- Email: [email protected]
The role of Ca2+ entry via synaptically activated NMDA receptors in the induction of long-term potentiation
Neuron,
Vol. 11, 817423,
November,
1993, Copyright
0 1993 by Cell Press
The Role of Ca*+ Entry via Synaptically Activated NMDA Receptors in the Induction of Long-Term Potentiation David J. Perkel,*+ Jeffrey J. Petrozzino,* Roger A. Nicoll,* and John A. Connor* *Departments of Pharmacology and Physiology University of California San Francisco, California 94143-0450 *Roche Institute of Molecular Biology Roche Research Center Nutley, New Jersey 07110-1199
Summary Influx of Ca*+ through the NMDA subtype of glutamate receptor is widely accepted as a trigger for many forms of neural plasticity. However, direct support for this model has been elusive, since indirect activation of dendritic voltage-sensitive Ca*+ channels is difficult to exclude. We have optically measured synaptically induced changes in cytoplasmic free Ca*+ concentration in pyramidal cell dendrites in hippocampal slices. Steady postsynaptic depolarization to the synaptic reversal potential eliminated the effect of voltage-sensitive Ca*+ channels. Under these conditions, synaptically induced Ca*+ transients were observed, which were blocked by the NMDA receptor antagonist APV. In addition, the magnitude of LTP was diminished when induced with the postsynaptic membrane held at progressively more positive potentials. LTP could be completely suppressed at potentials near +lOO mV. These results provide important experimental support for a role for Ca*+ influx through NMDA receptors in synaptic plasticity. Introduction One of the most important distinguishing features of the N-methyl-o-aspartate (NMDA) subtype of glutamate receptor is its abilityto signal to the postsynaptic cell the temporal coincidence of pre-and postsynaptic activity (Bourne and Nicoll, 1993). The demonstrated Ca*+ permeability of the NMDA receptor (Ascher and Nowak, 1988; Mayer et al., 1987; Mayer and Westbrook, 1987) allows for transmission of a spatially specific chemical signal to the postsynaptic cytoplasm in the vicinity of the activated receptors, and this appealing feature has been proposed as critical for such forms of plasticity as long-term potentiation (LTP) in the hippocampus (Nicoll et al., 1988; Wigstrijm and Gustafsson, 1985). While data consistent with such a role have come from a variety of sources, it has been difficult to demonstrate a rise in postsynaptic Ca*+ concentration clearly attributable to synaptic activation of NMDA receptors. For example, synaptically evoked Ca*+ tran+Present address: Division of Technology, Pasadena,
of Biology, 216-76, California California 91125.
Institute
sients that are sensitive to the NMDA receptor antagonist 2-aminc&phosphonovaleric acid (APV) have been observed (Bliss and Collingridge, 1993; Miyakawa et al., 1992; Regehr and Tank, 1990, 1992). As has been pointed out, however, it is problematic in these studies to exclude unequivocally the activation of voltagesensitive Ca*+ channels (Miyakawa et al., 1992; Regehr and Tank, 1992). It is therefore difficult to determine what fraction of the observed APV-sensitive Ca*+ signal was a direct result of influx through the NMDA receptor, and what fraction was due to NMDA receptor-mediated depolarization and thus indirect activation of voltage-sensitive Ca2+channels. Moreover, the available data strongly suggest that the majority of Ca*+ entry is due to activation of Ca*+ channels. For example, an experiment designed to reduce the contribution of voltage-sensitive Ca2+ channels by bathing cells in medium containing no added Mg*’ and hyperpolarizing the postsynaptic cell yielded very small synaptically evoked rises in postsynaptic Ca*+ concentration (Miyakawa et al., 1992). We have taken a different approach to eliminate Ca*+ entry via voltage-sensitive Ca*+ channels. By blocking postsynaptic K+ channels with Cs+ included in the intracellular solution, we have been able to depolarize the postsynaptic neuron to, or beyond, the reversal potential for excitatory synaptic events (approximately 0 mV). Thus, activation of afferent fibers led either to no change in the postsynaptic membranepotentialortoa hyperpolarization. Underthese conditions, Ca*+ imaging revealed synaptically activated dendritic Ca*+ transients that were largely and reversibly blocked by APV. In some cases, we observed that the amplitude of the Ca*+ transient was reduced with shifts of the membrane potential toward the Ca2+ equilibrium potential. These results provide strong evidence that synaptic activation of the NMDA receptor can lead to Ca*+ influx through the receptor channel andcause a rise in postsynaptic Ca*+ concentration. While this manuscript was in preparation, we received a manuscript in which evidence is presented regarding Ca *+ influx, and a similar conclusion was reached (Alford et al., 1993). To address whether such a Ca*+ signal is necessary for induction of LTP, we performed a separate set of experiments in which wedelivered LTP-inducing stimuli while holding the postsynaptic neuron at different membrane potentials. When afferent fibers were stimulated at low frequency and the postsynaptic cell was held near 0 mV, robust LTPwas elicited. If the holding potential was near +I00 mV, LTPwas blocked. At intermediate potentials between +I00 mV and 0 mV, intermediate degrees of potentiation were observed. These results extend the previous results of Malenka et al. (1988) and provide strong support for an essential role for Ca*+ influx through the NMDA receptor in the induction of LTP.
Neuron
818
DEPOLARIZATION T O 0 mV
Figure 1. Effect of Steady Depolarization on Intracellular Free Ca2+ Concentration in a CA3 Pyramidal Cell (A) Fluorescence of injected fura-2,380 nm excitation wavelength. (B) Ca2+ concentration based on fluorescence ratio (360 nmi 380 nm excitation) when the membrane potential was -70 mV. (C) Peak Ca’+ levels after depolarization to approximately 0 mV. The recovery of free Ca’+ concentration upon maintained depolarization of the same cell is illustrated in Figure 2A. Bar, 25 urn.
-I 50 nM
I 950 nM
SYNAPTIC
I 2.7 pt.A
STIMULATION Stimulation
Control
AT 0 mV
Figure 2. NMDA Receptor-Dependent Ca’+ Changes in the Apical Dendrites of a CA3 Pyramidal Neuron in Response to Stimulation of AssociationallCommissural Fibers during Steady Depolarization to 0 mV
Recoverv
20 mV/ e
The same cell as in Figure 1 was used. (A) Ca’+ levels 13.5 min after the beginning of steady depolarization to 0 mV. Bar, 25 urn. (B) Ca2’ responseto a train of synaptic stimuli delivered to associationallcommissural afferent fibers. The proximal apical dendritic region, the soma, and the basal dendritic region did not show Ca2’ changes during the stimulus. (C) Recovery, 30 s after the stimulus train. (D-F) Same as in (A)-(C), except that the bathing solution contained 100 uM DL-APV. Traces below images (A)(C) and (D)-(F) aresimultaneous membrane potential recordings, verifying that the cell was being held at the excitatory reversal potential.
j!-.. 1
SYNAPTIC
set
l;o
Sl TIMULATION Stimulation
Control
20mVI
I 6 pt.4
-b 1 set
3bO
d0
80b nM
IN APV AT 0 mV Recoverv
Role of Ca2+ Influx through 819
NMDA
Receptors
in LTP Induction
Results Conventional intracellular recordings were obtained from CA1 and CA3 pyramidal cells in guinea pig hippocampal slices. To measure free cytoplasmic Ca*+ concentration, cells were filled with the Ca2+ indicator furadissolved in CsCI, which blocks K+ channels and facilitates depolarization. Final indicator concentration was 200-250 FM (see Experimental Procedures). Ca2+ levels were determined ratiometrically from paired digitally acquired fluorescence images using excitation illumination of wavelength 360 nm and 380 nm (Grynkiewicz et al., 1985). When an impaled pyramidal cell was adequately filled with fura(Figure IA) and held at -70 mV, it displayed low resting Ca*+ concentrations, in the range of 50-100 nM (Figure IB). The cell was then steadily depolarized to approximately 0 mV, and Ca 2+ levels rose dramatically (to greater than 1 PM) throughout the cell during the initial portion of the depolarization, as the membrane fired repeated Ca*+ action potentials (Figure IQ Accompanying stabilization of the membrane potential, the Ca2+ concentration fell to a level not dramatically higher than the resting level observed at -70 mV (Figure 2A). When the Ca2+ concentration was stable, a train of stimuli (50 Hz; Is) was delivered to afferent fibers in the stratum radiatum. Starting l-2 s before the stimulus train and continuing for 5-10 s afterward, image pairs were collected at approximately 3.5 Hz. In portions of the dendrites, Ca2+ concentration rose briefly, often in a punctate fashion, and recovered within several seconds (Figures 2A-2C). These transients were repeatable several times within a single bout of steady depolarization. To test whether this rise of Ca2+ concentration depended on the activation of NMDA receptors, the depolarization and stimuli were repeated following addition of the NMDA receptor antagonist APV (100 PM) to the bathing solution. As illustrated in Figures 2D2F, stimulus trains in the presence of APV led to little detectable change in Ca2+ concentration. In 3 cells (including the one shown in Figure 2), the steady-state Ca2+ concentration was reduced in the presence of APV. It is possible that the APV sensitivity of steadystate Ca2+ concentration reflects Ca2+ influx through tonically activated NMDA receptors (Sah et al., 1989). In addition, this could reflect a gradual decline of Ca*+concentration during maintained depolarization, which was also observed in the absence of APV. In another pyramidal cell, we selected a small portion of the apical dendritic field that showed a clear Ca2+ transient and measured the Ca2+ concentration in that region as a function of time, so that the time course of the change in Ca2+ concentration in the dendrites could be examined quantitatively. Figure 3 (open squares) illustrates the change in Ca2+ concentration in response to a tetanic stimulus while the postsynaptic cell was held at +I0 mV. A rapid rise in Ca2+ concentration was observed, and the concentration decayed to baseline values over several seconds.
