Protein kinase C (PKC) potentiates NMDA receptors in hippocampal, trigeminal, and spinal neurons. Although PKC phosphorylates the NMDA receptor subunit NR1 at four residues within the C terminal splice cassette C1, the molecular mechanisms underlying PKC potentiation of NMDA responses are not yet known. The present study examined the role of Ca2+ in PKC potentiation of recombinant NMDA receptors expressed in Xenopus oocytes. We found that Ca2+ influx through PKC-potentiated NMDA receptors can further increase the NMDA response (“Ca2+ amplification”). Ca2+amplification required a rise in intracellular Ca2+concentration at or near the intracellular end of the channel and was independent of Ca2+-activated Cl− current. Ca2+ amplification depended on extracellular Ca2+ concentration during NMDA application and not during PKC activation. Ca2+amplification was reduced by the membrane-permeant Ca2+-chelating agent BAPTA-AM. Mutant receptors with greatly reduced Ca2+ permeability did not exhibit Ca2+ amplification. Receptors containing the NR1 N-terminal splice cassette showed more Ca2+amplification, possibly because of their larger basal current and therefore greater Ca2+ influx. Contrary to expectation, splicing out the two C-terminal splice cassettes of NR1 enhanced PKC potentiation in a manner independent of extracellular Ca2+. This observation indicates that PKC potentiation does not require phosphorylation of the C1cassette of the NR1 subunit. PKC potentiation of NMDA receptorsin vivo is likely to be affected by Ca2+ amplification of the potentiated signal; the degree of amplification will depend in part on alternative splicing of the NR1 subunit, which is regulated developmentally and in a cell-specific manner.
The NMDA class of glutamate receptors plays a critical role in synaptic plasticity, formation of neuronal circuitry, and excitotoxicity (for review, see Choi and Rothman, 1990; Constantine-Paton, 1990; Bliss and Collingridge, 1993). Protein phosphorylation and dephosphorylation are thought to be important mechanisms of regulation of synaptic activity. The PKC-activating phorbol ester 12-O-tetradecanoyl phorbol-13-acetate (TPA) enhances whole-cell currents (Gerber et al., 1989; Aniksztejn et al., 1992; Chen and Huang, 1992) and channel open probability (Chen and Huang, 1992) of NMDA receptors in hippocampal, trigeminal, and spinal neurons. TPA also potentiates NMDA responses inXenopus oocytes expressing rat brain mRNA (Durand et al., 1992; Kelso et al., 1992; Urushihara et al., 1992) and recombinant NMDA receptors (Durand et al., 1992, 1993; Kutsuwada et al., 1992; Yamazaki et al., 1992; Mori et al., 1993). Agonists of μ opioid receptors (Chen and Huang, 1991), phosphoinositol-coupled metabotropic glutamate receptors (Aniksztejn et al., 1991, 1992; Kelso et al., 1992; Shen et al., 1995), and muscarinic acetylcholine receptors (Markram and Segal, 1990) potentiate neuronal and recombinant NMDA receptors via activation of PKC. PKC phosphorylates specific residues in the C1splice cassette in the cytoplasmic C terminus of the NMDA receptor subunit NR1 (Tingley et al., 1993). Phosphorylation at these sites may regulate receptor interactions with the cytoskeleton and clustering at the membrane (Ehlers et al., 1995) but may decrease PKC potentiation (Durand et al., 1993; Sigel et al., 1994).
Ca2+ modulates NMDA receptor activity by several mechanisms. Ca2+ reduces single-channel conductance by binding in the ion channel (Jahr and Stevens, 1993). Ca2+ influx reduces open probability without change in single-channel conductance (producing “slow inactivation”) (Legendre et al., 1993; Rosenmund et al., 1995). Binding of Ca2+–calmodulin to the cytoplasmic C terminus decreases open probability and may contribute to Ca2+-dependent inactivation (Ehlers et al., 1996). High extracellular Ca2+ was reported to increase responses of recombinant NMDA receptors expressed inXenopus oocytes (Koltchine et al., 1996).
NMDA receptors are assembled from NR1 and NR2 subunits (for review, seeNakanishi, 1992; Westbrook, 1994; Mori and Mishina, 1995). The NR1 subunit is encoded by a single gene, and alternative RNA splicing gives rise to eight possible splice variants (Sugihara et al., 1992). NR2 subunits are of four types, NR2A–D, encoded by four different genes (Meguro et al., 1992; Monyer et al., 1992). NMDA receptors are likely to be NR1/NR2 heteromeric complexes in the nervous system (Sheng et al., 1994; Behe et al., 1995; Blahos and Wenthold, 1996). Although injection of NR1 subunit mRNA into oocytes leads to formation of functional receptors with different properties dependent on the splice variant (Durand et al., 1993; Zhang et al., 1994; Zukin and Bennett, 1995), oocytes may provide an NR2 subunit. Oocytes provide a necessary subunit for two other heteromeric channels (Buller and White, 1990;Hedin et al., 1996).
