The NMDARs are heterotetrameric complexes composed of eight different subunits in humans (and in rodents) including GluN1, GluN2(A-D), and GluN3(A/B; Collingridge et al., 2004). While the two GluN1 subunits are essential for the formation of functional NMDAR complexes, the remaining GluN2/3 subunits determine the receptor’s functional diversity (Paoletti et al., 2013). In the case of GluN2B-containing NMDARs, at least three different compositions have been studied using recombinant expression systems. These receptor subtypes include the diheteromeric GluN1/GluN2B (2B/2B)- (Hansen et al., 2014; Yi et al., 2019), triheteromeric GluN1/GluN2A/GluN2B (2A/2B)- (Hansen et al., 2014), and triheteromeric GluN1/GluN2B/GluN2D (2B/2D)-NMDARs (Yi et al., 2019; Figure 1). Despite some recent advancements in the pharmacological understanding of each receptor subtype, the proportion of diheteromeric and triheteromeric GluN2B-containing NMDARs expressed in CNS neurons is still a subject of debate.

Although triheteromeric GluN2B-containing NMDARs may be an important subpopulation of GluN2B-containing NMDARs in CNS neurons, most physiological studies overlook their contribution, primarily due to the inability of currently used experimental techniques to separate them from diheteromeric 2B/2B-NMDARs. To further complicate this issue, when used at their typical working concentrations, subtype-specific antagonists that are known to almost fully block the diheteromeric 2B/2B-NMDAR currents generally block a proportion (~30%–60%) of the triheteromeric GluN2B-containing NMDARs currents (Hansen et al., 2014; Stroebel et al., 2018). Thus, reduction of NMDAR function by GluN2B-specific antagonists could theoretically be due to either a full blockade of diheteromeric 2B/2B-NMDAR function or a partial blockade of triheteromeric GluN2B-containing NMDAR function. Nevertheless, accumulating evidence supports that at least some diheteromeric 2B/2B-NMDARs are present throughout the life span of neurons, and they have channel functions and downstream signaling properties that are different from non-GluN2B-containing NMDARs. Consequently, the physiological functions that will be discussed in this review are thought to be contributed largely by the diheteromeric 2B/2B-NMDARs. It is important to acknowledge, however, that the same physiological functions may be contributed by triheteromeric GluN2B-containing NMDARs.

The GluN2B subunit shows distinct temporospatial expression patterns from other GluN2/3 subunits (Wenzel et al., 1997). Sharp contrasts have been drawn between the expression patterns of the GluN2B and GluN2A subunits. For example, the GluN2B subunit is by far the most expressed GluN2/3 subunit in embryonic murine brains; however, its expression peaks at around the first postnatal week, after which it is partially and gradually replaced by the increasingly expressed GluN2A subunit (Wenzel et al., 1997). In adult murine forebrain neurons, one commonly cited finding is that the GluN2B subunit is mainly located extrasynaptically, whereas the GluN2A subunit is primarily expressed at the neuronal synapses (Stocca and Vicini, 1998). Nevertheless, this model of subunit-dependent localization of NMDAR subtypes is likely too simplistic, since some studies have reported a substantial proportion of GluN2B-containing NMDAR-mediated synaptic currents in more mature neurons (Thomas et al., 2006; Pegasiou et al., 2020).

The temporospatial expression of diheteromeric vs. triheteromeric GluN2B-containing NMDAR subtypes is still a subject of immense debate. Based on the predominant expression of the GluN2B subunit during early postnatal days, it is reasonable to hypothesize that diheteromeric 2B/2B-NMDAR is the dominant NMDAR complex during early development (Wenzel et al., 1997). As the expression of the GluN2A subunit increases (particularly at the postsynaptic density), synaptic NMDARs in the cortical neurons may start to be populated by diheteromeric 2A/2A-or triheteromeric 2A/2B-NMDARs (Paoletti et al., 2013). Nevertheless, the exact composition of the NMDARs throughout the developmental time course has yet to be determined by using more sophisticated structural biology methods, such as using cryogenic electron microscopy (cryo-EM) to investigate the subunit stoichiometry of native NMDARs in the CNS (Zhao et al., 2019).