The membrane potential during the stimulus was recorded to ensure that the synaptic stimulus did not cause depolarization of the membrane. As illustrated by the trace shown in the inset of Figure 3 (trace a), the synaptic response was reversed, indicating that voltage-sensitive Ca2+ channels could not have contributed subtantially to the observed rise in intracellular Ca*+. The effect of APV on the Ca2+ response to the same stimulus (Figure 3, trace b) in the same cell is shown in Figure 3 (closed circles). The synaptic stimulation led to only a small increase in Ca2’ concentration, which measured 5% of the control response. The effect of APV was partially reversible upon removal from the perfusion medium, recovering to 37% of the control response, as seen in Figure 3 (closed squares, trace c). The average synaptically induced rise in intradendritic Ca2+, measured with the postsynaptic membrane held between 0 mV and +20 mV, was 1400 f 700 nM (mean + SEM; n = 4 CA1 cells and 4 CA3 cells). In neurons in which the effects of APV were studied, there was a 85% +- 8% reduction in the Ca2+ transient in the presence of the antagonist (n = 8). In all cases in which synaptic stimulation was attempted following washout of APV, a partial recovery of the Ca*+ transient (to 71% + 16% of control) was observed (n = 4). The average tetanus-induced change in membrane potential was similar in control conditions (-4.3 k 1.5 mV) and in the presence of APV (-5.5 +
CALCIUM RESPONSE TO SYNAPTIC STIMULATION AT +lO mV
10mV
WASH
L IS
I
i! I
0
1 .o
2.0
3:.
1
%(,
Time(s)
Figure 3. Quantitative Analysis of Regions of the Apical Dendrite of a Different CA3 Neuron Showing Ca2+ Transients Open squares represent the time course of the cytoplasmic Ca2+ concentration in a small (2.5 x 2.5 pm) region of the dendrite following a tetanic stimulus (1 s, 100 Hz) delivered to afferent fibers. Closed circles represent the Ca2+ concentration following the addition of 100 PM DL-APV to the bath. Recovery of CaZ+ transient following removal of APV from the bath is indicated by the open squares. The postsynaptic cell was steadily held at +I0 mV prior to the tetanus. Insets illustrate the negative-going response of membrane potential, indicating that the synaptic stimulation (indicated by heavy horizontal bar) led to hyperpolarization (a, control; b, APV; c: after removal of APV).
Neuron 820
A5 l
Paired at +I0 mV
in 5 of the 7 cells in which tetanic stimulation was delivered at different somatic potentials, the Ca2+ transient was clearly decreased at more positive voltages. Between 0 mV and around +50 mV, there was, on average, a 40% + 16% reduction in the amplitude of the transient (p < 0.05; n = 7). This effect is consistent with a reduced driving force on Ca2+ entry associated with extreme depolarization and provides further evidence that the observed Ca2+ transients are due to influx through the NMDA receptors.
k. : ’ ?
..
- Paired at +70 mV
Pairing -
I
2 0
10 Time (min)
I 30
20
Bl.r-TK= -I 1
50
pA
20 ms
Figure 4. Blockadeof LTP When Low Frequency Synaptic Were Paired with Extreme Depolarization to +70 mV
Stimuli
Two groups of afferents were stimulated alternately in the baseline period, during which the postsynaptic cell was held at -80 mV. The cell was depolarized to an extreme potential and one pathway, the “test” pathway, was stimulated 50 times at 0.5 Hz. The membrane potential was shifted to the synaptic reversal potential (approximately +I0 mV), and the other pathway, the “reference” pathway, was given the same stimulus. The membrane potential was shifted back to -80 mV, and alternate stimuli were given to the two pathways to assess the presence of LTP. (A) Circles indicate EPSC amplitudes for the reference pathway. Triangles represent EPSC amplitudes for the test pathway. (B) Sample EPSCs (averages of 10 consecutive sweeps) in the test pathway taken before and after the pairing (indicated by 1 and 2 in [A]). (C) Sample EPSCs from the reference pathway before and after the pairing.
Role of Ca2+ in LTP To address whether Ca2’ influx through the NMDA receptor is necessary for the induction of LTP in the CA1 region, we took advantage of the increased control of membrane potential available with whole-cell recording. We reasoned that if Ca2+entry were necessary, then a reduced driving force on Ca2+ during activation of NMDA receptors should result in little or no LTP. During a baseline recording period (holding potential -80 mV), excitatory postsynaptic currents (EPSCs) were evoked alternately in two independent afferent pathways at 0.1 Hz. Stimulation was stopped and the potential was shifted to a test value between -30 mV and i-100 mV. When the holding current had stabilized,50stimuliweredeliveredtooneofthepathways at 0.2-0.7 Hz. Because LTP may be variable under whole-cell recording conditions, it was important to test whether the cell was capable of exhibiting potentiation. Therefore, after the 50 pulses were delivered to the test pathway, the membrane potential was shifted to the synaptic reversal potential (approxi-
ioo-
80 -
P 2 0 8 F 'yfj a, cc
2.3 mV). These data indicate that activation of synapses can directly increase the cytoplasmic CaZ+ concentration in a fashion directly dependent on NMDA receptors. We observed that the steady-state Ca2+ concentration measured at depolarized potentials depended on the holding potential. In each of 7cells examined, the intradendritic Ca*+ concentration was lower when the membrane potential was held near +50 mV than near 0 mV. Steady-state dendritic Ca2+ concentration averaged 330 + 120 nM near 0 mV and was reduced by 22% k 9% near +50 mV (p < 0.001; n = 7). These data provide evidence that Ca2+ influx contributes to the steady-state Ca2+ concentration. Although it was not possible to clamp the dendritic membrane potential during the synaptic stimulation,
60-
40-
20 -
90 0
I 50
100
Membrane potential (mV) Figure 5. Voltage Dependence of Induction of LTP by Pairing Low Frequency Stimulation with Depolarization The amount of potentiation in the test pathway of each experiment was normalized by the amount of potentiation observed in the reference pathway, which was stimulated while the cell was held at 0 mV. With extreme depolarization to approximately +I00 mV, little potentiation was observed. With intermediate potentials between 0 mV and +I00 mV, there was a gradual reduction in the amount of potentiation observed. Relative potentiation values were as follows: near +I00 mV, 5% + 23%, n = 3; near +60 mV, 25% + 13%, n = 8; near f30 mV, 57% + 12%, n = 5; near -35 mV, 47% * 34%, n = 3.
Role of Ca’+ Influx through 821
NMDA
Receptors
in LTP Induction
mately +I0 mV). In each experiment, regardless of the test potential, the second pathway was given 50 stimuli with the membrane potential held at the reversal potential. Results from an example experiment are illustrated in Figure 4. Summary data based on 19 experiments are illustrated in Figure 5. For each experiment in which clear potentiation was observed in the reference pathway (at least 30% enhancement measured IO-20 min after pairing), the degree of potentiation observed in the test pathway was expressed as a percentage of the potentiation observed in the reference pathway. The resulting relative potentiation value was 0% if no potentiation occurred in the test pathway and 100% if the test pathway showed the same amount of potentiation as the reference pathway. The EPSC amplitude in the reference pathway was increased, on average, to 226% + 22% (n = 19) of control. As seen in Figure 5, when the membrane potential was held near +I00 mV, synaptic stimulation resulted in littleor no potentiation (5% + 23%; n = 3; p < 0.03 versus reference pathway). The amount of enhancement was gradually reduced as the potential was shifted from 0 mV to +I00 mV.