In the present study, we analyzed the effect of extracellular Ca2+ on PKC potentiation of recombinant NMDA receptors expressed in Xenopus oocytes. PKC potentiation of specific splice variants was greater when measured in 1 mmCa2+ than when measured in 1 mmBa2+. The increase in potentiation observed in extracellular Ca2+ required that the Ca2+ be present during opening of the NMDA receptor, not during activation of PKC. The findings are consistent with a process of Ca2+-dependent amplification in which Ca2+ enters through the potentiated NMDA receptor and binds to its inner face or a closely associated molecule to increase channel open time or conductance. Modulation of PKC potentiation of neuronal NMDA receptors by extracellular Ca2+ and by alternative splicing could have important roles in synaptic plasticity and in formation of neural circuitry.
MATERIALS AND METHODS
Site-directed mutagenesis. Site-directed mutations were made with the oligonucleotide-directed in vitromutagenesis system, version 2 (Amersham, Arlington Heights, IL) as described (Zheng et al., 1994). In brief, NR1011 and NR1100 (nomenclature of NR1 splice variants according toDurand et al., 1993) were subcloned into the pBluescript SK(−) vector and used to transform the Escherichia coli strain DH5αF′IQ (Life Technologies, Gaithersburg, MD). Single-stranded DNA template was rescued with M13K07 helper phage (Bio-Rad, Hercules, CA). To make the channel mutants NR1011(N598R) and NR1100(N619R), we engineered a single codon substitution into the cloned NR1011 and NR1100 cDNAs to generate subunits carrying an arginine in place of asparagine in the second membrane domain M2. The following oligonucleotide was used for introduction of this change into the coding sequence 5′-CCTTCCCCAATGCCGGAGCGGAGCAGGACGCC-3′. To engineer the channel mutant NR1111(N619R), a BsmI–BamHI fragment containing exons 21 and 22 was shuttled from wild-type NR1111 into the mutant NR1100(N619R) cDNA.
To introduce N1 insert mutations with reduced positive charges in the NR1111 receptor, aBsmI–BamHI fragment containing exons 21 and 22 was shuttled from wild-type NR1111 into the mutant M1, M2, and M3 (NR1100) cDNAs (Zheng et al., 1994). Mutant NR1111 receptors were as follows: N1–1 (K192A, K193A, and R194A), N1–2 (R207A, R208A, and K211A), and N1–3 (K192A, K193A, R194A, R207A, R208A, and K211A). Codon substitutions were confirmed by DNA sequencing across mutationally altered and ligation regions with Sequenase version 2.1 (United States Biochemical, Cleveland, OH).
RNA synthesis. NR1011 cDNA was a gift from Dr. S. Nakanishi (Kyoto University, Kyoto, Japan); NR1111 cDNA was a gift from Dr. V. R. Anantharam (University of Massachusetts, Worcester, MA) and subcloned to pBluescript SK(−) vector; NR1100 cDNA was cloned in this laboratory (Durand et al., 1992). Mutant NR1 receptors were generated as described above. NR2A was a gift from Dr. M. Mishina (Niigata University, Niigata, Japan). To generate templates for transcription, circular plasmid cDNAs were linearized with NotI (wild-type and mutant NR1011, NR2A) or BamHI (wild-type and mutant NR1100, NR1111). Transcription reactions were performed with T7 and T3 polymerase (Ambion MEGAscript transcription kit, 4 hr at 37°C) in the presence of capped analog or a mMessage mMachine transcription kit (2 hr at 37°C).
Electrophysiological experiments in Xenopusoocytes. Adult female Xenopus laevis (Xenopus I, Ann Arbor, MI) were anesthetized by immersion in ice water or 0.15% aminobenzoic acid ethyl ester, and oocytes were isolated and prepared as described (Kushner et al., 1988, 1989). Selected stage V and VI oocytes were injected with in vitro transcribed RNA (∼20 ng/cell; for heteromeric receptor expression, NR1 and NR2 were mixed in a ratio of 1:3) and maintained at 18°C in culture buffer (in mm: 103 NaCl, 2.5 KCl, 2 MgCl2, 2 CaCl2, and 5 HEPES, pH 7.5). Three to 7 d after injection, oocytes were clamped at −60 mV in Mg2+-free Ca2+ Ringer’s solution (in mm: 116 NaCl, 2.0 KCl, 1.0 CaCl2, and 10 HEPES, pH 7.2), unless otherwise specified, with a two-electrode voltage-clamp amplifier (Dagan or Axon Instruments). NMDA (300 μm) together with glycine (10 μm) was bath-applied to elicit responses. Current amplitude was measured at the minimum immediately after the initial peak (indicated byarrows in sample records in Figs.1, and 5). In some experiments, 1.0 mm CaCl2 was replaced by 1.0 mmBaCl2 (Ba2+ Ringer’s solution), or the concentration of CaCl2 was changed. In experiments involving 5 and 10 mm Ca2+ Ringer’s solution, 10 and 20 mm Na+-gluconate was substituted for equimolar NaCl to maintain constant concentrations of Na+ and Cl−.
