Modulation of the Channel Activity of the ε2/ζ1-Subtype N-Methyl d-Aspartate Receptor by PSD-95*

A channel-associated protein PSD-95 has been shown to induce clustering of N-methyld-aspartate (NMDA) receptors, interacting with the COOH terminus of the ε subunit of the receptors. The effects of PSD-95 on the channel activity of the ε2/ζ1 heteromeric NMDA receptor were examined by injection of PSD-95 cRNA into Xenopus oocytes expressing the NMDA receptors. Expression of PSD-95 decreased the sensitivity of the NMDA receptor channels to l-glutamate. Mutational studies showed that the interaction between the COOH terminus of the ε2 subunit of the NMDA receptor and the second PSD-95/Dlg/Z0-1 domain of PSD-95 is critical for the decrease in glutamate sensitivity. It is known that protein kinase C markedly potentiates the channel activity of the NMDA receptor expressed in oocytes. PSD-95 inhibited the protein kinase C-mediated potentiation of the channels. Thus, we demonstrated that PSD-95 functionally modulates the channel activity of the ε2/ζ1 NMDA receptor. PSD-95 makes signal transmission more efficient by clustering the channels at postsynaptic sites. In addition to this, our results suggest that PSD-95 plays a protective role against neuronal excitotoxicity by decreasing the glutamate sensitivity of the channels and by inhibiting the protein kinase C-mediated potentiation of the channels.

A channel-associated protein PSD-95 has been shown to induce clustering of N-methyl D-aspartate (NMDA) receptors, interacting with the COOH terminus of the ⑀ subunit of the receptors. The effects of PSD-95 on the channel activity of the ⑀2/1 heteromeric NMDA receptor were examined by injection of PSD-95 cRNA into Xenopus oocytes expressing the NMDA receptors. Expression of PSD-95 decreased the sensitivity of the NMDA receptor channels to L-glutamate. Mutational studies showed that the interaction between the COOH terminus of the ⑀2 subunit of the NMDA receptor and the second PSD-95/Dlg/Z0-1 domain of PSD-95 is critical for the decrease in glutamate sensitivity. It is known that protein kinase C markedly potentiates the channel activity of the NMDA receptor expressed in oocytes. PSD-95 inhibited the protein kinase C-mediated potentiation of the channels. Thus, we demonstrated that PSD-95 functionally modulates the channel activity of the ⑀2/1 NMDA receptor. PSD-95 makes signal transmission more efficient by clustering the channels at postsynaptic sites. In addition to this, our results suggest that PSD-95 plays a protective role against neuronal excitotoxicity by decreasing the glutamate sensitivity of the channels and by inhibiting the protein kinase C-mediated potentiation of the channels.
The NMDA 1 receptor is a subclass of ionotropic glutamate receptors in the mammalian brain. The NMDA receptors exhibit several channel properties distinct from other non-NMDA glutamate receptors, including a high Ca 2ϩ /Na ϩ permeability ratio, a voltage-dependent Mg 2ϩ block, and a requirement for glycine as co-agonist (1)(2)(3)(4). The receptors are modulated by phosphorylation catalyzed by protein kinase C (5, 6) and by tyrosine kinases (7,8). In brain, functional NMDA receptors are thought to exist as heteromultimers of the 1 subunit (NR1) with ⑀ subunits (NR2 subunits). Different combinations of these subunits exhibit distinct channel properties and characteristic regional and developmental expression in vivo (9 -11). Disruption of the 1 gene is lethal in mice (12), whereas that of the ⑀1 and ⑀2 genes results in reduced channel activity and reduced LTP in the hippocampus (13,14). A number of evi-dences indicate that the activation of NMDA receptors is essential for induction of LTP, which underlies the formation and storage of some forms of memories (15).
