Protein kinase C transiently activated heteromeric N-methyl-D-aspartate receptor channels independent of the phosphorylatable C-terminal splice domain and of consensus phosphorylation sites.

We have expressed dual subunit combinations of isoforms of the N-methyl-D-aspartate receptor, NR1A-NR2A and NR1C-NR2A, in Xenopus oocytes. We show that both forms of the receptor are stereospecifically activated by low concentrations (10 nM) of the phorbol ester 4-beta-phorbol 12-myristate 13-acetate, known to activate protein kinase C (PKC). The activation is transient, and, after reaching a maximum in about 10 min, it decreases rapidly in spite of the continuous presence of phorbol ester. The addition of 2 microM oleoylacetylglycerol had similar consequences. NR1C differs from NR1A by a deletion of 37 amino acids that include four consensus phosphorylation sites for PKC in the C-terminal region. The corresponding peptide has been shown to become phosphorylated upon activation of PKC in neurons (Tingley, W. G., Roche, K. W., Thompson, A. K., and Huganir, R. L. (1993) Nature 364, 70-73). However, the activity of NR1C-NR2A receptors was stimulated 7-fold, twice the potentiation observed for NR1A-NR2A. By site-specific mutagenesis of NR1C and NR2A, we removed additional consensus PKC phosphorylation sites located between TM3 and TM4. Coexpression of these mutant subunits showed a similar response to phorbol esters as wild type receptors. Our results indicate that neither the predicted consensus phosphorylation sites between transmembrane sequences TM3 and TM4 nor the phosphorylatable C-terminal splice domain is essential for the modulation of N-methyl-D-aspartate receptors by PKC.

W e have expressed dual subunit combinations of isoforms of the N-methyl-D-aspartate receptor, NRlA-NR2A and NRlC-NR!U, in Xenopus oocytes. W e show that both forms of the receptor are stereospecifically activated by low concentrations (10 m) of the phorbol ester 4-P-phor-bo1 12-myristate 13-acetate, known to activate protein kinase C (PKC). The activation is transient, and, after reaching a maximum in about 10 min, it decreases rapidly in spite of the continuous presence of phorbol ester. The addition of 2 p m oleoylacetylglycerol had similar consequences. NRlC differs from NRlA by a deletion of 37 amino acids that include four consensus phosphorylation sites for PKC in the C-terminal region. The corresponding peptide has been shown to become phosphorylated upon activation of PKC in neurons (Tingley, W. G., Roche, K. W., Thompson, A. K., and Huganir, R. L. (1993) Nature 364,70-73). However, the activity of NRlC-NR2A receptors was stimulated 7-fold, twice the potentiation observed for NRlA-NR2A. By site-specific mutagenesis of NRlC and = A , we removed additional consensus PKC phosphorylation sites located between TM3 and TM4. Coexpression of these mutant subunits showed a similar response to phorbol esters as wild type receptors. Our results indicate that neither the predicted consensus phosphorylation sites between transmembrane sequences TM3 and TM4 nor the phosphorylatable C-terminal splice domain is essential for the modulation of N-methy1-D-aspartate receptors by PKC.
Glutamate is the major excitatory neurotransmitter in vertebrate brain where it acts as a n agonist a t ligand-gated ion channels and at metabotropic receptors. Glutamate neurotransmission has been implicated in neuronal plasticity and in neurodegeneration (for review, see Ref. 1). NMDA' receptors constitute a subgroup of ligand-gated glutamate channels, which are essential for inducing long-term potentiation and, due to their Ca2+ permeability, for the neurotoxicity of glutamate. cDNAs encoding a large number of protein subunits have * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Evidence has been provided that PKC enhances NMDA-mediated current responses (3,6,8,(10)(11)(12). Intracellularly applied PKC enhances such currents, in part by reducing the Mg2+-dependent block in isolated trigeminal neurons (lo), and phorbol esters stereoselectively enhance NMDA-induced currents measured in Xenopus oocytes after injection with rat brain RNA (11). This potentiation of NMDA-induced currents is probably an important step in the induction of long-term potentiation (13), a phenomenon thought to underlie memory and learning.
