The Protein-tyrosine Phosphatase PTPMEG Interacts with Glutamate Receptor δ2 and ε Subunits*

Glutamate receptor (GluR) δ2 is selectively expressed in cerebellar Purkinje cells and plays a crucial role in cerebellum-dependent motor learning. Although GluRδ2 belongs to an ionotropic GluR family, little is known about its pharmacological features and downstream signaling cascade. To study molecular mechanisms underlying GluRδ2-dependent motor learning, we employed yeast two-hybrid screening to isolate GluRδ2-interacting molecules and identified protein-tyrosine phosphatase PTPMEG. PTPMEG is a family member of band 4.1 domain-containing protein-tyrosine phosphatases and is expressed prominently in brain. Here, we showed by in situhybridization analysis that the PTPMEG mRNA was enriched in mouse thalamus and Purkinje cells. We also showed that PTPMEG interacted with GluRδ2 as well as with N-methyl-d-aspartate receptor GluRε1 in cultured cells and in brain. PTPMEG bound to the putative C-terminal PDZ target sequence of GluRδ2 and GluRε1 via its PDZ domain. Examination of the effect of PTPMEG on tyrosine phosphorylation of GluRε1 unexpectedly revealed that PTPMEG enhanced Fyn-mediated tyrosine phosphorylation of GluRε1 in its PTPase activity-dependent manner. Thus, we conclude that PTPMEG associates directly with GluRδ2 and GluRε1. Moreover, our data suggest that PTPMEG plays a role in signaling downstream of the GluRs and/or in regulation of their activities through tyrosine dephosphorylation.

methyl-4-isoxazole-propionate. N-methyl-D-aspartate (NMDA) receptors, consisting of heterooligomers of one or two subunit(s) and two or three ⑀ subunits, are sensitive to NMDA.
Regarding the ␦ subfamily, both the pharmacological and structural features remain to be understood (2). The GluR␦2 subunit is expressed in cerebellar Purkinje cells, specifically at the synapses between parallel fibers and Purkinje cells (3,4). This restricted expression pattern suggests that the GluR␦2 subunit plays an important role in cerebellum-dependent motor learning. A study with GluR␦2 knock-out mice revealed that GluR␦2 is crucial in the synapse formation between parallel fibers and Purkinje cells, switching of climbing fiber innervation from multiple to mono-innervation, LTD induction, and motor coordination (5). In addition, the phenotype of Lurcher mice, which is characterized by degeneration of cerebellar Purkinje cells, is caused by a point mutation in the third transmembrane domain of GluR␦2 (6). Thus, it is important to define the pharmacological feature of GluR␦2 and GluR␦2-mediated signaling.
Intensive studies in the last decade have revealed that protein phosphorylation is one of the most important mechanisms regulating synaptic plasticity, learning, and memory. For example, induction of LTP/LTD at the CA1 region of the hippocampus is regulated by calmodulin kinase II and calciumcalmodulin dependent phosphatase (calcineurin) activities (7). Recently, tyrosine phosphorylation mediated by Src family PTKs was also reported to be involved in synaptic plasticity and downstream signaling subsequent to GluR activation (8 -10). However, in contrast to Ser/Thr phosphatases, little is known about how PTPases might function in GluR signal transduction.
PTPMEG (11), PTPH1 (12), PTPBAS (13), and PTPD1 (14) are members of a PTPase subfamily that contains, from the N terminus to the C terminus, a band 4.1 domain, one or more PDZ domains, and a PTPase catalytic domain. The band 4.1 domain is considered to be responsible for targeting proteins to the cytoskeleton-membrane interface. Consistently, PTPMEG is associated with membrane structures in cells (15). PDZ domains are present in some scaffold proteins and recognize C-terminal end (S/T)X(V/L/I) sequences (16,17) or bind to other PDZ domains in their targets (18). Because of segment-specific expression of PTPH1 in diencephalon, PTPH1 is suggested to play a crucial role in thalamocortical connections (19). Although highly expressed in brain (15), neither the expression pattern nor the role of PTPMEG has been elucidated.
