Regulation of NMDA receptors by the tyrosine kinase Fyn

Authors

  • Catherine H. Trepanier,

    1.  Department of Pharmacology and Toxicology, University of Toronto, ON, Canada
    Search for more papers by this author
  • Michael F. Jackson,

    1.  Molecular Brain Research Group, Robarts Research Institute, University of Western Ontario, London, ON, Canada
    2.  Department of Physiology and Pharmacology, University of Western Ontario, London, ON, Canada
    Search for more papers by this author
  • John F. MacDonald

    1.  Department of Pharmacology and Toxicology, University of Toronto, ON, Canada
    2.  Molecular Brain Research Group, Robarts Research Institute, University of Western Ontario, London, ON, Canada
    3.  Department of Physiology and Pharmacology, University of Western Ontario, London, ON, Canada
    Search for more papers by this author

J. F. MacDonald, Molecular Brain Research Group, Robarts Research Institute, University of Western Ontario, 100 Perth Dr., London, ON, Canada N6A 5K8
Fax: +1 519 931 5721
Tel: +1 519 931 5777
E-mail: jfmacdonald@robarts.ca

Abstract

The phosphorylation and trafficking of N-methyl-d-aspartate (NMDA) receptors are tightly regulated by the Src family tyrosine kinase Fyn, through dynamic interactions with various scaffolding proteins in the NMDA receptor complex. Fyn acts as a point of convergence for many signaling pathways that upregulate GluN2B-containing NMDA receptors. In the following review, we focus on Fyn signaling downstream of different G-protein-coupled receptors: the dopamine D1 receptor, and receptors cognate to the pituitary adenylate cyclase-activating polypeptide. The net result of activation of each of these signaling pathways is upregulation of GluN2B-containing NMDA receptors. The NMDA receptor is a major target of ethanol in the brain, and accumulating evidence suggests that Fyn mediates the effects of ethanol by regulating the phosphorylation of GluN2B NMDA receptor subunits. Furthermore, Fyn has been shown to regulate alcohol withdrawal and acute tolerance to ethanol through a GluN2B-dependent mechanism. In addition to its effects on NMDA receptor function, Fyn also modifies the threshold for synaptic plasticity at CA1 synapses, an effect that probably contributes to the effects of Fyn on spatial and contextual fear learning.

Abbreviations
AC

adenylyl cyclase

D1R

dopamine D1 receptor

LTP

long-term potentiation

NMDA

N-methyl-d-aspartate

NMDAR

N-methyl-d-aspartate receptor

PACAP

pituitary adenylate cyclase-activating polypeptide

PKA

protein kinase A

PSD

postsynaptic density

RACK1

receptor for activated C kinase-1

SFK

Src family kinase

STEP

striatal enriched tyrosine phosphatase

Introduction

During fast excitatory synaptic transmission in the central nervous system, glutamate activates a mixture of N-methyl-d-aspartate (NMDA) receptors (NMDARs) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) in the postsynaptic cell. The enhancement of synaptic transmission, known as long-term potentiation (LTP), is expressed as an increase in the number of AMPARs in the postsynaptic membrane [1]. LTP induction at Schaffer collateral–CA1 synapses in the hippocampus requires the activation of NMDARs [1–4]. NMDARs are heterotetramers that consist of two obligatory GluN1 subunits and two modulatory GluN2 subunits [5]. Among the four types of GluN2 subunit (GluN2A–GluN2D), GluN2A and GluN2B exhibit the most broad expression patterns in the postnatal forebrain [6]. The subunit composition of NMDARs confers distinct biophysical and pharmacological properties on the receptor (reviewed in [7]).

As reviewed by Salter and Pitcher, the function of the NMDAR is kept in balance by the activities of serine/threonine kinases and phosphatases [8,9] and of protein tyrosine kinases and phosphotyrosine phosphatases [9–11]. Although Src was identified as the principal endogenous protein tyrosine kinase that regulates NMDAR channel gating [12], other members of the Src family kinases (SFKs), such as Fyn and Lyn, have been shown to modulate NMDAR function and signaling [13–19]. Here, we focus on the Fyn tyrosine kinase as an important modulator of synaptic plasticity in the CA1 region of the hippocampus. The trafficking and surface expression of NMDARs is tightly regulated via direct phosphorylation by SFKs [13,20]. Thus, we propose that Fyn differentially regulates the trafficking of GluN2A-containing and GluN2B-containing NMDARs, and specifically modulates the function of GluN2B-containing NMDARs.