TPA (1 mm in 10% DMSO) was dissolved in either Ca2+ or Ba2+ Ringer’s solution at a final concentration of 100 nm and bath-applied to oocytes for 10 min, which achieves near-maximal PKC potentiation. The TPA solution was washed out before NMDA application. Acetoxymethyl ester of bis-(o-aminophenoxy)-ethane-N,N,N′,N′,-tetraacetic acid (BAPTA-AM; 50 mm stock solution in DMSO) was diluted to 0.25 mm in Ca2+ Ringer’s solution immediately before each experiment. Niflumic acid in 0.5 mstock solution in ethanol was diluted to 0.5 mm in Ca2+ Ringer’s solution immediately before each experiment. The extent of PKC potentiation varied from batch to batch of oocytes. Therefore, comparisons are within single batches of oocytes or between batches with the same degree of potentiation of a given receptor subtype. Data represent mean ± SEM. Statistical analyses were performed by one-way ANOVA (GraphPad Prism version 3.0).
Extracellular Ca2+ affects PKC potentiation of recombinant NMDA receptors
To examine the effect of extracellular Ca2+ on PKC potentiation of NMDA receptors, we used oocytes expressing recombinant NMDA receptors. We applied NMDA (300 μm with 10 μm glycine) to oocytes clamped at −60 mV in Ba2+ or Ca2+ Ringer’s solution before and after application of TPA. We focused on three NR1 splice variants that differ in their potentiation by PKC: NR1011, which is prominent in forebrain and exhibits a relatively low level of potentiation by PKC; NR1111, which is also prominent in forebrain and exhibits a moderate level of PKC potentiation; and NR1100, which is prominent in midbrain and exhibits a high level of potentiation by PKC (Fig. 1 A) (Durand et al., 1993). This choice of splice variants permitted us to evaluate the action of Ca2+ on receptors with and without the N-terminal splice cassette N1 by comparison of NR1011 and NR1111 receptors and on receptors with and without the C-terminal splice cassettes C1 andC2 by comparison of NR1100 and NR1111 receptors. (When C2 is spliced out, a new unrelated reading frame is opened, which encodes an alternative cassette,C2′, of 22 amino acids before a new stop codon is reached.)
In 1 mm Ca2+ Ringer’s solution, TPA (100 nm) potentiated NMDA responses of NR1011-, NR1111-, and NR1100-injected oocytes to 3.1 ± 0.2-fold (n = 19), 6.2 ± 0.4-fold (n = 21), and 12.2 ± 0.8-fold (n= 16) of control NMDA responses, respectively (Fig.1 A,B). In 1 mm Ba2+Ringer’s solution, NMDA responses of all three splice variants lacked the initial peak current observed in the presence of extracellular Ca2+ (Fig. 1 A, bottom row). In Ba2+, TPA potentiated NMDA responses of NR1011-, NR1111-, and NR1100-injected oocytes to 2.3 ± 0.3-fold (n = 8), 2.8 ± 0.3-fold (n = 8), and 5.7 ± 0.4-fold (n = 10) of the control responses, respectively (Fig. 1 A,B). Thus, PKC potentiation of the NR1 splice variants containing N1(NR1111 and NR1100) was markedly greater in Ca2+ than in Ba2+, whereas potentiation of the splice variant NR1011, which lacks N1, was only slightly greater in Ca2+ than in Ba2+ Ringer’s solution.
We show below that the greater PKC potentiation recorded in Ca2+ requires a rise in Ca2+ at the intracellular opening of the channel during responses to NMDA after activation of PKC by TPA; we term this effect “Ca2+ amplification.” The magnitude of the Ca2+ amplification is determined by comparison with potentiation measured in Ba2+, which we assume to have no amplifying action. Because potentiation is measured as the ratio of responses measured in the same Ringer’s solution after and before TPA application, differences in single-channel conductances in Ca2+ and Ba2+ Ringer’s solution should not be a factor, except insofar as PKC potentiation may change channel permeability.
In Ba2+ Ringer’s solution, potentiation of NR1011 receptors was virtually identical to that of NR1111 receptors and about half of that of NR1100 receptors (Fig. 1 A,B). Thus, PKC potentiation in Ba2+ is little affected by theN1 cassette but is greater for the NR1 splice variants lacking the C-terminal splice cassettes C1 and C2(see Durand et al., 1993).