Discussion This study focused on two ing the induction of LTP: sure the postulated entry NMDA receptor channel; try a trigger for LTP?
related questions concernfirst, can one directly meaof Capthrough the synaptic and second, is this Ca*+ en-
Measurement of Ca2+ Transients during Synaptic Activation of NMDA Receptors Weobserved transient increases in dendriticfreeCa*+ concentration in CA1 and CA3 pyramidal cells in response to stimulation of Schaffer collateral (CAI) or associationallcommissural (CA3) synapses while the postsynaptic membrane potential was at or beyond the synaptic reversal potential of approximately0 mV. These changes in Ca*+ were greatly reduced in the presence of the NMDA receptor antagonist APV and were partially restored upon washout of APV. These data indicate that synaptic stimulation can lead to NMDA receptor-mediated rises in intracellular Ca2+ without a contribution from voltage-sensitive Ca*+ channels. A crucial aspect of the work presented here was to design experiments in which entry of Ca*+via voltagesensitive Ca*+ channels could not contribute to the observed synaptically induced Ca*+ transients. Voltare known to be present age-sensitive Ca *+ channels at high densities in the dendrites (Llinds, 1988; Westenbroek et al., 1990) and can generate large Ca*+ transients independent of synaptic stimulation (Miyakawa et al., 1992; Muller and Connor, 1991; Miiller et al., 1992). As noted by Miyakawa et al. (1992), it is difficult to exclude voltage-sensitive Ca2+ channels as the source for an APV-sensitive Ca*+ signal observed un-
der normal conditions. To avoid this problem, we took advantage of the fact that when the postsynaptic membrane is held at the reversal potential for a synapse, activation of the synapse does not, by definition, change the membrane potential. Since, in our experiments, weobserved thatthe stimulus-induced change of membrane potential was hyperpolarizing or null during the rise in Ca2+ concentration, we conclude that the activation of voltage-sensitive Ca2+ channels did not contribute to the response. A different approach to this problem was taken by Alford et al., (1993)who relied primarily on adequate spatial control of membrane potential and complete washout of dendritic Ca*+ currents to reduce Ca*+ influx through voltage-dependent Ca2+ channels. While the nearly complete blockade of Ca*+ transients by APV indicates a strong dependence on activation of NMDA receptors, it is possible that not all of the Caz+ that we measured actually represents Ca*+ influx through this receptor. For example, we do not know the degree to which Ca*+-induced Ca2+ release from intracellular stores (Henzi and MacDermott, 1992) may have amplified the NMDA receptor-mediated Ca*+ influx. However, our data do constrain the possible contribution made by activation of metabotropic glutamate receptors or Ca*+-permeant nonNMDA receptors (Kohler et al., 1993; Murphy and Miller, 1988; Ozawa et al., 1988); since APV, which does not affect these other receptors, blocked the observed Ca*+ transients by 85%, we conclude that the non-NMDA ionotropic and metabotropic receptors can account for only a small fraction of the response. Several factors have allowed us to observe a synaptically mediated Ca2+ signal. First, we used depolarization to unmask NMDAcurrents, which has two principal advantages over removal of Mg*+ from the bathing solution: Virtually complete (>95%) removal of the Mg2’ blockadeof the NMDA receptor can be achieved when the cell is held at +I0 mV (Hestrin et al., 1990); additionally, if the decay time constant of the NMDA receptor-mediated conductance change is voltage dependent (Konnerth et al., 1990), significantly greater Ca*+ influx would be expected to occur at positive potentials. A second factor is that we used the dihydropyridine Ca*+ channel antagonist nimodipine to reduce the steady-state Ca2+ influx through noninactivating Ca2-+ channels, thus maintaining the backour experimental ground Ca *+ levels low. Therefore, conditions tended both to increase the transient signal and to reducethe background Ca*+concentration. It is important to point out some limitations of the present imaging study. First, we were not able to resolve dendritic spines and therefore do not know whether the Ca*+ signals that we observed reflected averages of signals from numerous spines or spillover from spines to dendritic shafts. Another limitation involves control of membrane potential. Even under voltage-clamp conditions, it would have been impossible to clamp the membrane potential at the synaptic
Neuron a22
membrane during high frequency repetitive stimulation. The present imaging experiments were performed under current-clamp conditions, and thus we were able to obtain only the approximate voltage dependence of the Ca2+ influx. The problem of electrotonic control likely had less impact on the determination of the voltage dependence of LTP induction, since low frequency stimulation and somatic voltageclamp conditions were used. We therefore cannnot compare quantitatively the dependence on membrane potential of Cal+entryand LTP induction. Nonetheless, shifts of membrane potential affected both the size of the Ca2+ signals and the degree of LTP in the same direction. Finally, the time course of the Ca2+ transients that we observed may not necessarily reflect the true time course that would occur under normal conditions; there could be some slowing effect due to the buffering of Ca2+ by fura(Regehr and Tank, 1992), prolonged Ca2+ entry through altered NMDA receptor kinetics (Konnerth et al., 1990), or alteration in the activity of the Na+/Ca2+ exchanger by depolarization (Blaustein, 1988). Nonetheless, these limitations do not affect the primary conclusions from our imaging experiments: that synaptic stimulation gives rise to increased intracellular Ca2+ in a fashion directly dependent on Ca2’ influx through NMDA receptors. The Voltage Dependence of LTP Induction Several studies have shown that the presence of postsynaptic Ca2’ buffers blocks LTP (Lynch et al., 1976; Malenka et al., 1988). It is not known, however, whetherthere is simply a requirement for some”baseline” level of Ca2+ to enable the biochemical processes for LTP, or whether a specific rise in intracellular Ca2+ is necessary. Our results provide evidence for adirect instructive role for Ca2+, since LTP was blocked simply by providing extreme depolarization, which we postulate would greatly reduce Ca2+ influx. In addition, it should be mentioned that this experiment does not formally rule out the possibility that Na’ entry is important in LTP, since extreme depolarization would be expected to reduce the influx of any permeant ion whose reversal potential is greater than +50 mV. In combination with the available data concerning Ca2+ buffers, however, the most likely explanation for our data is that Ca2+ influx through NMDA receptors serves as a trigger for at least some of the biochemical processes subserving the enhanced synaptic transmission of LTP. It appears clear, however, that some otherfactorprovided bysynapticstimulation mayalso play an important role in leading to persistent potentiation (Kullmann et al., 1992). It is not known whether LTP at individual synapses occurs in a graded or all-or-none fashion. The graded reduction in the degree of potentiation observed with increasingly positive membrane potential could reflect a graded reduction in the amount of potentiation at each synapse. Alternatively, it could reflect a reduced probability of some all-or-none process such
that only a subset of synapses were potentiated, but to the usual degree. In summary, our results demonstrate directly the accumulationofdendriticCa2+in thepostsynapticceil via entry through NMDA receptors. In addition the present data indicate thatthis Ca2’ source likely serves as an instructive, rather than simply a permissive, signal for generating LTP, since blocking this source of Ca2+ with extreme depolarization blocked LTP. Experimental
Procedures
Hippocampal slices (300-500 pm thick) were prepared from 3- to 5-week-old Hartley guinea pigs using a vibratome. Slices were stored in a holding chamber at the interface between recording solution and humidified atmosphere consisting of 95% 02, 5% COz. Recordings were carried out at room temperature (22OC-24OC) with the slice submerged beneath continuously superfusing bathing medium solution. In some of the imaging experiments, an interface recording configuration was used. This extracellular solution contained 119 m M NaCI, 2.5 m M KCI, 4 m M MgSO+ 4 m M CaC&, 26 m M NaHCOp, 1 m M NaHP02, nglucose, and 50-100 pM picrotoxin. Imaging experiments were performed on CA? and CA3 pyramidal neurons. In these experiments, the dihydropyridine Cal’ channel antagonist nimodipine was added to the bathing medium (IO PM; 0.1% dimethylsulfoxide) to reduce the steady-state CaZ’concentrationduringmaintaineddepolarization. Intracellutips conlar recordings were made using microelectrodeswhose tained 5-10 m M fura- (potassium salt, Molecular Probes, Eugene OR) and 0.2-1.5 M CsCl and were backfilled with 2.5 M CsC!. Total electrode resistance was 40-80 MO. Cells were dye loaded with steady hyperpolarizing current injection (0.5-1.0 nA) and Cs+ loaded with brief depolarizing current pulses (100-200 ms; 0.2-0.7 nA; every 5-10 s) for IO-20 min. Indicator concentration in the cells was approximately200-250 PM, as estimated by comparing the 360 nm excited fluorescence of the microinjected neurons with that of neurons filled using whole-cell recording pipettes containing 200 PM fura-2. Intracellular signals were amplified using an Axoclamp-2A (Axon Instruments, Burlingame CA) in the discontinuous current-clamp mode (1-5 kHz switching frequency). Membrane potential was recorded on magnetic tape, subsequently digitized at 10 kHz, and low pass filtered at 500 Hz for display. Synaptic stimuli were delivered via bipolar stainless steel electrodes (FHC, Brunswick, ME); tetani consisted of 1 s trains at 50 or 100 Hz. Cells were imaged from the top surfaceofthe slice using an upright microscope (Zeiss Axioskop) and long working distance 40x water-immersion or 20x dry objectives (Zeiss). A CCD camera system (Series 200, Photometrics Ltd., Tucson, AZ) similar to ones previously described (Connor,1986;MijllerandConnor,1991)wasusedtoacquiredigitized images of fura-2. The camera was operated in frame-transfer mode to acquire image pairs at 360 and 380 nm excitation wavelengths. Exposure times for single frames were 100-200 ms, correspondingtoacquisition ratesof2-4Hzper imagepair. Fluorescence data were background corrected and spatially filtered (3 x 3 pixels) before ratio images were constructed. Ca*‘concentrations were then determined from ratio images (Crynkiewicz et al., 1985). For illustration of images, a mask for the ratios was constructed by importing a single-wavelength image (resting cell, 380 nm) into Adobe Photoshop and obtaining contours of the most prominent dendritic structures. Whole-cell recordings were obtained using the blind technique (Blanton et al., 1989; Coleman and Miller, 1989). The intracellular solution contained 130 m M cesium gluconate, 15 m M CsCI, IO m M NaCI, 10 m M HEPES, 0.2 m M EGTA, 2 m M ATP, and 0.1 m M GTP. Nimodipine was not present in experiments designed to assess LTP. Input resistance and access resistance were monitored throughout the experiment; data were discarded if significant changes in passive properties of the cell or recording occurred. Signals were amplified using an Axoclamp-
Role of Ca* Influx through 823
NMDA
Receptors
in LTP Induction
ID amplifier, filtered at 1 kHz, and digitized at 2 kHz using a TLI data acquisition system and modified pCL4MP software (all from Axon Instruments).