Fyn regulation of NMDAR phosphorylation and trafficking

The GluN2 subunits of the NMDAR are phosphorylated by SFKs on tyrosines located within the long intracellular C-terminal domains [21]. As described by Groveman et al., neuronal Src interacts with the C-terminus of GluN2A as well other components of the NMDAR complex [22–24]. The Fyn tyrosine kinase, on the other hand, has been shown to phosphorylate both GluN2A and GluN2B in postsynaptic densities (PSDs) of the rat forebrain [16,25–28]. Tryptic peptide mapping identified seven major tyrosine-phosphorylated peptides from GluN2A and five from GluN2B [29]. Three major tyrosines in the GluN2B C-terminal tail have been identified as Fyn phosphorylation sites by site-directed mutagenesis: Tyr1252, Tyr1336, and Tyr1472, with the latter site being the most prominently phosphorylated site in vitro [27]. Meanwhile, the tyrosine sites in GluN2A phosphorylated by Fyn have not been identified.

Phosphorylation of GluN2 subunits by exogenous Fyn is dependent on Src homology 2 domain-mediated binding to the PSD [26]. The PSD proteins PSD-93 and PSD-95 were shown to promote the phosphorylation of NMDARs by Fyn [28,30]. Genetic deletion of PSD-93 caused a reduction in the expression of Fyn and phosphorylated GluN2A and GluN2B in the synaptosomal membrane fraction [30]. In HEK293 cells, PSD-95 was shown to promote the Fyn-mediated phosphorylation of GluN2A via an interaction with the Src homology 2 domain of Fyn [28]. Furthermore, Fyn phosphorylation was shown to enhance calpain-mediated truncation of both GluN2A and GluN2B [31,32]. Although calpain-mediated cleavage does not alter the channel properties of GluN2B-containing NMDARs, its functional effects on GluN2A-containing NMDARs remain less clear.

Phosphorylation of Tyr1472 in GluN2B by Fyn is proposed to stabilize synaptic NMDARs by preventing the interaction of the clathrin adaptor protein with the YEKL motif on GluN2B, thereby preventing endocytosis of NMDARs [33–37]. When a constitutively active form of Fyn was coexpressed with GluN1 and GluN2B in cerebellar granule cells, there was a significant increase in the amplitude of NMDA miniature excitatory postsynaptic currents with no change in the deactivation kinetics, suggesting an increase in the number of synaptic NMDARs [35]. In addition to preventing endocytosis of GluN2B-containing NMDARs, Fyn, activated downstream of dopamine D1 receptor (D1R) stimulation, is proposed to increase the trafficking of GluN2B-containing NMDARs [38,39]. In HEK293 cells, however, recombinant Fyn was shown to modulate glutamate-evoked currents in GluN1/GluN2A-transfected but not GluN1/GluN2B-transfected cells [40]. However, HEK293 cells lack many of the scaffolding proteins in the NMDAR complex that are found in neuronal tissues, and this may explain the discrepant GluN2 dependence. On the basis of studies in cerebellar cells, then, Fyn appears to selectively modulate the trafficking and channel properties of GluN2B-containing NMDARs.

The Fyn-mediated phosphorylation and upregulation of GluN2B-containing NMDARs is under tonic control by the inhibitory scaffolding protein receptor for activated C kinase-1 (RACK1), keeping the basal level of GluN2B phosphorylation to a minimum [41–43]. RACK1 binds directly to both Fyn and the C-terminus of GluN2B, allowing Fyn to be localized in close proximity to its substrate (Fig. 1) [41]. According to this model, if RACK1 is present, Fyn cannot phosphorylate GluN2B; however, once RACK1 is released from this complex, e.g. through protein kinase A (PKA) activation, Fyn is then free to phosphorylate GluN2B and increase NMDAR currents [43].

Figure 1.

Fyn-mediated phosphorylation of GluN2B-containing NMDARs is controlled by the inhibitory scaffolding protein RACK1 (adapted from [43]). (A) RACK1 binds to the C-terminus of the GluN1/GluN2B receptor and the tyrosine kinase Fyn. (B) Under conditions of increased cAMP formation and PKA activation, e.g. by forskolin, RACK1 is released from the complex and translocates to the nucleus. (C) Once RACK1 is released from the NMDAR, Fyn is then free to phosphorylate GluN2B on Tyr1472, the main Fyn phosphorylation site, resulting in enhanced NMDA channel activity.

