Extrasynaptic and synaptic NMDA receptors form stable and uniform pools in rat hippocampal slices

Abstract

N-methyl-d-aspartate receptor (NMDAR) activation can trigger both long- and short-term plasticity, promote cell survival, and initiate cell death. A number of studies suggest that the consequences of NMDAR activation can vary widely depending on whether synaptic or extrasynaptic receptors are activated. Here we have examined the spatial distribution of NMDARs of CA1 pyramidal neurons in acutely dissected hippocampal slices. Using a physiological definition of extrasynaptic receptors as those not accessible to single release events, we find that extrasynaptic NMDARs comprise a substantial proportion of the dendritic NMDAR pool (36%). This pool of extrasynaptic NMDARs is stable and does not shuttle into the synaptic receptor pool, as we observe no recovery of synaptic current after MK-801 synaptic blockade and washout. The subunit composition of synaptic and extrasynaptic NMDA receptor pools is similar at 3 weeks of age, with NR2B subunits present in both compartments. NR2B receptors are not enriched in the extrasynaptic compartment. Our data suggest that any role played by extrasynaptic NMDARs in synaptic transmission is dictated by their subcellular location rather than their subunit composition or mobility.

N-methyl-d-aspartate receptors (NMDARs) are a critical component of excitatory transmission in the central nervous system. Activating these receptors can trigger both long- and short-term plasticity, promote cell survival, and initiate cell death. Previous work suggests that NMDARs exist both at the synapse and on the extrasynaptic membrane and may be linked to different signalling pathways (Tovar & Westbrook, 1999; Hardingham & Bading, 2002; Hardingham et al. 2002; Li et al. 2002; Lozovaya et al. 2004; Thomas et al. 2005; Zhang et al. 2007). For example, while synaptic NMDAR activation can result in an anti-apoptotic cAMP response element binding protein (CREB) signal, activation of extrasynaptic NMDARs by exogenous glutamate may dephosphorylate CREB resulting in neuronal cell death (Hardingham et al. 2002). It has also been proposed that extrasynaptic NMDARs could serve as a reserve pool, putatively diffusing into the synapse over the course of minutes (Tovar & Westbrook, 2002).

In this study we have examined the pool size and composition of dendritic extrasynaptic NMDARs in acutely dissected hippocampal slices. Extrasynaptic NMDARs were isolated by blocking synaptic receptors with MK-801 (0.1 Hz electrical stimulation), and activated with focal photolysis of 4-methoxy-7-nitroindolinyl (MNI)-glutamate. We find that they represent a substantial portion of functional dendritic NMDARs (36%). The size of the extrasynaptic pool was similar for both proximal and distal dendrites. In contrast to previous reports, extrasynaptic NMDARs were not enriched in NR2B subunits, with both synaptic and extrasynaptic compartments comprised of roughly half NR2A-containing and half NR2B-containing NMDARs. Extrasynaptic NMDARs were relatively immobile and did not move rapidly from the extrasynaptic to the synaptic compartment. These results suggest that any difference in signalling by synaptic and extrasynaptic NMDARs must be due to intracellular signalling pathways rather than subunit identity or mobility.

Methods

Synaptic transmission

Animals were anaesthetized with trifluoroethane and decapitated to prepare 300–400 μm thick coronal slices from the hippocampus of P14–P22 rats (Yang et al. 2006). All experiments conformed with the guidelines laid down by the Albert Einstein College of Medicine animal welfare committee. Whole-cell voltage clamp recordings were made from CA1 pyramidal neurons. The patch pipettes were filled with a caesium gluconate solution containing (mm): 123 caesium gluconate, 8 NaCl, 1 CaCl₂, 10 EGTA, 10 Hepes, 10 glucose, 5 ATP, 0.4 GTP, 1 QX-314 (pH 7.2; 280–290 mOsmol l⁻¹). Slices were superfused (3–4 ml min⁻¹) at room temperature (25°C) with oxygenated physiological saline (mm: 119 NaCl, 2.5 KCl, 1.3 or 0.1 MgCl₂, 2.5 CaCl₂, 1 NaH₂PO₄, 26.2 NaHCO₃, 11 glucose). All mobility experiments were performed at 34°C.

Recordings were rejected if the holding current was greater than −100 pA when pyramidal cells were voltage clamped at −60 mV. NMDA receptor currents were isolated by including NBQX (5 or 10 μm) and picrotoxin (100 μm) in the external solution. NMDA currents were recorded by relieving Mg²⁺ block with depolarization. Initial experiments were carried out at +30 mV (N= 4), but we subsequently determined that cells were easier to maintain when they were subjected to less depolarization. As a result, we used depolarization sufficient to elicit substantial NMDAR currents (∼−20 mV) in all subsequent experiments. We did not see any relationship between the amount of depolarization and the percent of extrasynaptic NMDAR measured (r²= 0.15).