References Alford, S., Frenguelli, B. C., Schofield, J. G., and Collingridge, G. L. (1993). Characterization of CaZ+ signals induced in hippocampal CA1 neurons by the synaptic activation of NMDA receptors. J. Physiol. 469, 693-716. Ascher, P., and Nowak, L. (1988). The role of divalent the N-methyl-o-aspartate responses of mouse central in culture. J. Physiol. 399, 247-266.
cations in neurones
Blanton, M. C., Lo Turco, J. J., and Kriegstein, A. R. (1989). Whole cell recording from neurons in slices of reptilian and mammalian cerebral cortex. j. Neurosci. Meth. 30, 203-210. Blaustein, M. P. (1988). Calcium transport rons. Trends Neurosci. 77, 438-443.
and buffering
Bliss, T. V. P., and Collingridge, G. L. (1993). A synaptic of memory: long-term potentiation in the hippocampus. 361, 31-39. Bourne, H. R., and Nicoll, R. (1993). Molecular coincident synaptic signals. Cell 72/Neuron
in neumodel Nature
machines integrate 70 (suppl.), 65-75.
Coleman, P. A., and Miller, R. F. (1989). Measurement of passive membrane parameters with whole-cell recording from neurons in the intact amphibian retina. J. Neurophysiol. 67, 218-230. Connor, J. A. (1986). Digital imaging of free calcium changes and of spatial gradients in growing processes in single mammalian CNS cells. Proc. Natl. Acad. Sci. USA 83, 6179-6183. Grynkiewicz, C., Poenie, M., and Tsien, R. Y. (1985). A new generation of CaZ+ indicatorswith greatly improved fluorescence properties. J. Biol. Chem. 260, 3440-3450. Henzi, V., and MacDermott, A. B. (1992). Characteristics function of Ca*+- and inositol 1,4,5-trisphosphatereleasable stores of Ca2+ in neurons. Neuroscience 46, 251-273.
and
Hestrin, S., Nicoll, R. A., Perkel, D. J., and Sah, P. (1990). Analysis of excitatory synaptic action in the rat hippocampus usingwhole cell recording from thin slices. J. Physiol. 422, 203-225. Kohler, M., Burnashev, N., Sakmann, B., and Seeburg, P. H. (1993). Determinants of Ca* permeability in both TM1 and TM2 of high affinity kainate receptor channels: diversity by RNA editing. Neuron 70,491-500. Konnerth, A., Keller, B. U., Ballanyi, K., and Yaari, Y. (1990). Voltage sensitivity of NMDA-receptor mediated postsynaptic currents. Exp. Brain Res. 87, 209-212. Kullmann, D. M., Perkel, D. J., Manabe, T., and Nicoll, R. A. (1992). Calcium entry via postsynaptic voltage-sensitive calcium channels can transiently potentiate excitatory synaptic transmission in the hippocampus. Neuron 9, 1175-1183. Llinas, R. R. (1988). The intrinsic of mammalian neurons: insights tion. Science 242, 1654-1664.
electrophysiological properties into central nervous system func-
Lynch, G., Gribkoff, V. K., and Deadwyler, S. A. (1976). Long-term potentiation is accompanied by a reduction in dendritic responsiveness to glutamic acid. Nature 263, 151-153. Malenka, R. C., Kauer, J. A., Zucker, R. J., and Nicoll, R. A. (1988). Postsynaptic calcium is sufficient for potentiation of hippocampal synaptic transmission. Science 242, 81-84. Mayer, M. L., and Westbrook, C. L. (1987). Permeation and block of N-methyl-o-asparticacid receptor channels bydivalent cations in mouse cultured central neurones. J. Physiol. 394, 501-527. Mayer, M. L., MacDermott, A. B., Westbrook, G. L., Smith, S. J., and Barker, J. L. (1987). Agonist- and voltage-gated calcium entry in cultured mouse spinal cord neurons under voltage clamp measured using arsenazo Ill. J. Neurosci. 7, 3230-3244. Miyakawa, H., Ross, W. N., Jaffe, D., Callaway, J. C., Lasser-Ross, N., Lisman, J. E., and Johnston, D. (1992). Synaptically activated increases in Ca2+ concentration in hippocampal CA1 pyramidal
cells are primarily 9, 1163-1173.
due to voltage-gated
Ca*+ channels.
Neuron
Miiller, W., and Connor, J. A. (1991). Dendritic spines as individual neuronal compartments for synaptic Ca2+ responses. Nature 354, 73-76. Miiller, W., Petrozzino, 1. I., Griffith, L. C., Danho, W., and Connor, J. A. (1992). Specific involvement of Ca2+-calmodulin kinase II in cholinergic modulation of neuronal responsiveness. J. Neurophysiol. 68, 2264-2269. Murphy, S. N., and Miller, R. J. (1988). A glutamate receptor regulates Ca2+ mobilization in hippocampal neurons. Proc. Natl. Acad. Sci. USA 85, 8737-8741. Nicoll, R. A., Kauer, J. A., and Malenka, excitement in long-term potentiation.
R. C. (1988). The current Neuron 7, 97-103.
Ozawa, S., Nakamura, T., and Yuzaki, M. (1988). Cation permeabilitychangecaused by L-glutamate in cultured rat hippocampal neurons. Brain Res. 443, 85-94. Regehr, W. G.,andTank, D. W. (1990). Postsynaptic tor-mediated calcium accumulation in hippocampal dal cell dendrites. Nature 345, 807-810.
NMDA recepCA1 pyrami-
Regehr, W. C., and Tank, D. W. (1992). Calcium concentration dynamics produced by synaptic activation of CA1 hippocampal pyramidal cells. J. Neurosci. 72, 4202-4223. Sah, P., Hestrin, S., and Nicoll, R. A. (1989). Tonic activation NMDA receptors by ambient glutamate enhances excitability neurons. Science 246, 815-818.
of of
Westenbroek, R. E., Ahlijanian, M. K., and Catterall, W. A. (1990). Clustering of L-type Ca2+ channels at the base of major dendrites in hippocampal pyramidal neurons. Nature 347, 281-284. Wigstrom, H., and Custafsson, B. (1985). On long-lasting potentiation in the hippocampus: a proposed mechanism for its dependence on coincident pre-and postsynaptic activity. Acta Physiol. Stand. 723, 510-522.
