NMDA Receptor Trafficking at Recurrent Synapses ... - Semantic Scholar

J Neurophysiol 98: 2818 –2826, 2007. First published August 29, 2007; doi:10.1152/jn.00346.2007.

NMDA Receptor Trafficking at Recurrent Synapses Stabilizes the State of the CA3 Network Jennifer L. Hellier,1 David R. Grosshans,2 Steven J. Coultrap,2 Jethro P. Jones,1 Peter Dobelis,1 Michael D. Browning,2,3 and Kevin J. Staley1,3 1

Departments of Pediatrics and Neurology and 2Pharmacology and 3Program in Neuroscience, University of Colorado Health Sciences Center, Denver, Colorado

Submitted 27 March 2007; accepted in final form 28 August 2007

Hellier JL, Grosshans DR, Coultrap SJ, Jones JP, Dobelis P, Browning MD, Staley KJ. NMDA receptor trafficking at recurrent synapses stabilizes the state of the CA3 network. J Neurophysiol 98: 2818 –2826, 2007. First published August 29, 2007; doi:10.1152/jn.00346.2007. Metaplasticity describes the stabilization of synaptic strength such that strong synapses are likely to remain strong while weak synapses are likely to remain weak. A potential mechanism for metaplasticity is a correlated change in both N-methyl-D-aspartate (NMDA) receptor-mediated postsynaptic conductance and synaptic strength. Synchronous activation of CA3–CA3 synapses during spontaneous bursts of population activity caused long-term potentiation (LTP) of recurrent CA3–CA3 glutamatergic synapses under control conditions and depotentiation when NMDA receptors were partially blocked by competitive antagonists. LTP was associated with a significant increase in membrane-bound NMDA receptors, whereas depotentiation was associated with a significant decrease in membrane-bound NMDA receptors. During burst activity, further depotentiation could be induced by sequential reductions in antagonist concentration, consistent with a depotentiation-associated reduction in membrane-bound NMDA receptors. The decrease in number of membrane-bound NMDA receptors associated with depotentiation reduced the probability of subsequent potentiation of weakened synapses in the face of ongoing synchronous network activity. This molecular mechanism stabilizes synaptic strength, which in turn stabilizes the state of the CA3 neuronal network, reflected in the frequency of spontaneous population bursts.

Reduction of GABAA receptor-mediated inhibition in the CA3 area of the hippocampus leads to periodic bursts of synchronous population activity that strongly activate the recurrent collateral glutamatergic synapses between CA3 pyramidal cells (Aradi and Maccaferri 2004; Jefferys 1998; Miles and Wong 1987). These spontaneous CA3 population bursts are a widely studied model of hippocampal sharp-waves (Behrens et al. 2005; Buzsaki 1986) and interictal epileptiform activity (Dzhala and Staley 2003; Traub and Miles 1991). Each burst is initiated by a build-up of excitatory postsynaptic potentials (EPSPs) that trigger action-potential-driven glutamate release and thus more EPSPs due to the extensive feedback of CA3 pyramidal cells (Chamberlin et al. 1990; Li et al. 1994; Miles and Poncer 1995). Activation of CA3 recurrent collateral synapses results in a large postsynaptic depolarization that, in combination with the synaptic activation, is sufficient to induce LTP at CA3–CA3 synapses (Bains et al. 1999;

Behrens et al. 2005) and CA3–CA1 synapses (Abegg et al. 2004; Buzsaki 1989). Increases in EPSP amplitude induced by LTP accelerate this positive feedback cycle and thereby increase burst probability and frequency (Bains et al. 1999). Conversely, depotentiation-induced decreases in EPSP amplitude reduce spontaneous CA3 burst probability (Bains et al. 1999; Behrens et al. 2005; Staley et al. 2001). The stability of the reduced CA3 burst probability after depotentiation is surprising because each subsequent burst should re-potentiate the CA3–CA3 synapses so that burst probability should gradually increase back to the initial, predepressed value. The stability of the weakened CA3–CA3 synapses is in line with prior observations that synapses that have undergone long-term depression (LTD) cannot subsequently undergo LTP (Delgado and O’Dell 2005). Metaplasticity refers to this stabilization of synaptic strength such that strengthened synapses are likely to remain strong and weakened synapses are likely to remain weak (Abraham et al. 2001). The unexpected stability of some CA3–CA3 synapses (Debanne et al. 1999; Montgomery and Madison 2002) strongly supports the existence of metaplasticity at these synapses, but the physiological basis for metaplasticity has remained unclear. Activity-dependent modulation of calcium-permeable NMDA receptors is an attractive mechanism of metaplasticity (PerezOtano and Ehlers 2005; Philpot et al. 2003) and is supported by the finding that N-methyl-D-aspartate (NMDA) receptors, the activation of which is needed for activity-dependent modulation of the strength of CA3 efferent synapses (Malenka and Bear 2004), are inserted into the membrane after LTP at CA3–CA1 synapses in adult animals (Grosshans et al. 2002a; Vissel et al. 2001), although this has been controversial (Bashir et al. 1991; Kauer et al. 1988). If depotentiation caused the reverse to occur so that NMDA receptor-mediated conductances were reduced (Morishita et al. 2005) for example by withdrawal of NMDA receptor subunits from the membrane (Montgomery et al. 2005), then NMDA receptor-induced potentiation would be more likely to occur at strong synapses and less likely to occur at weak synapses. In this investigation, we tested the hypotheses that LTP at CA3–CA3 synapses is associated with increases in membranebound NMDA receptors, the fraction of membrane-bound NMDA receptors is decreased during depotentiation produced by partial blockade of NMDA receptors during synchronous

Present address and address for reprint requests and other correspondence: K. J. Staley, Dept. of Neurology, Massachusetts General Hospital, 114 16th St., B114-2625, Cambridge, MA 02129 (E-mail: [email protected]).