Here, we report that PTPMEG interacts with GluR␦2 and GluR⑀1. Moreover, our study suggests involvement of PTP-MEG in the function of GluR␦2 and GluR⑀1 subunits.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screening-Yeast two-hybrid screens were performed using the L40 yeast strain harboring the reporter genes HIS3 and ␤-galactosidase under the control of upstream LexA-binding sites. According to the topological model of ionotropic glutamate receptors (20 -22), the C terminus of GluR␦2 (amino acids 922-1007) was used to screen a human brain cDNA library (CLONTECH Laboratories) in vector pACT2.
In Situ Hybridization -In situ hybridization was performed as described previously (26,27). In short, the cDNA fragments corresponding to the N terminus proximal region of mouse PTPMEG (amino acids 11-142) was amplified by the reverse transcription-polymerase chain reaction method. The amplified cDNA fragment was subcloned into the pBluescript II vector. After confirming the DNA sequence, we generated antisense and sense cRNA probes from the plasmid by in vitro transcription in the presence of [ 35 S]UTP. Parasagittal sections of brain (a thickness of 10 m) were prepared from C57BL/6 mice at postnatal day 17 (P17). cRNA probes (about 10 7 cpm) were applied to each slide for hybridization. Coverslip slides were then incubated overnight in humidified chambers at 55°C. After washing, slides were air-dried and exposed to Kodak X-OMAT film for 5 days. The emulsion radioautogram was done for 4 weeks.
Cell Culture, Transfection, Immunoprecipitation, and Immunoblotting-293T cells were maintained in 10% fetal bovine serum/Dulbecco's modified Eagle's medium under 37°C, 5% CO 2 condition. 293T cells (1 ϫ 10 6 /10-cm dish) were transfected with combinations of expression plasmids (total, 20 g) by the standard calcium phosphate method. Two days after transfection, cells were lysed in TNE buffer (50 mM Tris-HCl, 1% Nonidet P-40, 5 mM EDTA, 145 mM NaCl) containing 100 M Na 3 VO 4 , 100 M phenylmethylsulfonyl fluoride, 10 units/ml aprotinin, and 20 M Pefabloc SC (Merck). Insoluble fraction was excluded by centrifugation at 15 Krpm for 20 min. Total cell lysates (TCLs) were boiled in the presence of sample buffer. Affinity purified antibodies (1-3 g) were added to the precleared cell lysates to obtain protein immunoprecipitates. Immunoprecipitates and TCL were separated by SDSpolyacrylamide gel electrophoresis (7.5% gel) and transferred onto polyvinylidene difluoride membranes (Bio-Rad). The membrane was then blocked with 5% bovine serum albumin/TBST solution for 2 h at room temperature and treated with primary antibodies. Horseradish peroxidase-conjugated secondary antibodies and Renaissance Plus reagent (NEN Life Science Products) were used to visualize the immunoreactive proteins.
Preparation of Subcellular Fractions and Lysate from Brain-Isolation of subcellular fractions of brain was performed as described previously (31). Equal amounts (total protein, 15 g) of fractions were separated by SDS-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane. For immunoprecipitation, cerebella or telencephalons from 12-18-week-old C57BL/6 or GluR␦2 knockout mice were homogenized with 10 volumes (v/w) of homogenization buffer (0.32 M sucrose, 1 mM NaHCO 3 , 1 mM MgCl 2 , 100 M Na 3 VO 4 , 100 M phenylmethylsulfonyl fluoride, 10 units/ml aprotinin, and 20 M Pefabloc SC). The synaptosome and mitochondria fractions were prepared as described previously (31) and were lysed in 1% deoxycholate buffer (32). Immunoprecipitation of proteins from the lysates was performed as described above.