Native NMDA receptors are likely to be heteromeric receptors containing both NR1 and NR2 subunits (Sheng et al., 1994; Behe et al., 1995;Blahos and Wenthold, 1996). To determine whether (1) alternative splicing of NR1 affects PKC potentiation of heteromeric receptors, and (2) there is Ca2+ amplification of their responses, similar experiments were performed on oocytes expressing NR2A and one of the three NR1 subunits, NR1011, NR1111, and NR1100. In 1 mmCa2+ Ringer’s solution, TPA potentiated NMDA responses of heteromeric NR1011/NR2A, NR1111/NR2A, and NR1100/NR2A receptors to 3.6 ± 0.5-fold (n = 4), 8.2 ± 0.3-fold (n = 4), and 13.7 ± 1.8-fold (n = 4) of control NMDA responses, respectively (Fig.1 C, open bars). In Ba2+ Ringer’s solution, TPA potentiated NMDA responses of NR1011/NR2A, NR1111/NR2A, and NR1100/NR2A receptors to 3.6 ± 0.4-fold (n = 4), 4.7 ± 0.6-fold (n = 5), and 7.0 ± 0.7-fold (n = 5) of the control responses, respectively (Fig. 1 C, filled bars). These findings indicate that PKC potentiation of heteromeric NR1/NR2A receptors is affected by alternative splicing and exhibits Ca2+ amplification of the potentiated responses.
The dependence of PKC potentiation on the extracellular Ca2+ concentration was investigated over the range of 0.1–10 mm extracellular Ca2+ (Fig.2). At higher extracellular Ca2+ concentrations (5–10 mm), the control currents of each of the NR1 splice variants (but not necessarily Ca2+ flux through the channel) were reduced compared with those observed at 0.1 mmCa2+ (Fig. 2 A; also see Jahr and Stevens, 1993). However, PKC potentiation was markedly increased for both receptors (Fig. 2). Much of the increase in potentiation of NR1011 was a result of decrease in control responses rather than increase in the potentiated responses; however, if single-channel conductance was unaffected by potentiation, open probability must have been increased at elevated Ca2+. PKC-induced potentiation of NR1100 responses was decreased at 10 mm Ca2+, relative to that at 5 mm Ca2+; the reduction may reflect greater Ca2+-induced inactivation. These findings suggest that Ca2+ amplification of PKC potentiation increases with Ca2+ concentration over a considerable range. Because potentiation of NR1011 is increased by increasing external Ca2+, bothN1 containing and N1 lacking splice variants may share a common mechanism of Ca2+ amplification.
Ca2+ amplification is not a result of Ca2+-activated Cl− currents
Xenopus oocytes express endogenous chloride channels that are activated by cytoplasmic free Ca2+ [to generate I Cl(Ca)]. We used the Cl− channel blocker niflumic acid to test for possible contamination of the Ca2+ amplification of NMDA evoked responses by I Cl(Ca). In Figure3 A niflumic acid at 0.5 mm, a concentration that largely blocksI Cl(Ca) evoked by Ca2+ entry mediated by ionophore or through voltage-dependent Ca2+ channels (Leonard and Kelso, 1990; White and Aylwin, 1990), blocked 17% of the peak (measured to baseline) and 15% of the steady-state NMDA evoked currents in oocytes expressing NR1100/NR2A receptors at 1 mm external Ca2+ and −60 mV holding potential (first pair of records). Niflumic acid acted rapidly, and we saw no difference between its action whether or not it was applied 20 sec before NMDA plus niflumic acid; the action was also rapidly reversible (see White and Aylwin, 1990). After application of TPA, the same concentration of niflumic acid blocked 21% of the peak and 8% of the plateau phase of the potentiated NMDA responses (second pair of records). In another batch of oocytes, PKC potentiation was to 7.7 ± 0.4-fold (n = 3) and 8.2 ± 0.5-fold (n = 3) of control in the absence and presence of niflumic acid, respectively (p > 0.05), indicating a negligible contribution of I Cl(Ca)to the measured PKC potentiation. The persistence of an initial peak in niflumic acid suggests that there is a contribution of receptor desensitization to this peak.
We also tested for a contribution of I Cl(Ca) by measuring PKC potentiation near the Cl− reversal potential, E Cl. E Cl is approximately −20 mV in the oocytes, as shown by reversal of a component of the early peak at this potential (Fig. 3 B). At −12 mV, the NMDA response has three phases, an inward current through the NMDA receptors, a somewhat delayed outward current ascribable toI Cl(Ca), and a later plateau current that is again through the NMDA receptors. The data in Figure 3 Aindicate that the plateau phase has littleI Cl(Ca) component. At −18 mV, the same components are seen, but the net current at the time of the peak inI Cl(Ca) is inward. At potentials of −20 mV or more negative, the initial NMDA and I Cl(Ca)peaks are not distinguishable in these records. The presence nearE Cl of an initial NMDA peak greater than the plateau indicates that desensitization contributes to the peak at more negative voltages, as is also indicated by the persistence of the peak in niflumic acid (Fig. 3 A). We compared the PKC potentiation observed at −20 and −60 mV of NMDA splice variants that do and do not show Ca2+ amplification in 1 mmCa2+. In responses of oocytes expressing NR1011, which do not show Ca2+amplification, the degree of potentiation at −20 mV did not differ significantly from that at −60 mV (2.9 ± 0.3 at −20 mV versus 3.1 ± 0.7 at −60 mV), indicating thatI Cl(Ca) does not contribute significantly to PKC potentiation of these receptors (Fig. 3 C,D). Responses of oocytes expressing NR1111 and NR1100 exhibited some reduction (35–45%) in PKC potentiation at −20 mV compared with −60 mV (p < 0.05 for NR1111; p < 0.01 for NR1100), which is consistent with reduced Ca2+ influx attributable to the reduced driving force at the more depolarized potential. PKC potentiation of NR1111 receptors did remain greater than that of NR1011 receptors (p < 0.005), although potentiation of the two receptors was essentially the same in Ba2+ (Fig. 1 B). Because Ca2+ amplification was not abolished at −20 mV, we conclude that it is not a result of Ca2+-activated Cl− current. Although splicing in the C1and C2 cassettes reduced potentiation in Ba2+ at −60 mV, this effect was not present at −20 mV.
