Increased Activity of Src Homology 2 Domain Containing Phosphotyrosine Phosphatase 2 (Shp2) Regulates Activity-dependent AMPA Receptor Trafficking*

Long term synaptic plasticity, such as long term potentiation (LTP), has been widely accepted as a cellular mechanism underlying memory. Recently, it has been unraveled that Shp2 plays a role in synaptic plasticity and memory in Drosophila and mice, revealing significant and conserved effects of Shp2 in cognitive function. However, the exact mechanism underlying this function of Shp2 in synaptic plasticity and memory still remains elusive. Here, we examine the regulation of Shp2 in hippocampal LTP and contextual fear conditioning. We find that Shp2 is rapidly recruited into spines after LTP induction. Furthermore, the phosphorylation level of Shp2 at Tyr-542 is elevated after LTP stimuli either in cultured hippocampal neurons or acute slices. Notably, contextual fear conditioning also regulates the phosphorylation level of Shp2 at Tyr-542, suggesting fine-tuned regulation of Shp2 in LTP and memory formation. By using a Shp2-specific inhibitor and adeno-associated virus-Cre mediated Shp2 knock-out in cultured neurons, we provide evidence that the phosphatase activity of Shp2 is critical for activity-dependent AMPA receptor surface trafficking. Collectively, our results have revealed a regulatory mechanism of Shp2 underlying LTP and memory, broadening our understanding of Shp2 in cognitive function.

Shp2, encoded by the gene PTPN11, is a ubiquitously expressed phosphatase possessing a structure containing two tandem SH2 2 (Src homology 2) domains flanked on its N terminus, followed by a protein-tyrosine phosphatase domain and a C-terminal hydrophilic tail containing phosphorylation sites (1,2). The distinctive structural feature of Shp2 confers it playing two functional roles in signaling transduction. One function of Shp2 is conducting as an adaptor protein to assemble the multiple protein complex by its SH2 domain associating other phosphorylated tyrosine-containing proteins (2), the other is acting as a phosphatase to dephosphorylate its substrate upon certain stimuli. Tyrosine 542 of Shp2 at its hydrophilic tail is identified as a major phosphorylation site in response to growth factor stimuli (3). Previous studies argued that phosphorylation of Tyr-542 correlated with its phosphatase activity and promoted the activity of Shp2 by relieving the basal inhibition of protein-tyrosine phosphatase via disruption of intramolecular interaction of its tail with its N-SH2 domain (1,(3)(4)(5).
Long term synaptic plasticity, i.e. long term potentiation (LTP), and its opposing effect, long term depression are widely accepted as a cellular mechanism of memory (6). At glutamatergic synapses, LTP is expressed by a synapse-specific increase in the number of AMPA receptors at postsynaptic sites (7). In recent years, several studies unveiled the significance of Shp2 in cognitive function and long term memory. Genetic manipulation, e.g. knock-in mice with Noonan syndrome (NS) mutation of Shp2, and ablation of Shp2, no matter in Drosophila or rodents, both reveal that Shp2 is profoundly involved in synaptic plasticity and memory (8 -10). However, genetic approaches in vivo can hardly avoid some unpredicted defects as well as side effects that were exemplified by excess weight displayed in the Shp2 knock-out (KO) mice (8) and other similar NS patient defects associated with NS mouse models derived from the knock-in Shp2 mutation (9). Although previous studies provide convincing evidences that proper Shp2 function is requisite of normal cognitive function and memory, the exact mechanism of Shp2 underlying synaptic plasticity and memory yet remains to be resolved in detail. Besides, the regulation of Shp2 in response to synaptic plasticity, such as its phosphatase activity, which is represented by the phosphorylation level of Tyr-542, has thus far been unknown. Importantly, advancement of understanding in regulation of Shp2 activity in synaptic plasticity and memory should throw light on the development of therapeutic approaches to mental diseases associated with dysfunctional Shp2. Therefore, we determined to examine the regulation of Shp2 in synaptic plasticity, especially in LTP.
In the present study, we investigated the regulation of Shp2 activity in synaptic plasticity. We found up-regulation of Shp2 phosphatase activity during LTP. Furthermore, we demonstrated that Shp2 was required for synaptic delivery of AMPA receptors using both pharmacological and genetic procedures. Combined with previous studies on loss of function of Shp2, we suggest that the activity of Shp2 is delicately regulated in synaptic plasticity. Additionally, at least to some extent, the phos-phatase activity of Shp2 is required for AMPA receptor delivery in LTP.

