Abstract

Cortical regions of the brain stand out for their high content in synaptic zinc, which may thus be involved in synaptic function. The relative number, chemical nature and transmitter receptor profile of synapses that sequester vesicular zinc are largely unknown. To address this, we combined pre-embedding zinc histochemistry and post-embedding immunogold electron microscopy in rat hippocampus. All giant mossy fibre (MF) terminals in the CA3 region and ~45% of boutons making axospinous synapses in stratum radiatum in CA1 contained synaptic vesicles that stained for zinc. Both types of zinc-positive boutons selectively expressed the vesicular zinc transporter ZnT-3. Zinc-positive boutons further immunoreacted to the vesicular glutamate transporter VGLUT-1, but not to the transmitter γ-aminobutyric acid. Most dendritic spines in CA1 immunoreacted to α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptor (AMPAR) subunits GluR1–3 (~80%) and to N-methyl-d-aspartate receptor (NMDAR) subunits NR1 + NR2A/B (~90%). Synapses made by zinc-positive boutons contained 40% less AMPAR particles than those made by zinc-negative boutons, whereas NMDAR counts were similar. Further analysis indicated that this was due to the reduced synaptic expression of both GluR1 and GluR2 subunits. Hence, the levels of postsynaptic AMPARs may vary according to the presence of vesicular zinc in excitatory afferents to CA1. Zinc-positive and zinc-negative synapses may represent two glutamatergic subpopulations with distinct synaptic signalling.

Introduction

The mechanisms underlying functional diversity among excitatory synapses are poorly understood. In the hippocampus, presynaptic terminals from a single pyramidal cell display specific properties according to the target cell type (Shigemoto et al., 1996; Scanziani et al., 1998). Moreover, synapses made by CA3 Schaffer collaterals onto CA1 pyramidal neurons show a large variability in the initial probability of transmitter release (Dobrunz and Stevens, 1997) and synaptic plasticity (Debanne et al., 1999). The molecular and structural differentiation of synapses are likely to explain this functional diversity. For instance, the density of postsynaptic AMPA receptors (AMPARs) varies greatly even between synapses of similar size and location in the CA1 region (Nusser et al., 1998; Takumi et al., 1999), which may contribute to differences in quantal amplitude (Golshani et al., 2001). This AMPAR variability may either reflect unrecognized differences in neurochemistry or origin of the corresponding presynaptic boutons, different histories of synaptic activity (Luscher et al., 2000), or both.

We tested the former hypothesis by studying the distribution of vesicular zinc among hippocampal boutons (ZN+ boutons), which form an undetermined proportion of asymmetric synapses (ZN+ synapses) (Haug, 1967; Pérez-Clausell and Danscher, 1985). Depolarization of mossy fibre (MF) axons locally increases extracellular zinc levels in hippocampal slices (Assaf and Chung, 1984; Howell et al., 1984; Li et al., 2001a), suggesting synaptic release of vesicular zinc. In addition, zinc chelators disinhibit synaptic NMDARs (Vogt et al., 2000; Ueno et al., 2002) and block the induction of MF long-term potentiation (Li et al., 2001b), indicating that zinc modulates synaptic plasticity. Likewise, zinc has been implicated in the neurodegeneration associated with epilepsy (Cole et al., 2000) and ischaemia (Koh et al., 1996), but its net effect is as yet unclear.

To test whether vesicular zinc is located in a subpopulation of glutamatergic boutons, the proportion of ZN+ boutons was assessed in serial ultrathin sections from CA3 and CA1 regions. To identify the molecular signatures of ZN+ boutons, pre-embedding zinc histochemistry and post-embedding immuno-gold electron microscopy were combined.

Materials and Methods

Perfusion and Staining for Zinc

All procedures followed present regulations for animal care and handling and were approved by the University of Barcelona Ethical Committee. Male Wistar rats (250–400 g) were used. Two rats were perfused under anaesthesia (chloral hydrate, 600 mg/kg), first with heparinized saline, then with 200 ml of 0.1% Na2S, 2.5% glutaraldehyde, 0.002% CaCl2 in 0.1 M phosphate buffer (PB; pH = 7.4) and finally with 400 ml of 2.5% glutaraldehyde, 1% formaldehyde, 0.2% picric acid and 0.002% CaCl2 in PB, at a rate of 50 ml/min. Na2S renders zinc precipitates that become visible by silver amplification. The brains were taken out, postfixed, frontally cut on a Vibratome at 50 μm, rinsed in PB and processed free-floating for the visualization of zinc (Danscher, 1981). Briefly, sections were developed in a solution containing gum arabic, citrate buffer, hydroquinone and silver lactate for 75 min in the dark at 26°C.