MK-801 (50 μm in bath or 500 μm via picospritzer), d-APV (50 μm), ifenprodil (3 μm), and Ro 25-6981 (1 μm; Tocris) were used to block NMDA receptors. Synaptic stimulation was achieved by placing a monopolar or bipolar theta glass stimulating electrode adjacent to the dendrite to activate synapses (every 10–30 s; 40–200 μA, 100–200 μs). Repeated electrical stimulation (0.1 or 0.2 Hz) in the presence of MK-801 was used to block synaptic NMDARs. The different stimulating electrodes yielded similar results.

For comparisons of decay kinetics, currents were fitted with double exponentials (y=y₀+A₁e^−τ1+A₂e^−τ2). The weighted tau was then calculated as τ_w=A₁τ₁/(A₁+A₂) +A₂τ₂/(A₁+A₂) (Stocca & Vicini, 1998). Where A1 and A2 are the amplitudes of the fast and slow decay constants and τ1 and τ2 are their associated time constants. The decays of synaptic currents (τ_w= 88 ± 6 ms; N= 23), photolytic EPSCs (phEPSCs; τ_w= 98 ± 14 ms; N= 15), single spontaneous NMDAR currents (τ_w= 81 ± 6 ms; mean amplitude 29 ± 5 pA; N= 7), and currents from cells in ifenprodil (τ_w= 90.3 ± 9.1 ms; N= 15) did not differ significantly when fitted with double exponentials (ANOVA, P > 0.6). As the mobility experiments were performed at 34°C, we saw shorter decay times for these currents (τ_w= 52 ± 6 ms; N= 4).

Estimation of the extrasynaptic pool was corrected for any residual synaptic current, using the calculation: E= (T′−fT)/(1 −f), where E is the extrasynaptic NMDAR current, T′ the peak photolytic current amplitude after synaptic blockade, f the fraction of residual synaptic current after synaptic blockade and T the peak photolytic current amplitude prior to synaptic blockade. We observed 100% block of synaptic NMDAR current in 7 of 12 of experiments used for estimates of extrasynaptic pool size. For the experiments with incomplete block, 6.5 ± 0.8% (N= 5) of the current remained unblocked. Estimates of extrasynaptic NMDAR population were independent of percentage block (35 ± 5 versus 32 ± 5, P > 0.6, Student's t test).

Uncaging

Individual neurons were visualized with a 40× water immersion objective (Olympus). For the photolysis of caged glutamate (200 μm, MNI-glutamate, Tocris), the output of a continuous emission 5 W krypton ion laser (Coherent, Innova 302) with a 351 nm line was delivered, via a multimode optical fibre, through an Olympus 40× water-immersion objective to form an uncaging spot about 6 μm wide (this size uncaging area was used for all experiments except those in Fig. 2) with a final power of 5–10 mW (Pettit et al. 1997; Yang et al. 2007). We estimate this will produce a glutamate concentration of 100 μm (Wang & Augustine, 1995; Canepari et al. 2001). An acousto-optical modulator was used to vary the duration of the light pulse between 1 and 2 ms to produce phEPSCs with similar decay times to synaptic currents. The uncaging spot was positioned over a cellular process by including a fluorescent dye (Oregon Green, 200 μm; Molecular Probes, Eugene, OR, USA), in the patch pipette solution and then visualizing the cell with an Olympus Fluoview 300 confocal microscope. Data were collected and analysed off line with Igor Pro (Wavemetrics, Lake Oswego, OR, USA).

Figure 2. Activation of all functional synapses with local stimulation.

A, a transmitted light image of the stimulating pipette configuration. A stimulating pipette (S1) was placed ∼100 μm from the apical dendrite, and a second pipette was positioned over the dendrite (S2). Whole-cell voltage clamp recordings of NMDAR currents were recorded (R) every 15 s by interleaving stimulation of S1 and S2. B, a plot of current amplitude versus time. NMDAR currents were evoked with S1 and S2 from a CA1 pyramidal cell. Synaptic receptors activated by S1 were blocked with the use-dependent NMDA antagonist MK-801 (50 μm). After block by MK-801, stimulation with S2 elicited no synaptic currents demonstrating that no unblocked synapses remained within the stimulation area (N= 3).