Ca2+ amplification requires extracellular Ca2+ during opening of the NMDA channel, not during TPA application
To determine whether extracellular Ca2+ acts during TPA application and/or in a subsequent step (e.g., opening of the NMDA channel), we substituted 1 mmBa2+ for 1 mm Ca2+during (1) TPA application; (2) both TPA and NMDA application; and (3) NMDA application only. For the oocytes illustrated in Figure4, A and B, application of TPA in Ba2+ Ringer’s solution potentiated the NMDA response measured in Ca2+Ringer’s solution (Fig. 4 B) to a level similar to that observed when TPA was applied in Ca2+ Ringer’s solution (Fig. 4 A; to 7.7 and 8.3 times control, respectively). In Figure 4, C and D, application of TPA in Ba2+ Ringer’s solution potentiated the NMDA response measured in Ba2+ (Fig. 4 C) to a level similar to that observed when TPA was applied in Ca2+ Ringer’s solution (Fig. 4 D; to 5.2 and 4.3 times control, respectively) In Figure4 E, control responses were obtained in Ca2+ and in Ba2+ Ringer’s solution; then TPA was applied in Ba2+, and a test response was obtained. Next the oocyte was transferred to Ca2+ Ringer’s solution, and a test response was obtained within 20 sec. The degree of potentiation measured in Ca2+ was greater than that for Ba2+ (14- vs 6-fold) and as great as it would have been if the entire experiment had been performed in Ca2+. These findings indicate that the Ca2+ amplification requires Ca2+during the response to NMDA rather than during the action of TPA.
To examine the time course of onset and reversal of Ca2+ amplification, we first treated an oocyte expressing NR1100 receptors with TPA in 1 mmBa2+ Ringer’s solution. We then applied NMDA in 1 mm Ba2+, changed the solution to NMDA in 1 mm Ca2+, and then returned to NMDA in Ba2+ (Fig. 4 F). Transfer of the oocyte from Ba2+ to Ca2+ resulted in a rapid rise to the potentiation level observed in Ca2+; the response was characterized by a peak, followed by a decay to a plateau value. Transfer of the oocyte from Ca2+ to Ba2+ resulted in a rapid return to near the potentiated level in Ba2+. These findings suggest that Ca2+ amplifies the NMDA response, and that onset and decay of the Ca2+amplification are too rapid to be resolved within the limitations of perfusion in the oocyte system.
Ca2+ amplification requires a rise in intracellular free Ca2+
To test whether a rise in intracellular Ca2+ is required for Ca2+ amplification of NMDA responses, we measured TPA potentiation with and without pretreatment of oocytes with the membrane-permeant Ca2+ chelator BAPTA-AM. In Figure 5 A, PKC potentiation was to 12.5 times control in Ca2+ Ringer’s solution and to 6.8 times control in Ba2+ Ringer’s solution. Pretreatment with BAPTA-AM for 20 min had little effect on steady-state currents in Ca2+ or in Ba2+(although it did reduce the initial peak in Ca2+), but after BAPTA, PKC potentiation of NR1100 receptors measured in Ca2+ was reduced to 7.8 times control, a value near that obtained in Ba2+ (6.4 times control; Fig. 5 B). To determine whether BAPTA-AM inhibits activation of PKC by TPA, BAPTA-AM was applied to another oocyte after treatment with TPA (Fig. 5 C). Under these conditions, PKC potentiation was similar in Ca2+ and in Ba2+, i.e., to 5.5 and 5.7 times control, respectively, and was close to that in Ba2+ without BAPTA treatment (Fig.5 A). In an additional batch of oocytes expressing NR1100, we compared potentiation in controls and after BAPTA treatment, as in Figure 5, A and B. BAPTA treatment reduced potentiation measured in Ca2+ to a level not significantly different from that in Ba2+ without BAPTA (Fig. 5 D). Thus, BAPTA treatment blocks Ca2+ amplification. BAPTA also reduced PKC potentiation measured in Ba2+compared with the control potentiation measured in Ba2+, but the effect was not significant (p > 0.05). These findings indicate that this degree of BAPTA treatment reduces or blocks Ca2+amplification of the PKC-potentiated response but has little effect on PKC action or on the response measured in Ba2+. Longer BAPTA treatment reduced potentiation in Ba2+to a greater extent, which may reflect a requirement for cytoplasmic Ca2+ in TPA action.