Osmium-free Postfixation and Epon Embedding

Following staining for zinc, slices were processed as described previously (Phend et al., 1995). Briefly, they were rinsed in PB and immersed in 0.1 M maleate buffer solutions (MB; pH = 6) containing 1% tannic acid, 1% uranyl acetate and 0.5% iridium tetrabromide, then in 30, 50 and 70% ethanol, 1% p-phenylenediamine in 70% ethanol, 1% uranyl acetate in 70% ethanol and further dehydrated in ethanol. Sections were immersed in propylene oxide and flat-embedded in epon resin (Embed-812; EMS).

Some slices were parallel processed for conventional electron microscopy. Briefly, they were postfixed in 2% OsO4 in 0.12 M PB, counterstained in 1% uranyl acetate in 70% ethanol, dehydrated and embedded in epon as described above.

Freeze-substitution and Acrylic Embedding

Four rats were perfused with saline followed by 500 ml of 4% formaldehyde, 0.1–0.2% glutaraldehyde, 0.2% picric acid and 0.002% CaCl2 in PB. The brains were postfixed and cut, as above. Slices were rinsed in 0.1 M Tris–maleate buffer (TM; pH = 7.4) and cryoprotected in glycerol (10, 20, 30% in TM). The rostral hippocampus was cut out, frozen in liquid propane, freeze-substituted in methanol containing 0.5% uranyl acetate and embedded in Lowicryl-HM20 (van Lookeren et al., 1991). Another two rats were processed as above, except that fixation was preceded by the addition of 0.1% Na2S and 0.2% glutaraldehyde in PB. Brain slices were silver-amplified as above.

Antibodies

Rabbit GluR1 (3 mg/ml), GluR2/3 (3 mg/ml), NR2A/B (4 mg/ml) and mouse GluR2 (4.5 mg/ml) antibodies (Chemicon, Temecula, CA), rabbit GABA antibody (0.05 mg/ml; Sigma, St Louis, MO), rabbit NR2A and NR2B antibodies (2.5 mg/ml; Sigma-RBI), mouse NR1-pan antibody (5 mg/ml; Pharmingen, San Diego, CA) and rabbit VGLUT-1 antibody at a dilution of 1:1000 (Varoqui et al., 2002) were used. The specificity of NR2A and NR2B antibodies was confirmed by immunoblotting of rat cortical synaptosomal fractions (data not shown). Rabbit ZnT-3 antibody (1:100) (Palmiter et al., 1996) was a kind gift from Dr T.B. Cole (University of Washington) to Dr G. Danscher (Aarhus University).

Post-embedding Immunolabelling

Ultrathin sections were picked up on formvar-coated gold grids, washed in Tris-buffered saline (50 mM Tris–HCl, 0.3% NaCl; TBS; pH = 7.6) and incubated in 1% sodium borohydride/TBS (10 min, room temperature), then in 0.005% Tergitol NP-10 in TBS (TBST), followed by 10% normal goat or horse serum, 0.5% bovine serum albumin (BSA), 50 mM glycine and 0.1% gelatin in TBST (30 min, room temperature), followed by primary antibodies in TBST containing 2% serum (overnight at 36°C). They were rinsed in TBST (4 × 5 min) and TBST/pH 8.2 (2 × 10 min), then incubated in secondary antibodies, goat anti-rabbit or goat anti-mouse IgG or F(ab)2 fragments coupled to 10 or 15 nm colloidal gold particles (British BioCell International), diluted 1:30 in TBST/pH 8.2 with 0.5% BSA (2 h, 36°C), rinsed, immersed for 5 min in TBS/2% glutaraldehyde, jet-washed and counterstained.

