the increases were smaller than for the GluN2B-based constructs ( Figure 1B). We found that neurons expressing GluN2A 2B(CTR) were significantly more vulnerable to NMDA-induced excitotoxicity than GluN2A WT -expressing neurons ( Figure 1G). Thus, for a given amount of NMDAR-mediated current, the presence of CTD 2B Figure 1. The GluN2B C-Terminal Domain Promotes NMDAR-Mediated Toxicity When Linked to Either Channel Portion of GluN2B or GluN2A (A) Schematic and linear representation of GluN2A, GluN2B, and the chimeric subunits in which the C-terminal domain (CTD) has been replaced (CTR). Constructs encoding these subunits were expressed in hippocampal neurons. ATD, amino-terminal domain; S1-S2, extracellular ligand-binding domains (LBD); M1-M4, intramembranous domains. (B) Acute expression of GluN2B WT or GluN2B 2A(CTR) has a similar effect on NMDA-induced whole-cell currents. Neurons were transfected with the indicated constructs (plus eGFP marker) and whole-cell steady-state NMDAR-mediated currents evoked by 20 mM NMDA (and normalized to cell capacitance, here and throughout) were compared to control-transfected neurons (b-globin, n = 12-14 cells per construct) * p < 0.05 (t test comparison to control-transfected neurons). Responses, here and throughout, were measured at 48 hr posttransfection. Mean ± SEM shown here and throughout the figure.
(C) Expression of the subunits described in (A) does not alter the overall proportion of extrasynaptic NMDARs (n = 5-10 cells for each construct). Right shows example trace of NMDAR-mediated currents before (whole cell) and after synaptic NMDAR blockade (extrasynaptic). See Supplemental Experimental Procedures for details. (D) GluN2B WT expression renders neurons more vulnerable to an excitotoxic insult (20 mM NMDA for 1 hr), but replacing the CTD to that of GluN2A reduces the level of toxicity (*p < 0.05; n = 7; 150-200 cells analyzed per condition). (E) Example pictures of (D) showing transfected cells with the relevant plasmid (+eGFP) pre-and post-NMDA treatment. White arrows indicate transfected neurons before NMDA treatment. Red/blue arrows in the ''posttreatment'' panels indicate dead/live cells, respectively. (F) Expression of GluN2A WT or GluN2A 2B(CTR) enhances NMDAR currents to similar levels compared to globin-expressing cells (n = 10-11 cells per construct). *p < 0.05 (t test comparison to control-transfected neurons). (G) NMDA-induced toxicity is significantly higher in GluN2A 2B(CTR) -transfected neurons than with GluN2A WT (*p < 0.05; n = 8). See also Figure S1.

Neuron
The GluN2 CTD Subtype Controls Excitotoxicity promotes neuronal death better than CTD 2A , regardless of whether they are linked to the channel portion of GluN2A or GluN2B. This result illustrates the independent influence of the identity of the CTD on excitotoxicity, acting in addition to the influence of the identity of the rest of the channel on downstream signaling events (e.g., because of different channel kinetics and ligand binding properties).
A Mouse Knockin Model Reveals the Influence of the GluN2 CTD Subtype In Vitro and In Vivo We next investigated the importance of the GluN2 CTD subtype by an independent approach: a genetically modified ''knockin'' mouse in which the protein coding portion of the C-terminal exon of GluN2B (encoding over 95% of the CTD) was exchanged for that of GluN2A (GluN2B 2A(CTR) ; Figure 2A; see Supplemental Experimental Procedures). The 3 0 UTR of GluN2B, which also forms part of the C-terminal exon, was unchanged apart from a 61 bp insertion at its beginning (a remnant of the excision of a neomycin resistance selection cassette). We wanted to determine whether equivalent Ca 2+ influx through GluN2B-containing and GluN2B 2A(CTR) -containing NMDARs would result in different levels of neuronal death. We studied DIV10 cultured cortical neurons from GluN2B +/+ and GluN2B 2A(CTR)/2A(CTR) littermates. These cultures exhibited similar levels of basal viability and levels of synaptic connectivity and strength, as measured by mini EPSC frequency/size, spontaneous EPSC frequency, and AMPA receptor currents ( Figures S2A-S2D), as well as unaltered cell capacitance ( Figure S2E).