NMDA receptors with reduced Ca2+ permeability do not exhibit the Ca2+ amplification
To characterize further the effect of Ca2+ on PKC-potentiated responses, we engineered the mutations N598R (forN1-lacking receptors) and N619R (forN1-containing receptors) in the M2 (channel lining or P) region of each of the three NR1 splice variants (Fig.6 A). Introduction of an Arg in place of N598 greatly reduces Ca2+permeability of recombinant NMDA receptors (Burnashev et al., 1992;Kawajiri and Dingledine, 1993; Sakurada et al., 1993). In Ca2+ Ringer’s solution, current amplitudes in oocytes expressing the NMDA receptor channel mutant were reduced compared with those of wild-type receptors and lacked the initial peak (Fig. 6 B; currents were measured in oocytes from the same batch injected with the same amount of RNA). PKC potentiation of the mutant NR1011(N598R) receptors did not differ significantly from that of wild-type NR1011 receptors (2.6 ± 0.4- vs 3.1 ± 0.2-fold in Ca2+Ringer’s solution; Fig. 6 C, wild-type data from Fig.1 B). However, PKC potentiation of NR1111(N619R) receptors was markedly reduced relative to that of the corresponding wild-type receptor (3.2 ± 0.4- vs 6.2 ± 0.4-fold). PKC potentiation of NR1100(N619R) receptors was also reduced relative to that of the corresponding wild-type receptor (5.1 ± 0.5- vs 12.2 ± 0.8-fold). In all three cases, PKC potentiation of the mutant NMDA receptor measured in Ca2+ Ringer’s solution was similar to that of the corresponding wild-type receptor measured in Ba2+Ringer’s solution (Fig. 6 C). Thus, the M2 channel mutation abolished the Ca2+ amplification.
PKC potentiation and Ca2+ amplification are independent of agonist concentration
To investigate further whether PKC potentiation and Ca2+ amplification are reduced under conditions of reduced Ca2+ influx, we compared PKC potentiation of NR1011 and NR1100 receptors at 5 and 300 μm NMDA (with 10 μm glycine). For NR1011 receptors, the current amplitude at 5 μm NMDA was 55 ± 8% of that at 300 μm NMDA; for NR1100 receptors, the current amplitude at 5 μm NMDA was 19 ± 4% of that at 300 μm NMDA. The degree of PKC potentiation, however, was similar at the two agonist concentrations (4.4 ± 0.4 at 5 μm NMDA vs 3.7 ± 0.3 at 300 μm NMDA for NR1011 receptors; 9.8 ± 1.9 at 5 μmNMDA vs 8.7 ± 0.4 at 300 μm NMDA for NR1100 receptors) (Fig. 7). The finding that PKC potentiation is unchanged at low agonist concentration suggests that Ca2+ amplification is the same at low and high mean current levels per channel. It follows that Ca2+ amplification requires neither spatial summation of Ca2+ influx from neighboring channels nor temporal summation of Ca2+ influx from successive periods when the receptor is fully occupied by agonists during NMDA and glycine application. (Temporal summation might occur between successive openings in a burst.) Ca2+inactivation of NMDA receptors also occurs at low agonist concentration (Legendre et al., 1993) and thus also appears not to require summation of Ca2+ influx from successive binding events or neighboring channels.
Neutralization of positively charged residues within the N1 splice cassette reduces PKC potentiation of NMDA receptors
The N1 splice cassette alters agonist affinity, current amplitude, spermine and Zn2+ potentiation, and proton inhibition of recombinant NMDA receptors (Durand et al., 1992,1993; Hollmann et al., 1993; Zhang et al., 1994; Zheng et al., 1994;Traynelis et al., 1995), and many of these effects are reversed by neutralization of positive charges in N1. To determine whether the greater PKC potentiation observed for splice variants containing the N1 cassette is affected by the presence of the six positively charged residues, we substituted the neutral amino acid alanine for each of three positively charged residues at either end of the N1 splice cassette (N1–1 and N1–2) or for all six positively charged residues (N1–3; Fig.8 A). The three mutant receptors exhibited reduced current amplitudes relative to that of the wild-type NR1111 receptor (data not illustrated). In Ca2+ Ringer’s solution, PKC potentiation of N1–1, N1–2, and N1–3 receptors was reduced markedly relative to that of wild-type NR1111 receptors (to 3.5 ± 0.4-, 2.9 ± 0.3-, and 4.1 ± 0.5-fold of control responses, respectively) and was similar to potentiation of wild-type NR1011receptors (3.1 ± 0.2-fold; Fig. 8 B). Because neutralization of the positive charges in the N1 insert reduces current amplitude (defined as the NMDA response obtained after injection of a constant amount of RNA) (Hollmann et al., 1993; Zheng et al., 1994), reduction in Ca2+ influx could account for the observed decrease in PKC potentiation.