RESULTS
Identification of PTPMEG as a GluR␦2-interacting Molecule-To identify GluR␦2-interacting molecules, we employed yeast two-hybrid system using the C terminus proximal resi- In lanes 1-3, 1 ⁄20 amounts of the lysates used for immunoprecipitation were loaded for TCL samples. Immunoblotting was performed using anti-Myc or anti-GluR␦2 antibodies to detect the interacting-proteins. The same filter was stripped and then reprobed with the indicated antibodies.
dues of GluR␦2 (amino acids 922-1007) as a bait (Fig. 1A). We screened approximately 10 6 yeast transformants and obtained six positive clones that activated expression of selection markers His3 and LacZ. Nucleotide sequence analysis revealed that they encoded a GTPase exchange factor, three PTPases, and two PDZ containing-scaffold proteins that did not belong to the PSD-95 family. One of the PTPase-encoding clones carried a sequence for C-terminal two-thirds of PTPMEG (Fig. 1B). PT-PMEG belongs to a family of intracellular protein-tyrosine phosphatases that contain a band 4.1 domain, a PDZ domain, and a catalytic PTP domain (Ref. 11 and Fig. 1B). PTPMEG is expressed highly in brain (15), suggesting that PTPMEG may have important functions in the central nervous system. To confirm the interaction between GluR␦2 and PTPMEG, 293T cells were transfected with Myc-tagged PTPMEG and/or GluR␦2 expression plasmids, and then protein lysates were prepared from the transfected cells. By probing the anti-Myc immunoprecipitates of the lysates with anti-GluR␦2 or anti-Myc antibody, we showed that GluR␦2 co-precipitated with Myc-tagged PTPMEG only when both proteins were expressed (Fig. 1C, left panel). Conversely, PTPMEG was present in the GluR␦2 immunoprecipitates (Fig. 1C, right panel), indicating that PTPMEG associated with GluR␦2 in heterologous 293T cells.
Expression Pattern of PTPMEG in Brain-Previous reports indicated that expression of GluR␦2 was confined to cerebellar Purkinje cells (3). In contrast, the precise distribution of PTP-MEG in brain remained to be established. Because GluR␦2 immunoreactivity in cerebellum increased dramatically during synaptogenesis (postnatal 2-3 weeks) (2), we compared expression pattern of PTPMEG with that of GluR␦2 in brain during this same stage of development. Using the cRNA probes corresponding to mouse PTPMEG (amino acids 11-142), we performed in situ hybridization analysis of P17 mouse brain parasagittal sections (Fig. 2A). The antisense probe showed intense signals in the thalamus and the cerebellar Purkinje cell layer. Moderate signals were detected in the olfactory bulb, cerebral cortex, and hippocampus, and weak signals were detected broadly. No specific hybridization was detected using the FIG. 2. Distribution of the PTPMEG mRNA in brain. In situ hybridization of the PTPMEG mRNA was performed using sections of the brain at P17. A, dark field photomicrographs of a parasagittal section of brain. B, dark field photomicrographs of cerebellum. C, bright field photomicrographs of cerebellum. Th, thalamus; Cb, cerebellum; Ob, olfactory bulb; Cx, cerebral cortex; Hi, hippocampus; Mo, molecular layer; Pr, Purkinje cells; Gr, granular layer.

FIG. 3. Interaction of PTPMEG with GluR␦2 in brain.