Results

Extrasynaptic receptors comprise a large proportion of dendritic NMDARs

We determined the synaptic and extrasynaptic distribution of dendritic NMDARs in the semi-intact preparation of hippocampal brain slices using a combination of electrophysiology, local photolysis and confocal imaging. Since we are interested in the consequences of activating functional NMDARs, we have taken a physiological rather than an anatomical approach to classifying synaptic and extrasynaptic NMDARs. We define synaptic NMDARs as those functional receptors that can respond to the glutamate released during low-frequency (0.1 Hz) synaptic events. In contrast, extrasynaptic NMDARs do not respond to low-frequency synaptic events within the time frame of our experiments. While making sharp distinctions between synaptic and extrasynaptic NMDARs is difficult, we have followed the convention of physiologists in the field (Chen & Diamond, 2002; Clark & Cull-Candy, 2002; Tovar & Westbrook, 2002; Lozovaya et al. 2004; Scimemi et al. 2004).

To determine the proportion of extrasynaptic receptors in the total dendritic NMDA receptor population in acute hippocampal slices, we recorded whole-cell currents in CA1 pyramidal cells from juvenile rats (P14–P21; 25°C). To isolate NMDARs, AMPA and GABA_A receptors were blocked with NBQX (5 μm) and picrotoxin (100 μm). Synaptic NMDARs were activated by electrically stimulating presynaptic inputs in stratum radiatum every 10–30 s (single shock). NMDA currents were recorded by relieving Mg²⁺ block with depolarization to ∼−20 mV. We used photolysis of MNI-glutamate (200 μm; 1–2 ms) to activate extrasynaptic and synaptic NMDARs on a focal region (6 μm) of dendrite contained within the area of synaptic stimulation. The decay kinetics of currents evoked by both photolysis (phEPSC) and electrical stimulation were well fitted with a double exponential and did not differ significantly (synaptic τ_w= 88 ± 6 ms, N= 23; phEPSC τ_w= 98 ± 14 ms, N= 15; P > 0.4).

To isolate extrasynaptic NMDARs from the mixed population of synaptic and extrasynaptic NMDARs activated by photolysis, we removed the synaptic NMDARs with the irreversible use-dependent NMDAR blocker MK-801 (50 μm). Repeated synaptic stimulation (∼100 stimuli) with MK-801 progressively blocked the NMDARs of even low probability synapses. To measure the per cent of NMDARs that are extrasynaptic, we compared the amplitude of the photolytic currents before and after removal of the synaptic receptors. As the distribution of receptors may not be uniform throughout the length of the dendrite (Dodt et al. 1998; Magee & Cook, 2000; Pettit & Augustine, 2000), we compared the extrasynaptic NMDAR content of the proximal region of the apical dendrite (< 100 μm) with that of the distal dendrite (> 100 μm; Fig. 1E and F). Following blockade of synaptic NMDARs, focal photolysis revealed a current mediated by extrasynaptic receptors representing 32 ± 6% (N= 5; proximal) and 38 ± 4% (N= 7; distal) of the total NMDAR pool (Fig. 1E and F; post-MK-801 phEPSC peak amplitude/control phEPSC amplitude). The difference in the size of the extrasynaptic NMDAR pool at proximal and distal regions of the dendrite was not significant (P > 0.2; Student's t test) and data from both sets of experiments were pooled to obtain an overall average of 36 ± 4% (N= 12) of the total NMDAR pool. These results demonstrate that a substantial proportion of functional NMDARs located on the dendrite do not access the glutamate released under baseline synaptic conditions.

Figure 1. Extrasynaptic NMDARs represent a large proportion of the total population.

A, a live confocal image of an individual CA1 pyramidal neuron filled with Oregon green–BAPTA (200 μm). Note the stimulating pipette adjacent to the secondary apical dendrite and cyan circle indicating location and size of photolytic area. B, a plot of current amplitude versus time for synaptic and phEPSCs. Currents were elicited every 30 s before and after MK-801 application. Synaptic currents were blocked by stimulating at 0.1 Hz in the presence of MK-801 (50 μm). APV (50 μm) was added at 30 min. Numbers correspond to traces in C. C, sample currents from the graph in B. D, another experiment showing the isolation of extrasynaptic receptors with incomplete synaptic blockade and the sensitivity of extrasynaptic NMDARs to MK-801. Baseline transmission was elicited every 20 s and synaptic currents were blocked with MK-801 (50 μm) at 0.1 Hz. E, an example of a cell where the stimulating electrode is positioned within 100 μm of the soma. F, an example of a cell where the stimulating electrode is positioned more than 100 μm from the soma. The two group means were not significantly different (Student's t test, P > 0.05; N= 12) and were pooled for an average of 36 ± 4%.