Image Acquisition and Data Analysis

Electron micrographs were recorded by a CCD camera (Gatan Inc.). We checked the penetration of the silver amplification solution into the slice. Thin sections were cut transversal to the Vibratome slice and adjacent micrographs were taken from each section at ~100 μm from the CA1 pyramidal layer. The penetration of the developer began to decrease at 4 μm under the slice surface, so that the frequency of ZN+ boutons shown decreased gradually (P < 0.05, n = 10 experiments, four slices). Hence, analysis in subsequent experiments was restricted to this 4 μm band. ZN+ boutons in CA1 were quantified by random systematic sampling along the axis of apical dendrites. Perforated spines and dendritic shafts were excluded. In the stratum lucidum, only giant MF terminals were analysed to avoid contamination from spiny nonpyramidal neurons (Soriano and Frotscher, 1993). Based on previous findings (Matsubara et al., 1996; Fabian-Fine et al., 2000; He et al., 2000), synapses immunopositive for glutamate receptors were defined as those presenting one or more gold particles located as far as 45 nm from the inner leaflet of the postsynaptic membrane associated with the PSD. To compare the efficiency of immunolabelling in each protocol, we calculated the average number of gold particles that labelled the CA1 stratum radiatum after incubation with GluR1 + GluR2/3 antibodies. In conventional epon material, efficiency of immunolabelling amounted to 7 ± 2 gold particles/μm2 (area examined = 87 mm2). When combined with staining for zinc, the efficiency remained similar: 10 ± 4 particles/mm2 (area = 148 mm2; P = 0.3) and was highly specific (only 0.3 ± 0.1 particles/μm2 in the absence of primary antibody; area = 39 μm2; P < 0.001). Similar values were found in Lowicryl-embedded CA1 (10 ± 1 AMPAR particles/μm2; area = 35 mm2), Lowicryl-embedded CA1 stained for zinc (9 ± 2 particles/mm2; area = 58.2 mm2) and in a previous work (Racca et al., 2000).

Differences between means were assessed with the Mann–Whitney U-test. Significance was set at 0.05. Values are given as means ± SD unless stated otherwise. Relative frequencies were obtained from samples of similar but not identical size, hence the arcsin of the square root of each value was calculated to homogenize variances and data were compared with the Wilcoxon signed ranks test. Observed frequencies were fit to a Poisson distribution by the least-squares method.

Results

ZN+ Boutons in the Hippocampus Form a Subpopulation of Synapses

Staining for vesicular zinc was intense in the neuropile of CA3 and CA1 regions, but absent in cell body layers, the stratum lacunosum-moleculare and the medial perforant-path of the dentate gyrus (Fig. 1A). At the ultrastructural level, zinc ions (detected by silver grains) were associated with round, clear synaptic vesicles in boutons that formed asymmetric synapses with dendritic spines. In the stratum lucidum in CA3, all large MF boutons were ZN+. In the stratum radiatum in CA1, 38 ± 1% (n = 174) of axospinous boutons were identified as ZN+ in single sections. This proportion remained constant between 100 and 300 μm from the pyramidal cell layer. The mean postsynaptic density (PSD) length (Table 1) or gross bouton morphology did not differ between ZN+ and zinc-negative (ZN−) synapses.

To test whether the lack of staining for zinc in the remaining boutons in CA1 was due to technical limitations, boutons were sampled through five serial sections to increase the efficiency of detection. In these conditions, 46 ± 2% of boutons reacted for vesicular zinc (n = 77). In addition, 87 ± 2% of the profiles from identified ZN+ boutons were labelled for zinc (n = 52 ZN+ boutons, data from a different sample; three to five serial sections). Thus, ZN+ boutons were underestimated in only ~6% of cases in single sections, a proportion far from that of ZN− boutons (~55%).

To confirm the histochemical results, we performed immuno-gold staining for the obligatory vesicular zinc transporter ZnT-3 on thin sections. As shown in Figure 1C1,C2, ZnT-3 particles were consistently found in ZN+ boutons, whereas virtually all ZN− boutons were devoid of immunostaining. Most ZnT-3 particles overlaid synaptic vesicles (Fig. 1F). Results are summarized in Figure 1I. As a positive control, heavy ZnT-3 staining was found at MF to pyramidal CA3 synapses (Fig. 1B). Altogether, data strongly pointed to two separate populations of afferent boutons (ZN+/ZnT-3+ versus ZN−/ZnT-3−) making synapses onto CA1 spines.