In neurons of both genotypes, we observed a blockade of around 60% ( Figure 2B), indicative of a high ($80%) level of GluN1/GluN2B heterodimeric receptors. Moreover, the proportion of extrasynaptic NMDARs was found to be the same for GluN2B 2A(CTR)/2A(CTR) and GluN2B +/+ neurons ( Figure 2C). Thus, any differential CTD subtype-specific effects on excitotoxicity could be studied without the potentially confounding factor of altered NMDAR location. We also investigated whether any differences in use-dependent run-down of whole-cell NMDAR currents were observed because this may be relevant to longterm exposure to NMDA. Having measured baseline whole-cell NMDAR currents, ten further 10 s applications of NMDA were applied over a 10 min period. We found no difference in rundown of steady-state NMDAR currents in GluN2B +/+ and GluN2B 2A(CTR)/2A(CTR) neurons (around 3% per application; Figure S2F). We also examined NMDAR single-channel properties. We excised outside-out patches from DIV9 GluN2B +/+ and GluN2B 2A(CTR)/2A(CTR) neurons and measured NMDA-evoked unitary currents, finding no difference in their mean singlechannel conductance of approximately 50 pS, which is typical for GluN2B-containing NMDARs ( Figure S2G).
Despite the aforementioned similarities, we found one important difference; whole-cell NMDAR currents in GluN2B 2A(CTR)/2A(CTR) neurons were around 30% lower than GluN2B +/+ ( Figure 2D). Levels of GluN2B protein were lower in DIV10 GluN2B 2A(CTR)/2A(CTR) cortical neurons ( Figure S2H) and in P7 cortical protein extracts ( Figure S2I; ruling out the possibility of an in vitro artifact). An explanation for this difference was found when we looked at GluN2B 2A(CTR) mRNA levels, which were lower both in DIV10

Neuron
The GluN2 CTD Subtype Controls Excitotoxicity this decrement appeared to be a developmental-stage-dependent effect because by adulthood, levels of forebrain GluN2B mRNA ( Figure 3A) and protein (p = 0.51, n = 5,5) were unaltered in GluN2B +/+ versus GluN2B 2A(CTR)/2A(CTR) mice. We hypothesize that GluN2B 2A(CTR) , compared to wild-type GluN2B, may be transcribed, processed, or exported slightly less efficiently, which manifests itself in a mRNA decrement in development when expression of many genes, including those encoding NMDAR subunits, is changing rapidly.
Satisfied that these doses of NMDA elicit equivalent NMDARdependent currents and Ca 2+ loads, we next studied their effects on neuronal viability. Strikingly, we found that NMDA C1 and NMDA C2 both promoted more death in GluN2B +/+ neurons than in GluN2B 2A(CTR)/2A(CTR) (Figures 2H and 2I). Thus, swapping the GluN2B CTD for that of GluN2A in the mouse genome reduces the toxicity of NMDAR-dependent Ca 2+ influx. This is in agreement with our studies based on the overexpression of GluN2A/2B-based wild-type and chimeric subunits (Figure 1), thus confirming the importance of the CTD subtype by two independent approaches. We also performed a similar set of experiments in DIV18 neurons. Because there remained a difference in whole-cell currents (around 25%), we again generated NMDAR current dose-response curves to allow us to pick pairs of NMDA concentrations (15 and 20 mM; 30 and 40 mM) which would trigger equivalent currents ( Figure S2K). Consistent with our observations at DIV10, we once again saw increased NMDA-induced death in GluN2B +/+ neurons compared to GluN2B 2A(CTR)/2A(CTR) neurons experiencing equivalent levels of NMDAR activity ( Figure S2L).
We next wanted to determine whether maximal levels of neuronal death could be achieved in neuronal populations devoid of CTD 2B if NMDAR activity were high enough. We treated GluN2B 2A(CTR)/2A(CTR) neurons with a high dose (100 mM) of NMDA and found that this triggered near-100% neuronal death, as it also did in GluN2B +/+ neurons ( Figures 2H  and 2I). Thus, the influence of excitotoxicity on the GluN2 CTD subtype is abolished when insults are very strong.