When calculating the extrasynaptic NMDAR pool size, the photolytic area must lie within the area of synaptic receptor activation. To ensure this, we stimulated synapses on a length of dendrite as large as possible. We estimated the length of dendrite stimulated by assuming a quantal amplitude of 5 pA (McAllister & Stevens, 2000; Pankratov & Krishtal, 2003), a release probability of ∼0.2 (Hessler et al. 1993; Rosenmund et al. 1993), and 1–3 spines (μm of dendrite)⁻¹ (Harris et al. 1992; Ishizuka et al. 1995; Trommald et al. 1995). Therefore, a 200 pA current would be generated over an area 67–200 μm in length (200 pA/5 pA/1 or 3 spines/0.2 probability). To directly test this estimate, we monitored the calcium signal generated by synaptic stimulation of NMDARs in separate experiments. Single pulses of synaptic stimulation increase Ca²⁺ levels at distances of at least 80 μm along the length of the dendrite (online Supplemental Fig. 1). While interpretation of this result is complicated by diffusion of the calcium signals, it is clear that synaptic stimulation activates a length of dendrite at least 10-fold larger than that activated by photolysis. Although we always place our uncaging spot over the dendrite closest to the stimulation pipette, we find that all dendrites parallel to the stimulation pipette show calcium influx (Supplemental Fig. 1). This result is expected as excitatory inputs to CA1 cells in stratum radiatum run approximately orthogonal to the pyramidal cell apical dendrites (Andersen et al. 1980; Amaral & Witter, 1989).

Additional experiments were performed to test whether our stimulation protocol activates all functional synapses on the targeted region of dendrite. Failure to do so would result in an overestimate of the extrasynaptic pool size as these unactivated synapses would remain unblocked by MK-801 and appear to be synaptic. We used two stimulating pipettes for these experiments, with one pipette (S1) placed within 100 μm of the apical dendrites of test neurons as described for experiments in Fig. 1. The second stimulating electrode (S2) was placed parallel to the first and as close as possible to the dendrite (Fig. 2A). The ratio of S1/S2 current amplitudes was set at 2–3 with an average of 2.5 ± 0.3 (mean amplitudes, S1 = 346 ± 35 pA and S2 = 141 ± 4 pA). Stimulation intensity was 200 μA for 100 μs (S1) and 40 μA for 100 μs (S2) and current decays for S1 and S2 were similar (62 ± 9 and for S2 71 ± 7 ms, respectively). After establishing baseline transmission for both electrodes (every 15 s with S1 and S2 stimulation interleaved), all synaptic current generated by the distal electrode was blocked with MK-801 (50 μm). Following blockade, we attempted to elicit synaptic currents with the proximal stimulating electrode. Given the proximity of this electrode to the dendrite, we expect it to activate all local functional synapses by direct depolarization. If there were a significant number of unblocked synapses we should see residual current. We found that block of synaptic currents generated by the distal stimulation pipette also blocked synaptic NMDAR currents evoked by the second proximal pipette (N= 3; Fig. 2B). This result suggests that our stimulation protocol effectively activates all functional synapses on the targeted region of dendrite.

Finally, NMDARs are subject to random stochastic channel openings. If extrasynaptic receptors were to open in the presence of MK-801, they could be blocked causing us to underestimate the pool size. To limit this problem, MK-801 blocking time was kept to a maximum of 30 min. This limit was chosen because a plot of extrasynaptic pool size versus time to block showed no correlation between these two measures (r²= 0.02; Fig. 3). These results suggest that random block of channels due to stochastic openings is limited for the period we apply MK-801.

Figure 3. Extrasynaptic pool size estimates were independent of time in MK-801.

The time from the beginning of MK-801 application to block of synaptic receptors versus the estimate of extrasynaptic pool size plotted for each cell (N= 12). There was no correlation between the time in MK-801 and pool size (r²= 0.02).

Photolysis activates synaptic and extrasynaptic receptors

Our method of isolating extrasynaptic NMDARs rests on the assumption that uncaged glutamate activates both the synaptic and extrasynaptic pool of NMDARs. However, the possibility exists that caged agonists have limited access to the synaptic compartment due to steric constraints. To test this possibility we used repeated uncaging combined with MK-801 to block all the phEPSC (Fig. 4). Subsequent stimulation of the synaptic inputs failed to elicit a NMDAR current (N= 5). These experiments require that the uncaging area be as large as possible and that activated synapses are contained within the uncaging area. This was achieved by increasing the magnitude of the phEPSC with longer uncaging times (3–5 ms) and decreasing the size of the stimulation current until the photolytically activated region was larger than the synaptic stimulation area (Fig. 4A). This experiment confirms that photolysis activates both synaptic and extrasynaptic receptors. It also provides additional evidence that positioning the uncaging beam orthogonal to the stimulation pipette correctly matches the two stimulation areas.