Expression of VGLUT-1 at ZN+ Boutons

To provide evidence relative to the transmitter content of ZN+ boutons, we studied the presence of the vesicular glutamate transporter isoform VGLUT-1 (which is mostly telencephalic) and the inhibitory transmitter γ-aminobutyric acid (GABA). Both ZN+ and ZN− boutons showed strong immunoreactivity for VGLUT-1 (Figs. 1D,E,I), the majority of which concentrated over synaptic vesicles and, to a lesser extent, on junctional membranes (Fig. 1G). VGLUT-1 particles were frequently associated with the luminal face of the vesicular membrane and seemed to distribute evenly over the vesicle pool. Notably, some vesicular VGLUT-1 particles were apposed to medium-sized (30–50 nm) silver grains in ZN+ boutons (Fig. 1H).

Immunoreactivity for GABA left all axospinous boutons unstained, but was abundant in symmetric synapses on shafts (Fig. 1J), in a few postsynaptic dendrites and in mitochondria (Fig. 1K).

Co-localization of Postsynaptic Ionotropic Glutamate Receptors with ZN+ Terminals

The identification of two axospinous populations of glutamatergic boutons in CA1 prompted us to analyse the postsynaptic expression of glutamate receptors relative to ZN+ and ZN− boutons. The efficiency and specificity of immunolabelling was unaffected by zinc histochemistry (see Material and Methods). Furthermore, the mean number of synaptic AMPAR (GluR1 + GluR2/3 antibodies) and NMDAR (NR1 + NR2A/B or NR1 + NR2A + NR2B antibodies, pooled data) particles coincided with that found in conventional immunocytochemical studies (Table 1) (Rubio and Wenthold, 1997; Kharazia and Weinberg, 1999; Petralia et al., 1999). Closer examination revealed the typical pattern of AMPAR labelling: 69% of particles (from a sample of 490 particles) associated with the postsynaptic membrane or density; 11.2% on presynaptic terminals, most of which were on synaptic vesicles; and 19.8% localized to the cytoplasm or extrasynaptic membrane of the spine. Except for the synaptic cleft, extracellular compartments were virtually unlabelled. Several synapses were immunopositive in single sections stained for zinc: 64 ± 6% (n = 273) and 66 ± 4% (n = 250) of synapses were labelled after incubation with NMDAR or AMPAR antibody mixtures, respectively. When analysed through two adjacent sections, the percentage of labelled synapses increased to 92 ± 2% for NMDARs (n = 108) and to 79 ± 3% for AMPARs (n = 103; different from NMDAR, P < 0.05).

NMDAR (Fig. 2AD) and AMPAR (Fig. 2FI) particles were found along the PSD of spines contacted by either ZN+ or ZN− boutons. Receptor particles and zinc precipitates were sometimes apposed to one another at individual ZN+ synapses (Fig. 2D,G). Similar proportions of ZN+ and ZN− synapses were immunolabelled for NMDAR (P = 0.13) and AMPAR subunits (P = 0.8), respectively (Table 1).

In CA3, all large MF boutons were profusely labelled for vesicular zinc and commonly enveloped immunostained thorny excrescences. As in CA1, both NMDAR (Fig. 2E) and AMPAR (Fig. 2L) particles were localized over the postsynaptic membrane and density of spines.

Further supporting the specificity of double staining, results were reproduced in lightly fixed, Lowicryl-embedded CA1 region: GluR2/3 protein and zinc ions showed post- and pre-synaptic distributions, respectively, and frequently co-localized to single synapses, although tissue preservation was largely compromised (Fig. 2J,K).

Lower Amounts of AMPAR, but not NMDAR, Immunoparticles at ZN+ Synapses

Synaptic immunostaining was quantified at 200–300 μm from the pyramidal cell layer in CA1. The mean number of NMDAR particles was similar at ZN+ and ZN− synapses (Table 1; particles per labelled synapse, P = 0.9), as were the frequency distributions of NMDAR particles found in the two populations (Fig. 3A). In contrast, AMPAR particles were significantly less abundant (~40%) at ZN+ synapses than at ZN− synapses (particles per labelled synapse, P < 0.01; Table 1). The frequency distribution of AMPAR labelling was much more skewed towards lower values at ZN+ synapses (Fig. 3B; skewness factor for ZN+ = 1.6; skewness factor for ZN− = 0.8; Shapiro–Wilk test). The distinct content in AMPAR particles was confirmed in an additional sample of synapses reacted through two serial sections: the mean number ± SEM of particles per synapse was 3 ± 0.3 for ZN− (n = 34) and 1.8 ± 0.1 for ZN+ synapses (n = 38; P < 0.05). As synapses were compared at the same distance from the pyramidal cell layer, data were not affected by the distance-dependent increase in dendritic AMPARs (Andrasfalvy and Magee, 2001). In CA3, the linear density of NMDAR particles at MF synapses was markedly lower than that of AMPARs (P < 0.05) and was also lower than that of NMDARs in CA1 (P < 0.05).