Some divalent cations, including Zn²⁺ and Mg²⁺, modulate the activities of NMDARs in a subtype-specific manner (Nowak et al., 1984; Paoletti et al., 2013). For example, Zn²⁺ is a more potent allosteric inhibitor on diheteromeric 2A/2A-NMDARs than diheteromeric 2B/2B-NMDARs, and, thus, a frequently used pharmacological tool for investigating GluN2A-containing NMDAR functions (Nozaki et al., 2011); however, its specificity for NMDARs containing the GluN2A subunit is limited by its similarly potent inhibition on triheteromeric 2A/2B-NMDAR currents and its modest efficacy of blocking (~50%) diheteromeric 2A/2A-NMDAR currents under physiological pH (Hansen et al., 2014). NMDAR subtypes also experience voltage-dependent channel blocking by Mg²⁺, and the Mg²⁺ sensitivity is roughly 10 times stronger for NMDARs containing the GluN2A/GluN2B subunits compared with those containing the GluN2C/GluN2D subunits (Paoletti et al., 2013). Importantly, the relief from Mg²⁺ inhibition of the GluN2A and/or GluN2B-containing NMDAR currents during repetitive or strong synaptic activities is critical for activity-dependent synaptic plasticity (Paoletti et al., 2013). The subtype-dependent modulation of the activities of GluN2B-containing and other NMDAR subtypes, by these divalent cations, are important physiological mechanisms for regulating neuronal functions.

Developing subtype-specific modulators for GluN2B-containing NMDARs has been a long-standing interest for both research and therapeutic purposes. Multiple GluN2B-containing NMDAR antagonists have been tested in preclinical studies and in clinical trials, most of which are analogous to a phenylethanolamine called ifenprodil, which binds to a GluN1-GluN2B subunit interface at the N-terminal domain (Karakas et al., 2011; Liu W. et al., 2020). Some of the antagonists in this class, such as Ro25-6981 and CP-101606, have achieved outstanding subtype specificity with over 1,000-fold higher affinity for diheteromeric GluN2B-containing NMDARs compared with other diheteromeric NMDAR subtypes (Chenard et al., 1995; Fischer et al., 1997). Although Ro25-6981, when applied at its working concentration, also blocks a proportion of the currents measured from recombinantly expressed triheteromeric 2A/2B-NMDARs (Hansen et al., 2014), its blocking effect on the 2A/2B-NMDAR currents is small (~30%) compared with that on diheteromeric 2B/2B-NMDAR currents (>90%). Therefore, this drug has been a useful tool for investigating the physiological functions of diheteromeric 2B/2B-NMDARs.

Several endogenous polyamines in the CNS, including spermine and spermidine, positively modulate the GluN2B-containing NMDAR currents (Williams et al., 1994). Importantly, the positive modulation is specific to the diheteromeric 2B/2B-NMDARs with little to no effect on other NMDAR subtypes, including the triheteromeric 2A/2B-NMDARs (Stroebel et al., 2014). It is important to note that polyamines generally have low potency toward modulating the diheteromeric 2B/2B-NMDAR currents (EC50 around hundreds of μM) and may also affect many other targets in the CNS at the concentration that is commonly used (Guerra et al., 2016). Further understanding of the structural biology of polyamine modulation on diheteromeric NMDARs, including the resolution of a crystal structure illustrating the binding of polyamines to 2B/2B-NMDARs, may facilitate the development of more potent and specific modulators to investigate the functions of diheteromeric 2B/2B-NMDARs.

Long-term potentiation (LTP) has been acknowledged as a well-investigated form of activity-dependent synaptic plasticity in mammalian brains. Although the roles of GluN2B-containing NMDARs in LTP may be ambivalent overall, many earlier lines of evidence support the importance of this NMDAR subgroup in the induction of LTP (Shipton and Paulsen, 2014). First, compared with the GluN2A subunit, the GluN2B subunit has a much higher affinity for calcium calmodulin-dependent kinase II (CaMKII), a key protein kinase that is involved in LTP induction (Barria and Malinow, 2005). Knockin mice with the GluN2B subunit’s CaMKII binding site residues replaced by the analogous ones on the GluN2A subunit have severely impaired LTP (Barria and Malinow, 2005). Furthermore, genetic ablation of the GluN2B subunit in the hippocampal CA1 region abolishes LTP in a subset of hippocampal circuits, whereas transgenic mice overexpressing the GluN2B subunit have dramatically enhanced LTP in the hippocampal CA3-CA1 synapse (Tang et al., 1999; Akashi et al., 2009). Together, these data support the hypothesis that GluN2B-containing NMDARs and their associated protein kinases play essential roles in the induction of LTP.