Fyn modulation of GluN2B-containing NMDARs

Although Fyn has been shown to phosphorylate both GluN2A-containing and GluN2B-containing NMDARs, several lines of evidence indicate that Fyn selectively modulates the activity of GluN2B-containing NMDARs and enhances the activity of these receptors. The neuropeptide pituitary adenylate cyclase-activating polypeptide (PACAP) has been shown to modulate GluN2B-containing NMDARs [41,43,44]. Whether the PACAP-mediated potentiation of NMDAR currents requires activation of Src or Fyn kinase activities remains controversial [43,44]. Activation of the Gαs-protein-coupled D1R recruits the Fyn tyrosine kinase in a signaling cascade that upregulates GluN2B-containing NMDAR activity [38,39,45,46]. Fyn also modulates the ethanol sensitivity of NMDARs in a GluN2B-dependent manner, and contributes to the development of acute tolerance to ethanol [47–49].

PACAP receptor regulation of NMDAR function

In a model proposed by Ron’s group, activation of PACAP-specific receptors enhances NMDAR function by inducing the displacement of Fyn from the Fyn–RACK1–NR2B complex, thereby allowing Fyn to phosphorylate GluN2B and increase NMDA channel activity [41,43,47]. Application of the neuropeptide PACAP(1–38) increases tyrosine phosphorylation of GluN2B and increases NMDAR-mediated field excitatory postsynaptic potentials in CA1 hippocampal slices [43]. This enhancement was abolished in slices from Fyn null mutant mice and by the GluN2B antagonist ifenprodil, suggesting that PACAP signals through Fyn to enhance GluN2B-containing NMDAR function. These findings are at odds with results from our laboratory showing that PACAP signals through a related tyrosine kinase, Src, to upregulate NMDA-evoked currents [44]. We showed that inclusion of a selective Src inhibitory peptide, Src(40–58), eliminated the PACAP38-induced potentiation of synaptic NMDA currents. Furthermore, PACAP treatment of CA1 hippocampal slices significantly increased tyrosine phosphorylation of Src, including Tyr416, which is required for activation of the kinase [44]. As a further testament to the Src-dependency of the PACAP regulation, we introduced peptides homologous with the binding sites on RACK1 for Fyn (R1) and Src (R6) into acutely isolated CA1 neurons, and found that they occluded the effects of PACAP on NMDA responses [44]. Intracellular infusion of Src(40–58) blocked the ability of these peptides to potentiate NMDA peak currents, suggesting that Src mediates the PACAP modulation of NMDAR activity.

D1R regulation of NMDAR function

Accumulating evidence supports the involvement of Fyn in the D1R regulation of GluN2B-containing NMDARs [38,39]. Activation of D1Rs increases the trafficking of NMDARs from intracellular to postsynaptic subcellular compartments in the striatum, an effect that is abolished in Fyn null mutant mice [38]. Furthermore, genetic deletion of Fyn prevented the D1R-mediated enhancement of tyrosine-phosphorylated GluN2A and GluN2B in the synaptosomal membrane fraction. A more recent study corroborated the Fyn dependence of this signaling pathway in cultured rat prefrontal cortical neurons [39]. The D1R-mediated enhancement of GluN2B expression and surface insertion was completely blocked by small interfering RNA knockdown of Fyn but not of Src.

The GluN2B dependence of the D1R modulation of NMDAR currents is specific to CA3–CA1 synapses [50]. At the more distal entorhinal–CA1 synapses, D1R activation depresses NMDAR currents through a GluN2A-dependent mechanism. The D1R-mediated bidirectional modulation of NMDAR currents depends on the subtype of GluN2 that predominates at the synapse. At CA3–CA1 synapses, GluN2B subunits predominate and underlie the D1R-mediated potentiation of NMDAR currents; however, entorhinal–CA1 synapses are enriched in GluN2A subunits, and therefore mediate the D1R-induced depression of NMDAR currents [50]. Furthermore, a direct protein–protein interaction between the C-terminus of D1R (t3 domain) and GluN2A was identified, and shown to attenuate NMDAR-gated currents in hippocampal cultured neurons [51].