We measured the variability (coefficient of variation, CV) of AMPAR counts in single sections from CA1. CV was 0.99 for ZN− synapses (n = 129) and 1.15 for ZN+ synapses (n = 121). To determine the source of this variability, we compared the AMPAR frequency distribution with the corresponding theoretical Poisson distribution. A good fit was observed for both ZN+ and ZN− synapses (Fig. 3C; P = 0.99 for ZN+; P = 0.98 for ZN−; χ2 goodness-of-fit test). Therefore, much of the variability in particle counts arose from the low probability of the immunoreaction. Indeed, large and small synapses of the same population had similar CVs, despite differences in the mean number of AMPAR particles (data not shown).

Both GluR1 and GluR2 Subunits are Selectively Down-regulated at ZN+ Synapses

The distinct number of AMPAR particles in ZN+ and ZN− synapses may reflect either reduced overall AMPAR levels or the specific down-regulation of one subunit at ZN+ synapses. To distinguish between the two possibilities, the synaptic expression of GluR1 and GluR2 subunits was studied separately in single sections. Using a GluR1 subunit-specific antibody, 66 ± 12% of ZN− synapses (n = 164) were labelled in CA1 versus 39 ± 15% of ZN+ synapses (n = 153; P < 0.05; Fig. 4AC). Similarly, 54 ± 8% of ZN− synapses (n = 94) reacted for GluR2, but only 37 ± 8% of ZN+ synapses did (n = 129; P < 0.05; Fig. 4DF). Immunoreactive ZN+ synapses tended to have fewer particles for both GluR1 (1.4 ± 0.2 particles at ZN+ synapses and 1.7 ± 0.4 at ZN−; P = 0.4) and GluR2 (1.4 ± 0.1 particles at ZN+ synapses and 1.6 ± 0.3 at ZN−; P = 0.3), but differences were not statistically significant.

Discussion

The main finding of the present experiments is that, in the stratum radiatum in CA1, two populations of glutamatergic boutons making morphologically similar synapses can be differentiated based on their vesicular zinc content and ZnT-3 expression. Furthermore, the levels of postsynaptic AMPARs, but not of NMDARs, correlate with the presynaptic bouton type. Thus ZN+ synapses showed a lower AMPAR content than ZN− synapses.

ZN+ Boutons in the Hippocampus: Abundance and Neurochemistry

The presence of histochemically chelatable zinc, corresponding to vesicular zinc, in the CA1 region is well established (Slomianka, 1992; Sanchez-Andres et al., 1997). However, the incidence of vesicular zinc among defined boutons has been unexplored. We identified vesicular zinc in ~45% of the boutons making axospinous synapses in the intact CA1 stratum radiatum. Despite the low number of silver grains per bouton profile, due to the small bouton size in this region, staining for zinc (or the lack of it) was consistent in serial sections. Furthermore, zinc stain was accompanied by ZnT-3 immunoreactivity of synaptic vesicles, in contrast to ZN− boutons. This provided independent evidence that zinc histochemistry accurately identifies vesicular zinc, even in small cortical boutons. Complementing these data, all large MF boutons in CA3 stained for zinc and ZnT-3. These findings strongly advocate a diversity of zinc content among synapses based on the differential expression of ZnT-3. Whether this presynaptic diversity reflects the heterogeneous origin of the parent cell bodies or an as yet overlooked diversity among CA3 pyramidal cells is currently under investigation. Functional experiments in vitro have failed to detect zinc release (Howell et al., 1984) or a post-synaptic effect of endogenous zinc (Lu et al., 2000) in CA1. Since vesicular zinc is widespread in this region, the discrepancy may be explained by (i) the substantial depletion of endogenous zinc observed in in vitro preparations ( Suh et al., 2000), (ii) the presence of contaminant zinc, which hinders the interpretation of negative results (Paoletti et al., 1997) and (iii) the distribution of vesicular zinc to a subset of boutons, as shown here. Hence, our data encourage the reassessment of the synaptic actions of zinc in the CA1 region.

