STEP activation by Gαq coupled GPCRs opposes Src regulation of NMDA receptors containing the GluN2A subunit

N-methyl-D-aspartate receptors (NMDARs) are necessary for the induction of synaptic plasticity and for the consolidation of learning and memory. NMDAR function is tightly regulated by functionally opposed families of kinases and phosphatases. Herein we show that the striatal-enriched protein tyrosine phosphatase (STEP) is recruited by Gαq-coupled receptors, including the M1 muscarinic acetylcholine receptor (M1R), and opposes the Src tyrosine kinase-mediated increase in the function of NMDARs composed of GluN2A. STEP activation by M1R stimulation requires IP3Rs and can depress NMDA-evoked currents with modest intracellular Ca2+ buffering. Src recruitment by M1R stimulation requires coincident NMDAR activation and can augment NMDA-evoked currents with high intracellular Ca2+ buffering. Our findings suggest that Src and STEP recruitment is contingent on differing intracellular Ca2+ dynamics that dictate whether NMDAR function is augmented or depressed following M1R stimulation.

these results present a conundrum in that mAChRs both enhance or depress NMDAR currents in hippocampal neurons.
We have resolved this conundrum by testing the hypothesis that Gα q GPCRs, including M1 mAChRs (M1Rs), recruit both Src kinase and the striatal-enriched protein tyrosine phosphatase (STEP) concurrently. The balance between kinase and phosphatase activity ultimately determines the degree of phosphorylation of NMDAR subunits and by extension their level of activity. We now report that a key factor in determining the balance between Src and STEP activation downstream of Gα q GPCR stimulation is the intracellular concentration of Ca 2+ . Low Ca 2+ (or high Ca 2+ buffering) favors the activity of Src and leads to a strong potentiation of NMDA-evoked currents by muscarine. This potentiation required Ca 2+ entry via NMDARs and was effected specifically through a Src-mediated increase of GluN2AR function, without participation of Fyn or GluN2BRs. Conversely, elevated intracellular Ca 2+ (or low Ca 2+ buffering) favours STEP activation and leads to a muscarine induced depression of NMDARs. This STEP-mediated depression of NMDA-evoked currents required inositol triphosphate receptor (IP3R) stimulation. Our results demonstrate that intracellular Ca 2+ dynamics play a key role in controlling Gα q-coupled GPCR-induced changes in NMDAR function through a differential recruitment of Src and STEP. The mechanisms described are likely to be important determinants of metaplasticity.

Results
NMDA receptor-mediated currents are constrained by a Gαq-coupled GPCR stimulated tyrosine phosphatase. We previously showed that muscarine potentiates NMDA-evoked currents in hippocampal CA1 neurons by recruiting Src 8 . In contrast, muscarine is reported to consistently inhibit NMDAR currents in hippocampal CA3 neurons 9,12 . When hippocampal CA3 neurons were treated with a non-specific tyrosine phosphatase inhibitor (pervanadate), the inhibition of NMDAR currents by muscarine was converted to a potentiation 12 . These finding suggest that M1Rs can either potentiate or inhibit NMDA-evoked currents through mechanisms involving Src or an unidentified protein tyrosine phosphatase (PTP). We therefore investigated whether PTPs limit the ability of Src kinase to enhance NMDAR currents downstream of Gα q-coupled GPCR stimulation in CA1 hippocampal neurons. Stimulation of mAChRs with the non-selective muscarinic agonist potentiates NMDAR currents (n = 7, 1.23 ± 0.08). The potentiation by CCh is increased when applied in the presence of the tyrosine phosphatase inhibitor orthovanadate (10 μ M; n = 7, 1.45 ± 0.10, P < 0.05 compared with CCh alone). In the absence of CCh treatment, NMDA currents remain stable (without CCh; n = 8, 1.01 ± 0.02). (b) Application of PACAP38 (1 nM; timing indicated by the shaded region) potentiates NMDAR currents (n = 6, 1.25 ± 0.05). The potentiation by PACAP38 is increased when applied in the presence of the tyrosine phosphatase inhibitor orthovanadate (10 μ M; n = 6, 1.47 ± 0.09, P < 0.05 compared with PACAP38 alone). (c) Summary bar graph of data in (a,b). *Indicates P < 0.05. Calibration bars: 3s; (a) without CCh 500 pA, CCh 300 pA, CCh + Ortho 400 pA; (b) PACAP38 300 pA, PACAP38 + Ortho 600 pA.