Results
Expression Profile of Shp2 and Phosphorylated Shp2 at Tyr-542 in Brain-Shp2 is highly expressed in the rat brain and widely distributed throughout numerous brain regions (11). Distinct phosphorylation levels of Tyr-542, which exists in several brain regions may imply a region-specific function. So we first set out to examine the status of phosphorylated Shp2 at Tyr-542 (named Tyr(P)-542) in vivo. Medial prefrontal cortex, hippocampus, thalamus, cerebellum, amygdala, pons, and striatum of adult mouse brain were collected and homogenized to analyze the expression level of Tyr(P)-542 in these brain regions. We found that total Shp2 was uniformly expressed throughout the brain with a phosphorylation level of Tyr-542 (evaluated as a ratio to total Shp2) exhibiting relatively much higher in the pons (Fig. 1, A and B). To ensure the accuracy of regions isolated in our experiments, the expression pattern of GluN2B was also assessed. As expected, it was in line with a previous study (Fig. 1, A and B) (12).
Shp2 was reported to exist in the synaptic plasma membrane (11), so we determined to explore the subcellular location of Shp2 and p-Shp2 at Tyr-542 with a successfully separated membrane fraction (P2), a Triton-insoluble postsynaptic density (also considered as PSD) fraction, and a non-PSD fraction of hippocampus. We found that Shp2 was enriched in the non-PSD fraction, whereas the Tyr(P)-542 level exhibited relatively much higher in the PSD fraction ( Fig. 1, C and D). Moreover, a confocal image revealed that Shp2 (0.70 Ϯ 0.012, n ϭ 36) and p-Shp2 at Tyr-542 (0.67 Ϯ 0.0089, n ϭ 36) was robustly colocalized with PSD95, a scaffolding protein as a postsynaptic marker, in cultured hippocampal neurons (Fig. 1, E and F). Three-dimensional rendering of Shp2 and p-Shp2 at Tyr-542 colocalized with PSD95 with Z-stack images deprived from confocal microscopy demonstrated that both Shp2 and Tyr(P)-542 were localized adjacent to postsynaptic sites (Fig. 1G). To further confirm this result, a super-resolution image technology called stochastic optical reconstruction microscopy (STORM) was employed and subsequently we found Shp2 partially colocalized with PSD95 in dendrites (Fig. 1H). This unique positioning of Shp2 and p-Shp2 at Tyr-542 suggests that Shp2 was situated at the postsynaptic site and possibly plays a crucial role in the synaptic signaling transduction in dendritic spines.
Gain of function of Shp2 has been implicated in relationship to Noonan syndrome, which is a systemic developmental disorder (13). Thus, we sought to examine the developmental profile of Shp2 and p-Shp2 at Tyr-542 (3,5). We evaluated the expression of GluN2A, Shp2, and p-Shp2 at the Tyr-542 level during development by immunoblotting extracts of hippocampus from mice at different ages. As a previous study showing a substantial increase of GluN2A expression level during development (14), our development profile of GluN2A was in accordance with that study. Remarkably, we found that the Tyr(P)-542 level was elevated with the total expression level of Shp2 being reduced during development (Fig. 1, I and J). The differences in expression levels between Shp2 and Tyr(P)-542 during development suggest the activity of Shp2 is finely tuned in neuron development.
Recruitment of Shp2 into Postsynaptic Density during LTP Induction-Proper function of Shp2 has been shown to be dispensable for normal LTP and memory (8 -10). Both theoretical and experimental work suggests that the movement of molecules and organelles between the spine and the adjacent dendritic shaft contributes to synaptic activity (15). Therefore, examining the translocation of Shp2 in different subcellular fractions during the activity may reveal its functional role. We analyzed the translocation of endogenous Shp2 and GluA1 by subcellular fractionation of the CA1 region of a hippocampal slice after high-frequency (100 Hz/1 s) stimulation (HFS). The field EPSP during the last 5 min of recording clearly demonstrated that HFS elicited a substantial increase in slope of EPSP. We then determined that Shp2 was significantly increased in the PSD fraction after LTP induction at 10 min (2.91 Ϯ 0.34, p Ͻ 0.005, two-tailed Student's t test, n ϭ 4) without change in the P2 fraction (1.10 Ϯ 0.14, n ϭ 4) (Fig. 2, B and C). We also observed accumulation of GluA1 of the AMPA receptor (1.76 Ϯ 0.16, p Ͻ 0.01, two-tailed Student's t test, n ϭ 4) in the PSD fraction after LTP induction, which was consistent with our previous report (16).