A, subcellular localization of PTPMEG in cerebellum and telencephalon. The fractionated samples (15 g) prepared from brain lysates were subjected to immunoblotting with anti-PTPMEG antibody. The same filter was stripped and then reprobed with antibodies against control proteins. The subcellular fractionation was verified by immunoblotting with antibodies for control proteins, such as GluR␦2 (a GluR subunit enriched to cerebellar PSD fraction), GluR⑀1 (a GluR subunit enriched to PSD fraction), PSD-95 (a PSD fraction marker), and synaptophysin (a presynaptic fraction marker). B, identification of GluR␦2 in PTPMEG immunoprecipitates (left panel) and PTPMEG in GluR␦2 immunoprecipitates (right panel) from mouse cerebella. The lysates of wild-type or GluR␦2 knock-out mice cerebella were immunoprecipitated with the indicated antibodies. The immunoprecipitates and synaptosome and mitochondria fractions (Sm; 1 ⁄18 amount of lysate used for anti-GluR␦2 immunoprecipitates and 1 ⁄36 amount of lysate used for anti-PTPMEG immunoprecipitates) were subjected to immunoblotting with anti-GluR␦2, anti-PTPMEG, and anti-Trk B antibodies. Sb, soluble fraction; Sm, synaptosome and mitochondria fraction; Sy, synaptosome fraction; Cb, cerebellum; Tel, telencephalon; IP, immunoprecipitation; WT, wild type; KO, GluR␦2 knock-out. sense cRNA probe (data not shown). Dark and bright field photomicrographs of cerebellum revealed that the PTPMEG mRNA was abundant in Purkinje cells (Fig. 2, B and C). These data suggested co-expression of GluR␦2 and PTPMEG mRNA in Purkinje cells during synaptogenesis, which was consistent with our observation that GluR␦2 interacted with PTPMEG. We also detected abundant PTPMEG mRNA in the regions of the anterior nucleus and the ventral anterior nucleus of the thalamus. The molecules that would interact with PTPMEG in these regions have not been known.
Subcellular Distribution of PTPMEG and Its Interaction with GluR␦2 in Cerebellum-GluR␦2 subunits are targeted to the postsynaptic fraction of Purkinje cells (4). To compare the subcellular localization of PTPMEG with that of GluR␦2 in the neural cells, we prepared subcellular fractions of adult mouse cerebellum and telencephalon and performed immunoblot analysis (33)(34)(35). As shown in Fig. 3A, PTPMEG was present in the postsynaptic density (PSD) fraction where glutamate receptors were concentrated (upper panel). PTPMEG in the PSD fraction was about 3% of PTPMEG in the total soluble fraction. Note that the proteins in the PSD fraction prepared as above corresponded to about 0.5% of total soluble proteins. The validity of subcellular fractionation was proved by immunoblotting the proteins in the subcellular fractions with antibodies against various brain specific proteins such as GluR␦2, GluR⑀1, PSD-95, and synaptophysin (Fig. 3A). To determine whether GluR␦2 and PTPMEG interact in cerebellum, anti-GluR␦2 immunoprecipitates from cerebellar lysates were probed with anti-PTPMEG antibodies. The data clearly showed that PTPMEG was co-precipitated with GluR␦2 (Fig. 3B). In reciprocal coimmunoprecipitation experiments, anti-PTPMEG immunoprecipitates from cerebellar lysates were probed with anti-GluR␦2 antibodies. The data showed that anti-PTPMEG immunoprecipitates contained GluR␦2.
To identify the amino acid sequences responsible for the interaction between GluR␦2 and PTPMEG, we constructed deletion mutants of PTPMEG and GluR␦2 (Fig. 4, A and B), and combinations of GluR␦2 and PTPMEG constructs were transfected into 293T cells. Co-immunoprecipitation experiments with the lysates of the transfectants showed that wild-type PTPMEG and its mutants containing the PDZ domain, except mutant "e" interacted with wild-type GluR␦2 (Fig. 4C). It is likely that the sequence between the band 4.1 domain and the PDZ domain had an inhibitory effect on the interaction when exposed at the N terminus. Co-immunoprecipitation experiments also revealed that removal of the putative PDZ target sequence Thr-Ser-Ile from the C terminus of GluR␦2 and the point mutation at the C terminus (Ala-1007 instead of Ile-1007) abolished interaction between the two proteins (Fig. 4D). Thus, we concluded that at least some PTPMEG proteins were colocalized with GluR␦2 and that these two proteins interacted with each other in brain. Our present data suggest that PTP-MEG associates directly with the putative C-terminal PDZ target sequence of GluR␦2 via its PDZ domain.