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

INTRODUCTION

2818

0022-3077/07 $8.00 Copyright © 2007 The American Physiological Society

www.jn.org

METAPLASTICITY INDUCED BY SPONTANEOUS CA3 BURSTS

network activity, and reduced membrane-bound NMDA receptors stabilize the depotentiated state of the synapse so that the network remains locked in a state of low burst probability.

2819

95% O2-5% CO2, and recovered for 1.5–2 h before initiating experiments.

Electrophysiological recordings METHODS

Hippocampal slice preparation Adult male Sprague-Dawley rats (4- to 8-wk-old; Harlan, Wilmington, MA) were treated according to the guidelines of the Animal Care and Use Committee at the University of Colorado Health Sciences Center. Rats were anesthetized (sodium pentobarbital, 50 mg/kg), and brains were quickly removed and sliced in ice-cold, high-sucrose solution (composition in mM: 87 NaCl, 2.5 KCl, 26 NaHCO3, 0.5 CaCl2, 7 MgCl2, 1.25 NaH2PO4, 25 glucose, and 75 sucrose). Coronal or horizontal slices of the hippocampus (350 – 400 ␮m) were prepared with a vibrating microtome (Leica, Bannockburn, IL). For CA3 mini-slices, hippocampi were trimmed by removing neocortex, dentate gyrus, and CA1 regions prior to tissue storage (Fig. 1A). Slices were stored at 32–34°C in a modified artificial cerebrospinal solution (ACSF, in mM: 106.5 NaCl, 2.5 KCl, 26 NaHCO3, 1.25 CaCl2, 4.5 MgCl2, 1.25 NaH2PO4, 17.5 glucose, 37.5 sucrose), humidified with

A

B

In both whole hippocampal and CA3 mini-slices, extracellular field potentials were recorded from the stratum pyramidale of the CA3 region using glass pipette electrodes filled with 150 mM NaCl. Whole cell experiments were guided by visualization using differential interference contrast. Voltage-clamp recordings were performed using a filling solution containing (in mM) 125 cesium methylsulfonate, 2 MgCl2, 8 NaCl, 0.5 cesium ethylene glycol-bis (␤-aminoethyl ether) N,N,N⬘,N⬘-tetraacetic acid (EGTA), 10 HEPES, 4 sodium adenosine 5⬘-triphosphate, 0.3 sodium guanosine 5⬘-triphosphate, and 1 QX314 (to block voltage-dependent sodium channels). To induce CA3 bursting, slices were bathed in a modified ACSF (Stasheff et al. 1989) (changes in mM: 3.3 KCl, 1.3 CaCl2, and 0.9 MgCl2) containing 100 ␮M picrotoxin (Sigma, St. Louis, MO) and 1 ␮M CGP-55845 (Tocris, Ellisville, MO) to block GABAA and GABAB-mediated synaptic activity and unmask CA3–CA3 recurrent collateral synapses. Depotentiation of recurrent CA3 synapses was induced by transient blockade of NMDA receptors through application of 2.5–10 ␮M D-(⫺)-2-amino-5-phosphonopentanoic acid (D-APV; Tocris, Ellisville, MO) or 20 – 80 ␮M SDZ-220-581 (SDZ; Tocris, Ellisville, MO), both competitive NMDA antagonists (Urwyler et al. 1996). D-APV and SDZ were applied in either decreasing or increasing concentrations. Specifically, 30 min after the initial dose, the concentration was decreased or increased and applied for 30 min. This process was continued until the final dose and/or wash (e.g., D-APV: 30 min at 10 ␮M then 30 min at 5 ␮M then 30 min at 2.5 ␮M then 30- to 60-min wash). To block NMDA receptor-mediated LTP, 80 –100 ␮M DL-2-amino-5-phosphonovaleric acid (DL-APV; Tocris, Ellisville, MO) was bath applied prior to application of picrotoxin and CGP-55845. Extracellular events were recorded with an Axoclamp-2B amplifier (Axon Instruments, Union City, CA), digitized at 2–10 kHz, and analyzed using routines written in VisualBasic 6.0 (Microsoft, Seattle, WA).

Surface expression assays

C

FIG. 1. Stable extracellular recordings were obtained from the CA3 pyramidal cell layer of the hippocampus. A: CA3 mini-slices were made by cutting and removing the CA1, dentate gyrus, and neocortex. Sample traces of spontaneous bursts used to measure interburst intervals (distance between the start of 1 burst to the beginning of the next) and burst durations (length of the burst measured above 3 times baseline noise) for analysis. B: interburst interval (‚) and burst duration (䊐) were stable for 2 h in this CA3 mini-slice preparation. C: interburst intervals remained stable from the initial recording (⬃15 min after application of picrotoxin and CGP) through the end of baseline recording (30 –120 min) in all experiments (P ⫽ 0.48, paired t-test, n ⫽ 22). REC, recording electrode.

J Neurophysiol • VOL

Surface expression assays were performed using a membraneimpermeable cross-linking reagent as previously described (Grosshans et al. 2002b; Hall and Soderling 1997a,b; Smith et al. 2006), and CA3 mini-slices were prepared as explained in the preceding text. The membrane-impermeable cross-linker, BS3 [BIS- (sulfosuccinimidyl) Suberate, Pierce, Rockford, IL), reacts with primary amine groups on proteins exposed to the extracellular space. This process covalently links the subunits of the NMDA receptors expressed at the cell surface, causing these receptors to run at a higher molecular weight than intracellular receptors when separated by gel electrophoresis. If the total amount of a given receptor subunit remains constant with a specific treatment, then a change in the intracellular pool of receptor is reflective of an opposite change in the membrane bound pool of receptor. To that end, the total amounts of each NMDAR subunit in the slices were measured from samples that were not treated with cross-linker. Because the total subunit composition did not change, the amount of membrane bound subunit was inferred from changes in the intracellular pool of receptors. Therefore in cross-linked samples the intracellular receptors still run at the molecular weight of the subunit after cross-linking treatment, whereas the membrane bound receptors run at much higher molecular weight. Direct measurement of the membrane pool of receptors is not feasible due to technical limitations in working with such high-molecular-weight proteins. Spontaneous bursting was induced in two to four slices, and an equal number of slices were maintained as matched controls. In a separate series, a third group of bursting slices was treated with

98 • NOVEMBER 2007 •

www.jn.org

2820

HELLIER, GROSSHANS, COULTRAP, JONES, DOBELIS, BROWNING, AND STALEY

sequentially decreasing concentrations of D-APV to induce depotentiation at CA3–CA3 synapses. Immediately after ⬎30-min wash of D-APV, all groups of slices (control, bursting, and D-APV) were immersed in ice-cold ACSF containing 1 mg/ml BS3 for 45 min at 4°C. The slices were kept on ice during the cross-linking to prevent further receptor trafficking. Slices were washed three times in cold ACSF containing 20 mM Tris (pH 7.6) to deactivate and remove remaining cross-linker, and protein concentrations were measured.