On the molecular level, it is not clear how the interaction of PKC with NMDA receptors occurs. Specifically, it is not known which PKC isoform is involved and what amino acids on NMDA receptors are responsible for the functional modulation. It has been shown that homomeric NMDA receptor function can be enhanced by p-PMA after expression in Xenopus oocytes. This is the case for mouse ('1 (homologous to 1A subunits from rat) (3), rat lA, and rat 1G subunits (called NRlB in the nomenclature chosen here) (8). Similarly, the function of ('161 a n d 5 1~2 but not ('le3 heteromeric NMDA channels has been reported to be enhanced by phorbol ester (6). In some of these experiments, it has been noted that the stimulation by p-PMA was transient (6,12), but, because p-PMA was always added only at the beginning of the experiments, the loss of stimulation might have been caused by the washout of p-PMA.
Very recently, Tingley et al. (14) suggested that a specific C-terminal splice domain occurring in some variants of the NR1 subunits, carrying several consensus phosphorylation sites for PKC, is responsible for the functional modulation of NMDA receptors by PKC. In addition, they proposed that alternative splicing of subunits might add to diversity in longand short-term modulation of synaptic efficiency. Structural evidence was provided for the use of the phosphorylation sites on the corresponding amino acids.
We describe here in detail the modulation of heteromeric NMDA receptor channels after continuous stimulation of PKC by p-PMA and OAG and compare the effects of the presence or the absence of the C-terminal splice domain. Furthermore, in the heteromeric channel composed with the NRlC subunit, which is lacking the 37 amino acids excoded by exon 21 (15), we removed all four putative substrate sites for PKC located on the putative intracellular domains of the two subunits between TM3 and TM4 and studied the functional consequences.

MATERIALS AND METHODS
Isolation of cDNAs-To isolate the NRlA and NRlC cDNAs, whole rat brain poly(A)+ RNA (1 pg) was reverse-transcribed to cDNA and then amplified by means of PCR (using a gene Amp RNA-PCR kit from Perkin-Elmer). For PCR, the following oligonucleotide primers derived from the rat NMDARl sequence (2) were used: sense (5'-?TCCTA-CACTGCCAACTTGGCAGC) and antisense (5'-GACCCCTGCCATGT-TCTCAAAAGT). The final PCR reaction product was extracted with phenolkhloroform and then precipitated with ethanol. The PCR products were electrophoresed on a 1% agarose gel. The major band of 513 base pairs was cut from the gel. The PCR-derived fragment was 32Pnick-translated and used as a probe for hybridization screening of a rat hippocampal Agtll cDNA library. Among 1.2 x lo6 recombinants examined, 14 positive clones were plaque-purified. The sequence analysis of the two largest inserts (designated NRlAand NRlC) revealed that both clones are identical to the rat NMDARlA and 1C subunits described by Sugihara et al. (7).
To isolate the NR2A subunit, a rat cortex A@,, cDNA library was screened with a 675-base pair rat cDNAprobe. This probe was obtained from reverse transcription PCR amplification of rat brain poly(A)+ RNA using PCR primers based on the sequence of NR2A published by Double and triple mutations of PKC sites were successively introduced into the NRlC cDNA. The complete DNA inserts of all the mutants were sequenced to ensure that the correct mutations were introduced without unwanted side mutations.
Expression of NMDA Receptors in Xenopus Oocytes-The in vitro transcription, capping, and polyadenylation of the rat brain subunit isoforms have been described elsewhere (17-191. The cRNA combinations were coprecipitated in ethanol, shipped, and stored below 0 "C. Isolation of follicles from the frogs, culturing of the follicles, injection with RNA, and removal of the follicular cell layers from the oocytes were all performed as described earlier (20). Follicles were injected with about 50 nl of the capped transcripts. This solution contained the transcripts coding for each of the different subunits a t a concentration of 20 m to allow injection of stoichiometric amounts. The follicular cell layers were removed from the oocytes 1 or 2 days prior to the electrophysiological experiments.