To promote excitotoxic neuronal loss, we stereotaxically injected a small (15 nmol) dose of NMDA into the hippocampus (just below the dorsal region of the CA1 layer) and quantified the resulting lesion volume 24 hr later. Consistent with the position of the injection site, the lesions were centered on the CA1 subregion, an effect potentially enhanced by the known vulnerability of this subregion to excitotoxic insults (Stanika et al., 2010). However the lesion also spread to other hippocampal subregions (CA3, dentate gyrus) as well as a small intrusion into the thalamus. Importantly, analysis revealed that GluN2B 2A(CTR)/2A(CTR) mice exhibited smaller lesion volumes in the hippocampus and the thalamic region (and smaller overall lesion volumes) than GluN2B +/+ mice ( Figures 3C-3F). Thus, the GluN2 CTD subtype also influences NMDAR-mediated excitotoxicity in vivo.
Differential Signaling to CREB Contributes to GluN2 CTD Subtype-Specific Excitotoxicity We next investigated the mechanistic basis for the observed GluN2 CTD subtype-dependent differences in vulnerability to excitotoxicity. NMDAR-dependent activation of CREB-dependent gene expression protects against excitotoxicity (Lee et al., 2005) and can act as a protective response to excitotoxic insults (Mabuchi et al., 2001). We found that basal levels of CREB (serine-133) phosphorylation (normalized to total CREB) were unaltered in GluN2B 2A(CTR)/2A(CTR) neurons (118% ± 12% compared to GluN2B +/+ neurons, p = 0.2). However we found that in response to treatment with NMDA C1 , CREB (serine-133) phosphorylation was more prolonged in GluN2B 2A(CTR)/2A(CTR) neurons than in GluN2B +/+ neurons, as assayed by western blot and immunohistochemistry ( Figures 4A-4C), and also that activation of a CRE-reporter gene and the CREB target gene Adcyap1 was stronger in GluN2B 2A(CTR)/2A(CTR) neurons than GluN2B +/+ (Figures 4D and 4E). These differences did not extend to all transcriptional events: no differences were seen in the

Neuron
The GluN2 CTD Subtype Controls Excitotoxicity NMDA C1 -induced activation of Srxn1, an AP-1 target gene , or suppression of the FOXO target gene Txnip (Al-Mubarak et al., 2009;Figures S3A and S3B). To confirm whether CREB-dependent gene expression causally influenced vulnerability to NMDAR-mediated excitotoxicity we utilized the inhibitory CREB family member ICER which we have previously confirmed blocks the induction of CRE-mediated gene expression when expressed in cortical neurons . ICER expression increased levels of NMDA C1 -induced death in both GluN2B 2A(CTR)/2A(CTR) and GluN2B +/+ neurons ( Figures  4F-4H). However, the effect of ICER on GluN2B 2A(CTR)/2A(CTR) neurons was greater than its effect on GluN2B +/+ neurons (Figure 4G), indicating that differential CREB activation is a contributing factor to the observed CTD subtype-dependent control of excitotoxicity.
The basis for stronger association of PSD-95 with GluN2B WT compared to GluN2B 2A(CTR) could be due to different sequences immediately upstream of the conserved C-terminal PDZ ligand. We generated a chimeric variant of GluN2B in which the final 12 amino acids of its CTD have been replaced by those of GluN2A (three amino acid differences, GluN2B (2A-PDZ) ). Coimmunoprecipitation studies revealed that GluN2B (2A-PDZ) had a similar affinity for PSD-95 as GluN2B ( Figure S4C), indicating that immediate upstream sequence differences are not the basis for differential association of PSD-95 with the CTDs of GluN2B and GluN2A. Recently, additional PSD-95 interaction domains have been discovered on internal regions of CTD 2B (1086-1157Cousins et al., 2009), which could contribute to the overall affinity of the CTD for PSD-95. The role of these additional regions in neurons is not yet clear, but could act to stabilize the primary interaction with the C-terminal PDZ ligand, or even act independently. Deletion of this region (creating GluN2B D(1086(creating GluN2B D( -1157 ) resulted in a small reduction in PSD-95 association ( Figure 5G). Importantly, NMDA-induced death following overexpression of GluN2B D(1086-1157) in primary rat hippocampal neurons (as per the assays used in Figure 1) was significantly lower than in neurons overexpressing GluN2B WT ( Figure 5H), even though whole-cell NMDAR currents were found to be the same in GluN2B D(1086-1157) as wildtype GluN2B WT -expressing neurons ( Figure 5I), implicating this region of the CTD in contributing to prodeath NMDAR signaling.