Many recent studies examining the roles of GluN2B-containing NMDARs in LTP have generated mixed results (Shipton and Paulsen, 2014). For example, several lines of lately generated transgenic mice conditionally lacking the GluN2B subunit in the hippocampus only have minor impairment in a weak form of tetanus-induced LTP, but no impairment in a stronger form of LTP in the CA3-CA1 synapse (von Engelhardt et al., 2008; Brigman et al., 2010). Moreover, GluN2B-containing NMDAR antagonists such as Ro25-6981 do not affect high-frequency stimulation-induced LTP in many studies (see Liu et al., 2004; Fox et al., 2006; Papouin et al., 2012 but Bartlett et al., 2007; Berberich et al., 2007). Several factors, including the age of the animals (Ito et al., 1996), LTP induction protocol (von Engelhardt et al., 2008), and neural circuit and brain region (Akashi et al., 2009) may affect the roles of GluN2B-containing NMDAR in LTP induction. Overall, the requirement of the GluN2B subunit for the induction of LTP is likely situation-dependent.

Several factors will need to be considered when interpreting the roles of GluN2B-containing NMDARs in LTP. First, most current studies ignore the contribution of triheteromeric NMDARs, especially those containing both GluN2A and GluN2B subunits. One study provided evidence for the importance of triheteromeric 2A/2B-NMDARs in LTP using carefully designed interrogation of NMDAR subtype functions (Delaney et al., 2012), but exclusive isolation of the triheteromeric NMDAR functions remains challenging with current experimental approaches. Second, transgenic mice studies overexpressing or deleting GluN2B may alter the neuronal development and/or trigger molecular mechanisms to compensate for the changes in the GluN2B subunit expression, thereby misrepresenting the true physiological functions of the subunit (Ito et al., 1996; Tang et al., 1999). Third, multiple forms of NMDAR-dependent LTP may be present (Liu A. et al., 2020), but most experiments testing the roles of GluN2B-containing NMDARs in LTP used a single form of induction protocol limited to the same neural circuit, thus considering LTP as a unified phenomenon. Nevertheless, as illustrated by a thoughtfully designed study using both genetic and pharmacological manipulations (Foster et al., 2010), it is generally accepted that the C-terminal of the GluN2B subunit plays a scaffolding role for recruiting proteins that are critical for LTP induction, whereas the channel function specific for diheteromeric 2B/2B-NMDAR may not be required for the induction of LTP.

The NMDAR-dependent form of long-term depression (LTD)-most commonly studied at the hippocampal CA3-CA1 circuit-has been one of the best-characterized mechanisms of activity-dependent reduction in synaptic strength (Collingridge et al., 2010). Increasing amount of evidence suggests an important role of GluN2B-containing NMDARs in the induction of NMDAR-dependent LTD. One of the earliest studies in this field, by using differential activation of NMDAR subpopulations, found that synaptic NMDAR activation mediated LTP, while extrasynaptic NMDAR activation mediated LTD (Lu et al., 2001). Since the GluN2B subunit is predominant expressed in the extrasynaptic location in adult neurons (Stocca and Vicini, 1998), this evidence indirectly supports the importance of GluN2B-containing NMDARs in LTD.

Subsequent pharmacological studies using subtype-specific modulators found that Ro25-6981 was sufficient to block LTD in the CA3-CA1 synapse using acutely prepared brain slices derived from adult animals (see Liu et al., 2004; Fox et al., 2006; Papouin et al., 2012 but Bartlett et al., 2007; Morishita et al., 2007) and in vivo using anesthetized animals (Fox et al., 2006; Ge et al., 2010). Notably, both homozygote Cre-dependent GluN2B knockout and heterozygote loss-of-function GluN2B knockin in mice reliably abolish LTD (Ito et al., 1996; Brigman et al., 2010; Shin et al., 2020), providing the strongest evidence for the critical roles of the GluN2B subunit in LTD. Despite the inconsistent results derived from pharmacological studies, a major proportion of evidence, particularly from studies using in vivo models, supports an essential role of GluN2B-containing NMDARs in LTD induction.