Electrophysiological Experiments-All electrophysiological measurements were carried out on denuded oocytes. Oocytes were placed in a 0.4-ml bath on a nylon grid, and the bath was perfused throughout the experiment at 6 mVmin with 90 m NaCl, 1 IILM KCl, 1 II~M CaCl,, 5 m Hepes-NaOH (pH 7.4). The perfusion medium could be switched (70% change in <0.5 s) (21) to both, one supplied with NMDNglycine or one supplied with the activator of PKC. Magnesium ions were omitted in order to avoid inhibition of the NMDA channel. In the medium containing NMDNglycine, Ca2+ was replaced by Ba2+ in order to prevent influx of Ca2+ through NMDAchannels and subsequent activation of the Ca2+dependent chloride current endogenous to Xenopus oocytes. The perfusion solution was applied through a glass capillary with an inner diameter of 1.35 mm, the mouth of which was placed about 0.4 mm from the surface of the oocyte (21). Phorbol esters (Sigma) and OAG (Novabiochem, CH-4448 Laufelfingen, Switzerland) were dissolved in dimethyl sulfoxide. Final concentration of dimethyl sulfoxide in an experiment was always ~0 . 1 % .
All experiments were carried out a t room temperature (21-25 "C). For the current measurements, oocytes were impaled with two microelectrodes, and the membrane potential was voltageclamped at -80 mV. NMDNglycine was used at concentrations of 100 and 10 p i , respectively. This resulted in about 40-50% of the maximal current amplitude, as determined with 100 pi glutamate, 100 m glycine routinely before every experiment. The effect of phorbol esters was determined as described earlier (22). Briefly, NMDNglycine was applied at intervals of 4 min for about 20 s. First, control applications were 5"AGATGAACTTGTCTGCGGGGTITCTGAGCC 3'; NRlC (S748A), This procedure was chosen in order to prevent both accumulation and washout of the hydrophobic drug. 100 NMDN10 1.1~ glycine (bar) was applied for 20 s every 4 min in order to measure the current amplitude. The times given aboue the bars indicate the time measured from the first exposure to p-PMA. made until the currents were constant and the latter standardized to 100%. 10 m p-PMAor 100 nM a-PMA were applied by bath perfusion for an initial period of 1 min. In order to prevent both washout and accumulation of the drugs, perfusion was thereafter switched to drug perfusion for a 1-min interval every 4 min. OAG was applied by continuous perfusion. Data are expressed in percent of the control current amplitude and are indicated as mean f S.D. Alternatively, the data were normalized to the current measured 10 min after the first application of p-PMA. Statistical significance was tested by Wilcoxon's rank test. In control experiments, the oocyte surface area was determined by measuring capacity transient currents, and it was ascertained that under the mentioned conditions p-PMA does not alter the membrane surface area (23).
Nomenclature of NMDA Receptor Subunits-In this paper, we are following the nomenclature suggested by Sugihara et al. (7) and Monyer et al. (5) and use the names NR1 and NR2 for the subunit groups also called NMDARl and NMDAR2 for rat subunits and ( and E for mouse NRlA, NRlC, and NR2A. NRlC differs from NRlA by the deletion of 37 subunits. The present work was performed with the subunit isoforms amino acid residues (D864-T900) encoded by exon 21 (7, 15) containing four consensus phosphorylation sites for PKC.

Functional NMDA Receptor Expression in Xenopus Oocytes-
The dual subunit combination NRlC-NR2A was expressed in Xenopus oocytes. Maximal current amplitudes amounted to 1 0 0 4 0 0 nA in a Ca-free medium. In the medium containing the agonists 100 NMDA and 10 p~ glycine, Ca2+ ions were replaced by Ba2+ ions in order to eliminate Ca2+-activated currents endogenous to the oocyte. Ca2+ was maintained in the medium during the intervals between agonist application in order to prevent depletion of Ca2+ in the oocyte. Agonist-induced desensitization was rapidly reversed within less than 1 min, such that our observations are not perturbed by desensitization.