Vol. 11, 817423,
November,
1993, Copyright
0 1993 by Cell Press
The Role of Ca*+ Entry via Synaptically Activated NMDA Receptors in the Induction of Long-Term Potentiation David J. Perkel,*+ Jeffrey J. Petrozzino,* Roger A. Nicoll,* and John A. Connor* *Departments of Pharmacology and Physiology University of California San Francisco, California 94143-0450 *Roche Institute of Molecular Biology Roche Research Center Nutley, New Jersey 07110-1199
Summary Influx of Ca*+ through the NMDA subtype of glutamate receptor is widely accepted as a trigger for many forms of neural plasticity. However, direct support for this model has been elusive, since indirect activation of dendritic voltage-sensitive Ca*+ channels is difficult to exclude. We have optically measured synaptically induced changes in cytoplasmic free Ca*+ concentration in pyramidal cell dendrites in hippocampal slices. Steady postsynaptic depolarization to the synaptic reversal potential eliminated the effect of voltage-sensitive Ca*+ channels. Under these conditions, synaptically induced Ca*+ transients were observed, which were blocked by the NMDA receptor antagonist APV. In addition, the magnitude of LTP was diminished when induced with the postsynaptic membrane held at progressively more positive potentials. LTP could be completely suppressed at potentials near +lOO mV. These results provide important experimental support for a role for Ca*+ influx through NMDA receptors in synaptic plasticity. Introduction One of the most important distinguishing features of the N-methyl-o-aspartate (NMDA) subtype of glutamate receptor is its abilityto signal to the postsynaptic cell the temporal coincidence of pre-and postsynaptic activity (Bourne and Nicoll, 1993). The demonstrated Ca*+ permeability of the NMDA receptor (Ascher and Nowak, 1988; Mayer et al., 1987; Mayer and Westbrook, 1987) allows for transmission of a spatially specific chemical signal to the postsynaptic cytoplasm in the vicinity of the activated receptors, and this appealing feature has been proposed as critical for such forms of plasticity as long-term potentiation (LTP) in the hippocampus (Nicoll et al., 1988; Wigstrijm and Gustafsson, 1985). While data consistent with such a role have come from a variety of sources, it has been difficult to demonstrate a rise in postsynaptic Ca*+ concentration clearly attributable to synaptic activation of NMDA receptors. For example, synaptically evoked Ca*+ tran+Present address: Division of Technology, Pasadena,
of Biology, 216-76, California California 91125.
Institute
sients that are sensitive to the NMDA receptor antagonist 2-aminc&phosphonovaleric acid (APV) have been observed (Bliss and Collingridge, 1993; Miyakawa et al., 1992; Regehr and Tank, 1990, 1992). As has been pointed out, however, it is problematic in these studies to exclude unequivocally the activation of voltagesensitive Ca*+ channels (Miyakawa et al., 1992; Regehr and Tank, 1992). It is therefore difficult to determine what fraction of the observed APV-sensitive Ca*+ signal was a direct result of influx through the NMDA receptor, and what fraction was due to NMDA receptor-mediated depolarization and thus indirect activation of voltage-sensitive Ca2+channels. Moreover, the available data strongly suggest that the majority of Ca*+ entry is due to activation of Ca*+ channels. For example, an experiment designed to reduce the contribution of voltage-sensitive Ca2+ channels by bathing cells in medium containing no added Mg*’ and hyperpolarizing the postsynaptic cell yielded very small synaptically evoked rises in postsynaptic Ca*+ concentration (Miyakawa et al., 1992). We have taken a different approach to eliminate Ca*+ entry via voltage-sensitive Ca*+ channels. By blocking postsynaptic K+ channels with Cs+ included in the intracellular solution, we have been able to depolarize the postsynaptic neuron to, or beyond, the reversal potential for excitatory synaptic events (approximately 0 mV). Thus, activation of afferent fibers led either to no change in the postsynaptic membranepotentialortoa hyperpolarization. Underthese conditions, Ca*+ imaging revealed synaptically activated dendritic Ca*+ transients that were largely and reversibly blocked by APV. In some cases, we observed that the amplitude of the Ca*+ transient was reduced with shifts of the membrane potential toward the Ca2+ equilibrium potential. These results provide strong evidence that synaptic activation of the NMDA receptor can lead to Ca*+ influx through the receptor channel andcause a rise in postsynaptic Ca*+ concentration. While this manuscript was in preparation, we received a manuscript in which evidence is presented regarding Ca *+ influx, and a similar conclusion was reached (Alford et al., 1993). To address whether such a Ca*+ signal is necessary for induction of LTP, we performed a separate set of experiments in which wedelivered LTP-inducing stimuli while holding the postsynaptic neuron at different membrane potentials. When afferent fibers were stimulated at low frequency and the postsynaptic cell was held near 0 mV, robust LTPwas elicited. If the holding potential was near +I00 mV, LTPwas blocked. At intermediate potentials between +I00 mV and 0 mV, intermediate degrees of potentiation were observed. These results extend the previous results of Malenka et al. (1988) and provide strong support for an essential role for Ca*+ influx through the NMDA receptor in the induction of LTP.
Neuron
818
DEPOLARIZATION T O 0 mV
Figure 1. Effect of Steady Depolarization on Intracellular Free Ca2+ Concentration in a CA3 Pyramidal Cell (A) Fluorescence of injected fura-2,380 nm excitation wavelength. (B) Ca2+ concentration based on fluorescence ratio (360 nmi 380 nm excitation) when the membrane potential was -70 mV. (C) Peak Ca’+ levels after depolarization to approximately 0 mV. The recovery of free Ca’+ concentration upon maintained depolarization of the same cell is illustrated in Figure 2A. Bar, 25 urn.
-I 50 nM
I 950 nM
SYNAPTIC
I 2.7 pt.A
STIMULATION Stimulation
Control
AT 0 mV
Figure 2. NMDA Receptor-Dependent Ca’+ Changes in the Apical Dendrites of a CA3 Pyramidal Neuron in Response to Stimulation of AssociationallCommissural Fibers during Steady Depolarization to 0 mV
Recoverv
20 mV/ e
The same cell as in Figure 1 was used. (A) Ca’+ levels 13.5 min after the beginning of steady depolarization to 0 mV. Bar, 25 urn. (B) Ca2’ responseto a train of synaptic stimuli delivered to associationallcommissural afferent fibers. The proximal apical dendritic region, the soma, and the basal dendritic region did not show Ca2’ changes during the stimulus. (C) Recovery, 30 s after the stimulus train. (D-F) Same as in (A)-(C), except that the bathing solution contained 100 uM DL-APV. Traces below images (A)(C) and (D)-(F) aresimultaneous membrane potential recordings, verifying that the cell was being held at the excitatory reversal potential.
j!-.. 1
SYNAPTIC
set
l;o
Sl TIMULATION Stimulation
Control
20mVI
I 6 pt.4
-b 1 set
3bO
d0
80b nM
IN APV AT 0 mV Recoverv
Role of Ca2+ Influx through 819
NMDA
Receptors
in LTP Induction
Results Conventional intracellular recordings were obtained from CA1 and CA3 pyramidal cells in guinea pig hippocampal slices. To measure free cytoplasmic Ca*+ concentration, cells were filled with the Ca2+ indicator furadissolved in CsCI, which blocks K+ channels and facilitates depolarization. Final indicator concentration was 200-250 FM (see Experimental Procedures). Ca2+ levels were determined ratiometrically from paired digitally acquired fluorescence images using excitation illumination of wavelength 360 nm and 380 nm (Grynkiewicz et al., 1985). When an impaled pyramidal cell was adequately filled with fura(Figure IA) and held at -70 mV, it displayed low resting Ca*+ concentrations, in the range of 50-100 nM (Figure IB). The cell was then steadily depolarized to approximately 0 mV, and Ca 2+ levels rose dramatically (to greater than 1 PM) throughout the cell during the initial portion of the depolarization, as the membrane fired repeated Ca*+ action potentials (Figure IQ Accompanying stabilization of the membrane potential, the Ca2+ concentration fell to a level not dramatically higher than the resting level observed at -70 mV (Figure 2A). When the Ca2+ concentration was stable, a train of stimuli (50 Hz; Is) was delivered to afferent fibers in the stratum radiatum. Starting l-2 s before the stimulus train and continuing for 5-10 s afterward, image pairs were collected at approximately 3.5 Hz. In portions of the dendrites, Ca2+ concentration rose briefly, often in a punctate fashion, and recovered within several seconds (Figures 2A-2C). These transients were repeatable several times within a single bout of steady depolarization. To test whether this rise of Ca2+ concentration depended on the activation of NMDA receptors, the depolarization and stimuli were repeated following addition of the NMDA receptor antagonist APV (100 PM) to the bathing solution. As illustrated in Figures 2D2F, stimulus trains in the presence of APV led to little detectable change in Ca2+ concentration. In 3 cells (including the one shown in Figure 2), the steady-state Ca2+ concentration was reduced in the presence of APV. It is possible that the APV sensitivity of steadystate Ca2+ concentration reflects Ca2+ influx through tonically activated NMDA receptors (Sah et al., 1989). In addition, this could reflect a gradual decline of Ca*+concentration during maintained depolarization, which was also observed in the absence of APV. In another pyramidal cell, we selected a small portion of the apical dendritic field that showed a clear Ca2+ transient and measured the Ca2+ concentration in that region as a function of time, so that the time course of the change in Ca2+ concentration in the dendrites could be examined quantitatively. Figure 3 (open squares) illustrates the change in Ca2+ concentration in response to a tetanic stimulus while the postsynaptic cell was held at +I0 mV. A rapid rise in Ca2+ concentration was observed, and the concentration decayed to baseline values over several seconds.