To reinforce this conclusion, we investigated colocalization between Shp2 and PSD95 in dendrites with and without cLTP stimuli using a standard protocol (17,18) that selectively activates synaptic NMDA receptors. Statistical analysis of the colocalization illustrated that the proportion of Shp2 clusters (without cLTP treatment, 0.69 Ϯ 0.0056, n ϭ 15 neurons; cLTP treatment, 0.73 Ϯ 0.0054, n ϭ 15 neurons; two-tailed Student's t test, p Ͻ 0.005) that were merged with PSD95 clusters was markedly increased in the cLTP group (Fig. 2, D and E). The results from biochemistry and confocal images mentioned above suggest increased incorporation of Shp2 into excitatory postsynaptic sites after LTP induction.
The Activity of Shp2 Is Up-regulated during LTP Stimulation and Contextual Fear Conditioning-Given enrichment of Shp2 in the postsynaptic fraction after LTP induction, we postulated that the activity of Shp2 was strictly regulated and thus it played an important role in synaptic activity. Therefore, we treated cultured hippocampal neurons with bicuculline, which is an antagonist of GABA receptor to increase basal excitatory synaptic activity of neurons. In our experiment, short term exposure to bicuculline (25 M; 5-15 min) elevated the Tyr(P)-542 level (5 min treatment, 2.71 Ϯ 0.37, p Ͻ 0.01; 10 min treatment, 3.27 Ϯ 0.27, p Ͻ 0.005; 15 min treatment, 2.90 Ϯ 0.49, p Ͻ 0.01, versus Ctl, one-way ANOVA, n ϭ 6) in neurons without affecting the total Shp2 level, indicating that the Tyr-542 level of Shp2 can be regulated rather quickly by neuronal activity (Fig.  3, A and B).
To further explore the regulation of Shp2 in synaptic activity, we examined the phosphorylation level of Shp2 at Tyr-542 after HFS in a acute slice preparation. We found increased average responses during the last 5 min of the recording indicating a successful LTP (Fig. 2C). We collected CA1 mini slices from male mice (5-6 weeks) after HFS, the lysate was subjected to immunoblot analysis. As shown in Fig. 3, D and E, phosphorylation of GluA1 Ser-845 was elevated during the acute phase (10 min, 2.43 Ϯ 0.11, p Ͻ 0.005, LTP versus Ctl, one-way ANOVA, n ϭ 4) and maintain phase (60 min, 2.16 Ϯ 0.14, p Ͻ 0.005, LTP versus Ctl, one-way ANOVA, n ϭ 4), indicating a successful LTP induction. Likewise, the Tyr(P)-542 level was substantially increased at both time points (10 min, 1.96 Ϯ 0.19, p Ͻ 0.05; 60 min 2.25 Ϯ 0.31, p Ͻ 0.01, LTP versus Ctl, one-way ANOVA, n ϭ 4), suggesting the Tyr-542 level was also regulated during LTP (Fig. 3, D and E). We further verified our findings in cultured hippocampal neurons and also found a remarkable increase of the Tyr(P)-542 level at 10 min after cLTP induction without altering the total Shp2 level (Tyr(P)-542 level, 2.51 Ϯ 0.40, p Ͻ 0.01, cLTP versus Ctl, one-way ANOVA, n ϭ 4) (Fig. 3, F and G). Considering that the NMDA receptor plays a critical role in LTP in the hippocampus, we wondered whether the elevation level of Tyr(P)-542 was NMDA receptor dependent. Therefore, we co-incubated APV, which is an antagonist of NMDA receptor, in the neurons during glycine-induced cLTP. Notably, an increase of the level of Tyr(P)-542 was diminished when APV was present during cLTP (1.15 Ϯ 0.10, p Ͻ 0.01, APV ϩ cLTP versus cLTP, one-way ANOVA, n ϭ 4), suggesting that the level change of Tyr(P)-542 in cLTP is NMDA receptor dependent (Fig. 3

, F and G).