Response of NRlC-NR2A Type NMDA Receptors to PMA and OAG-Oocytes expressing NRlC-NR2A were exposed to conditions known to result in activation of PKC. p-PMA was repeatedly applied for 1 min with intervals of 3 min in order to minimize both accumulation and washout during the experiment. Fig. 1 shows a typical experiment. After several control applications of NMDNglycine to ensure constant current amplitudes, 10 nM p-PMA was applied for 1 min. Within 2 min after the first exposure to p-PMA, the NMDA current amplitude was more than doubled, and, within about 10 min, a maximum amplitude was reached. At later times, current amplitudes decreased again. Fig. 2 shows the time course of the 6-PMA effect averaged from four oocytes of the same batch. The rising phase of the current response was studied in four additional batches of oocytes originating from three additional donor animals. Similar averaged data were obtained in all cases, using two to four oocytes from each batch. Some variation of the observed current decrease from the maximal current response was found between different batches of oocytes. The average current amplitudes in a total of five batches, 30 min after the first exposure to p-PMA, were 110% (3 oocytes), 547% (5 oocytes), 156% (4 oocytes) (Fig. 21, 115% (3 oocytes), and 162% (2 oocytes) of the response before application of p-PMA. In five oocytes from four different batches of oocytes, the response after 30 min was smaller than the initial response before application of p-PMA. The reason for the observed current decrease, in spite of the repeated application of p-PMA, is not clear from our experiments. In control experiments (not shown), the amplitude of NMDAinduced currents remained constant for more than 30 min, indicating that the current decrease mentioned above is not due to a run-down of NMDA receptor channels. The stereoisomeric control substance a-PMA (100 m) that does not activate PKC at a 10-fold higher concentration than P-PMA did not affect the NMDA curfent amplitude (Fig. 2). If the nonspecific protein kinase inhibitor staurosporine (5 p.~) was applied in conjunction with 0-PMA, the stimulation was strongly reduced from about 700 to 150% (Fig. 2) of the control current amplitude (100%). Surprisingly, staurosporine alone also slightly stimulated the control response. Staurosporine at 1 PM was also able to largely suppress p-PMA effects (not shown). ~ontinuously applied OAG (2 p~) , a synthetic diacylglycerol, had very similar effects as P-PMA on NMDA-induced currents, except that the onset of action was somewhat slower (Fig. 3). The effects of a-PMA and OAG were confirmed in additional experiments.
Effect of the C-terminal Splice Domain on the Modulation by PIE-In similar experiments, the effect of p-PMA on the NRlA-NR2A receptor channel was determined. The additional C-terminal amino acids did not significantly affect the response to the agonists (not shown). Fig. 4A shows the time course of the p-PMA effect in this subunit combination. Stimulation of the control current amplitude was about half that observed with NRlC-NR2A in the same batch of oocytes (Fig. 4-4). In three such experiments, relative average stimulation of NRlA-NR2A compared with NR1C-NR2A was 54, 51, and 62%. Subunit NRlC differs from NRlA by the absence of 37 amino acids encoded by exon 21 (15) containing four consensus phosphorylation sites for PKC. O u r data show, seemingly paradoxically, that removal of these sites increases the extent of activation by PKC.