Scientific RepoRts | 6:36684 | DOI: 10.1038/srep36684 carbachol (CCh, 5 μ M) potentiated peak responses to rapid applications of NMDA (50 μ M, 0.5 μ M glycine; Fig. 1a,c, n = 7, P < 0.05 compared with baseline). When applied in the presence of the non-specific PTP inhibitor sodium orthovanadate (10 μ M), the potentiation of NMDAR currents by CCh was substantially enhanced (Fig. 1a,c, n = 7, P < 0.05 compared with CCh alone). Sodium orthovanadate did not influence peak NMDAR currents when applied alone ( Supplementary Fig. 1, n = 4, P > 0.05 compared with baseline). This suggests that a PTP is recruited downstream of mAChR stimulation to oppose the potentiation by Src kinase. To determine if PTPs are recruited downstream of another Gα q-coupled GPCR we examined the effects of PTP inhibition on NMDA-evoked currents during applications of the selective PAC 1 R agonist PACAP38 (1 nM). Sodium orthovanadate enhanced the potentiation of NMDARs by PACAP38 implying that PAC 1 Rs also activate an endogenous PTP that limits the extent of NMDA-evoked current potentiation (Fig. 1b,c, n = 6, P < 0.05 compared with PACAP38 alone).

STEP activation by M1 muscarinic receptors limits Src potentiation of GluN2A containing
NMDARs. In considering the identity of the PTP that opposes the enhancement of NMDAR currents in response to Gα q-coupled GPCR stimulation, one candidate is STEP 16 . Tyrosine dephosphorylation of the STEP substrate GluN2B depresses NMDAR function in hippocampal neurons 17 and promotes internalization of GluN2B-containing NMDARs (GluN2BRs) following exposure to β-amyloid [18][19][20] . To determine if STEP is involved, we supplemented the patch pipette solution with anti-STEP (1:400 dilution), a functional inhibitory antibody previously shown to selectively inhibit the activity of STEP 17,21 . The presence of anti-STEP dramatically enhanced the potentiation of these currents by CCh and by the mAchR-selective agonist muscarine (10 μ M; Fig. 2a,b,d, CCh + anti-STEP: n = 5, P < 0.05 compared with CCh alone; muscarine + anti-STEP: n = 6, P < 0.05 compared with muscarine alone). In contrast, no time-dependent change in the amplitude of NMDAR currents occurred when anti-STEP was applied alone ( Supplementary Fig. 1, n = 4, P > 0.05 compared with baseline).
Previously we showed that the dopamine 1 receptor (D1R), a Gα s-dependent GPCR, regulates NMDA responses in CA1 neurons through Fyn kinase and GluN2BRs but not Src and GluN2ARs 6 . Although Fyn and GluN2BRs are well recognized STEP substrates, D1R stimulation has been shown to suppress STEP activity through PKA-mediated phosphorylation. We therefore predicted that augmented GluN2BR function resulting from D1R-mediated Fyn activation would not be subject to the opposing influence of STEP. Confirming this prediction, the presence of anti-STEP had no effect on the potentiation of NMDA responses resulting from D1R stimulation by SKF81297 ( Supplementary Fig. 3, control: n = 6, P < 0.05 compared with baseline; anti-STEP: n = 6, P > 0.05 compared with SKF81297).

Intracellular Ca 2+ regulates the direction of change in NMDAR function imposed by STEP or Src.