NR2 subunits (⑀ subunits) have been shown to interact specifically with PSD-95 (postsynaptic density-95) (16 -19). The latter protein is a member of the channel-associated proteins (PSD-95 family), including SAP97 (synaptic associated protein 97)/hdlg (20,21), chapsin-110/PSD-93 (22,23), and SAP102 (24). This family of proteins is characterized by the presence of three domains with a length of ϳ90 amino acids (PDZ domain) in the NH 2 -terminal region, followed by SH3 and guanylate kinase-like domains. PSD-95 was shown to interact with the COOH-terminal E(T/S)XV sequence motif of NMDA receptor ⑀ subunits (18,19) and of K ϩ channels through the NH 2 -terminal PDZ domains (25,26). The interaction of these proteins induces the clustering of the channel proteins (23,25,27). Thus, PSD-95 plays an important role in the molecular organization of NMDA receptors at postsynaptic sites. Although PSD-95 was shown to enhance the channel activity of the inward rectifier K ϩ channel Kir4.1 (27), it is not known whether PSD-95 directly affects the channel activities of the NMDA receptors.
In this study, we examined the effects of PSD-95 on ⑀2/1 heteromeric NMDA receptor channels expressed in Xenopus oocytes. We found that PSD-95 modulates the channel activity of the NMDA receptor. PSD-95 decreases the sensitivity of the channels to L-glutamate and inhibits the protein kinase Cmediated potentiation of the channels.

EXPERIMENTAL PROCEDURES
Preparation of cRNA and Oocytes-Plasmids pBKSA1, pBKSA⑀1, and pBKSA⑀2 containing cDNAs encoding mouse brain 1, ⑀1, and ⑀2 subunits of the NMDA receptor, respectively, were kindly provided by Dr. Masayoshi Mishina (6,28). Complementary RNAs of NMDA receptors were synthesized in vitro with T3 RNA polymerase using NotIcleaved plasmids as templates. The coding region of PSD-95 was amplified by polymerase chain reaction from a rat brain cDNA library. The resulting product was subcloned into pSP64 poly(A) vector (Promega, Madison, WI). The construction was confirmed by dideoxynucleotide sequencing. The PSD-95 cRNA was synthesized in vitro with SP6 RNA polymerase using EcoRI-cleaved plasmid. Deletion and site-specific mutants of the NMDA receptor ⑀2 subunit and PSD-95 were constructed by polymerase chain reaction strategy using Pfu polymerase (Stratagene, La Jolla, CA) (see Fig. 7A). The wild-type ⑀2 subunit and PSD-95 were used as polymerase chain reaction templates for mutation, except the C(3,5)S⌬PDZ(1ϩ2) construct, for which the ⌬PDZ(1ϩ2) construct was used as a template. Mutations were verified by dideoxynucleotide sequencing.
Stage V and VI oocytes were obtained from anesthetized Xenopus laevis and incubated for 2 h at room temperature with 2 mg/ml collagenase in Barth's medium (88 mM NaCl, 1 mM KCl, 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 0.82 mM MgSO 4 , 2.4 mM NaHCO 3 , and 7.7 mM Tris-HCl, pH 7.2) containing 18 units/ml penicillin and 18 g/ml streptomycin without Ca 2ϩ . The follicular cell layer was then removed with forceps. The oocytes were injected with ⑀2 and 1 cRNAs at a molar ratio of 1:2. The total amounts of cRNA injected were 5 ng/oocyte. For co-expression of PSD-95, the synthesized PSD-95 cRNA (1-25 ng) was injected 24 h after injection of the NMDA receptor cRNAs. Before recording, oocytes were incubated at 19°C for 18 -26 h in Barth's medium.
Electrophysiological Recordings-Currents were recorded with twoelectrode voltage-clamp techniques using a CA-1a high performance oocyte clamp (Dagan Corp., Minneapolis, MN). Electrodes were filled with 3 M KCl and had resistances of 1-5 megaohms. Oocytes were perfused by a constant stream of Ba 2ϩ Ringer solution (115 mM NaCl, 2.5 mM KCl, 1.8 mM BaCl 2 , and 10 mM HEPES, pH 7.2) at 23-25°C. The oocyte membrane was voltage-clamped at Ϫ70 mV. In the standard assay, currents were evoked by bath perfusion of Ba 2ϩ Ringer solution containing 100 M L-glutamate and 10 M glycine for 20 s, followed by a washout with standard Ba 2ϩ Ringer solution. Current signals were digitized for analyses; statistical significance was determined using Student's t test.