Semi-quantitative Western blotting Equal amounts of protein from each condition were loaded onto SDS-PAGE gels then transferred to PVDF membranes as previously described (Coultrap et al. 2005). A standard curve of hippocampal homogenate was included on each gel to allow comparison of samples across gels and ensure that the samples were within the linear range of detection for each antibody. Blots were probed with anti-NR1 (BD Pharmingen), anti-NR2A or anti-NR2B (Snell et al. 1996). Antibodies were diluted 1:5,000, 1:1,000, and 1:3,000 in 1% BSA, respectively. To make certain that no BS3 was entering the cells, blots were also probed for the intracellular protein anti-␣Actinin (Chemicon) diluted 1:1000 in 1% BSA. Immunoreactivity of each band was reported relative to the standard curve (for detailed description of Western blotting and analysis see Coultrap et al. 2005).

survival function that is equal to one minus the cumulative binomial distribution. Eq. 1 was used to represent the time dependence of the single-trial probability p1 P(N, K, p1) ⫽ 1 ⫺

冉

冊

N! (p1)K (1 ⫺ p1)(N ⫺ K) K! ⫻ (N ⫺ K)!

N and K were used to fit the cumulative probability of burst initiation versus time in control conditions and after induction of depotentiation, at which time the preparation was back in the control conditions under which the first fit was performed.

Statistical analysis ANOVA (multiple-comparison test using Newman-Keuls) was performed to determine if significant differences existed among control, bursts, and bursts after depotentiation (i.e., transient application of D-APV or SDZ). We compared: interburst intervals, burst durations, and NR2A and NR1 protein subcellular distributions in control, bursting, and D-APV experiments. Student’s t-tests were performed to determine significant differences in protein distributions between control and bursting slices and burst frequencies between the initial and end of baseline activity. Significance for all statistical analyses was accepted when P ⬍ 0.05.

Data analysis

RESULTS

Interburst intervals were measured from the start of one burst to the beginning of the next burst. Burst durations were calculated as the time during which the absolute value of the burst amplitude remained greater than three times baseline noise (Fig. 1) (Staley et al. 1998). Although partial blockade of NMDA receptors with either SDZ or D-APV significantly decreased burst durations by 15–20% (data not shown), we observed a more significant change in interburst intervals in response to both single and multiple doses. Therefore we focused on interburst intervals in our analysis. For CA3 mini-slices, changes in surface expression of NMDA receptors were assayed by semi-quantitative Western blot analysis of the intracellular population of receptor after treatment of samples with the membrane-impermeable cross-linking reagent (Grosshans et al. 2002a,b). A five-point dilution series of rat hippocampal homogenate was included on each gel to ensure that all samples were within the linear range of detection. Images were obtained using SuperSignal chemiluminescent substrate and an Alpha Innotech imaging system.

Activity-dependent NMDA receptor trafficking at CA3–CA3 synapses

Burst interval analysis Modeling the effects of depotentiation of CA3–CA3 synapses on burst probability was based on binomial algorithms previously described and tested (Staley et al. 2001). Briefly, the probability of burst initiation versus time elapsed since the last burst ended is analyzed in terms of the recovery from burst-induced short-term depression of the pool of CA3–CA3 synapses that are critical for burst initiation. The total number of synapses capable of participating in the next burst ignition is denoted by N, and the number of synapses that must have recovered from the depression to initiate the next burst is denoted by K. Each synapse was assumed to recover from burst-induced shortterm depression monoexponentially with a time constant that was stable across all N synapses (Eq. 1) p 1 ⫽ 1 ⫺ e(⫺t/␶)

(1)

where ␶ is the time constant describing the recovery rate and t is the time since the onset of depression (end of burst). When fully recovered, all synapses were assumed to have a steady-state glutamate release probability, p, which was set to 1 in this analysis. The probability of recovery of K of N synapses was estimated from the J Neurophysiol • VOL

(2)

We tested the hypothesis that LTP is associated with increases in membrane-bound NMDA receptors by inducing spontaneous bursts of CA3 population activity. Previous studies have shown that CA3 population bursts induce LTP at CA3–CA3 recurrent collateral synapses (Bains et al. 1999; Behrens et al. 2005). Burst probability is dependent on the strength of the recurrent collateral synapses (Bains et al. 1999; Staley et al. 2001), so we waited until the burst probability was stable (⬎30 min of stable bursting activity) and then assayed for changes in the subcellular distribution of NMDA receptors (see Grosshans et al. 2002a,b). As previously observed at CA3–CA1 synapses (Grosshans et al. 2002a), LTP induced by CA3 bursting was associated with a significant decrease in the intracellular fraction of NR2A and NR1 subunits of the NMDA receptor (by 18.9 ⫾ 5.8 and 22.6 ⫾ 7.8%, respectively; Fig. 2A, n ⫽ 10). These results are consistent with net movement of NMDA receptor subunits to the surface membrane. To test whether NMDA receptor subunits were withdrawn from the membrane during long-term decreases in synaptic strength, we weakened CA3–CA3 synapses by partially blocking NMDA receptors during bursting. This manipulation is considered to reduce calcium influx to the range in which long-term decreases in synaptic strength are observed (Bains et al. 1999; Cummings et al. 1996). Bath application of D-APV (10 – 5 – 2.5 ␮M; see METHODS for details regarding application) during CA3 bursts increased the intracellular fraction of NR2A by 14.8 ⫾ 6.4% and NR1 by 15.2 ⫾ 5.1% compared with control (Fig. 2B, n ⫽ 10) with no change in total amount of receptor subunits (Fig. 2C). However, in the absence of bursting, the fraction of membrane-bound NMDA receptor subunits was not changed by partial NMDA receptor blockade using the same sequence of concentrations of D-APV as in the experiments during bursting (Fig. 2D, n ⫽ 8). This suggests that this internalization of NMDA receptors requires both