PKC potentiation of recombinant NMDA receptors is amplified by Ca2+ influx
In the present study, alternative splicing and extracellular Ca2+ were shown to affect PKC potentiation of recombinant NR1 receptors expressed in Xenopus oocytes. ForN1-containing receptors, PKC potentiation was greater when measured in Ca2+ Ringer’s solution than when measured in Ba2+ Ringer’s solution. The data are consistent with a mechanism in which Ca2+ entering through the activated receptor binds to the channel or a closely associated molecule to increase channel opening; this amplification would be greater after PKC action. Contamination by Ca2+-activated Cl− currents was minimal as discussed below. Ca2+ was effective when present during the application of NMDA and had no effect during activation of PKC by TPA. The evidence that the amplification requires a rise in intracellular Ca2+ is the following: (1) amplification is increased by increasing extracellular Ca2+ (Fig. 2); (2) amplification is reduced by depolarizing to reduce the driving force for Ca2+(Fig. 3); (3) amplification is reduced in mutant receptors with reduced Ca2+ permeability (Fig. 6); and (4) amplification is prevented by applying a membrane-permeant form of the Ca2+ chelator BAPTA (Fig. 5). In addition, the splice variants containing the N1 cassette exhibit greater Ca2+ amplification (Fig. 1), possibly because they generate larger currents, which could be associated with greater Ca2+ influx. The reduced amplification observed after neutralization of positive charges in the N1 cassette (Fig. 8) is consistent with this interpretation, because these mutations reduce current amplitude. Furthermore, theN1-lacking receptor, NR1011, which has a smaller current amplitude, exhibited Ca2+amplification at increased extracellular Ca2+; thus, the N1 cassette is not required for Ca2+amplification.
The degree of Ca2+ amplification was evaluated by comparison with responses in Ba2+. Ba2+ is unlikely to have any amplifying effect, because potentiation in Ba2+ was similar to that in Ca2+ after BAPTA (Fig. 5) and to that of the mutant receptor with reduced Ca2+ permeability (Fig. 6). In the simplest model, PKC does not affect the Ca2+amplification directly but increases channel open time and/or conductance, thereby increasing Ca2+ influx and the amount of amplification. PKC may also increase sensitivity of the receptor to intracellular Ca2+. Alternatively, PKC action may increase single-channel conductance by reducing divalent ion binding in the channel (as has been suggested for PKC action on Mg2+ block of NMDA receptors in trigeminal neurons;Chen and Huang, 1992); if Ca2+ ions were affected more than Ba2+ ions, it could account for the greater potentiation in Ca2+. Because BAPTA blocks the Ca2+ amplification after PKC, it is unlikely that PKC increases channel conductance. Another possibility might be that PKC increases the plateau response by reducing Ca2+-dependent inactivation. This mechanism is unlikely, because the action of BAPTA indicates that there is little Ca2+-dependent inactivation either before or after TPA application. (Moreover, the inactivation would have to be rapid enough to be undetected in the rising phase of the response to bath applied agonists.) A final possibility to be considered is that Ca2+ causes insertion of new receptor molecules. This mechanism also appears unlikely because of the rapid onset and reversibility of the Ca2+ amplification (Fig.4).
Ca2+ amplification is not an artifact of Ca2+-activated chloride current
A number of observations indicate thatI Cl(Ca) does not contribute significantly to PKC potentiation or Ca2+ amplification, as measured in this study. (1) The chloride channel blocker, niflumic acid, had only a small effect on the amplitude of the plateau responses and did not reduce PKC potentiation measured in 1 mmCa2+ Ringer’s solution at −60 mV (Fig.3 A). Leonard and Kelso (1990) reported a greater effect of niflumic acid on NMDA responses in oocytes expressing rat brain mRNA, but this difference may have resulted from use of brain mRNA, which may encode additional Ca2+-activated Cl− channels and from recording at −80 mV in 2 mm external Ca2+, which would increaseI Cl(Ca). (2) Ca2+amplification is present at −20 mV, a value near ECl in oocytes (Fig. 3 B,C). At this holding potential, the chloride current is small, but Ca2+ amplification persists, although reduced, presumably because of the reduced driving force for Ca2+ influx. (3) We measure in oocytes a reversal potential for recombinant NMDA receptors of about −5 mV (data not illustrated; also see Hollmann et al., 1993). This value is close to that reported for neuronal NMDA receptors (cf. Mayer and Westbrook, 1987), which indicates that there is littleI Cl(Ca) contribution. (4) The degree of Ca2+ amplification was not changed when Ca2+ influx was reduced by reducing agonist concentration (Fig. 7). Also, the degree of Ca2+amplification was the same for oocytes generating large and small currents in response to saturating NMDA, reflecting differences in the level of receptor expression (data not illustrated). I(Ca)Cl would be expected to decrease with decrease in NMDA elicited currents.
Are Ca2+ influx and Ca2+amplification during NMDA application regenerative?