GluN2B-containing NMDARs play important roles in a wide range of cognitive functions in rodent models, and this phenomenon is thought to be mediated by their roles in LTP and LTD (von Engelhardt et al., 2008; Brigman et al., 2010). For example, transgenic mice overexpressing the GluN2B subunit in forebrain neurons have improved object recognition memory (Tang et al., 1999), spatial memory (Tang et al., 1999), working memory (Cui et al., 2011), social recognition memory (Jacobs and Tsien, 2012), and motor skill learning (Duan et al., 2018). These enhanced learning abilities may be related to the enhanced LTP in the transgenic mice (Tang et al., 1999; Cui et al., 2011). By contrast, knocking out the GluN2B subunit in mice cortex and hippocampal CA1 regions impaired spatial learning, working memory, and trace fear conditioning (Brigman et al., 2010). Such learning deficits may be related to the impaired LTD (Brigman et al., 2010). Altogether, these pieces of evidence support the indispensable roles of GluN2B-containing NMDARs in many types of learning and memory and the possibility of memory enhancement by enhancing GluN2B-containing NMDAR functions. It should be noted that whether these effects on learning and memory, particularly those observed from the GluN2B subunit-overexpressed transgenic mice, are due to the functional (channel opening) and/or structural (recruiting critical molecules such as CaMKII) roles of GluN2B-containing NMDARs remains to be determined.

Recent behavioral experiments indicate that certain types of cognitive performances, especially those that require flexible adaptations to learned responses, are selectively affected by the manipulation of GluN2B-containing NMDAR functions. Pharmacological inhibition of GluN2B-containing NMDAR function impaired fear memory extinction (Milton et al., 2013), strategy shift during a discrimination task (Dalton et al., 2011), and reversal learning during Morris Water maze (Dong et al., 2013) in rodents, but not their performances on fear conditioning, visual discrimination, and initial spatial learning. Moreover, the impaired behavioral flexibility following GluN2B-antagonism may be mediated by the deficits in LTD, which could be critical for weakening the previously formed associations before new learning can takes place. Indeed, impairment of fear extinction learning in rats by a GluN2B-containing NMDAR antagonist is associated with their impaired LTD in an amygdala circuit (Dalton et al., 2012), while enhancement of spatial reversal learning in mice by NMDAR co-agonist D-serine is associated with their enhanced LTD in the hippocampus (Duffy et al., 2008). Although GluN2B-containing NMDARs have been implicated in other cognitive functions related to memory acquisition (Howland and Cazakoff, 2010; Dong et al., 2012), retrieval (Wong et al., 2007), and forgetting (Sachser et al., 2016), their role in learning tasks that require behavioral flexibility has been the most consistently replicated phenomenon across different behavioral assays and research settings.

Overactivation of NMDARs during pathological conditions is known to cause excitotoxic neuronal death (known as excitotoxicity), during which different NMDAR subtypes may have distinct, even opposite functions (Wu and Tymianski, 2018; Ge et al., 2020). A large body of literature has suggested that pathological overactivity of GluN2B-containing NMDARs is associated with neuronal death (Ge et al., 2020). For example, neurons expressing chimeric NMDARs with the C-terminal region of the GluN2A subunit replaced by that of the GluN2B subunit experienced increased susceptibility to NMDA-induced cell death, an in vitro model of excitotoxic neuronal injury (Martel et al., 2012). By contrast, transgenic mice with parts of the GluN2B C-terminal domain (CTD) region deleted or modified experienced decreased susceptibility to in vitro NMDA-induced cell death and an in vivo middle cerebral artery occlusion (MCAO) model of stroke (Martel et al., 2012; Vieira et al., 2016; Tang et al., 2018). Furthermore, antagonists that are selective for GluN2B-containing NMDARs, but not those that are preferential for GluN2A-containing NMDARs, protected neurons from excitotoxic death during multiple forms of neuronal injury (DeRidder et al., 2006; Liu et al., 2007; Chen et al., 2008). Although the strict “NMDAR subtype” hypothesis in neuronal survival and death has been challenged by some inconsistent findings (Papouin et al., 2012; Zhou et al., 2013), and by the recent evidence suggesting the roles of triheteromeric 2A/2B-NMDAR and other GluN2-containing NMDAR subtypes in neuronal death (Doyle et al., 2018; Ma et al., 2019), it is generally accepted that GluN2B-containing NMDARs, by coupling to multiple neuronal death signaling complexes, play a dominant role in mediating neuronal toxicity (Wu and Tymianski, 2018; Ge et al., 2020).