www.jn.org


2821

A

FIG. 2. Western blots of N-methyl-D-aspartate (NMDA) receptor subunits in CA3 mini-slices depicting net movement after long-term potentiation (LTP) and depotentiation. A: induction of LTP by bursting led to a significant decrease in the intracellular pools of NR2A and NR1 subunits to control slices as previously shown to correlate with an increase in surface expression (P ⬍ 0.05, Student’s t-test, n ⫽ 10). B: after inducing depotentiation [sequentially decreasing the application of D-(⫺)-2-amino-5-phosphonopentanoic acid (D-APV) and wash], the intracellular pools of NR2A and NR1 were increased significantly as compared with bursting alone (P ⬍ 0.05, repeated-measures ANOVA, n ⫽ 10). C: total protein for NR2A, NR2B, and NR1 did not change over time after depotentiation (P ⬎ 0.5 for each comparison, repeated-measures ANOVA, n ⫽ 3). D: in the absence of spontaneous bursting, however, sequentially decreasing the concentrations of D-APV to CA3 mini-slices does not change the total amount of NR2A or NR1 subunits in the intracellular pool compared with control (P ⫽ 0.82, Student’s t-test, n ⫽ 8). *, significant from control; **, significant from each other.

B

C

D

partial blockade of NMDA receptors as well as bursting activity. These results are consistent with NMDA receptor subunit withdrawal from the cellular membrane associated with longterm decreases in synaptic strength. Coincident with the increased intracellular fraction of NR2A and NR1 in slices exposed to decreasing concentrations of D-APV during CA3 population bursts, the burst frequency was also reduced (Fig. 3, A and C). The decrease in burst frequency under these conditions was previously shown to be due to depotentiation of CA3–CA3 synapses (Bains et al. 1999), consistent with the finding that burst frequencies are reduced by partial blockade of AMPA receptors (Chamberlin et al. 1990; Staley et al. 2001), and reduced sEPSC amplitudes (Fig. 3D). Presynaptic effects such as reductions in glutamate release probability by adenosine A1 receptor activation can also reduce burst frequency (Dulla et al. 2005; Dunwiddie 1980). However, the rate of increase in burst probability versus time elapsed since the prior burst is due to recovery of synapses from activity-dependent depression incurred during the prior burst. Under these conditions, pharmacological reductions in glutamate release probability produce substantially different rates of increase in burst probability versus time compared with postsynaptic reductions in synaptic strength (Staley et al. 2001). The pattern of increase in burst probability in these experiments strongly supports a postsynaptic effect (Fig. 3, A J Neurophysiol • VOL

and B) consistent with NMDA receptor-dependent postsynaptic reduction in synaptic strength (Cummings et al. 1996). Stabilizing the strength of CA3–CA3 synapses Withdrawal of NMDA receptor subunits from the surface membrane during depotentiation removes receptors that are available for calcium entry. In turn, this should make it more difficult to induce further synaptic plasticity due to decreased calcium influx (Cummings et al. 1996; Lisman 2001). We have previously shown that once depotentiation is induced at CA3– CA3 synapses, it is not possible to repotentiate these synapses to the point that burst probability returns to the initial state (Bains et al. 1999). The NMDA receptor membrane-trafficking data raise the possibility that this synaptic stabilization is due to reduced NMDA receptor-mediated calcium influx as a consequence of a reduced number of NMDA receptors in the cell membrane. In turn, this raises the possibility that depotentiation under these conditions is also limited by NMDA receptor withdrawal, because calcium influx falls below the threshold for depotentiation (Lisman 2001) in the face of both competitive antagonist and reduced number of surface receptors. If so, then reducing the concentration of the competitive NMDA antagonist should increase the NMDA conductance back to the point that additional synaptic weakening takes


www.jn.org

2822


A

B

C

D

FIG. 3. Long-term reduction in CA3 burst probability after partial blockade of NMDA receptors. A: cumulative burst probability as a function of time elapsed since the termination of the last burst. In control conditions, the cumulative probability of burst initiation is well-fit by a survival function (Eq. 2) (Staley et al. 2001) in which N ⫽ 38 and K ⫽ 19. After return to control conditions after a 1-h exposure to sequentially decreasing concentrations of NMDA receptor antagonists during bursting (see METHODS), the burst interval is increased, the cumulative probability curve is flatter, and the data are best fit by Eq. 2 using N ⫽ 20 and K ⫽ 17. B: group-averaged ratios of K to N before (control) vs. after (wash) exposure to NMDA antagonists during bursting. After antagonist exposure, the reduction in N, the number of synapses that are capable of participating in burst initiation, exceeds the reduction in K, the minimum number of synapses that must be available to initiate a burst, significantly increasing the ratio of K to N (P ⬍ 0.001, 2-tailed t-test, n ⫽ 12 slices). The increase in burst interval and relative decrease in the fit parameter N is most consistent with a postsynaptic reduction in strength of CA3–CA3 synapses, and is also seen in the presence of AMPA antagonists (Staley et al. 2001). C and D: whole cell recording before and after D-APV application during bursting demonstrates a reduction in both CA3 burst frequency (C) and sEPSC amplitude (D) consistent with activity and NMDA receptor-dependent reduction in synaptic strength.