Our results show that M1R stimulation leads to a Src-dependent potentiation of GluN2ARs in CA1 pyramidal neurons in spite of a paradoxical stimulation of STEP. In recordings from isolated CA1 neurons, we employed a relatively high concentration of the slow Ca 2+ chelator EGTA (11 mM EGTA). In contrast, Jo and colleagues 2010 employed a much lower concentration of EGTA (0.5 mM) in their patch recordings and demonstrated that CCh induced an inhibition of synaptic NMDARs. This suggested the possibility that the balance between effects of Src and STEP on NMDAR function may be influenced by the degree of Ca 2+ buffering. Therefore, we performed a series of recordings in which we varied the amount of Ca 2+ buffering in the patch solutions. When EGTA was reduced to 1 mM, applications of muscarine failed to potentiate peak NMDARs ( Fig. 4a,d, n = 5, P > 0.05 compared with baseline). However, when Src(40-58) was included in the patch pipette to block Src, muscarine induced a robust inhibition of NMDARs ( Fig. 4a,d, n = 5, P < 0.05 compared with muscarine alone). Conversely, when anti-STEP was administered to the cell interior, muscarine now potentiated NMDAR currents (Fig. 4a,d, n = 6, P < 0.05 compared with muscarine alone). Potentiation by anti-STEP was blocked by Src(40-58) (Fig. 4a,d, n = 6, P > 0.05 compared with baseline). Given these results, we further reduced the concentration of EGTA to 0.1 mM and under these recording conditions muscarine exclusively inhibited peak NMDARs (Fig. 4b,d, n = 5, P < 0.05 compared with baseline). Importantly, this inhibition was blocked by including anti-STEP in the patch pipette ( Fig. 4b,d, n = 7, P > 0.05 compared with baseline) and was absent in neurons isolated from STEP −/− mice (Fig. 4c,d, n = 7, P > 0.05 compared with baseline). These results demonstrate that Ca 2+ buffer capacity in neurons determines whether the M1R stimulation predominantly activates STEP or Src, leading to inhibition or potentiation, respectively.
We next sought to determine the source of the Ca 2+ transients that regulate Src and STEP signalling, albeit at different intracellular Ca 2+ concentrations. Two primary sources likely contribute to Ca 2+ elevations in our recordings; entry via NMDARs and release from intracellular stores by M1R-mediated activation of IP 3 Rs. In previous experiments, muscarine and NMDA were simultaneously applied to isolated neurons. To determine the contribution of Ca 2+ influx via NMDA receptors in this paradigm, we stopped NMDA exposure during muscarine application and waited another 5 min to thoroughly wash out any residual muscarine before re-commencing applications of NMDA. With these conditions the muscarine potentiation of NMDAR currents was absent ( Fig. 5a,d, control: n = 5, P > 0.05 compared with baseline). This implies that the potentiation by muscarine is dependent in part upon the entry of Ca 2+ via NMDARs. When repeated in the presence of anti-STEP, a substantial potentiation of NMDAR currents was now observed (Fig. 5a,d, anti-STEP: n = 7, P < 0.05 compared with baseline). As the application of anti-STEP alone does not potentiate NMDAR currents ( Supplementary Fig. 1, n = 4, P > 0.05 compared with baseline), this suggested to us that modest activation of Src likely occurred when muscarine was applied in the absence of concurrent NMDAR stimulation. This was confirmed by our finding that anti-STEP facilitated potentiation of NMDARs by muscarine (without NMDAR co-stimulation) could be prevented by Src40-58 (Fig. 5b,d, anti-STEP + Src(40-58): n = 6, P > 0.05 compared with baseline), but not by Fyn39-57 (Fig. 5b,d, anti-STEP + Fyn(39-57): n = 6, P < 0.05 compared with baseline). Given that Ca 2+ influx via NMDARs is important for robust Src recruitment but not for STEP, we questioned whether IP 3 R-regulated Ca 2+ stores contributed to STEP recruitment following M1R stimulation. In the presence of xestospongin C (10 μ M; an IP 3 R inhibitor), muscarine applied without concurrent NMDAR stimulation now potentiated NMDA responses (Fig. 5c,d, xestos: n = 5, P < 0.05 compared with baseline). Potentiation of NMDARs by M1R stimulation in the presence of xestospongin C was not influenced by intracellular anti-STEP application, suggesting that block of IP 3 Rs prevents STEP recruitment (Fig. 5c,d, xestos + anti-STEP: n = 5, P > 0.05 compared with group without anti-STEP). This suggests that the downstream effect of mAchR stimulation on GluN2ARs is Ca 2+ source-specific; entry through NMDARs favors Src activation, whilst release from intracellular stores favors STEP activation.