Immunoblotting-Polyclonal antibodies against the NMDA receptor ⑀2 and 1 subunits were obtained from Calbiochem and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Monoclonal antibody against the central region (amino acids 353-504) of rat PSD-95 was obtained from Transduction Laboratories (Lexington, KY). Polyclonal antibodies against PSD-95 were raised in rabbits using a glutathione S-transferase fusion protein with the NH 2 -terminal region (amino acids 4 -404) of rat PSD-95 as immunogen and were affinity-purified. For immunoblotting of PSD-95, oocytes were solubilized with Laemmli gel sample buffer containing 2% SDS (50 l/oocyte) (29). For immunoblotting of the NMDA receptor, the total membrane fraction was prepared from oocytes. After measurement of current responses, oocytes were homogenized with 100 mM NaCl, 5 mM EDTA, and 20 mM Tris-HCl, pH 7.5, and then centrifuged at 1000 ϫ g for 10 min. The supernatant was centrifuged at 200,000 ϫ g for 30 min. The obtained pellet (total membrane fraction) was solubilized with Laemmli gel sample buffer. The samples were separated by SDS-polyacrylamide gel electrophoresis on 5 or 7.5% gels and transferred to nitrocellulose sheets. After blocking with 5% skim milk and 0.05% Tween 20 in Tris-HCl-buffered saline, the sheets were incubated with primary antibodies in the blocking solution. Labeled bands were visualized using enhanced chemiluminescence (Pierce). The bands were analyzed by densitometric scanning using Densitograph AE-6900M (Atto Corp., Tokyo, Japan). The amounts of the proteins were quantified from the intensity of the bands, which has a linearity to the amounts of the samples applied to the gel.

RESULTS
Characterization of ⑀2/1 Heteromeric NMDA Receptor Channels in Oocytes Co-expressing PSD-95-To investigate the effects of PSD-95 on the channel activities of the ⑀2 (NR2B)/1 (NR1a) heteromeric NMDA receptor, we expressed the NMDA receptor and PSD-95 in Xenopus oocytes by injection of in vitro synthesized cRNAs. When the expression of the proteins was monitored by immunoblotting, the level of the NMDA receptor reached a plateau ϳ40 h after injection of the cRNAs and was then constant for several days (data not shown). On the other hand, the level of PSD-95 rapidly increased to the maximum within 24 h after injection and then decreased (data not shown). Therefore, in this study, PSD-95 cRNA was injected into oocytes that were already expressing the NMDA receptor, and the effects of PSD-95 on the NMDA receptor channels were examined 18 -26 h after injection of PSD-95 cRNA, measuring current responses to 100 M L-glutamate and 10 M glycine in Ba 2ϩ Ringer solution under voltage clamp at Ϫ70 mV. However, the results obtained with these oocytes were qualitatively the same as those obtained with oocytes in which cRNAs of the NMDA receptor and PSD-95 were co-injected.
To characterize the channels in oocytes co-expressing PSD-95, we first examined the effects of ifenprodil on the channel activity. Ifenprodil is known to selectively block ⑀2/1 heteromeric NMDA receptor channels (30). Ifenprodil completely inhibited the channels in oocytes co-expressing the receptor and PSD-95, as in oocytes expressing the receptor alone, whereas it had no effect on other subtypes of NMDA receptor channels such as the ⑀1/1 heteromeric receptor (Fig. 1). These results indicate that the current responses in oocytes co-expressing the receptor with PSD-95 are indeed through ⑀2/1 NMDA receptor channels. The average of the current responses from 15 oocytes was 1.34 Ϯ 0.56 A (mean Ϯ S.D., n ϭ 15) in oocytes expressing the NMDA receptor and 2.56 Ϯ 1.00 A (mean Ϯ S.D., n ϭ 15) in oocytes co-expressing PSD-95. When the expression of the receptor was compared in these oocytes by immunoblotting, the expression level of the receptor was proportional to the current response, and the ratio between ⑀2 and 1 subunits was not changed by PSD-95 (Fig. 2). These results indicate that PSD-95 does not inhibit the expression or plasma membrane insertion of the receptor. A higher expression level of the receptor in oocytes co-expressing PSD-95 might be due to the stabilization of the receptor by interacting with PSD-95. However, the amplitudes of evoked currents varied from oocyte to oocyte and could not simply be compared between oocytes expressing the receptor and co-expressing PSD-95. In this study, we focused on the qualitative differences in the channel properties between them. When the channel activities were measured at various concentrations of glutamate, the dose-response curve shifted to the right in oocytes co-expressing the receptor and PSD-95 (Fig. 3). The EC 50 was increased from 1.39 to 5.76 M by injection of PSD-95 cRNA. Thus, PSD-95 significantly decreased the sensitivity of the channels to glutamate. On the other hand, PSD-95 did not change other basic properties of the NMDA receptor such as a requirement for glycine as co-agonist, the current-voltage relationship, and a voltage-dependent Mg 2ϩ block of the channels (data not shown).