place. Sequential reductions in the concentration of competitive NMDA antagonists (either SDZ-220-581 or D-APV) resulted in sequential reductions in burst probability, consistent with additional depotentiation of CA3–CA3 synapses (Fig. 4). This additive depotentiation could not be replicated by prolonged exposure to antagonist at a constant concentration (Fig. 4, A vs. B). Order of antagonist application is crucial for maximal depotentiation Variations in postsynaptic calcium transients at different CA3–CA3 synapses could arise from heterogeneity of synaptic glutamate concentrations, postsynaptic NMDA receptor densities, NMDA receptor conductances, or the link between calcium influx and depotentiation (Grishin et al. 2004). If so, then it might be necessary to use a variety of concentrations of competitive NMDA receptor antagonist to achieve the calcium influx that induced LTD at all recurrent synapses in the CA3 network. However, the degree of depotentiation would not be sensitive to the order of application of the different antagonist concentrations. When the order of application of different concentrations of NMDA receptor antagonists was reversed using identical exposure times (i.e., low concentrations followed by higher concentrations; Fig. 4, C and D), the induction of LTD was significantly reduced compared with the application of high concentrations followed by lower concentrations. These results support the hypothesis that NMDA receptors are removed from the membrane during long-term synaptic weakening and that the withdrawn receptors are part of the pool of NMDA receptors that contribute to synaptic plasticity. J Neurophysiol • VOL

Depotentiation versus LTD of synaptic strength If stabilization of synaptic strength in area CA3 occurs in vivo, then not all CA3-CA3 synapses in an acute slice preparation should be equally plastic (Debanne et al. 1999; Montgomery and Madison 2002; Montgomery et al. 2005). To test whether the synapses that were weakened by partial blockade of NMDA receptors were those that had recently been potentiated during burst induction versus those that were not, we induced bursting by GABAA and GABAB receptor blockade after preventing potentiation by completely blocking NMDA receptors with 80 –100 ␮M DL-APV. Exposure to concentrations of antagonist that completely block the NMDA receptor had no long-term effects on burst probability (Fig. 5, A and B; n ⫽ 7). Subsequent reductions in NMDA receptor antagonist concentrations had no significant effect on burst probability (Fig. 5, C and D; n ⫽ 10). This indicates that partial NMDA receptor blockade during bursting depotentiates synapses that had recently been potentiated during bursting and does not induce LTD of synapses that were not recently potentiated. Delaying the application of NMDA antagonists for as long as 2 h after burst induction did not change the reduction in burst probability observed during sequential application of NMDA receptor antagonist (data not shown; n ⫽ 4). This suggests that the process of conversion of synapses from “recently potentiated” to “stable” is longer than can be observed in the time during which acutely prepared slices are stable. If synapses remain in a highly plastic state for hours after potentiation, and synapses that have not been potentiated are not amenable to NMDA receptor-mediated LTD, then during synchronous network activity all synapses should eventually become trapped in a fully potentiated state. However, even a single application of NMDA antagonist, which did not maxi-


www.jn.org


A

2823

B

C

D

E

F

FIG. 4. Partial blockade of NMDA receptors significantly reduced bursting probability in CA3. A: representative data set showing increased interburst intervals during application of NMDA antagonist (SDZ, 80 ␮M). B: sequential application of lower SDZ concentrations resulted in progressively decreased burst frequencies. C: summary of SDZ data (‚ ⫽ 20 ␮M, 䊐 ⫽ 40 ␮M, E ⫽ 80 ␮M). Interburst intervals progressively increase when the concentration of antagonist is sequentially decreased (P ⫽ 0.03, ANOVA Newman-Keuls, n ⫽ 5). Burst interval, however, is unchanged when identical antagonist concentrations are applied in a sequentially increasing order (P ⫽ 0.63, ANOVA Newman-Keuls, n ⫽ 5). D: summary of D-APV data demonstrating the same effect as observed with SDZ (decreasing D-APV concentrations: P ⫽ 0.008, ANOVA NewmanKeuls, n ⫽ 17; increasing D-APV concentrations: P ⫽ 0.09, ANOVA Newman-Keuls, n ⫽ 5). ‚ ⫽ 2.5 ␮M, 䊐 ⫽ 5 ␮M, E ⫽ 10 ␮M; *, significant from baseline only in the decreasing concentrations groups. E: sample experiment showing increased interburst intervals with application and wash of 10 ␮M D-APV. F: in the grouped data, the increase in mean interburst interval is significant from baseline (P ⫽ 0.01, ANOVA Newman-Keuls, n ⫽ 7). IBI, interburst interval.

mally reduce burst probability, significantly stabilized the burst frequency (Fig. 6; n ⫽ 9). This suggests that under these conditions, submaximal depotentiation induces sufficient NMDA receptor withdrawal to prevent further strength-

ening of the synapse. The data presented in Fig. 2B indicate that this occurs at least in part via withdrawal of NMDA receptors from the membrane. DISCUSSION

A

B

C

D

FIG. 5. Subsequent reductions of D-APV concentrations do not induce de novo LTD. A: blocking NMDA receptor-mediated LTP with 80 –100 ␮M DL-APV prior to burst initiation induces bursting that is stable for hours. B: interburst intervals remained stable from the initial recording (20 min after application of picrotoxin and CGP) through the end of baseline recording (120 –180 min) in all experiments (P ⫽ 0.1, paired t-test, n ⫽ 7). C: after burst induction with complete GABAA, GABAB, and NMDA receptor blockade, subsequent reductions in D-APV concentrations produced no change on interburst interval. D: summary data showing no significant effect on burst induction probability (P ⫽ 0.69, ANOVA NewmanKeuls, n ⫽ 10); 80 ␮M ⫽ DL-APV; all other concentrations ⫽ D-APV.