Figure 4. Synaptic stimulation and photolysis activate the same region of dendrite.

A, a live confocal image of an individual CA1 pyramidal neuron, stimulating and whole-cell pipette. Circles indicate locations of photolysis. B, synaptic and phEPSCs before and after MK-801 (50 μm) application. Numbers correspond to locations on graph in C. C, a plot of current amplitude versus time for phEPSC (red circles) and synaptic currents (black circles). Currents acquired during MK-801 wash-in are not plotted. MK-801 block of phEPSCs also blocks synaptically stimulated NMDARs. Photolysis at another location generates current (green circle). Some data points collected during MK-801 blockade are omitted due to axis break.

Extrasynaptic NMDARs are not enriched with NR2B subunits

Previous work has suggested that NR2A- and NR2B-containing receptors can be found in both synaptic and extrasynaptic compartments (Kirson & Yaari, 1996; Li et al. 2002; Thomas et al. 2005). To investigate NMDAR subunit localization in acute slices we examined the sensitivity of NMDAR field potentials to the NR2B-selective antagonists ifenprodil (3 μm) and Ro 25-6981 (Ro25; 1 μm) in slices from rats aged P5–P92 (Fig. 5A). Both antagonists blocked significantly more synaptic NMDARs in young animals (P5 and 2–3 weeks) than in older animals (> 5 weeks), consistent with studies reporting high NR2B expression early in postnatal development (Monyer et al. 1994; Li et al. 1998; Tovar & Westbrook, 1999). These data demonstrate the presence of significant numbers of synaptic NR2B subunits in animals 2–3 weeks of age.

Figure 5. Extrasynaptic NMDARs are not enriched in NR2B subunits in rats 2–3 weeks of age.

A, a plot of the reduction in NMDAR field potentials following application of the NR2B selective antagonists ifenprodil (3 μm) or Ro 25-6981 (1 μm). Low Mg²⁺ (0.1 mm) was used to elicit field EPSPs. No differences between the two antagonists were observed. Neonate (P5), juvenile (P13–21) and adult (P39–P92) animals showed a 45% (41 ± 11, 48 ± 8), 20% (24 ± 3, 17 ± 6) and 0% (−2.3 ± 3, 2.5 ± 2.5) block, respectively. The degree of block by ifenprodil was significantly different across all groups (one-way ANOVA, P < 0.001; N= 33). Insets, averages of 60 field potentials before (black) and after (red) Ro 25-6981 application. B, individual synaptic spines are pictured with NMDARs. Whole-cell NMDAR currents were elicited by electrical stimulation, photolysis over the soma (red star), photolysis over the dendrite, and photolysis over the dendrite after removal of synaptic NMDARs with MK-801. C, single traces before (white) and after bath application of the NR2B-selective antagonist ifenprodil (red; 3 μm) from P13–P21 rats. Stable ifenprodil blockade was achieved in 11 ± 1 min (N= 16; s.d.= 4). Synaptic and extrasynaptic currents were reduced to a similar extent by ifenprodil (3 μm). Synaptic currents were reduced by 46.6 ± 7.7%, somatic (extrasynaptic) by 42.2 ± 9.3%, dendritic (synaptic + extrasynaptic) by 45.8 ± 3.8%, and extrasynaptic (photolysis following MK-801 application) by 38.2 ± 5.8%. There was no significant difference in degree of block across all currents (ANOVA, P > 0.05) suggesting that NR2B subunits are not enriched in the extrasynaptic receptor pool.

To test whether the extrasynaptic pool is enriched for NR2B receptors, we examined the sensitivity of whole-cell extrasynaptic NMDAR currents to bath application of ifenprodil (3 μm). As there are no excitatory synapses on the soma (Harris et al. 1992; Megias et al. 2001), we uncaged glutamate (200 μm) over the soma as an initial measure of extrasynaptic receptors. Ifenprodil reduced somatic phEPSC amplitude 42.2 ± 9.3% (Fig. 5B and C; N= 5). Given that somatic NMDARs may not have the same subunit composition as dendritic extrasynaptic NMDARs, we next examined the NR2B content of mixed extrasynaptic and synaptic dendritic NMDARs. Ifenprodil reduced dendritic phEPSCs 45.8 ± 3.8% (Fig. 5B and C; N= 5).