DISCUSSION
We have demonstrated distinct roles for the CTDs of GluN2B and GluN2A in determining the dose response of NMDAR-mediated excitotoxicity. CTD 2B promotes neuronal death more efficiently than CTD 2A , an effect which is observed regardless of whether the CTD is tethered to the channel portion of GluN2B or of
At developmental stages where GluN2B-containing NMDARs dominate at all locations, differential synaptic versus extrasynaptic NMDAR signaling is still observed (Hardingham et al., 2002). Importantly, the strong trans-synaptic activation of synaptic GluN2B-containg NMDARs is neuroprotective (Martel et al., 2009;Papadia et al., 2008). Our current study shows that the identity of the GluN2 CTD profoundly influences excitotoxicity in the context of chronic activation of all (synaptic and extrasynaptic) NMDARs, scenarios that are likely to exist in pathological situations such as ischemia, traumatic brain injury, or glutamate dyshomeostasis triggered by disease-causing agents. Thus, location/stimulus-specific effects can be uncoupled from GluN2 subunit-specific effects, suggesting that subunit/CTD composition represents an additional factor that determines the level of excitotoxicity following chronic NMDAR activation. This is further supported by the fact that recent electrophysiological and immuno-EM studies have shown that GluN2 subunit composition may not be dramatically different at synaptic versus extrasynaptic sites (Harris and Pettit, 2007;Petralia et al., 2010;Thomas et al., 2006). Our observations that swapping CTD 2B for CTD 2A has little effect on whether a subunit ends up at a synaptic or extrasynaptic site is consistent with the aforementioned studies reporting that subunits do not have a strong location preference. Any apparent enrichment of synaptic sites for GluN2A may reflect the fact that GluN2A upregulation coincides developmentally with increased synaptogenesis (Liu et al., 2004), or be due to the influence of sequences outside of the CTD.
That notwithstanding, GluN2B has been reported to be partly enriched at extrasynaptic locations in neurons (Groc et al., 2006;Martel et al., 2009;Tovar and Westbrook, 1999), which suggests that GluN2 subtype effects and location effects may cooperate to exacerbate excitotoxicity under certain circumstances. Of note, recent work has revealed a causal role for enhanced GluN2B-containing extrasynaptic NMDARs in ischemic neuronal death (Tu et al., 2010). Also, a specific increase in GluN2Bcontaining NMDARs in medium-sized spiny striatal neurons, specifically at extrasynaptic locations, contributes to phenotype onset in a model of Huntington's disease Milnerwood et al., 2010), where the synaptic/extrasynaptic NMDAR balance controls mutant Huntingtin toxicity (Okamoto et al., 2009).