We demonstrate that NMDA receptor membrane trafficking accompanies both LTP and depotentiation at CA3–CA3 synapses and that these changes are associated with stabilization of synaptic strength and thereby the state of the neural network. Stabilization of the CA3 network in low-probability burst states is unexpected because each burst provides sufficient activation of CA3–CA3 recurrent synapses to induce LTP. This LTP should progressively strengthen synapses so that the only stable state of the synapses is maximum potentiation, and as a consequence, the only stable state of the network is maximal burst probability. However, the recent discoveries that NMDA receptor location changes after LTP (Grosshans et al. 2002a) and LTD (Montgomery et al. 2005; Morishita et al. 2005) raise the possibility of stabilization of intermediate synaptic strengths despite ongoing activity at intensity levels that could induce LTP in naı¨ve synapses. Lisman (2001) proposed that synaptic plasticity is modulated by the amount of calcium influx into dendritic spines: a large influx produces LTP, whereas a smaller influx produces depotentiation and still smaller influxes have no long-term effects on synaptic strength (Fig. 7A). This idea has been confirmed by direct visualization of calcium transients during induction of synaptic plasticity (Ismailov et al. 2004). Thus


www.jn.org

2824


A

B

FIG. 6. Fully potentiated synapses can be depotentiated with a partial block of NMDA receptors. A: sample experiment showing that after 2 h of spontaneous CA3 bursting, a single application of 10 ␮M D-APV increases interburst interval. After complete wash of D-APV, interburst intervals do not return to baseline frequency. B: interburst intervals significantly increased compared with baseline (P ⫽ 0.003, repeated-measures ANOVA Newman-Keuls, n ⫽ 9).

partial blockade of calcium-permeable NMDA receptor-mediated conductances during stimuli that normally induce LTP will instead induce LTD or no change in synaptic strength, depending on the degree of reduction of NMDA receptormediated conductances (Bains et al. 1999; Cummings et al. 1996; Yang et al. 1999). This process is illustrated in Fig. 7. These data support the idea that NMDA receptor trafficking contributes to metaplasticity or the stabilization of altered synaptic strengths (Bear and Abraham 1996; Malenka and Bear 2004; Perez-Otano and Ehlers 2005; Roche et al. 2001). When the density of synaptic NMDA receptors is reduced, it may be

possible to “freeze” a synapse into a state where either no LTP is likely (at intermediate levels of reduction of synaptic NMDA receptor density) or to a state where neither LTP nor LTD could be induced (with more extensive synaptic NMDA receptor reduction; Fig. 7). Our data suggest that these two states can be achieved sequentially. The stability of burst probability after transient, partial NMDA receptor block is consistent with a state in which LTP cannot be induced by stimuli (i.e., spontaneous bursting) that normally induce LTP (Fig. 4). The burst probability minima demonstrated in Fig. 4, C and D, during sequential reductions in NMDA receptor antagonists are consistent with a state in which no further activity and NMDA receptor-dependent long-term weakening is possible. Although it would be of interest to demonstrate that the duration of bursting and/or sequence of exposure to different D-APV concentrations are associated with different levels of membrane-bound receptor expression, these experiments are currently not feasible because the maximum NMDA receptor protein changes we demonstrated were already at the lower limits of resolution by Western blotting. Although we see only a small change (15%) in the subcellular distribution of NMDA receptor subunits, the changes are significant (Fig. 2). Similar significant small changes (10 –20%) have been shown previously using the same method (i.e., cross-linking and subcellular fractionation combined with semi-quantitative Western blot detection) (Goebel et al. 2005; Grosshans et al. 2002a). Synapses that are not subject to NMDA receptor-dependent plasticity have been demonstrated in CA3 (Debanne et al. 1999; Montgomery and Madison 2002) although NMDA receptor-mediated conductances have been demonstrated at these synapses (Debanne et al. 1999). Our data do not exclude the possibility that synaptic NMDA receptors are still present after maximal depotentiation; rather the data predict that the calcium transients produced by the remaining NMDA receptors are not sufficient to induce synaptic plasticity. Published concentration-response curves for D-APV suggest that 10 ␮M D-APV blocks ⬎90% of the NMDA receptorFIG. 7. Illustration of the interaction between NMDA receptor cycling and the sliding calcium scale of synaptic plasticity. A: sliding scale of calcium flux into active synapses where large calcium fluxes produce LTP (green). As calcium influx is reduced, longterm depression (LTD) or depotentiation is induced (red). B: partially blocking NMDA receptors with a competitive antagonist reduces the influx of calcium into the dendritic spine and induces depotentiation (B1), near point a on the graph. In parallel with this 1st depotentiation, a portion of membrane-bound NMDA receptor subunits is endocytosed, decreasing the total NMDA receptor-mediated calcium influx to the spine (B2). If the same amount of D-APV (10 ␮M) is continually applied to this decreased number of available NMDA receptors (B2), the influx of calcium is reduced below the value at which depotentiation occurs, illustrated as a movement to point b on the graph. To elicit further depotentiation, the concentration of NMDA receptor antagonist must be decreased (B3), which increases the flux of calcium, illustrated as movement back toward point a on the graph. This increase in calcium influx further induces both depotentiation and internalization of membrane-bound NMDA receptors (B3). Yellow squares, Ca2⫹; red circles, D-APV.