All ZN+ boutons stained for VGLUT-1 and lacked GABA, confirming their excitatory nature (Martínez-Guijarro et al., 1991; Beaulieu et al., 1992). The hypothesis that vesicular zinc is synaptically released together with glutamate to modulate postsynaptic responses is based on the assumption that both are contained in the same synaptic vesicle, at least in synapses where exocytosis is uniquantal. We observed several silver grains overlying or contained within synaptic vesicles that were juxtaposed to VGLUT-1 immunoparticles. It is possible, then, that ionic zinc and glutamate co-exist in a number of vesicles. About 25% of VGLUT-1 labelling was also found on pre- and postsynaptic junctional membranes. This may correspond to VGLUT-1 from docked presynaptic vesicles, but also to some VGLUT-1 expressed on the plasma membrane, as suggested elsewhere (Bellocchio et al., 1998).

Expression of AMPARs at ZN+ Synapses

The majority of dendritic spines in CA1 and CA3 reacted for AMPAR and NMDAR subunits after being processed for the visualization of zinc. The percentages of immunolabelled synapses are similar to those described in single sections of epon-embedded sensory cortex (Kharazia and Weinberg, 1999), although equivalent studies in the hippocampus are not available. Percentages increased when synapses were immuno-reacted in paired sections to levels consistent with those found in Lowicryl (Nusser et al., 1998; Takumi et al., 1999; Racca et al., 2000). AMPAR label was not detected in ~20% of spines, which might correspond to functionally ‘silent’ synapses at resting membrane potentials or to synapses with AMPARs expressed at levels below the detection limit of the immunogold method (Groc et al., 2002). The subsynaptic distribution of AMPARs matched that reported in synapses of CA3 radiatum (Fabian-Fine et al., 2000). Likewise, in MF synapses the density of AMPARs was higher than that of NMDARs, coincident with a previous study (Takumi et al., 1999) and in agreement with electrophysiological data (Weisskopf and Nicoll, 1995). AMPAR counts, however, were not higher than in CA1, in contrast with other work (Nusser et al., 1998; Takumi et al., 1999). Differences in the antibody mixtures used here (GluR1–3 antibodies) and elsewhere (pan-AMPAR + GluR2/3 and GluR1–4 antibodies) may account for this discrepancy and possibly reflect a distinct AMPAR subunit enrichment at MF synapses. Altogether, data confirm the good performance of the double staining method used here.

Gold particles located over the postsynaptic junctional membrane most likely identified subunits that were assembled into synaptic receptors. This conclusion is based on the correlation between the number of receptor particles and the amplitude of miniature synaptic currents reported elsewhere (Nusser et al., 1997; Golshani et al., 2001). Thus specific immunolabelling for glutamate receptor subunits provided strong anatomical–molecular evidence supporting the expression of functional glutamate receptors at ZN+ synapses.

The population of ZN+ synapses presented, on average, significantly lower AMPAR staining than the ZN− population. In principle, reduced GluR1–3 labelling can be caused by either a lower receptor number or a different subunit composition of receptors. Immunoprecipitation studies indicate that these subunits form most of the receptor complexes expressed by CA1 pyramidal neurons, with little presence of GluR4 (Wenthold et al., 1996). We found that GluR1 and GluR2 subunits were similarly widespread at ZN− synapses in CA1, much like combined GluR1–3. However, both GluR1 and GluR2 were detected in substantially fewer ZN+ synapses, again in very similar proportions. Hence, these comparisons reveal that ZN+ synapses most likely express lower levels of AMPAR than ZN− synapses, rather than being enriched in different subunits. Differential AMPAR expression according to these afferent subtypes may explain, at least in part, the variability in AMPAR content that is not due to differences in synaptic size (Nusser et al., 1998). On top of this, synaptic activity may modulate receptor levels within a given range (Heynen et al., 2000). It remains to be determined whether ZN+ and ZN− boutons have different exo/endocytotic properties, which are key to all forms of synaptic plasticity that require high-frequency presynaptic use.