We had shown that the activity of Shp2 was tightly regulated during LTP. Given that LTP is the cellular mechanism underlying memory, we sought to investigate whether the Tyr(P)-542 level was changed in memory formation in vivo. We utilized a well established hippocampus-related memory test, which is contextual fear conditioning. After a training trial containing three times of foot-shocks at 30-s intervals, freezing time within the last 2 min in chambers were recorded and shown to be significantly increased after three foot-shocks (Ctl, 3.39 Ϯ 1.20 s, n ϭ 4; CFC 10 min, 92.51 Ϯ 10.78 s, p Ͻ 0.005, one-way ANOVA, n ϭ 4; CFC 60 min, 106.20 Ϯ 11.62 s, p Ͻ 0.001, one-way ANOVA, n ϭ 4) indicating a successful fear learning (Fig. 3H). The results showed that the Tyr(P)-542 level was increased at 10 min (1.39 Ϯ 0.054, p Ͻ 0.05, 10 min versus Ctl, one-way ANOVA, n ϭ 4) and more remarkably enhanced at 60 min (1.61 Ϯ 0.14, p Ͻ 0.01, 60 min versus Ctl, one-way ANOVA, n ϭ 4) after contextual fear conditioning, without affecting the total Shp2 level. As a positive control, both Tyr-1472 (10 min, 2.34 Ϯ 0.14, p Ͻ 0.01; 60 min, 2.19 Ϯ 0.31, p Ͻ 0.01, versus Ctl, one-way ANOVA, n ϭ 4) (19) and Tyr-1070 levels (10 min, 1.66 Ϯ 0.17, p Ͻ 0.01; 60 min, 1.66 Ϯ 0.11, p Ͻ 0.01, versus Ctl, one-way ANOVA, n ϭ 4) (12) of GluN2B were increased in this process, indicating a successful contextual fear conditioning in our paradigm (Fig. 3, I and J). Taken together, the results from cultured neurons, acute slices, and mice convergently point  toward to the suggestion that the activity of Shp2 is tightly regulated in LTP and memory formation.
To confirm the result, surface staining of GluA1 was assessed with and without cLTP induction in the groups. We found a similar result in the AAV-GFP group, showing that surface GluA1 dramatically increased after cLTP induction (without cLTP 1.00 Ϯ 0.027, n ϭ 15; with cLTP 1.24 Ϯ 0.037, n ϭ 19, one-way ANOVA, p Ͻ 0.005). However, surface GluA1 in the AAV-Cre group after cLTP was similar to that without cLTP (without cLTP, 0.92 Ϯ 0.050, n ϭ 16; with cLTP, 0.87 Ϯ 0.043, n ϭ 19, one-way ANOVA, p Ͻ 0.005) (Fig. 4, E and F). Collectively, our results suggest Shp2 is required for trafficking of AMPA receptor during cLTP.

Discussion
In this study, we provided evidence that Shp2 was regulated in LTP. We demonstrated that Shp2 was translocated into spines during LTP and was up-regulated, which was represented by a marked increase of the Tyr(P)-542 level. Despite an extensive attempt in earlier studies utilizing genetic approaches that lead to fruitful and insightful results, there is limited understanding of the exact Shp2 functional roles in synaptic plasticity and memory. Besides being a signaling node upstream of the MEK-ERK pathway to activate ERK in LTP, Shp2 thereof is also regulated in LTP and memory formation. The data presented show that Shp2 quickly accumulated and responded to LTP stimuli in an acute slice receiving HFS as well as in cultured hippocampal neurons accepting synaptic NMDA receptor activation (Fig. 2). These results, and additional with previous studies, give a more comprehensive understanding of role of Shp2 in long term synaptic plasticity and memory.
Our result demonstrated that Shp2 was uniformly expressed throughout the mice brain, implying numerous and complicated functions of Shp2 in brain. Moreover, we found an appar-Shp2 Regulates Activity-dependent AMPA Receptor Trafficking ently higher level of Tyr(P)-542 in pons. Within the pons it is the pneumotaxic center that regulates the change from inhalation to exhalation (21). Thus, we surmised that the activity of Shp2 was possibly related to controlling the respiratory activity by pons.
Previous studies have shown that Shp2 physically associates with two important components of PSD, PSD95 and GluN2B (22)(23)(24). In line with those data, we demonstrated that Shp2 was located adjacent to PSD95 in cultured neurons by confocal microscopy and STORM. Notably, the super-resolution microscopy offered a much clearer image of Shp2 colocalization with PSD95 in dendrite spines (Fig. 1). Moreover, during LTP, Shp2 was rapidly translocated into spines, suggesting recruitment of Shp2 into spines serving as some significant functional roles in LTP (Fig. 2). Given Shp2 being physically associated with PSD95 and GluN2B, it will be intriguing to explore whether PSD95 or GluN2B act as a vehicle to mediate the translocation of Shp2 in LTP in the future.