Interaction between GluR⑀1 and PTPMEG in Cultured Cells and in Telencephalon-Because the Ser-Asp-Val sequence of the C terminus of GluR⑀1 is a typical target of the PDZ domain, GluR⑀1 may also interact with PTPMEG. To examine this possibility, 293T cells were transfected with expression plasmids encoding Myc-tagged PTPMEG and/or GluR⑀1, and the lysates of the transfectants were subjected to co-immunoprecipitation experiments. By probing anti-Myc immunoprecipi- tates with anti-GluR⑀1 or anti-Myc antibodies, we showed that GluR⑀1 co-precipitated with Myc-tagged PTPMEG only when both proteins were expressed (Fig. 5A). To demonstrate the interaction between GluR⑀1 and PTPMEG in brain, we carried out co-immunoprecipitation experiments using the lysates from mouse telencephalons. As shown in Fig. 5B, we detected PTPMEG in the anti-GluR⑀1 immunoprecipitates (left panel) and GluR⑀1 in the anti-PTPMEG immune-complex (right panel). The data suggested that PTPMEG interacted with GluR⑀1 in vivo. To address the mechanism of the interaction, we expressed various PTPMEG mutants (Fig. 4A) together with GluR⑀1 in 293T cells. From the lysates of the transfectants, we could co-immunoprecipitate GluR⑀1 with wild-type PTPMEG and with mutants that carried the PDZ domain (Fig. 5D). Mutant e of PTPMEG, which did not interact with GluR␦2, could associate with GluR⑀1. We do not have good explanation for this observation. However, it is possible that the threedimensional structure around the C terminus of GluR⑀1 is different from that of GluR␦2, which could cause different affinities of mutant e construct to the GluRs. In reciprocal immunoprecipitation experiments with various HA-tagged GluR⑀1 mutants (Fig. 5C), we could detect wild-type GluR⑀1 and its mutants containing the C-terminal PDZ target sequence in the PTPMEG immunoprecipitates (Fig. 5E). Furthermore, the point mutant of the C terminus of GluR⑀1 (GluR⑀1HA-V1464A) did not associated with PTPMEG in 293T cells. Thus, we conclude that the PDZ domain of PTPMEG and the C terminus of GluR⑀1 are critically important for the interaction between PTPMEG and GluR⑀1.
Enhancement of Fyn-mediated Tyrosine Phosphorylation of GluR⑀1 by PTPMEG-Because the biochemical and pharmacological features of NMDA receptors had been better characterized than those of GluR␦2, we decided to explore the biological significance of PTPMEG-GluR⑀1 interaction. We first examined the effect of PTPMEG on tyrosine phosphorylation of GluR⑀1. Because PTPMEG has PTPase activity, we expected that PTPMEG might compete with Fyn in tyrosine phosphorylating GluR⑀1. To test this possibility, the expression vectors encoding constitutively active Fyn (FynF) and/or PTPMEG were transfected into 293T cells together with GluR⑀1, and the level of tyrosine phosphorylation of GluR⑀1 was examined. Unexpectedly, coexpression of PTPMEG enhanced Fyn-mediated tyrosine phosphorylation of GluR⑀1 (Fig. 6). Furthermore, FIG. 5. Interaction of PTPMEG with GluR⑀1 in vivo. A, the interaction between GluR⑀1 and PTPMEG in heterologous 293T cells. 