The membrane potential during the stimulus was recorded to ensure that the synaptic stimulus did not cause depolarization of the membrane. As illustrated by the trace shown in the inset of Figure 3 (trace a), the synaptic response was reversed, indicating that voltage-sensitive Ca2+ channels could not have contributed subtantially to the observed rise in intracellular Ca*+. The effect of APV on the Ca2+ response to the same stimulus (Figure 3, trace b) in the same cell is shown in Figure 3 (closed circles). The synaptic stimulation led to only a small increase in Ca2’ concentration, which measured 5% of the control response. The effect of APV was partially reversible upon removal from the perfusion medium, recovering to 37% of the control response, as seen in Figure 3 (closed squares, trace c). The average synaptically induced rise in intradendritic Ca2+, measured with the postsynaptic membrane held between 0 mV and +20 mV, was 1400 f 700 nM (mean + SEM; n = 4 CA1 cells and 4 CA3 cells). In neurons in which the effects of APV were studied, there was a 85% +- 8% reduction in the Ca2+ transient in the presence of the antagonist (n = 8). In all cases in which synaptic stimulation was attempted following washout of APV, a partial recovery of the Ca*+ transient (to 71% + 16% of control) was observed (n = 4). The average tetanus-induced change in membrane potential was similar in control conditions (-4.3 k 1.5 mV) and in the presence of APV (-5.5 +
CALCIUM RESPONSE TO SYNAPTIC STIMULATION AT +lO mV
10mV
WASH
L IS
I
i! I
0
1 .o
2.0
3:.
1
%(,
Time(s)
Figure 3. Quantitative Analysis of Regions of the Apical Dendrite of a Different CA3 Neuron Showing Ca2+ Transients Open squares represent the time course of the cytoplasmic Ca2+ concentration in a small (2.5 x 2.5 pm) region of the dendrite following a tetanic stimulus (1 s, 100 Hz) delivered to afferent fibers. Closed circles represent the Ca2+ concentration following the addition of 100 PM DL-APV to the bath. Recovery of CaZ+ transient following removal of APV from the bath is indicated by the open squares. The postsynaptic cell was steadily held at +I0 mV prior to the tetanus. Insets illustrate the negative-going response of membrane potential, indicating that the synaptic stimulation (indicated by heavy horizontal bar) led to hyperpolarization (a, control; b, APV; c: after removal of APV).
Neuron 820
A5 l
Paired at +I0 mV
in 5 of the 7 cells in which tetanic stimulation was delivered at different somatic potentials, the Ca2+ transient was clearly decreased at more positive voltages. Between 0 mV and around +50 mV, there was, on average, a 40% + 16% reduction in the amplitude of the transient (p < 0.05; n = 7). This effect is consistent with a reduced driving force on Ca2+ entry associated with extreme depolarization and provides further evidence that the observed Ca2+ transients are due to influx through the NMDA receptors.
k. : ’ ?
..
- Paired at +70 mV
Pairing -
I
2 0
10 Time (min)
I 30
20
Bl.r-TK= -I 1
50
pA
20 ms
Figure 4. Blockadeof LTP When Low Frequency Synaptic Were Paired with Extreme Depolarization to +70 mV
Stimuli
Two groups of afferents were stimulated alternately in the baseline period, during which the postsynaptic cell was held at -80 mV. The cell was depolarized to an extreme potential and one pathway, the “test” pathway, was stimulated 50 times at 0.5 Hz. The membrane potential was shifted to the synaptic reversal potential (approximately +I0 mV), and the other pathway, the “reference” pathway, was given the same stimulus. The membrane potential was shifted back to -80 mV, and alternate stimuli were given to the two pathways to assess the presence of LTP. (A) Circles indicate EPSC amplitudes for the reference pathway. Triangles represent EPSC amplitudes for the test pathway. (B) Sample EPSCs (averages of 10 consecutive sweeps) in the test pathway taken before and after the pairing (indicated by 1 and 2 in [A]). (C) Sample EPSCs from the reference pathway before and after the pairing.
Role of Ca2+ in LTP To address whether Ca2’ influx through the NMDA receptor is necessary for the induction of LTP in the CA1 region, we took advantage of the increased control of membrane potential available with whole-cell recording. We reasoned that if Ca2+entry were necessary, then a reduced driving force on Ca2+ during activation of NMDA receptors should result in little or no LTP. During a baseline recording period (holding potential -80 mV), excitatory postsynaptic currents (EPSCs) were evoked alternately in two independent afferent pathways at 0.1 Hz. Stimulation was stopped and the potential was shifted to a test value between -30 mV and i-100 mV. When the holding current had stabilized,50stimuliweredeliveredtooneofthepathways at 0.2-0.7 Hz. Because LTP may be variable under whole-cell recording conditions, it was important to test whether the cell was capable of exhibiting potentiation. Therefore, after the 50 pulses were delivered to the test pathway, the membrane potential was shifted to the synaptic reversal potential (approxi-
ioo-
80 -
P 2 0 8 F 'yfj a, cc
2.3 mV). These data indicate that activation of synapses can directly increase the cytoplasmic CaZ+ concentration in a fashion directly dependent on NMDA receptors. We observed that the steady-state Ca2+ concentration measured at depolarized potentials depended on the holding potential. In each of 7cells examined, the intradendritic Ca*+ concentration was lower when the membrane potential was held near +50 mV than near 0 mV. Steady-state dendritic Ca2+ concentration averaged 330 + 120 nM near 0 mV and was reduced by 22% k 9% near +50 mV (p < 0.001; n = 7). These data provide evidence that Ca2+ influx contributes to the steady-state Ca2+ concentration. Although it was not possible to clamp the dendritic membrane potential during the synaptic stimulation,
60-
40-
20 -
90 0
I 50
100
Membrane potential (mV) Figure 5. Voltage Dependence of Induction of LTP by Pairing Low Frequency Stimulation with Depolarization The amount of potentiation in the test pathway of each experiment was normalized by the amount of potentiation observed in the reference pathway, which was stimulated while the cell was held at 0 mV. With extreme depolarization to approximately +I00 mV, little potentiation was observed. With intermediate potentials between 0 mV and +I00 mV, there was a gradual reduction in the amount of potentiation observed. Relative potentiation values were as follows: near +I00 mV, 5% + 23%, n = 3; near +60 mV, 25% + 13%, n = 8; near f30 mV, 57% + 12%, n = 5; near -35 mV, 47% * 34%, n = 3.
Role of Ca’+ Influx through 821
NMDA
Receptors
in LTP Induction
mately +I0 mV). In each experiment, regardless of the test potential, the second pathway was given 50 stimuli with the membrane potential held at the reversal potential. Results from an example experiment are illustrated in Figure 4. Summary data based on 19 experiments are illustrated in Figure 5. For each experiment in which clear potentiation was observed in the reference pathway (at least 30% enhancement measured IO-20 min after pairing), the degree of potentiation observed in the test pathway was expressed as a percentage of the potentiation observed in the reference pathway. The resulting relative potentiation value was 0% if no potentiation occurred in the test pathway and 100% if the test pathway showed the same amount of potentiation as the reference pathway. The EPSC amplitude in the reference pathway was increased, on average, to 226% + 22% (n = 19) of control. As seen in Figure 5, when the membrane potential was held near +I00 mV, synaptic stimulation resulted in littleor no potentiation (5% + 23%; n = 3; p < 0.03 versus reference pathway). The amount of enhancement was gradually reduced as the potential was shifted from 0 mV to +I00 mV.