Furthermore, we provided evidence for the first time that the Tyr(P)-452 level of Shp2 was up-regulated in response to cLTP, HFS-induced LTP stimuli, and contextual fear conditioning (Fig. 3). The phosphorylation level of Tyr-542 of Shp2 has been shown to be a concomitant regulation with the activity of Shp2 and at least to some extent, very likely represented the enzymatic activity of Shp2. Thus, we proposed that phosphatase activity of Shp2 was rapidly modulated in LTP and memory formation, rather than being a necessary constituting element required for LTP and memory. Following this suggestion, we showed that the inhibiting activity of Shp2 by its inhibitor NSC87877 resulted in a dampened increase of GluA1 at Ser-845 as well as GluA1 delivery into synapse during cLTP (Fig. 4). Taking advantage of Shp2 flox/flox mice, we prepared cultured hippocampal neurons lacking Shp2 via AAV-Cre-mediated KO. The results from Shp2 KO neurons are consistent with other reports (8 -10) and our unpublished data, 3 reinforcing the conclusion that Shp2 is critical for AMPA receptor delivery during LTP. Additionally, we found concomitant changes of ERK activity and phosphorylation of GluA1 in cLTP after inhibition of Shp2 activity either by NSC87877 or AAV-mediated KO of Shp2. Activation of ERK signaling facilitates AMPA receptor surface delivery during LTP (25). Furthermore, previous studies indicate that improper Shp2 function results in impaired LTP and memory via deregulated ERK activity (9, 10). Combined using our results and previous studies, ERK has been implicated as a crucial signaling node in the Shp2-related pathway that regulates GluA1 trafficking in LTP.
Heretofore, several potent and specific Shp2 inhibitors, including NSC87877, PHPS1, and GS-493, have been designed and developed for therapeutic treatment targeting deregulated Shp2-related diseases, such as glioblastoma, acute myelogenous leukemia, and gastric cancer (20, 26 -29). However, our results indicated that Shp2 was finely regulated and its phosphatase activity was a requisite for AMPA receptor trafficking in LTP. Given a Shp2 inhibitor affecting AMPA receptor surface expression in LTP and thus possibly harming memory, much more caution should be paid to application of the Shp2 inhibitor for treating Shp2-related diseases. In this regard, interruption of Shp2 function by interfering with the association of interacted proteins with Shp2 using peptide-based drugs or impeding upstream/downstream signaling nodes of the Shp2 pathway may be an alternative choice in targeting Shp2related diseases without afflicting the patients' memory.
Collectively, our results supplemented the illustration of the functional roles of Shp2 in LTP and memory. We not only show that Shp2 is rapidly translocated into spines but also was activated during LTP and contextual fear conditioning. We further revealed that the phosphatase activity of Shp2 was required for AMPA receptor surface expression in LTP by a specific inhibitor of Shp2 and AAV-Cre mediated KO of Shp2 in neurons. Thus, we proposed that it might be cautious when treating patients with Shp2 inhibitors. Last but not least, more comprehensive understanding of Shp2 is vital for our development of therapeutic approaches for Shp2-related diseases.

Experimental Procedures
Hippocampal Neuron Culture and Cell Line-All protocols in our experiments were performed followed Zhejiang University Animal Experimentation Committee guidelines and the guidelines conform to the Guideline for the Care and Use of Laboratory Animals. Efforts were made to minimize the number of animals used and reduce animal suffering. Hippocampal neurons cultures from Sprague-Dawley rats of embryonic day 18 (E18) or newborn (P1-P2) of Shp2 flox/flox mice were prepared following the protocol described previously (30) in our laboratory. In general, hippocampal tissue was quickly dissected and then digested in 0.5% trypsin for 15 min at 37°C. Dissociated neurons were seeded on poly-L-lysine (Sigma)coated 12-mm coverslips (Deckglaser) at a density of 100,000/ dish for immunocytochemistry or 35-mm dish (Corning) at a density of 300,000/dish for biochemical analysis in plating medium (Neurobasal containing 10% horse serum). Four hours later, the plating medium was substituted by culture medium (Neurobasal media containing 0.5 mM GlutaMax supplemented with 2% B27 and 1% penicillin/streptomycin). Subsequently, half-culture medium was replaced by fresh medium every 4 days. At DIV 4 cells were treated with cytosine arabinoside at a final concentration of 2.5 M. All cell culture reagents were purchased from GIBCO (Invitrogen) unless stated otherwise.