The striatal enriched tyrosine phosphatase (STEP) has been implicated in the D1R regulation of NMDARs and in the regulation of Fyn activity, and thus may be perfectly positioned in the signaling cascade linking D1Rs, Fyn, and NMDARs (Fig. 2) [45,46]. Activation of D1Rs leads to PKA-mediated phosphorylation of both STEP61 and STEP46 at regulatory serines within their KIM domains, which decreases the affinity of STEP for its substrates [46]. Phosphorylated STEP prevents the dephosphorylation of GluN2B at Tyr1472 and prevents the dephosphorylation of Fyn at Tyr420, thus increasing Fyn activity and increasing surface trafficking of GluN2B-containing NMDARs [45,52].

Figure 2.

Proposed model for the D1R-mediated upregulation of GluN2B-containing NMDARs. According to Lombroso’s model, activation of D1Rs leads to PKA-mediated phosphorylation and inactivation of STEP [46]. Phosphorylated STEP prevents the dephosphorylation and inactivation of Fyn [45], allowing Fyn to increase the trafficking of GluN2B-containing NMDARs to the cell surface [35,38,39]. Furthermore, phosphorylated STEP prevents the dephosphorylation of GluN2B at Tyr1472 [60], which further stabilizes GluN2B-containing NMDARs at the synaptic membrane [35].

Ethanol regulation of NMDAR function

Fyn regulates the ethanol sensitivity of NMDARs in a GluN2B-dependent manner, and is proposed to modulate acute tolerance to ethanol [47–49]. Fyn-deficient mice show enhanced sensitivity to the hypnotic effects of ethanol, as indicated by an enhanced duration of ethanol-induced loss of righting reflex as compared with control littermates [47–49]. GluN2B was shown to mediate acute sedative effects of ethanol, as preinjection with a GluN2B antagonist increased the duration of loss of righting reflex in Fyn wild-type mice to levels similar to those in Fyn null mutant mice [47]. Moreover, deletion of Fyn prevented acute tolerance to ethanol inhibition of NMDA-mediated excitatory postsynaptic potentials in hippocampal slices [48]. Taken together, these data suggest that Fyn interacts with GluN2B of the NMDAR to mediate ethanol’s acute sedative effects.

Acute ethanol treatment significantly increases the tyrosine phosphorylation of GluN2B subunits in a Fyn-dependent and brain region-specific manner [42,48,53]. Acute ethanol treatment increased the tyrosine phosphorylation of GluN2B in the hippocampus but not in the cerebral cortex [42,53]. Furthermore, acute and chronic ethanol treatments did not alter Fyn or Csk levels in cortical neurons [53]. The differential effects of ethanol on GluN2B phosphorylation in the hippocampus and the cerebral cortex may be explained by the hippocampus-specific compartmentalization of Fyn to the NMDAR via the scaffolding protein RACK1 [42]. In the hippocampus, RACK1 forms a trimolecular complex with Fyn and GluN2B of the NMDAR; however, in the cerebral cortex, RACK1 associates only with Fyn and not GluN2B. In the hippocampus, acute administration of ethanol induces the dissociation of RACK1 from GluN2B and Fyn, thereby facilitating the Fyn-mediated phosphorylation of GluN2B, which enhances NMDA channel activity and counteracts the inhibitory effects of ethanol [42].

Fyn-mediated phosphorylation of GluN2B has been shown to prevent the behavioral changes associated with alcohol withdrawal [54]. A light–dark test was used to assess the withdrawal behaviors of wild-type and Fyn transgenic mice overexpressing a native or mutated, constitutively active Fyn. Alcohol-withdrawn wild-type mice show less exploration of the light compartment and fewer transitions between light and dark than alcohol-naïve mice, indicative of increased anxiety-like behaviors [54]. Alcohol withdrawal from Fyn transgenic mice, however, did not elicit the same behavioral patterns, suggesting that increasing Fyn expression can prevent alcohol withdrawal-induced anxiety-like behaviors. Treatment of Fyn transgenic mice with the GluN2B antagonist ifenprodil, however, restored the alcohol withdrawal-induced behaviors to wild-type levels, suggesting that Fyn modulates alcohol withdrawal through a GluN2B-dependent mechanism [54].