A

B



www.jn.org


mediated conductance (Wang et al. 2002; Williamson and Wheal 1992), although the precise reduction at each synapse will vary as a function of the concentration of glutamate in the synaptic cleft. Such heterogeneity does not explain the increase in depotentiation seen with sequential application of different concentrations of antagonist because a decreasing sequence of concentrations was necessary for increased depotentiation (Fig. 4). This is most consistent with an altered threshold for depotentiation induced by NMDA receptor trafficking as illustrated in Fig. 7. When LTP was prevented by complete NMDA receptor blockade (i.e., 80 –100 ␮M DL-APV) during burst induction, subsequent partial NMDA receptor blockade had no effect (Fig. 5). This suggests that long-term weakening of recurrent collateral CA3 synapses induced by partial blockade of NMDA receptors during population bursts is due to depotentiation of recently potentiated synapses mediated by NMDA receptors rather than LTD of synapses that were not mediated by NMDA receptor potentiation during the initial bursting. This raises the possibility that partial NMDA receptor antagonism induces activity-dependent, long-term synaptic weakening only at synapses with large numbers of NMDA receptors, such as those synapses that have experienced a recent LTP-associated increase in subsynaptic NMDA receptors (Grosshans et al. 2002a) (Fig. 2A). Alternatively, recently potentiated synapses may be in a unique state with respect to this form of synaptic plasticity (Montgomery and Madison 2004). In the present study, bursts could not be completely stopped by successive applications of lower concentrations of NMDA antagonists. This is consistent with withdrawal of NMDA receptors from recently depotentiated synapses such that no further NMDA receptor-dependent plasticity could be elicited. In prior experiments using submerged slices, bursts could sometimes be stopped completely (Bains et al. 1999), but this likely reflects the slightly diminished excitability of submerged slice preparations and differences in burst induction protocols. Bains et al. tetanized the pyramidal cell layer to potentiate recurrent collateral synapses while we blocked GABAA and GABAB receptors. There are limitations to this study. We did not resolve the subcellular location of membranous NMDA receptors, so we cannot exclude the possibility that the receptors that are added or removed from the membrane during LTP and depotentiation are extrasynaptic (Fig. 2, A and B). However, the inability to increase burst probability after washout of NMDA receptor antagonists is most consistent with a change in the synaptic concentration of the NMDA receptors. Here we show that an increase in CA3 burst frequency is associated with an increase in surface expression of NMDA receptor (Fig. 2A), and this may provide a mechanism for the stabilization of potentiated synapses. Likewise synaptic depotentiation and corresponding decrease in burst frequency are coupled with an increase in the intracellular pool of NMDA receptors (Fig. 2B). This is consistent with withdrawal of NMDA receptors from the postsynaptic membrane as a mechanism for the stabilization of synapses in a depotentiated state with consequent stabilization of the CA3 neuronal network’s burst probability. This could represent a mechanism for persistence of memory. J Neurophysiol • VOL

2825

ACKNOWLEDGMENTS

We thank the members of the Staley and Browning laboratories for critical review and discussion of this work. GRANTS

This research was supported by grants from National Institutes of Health and The Epilepsy Foundation of America to K. J. Staley, and D. R. Grosshans was funded by an individual award from the NIH. REFERENCES

Abegg MH, Savic N, Ehrengruber MU, McKinney RA, Gahwiler BH. Epileptiform activity in rat hippocampus strengthens excitatory synapses. J Physiol 554: 439 – 448, 2004. Abraham WC, Mason-Parker SE, Bear MF, Webb S, Tate WP. Heterosynaptic metaplasticity in the hippocampus in vivo: a BCM-like modifiable threshold for LTP. Proc Natl Acad Sci USA 98: 10924 –10929, 2001. Aradi I, Maccaferri G. Cell type-specific synaptic dynamics of synchronized bursting in the juvenile CA3 rat hippocampus. J Neurosci 24: 9681–9692, 2004. Bains JS, Longacher JM, Staley KJ. Reciprocal interactions between CA3 network activity and strength of recurrent collateral synapses. Nat Neurosci 2: 720 –726, 1999. Bashir ZI, Alford S, Davies SN, Randall AD, Collingridge GL. Long-term potentiation of NMDA receptor-mediated synaptic transmission in the hippocampus. Nature. 349: 156 –158, 1991. Bear MF, Abraham WC. Long-term depression in hippocampus. Annu Rev Neurosci 19: 437– 462, 1996. Behrens CJ, van den Boom LP, de Hoz L, Friedman A, Heinemann U. Induction of sharp wave-ripple complexes in vitro and reorganization of hippocampal networks. Nat Neurosci 8: 1560 –1567, 2005. Buzsaki G. Hippocampal sharp-waves: their origin and significance. Brain Res 398: 242–252, 1986. Buzsaki G. Two-stage model of memory trace formation: a role for “noisy” brain states. Neuroscience 31: 551–570, 1989. Chamberlin NL, Traub RD, Dingledine R. Role of EPSPs in initiation of spontaneous synchronized burst firing in rat hippocampal neurons bathed in high potassium. J Neurophysiol 64: 1000 –1008, 1990. Coultrap SJ, Nixon KM, Alvestad RM, Valenzuela CF, Browning MD. Differential expression of NMDA receptor subunits and splice variants among the CA1, CA3 and dentate gyrus of the adult rat. Brain Res Mol Brain Res 135: 104 –111, 2005. Cummings JA, Mulkey RM, Nicoll RA, Malenka RC. Ca2⫹ signaling requirements for long-term depression in the hippocampus. Neuron 16: 825– 833, 1996. Debanne D, Gahwiler BH, Thompson SM. Heterogeneity of synaptic plasticity at unitary CA3–CA1 and CA3–CA3 connections in rat hippocampal slice cultures. J Neurosci 19: 10664 –10671, 1999. Delgado JY, O’Dell TJ. Long-term potentiation persists in an occult state following mGluR-dependent depotentiation. Neuropharmacology 48: 936 – 948, 2005. Dulla CG, Dobelis P, Pearson T, Frenguelli BG, Staley KJ, Masino SA. Adenosine and ATP link PCO2 to cortical excitability via pH. Neuron 48: 1011–1023, 2005. Dunwiddie TV. Endogenously released adenosine regulates excitability in the in vitro hippocampus. Epilepsia 21: 541–548, 1980. Dzhala VI, Staley KJ. Transition from interictal to ictal activity in limbic networks in vitro. J Neurosci 23: 7873–7880, 2003. Goebel SM, Alvestad RM, Coultrap SJ, Browning MD. Tyrosine phosphorylation of the N-methyl-D-aspartate receptor is enhanced in synaptic membrane fractions of the adult rat hippocampus. Brain Res Mol Brain Res 142: 65–79, 2005. Grishin AA, Gee CE, Gerber U, Benquet P. Differential calcium-dependent modulation of NMDA currents in CA1 and CA3 hippocampal pyramidal cells. J Neurosci 24: 350 –355, 2004. Grosshans DR, Clayton DA, Coultrap SJ, Browning MD. LTP leads to rapid surface expression of NMDA but not AMPA receptors in adult rat CA1. Nat Neurosci 5: 27–33, 2002a. Grosshans DR, Clayton DA, Coultrap SJ, Browning MD. Analysis of glutamate receptor surface expression in acute hippocampal slices. Sci STKE 137: PL8, 2002b. Hall RA, Soderling TR. Differential surface expression and phosphorylation of the N-methyl-D -aspartate receptor subunits NR1 and NR2 in cultured hippocampal neurons. J Biol Chem 272: 4135– 4140, 1997a.