Finally, we determined the sensitivity of the dendritic extrasynaptic NMDAR pool to ifenprodil. Extrasynaptic NMDARs were isolated by removing synaptic receptors with MK-801, and extrasynaptic NMDAR currents were elicited by photolysis over the dendrite. Given that whole-cell synaptic NMDAR currents were reduced by 46.6 ± 7.7% (Fig. 5B and C; N= 7), ifenprodil should reduce extrasynaptic current amplitude more than 45% if these receptors were enriched for NR2B. We found that extrasynaptic NMDAR currents were blocked by 38.2 ± 5.8% (Fig. 5B and C; N= 6). There were no significant differences in current block across all whole-cell experiments (ANOVA, P > 0.8; N= 23). These data demonstrate that NR2B subunits are not enriched in the extrasynaptic receptor compartment. Given that the NR2B-sensitive component of the extrasynaptic pool is equal to that of the synaptic, these results argue against a model that strictly compartmentalizes NR2A- and NR2B-containing NMDARs to synaptic and extrasynaptic locations.

Recombinant NR2B-containing receptors decay more slowly than NR2A-containing receptors (Monyer et al. 1994). The data for native receptors appear much more controversial with some groups reporting changes in decay kinetics following NR2B antagonists (Steigerwald et al. 2000; Lopez de Armentia & Sah, 2003; Scimemi et al. 2004), while others report either no difference in the decay kinetics of ifenprodil-sensitive and -insensitive NMDAR currents (Kirson & Yaari, 1996; Stocca & Vicini, 1998; Dalby & Mody, 2003; Massey et al. 2004), or a dissociation between the sensitivity to ifenprodil and the decay kinetics (Steigerwald et al. 2000; Barth & Malenka, 2001). The decays of synaptic, somatic and dendritic currents were well fitted by double exponentials, and pairwise comparison of kinetics before and after ifenprodil application revealed no significant differences (Student's t test, P > 0.7; N= 15). The absence of a change in the kinetics could be the result of triheteromeric (NR1–NR2A–NR2B) NMDARs, the presence of NR2D receptors, or of post-translational modifications of native receptors.

Synaptic and extrasynaptic NMDARs form stable populations

Previous work suggests that NMDARs are mobile and can shuttle between the synaptic and extrasynaptic membrane within minutes (Tovar & Westbrook, 2002; Groc et al. 2006). To evaluate extrasynaptic NMDAR mobility in acute slices we applied MK-801 by picospritzer (2 min) during electrical stimulation (0.1 Hz, ∼10 stimulations) to block whole-cell synaptic NMDAR currents. This reduced response amplitudes by 20–50%. MK-801 was then washed out for at least 20 min. This time period was sufficient to wash out MK-801 as demonstrated by stable current amplitudes and comparable decay kinetics post-wash (Fig. 6A and B; Student's t test, P > 0.2; N= 4). Total recovery time across all experiments (wash time + stimulation time) ranged from 31 to 58 min with an average of 45.8 ± 5.9 min (last 3 traces of recording). Recovery was measured as the difference between the average current amplitude before (last 3 traces before wash) and after MK-801 washout. We observed no significant recovery (1.8 ± 4%; paired Student's t test, P > 0.3; N= 4).

Figure 6. Extrasynaptic NMDARs in acute slices are immobile.

A, single traces before (1) brief MK-801 application and after (2) MK-801 washout. When scaled, the two traces overlap. B, brief exposure to MK-801 (2 min) during synaptic stimulation (45–100 μA, 100 μs) resulted in partial block of synaptic currents. Following washout of MK-801 (20 min), no significant recovery was observed (inset, 1.8 ± 4%; N= 4). Average peak amplitude before block was 139.7 ± 13.7 pA and 72.9 ± 18.6 pA post-block (N= 4). C, single traces before (1), during (2) and after (3) MK-801 application. When traces 1 and 2 are scaled, the faster decay time in MK-801 is revealed. D, following block of synaptic receptors, 20 min were required to washout MK-801. Inset, current recovery was negligible at 2.5 ± 1% following washout (N= 4).

Given that the mobile receptor population could be a small fraction of the total, we also tested for mobility following complete block of synaptic NMDAR current (Fig 6C and D; mean block time 14 ± 3 min). Current decay times decreased significantly in the presence of MK-801 due to ongoing block of NMDARs (paired t test, P < 0.02; N= 10; Fig. 6C). Complete block of synaptic NMDARs should allow for detection of a very small increase in functional synaptic receptors. Again, recovery was limited to 2.5 ± 1% following MK-801 washout (last three traces for a mean recovery time of 23 ± 1 min; range 21–25 min; N= 4). The absence of synaptic currents was not a result of complete NMDAR blockade, as it was possible to evoke stable EPSCs by increasing the stimulation intensity to recruit additional unblocked synaptic inputs (data not shown). These results suggest that although extrasynaptic receptors form a substantial NMDAR population, they represent a stable pool of receptors. Mobility is either limited or very slow in acute slices. This stability implies that extrasynaptic NMDARs do not simply form a reserve pool waiting to replace synaptic receptors.