293T cells were transfected with combinations of the Myc-tagged PTP-MEG and GluR⑀1 expression plasmids. The lysates of the transfectants were immunoprecipitated (IP) with anti-Myc monoclonal antibody. Im munoprecipitates and TCL are as indicated in Fig. 1. Immunoblotting was performed using anti-Myc or anti-GluR⑀1 monoclonal antibody to detect the interacting proteins. The same filter was stripped and then reprobed with the indicated antibodies. B, identification of GluR⑀1 in PTPMEG immunoprecipitates (left panel) and PTPMEG in GluR⑀1 immunoprecipitates (right panel) from mice telencephalons. The lysates of mouse telencephalons were immunoprecipitated with the indicated antibodies. The immunoprecipitates and synaptosome and mitochondria fractions (Sm; 1 ⁄18 amount of lysate used for anti-GluR⑀1 immunoprecipitates and 1 ⁄36 amount of lysate used for anti-PTPMEG immunoprecipitates) were subjected to immunoblotting with anti-GluR⑀1, anti-PTPMEG, anti-Trk B, and anti-PSD-95 antibodies. C, schematic diagram of the GluR⑀1 deletion mutants. TM1-4 indicate transmembrane regions. HA indicates the influenza HA tag. Residue numbers correspond to those in the amino acid sequence of GluR⑀1. The ⌬A mutant contains an internal deletion of 125 amino acids (amino acids 1220 -1345). The ⌬B and ⌬C mutants lack 348 and 607 C-terminal amino acids, respectively. VA indicates the point mutant of GluR⑀1 (GluR⑀1HA-V1464A). D and E, binding of PTPMEG to the C terminus of GluR⑀1 through its PDZ domain. The expression constructs of PTP-MEG deletion mutants were as shown in Fig. 4A. Various combinations of expression constructs, as indicated above each lane, were transfected into 293T cells. The lysates of the transfectants were subsequently immunoprecipitated with the indicated antibodies to test for co-precipitation of associated proteins. Immunoprecipitates and TCL are as indicated in Fig. 1. The positions of the PTPMEG deletion mutants are indicated by arrowheads. The same filter was stripped and then reprobed with the indicated antibodies. wt, wild type. coexpression of PTPase inactive mutant of PTPMEG (termed PTPMEG-DA) did not increase Fyn-mediated tyrosine phosphorylation of GluR⑀1. We also showed that the PTPase active mutant (Fig. 4A, construct f) facilitated phosphorylation of GluR⑀1 more effectively than wild-type PTPMEG. These data showed that PTPMEG enhanced Fyn-mediated tyrosine phosphorylation of GluR⑀1 in a PTPase activity-dependent manner. Furthermore, introduction of the V1464A point mutation of GluR⑀1 significantly reduced Fyn-mediated GluR⑀1 phosphorylation in the presence of PTPMEG. Because the GluR⑀1 mutant was unable to interact with PTPMEG, the data supported our conclusion that Fyn-mediated tyrosine phosphorylation of GluR⑀1 was enhanced by the interaction between PTPMEG and GluR⑀1. A slight enhancement of tyrosine phosphorylation of the GluR⑀1 mutant observed in the presence of PTPMEG could be due to nonspecific microenvironmental changes induced by the phosphatase activity. The overall level of proteintyrosine phosphorylation in the cells expressing GluR⑀1, FynF in the presence of PTPMEG was significantly lower than that in the absence of PTPMEG (data not shown), suggesting that PTPMEG was active in the cells. Moreover, the effect of PTP-MEG on tyrosine-phosphorylation of GluR⑀1 was not caused by the further activation of FynF, because in vitro kinase assays showed that the activity of FynF was not dependent on the PTPase activity of PTPMEG (data not shown).