Effects of PSD-95 on the Protein Kinase C-mediated Potentiation of the Channels-It was reported that the channel activity of the ⑀2/1 NMDA receptor in oocytes is markedly potentiated by treatment with TPA, due to activation of protein kinase C (28,31,32). Under our assay conditions, the evoked currents were ϳ4-fold increased by 1 M TPA in oocytes expressing the NMDA receptor (Figs. 4 and 5). In oocytes coexpressing PSD-95, however, little potentiation of the channel activity (ϳ1.3-fold) was observed with TPA treatment (Figs. 4  and 5). This inhibitory effect of PSD-95 on the potentiation of the channels by TPA was dependent on the amounts of PSD-95 cRNA injected into an oocyte (Fig. 5). To rule out the possibility that PSD-95 directly modulates protein kinase C in oocytes, we examined the effects of PSD-95 on the metabotropic glutamate receptor. The activity of metabotropic glutamate receptors in oocytes can be estimated by the channel activity of Ca 2ϩ -activated chloride channels. When expressed in oocytes, metabotropic glutamate receptor-5 showed pronounced desensitization in response to an activation by 1 mM glutamate (data not shown), as reported previously (33). This desensitization has been shown to require protein kinase C-catalyzed phosphorylation of the receptor (33). Indeed, a brief treatment with TPA after activation by glutamate prolonged the desensitization. Co-expression of PSD-95 had no effect on the desensitization upon activation by 1 mM glutamate or on the prolonged desensitization by TPA (data not shown), indicating that expression of PSD-95 does not change the activity of protein kinase C in oocytes.
Analyses of the Channel Modulation by PSD-95 Using the Deletion Mutant of the NMDA Receptor-PSD-95 has been shown to interact with the COOH terminus of the NMDA receptor ⑀2 subunit through the first two PDZ domains (18,19,25). To clarify a role of this interaction in the modulation of glutamate sensitivity, the effects of PSD-95 were examined on the mutant NMDA receptor that lacks the COOH-terminal four amino acid residues of the ⑀2 subunit. The change in the sensitivity to glutamate was monitored by the ratio of current response at 100 M glutamate to that at 10 M. In accord with the dose-response curves (Fig. 3), expression of PSD-95 increased the ratio from 1 to ϳ1.5 in oocytes expressing the wild-type NMDA receptor (Fig. 6). In oocytes expressing the mutant receptors, however, the ratio remained ϳ1 regardless of PSD-95 expression (Fig. 6). These data indicate that the interaction of the COOH-terminal region of the ⑀2 subunit with the PDZ domain of PSD-95 is essential for a decrease in sensitivity to glutamate.