Discussion

We find a substantial pool of extrasynaptic receptors (36%), in acutely dissected hippocampal slices. These extrasynaptic NMDARs are either immobile or move very slowly in acutely dissected hippocampal slices. Earlier experiments using cultured hippocampal neurons showed that extrasynaptic NMDARs were highly mobile and able to diffuse into the synaptic compartment within minutes (Tovar & Westbrook, 2002; Groc et al. 2006), suggesting they might play a role in modulating synaptic NMDAR responses. A possible explanation for this difference may be that receptors in intact tissue are stabilized by extracellular matrix cues. While lack of mobility may at first seem limiting with respect to their physiological role, it is possible that periods of high activity under normal or pathological conditions could activate them, making mobility unnecessary. In addition, NMDARs have recently been shown to dynamically traffic in response to neuronal activity providing an alternate mechanism for modulation of synaptic strength (Lau & Zukin, 2007). It is also possible that some pathological or plasticity-inducing conditions might increase NMDAR mobility, as they have been shown to cause NMDAR insertion (Grosshans et al. 2002; Carpenter-Hyland et al. 2004).

Previous work has suggested that receptor localization determines subunit composition, leading to the activation of different intracellular signalling pathways (Hardingham et al. 2002; Kohr et al. 2003; Krapivinsky et al. 2003; Liu et al. 2004; Massey et al. 2004; White & Youngentob, 2004; Zhang et al. 2007). We have compared the subunit composition of synaptic and extrasynaptic NMDARs in acute slices and find 46.6 ± 7.7% of synaptic NMDAR currents are ifenprodil sensitive (NR2B containing). These results support earlier findings that reported the presence of ∼45% ifenprodil-sensitive synaptic NMDARs (Kirson & Yaari, 1996; Sobczyk et al. 2005; Thomas et al. 2006). We also find that ifenprodil produces an equivalent block of synaptic and extrasynaptic NMDAR currents (38.2 ± 5.8%). This suggests that NR2B receptors are not enriched in the extrasynaptic compartment in 2- to 3-week-old animals, consistent with recent results from cultured hippocampal neurons (Thomas et al. 2006). It is interesting to note that a comparison between synaptic and extrasynaptic NMDARs in cerebellar granule cells found them to have remarkably similar single-channel properties (Clark et al. 1997). Our results suggest dissociation between the signalling properties associated with NMDAR location and subunit composition.

We used NMDAR field potentials to probe the sensitivity of synaptic NMDARs to NR2B-selective antagonists. Consistent with previous reports, we found a developmental shift in synaptic NR2B contents (Monyer et al. 1994; Kirson & Yaari, 1996; Li et al. 1998; Tovar & Westbrook, 1999). Although the total block of field potentials was less than that seen in whole-cell experiments using juvenile animals, this difference is probably due to the superior relief of Mg²⁺ block afforded by the whole-cell voltage clamp configuration. As a result, NMDAR field recordings are not ideal for determination of the absolute magnitude of block by subunit-selective antagonists. Nonetheless, these experiments reveal relative differences in NR2B content at different ages. While there was little block in adult animals, this does not mean there are no synaptic NR2B subunits in adult animals. In fact, previous work testing for ifenprodil sensitivity in adult hippocampal slices under whole-cell voltage clamp conditions found approximately 20% block of synaptic current amplitudes by ifenprodil (Kirson & Yaari, 1996). As a result, it seems likely that a small portion of NR2B subunits remain in the synapse well into adulthood.

Our estimate depends upon the near-complete activation of all synapses within the uncaging area. If a significant fraction of the synapses remain unactivated, and unblocked by MK-801, we would overestimate the size of the extrasynaptic pool. To achieve overlap we activated a long stretch of dendrite with electrical/synaptic stimulation, as shown by the calcium signal generated following electrical stimulation of afferents (∼80 μm; Supplemental Fig. 1). We then positioned the much smaller uncaging spot (6 μm) in the centre of this stimulation area. Further, MK-801 block by a large photolysis spot also blocked synaptic currents evoked by an orthogonal stimulating pipette (Fig. 4). This provides evidence that positioning the uncaging beam orthogonal to the stimulation pipette correctly aligns the two stimulation areas. Our results are in line with a substantial body of elegant work demonstrating that the axon fibres run orthogonal to the apical dendrites (Andersen et al. 1980; Amaral & Witter, 1989).