www.jn.org

2826


Hall RA, Soderling TR. Quantitation of AMPA receptor surface expression in cultured hippocampal neurons. Neuroscience 78: 361–371, 1997b. Ismailov I, Kalikulov D, Inoue T, Friedlander MJ. The kinetic profile of intracellular calcium predicts long-term potentiation and long-term depression. J Neurosci 24: 9847–9861, 2004. Jefferys JG. Mechanisms and experimental models of seizure generation. Curr Opin Neurol 11: 123–127, 1998. Kauer JA, Malenka RC, Nicoll RA. A persistent postsynaptic modification mediates long-term potentiation in the hippocampus. Neuron 1: 911–917, 1988. Li XG, Somogyi P, Ylinen A, Buzsaki G. The hippocampal CA3 network: an in vivo intracellular labeling study. J Comp Neurol 339: 181–208, 1994. Lisman JE. Three Ca2⫹ levels affect plasticity differently: the LTP zone, the LTD zone and no man’s land. J Physiol 532: 285, 2001. Malenka RC, Bear MF. LTP and LTD: an embarrassment of riches. Neuron 44: 5–21, 2004. Miles R, Poncer JC. Recurrent excitatory synapses and circuits in the CA3 region of the hippocampus. In: Excitatory Amino Acids and Synaptic Transmission (2nd ed.). San Diego, CA: Academic, 1995, p. 191–201. Miles R, Wong RK. Inhibitory control of local excitatory circuits in the guinea pig hippocampus. J Physiol 388: 611– 629, 1987. Montgomery JM, Madison DV. State-dependent heterogeneity in synaptic depression between pyramidal cell pairs. Neuron 33: 765–777, 2002. Montgomery JM, Madison DV. Discrete synaptic states define a major mechanism of synapse plasticity. Trends Neurosci 27: 744 –750, 2004. Montgomery JM, Selcher JC, Hanson JE, Madison DV. Dynamin-dependent NMDAR endocytosis during LTD and its dependence on synaptic state. BMC Neurosci 6: 48, 2005. Morishita W, Marie H, Malenka RC. Distinct triggering and expression mechanisms underlie LTD of AMPA and NMDA synaptic responses. Nat Neurosci 8: 1043–1050, 2005. Perez-Otano I, Ehlers MD. Homeostatic plasticity and NMDA receptor trafficking. Trends Neurosci 28: 229 –238, 2005. Philpot BD, Espinosa JS, Bear MF. Evidence for altered NMDA receptor function as a basis for metaplasticity in visual cortex. J Neurosci 23: 5583–5588, 2003.


Roche KW, Standley S, McCallum J, Dune Ly C, Ehlers MD, Wenthold RJ. Molecular determinants of NMDA receptor internalization. Nat Neurosci 4: 794 – 802, 2001. Smith KE, Gibson ES, Dell’Acqua ML. cAMP-dependent protein kinase postsynaptic localization regulated by NMDA receptor activation through translocation of an a-kinase anchoring protein scaffold protein. J Neurosci 26: 2391–2402, 2006. Snell LD, Nunley KR, Lickteig RL, Browning MD, Tabakoff B, Hoffman PL. Regional and subunit specific changes in NMDA receptor mRNA and immunoreactivity in mouse brain following chronic ethanol ingestion. Brain Res Mol Brain Res 40: 71–78, 1996. Staley KJ, Bains JS, Yee A, Hellier J, Longacher JM. Statistical model relating CA3 burst probability to recovery from burst-induced depression at recurrent collateral synapses. J Neurophysiol 86: 2736 –2747, 2001. Staley KJ, Longacher M, Bains JS, Yee A. Presynaptic modulation of CA3 network activity. Nat Neurosci 1: 201–209, 1998. Stasheff SF, Anderson WW, Clark S, Wilson WA. NMDA antagonists differentiate in epileptogenesis from seizure expression in an in vitro model. Science 245: 648 – 651, 1989. Traub R, Miles R. Neuronal Networks of the Hippocampus. Cambridge, UK: Cambridge Univ. Press, 1991. Urwyler S, Laurie D, Lowe DA, Meier CL, Muller W. Biphenyl-derivatives of 2-amino-7-phosphonoheptanoic acid, a novel class of potent competitive N-methyl-D -aspartate receptor antagonists. I. Pharmacological characterization in vitro. Neuropharmacology 35: 643– 654, 1996. Vissel B, Krupp JJ, Heinemann SF, Westbrook GL. A use-dependent tyrosine dephosphorylation of NMDA receptors is independent of ion flux. Nat Neurosci 4: 587–596, 2001. Wang G, Ding S, Yunokuchi K. Glutamate-induced increases in intracellular Ca2⫹ in cultured rat neocortical neurons. Neuroreport 13: 1051–1056, 2002. Williamson R, Wheal HV. The contribution of AMPA and NMDA receptors to graded bursting activity in the hippocampal CA1 region in an acute in vitro model of epilepsy. Epilepsy Res 12: 179 –188, 1992. Yang SN, Tang YG, Zucker RS. Selective induction of LTP and LTD by postsynaptic [Ca2⫹]i elevation. J Neurophysiol 81: 781–787, 1999.


www.jn.org