Discussion This study focused on two ing the induction of LTP: sure the postulated entry NMDA receptor channel; try a trigger for LTP?
related questions concernfirst, can one directly meaof Capthrough the synaptic and second, is this Ca*+ en-
Measurement of Ca2+ Transients during Synaptic Activation of NMDA Receptors Weobserved transient increases in dendriticfreeCa*+ concentration in CA1 and CA3 pyramidal cells in response to stimulation of Schaffer collateral (CAI) or associationallcommissural (CA3) synapses while the postsynaptic membrane potential was at or beyond the synaptic reversal potential of approximately0 mV. These changes in Ca*+ were greatly reduced in the presence of the NMDA receptor antagonist APV and were partially restored upon washout of APV. These data indicate that synaptic stimulation can lead to NMDA receptor-mediated rises in intracellular Ca2+ without a contribution from voltage-sensitive Ca*+ channels. A crucial aspect of the work presented here was to design experiments in which entry of Ca*+via voltagesensitive Ca*+ channels could not contribute to the observed synaptically induced Ca*+ transients. Voltare known to be present age-sensitive Ca *+ channels at high densities in the dendrites (Llinds, 1988; Westenbroek et al., 1990) and can generate large Ca*+ transients independent of synaptic stimulation (Miyakawa et al., 1992; Muller and Connor, 1991; Miiller et al., 1992). As noted by Miyakawa et al. (1992), it is difficult to exclude voltage-sensitive Ca2+ channels as the source for an APV-sensitive Ca*+ signal observed un-
der normal conditions. To avoid this problem, we took advantage of the fact that when the postsynaptic membrane is held at the reversal potential for a synapse, activation of the synapse does not, by definition, change the membrane potential. Since, in our experiments, weobserved thatthe stimulus-induced change of membrane potential was hyperpolarizing or null during the rise in Ca2+ concentration, we conclude that the activation of voltage-sensitive Ca2+ channels did not contribute to the response. A different approach to this problem was taken by Alford et al., (1993)who relied primarily on adequate spatial control of membrane potential and complete washout of dendritic Ca*+ currents to reduce Ca*+ influx through voltage-dependent Ca2+ channels. While the nearly complete blockade of Ca*+ transients by APV indicates a strong dependence on activation of NMDA receptors, it is possible that not all of the Caz+ that we measured actually represents Ca*+ influx through this receptor. For example, we do not know the degree to which Ca*+-induced Ca2+ release from intracellular stores (Henzi and MacDermott, 1992) may have amplified the NMDA receptor-mediated Ca*+ influx. However, our data do constrain the possible contribution made by activation of metabotropic glutamate receptors or Ca*+-permeant nonNMDA receptors (Kohler et al., 1993; Murphy and Miller, 1988; Ozawa et al., 1988); since APV, which does not affect these other receptors, blocked the observed Ca*+ transients by 85%, we conclude that the non-NMDA ionotropic and metabotropic receptors can account for only a small fraction of the response. Several factors have allowed us to observe a synaptically mediated Ca2+ signal. First, we used depolarization to unmask NMDAcurrents, which has two principal advantages over removal of Mg*+ from the bathing solution: Virtually complete (>95%) removal of the Mg2’ blockadeof the NMDA receptor can be achieved when the cell is held at +I0 mV (Hestrin et al., 1990); additionally, if the decay time constant of the NMDA receptor-mediated conductance change is voltage dependent (Konnerth et al., 1990), significantly greater Ca*+ influx would be expected to occur at positive potentials. A second factor is that we used the dihydropyridine Ca*+ channel antagonist nimodipine to reduce the steady-state Ca2+ influx through noninactivating Ca2-+ channels, thus maintaining the backour experimental ground Ca *+ levels low. Therefore, conditions tended both to increase the transient signal and to reducethe background Ca*+concentration. It is important to point out some limitations of the present imaging study. First, we were not able to resolve dendritic spines and therefore do not know whether the Ca*+ signals that we observed reflected averages of signals from numerous spines or spillover from spines to dendritic shafts. Another limitation involves control of membrane potential. Even under voltage-clamp conditions, it would have been impossible to clamp the membrane potential at the synaptic
Neuron a22
membrane during high frequency repetitive stimulation. The present imaging experiments were performed under current-clamp conditions, and thus we were able to obtain only the approximate voltage dependence of the Ca2+ influx. The problem of electrotonic control likely had less impact on the determination of the voltage dependence of LTP induction, since low frequency stimulation and somatic voltageclamp conditions were used. We therefore cannnot compare quantitatively the dependence on membrane potential of Cal+entryand LTP induction. Nonetheless, shifts of membrane potential affected both the size of the Ca2+ signals and the degree of LTP in the same direction. Finally, the time course of the Ca2+ transients that we observed may not necessarily reflect the true time course that would occur under normal conditions; there could be some slowing effect due to the buffering of Ca2+ by fura(Regehr and Tank, 1992), prolonged Ca2+ entry through altered NMDA receptor kinetics (Konnerth et al., 1990), or alteration in the activity of the Na+/Ca2+ exchanger by depolarization (Blaustein, 1988). Nonetheless, these limitations do not affect the primary conclusions from our imaging experiments: that synaptic stimulation gives rise to increased intracellular Ca2+ in a fashion directly dependent on Ca2’ influx through NMDA receptors. The Voltage Dependence of LTP Induction Several studies have shown that the presence of postsynaptic Ca2’ buffers blocks LTP (Lynch et al., 1976; Malenka et al., 1988). It is not known, however, whetherthere is simply a requirement for some”baseline” level of Ca2+ to enable the biochemical processes for LTP, or whether a specific rise in intracellular Ca2+ is necessary. Our results provide evidence for adirect instructive role for Ca2+, since LTP was blocked simply by providing extreme depolarization, which we postulate would greatly reduce Ca2+ influx. In addition, it should be mentioned that this experiment does not formally rule out the possibility that Na’ entry is important in LTP, since extreme depolarization would be expected to reduce the influx of any permeant ion whose reversal potential is greater than +50 mV. In combination with the available data concerning Ca2+ buffers, however, the most likely explanation for our data is that Ca2+ influx through NMDA receptors serves as a trigger for at least some of the biochemical processes subserving the enhanced synaptic transmission of LTP. It appears clear, however, that some otherfactorprovided bysynapticstimulation mayalso play an important role in leading to persistent potentiation (Kullmann et al., 1992). It is not known whether LTP at individual synapses occurs in a graded or all-or-none fashion. The graded reduction in the degree of potentiation observed with increasingly positive membrane potential could reflect a graded reduction in the amount of potentiation at each synapse. Alternatively, it could reflect a reduced probability of some all-or-none process such
that only a subset of synapses were potentiated, but to the usual degree. In summary, our results demonstrate directly the accumulationofdendriticCa2+in thepostsynapticceil via entry through NMDA receptors. In addition the present data indicate thatthis Ca2’ source likely serves as an instructive, rather than simply a permissive, signal for generating LTP, since blocking this source of Ca2+ with extreme depolarization blocked LTP. Experimental
Procedures
Hippocampal slices (300-500 pm thick) were prepared from 3- to 5-week-old Hartley guinea pigs using a vibratome. Slices were stored in a holding chamber at the interface between recording solution and humidified atmosphere consisting of 95% 02, 5% COz. Recordings were carried out at room temperature (22OC-24OC) with the slice submerged beneath continuously superfusing bathing medium solution. In some of the imaging experiments, an interface recording configuration was used. This extracellular solution contained 119 m M NaCI, 2.5 m M KCI, 4 m M MgSO+ 4 m M CaC&, 26 m M NaHCOp, 1 m M NaHP02, nglucose, and 50-100 pM picrotoxin. Imaging experiments were performed on CA? and CA3 pyramidal neurons. In these experiments, the dihydropyridine Cal’ channel antagonist nimodipine was added to the bathing medium (IO PM; 0.1% dimethylsulfoxide) to reduce the steady-state CaZ’concentrationduringmaintaineddepolarization. Intracellutips conlar recordings were made using microelectrodeswhose tained 5-10 m M fura- (potassium salt, Molecular Probes, Eugene OR) and 0.2-1.5 M CsCl and were backfilled with 2.5 M CsC!. Total electrode resistance was 40-80 MO. Cells were dye loaded with steady hyperpolarizing current injection (0.5-1.0 nA) and Cs+ loaded with brief depolarizing current pulses (100-200 ms; 0.2-0.7 nA; every 5-10 s) for IO-20 min. Indicator concentration in the cells was approximately200-250 PM, as estimated by comparing the 360 nm excited fluorescence of the microinjected neurons with that of neurons filled using whole-cell recording pipettes containing 200 PM fura-2. Intracellular signals were amplified using an Axoclamp-2A (Axon Instruments, Burlingame CA) in the discontinuous current-clamp mode (1-5 kHz switching frequency). Membrane potential was recorded on magnetic tape, subsequently digitized at 10 kHz, and low pass filtered at 500 Hz for display. Synaptic stimuli were delivered via bipolar stainless steel electrodes (FHC, Brunswick, ME); tetani consisted of 1 s trains at 50 or 100 Hz. Cells were imaged from the top surfaceofthe slice using an upright microscope (Zeiss Axioskop) and long working distance 40x water-immersion or 20x dry objectives (Zeiss). A CCD camera system (Series 200, Photometrics Ltd., Tucson, AZ) similar to ones previously described (Connor,1986;MijllerandConnor,1991)wasusedtoacquiredigitized images of fura-2. The camera was operated in frame-transfer mode to acquire image pairs at 360 and 380 nm excitation wavelengths. Exposure times for single frames were 100-200 ms, correspondingtoacquisition ratesof2-4Hzper imagepair. Fluorescence data were background corrected and spatially filtered (3 x 3 pixels) before ratio images were constructed. Ca*‘concentrations were then determined from ratio images (Crynkiewicz et al., 1985). For illustration of images, a mask for the ratios was constructed by importing a single-wavelength image (resting cell, 380 nm) into Adobe Photoshop and obtaining contours of the most prominent dendritic structures. Whole-cell recordings were obtained using the blind technique (Blanton et al., 1989; Coleman and Miller, 1989). The intracellular solution contained 130 m M cesium gluconate, 15 m M CsCI, IO m M NaCI, 10 m M HEPES, 0.2 m M EGTA, 2 m M ATP, and 0.1 m M GTP. Nimodipine was not present in experiments designed to assess LTP. Input resistance and access resistance were monitored throughout the experiment; data were discarded if significant changes in passive properties of the cell or recording occurred. Signals were amplified using an Axoclamp-
Role of Ca* Influx through 823
NMDA
Receptors
in LTP Induction
ID amplifier, filtered at 1 kHz, and digitized at 2 kHz using a TLI data acquisition system and modified pCL4MP software (all from Axon Instruments).