The idea that subunit composition influences excitotoxicity independently or additively to the influence of receptor location raises the possibility of a hierarchy of NMDARs when it comes to promoting excitotoxicity, based on the combination of composition (2A versus 2B) and location (synaptic versus extrasynaptic). Whereas strong activation of synaptic GluN2Bcontaining NMDARs is well-tolerated and neuroprotective (Martel et al., 2009;Papadia et al., 2008), the current study raises the possibility that activation of synaptic GluN2B-containing NMDARs (but not GluN2A-containing) could augment excitotoxicity in the context of chronic extrasynaptic NMDAR activation, for example, through enhanced NO production. This would explain the antiexcitotoxic effect of TAT-NR2B9c, PSD-95 knockdown, or disrupting the PSD-95-nNOS interface (Aarts et al., 2002;Cao et al., 2005;Sattler et al., 1999;Soriano et al., 2008b;Zhou et al., 2010), and the reversal of CTD 2B -dependent CREB inactivation by TAT-NR2B9c and nNOS inhibition (Figure 5). However, because PSD-95 clusters have been observed at extrasynaptic sites (Carpenter-Hyland and Chandler, 2006), colocalizing with extrasynaptic NMDARs (Petralia et al., 2010), the possibility that extrasynaptic CTD 2B also contributes to this pathway should not be ruled out. Regardless of these issues, targeting GluN2B-PSD95 signaling to neurotoxic pathways offers genuine translational potential because it has been recently shown that stroke-induced damage and neurological deficits can be prevented in nonhuman primates by the administration of TAT-NR2Bc as late as 3 hr after stroke onset (Cook et al., 2012).
Investigations into why PSD-95 association with GluN2B WT is stronger than its association with GluN2B 2A(CTR) implicated a previously identified internal region (Cousins et al., 2009) as a contributing factor, although deleting it had a relatively small effect on PSD-95 association, indicating that other determinants may also be relevant. Also, differing affinities of CTD 2B and CTD 2A for PSD-95 may be partly due to other factors binding CTD 2A , occluding PSD-95 binding.
It is also possible that signals other than NO underlie the differential CTD subtype prodeath signaling, or that NO affects pathways other than CREB. One known NO target is the PI3K-Akt pathway, which is induced by NMDAR activity and neuroprotective in this context (Lafon-Cazal et al., 2002;Papadia et al., 2005). Modest NO levels promote PTEN S-nitrosylation, boosting Akt activity, whereas excessive NO also S-nitrosylates Akt itself, inactivating it (Numajiri et al., 2011). We have preliminary evidence that NMDA-induced Akt activation is enhanced in GluN2B 2A(CTR)/2A(CTR) neurons (M.A. Martel and G.E. Hardingham, unpublished data), and it will be of interest to determine any role of differential NO production. Also, it would be of interest to know whether NMDAR signaling to protective transcriptional responses other than CREB are sensitive to GluN2 CTD subtype (e.g., Iduna; Andrabi et al., 2011). These, and other issues surrounding subunit-specific signaling could benefit from a future systematic analysis of the NMDAR signaling complex in GluN2B +/+ versus GluN2B 2A(CTR)/2A(CTR) neurons.

EXPERIMENTAL PROCEDURES
Neuronal Culture and Induction of Excitotoxicity Cortical mouse and hippocampal rat neurons were cultured as described (Papadia et al., 2008) at a density of between 9 and 13 3 10 4 neurons per cm 2 from E17.5 mice or E21 rats with neurobasal growth medium supplemented with B27 (Invitrogen, Paisley, UK). Stimulations of cultured neurons were done in most cases after a culturing period of 9-11 days, during which neurons develop a network of processes, express functional NMDA-type and AMPA/kainate-type glutamate receptors, and form synaptic contacts. Other experiments were performed at DIV 18. To apply an excitotoxic insult, neurons were first placed overnight into a minimal-defined medium  containing 10% MEM (Invitrogen) and 90% salt-glucose-glycine (SGG) medium (Bading et al., 1993; SGG: 114 mM NaCl, 0.219% NaHCO 3 , 5.292 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 10 mM HEPES, 1 mM Glycine, 30 mM Glucose, 0.5 mM sodium pyruvate, 0.1% Phenol Red; osmolarity 325 mosm/l; Papadia et al., 2005). Neurons were then treated with NMDA (Tocris Bioscience, Bristol, UK) at the indicated concentrations for 1 hr, after which NMDARs were blocked by adding the antagonist MK-801 (10 mM). After a further 23 hr, neurons were fixed and subjected to DAPI staining, and cell death was quantified by counting (blind) the number of shrunken, pyknotic nuclei as a percentage of the total. For analysis of excitotoxicity in GluN2B +/+ versus GluN2B 2A(CTR)/2A(CTR) neurons, approximately 800-1,200 cells were analyzed per condition, per replicate (repeated across several replicates).