Expression of NMDARs at ZN+ Synapses

Exogenously added zinc inhibits the NMDAR component of synaptic responses (Westbrook and Mayer, 1987). Double labelling showed that NMDARs were found on the majority of ZN+ synapses in CA1 and CA3 regions, suggesting that the synaptic co-release of zinc may mediate a similar inhibition in vivo. There is strong evidence for such an interplay at MF synapses, where endogenous zinc tonically inhibits postsynaptic NMDARs (Vogt et al., 2000), but not in CA1 (Lu et al., 2000). Conversely to functional results, the linear densities of NMDAR particles were clearly higher at CA1 synapses than at MF synapses (Table 1). Other factors, such as subunit-specific receptor affinity (Chen et al., 1997; Paoletti et al., 1997), carboxyl-tail interactions (Yamada et al., 2002), or peak concentration of released zinc may thus contribute to the inhibition of NMDARs by zinc.

In conclusion, due to the presence of vesicular zinc in about half of spine synapses and the associated selectivity of glutamate receptor content, synaptic transmission and plasticity in CA1 neurons may be considerably affected by the particular excitatory input.

Notes

We thank the staff of the Cryomethods and Electron Microscopy Units at the Serveis Cientificotècnics of the Universitat de Barcelona. We are indebted to Dr Gorm Danscher (University of Aarhus) for his support and critical comments on the manuscript. C.B.S. is the recipient of a pre-doctoral grant from the Ministry of Science and Technology of Spain. This study was supported by the Ministries of Education (PM97-0048-C02-02) and Health (FIS 01/0096-02) of Spain and the Fundació Marató de TV3 (97/1022).

Address correspondence to Carlos Balet Sindreu, Departament de Biologia Cellular, Universitat de Barcelona, Av. Diagonal 645, ES-08071 Barcelona, Spain. Email: carbalet@bio.ub.es.

Table 1

Labelling for AMPAR and NMDAR subunits in single sections of ZN+ and ZN− synapses in strata radiatum (CA1) and lucidum (CA3)

 ZN− (CA1) ZN+ (CA1) ZN+ (CA3) 
aDifference between ZN− and ZN+ synapses in CA1. 
bDifference between CA3 and CA1 regions. 
cDifference between AMPAR and NMDAR label. 
GluR1–3 
 Synapses examined 129 121 118 
 Immunolabelled synapses (%) 67 ± 10 60 ± 11 67 ± 8 
 Particles/labelled synapse 2.6 ± 0.3a 1.7 ± 0.3a 2.1 ± 0.2 
 Postsynaptic density length (nm) 215 ± 15b 216 ± 22b 309 ± 32b 
 Linear density (particles/μm PSD) 12.3 ± 0.7a,b 8.9 ± 1.3a 7.1 ± 1.3b,c 
NR1+NR2A/B 
 Synapses examined 126 147 110 
 Immunolabelled synapses (%) 70 ± 6 60 ± 14 56 ± 7 
 Particles/labelled synapse 2.11 ± 0.16b 2.0 ± 0.4b 1.6 ± 0.2b 
 Postsynaptic density length (nm) 212 ± 15b 235 ± 12b 327 ± 44b 
 Linear density (particles/μm PSD) 10.4 ± 1.1b 9 ± 2b 4.91 ± 0.18b,c 
 ZN− (CA1) ZN+ (CA1) ZN+ (CA3) 
aDifference between ZN− and ZN+ synapses in CA1. 
bDifference between CA3 and CA1 regions. 
cDifference between AMPAR and NMDAR label. 
GluR1–3 
 Synapses examined 129 121 118 
 Immunolabelled synapses (%) 67 ± 10 60 ± 11 67 ± 8 
 Particles/labelled synapse 2.6 ± 0.3a 1.7 ± 0.3a 2.1 ± 0.2 
 Postsynaptic density length (nm) 215 ± 15b 216 ± 22b 309 ± 32b 
 Linear density (particles/μm PSD) 12.3 ± 0.7a,b 8.9 ± 1.3a 7.1 ± 1.3b,c 
NR1+NR2A/B 
 Synapses examined 126 147 110 
 Immunolabelled synapses (%) 70 ± 6 60 ± 14 56 ± 7 
 Particles/labelled synapse 2.11 ± 0.16b 2.0 ± 0.4b 1.6 ± 0.2b 
 Postsynaptic density length (nm) 212 ± 15b 235 ± 12b 327 ± 44b 
 Linear density (particles/μm PSD) 10.4 ± 1.1b 9 ± 2b 4.91 ± 0.18b,c 
Figure 1.