The mechanism of the proposed Ca2+amplification is as yet unknown, but single-channel studies and mutational analysis will undoubtedly further elucidate the molecular basis. Ca2+ may act directly on the channel protein or on a closely associated accessory protein expressed by theXenopus oocyte. Candidate proteins are calmodulin and Ca2+- and calmodulin-dependent kinases, implicated in NMDA-dependent LTP. The rapid rise in current amplitude on transfer of the oocyte from Ba2+ to Ca2+Ringer’s solution and rapid decay on transfer from Ca2+ to Ba2+ Ringer’s solution argues against a covalent modification, such as phosphorylation or dephosphorylation and in favor of a direct action of Ca2+ (or Ca2+–calmodulin complex) on the cytoplasmic face of the receptor. Ca2+–calmodulin binds to a site in theC1 cassette and a site in C0, the region between TM4 and C1 of the NR1 subunit. However, binding to these sites decreases channel open time in heteromeric receptors with NR2A expressed in HEK 293 cells (Ehlers et al., 1996).
The Ca2+ amplification should tend to be regenerative. However, in our experiments the amplification was independent of agonist concentration (Fig. 7). This finding, together with our observation that Ca2+ amplification is independent of the level of receptor expression (data not shown), suggests that there was neither spatial summation of Ca2+ influx from neighboring channels nor temporal summation of Ca2+ influx from successive bursts of openings caused by periods of full occupancy of the receptor by NMDA and glycine. It follows that under these conditions there was no regenerative response.
As noted above, neuronal receptors can show Ca2+-dependent inactivation (Legendre et al., 1993;Rosenmund et al., 1995). In recombinant receptors expressed in HEK 293 cells, Ca2+-dependent inactivation is affected by the NR2 subunit, being greatest in NR1/NR2A receptors, less in NR1/NR2B receptors, and virtually absent in NR1/NR2C receptors (Krupp et al., 1996). In Xenopus oocytes, increase in extracellular Ca2+ was reported to increase NMDA response amplitude comparable to the Ca2+ amplification proposed here; however, unlike the results in Figure 2, the response amplitude was not reduced at higher concentrations of Ca2+ (Koltchine et al., 1996).
Recombinant NMDA receptors expressed in Xenopus oocytes exhibit a late, slowly rising phase in NMDA responses, particularly evident in N1-containing NR1 receptors, in the presence or absence of activators of PKC (Figs. 1 A, 5) (Koltchine et al., 1996). This slow, Ca2+-dependent response is more likely to occur at higher agonist concentrations and after PKC potentiation. It may represent the release of Ca2+from intracellular pools, although the persistence of the response in BAPTA argues against this mechanism.
Does PKC potentiation involve phosphorylation of the NR1 protein?
PKC phosphorylates specific serine residues in the NR1011 receptor expressed transiently in HEK 293 cells (Tingley et al., 1993). The four identified series are contained in theC1 cassette, although our electrophysiological findings indicate that splicing out the C1 and C2cassettes increases PKC potentiation (Fig. 1) (Durand et al., 1992,1993). Thus, PKC potentiation may not require phosphorylation of the receptor itself, and phosphorylation of C1 may even reduce PKC potentiation. In agreement, mutation of the phosphorylatable serines to alanines increases TPA potentiation (Zhang et al., 1996). It is then reasonable to suggest that there is another molecule associated with NR1, phosphorylation of which is responsible for the potentiation. An obvious possibility is NR2A and/or B, which apparently are phosphorylated by PKC (Hall and Soderling, 1997; Leonard and Hell, 1997)
Functional significance of the Ca2+ amplification
Although our experiments were performed on recombinant NMDA receptors expressed in Xenopus oocytes, Ca2+ amplification is likely to occur after PKC potentiation of neuronal NMDA receptors. As noted above, PKC potentiates NMDA receptors in neurons from hippocampus (Aniksztejn et al., 1992), trigeminal nucleus (Chen and Huang, 1992), and spinal cord (Gerber et al., 1989). In our study, extracellular Ca2+ and alternative splicing were shown to affect PKC potentiation. These results together with the finding that expression of NR1 splice variants is temporally and spatially regulated (Laurie and Seeburg, 1994; Zukin and Bennett, 1995; Paupard et al., 1997) provide a mechanism whereby NMDA signals can be amplified in specific neuronal populations at times of enhanced synaptic activity. Thus, PKC potentiation of NMDA responses may have a significant impact on cellular events including induction of long-term potentiation and synaptogenesis. Ca2+ amplification, if present in neurons, may have particularly important actions in small structures such as dendritic spines, where influxes through neighboring receptor channels would be more likely to interact.
This work was supported by National Institutes of Health Grants NS 20752 (R.S.Z.) and NS 07412 (M.V.L.B.). M.V.L.B. is the Sylvia and Robert S. Olnick Professor of Neuroscience. We thank J. Zavilowitz for technical assistance. We are grateful to Drs. S. Nakanishi, V. R. Anantharam, and M. Mishina for providing the NR1011, NR1111, and mouse ε-1 (equivalent to rat NR2A) cDNAs, respectively. We thank Dr. T. Opitz and R. Araneda for helpful comments on this manuscript.
Correspondence should be addressed to Dr. R. Suzanne Zukin, Department of Neuroscience, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461.