HEK293 cells were grown in DMEM containing 10% fetal bovine serum plus 1% penicillin/streptomycin. The indicated concentration of NSC87877 was added directly into the culture media for 30 min incubation time before being homogenized.
Western Blotting-Western blotting was performed according to conventional methods with minor modifications (30). Briefly, hippocampal tissue from mice or cultured cells was homogenized in RIPA buffer containing 10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, and 1% sodium deoxycholate (Beyotime) and centrifuged at 16,000 ϫ g for 20 min. Proteins in the supernatant were quantified using a BCA Protein Assay Kit (Thermo), and denatured by boiling in 4ϫ loading buffer for 5 min. Equal amounts of total protein were loaded onto 10% polyacrylamide-SDS gel and transferred to nitrocellulose membranes (Whatman, GE Healthcare). The membranes were incubated in blocking buffer (2.5% BSA in TBST) for at least 60 min at room temperature, and then incubated with the respective primary antibodies in TBST overnight at 4°C. After an extensive wash with TBST, the blots were incubated with HRP-conjugated secondary antibody for 1 h in TBST at room temperature. Following washes for 1 h with TBST, immunoreactivity was detected by chemiluminescence with ECL reagent (Bio-Rad). Protein bands were analyzed with Quantity One software (Bio-Rad).
Extraction of Subcellular Fraction-Subcellular fractionation was performed as described (12). Hippocampus from adult mice or the CA1 region of the hippocampal slice was homogenized in cold buffer A (0.32 M sucrose, 10 mM HEPES, pH 7.4) and centrifuged at 1,000 ϫ g for 4 min to remove nuclei and large debris (P1). The supernatant was subsequently centrifuged at 12,000 ϫ g for 20 min to obtain a crude membrane fraction (P2), the pellet was then rinsed twice in buffer B (4 mM HEPES, 1 mM EDTA, pH 7.4) and centrifuged at 12,000 ϫ g for another 20 min, the pellet was further dissolved in buffer C (20 mM HEPES, 100 mM NaCl, 0.5% Trition X-100, pH 7.2). Subsequently, the homogenate was incubated for 15 min and centrifuged at 12,000 ϫ g for 20 min, and this resulting supernatant was considered as the non-postsynaptic density membrane fraction (non-PSD). The pellet was then dissolved in buffer D (20 mM HEPES, 0.15 mM NaCl, 1% Triton X-100, 1% deoxycholic acid, 1% SDS, 1 mM DTT, pH 7.5) for 1 h at 4°C and subjected to centrifugation at 10,000 ϫ g for 15 min, The supernatant was regarded as the PSD or Triton-insoluble fraction. At last, P2, non-PSD, and PSD were used for Western blot analysis. All buffers were supplemented with protease inhibitors and phosphatase inhibitors.
Induction of Chemical LTP-The cLTP protocol was adopted from previous publications (17,18). Hippocampal neurons were rinsed twice in non-magnesium ECS (143 mM NaCl, 5 mM KCl, 3 mM CaCl 2 , 10 mM HEPES, 10 mM glucose, pH 7.4), and incubated in ECS ϩ (non-magnesium ECS containing 0.5 M tetrodotoxin, 1 M strychnine, 20 M bicuculline) for 20 min. Afterward, glycine was applied into the ECS ϩ for 5 min. After rinsing in non-magnesium ECS twice, neurons were incubated in ECS ϩ for the indicated time. In the control group, neurons were treated with ECS ϩ throughout the experiment. In the inhibitor group, APV or NSC87877 were applied throughout the process.
Slice Preparation and Field-EPSP Recording-The experiment was carried out as described previously (17). 5-6-Weekold male Sprague-Dawley rats were anesthetized with ether and quickly decapitated. The brain was quickly transferred to icecold ACSF (119 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl 2 , 1.3 mM MgSO 4 , 1.0 mM NaH 2 PO 4 , 26.2 mM NaHCO 3 , 11 mM glucose) saturated with 95% O 2 and 5% CO 2 . The hippocampus was carefully separated and 400-m thick transverse slices were obtained with a vibratome (Leica VT1200S) in ice-cold ACSF and recovered in ACSF at 25°C for at least 1.5 h before recordings.