The GluN2B-CaMKII has been the best-characterized GluN2B-dependent signaling complex by far, largely due to the abundance of CaMKII holoenzyme and its importance in synaptic plasticity, especially LTP, and the formation and maintenance of glutamatergic synapses (Lisman et al., 2012; Bayer and Schulman, 2019; Yasuda et al., 2022). During the induction of LTP, a high calcium concentration mediated by the calcium influx through the NMDAR activates calcium calmodulin (CaM), which triggers the CaMKII kinase activity, autophosphorylation of the threonine (Thr)286 residue, and binding to the GluN2B-subunit of the NMDAR (Strack and Colbran, 1998; Bayer et al., 2001; Barria and Malinow, 2005). The binding of CaMKII to GluN2B is critical for inducing the downstream signaling cascades, including the phosphorylation of multiple protein substrates for the CaMKII-dependent potentiation of the synaptic strength (Barria et al., 1997; Tomita et al., 2005; Figure 2). It is important to acknowledge that CaMKII also plays important roles in LTD (Bayer and Schulman, 2019). Notably, the effects of CaMKII on LTD may not be dependent on its binding with GluN2B, but on the autonomous activity of CaMKII (Pi et al., 2010). This autonomous activity is inhibited by a high concentration of CaM but favors a prolonged level of intracellular calcium at low to moderate concentrations (Bayer and Schulman, 2019), and is dependent on the autophosphorylation of Thr305/306 on CaMKII (Pi et al., 2010; Cook et al., 2021; Figure 2). Additionally, LTD stimuli in the excitatory synapse may result in CaMKII translocation to the inhibitory synapse and the subsequent inhibitory LTP (iLTP; Cook et al., 2021). Overall, the roles of CaMKII in bidirectional plasticity are highly consistent with the type of stimulation that is necessary to induce LTP (strong and brief) and LTD (moderate and long) in neurons.

CaMKII has long been implicated in neuronal survival and death, but the direction to which it contributes to neuronal viability has been controversial (Coultrap et al., 2011). Recently, it has been found that a peptide inhibitor tat-CN21, which blocks the autonomous activity of CaMKII, is neuroprotective against glutamate-induced excitotoxicity when applied before or shortly after the insult, whereas CaMKII inhibitor KN-93, which does not block the autonomous activity of CaMKII, has no neuroprotective effect (Vest et al., 2010; Ashpole and Hudmon, 2011). These studies suggest that the autonomous activity of CaMKII, in addition to mediating LTD, may also mediate a critical signaling pathway underlying neuronal death. While the autonomous activity of CaMKII has been increasingly acknowledged as a target for neuroprotection, it is important to note that prolonged inhibition of CaMKII autonomous activity by tat-CN21 for longer than 8 h increased, rather than decreased the susceptibility of neurons to subsequence excitotoxic insults (Ashpole et al., 2012). Moreover, compared to wild-type mice, transgenic mice without CaMKII are more susceptible to ischemic neuron insults, suggesting that the same pathway may also be implicated in neuronal survival signaling (Waxham et al., 1996). Since CaMKII protein may be critical for both neuronal survival and neuronal death, further evidence clarifying the CaMKII-dependent neuronal survival signaling from neuronal death signaling pathways, including the involvement of GluN2B-CaMKII binding in the signaling, will be needed to optimize the CaMKII-based neuroprotective strategy.

The GluN2B-death-associated protein kinase 1 (DAPK1) complex plays a critical role in mediating downstream signaling cascades following GluN2B-containing NMDAR activation (Wu and Tymianski, 2018; Ge and Wang, 2022). To begin with, the GluN2B-DAPK1 has been discovered as a neuronal death signaling complex which rapidly assemblies upon the activation of extrasynaptic NMDAR during excitotoxic insult (Tu et al., 2010). Transgenic mice lacking the DAPK1 gene or with mutated GluN2B C-terminal region abolishing GluN2B-DAPK1 binding is resistant to ischemic insult during in vitro and in vivo models of stroke (see Tu et al., 2010; Tang et al., 2018 but McQueen et al., 2017), while interfering the formation of the GluN2B-DAKP1 complex using cell-penetrant peptides Tat-GluN2B_ct and Tat-GluN2B_ct-CTM is neuroprotective against mice models of ischemic stroke (Tu et al., 2010; Fan et al., 2014). Although conflicting evidence regarding the importance of this signaling pathway in neuronal death has been reported, especially regarding the requirement of GluN2B S1303 phosphorylation by DAPK1 during neuronal death signaling (McQueen et al., 2017; Tullis et al., 2021), the GluN2B-DAPK1 complex has still been well-recognized as one of the critical signaling pathways for NMDAR-receptor mediated excitotoxicity and, consequently, an important drug target for treating ischemic stroke.