The activity of STEP and Src is differentially regulated by the concentration and source of intracellular
Ca 2+ . Our findings show that the direction of change in GluN2AR function resulting from M1R stimulation is determined by intracellular Ca 2+ dynamics, presumably by dictating the relative strength of Src vs STEP activity. To confirm the parallel stimulation of Src and STEP activity by M1Rs and consequent change in GluN2A tyrosine phosphorylation, we used Western blotting to examine: (1) Phosphorylation of STEP at Ser221 within the substrate binding domain 28 . Phosphorylation of this site sterically prevents the association of STEP with substrates, resulting in an increase in Tyr phosphorylation of these substrates [29][30][31][32] . (2) Phosphorylation of Src at Tyr416. Residing within the activation loop, phosphorylation at this site activates Src kinase. (3) Phosphorylation of GluN2AR.
We first treated primary hippocampal neurons with muscarine (10 μ M) in the presence of NMDA (50 μ M) in ECS for 10 min (Mus/NMDA), without exogenous Ca 2+ buffers added. Mus/NMDA treatment resulted in decreased phosphorylation of STEP 61 at Ser221 (Fig. 6a, Control and Veh lanes, n = 5, P < 0.01), as revealed by the reduced upper band in the doublets 29 . To confirm this finding we probed using an antibody recognizing STEP 61 only when not phosphorylated at Ser221. Consistently, we found increased non-phosphorylated STEP 61 (Fig. 6a, Control and Veh lanes, n = 5, P < 0.01), suggesting that Mus/NMDA treatment led to the activation of STEP 61 . In contrast, no change in the phosphorylation of Src at Tyr416 was observed after Mus/NMDA treatment (Fig. 6b, Control and Veh lanes, n = 5, P > 0.05). To investigate changes in GluN2A upon Mus/NMDA treatment, we immunoprecipitated GluN2A followed by probing with anti-pan-Tyr antibody. We found decreased Tyr phosphorylation of GluN2A (Fig. 6b, Control and Veh lanes, n = 5, P < 0.05), in agreement with the inhibition of NMDAR response. The effects of muscarine were prevented when applied in the presence of the M1R antagonist pirenzepine (Fig. 6a,b, Prz lanes, n = 5, P < 0.05, P < 0.01, respectively).
To demonstrate the role of Src in mediating the increase in GluN2A Tyr phosphorylation when STEP is inhibited by TC-2153 or xestospongin C in neurons pre-treated with 1 mM EGTA/AM, we repeated Mus/  anti-non-phospho-STEP 61 antibodies. Phospho-and non-phospho-STEP 61 levels were normalized to total STEP 61 protein levels and then to β -actin as loading control. (b) GluN2A and Src were immunoprecipitated from treated lysates using specific antibodies, followed by probing with anti-Tyr, anti-pY416 Src and panprotein antibodies, respectively. Phospho-protein levels were normalized to pan-protein levels. (c) In addition to the inhibitors used in (a), PP2 (10 μ M) and PP3 (10 μ M) were added to ECS for stimulation. Phospho-and non-phospho-STEP 61 levels were assessed. (d) After treatments, phosphorylation of GluN2A and Src were measured on immunoprecipitated samples. Phospho-protein levels were normalized to pan-protein levels. All data were expressed as mean ± SEM. Statistical significance was determined using one-way ANOVA with post hoc Tukey test (*P < 0.05, **P < 0.01, ***P < 0.001, n = 5).
Previous studies have shown that Fyn, but not Src, is a STEP substrate 32,34 . Evidence also suggests an interaction between STEP 61 and GluN2A in heterologous cell lines 35 . Thus we set out to test whether there is a direct link between STEP 61 and GluN2A. We immunoprecipitated STEP using a well-validated anti-STEP (23E5) antibody and looked for co-immunoprecipitation of interacting proteins. We confirmed previous findings that GluN2A, but not Src, is associated with STEP 61 in hippocampal lysates (Fig. 7c, first 2 lanes, n = 3). As expected, known substrates GluN2B and Fyn also bind to STEP 61 (Fig. 7c, first 2 lanes, n = 3). We also used STEP KO hippocampal lysates to exclude possible non-specific binding (Fig. 7c, last 2 lanes, n = 3). To confirm the interaction between STEP 61 and GluN2A we used an in vitro binding assay 31,34 . A substrate-trapping mutant of the longer isoform (STEP 61 ) and the shorter isoform (STEP 46 ) were immobilized on sepharose matrix, and incubated with mouse hippocampal lysates. We found co-purification of GluN2A with STEP 61 but not STEP 46 , suggesting the extra N-terminus in STEP 61 is required for the binding (Fig. 7d). In agreement with previous findings 31,34 GluN2B and Fyn also showed higher affinity with STEP 61 (Fig. 7d). We didn't find interaction of Src and STEP 61 under these conditions (Fig. 7d). These findings confirm that not only GluN2B and Fyn but also GluN2A are substrates of STEP.