Basal ~M~A Receptor Activity-We find that PKC activates NMDA receptor channels. It may be thought that full activity of B DISCUSSION We have studied in detail the modulation of heteromeric NMDA channels with the composition NRlC-NR2A by agents that activate PKC. So far, very little is known about the modulation of heteromeric NMDA channels. A study (6) with mouse 4161, 4'162, and 4163 channels reported modulation of these channels by p-PMA in the presence of €1 and €2 but not in the the channels is revealed only after stimulation of PKC. Following this assumption, we replotted the data shown in Fig. 4 A , assigning 100% activity to the current amplitudes measured 10 min after application of p-PMA when the maximal current amplitude was reached (Fig. 4B ). Except for the initial phase, the resulting curves from the two subunit combinations overlap over the whole time course studied. The subunit combination including the subunit containing the additional amino acids displayed a basal activity of 30% of the maximal current amplitude, as opposed to the one lacking the extra amino acids, which had 15%. In two additional experiments with four oocytes each from a different donor animal, the average values were 28 and 14% and 35 and 23%, respectively. Thus, the basal activity of the two subunit combinations differs significantly (p < 0.01, pooled data from three experiments). These data indicate that the extra amino acids present in NRlA carrying the phosphorylation sites could confer a higher basal activity to the receptor.
Mutation of Putative PKC Substrate Sites on the Domain between TM3 a n d TM4-The phosphorylation sites situated on the extra amino acids close to the C terminus are obviously not essential for activation of NMDA receptor currents by PKC. We speculate that PKC consensus phosphorylation sites (24, 25) situated on the putative intracellular domain between the predicted transmembrane sequences TM3 and TM4 could be responsible for the modulation as is the case in other ligandactivated channels (26).
It has also been shown that homologous parts of the GAE%AA receptor confer modulation by protein kinase A (27) and PKC (22). Therefore, we investigated the effect of the mutations NR2A(S747A), NRlC(S658A), NRlC(S748A), and NRlC(T773A) on the modulation by activation of PKC. Mutated subunits were combined with a wild type subunit to give subunit combinations NRlC-NR2A carrying one mutation. None of the mutant receptors showed a deviant behavior of the wild type receptor (not shown). Subsequently, NR2AcS747A) was combined with NRlC, carrying one or two of the shown mutations. Again, the modulation by p-PMA was the same as in the wild type (not shown). Finally, NR2A(S747A) was combined with a NRlC subunit, carrying all three mutations. Fig. 5 shows that this mutated channel is modulated by P-PMA to the same extent as the wild type channel. A second experiment (three oocytes from a different donor animal) gave comparable results, which indicates that the chosen sites are not essential for the modulation of NMDA receptor channels by PKC.
presence of €3. The E family of subunits in mouse corresponds to the NR2 family in rat. We show that the NRlC-NR2A current is stimulated after application of OAG (Fig. 3) and stereospecifically after the application of very low concentrations of p-PMA ( Figs. 1 and 2). Stimulation by p-PMA occurs very rapidly (<<2 min) but is strongly inhibited by the simultaneous application of the nonspecific kinase inhibitor staurosporine (Fig. 2). Together with the finding that p-PMA increases phosphorylation of NMDA receptors (141, our observations suggest that the modulation may be mediated by direct effects of PKC. But, on the basis of our experiments, we cannot exclude an indirect action of PKC on NMDA receptors via other proteins.