Other than playing a central role in neuronal death signaling, the GluN2B-DAPK1 complex has recently been found to mediate NMDAR-dependent LTD (Goodell et al., 2017; Figure 2). Interestingly, the signaling of GluN2B-DAPK1 during LTD is dependent on the protein phosphatase calcineurin (Tu et al., 2010; Goodell et al., 2017), and this new evidence is highly coherent with the well-recognized roles of protein phosphatases in LTD. Moreover, DAPK1 competitively prevents the formation of the GluN2B-CaMKII complex, and this process blocks the induction of LTP by CaMKII and favors the induction of LTD (Goodell et al., 2017). The unique roles of DAPK1 in regulating GluN2B-CaMKII complex activity may be critical for determining the direction of synaptic plasticity (Bayer and Schulman, 2019). These studies illustrating the roles of the GluN2B-DAPK1 complex in LTD, together with the recent studies describing the roles of CaMKII in LTD, help provide a coherent explanation regarding the signaling pathways underlying NMDAR-dependent LTP and LTD (Figure 2).

The GluN2B-Ras Protein Specific Guanine Nucleotide Releasing Factor 1 (RasGRF1) complex mediates another highly important signaling pathway for the induction of LTD, whereas the GluN2A-RasGRF2 signaling favors the production of LTP (Li et al., 2006; Feig, 2011; Miller et al., 2013). RasGRF1 is a small GTPase exchange factor (GEF) that binds specifically to the C-terminal region of the GluN2B subunit, but not that of the GluN2A or GluN1 subunits (Krapivinsky et al., 2003). The binding of RasGRF1 with GluN2B-containing NMDARs allows it to function as a calcium sensor and an activator of the p38 mitogen-activated protein kinase (MAPK) signaling pathway, which plays critical roles in LTD (Kim et al., 2005; Falcicchia et al., 2020). Interestingly, transgenic mice lacking RasGRF1 are specifically impaired in NMDAR-dependent LTD but not LTP, further supporting that the activation of RasGRF1 is more critical for this form of synaptic plasticity (Li et al., 2006). Besides mediating an important signaling pathway for LTD, some evidence suggests that the GluN2B-RasGRF1 complex may regulate neuronal fate. In comparison with their wild-type counterparts, transgenic mice lacking both RasGRF1 and RasGRF2 are more susceptible to neuronal infarction induced by MCAO (Tian et al., 2004). Moreover, p38 MAPK, the downstream effector of RasGRF1, is critical for mediating neuronal excitotoxicity (Cao et al., 2004). Given that LTD and neuronal death often involve similar signaling cascades, the role of the GluN2B-RasGRF1 complex in neuronal fate, particularly in neuronal death signaling, may warrant further investigation.

The GluN2B-postsynaptic density protein 95 (PSD95)-neuronal nitric oxide synthase (nNOS) complex has been an established signaling complex mediating neuronal death and an important drug target for neuroprotection (Sattler et al., 1999; Wu and Tymianski, 2018; Ge and Wang, 2022). Multiple drugs or interfering peptides, including NA-1 (also known as Tat-NR2B^C9; Aarts et al., 2002), Tat-N-dimer (Bach et al., 2012), and ZL006 (Zhou et al., 2010), were designed to disrupt the formation of this protein complex during neuronal injury, and these therapeutics achieved neuroprotection in rodent models of ischemic stroke (Zhou et al., 2010; Bach et al., 2012), traumatic brain injury (Qu et al., 2020), and epilepsy (Colciaghi et al., 2019). Notably, in a phase III clinical trial (NCT02930018), NA-1 effectively improved neurological outcomes and reduced the brain infarct volume in ischemic stroke patients undergoing endovascular treatment (Hill et al., 2020). This clinical evidence further validates the GluN2B-PSD95-nNOS complex as an important therapeutic target for neuroprotection and a key mediator of neuronal death signaling. Limited evidence also supports the role of GluN2B-PSD95-nNOS in synaptic plasticity. For example, disrupting this protein complex with ZL006 impaired LTP that was measured from the amygdala in rat brain slices (Li et al., 2018). Nevertheless, the definitive role of this protein complex in bidirectional synaptic plasticity has yet to be verified. Given that several other neuronal death signaling complexes have been implicated in synaptic depression, it may be worthwhile to further explore the roles of the GluN2B-PSD95-nNOS complex in LTD of the glutamatergic synapses.