Analyses of the Channel Modulation by PSD-95 Using Mutant PSD-95-The modulation of glutamate sensitivity by PSD-95 was further analyzed using a series of PSD-95 mutants (Fig. 7A). The expression of these constructs was compared with that of wild-type PSD-95 by immunoblotting with antibodies against the NH 2 -terminal region (amino acids 4 -404) and against the central region (amino acids 353-504) of rat PSD-95. When 25 ng of cRNA was injected into an oocyte for each mutant, there was no significant difference in the expres- sion of these constructs at the protein level (Fig. 7, B and C). The deletion mutant lacking the SH3 domain (⌬SH3), the COOH-terminal half-region including the SH3 and guanylate kinase-like domains (⌬C-half), or the PDZ1 domain (⌬PDZ1) had the same effect on the glutamate sensitivity of the channels as full-length PSD-95 (Fig. 8). Deletion of the PDZ2 domain (⌬PDZ2) significantly decreased the ratio of current responses between 100 and 10 M L-glutamate from that of fulllength PSD-95. A double PSD domain deletion mutant lacking both PDZ1 and PDZ2 domains (⌬PDZ(1ϩ2)) and a triple PDZ domain deletion mutant (⌬PDZ(1ϩ2ϩ3)) further decreased the ratio to that observed in oocytes expressing the channels without PSD-95 (Fig. 8). Since two cysteine residues near the NH 2 terminus at positions 3 and 5 in PSD-95 have been shown to be critical for the in vivo interaction with the K ϩ channel (34, 35), we examined the role of these two cysteine residues in the channel modulation by PSD-95 using the mutant in which both of these cysteine residues were substituted to serine (C(3,5)S). This construct partially decreased the ratio from that of wildtype PSD-95 (Fig. 8). Deletion of both PDZ1 and PDZ2 domains from this mutant (C(3,5)S⌬PDZ(1ϩ2)) further decreased the ratio to the level without PSD-95 (Fig. 8). Thus, the PDZ2 domain is critical for the action of PSD-95 on the glutamate sensitivity of the channels, although the PDZ1 domain and the NH 2 -terminal two cysteine residues play an additional role in it. DISCUSSION PSD-95 has been shown to induce clustering of the ⑀2 (NR2B) NMDA receptor (23). In addition to this function of PSD-95, the present study demonstrated that it functionally modulates the channel activity of the ⑀2/1 heteromeric NMDA receptor. PSD-95 decreased the sensitivity of the channels to L-glutamate (Fig. 3) and inhibited the protein kinase C-mediated potentiation of the channels (Figs. 4 and 5). Furthermore, we demonstrated that the interaction between the COOH terminus of the ⑀2 subunit and the PDZ2 domain of PSD-95 plays a critical role in the decrease in glutamate sensitivity (Fig. 8). On the other hand, other basic channel properties of the ⑀2/1 NMDA receptor such as a voltage-dependent Mg 2ϩ block, a requirement for glycine as co-agonist, and inhibition of the channels by ifenprodil are not significantly altered by PSD-95.
Our results indicate that PSD-95 has inhibitory effects on ⑀2/1 NMDA receptor channels, in contrast to the inward rectifier K ϩ channel Kir4.1. The channel activity of the latter channel is enhanced by PSD-95 in co-expressed cells (27). For ⑀2/1 NMDA receptor channels, PSD-95 not only makes signal transmission more efficient by clustering the channels at postsynaptic sites, but also protects neuronal cells from excitotoxicity by decreasing the glutamate sensitivity and by inhibiting the protein kinase C-mediated potentiation.
Our mutational studies indicate that a direct interaction between the COOH terminus of the NMDA receptor ⑀2 subunit and the PDZ2 domain of PSD-95 is responsible for the decrease in the glutamate sensitivity of the channels because PSD-95 could not decrease the sensitivity of the mutant NMDA receptor lacking the COOH-terminal four amino acid residues (Fig.  6) and because the activity to decrease the sensitivity was significantly reduced in any mutants of PSD-95 lacking the PDZ2 domain (Fig. 8). It has been reported that the ⑀2 subunit (NR2B) has a strong preference for the PDZ2 domain over the PDZ1 domain (19). In accord with this, deletion of the PDZ2 domain did not completely abolish the effect of PSD-95, whereas a double PSD domain deletion of both PDZ1 and PDZ2 domains did (Fig. 8). The interaction between the COOH terminus of the channels and the PDZ2 domain also induces channel clustering (23,34). Does channel clustering decrease glutamate sensitivity? The NH 2 -terminal two cysteine residues of PSD-95 have been shown to be essential for clustering of Shaker-type K ϩ channel Kv1.4, but not for the binding of the COOH terminus of the channel to the PDZ1 or PDZ2 domain (34). If this is true of the NMDA receptor, our results suggest that the binding (but not the clustering) decreases glutamate sensitivity because the activity to decrease the glutamate sensitivity was still retained to some extent (Fig. 8) when cysteines 3 and 5 of PSD-95 were substituted with serine, which completely abolishes the channel clustering activity (34). This intermediate effect of the cysteine substitution might be explained by the finding that cysteines 3 and 5 are palmitoylated, which is essential for membrane targeting of PSD-95 (35). The PSD-95 mutant, which is not recruited to the plasma membrane, might not efficiently exert its action on the channels even though it has a PDZ2 domain.