References Alford, S., Frenguelli, B. C., Schofield, J. G., and Collingridge, G. L. (1993). Characterization of CaZ+ signals induced in hippocampal CA1 neurons by the synaptic activation of NMDA receptors. J. Physiol. 469, 693-716. Ascher, P., and Nowak, L. (1988). The role of divalent the N-methyl-o-aspartate responses of mouse central in culture. J. Physiol. 399, 247-266.
cations in neurones
Blanton, M. C., Lo Turco, J. J., and Kriegstein, A. R. (1989). Whole cell recording from neurons in slices of reptilian and mammalian cerebral cortex. j. Neurosci. Meth. 30, 203-210. Blaustein, M. P. (1988). Calcium transport rons. Trends Neurosci. 77, 438-443.
and buffering
Bliss, T. V. P., and Collingridge, G. L. (1993). A synaptic of memory: long-term potentiation in the hippocampus. 361, 31-39. Bourne, H. R., and Nicoll, R. (1993). Molecular coincident synaptic signals. Cell 72/Neuron
in neumodel Nature
machines integrate 70 (suppl.), 65-75.
Coleman, P. A., and Miller, R. F. (1989). Measurement of passive membrane parameters with whole-cell recording from neurons in the intact amphibian retina. J. Neurophysiol. 67, 218-230. Connor, J. A. (1986). Digital imaging of free calcium changes and of spatial gradients in growing processes in single mammalian CNS cells. Proc. Natl. Acad. Sci. USA 83, 6179-6183. Grynkiewicz, C., Poenie, M., and Tsien, R. Y. (1985). A new generation of CaZ+ indicatorswith greatly improved fluorescence properties. J. Biol. Chem. 260, 3440-3450. Henzi, V., and MacDermott, A. B. (1992). Characteristics function of Ca*+- and inositol 1,4,5-trisphosphatereleasable stores of Ca2+ in neurons. Neuroscience 46, 251-273.
and
Hestrin, S., Nicoll, R. A., Perkel, D. J., and Sah, P. (1990). Analysis of excitatory synaptic action in the rat hippocampus usingwhole cell recording from thin slices. J. Physiol. 422, 203-225. Kohler, M., Burnashev, N., Sakmann, B., and Seeburg, P. H. (1993). Determinants of Ca* permeability in both TM1 and TM2 of high affinity kainate receptor channels: diversity by RNA editing. Neuron 70,491-500. Konnerth, A., Keller, B. U., Ballanyi, K., and Yaari, Y. (1990). Voltage sensitivity of NMDA-receptor mediated postsynaptic currents. Exp. Brain Res. 87, 209-212. Kullmann, D. M., Perkel, D. J., Manabe, T., and Nicoll, R. A. (1992). Calcium entry via postsynaptic voltage-sensitive calcium channels can transiently potentiate excitatory synaptic transmission in the hippocampus. Neuron 9, 1175-1183. Llinas, R. R. (1988). The intrinsic of mammalian neurons: insights tion. Science 242, 1654-1664.
electrophysiological properties into central nervous system func-
Lynch, G., Gribkoff, V. K., and Deadwyler, S. A. (1976). Long-term potentiation is accompanied by a reduction in dendritic responsiveness to glutamic acid. Nature 263, 151-153. Malenka, R. C., Kauer, J. A., Zucker, R. J., and Nicoll, R. A. (1988). Postsynaptic calcium is sufficient for potentiation of hippocampal synaptic transmission. Science 242, 81-84. Mayer, M. L., and Westbrook, C. L. (1987). Permeation and block of N-methyl-o-asparticacid receptor channels bydivalent cations in mouse cultured central neurones. J. Physiol. 394, 501-527. Mayer, M. L., MacDermott, A. B., Westbrook, G. L., Smith, S. J., and Barker, J. L. (1987). Agonist- and voltage-gated calcium entry in cultured mouse spinal cord neurons under voltage clamp measured using arsenazo Ill. J. Neurosci. 7, 3230-3244. Miyakawa, H., Ross, W. N., Jaffe, D., Callaway, J. C., Lasser-Ross, N., Lisman, J. E., and Johnston, D. (1992). Synaptically activated increases in Ca2+ concentration in hippocampal CA1 pyramidal
cells are primarily 9, 1163-1173.
due to voltage-gated
Ca*+ channels.
Neuron
Miiller, W., and Connor, J. A. (1991). Dendritic spines as individual neuronal compartments for synaptic Ca2+ responses. Nature 354, 73-76. Miiller, W., Petrozzino, 1. I., Griffith, L. C., Danho, W., and Connor, J. A. (1992). Specific involvement of Ca2+-calmodulin kinase II in cholinergic modulation of neuronal responsiveness. J. Neurophysiol. 68, 2264-2269. Murphy, S. N., and Miller, R. J. (1988). A glutamate receptor regulates Ca2+ mobilization in hippocampal neurons. Proc. Natl. Acad. Sci. USA 85, 8737-8741. Nicoll, R. A., Kauer, J. A., and Malenka, excitement in long-term potentiation.
R. C. (1988). The current Neuron 7, 97-103.
Ozawa, S., Nakamura, T., and Yuzaki, M. (1988). Cation permeabilitychangecaused by L-glutamate in cultured rat hippocampal neurons. Brain Res. 443, 85-94. Regehr, W. G.,andTank, D. W. (1990). Postsynaptic tor-mediated calcium accumulation in hippocampal dal cell dendrites. Nature 345, 807-810.
NMDA recepCA1 pyrami-
Regehr, W. C., and Tank, D. W. (1992). Calcium concentration dynamics produced by synaptic activation of CA1 hippocampal pyramidal cells. J. Neurosci. 72, 4202-4223. Sah, P., Hestrin, S., and Nicoll, R. A. (1989). Tonic activation NMDA receptors by ambient glutamate enhances excitability neurons. Science 246, 815-818.
of of
Westenbroek, R. E., Ahlijanian, M. K., and Catterall, W. A. (1990). Clustering of L-type Ca2+ channels at the base of major dendrites in hippocampal pyramidal neurons. Nature 347, 281-284. Wigstrom, H., and Custafsson, B. (1985). On long-lasting potentiation in the hippocampus: a proposed mechanism for its dependence on coincident pre-and postsynaptic activity. Acta Physiol. Stand. 723, 510-522.