Discussion
Here we show that stimulation of Gα q-coupled receptors for muscarine (M1R) or pituitary adenylate cyclase activating peptide (PAC 1 R) potentiate NMDAR-mediated currents. The enhancement of NMDA currents was mediated by Src acting specifically upon NMDARs composed of the GluN2A receptor subunit. A major finding of our study is that the Src-mediated increase in GluN2AR function is counterbalanced by STEP activated concurrently by Gα q receptors. Under conditions that favor Src activation, STEP limits the potentiation of GluN2ARs by Src. Conversely, under conditions that favor STEP activation, STEP depresses the function of GluN2ARs (Fig. 8). Importantly, we show that potentiation of NMDAR currents by Src and inhibition by STEP downstream of M1Rs have discreet Ca 2+ requirements; Src requires entry of Ca 2+ via NMDARs whereas STEP requires release of Ca 2+ from IP 3 R-sensitive stores. The balance of Src and STEP activation, and consequent impact on GluN2AR function, is dictated by the dynamic balance between source specific intracellular Ca 2+ elevations. More conclusively, Figure 7. STEP binds to and dephosphorylates GluN2A. Synaptosomal fractions from WT and STEP KO mouse hippocampus were used for immunoprecipitation with anti-GluN2A (a) or anti-phophos-Tyr (b), followed by immunoblotting with anti-phopho-Tyr or anti-GluN2A, anti-GluN2B and anti-Src, respectively. Phospho-protein levels were normalized to total proteins, repectively. All data were expressed as mean ± SEM. Statistical significance was determined using two-tailed Student's t test (*P < 0.05, **P < 0.01, n = 6). (c) STEP was immunoprecipitated from WT and STEP KO hippocampal lysates using anti-STEP (23E5) antibody. Potential interacting proteins were verified using selective antibodies indicated in the figure. Representative blots were shown from three independent replicates. (d) GST-STEP fusion proteins were bound to glutathionesepharose 4B beads and incubated with hippocampal lysates. Co-purified proteins were verified using selective antibodies indicated in the figure. Representative blots were shown from three independent replicates.
we monitored changes in the phosphorylation of STEP and Src at key sites that regulate their enzyme activity and show that co-stimulation of M1R and NMDARs can cause both STEP and Src activation. Consistent with electrophysiological findings, the direction of change in GluN2A tyrosine phosphorylation was determine by the relative strength of Src or STEP activation; conditions that caused increased NMDAR current with M1R stimulation favored Src activation and increased GluN2A tyrosine phosphorylation and, conversely, conditions that decreased NMDAR currents with M1R stimulation favored STEP activation and decreased GluN2A tyrosine phosphorylation.
In contrast, when Gα s-coupled D1Rs were stimulated the resulting Fyn-dependent enhancement of GluN2BRs was not influenced by STEP. We attribute this to several factors. Firstly, past work has shown that Gα s-coupled D1Rs inhibit STEP via PKA-mediated phosphorylation at Ser221 28,36 within the kinase-interacting motif (KIM) domain important for STEP substrate recognition. Secondly, our own results demonstrate that basal NMDAR function is not influenced by STEP, consistent with evidence that STEP activity is low under resting conditions 29,37 . Thus, the parallel recruitment of Fyn, in concert with suppression of low basal STEP activity, accounts for the observed D1R-mediated enhancement of GluN2BR function that is unopposed by STEP ( Supplementary Fig. 3). Accordingly, key to reconciling divergent NMDAR subunit-and SFK-selective actions of STEP is to consider the activation context. For example, β -amyloid provokes increased STEP levels in Alzheimer's disease through inhibition of the proteasome that normally degrades STEP, resulting in GluN2B internalization as a consequence of STEP-mediated dephosphorylation of Tyr1472 [18][19][20] . Our findings suggest an additional and previously overlooked context in which STEP is recruited by Gα q receptors (e.g. PAC 1 R and M1R). Unlike the D1R pathway in which Fyn signalling is augmented through inhibition of STEP, pathways downstream of PAC 1 R and M1R initiate an increase in the activity of both Src and STEP. In this way STEP provides feedback inhibition that constrains enhancement of NMDAR function through concurrent Src activation targeting GluN2ARs.