Transient Stimulation-We observe that the stimulation by activators of PKC is transient, reaching a maximum of about 6-7-fold a t about 10 min (Fig. 2). Subsequent decline of the response was somewhat variable and reached, in some cases, amplitudes below the initial control response within 30 min. In our protocols, we ensured a continuous presence of activators of PKC. A transient stimulation of the NMDA-induced currents has recently been reported (6, 121, but, in these cases, high concentrations of p-PMA were applied initially only, followed by a washout period. Thus, it is not clear whether the transient nature of the stimulation was caused by the washout of p-PMA. Our work can only suggest some hypotheses for the mechanism leading to the transient nature of the stimulation of NMDAinduced currents by PKC. It is possible that the isofonn of PKC involved is rapidly down-regulated and the NMDA channel dephosphorylated by a protein phosphatase within the duration of our experiment. Alternatively, the overphosphorylation of NMDA receptors could inhibit the channel function. Finally, PKC could turn on another enzyme acting in an inhibitory fashion on NMDA channels. Role of the C-terminal Phosphorylation Sites-A previous biochemical experiment has shown that the NRlA subunit transfected into a permanent cell line is phosphorylated in the basal state and that this phosphorylation can be enhanced by exposure of the cells t o p-PMA (14). P-PMA-induced phosphorylation was strongly reduced after deletion of C-terminal amino acids (construct identical to NRlC) or after converting four serine residues of PKC consensus phosphorylation sites to alanine (14). Such findings strongly suggested the involvement of these amino acids in the modulation of NMDA receptors by PKC. The fact shown here that NMDA receptors containing the NRlC subunit can be functionally modulated by PKC is in itself interesting. Previously, it has been shown that homomeric channels made of subunits lacking the C-terminal amino acid domain can also be stimulated by p-PMA (12). These observations are unexpected in the light of the biochemical work (14) and show that the four mentioned serine residues are dispensable for the modulation of NMDA channels by PKC. Direct comparison of NRlA-NR2A and NRlC-NR2A channels showed that removal of the extra amino acids does not lead to loss of stimulation by PKC but, paradoxically, to an increased stimulation.
The Phosphorylation Sites Lacking in NRlC May Be Znvolved in Defining Basal Activity-As indicated in the previous section, one could assume that full activity of the NMDA channels is revealed only after stimulation of PKC and that the C-terminal splice domain is involved in defining the basal ac-tivity. It is therefore interesting to compare this result with those of a recent publication on homomeric NMDA receptors in Xenopus oocytes and their modulation by PKC (12). The basal activity can be calculated from the data on the maximal stimulation by p-PMA. In spite of the different experimental protocols, the basal activity of the NRlA channel (33%) compares quantitatively very well with the basal activity found here for NRlA-NR2A channels (30%). If our model is correct, it can be deduced from the data of Durand et al. (12) that about twothirds of the basal activity is contributed by the presence of the first and about one-third by the second C-terminal splice domain.
We do not show here whether the enhanced basal activity of the receptor containing the additional 37 amino acids is a consequence of a phosphorylation of the additional sites. If it is the case, the basal phosphorylation does not necessarily have to occur by a n isoform of PKC but could be achieved via another protein kinase recognizing the additional amino acids in NRlA. I t should be borne in mind that our data were collected in Xenopus oocytes and that the steady state of basal phosphorylation and activity could be different in neurons.
Role of the Consensus Phosphorylation Sites for PKC on the Intracellular Loops between TM3 and TM4-We chose the subunit combination NRlC-NR2A for further work. In analogy to other ligand-activated ion channels, we assumed an involvement of consensus phosphorylation sites located on the putatively intracellular regions between TM3 and TM4 of the two subunits. Three such possible phosphate-accepting amino acids on NRlC and one on NR2A were replaced by alanine. In our assays, the mutant subunit combination showed a behavior very similar to that of the wild type combination, indicating that the four amino acids are not involved in the modulation of NMDA receptors by PKC.
In conclusion, we present here a detailed study of the modulation of heteromeric NMDA receptor channels by PKC. We show that the transient nature of the stimulation is a true effect, which is not just due to the washout of stimulators of PKC. We provide evidence that the modulation can also occur in the absence of the amino acids encoded in exon 21 (7, 151, which have been shown to be phosphorylated by PKC in neurons (14). However, we put forward the notion that the extra amino acids in NRlA may help to determine the basal activity of the channel. Four predicted phosphorylation sites between TM3 and TM4 are not required for the modulation or determination of basal activity. Thus, the molecular basis of the modulation of NMDA channel by PKC remains to be established. paper by Yamakura, T., Mori, H., Shimoji, K., and Mishina, M. (Bio-Note Added in Proof-mer final submission of this manuscript, a chem. Biophys. Res. Commun. 196, 1537-1544(1993) came to our attention. They report similar findings concerning the lack of a role of the C-terminal domain of the mouse 51 subunit.