The EC 50 of NMDA receptor channels for L-glutamate in neurons has been measured in cultured neurons. The reported values vary from 2.3 to 16 M (36, 37). Lower EC 50 values (0.8 -1.5 M) were reported, including this study, in oocytes expressing ⑀2/1 (or NR1/NR2B) NMDA receptor channels (28,38). These values cannot simply be compared because a neuron may not express only one type of NMDA receptors. Furthermore, the kinetics of the channel in neurons are not simple since the desensitization occurs during measurement. Despite these uncertainties, the EC 50 obtained from oocytes co-expressing the ⑀2/1 NMDA receptor with PSD-95 (Fig. 3) is in good agreement with EC 50 values obtained from neurons. We do not know the physiological significance of this effect of PSD-95 on the channels. PSD-95 may play a protective role in neuronal excitotoxicity, lowering the sensitivity of the NMDA receptor channels to glutamate.
In this study, we demonstrated that PSD-95 suppresses the protein kinase C-mediated potentiation of NMDA receptor channels (Figs. 4 and 5). It is unlikely that PSD-95 directly inhibits the kinase activity of protein kinase C in oocytes because PSD-95 had no effect on the desensitization of metabotropic glutamate receptor-5 in oocytes co-expressing both proteins, which was shown to require phosphorylation of metabotropic glutamate receptor-5 by protein kinase C (33). Further study is necessary to elucidate how the NMDA receptor channels are potentiated by protein kinase C and how this potentiation is modulated by PSD-95.
Although ⑀2/1 NMDA channels in oocytes are potentiated by the treatments that activate protein kinase C (Fig. 4) (31,32), it is still controversial whether phosphorylation by protein kinase C potentiates the NMDA receptor channels in vivo. The protein kinase C-mediated potentiation of the channels is observed in dorsal horn neurons in spinal cord and caudal brain stem (5,39), whereas activation of the kinase does not potentiate NMDA receptor-mediated responses in CA1 hippocampal neurons (40) where the ⑀2 subunit is highly expressed (9,10,28). The latter observation can be explained in part by our finding that PSD-95 suppresses the protein kinase C-mediated potentiation of ⑀2/1 NMDA receptor channels.
The induction of LTP at synapses requires Ca 2ϩ entry into the postsynaptic dendritic spine via NMDA receptors (15). LTP is triggered by delivering synchronous high frequency stimulation, a tetanus, to the pathway. In the widely accepted model, tetanic stimulation sufficiently depolarizes the postsynaptic membranes, which reduces the extent of the Mg 2ϩ -induced block of NMDA receptors and allows Ca 2ϩ influx (15). Consistent with this model, the PSD-95-interacting NMDA receptors also showed a voltage-dependent Mg 2ϩ block of the channels (data not shown). A number of evidences indicate that the activation of protein kinase C is necessary for induction of LTP (15,41). If the activation of protein kinase C is induced by Wild-type or mutant PSD-95 cRNA (25 ng) was injected into an oocyte expressing the ⑀2/1 NMDA receptor. Immunoblotting was performed as described under "Experimental Procedures" using polyclonal antibodies against the NH 2 -terminal region (amino acids 4 -404) (B) and a monoclonal antibody against the central region (amino acids 353-504) (C) of rat PSD-95. tetanic stimulation, PSD-95 may play a protective role regarding the NMDA receptor channels, preventing further potentiation of the channels and excess Ca 2ϩ influx.