The concurrent stimulation of Src and STEP by Gα q receptors allows for bidirectional modulation of NMDAR function. We find that intracellular Ca 2+ dynamics and the source contributing to intracellular Ca 2+ elevations determines the direction of change in GluN2AR function. This was evident from M1R stimulation experiments in which we varied Ca 2+ buffering by EGTA allowing the extent of intracellular Ca 2+ elevations to be experimentally determined. Whereas robust Src-dependent potentiation is observed when large elevations of intracellular Ca 2+ are prevented (11 mM EGTA), M1R stimulation fails to potentiate the NMDA responses when modest intracellular Ca 2+ concentrations are achieved (1 mM EGTA; Fig. 4a,d). Nevertheless, when modest intracellular Ca 2+ elevations are permitted, treatment with anti-STEP enables muscarine to now potentiate the NMDA response, whereas treatment with Src(40-58) enables muscarine to now depress the NMDA response. When large intracellular Ca 2+ elevations are permitted (0.1 mM EGTA; Fig. 4b,d), muscarine now depresses NMDA responses and this depression can be prevented by treatment with anti-STEP. In considering the functional outcome for GluN2ARs, these results indicate that the balance between Src-mediated potentiation and STEP-mediated depression is determined by intracellular Ca 2+ levels. This model is precisely supported by our biochemical findings in which we varied intracellular Ca 2+ buffering using cell-permeable EGTA/AM. Mus/NMDA treatment depresses GluN2A tyrosine phosphorylation in neurons pre-treated with 0.1 mM EGTA/AM, but not in neurons pre-treated with 1 mM EGTA. Of note, the difference between these two conditions resides in the sensitivity of Src activity to the concentration of applied EGTA/AM. In the presence of 1 mM EGTA/AM, Mus/NMDA increased Src Tyr416 phosphorylation whereas no change in Src Tyr416 was observed when Mus/NMDA was applied in the presence of 0.1 mM EGTA/AM. As increasing the concentration of EGTA/AM from 0.1 to 1 mM is anticipated to reduce the rise in intracellular Ca 2+ achieved during Mus/NMDA treatment, these findings suggest that a Ca 2+ -dependent tyrosine phosphatase may limit Src activity. Although the phosphatase identity remains to be determined, collectively our biochemical findings rule out a role for STEP given convincing evidence that Src is not a STEP substrate.
A noteworthy aspect of our findings is the observed requirement of coincident mAchR and GluN2AR stimulation for Src-mediated potentiation of GluN2ARs. A similar requirement was previously reported for the mGluR5-induced enhancement of NMDAR function 38 . This was demonstrated through experiments in which muscarine was applied in the absence of co-incident NMDAR stimulation. Here, muscarine did not potentiate NMDA responses but could do so in the presence of anti-STEP. This suggests that STEP was dominant under these conditions. In considering the Ca 2+ source contributing to STEP recruitment in the absence of NMDAR stimulation, a likely source was through mobilization of internal Ca 2+ stores by IP 3 Rs. This was confirmed by demonstrating that muscarine, applied without concurrent NMDAR stimulation, could potentiate NMDAR currents in the presence of the IP 3 R blocker xestospongin C. Confirming that STEP was inactive, anti-STEP treatment was ineffective when applied with IP 3 Rs blocked. Our findings suggest a previously overlooked aspect of STEP function, namely the feedback regulation of GluN2AR function augmented by Src signalling downstream of Gα q-coupled GPCRs. Although we show that STEP activation reduces GluN2A tyrosine phosphorylation, the mechanism by which this biochemical change inhibits GluN2AR function remains to be determined. Additional experiments (Supplementary Fig. 4) suggest that it does not involve changes in tyrosine phosphorylation of GluN2A Tyr1325. Speculatively, a mechanism can be proposed on the basis of past work demonstrating that dephosphorylation of GluN2A Tyr842 depresses NMDA currents through AP2 mediated receptor internalization 39 . Such a mechanism would be analogous to that by which STEP supresses the function of GluN2BRs through dephosphorylation of GluN2B at Tyr1472 and receptor internalization 18,20 .