To show that all functional synapses were activated, we performed experiments using two parallel stimulating pipettes to evoke synaptic NMDAR currents (Fig. 2). We found that MK-801 block of synaptic currents with the first stimulation pipette also blocked synaptic NMDAR currents evoked by the second pipette. Given the proximity of the second pipette to the cell dendrites, we would expect that all synapses would be directly depolarized and release glutamate. The inability to elicit synaptic currents with this pipette suggests that the vast majority of synapses are activated by our stimulation protocol. While release probability is thought to be low at CA1 pyramidal neuron synapses, repeated stimulation allows the cumulative probability to approach 1. However, it is possible that there are non-functional synapses which cannot be blocked. We have controlled for this problem by choosing neurons deep within the slice to avoid cells with damaged dendritic trees. Further, our measurements of extrasynaptic pool size are remarkably consistent with previous estimates (synaptic pool size of 81 ± 4% and 71 ± 3%) obtained in autaptic cultures in which all functional synapses are blocked and the number of non-functional synapses is limited (Rosenmund et al. 1995; Thomas et al. 2006). The similarity between these results also suggests that unblocked/non-functional synaptic NMDARs do not contribute significantly to our determination of extrasynaptic NR2B content.

Experiments to estimate the extrasynaptic pool size were done at room temperature (25°C) to facilitate comparison with previous reports (Tovar & Westbrook, 1999; Thomas et al. 2006). This could result in underestimation of pool size since glutamate transporters operate less efficiently at this temperature (Asztely et al. 1997; Diamond & Jahr, 2000). However, this source of potential error in our estimation of pool size does not impact on our conclusions that extrasynaptic NMDARs form a large pool of receptors, with a similar molecular identity to synaptic receptors.

Our definition of extrasynaptic receptors does not distinguish between synaptic NMDARs that frequently respond to glutamate and those that do so infrequently. NMDARs located at low release probability synapses seldom open, while those at high release probability synapses commonly participate in synaptic communication. Such a situation could also occur for perisynaptic receptors which infrequently encounter glutamate. However, immunogold staining of NMDARs suggests the majority of NMDARs are located at the centre of the postsynaptic density, with fewer than 5% of receptors located perisynaptically (Racca et al. 2000; Valtschanoff & Weinberg, 2001). A ‘back of the envelope’ calculation from our data suggest that perisynaptic receptors are unlikely to contribute greatly to our synaptic fraction. We stimulated ∼100 times to achieve MK-801 block of synaptic receptors. Given an average release probability of 0.2 (Hessler et al. 1993; Rosenmund et al. 1993) each synapse has only 20 opportunities to activate these perisynaptic NMDARs. Given that NMDARs have an open probability of ∼1% (Jahr, 1992; Rosenmund et al. 1995; Dzubay & Jahr, 1996) and a probability of MK-801 block less than 1 (Huettner & Bean, 1988), it is clear that a substantial portion of these receptors will remain unblocked by our MK-801 protocol. The critical distinction between synaptic and extrasynaptic NMDARs is not the frequency with which they activate, but rather that synaptic receptors respond to single synaptic events while extrasynaptic NMDARs do not respond within the time frame of our experiments. However, it is possible that this substantial and stable pool of extrasynaptic NMDARs exists to shape synaptic transmission when activity increases beyond single synaptic events.

It has been suggested that activating extrasynaptic NMDARs has profound consequences for the cell, either leading to apoptosis (Hardingham et al. 2002; Ivanov et al. 2006; von Engelhardt et al. 2007) or long-term depression of synaptic transmission (Liu et al. 2004). It remains controversial, however, whether these effects are due to the subunit composition or location of extrasynaptic receptors (Rusakov et al. 2004; Kim et al. 2005; Kohr, 2006; Liu et al. 2007). One implication of our findings is that when studying these processes, NR2B blockade is not an appropriate experimental tool for selectively blocking extrasynaptic NMDARs. Since synaptic and extrasynaptic NMDARs have similar subunit composition, our results suggest that any effects of extrasynaptic NMDARs are probably due to the coupling of these receptors to highly local signalling pathways.

Supplementary material

Online supplemental material for this paper can be accessed at:

http://jp.physoc.org/cgi/content/full/jphysiol.2007.137679/DC1 and http://www.blackwell-synergy.com/doi/suppl/10.1113/jphysiol.2007.137679