DISCUSSION
In this report, we have shown that a protein-tyrosine phosphatase PTPMEG interacts with GluR␦2 as well as GluR⑀1 both in vivo and in vitro. The interaction is mediated by the PDZ domain of PTPMEG and the C-terminal PDZ target sequences of these GluRs. Because the C termini of the four GluR⑀ subunits contain Glu-Ser-(Asp/Glu)-Val sequences, we assume that all of them would interact with PTPMEG. In fact, we observed the interaction between PTPMEG and GluR⑀2 in 293T cells (data not shown). Both the GluR⑀ subunits and GluR␦2 also bind to other PDZ domain-containing proteins, such as PSD-95 family proteins (16,17,32,36). These proteins are thought to be important to localize the GluR subunits to the postsynaptic cell membrane, serving as scaffold proteins. Unlike these, PTPMEG is a catalytical protein associated with PTPase activity. Therefore, the biological significance of its interaction with GluRs ought to be distinct from the scaffold function. We showed here that PTPMEG stimulated Fyn-mediated tyrosine phosphorylation of GluR⑀1 in a manner dependent on its PTPase activity. Although we previously reported that PSD-95 could stimulate Fyn-mediated tyrosine phosphorylation of GluR⑀1 (24), the underlying mechanisms of stimulation mediated by PSD-95 and by PTPMEG would be different from each other. Apparently, it is important to clarify the mechanism by which the interaction between GluRs and the PDZ domain-containing proteins, including PTPMEG, is regulated.
The band 4.1 domain is observed in several membrane-cytoskeleton linker proteins such as ezrin, radixin, and moesin (ERM proteins). ERM proteins associate with the transmembrane protein CD44 through their band 4.1 domains (37). The band 4.1 domain of ERM proteins can also interact with Rho GDI and Dbl and links ERM proteins to the signal transduction pathways controlled by Rho GTPases (38,39). These results suggest that the band 4.1 domain of PTPMEG functions as the binding domain for another transmembrane protein or as the adapter domain involved in the Rho GTPase signaling cascade to regulate the organization of actin cytoskeleton. We propose that the rearrangement of the cytoskeleton around GluR⑀1 might make GluR⑀1 more susceptible to tyrosine phosphorylation by Fyn PTK. It would be a likely mechanism by which PTPMEG exerts its effect on Fyn-mediated tyrosine phosphorylation of GluR⑀1. Tyrosine phosphorylation of GluR␦2 was not observed so far, suggesting that PTPMEG would play another role by interacting with GluR␦2.
NMDA receptors play crucial roles in the neuronal functions, such as development, synaptic plasticity, and neurotoxicity (28,40). NMDA receptor activation induces calcium influx, followed by activation of Ca 2ϩ -dependent enzymes including a calciumactivated neutral protease (calpain). In platelets, the phosphatase activity of PTPMEG is activated upon cleavage by calpain in response to calcium ionophore and thrombin (15). Thus, NMDA receptor stimulation may activate PTPMEG through the activation of calpain. This would result in further activation of NMDA receptor, because PTPMEG enhances Fyn-mediated GluR⑀1 phosphorylation and because tyrosine phosphorylation of GluR⑀1 could stimulate the channel activity of the NMDA receptor (41). However, these possibilities need clarification by experimentation in primary neurons.
We have shown that PTPMEG is prominently expressed in cerebellar Purkinje cells and in the cells at thalamus where GluR␦2 (3) and the four GluR⑀ subunits (42), respectively, are rich. Therefore, PTPMEG could interact with distinct GluRs depending on the cell types in which it is expressed. In addition, because PTPH1, a PTPase highly similar to PTPMEG, is also expressed significantly in thalamus, it is intriguing to address whether PTPH1 also interacts with GluR⑀ subunits. Finally, our present study suggests that PTPMEG could function as a regulator as well as a downstream signal transducer of GluR␦2 and GluR⑀ subunits. Further investigations, such as those aimed at elucidating the roles of PTPMEG in GluRs signaling pathway, will provide valuable insight into the molecular mechanisms of GluR␦2-dependent motor learning as well as NMDA receptor functions.
Fields, and S. Hollenberg for pBTM116, K. Mikoshiba for the mouse anti-synaptophysin monoclonal antibody, and T. Akiyama for the rabbit anti-PSD-95 polyclonal antibodies. We also thank H. Mori, C. Hisatsune, Y. Yoshida, and M. Ohsugi for valuable discussions and technical advice.