The present findings and past work, which has demonstrated the important role of STEP in offsetting Fyn-mediated phosphorylation of GluN2B subunits, provides a broader context in which STEP can be recruited to limit NMDAR function augmented under different physiological and pathophysiological conditions by SFK signalling. Our study used fast NMDA applications to acutely isolated hippocampal neurons as a means of assessing the consequence of Src and STEP recruitment on NMDAR function. The NMDA responses were therefore likely mediated primarily by extrasynaptic NMDARs. Nevertheless, all of the signalling components examined regulate synaptic NMDARs 17 . Therefore, contingent on the basal phosphorylation status of NMDAR subunits, Src and/or STEP, our findings suggest that stimulation of mAchR (or other Gα q-coupled GPCRs) at quiescent glutamate synapses will favor STEP recruitment leading to reduced function of GluN2ARs. Conversely, Src activation will be favored at active glutamate synapses leading to augmented GluN2AR function. Additionally, spontaneous synaptic events promoting Ca 2+ entry via NMDARs may participate in maintaining basal Src activity necessary for homeostatic maintenance of synaptic NMDAR functions. Considered in light of past evidence demonstrating that D1Rs augment GluN2BR function via Fyn stimulation and STEP inhibition, these findings suggest that STEP, acting in concert with SFKs, dynamically orchestrates NMDAR subunit-dependent signalling downstream of GPCRs. The functional interplay between SFKs and STEP can be finely regulated at individual synapses based on the level of activity and whether such activity converges with input from transmitter systems acting upon their cognate GPCRs. This functional interplay is likely to have important implications for regulating the direction of plasticity and thus metaplasticity at excitatory synapses.

Methods
Cell isolation and whole-cell recordings. Hippocampal CA1 neurons were isolated from Wistar rats (14-22 days old) or from STEP KO mice and WT littermates (21-28 days old) using previously described procedures 6,40 . All animal experimentations were conducted in accordance with standards established by the Canadian Council on Animal Care (CCAC) and approved by the Animal Care Committee at the University of Western Ontario. The extracellular solution (pH 7.4, osmolality between 315 and 325 mOsm) consisted of the following (in mM): 140 NaCl, 1.3 CaCl 2 , 5 KCl, 25 HEPES, 33 glucose, 0.0005 glycine and 0.0005 tetrodotoxin. Recording electrodes (resistance between 3 and 5 MΩ ) were constructed from borosilicate glass (1.5 mm in diameter, World Precision Instruments, Sarasota, FL) using a two-stage puller (PP83, Narishige, Tokyo, Japan) and were filled with intracellular solution (pH 7.2, osmolality between 290 and 300 mOsm ) containing the following (in mM): 140 CsF, 1 CaCl 2 , 2 MgCl 2 , 10 HEPES, 2 tetraethylammonium, and 2 K 2 ATP. Unless otherwise indicated, the EGTA concentration used in intracellular solution is 11 mM EGTA. Where indicated, 1 mM or 0.1 mM EGTA was used instead. For some experiments, the intracellular solution was supplemented with Src(40-58), Fyn(39-57), anti-STEP, orthovanadate or xestospongin C, with all concentrations indicated in the text. Recordings were conducted at room temperature between 20-22 °C. After formation of the whole-cell configuration, the neurons were voltage clamped at − 60 mV and lifted into the stream of solution supplied by a computer-controlled multi-barreled fast perfusion system (SF-77 B, Warner Instrument Corporation). The solution exchange time was 3-5 ms. NMDA (50 μ M) was applied 1/60 sec for 3 sec. To monitor access resistance, a voltage step of − 10 mV was applied before each application of NMDA (50 μ M). When series resistance increased to > 20 MΩ , the cell was discarded. Currents were recorded using an Axopatch 1D amplifier. Data were filtered at 2 kHz and digitized at 10 kHz using Clampex software.
Transgenic mice and sample preparation. The STEP knockout (KO) mice and wild-type (WT) littermates were generated and maintained at Yale University as described previously 41 . All procedures were performed in accordance to the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Yale University. Mouse (both male and female, 3-6 months old) hippocampi were dissected out and homogenized in Dounce tissue grinders (Wheaton, Millville, NJ) in ice-cold TEVP buffer (10 mM Tris pH 7.4, 1 mM EDTA, 1 mM EGTA, 1 mM Na 3 VO 4 , 5 mM NaF, 320 mM