GluN2B-2A(CTR) Knockin Mouse
GluN2B-2A(CTR) knockin mice contain a GluN2B gene in which the protein coding portion of the C-terminal exon has been replaced with the protein coding region of the C-terminal exon of GluN2A (C-terminal domain replacement, CTR). The C-terminal exon encodes amino acids 867G to 1482V (GluN2B) and 866G to 1464V (GluN2A), which represents over 95% of the CTD, beginning at position 838E (GluN2A) and 839E (GluN2B). All other regions of the GluN2B gene are unaltered, including the 3 0 UTR, although there remains a 61 bp insert containing a loxP site located after the STOP codon at the beginning of the 3 0 UTR (a remnant of the excision of the Neo-selection cassette). To obtain cultured neurons from GluN2B 2A(CTR)/2A(CTR) mice, male and female heterozygous GluN2B +/2A(CTR) mice were mated, and the cortices from individual E17.5 mice were cultured as above. See Supplemental Experimental Procedures for further details.

Transfection and Following the Fate of Transfected Cells
Neurons were transfected at DIV8 using Lipofectamine 2000 (Invitrogen), using an established protocol (McKenzie et al., 2005). Transfection efficiency was approximately 5%. Greater than 99% of eGFP-expressing transfected neurons were NeuN-positive, and <1% were GFAP-positive (Soriano et al., 2008a), confirming their neuronal identity. For studying the effects of expressing wild-type and chimeric receptors based on GluN2A and GluN2B, constructs were cotransfected with peGFP (ratio 1:1) to identify transfected cells. Coexpression at this ratio was confirmed in the case of pRFP (Papadia et al., 2008). After 48 hr, the transfected neurons were then either subjected to electrophysiological analysis or their fate following an excitotoxic insult was studied. Pictures of GFP-expressing neurons were taken on a Leica AF6000 LX imaging system, with a DFC350 FX digital camera. Using the automated cell-finder function within the Leica AF6000 software, images of transfected neurons were taken both before and 24 hr after a 1 hr treatment with NMDA (20 mM). Cell death was assessed by counting the number of surviving GFP-positive neurons. In the vast majority of cases, death was easily spotted as an absence of a healthy GFP-expressing cell where one once was. In place of the cell, there was in most cases (>90%) evidence of death in the form of fragmented neurites, fluorescent cell debris, and a pyknotic nucleus (Papadia et al., 2008). This confirmed that the cells were genuinely dying as opposed to more unlikely scenarios, such as quenching of eGFP fluorescence in a subpopulation of neurons. For each condition, 150-200 neurons were studied over several independent experiments. An identical experimental regime was employed for studying the influence of ICER expression on vulnerability of GluN2B 2A(CTR)/2A(CTR) and GluN2B +/+ neurons to NMDA-induced excitotoxicity. Neurons were transfected with vectors encoding eGFP and the inhibitory CREB family member ICER1 (Stehle et al., 1993), or a control vector (encoding b-globin). We have previously confirmed that ICER1 expression inhibits CREmediated gene expression in neurons . The fate of transfected neurons following exposure to NMDA was then studied as described previously.

Other Procedures
See Supplemental Experimental Procedures for details of genotyping, plasmid generation, electrophysiological recording conditions, qPCR analysis, Ca 2+ imaging, stereotaxic NMDA administration, NO assays, western blotting and immunofluorescence, co-immunoprecipitation, and equipment settings. All procedures were authorized under a UK Home Office approved project licence and adhered to regulations specified in the Animals (Scientific Procedures) Act (1986) and approved by the University of Edinburgh's Local Ethical Review Committee. Statistical testing involved a 2-tailed paired Student's t test. For studies employing multiple testing, we used a one-way ANOVA followed by Fisher's LSD or Tukey's post hoc test.

SUPPLEMENTAL INFORMATION
Supplemental Information includes four figures and Supplemental Experimental Procedures and can be found with this article online at doi:10.1016/j. neuron.2012.03.021.