Fyn regulation of synaptic plasticity and learning

Evidence that Fyn modulates the threshold for LTP induction comes from studies using Fyn transgenic mice [15,55,56]. In Fyn null mice, LTP is impaired at weak and moderate intensities of presynaptic stimulation [15]. However, this blunting of LTP can be overcome by strong-intensity stimulation and by using a pairing protocol in Fyn null mice. To determine whether the disruption of LTP in Fyn null mice was caused directly by the absence of Fyn or indirectly by impairment of neuronal development, Fyn rescue mice were generated [55]. In these mice, LTP was restored to levels comparable to those in Fyn wild-type mice, suggesting that Fyn can directly modulate the threshold for LTP induction.

Further evidence that Fyn influences the threshold for LTP induction comes from transgenic mice overexpressing a constitutively active mutant of Fyn [56]. A weak theta-burst stimulation, which is subthreshold for LTP induction in control slices, was able to elicit LTP in active Fyn mutant slices; however, a stronger stimulation protocol induced similar levels of LTP in control and mutant mice. Fyn is proposed to reduce the threshold for LTP induction by increasing transmitter release or enhancing the postsynaptic response [56]. Reduced GABAergic inhibition in active Fyn mutant mice may also contribute to the altered LTP and synaptic transmission in these mice. Overexpression of wild-type Fyn, however, did not modify the threshold for LTP induction or basal synaptic transmission [56]. Together, these studies suggest that increasing Fyn activity shifts the modification threshold to the left, lowering the threshold for LTP induction, whereas Fyn deletion shifts the modification to the right, increasing the threshold for LTP induction (Fig. 3).

Figure 3.

 Fyn modulates the threshold for LTP induction. Increasing Fyn expression, e.g. by overexpressing a constitutively active mutant of Fyn, shifts the modification threshold to the left, thereby shifting the threshold for LTP to lower frequencies [56]. Conversely, genetic deletion of Fyn shifts the modification threshold to the right, increasing the threshold frequency for the induction of LTP [15,55]. LTD, long-term depression.

The Fyn-mediated alterations in synaptic plasticity are proposed to underlie the effects of Fyn on learning in different behavioral paradigms. In the Morris Water Maze task, Fyn knockouts showed significantly longer latencies to reach the hidden platform over time than wild-type mice [15]. The impaired spatial learning ability of Fyn null mice was also reflected in the transfer test, as they spent significantly less time in the trained quadrant than wild-type littermates.

Fyn-mediated phosphorylation of GluN2B at Tyr1472 is proposed to mediate the formation of contextual and auditory fear conditioning. Both short-term and long-term contextual fear memory are impaired in Fyn-deficient mice; conversely, overexpression of native or constitutively active Fyn does not influence contextual memory formation [57]. Furthermore, contextual fear conditioning results in transient activation of Fyn kinase activity and increases the phosphorylation of Tyr1472 on GluN2B in the dorsal hippocampus of wild-type but not Fyn null mice [58]. The extinction of contextual fear, however, results in significant downregulation of Fyn activity, as demonstrated by decreased phosphorylation of Tyr418 in Fyn-containing immunoprecipitates [59]. Thus, regulation of Fyn activity differentially modulates the acquisition and extinction of contextual fear memory. Phosphorylation of Tyr1472 also appears to be essential for the formation of auditory fear conditioning [18]. Mice expressing a knock-in mutation of the Tyr1472 site to phenylalanine (Y1472F) showed impaired freezing responses 1 h after conditioning, as well as impaired amygdaloid LTP and CAMKII signaling. Together, these studies suggest that Fyn-mediated alterations in synaptic plasticity underlie the effects of Fyn on different types of learning.

Conclusion

The Src family tyrosine kinase Fyn regulates the phosphorylation and trafficking of NMDARs, and acts as a point of convergence for many signaling pathways regulating NMDARs. Although Fyn has been shown to phosphorylate both GluN2A and GluN2B of the NMDAR, several lines of evidence suggest that Fyn selectively modulates the function of GluN2B-containing NMDARs. Furthermore, Fyn influences the modification threshold for synaptic plasticity and is thought to modulate learning through a GluN2B-dependent mechanism. The Fyn-mediated behavioral adaptations suggest that Fyn may represent a viable target in the treatment of alcohol dependence.

Ancillary