The transverse hippocampal slice was then transferred to a submerged recording chamber continuously perfused with ACSF saturated with 95% O 2 and 5% CO 2 at a rate of 4 ml/min. After removing the CA3 area, 100 M picrotoxin was added in ACSF during recording. Field-EPSPs were recorded in the SC-CA1 region at 0.05 Hz for 30 min of stable baseline recording and 100 Hz/1-s induction stimulation were used to induce LTP and recording was continued for 10 or 60 min. Electrical stimuli were delivered through a bipolar concentric electrode (CBARC75, FHC).
Contextual Fear Conditioning-Before the contextual fear conditioning stimulation, male adult mice (C57BL/6, 8 -10 weeks age) were routinely handled for at least 3 days. All fear conditioning stimulations were performed in an NIR Video Fear Conditioning System (Med Associates). After 2 min of freely exploring the chamber, three foot-shocks (2 s duration, 0.6 mA intensity) were delivered at 30-s intervals through the stainless steel rods. After another 2 min habituation in the chambers, mice were transferred to the homecage for 10 or 60 min, then mice were sacrificed for hippocampi collection. The tissue was immediately stored in liquid nitrogen. The chambers were cleaned with 70% ethanol between trials.
Immunocytochemistry-We followed the protocol described previously for surface staining of GluA1 (31).
For staining of endogenous Shp2, pShp2 Tyr-542, and PSD95, neurons on coverslips were fixed with 4% paraformaldehyde for 10 min. After a brief rinse in PBS, neurons were then permeabilized and blocked simultaneously in the blocking buffer (0.2% Triton X-100 and 5% BSA in PBS) for 0.5 h. Primary antibodies were then applied in PBS containing 5% BSA overnight at 4°C. The following day, after three times washing with PBS, secondary antibodies were applied in PBS containing 5% BSA either for 1 h at room temperature or overnight at 4°C. For staining of GluA1, donkey anti-mouse Alexa 546-conjugated and donkey anti-rabbit Alexa 633-conjugated secondary antibody were used; for colocalization detection between Shp2 and PSD95, donkey anti-mouse Alexa 488-conjugated and donkey anti-rabbit Alexa 546-conjugated secondary antibody were used.
For STORM image, we followed the protocol previously described (29). After fixation, permeability, and blocking, DIV 18 hippocampus neurons were incubated with primary antibody in PBS containing 5% BSA (anti-PSD95, 1:60; anti-Shp2, 1:20) overnight at 4°C, and washed five times with PBS with each interval of 10 min. After an extensive wash, secondary antibodies were then applied in PBS containing 5% BSA (donkey anti-mouse Alexa 647, 1:50; donkey anti-rabbit Cy3B, 5 g/ml) overnight at 4°C. After five times washing in PBS, neurons were further fixed for another 10 min with 4% paraformaldehyde, then the coverslips were kept in PBS for immediate STORM microscopy.
Image Analysis and Quantification-Images were acquired with a confocal microscope (Fluoview FV1000, Olympus). Colocalization ratios of clusters were analyzed using ImageJ with plug-in of JACoP. For surface AMPA receptor intensity, the integrated intensity of individual puncta of the endogenous surface GluA1 on the dendritic protrusions was measured. For Z-stack images of hippocampal neurons, the images were recorded, processed, and analyzed with the help of Imaris software (Imaris software package, Bitplane AG, Zurich, Switzerland) as previously described (32). Acquisition of three-dimensional STORM images was performed using a protocol previously described (33) on a Nikon N-STORM microscope equipped with a ϫ100 oil immersion objective (CFI Apo ϫ100 oil NA 1.49 TIRF WD 0.12) and an Andor camera (iXon 897 back illuminated EMCCD). Each image acquisition cycle contained one activation frame (405 nm laser) and five imaging frames (561-or 647-nm lasers). The integration time of the EMCCD was set to 1 frame mode with an EM gain of 30, 5,000 -10,000 cycles per channel were taken for the reconstruction of each three-dimensional super-resolution image using Nikon NIS Elements 4.30 software.
Statistical Analysis-All data were analyzed using GraphPad Prism 6 software and represented as mean Ϯ S.E. The comparisons of two groups were compared by Student's t test. Data of multiple groups (Ն3 groups) were tested using one-way ANOVA followed with Tukey's post hoc test. p Յ 0.05 was considered as significant differences.