ZN+ presynaptic boutons constitute a glutamatergic subpopulation in the hippocampus. (A) Bright-field micrograph of a silver-amplified slice showing histochemically reactive zinc in adult rat hippocampus. Staining for zinc is found in strata radiatum (s.r.), oriens (s.o.), lucidum (s.l.) and hilus (h), but absent in strata pyramidale (s.p.), lacunosum-moleculare (s.l.-m) and perforant-path field of dentate gyrus (p.p.). Rectangles illustrate regions analysed for electron microscopy. (B–C) Co-localization of vesicular zinc and ZnT-3 in large MF terminals in CA3 (B) and boutons (b+) in CA1, cut in paired sections (C1,C2) from osmium-free, epon-embedded hippocampus. Arrowheads indicate ZnT-3 immunogold particles (10 nm). (D,E) Co-localization of vesicular zinc and VGLUT-1 in CA1 synapses. VGLUT-1 immunoparticles (arrowheads) labelled both ZN+ (D) and ZN− (E) terminals (b+, b−, respectively). (F) ZnT-3 immunoparticles were mostly distributed over synaptic vesicles (89 ± 3%; n = 133 particles sampled) of presynaptic boutons. (G) The majority of VGLUT-1 was found on synaptic vesicles, but a significant fraction was also associated with the plasma membrane (n = 895 particles sampled). (H) Several examples in which zinc and VGLUT-1 labels were seen in close apposition to one another, with underlying synaptic vesicles. Scale bars = 50 nm. (I) Differential expression of ZnT-3 and VGLUT-1 at ZN+ and ZN− synapses in CA1. VGLUT-1 stained all ZN+ (98 ± 3%) and ZN− boutons (96 ± 4%; n = 242). In contrast, ZnT-3 was abundant in ZN+ boutons (92 ± 6%) but scarce among ZN− boutons (7 ± 2%; n = 55; P < 0.01), the latter likely corresponding to misidentified ZN+ boutons. (J,K) Immunoreactivity for GABA (arrowheads) was absent from synaptic vesicles of ZN+ boutons in CA1 (J) and MFs in CA3 (K), but present in boutons (b−) making symmetric synapses on shafts (d) and in mitochondria (m). s, spine; t, thorny excrescence. Scale bars = 0.2 μm.

Figure 1.

ZN+ presynaptic boutons constitute a glutamatergic subpopulation in the hippocampus. (A) Bright-field micrograph of a silver-amplified slice showing histochemically reactive zinc in adult rat hippocampus. Staining for zinc is found in strata radiatum (s.r.), oriens (s.o.), lucidum (s.l.) and hilus (h), but absent in strata pyramidale (s.p.), lacunosum-moleculare (s.l.-m) and perforant-path field of dentate gyrus (p.p.). Rectangles illustrate regions analysed for electron microscopy. (B–C) Co-localization of vesicular zinc and ZnT-3 in large MF terminals in CA3 (B) and boutons (b+) in CA1, cut in paired sections (C1,C2) from osmium-free, epon-embedded hippocampus. Arrowheads indicate ZnT-3 immunogold particles (10 nm). (D,E) Co-localization of vesicular zinc and VGLUT-1 in CA1 synapses. VGLUT-1 immunoparticles (arrowheads) labelled both ZN+ (D) and ZN− (E) terminals (b+, b−, respectively). (F) ZnT-3 immunoparticles were mostly distributed over synaptic vesicles (89 ± 3%; n = 133 particles sampled) of presynaptic boutons. (G) The majority of VGLUT-1 was found on synaptic vesicles, but a significant fraction was also associated with the plasma membrane (n = 895 particles sampled). (H) Several examples in which zinc and VGLUT-1 labels were seen in close apposition to one another, with underlying synaptic vesicles. Scale bars = 50 nm. (I) Differential expression of ZnT-3 and VGLUT-1 at ZN+ and ZN− synapses in CA1. VGLUT-1 stained all ZN+ (98 ± 3%) and ZN− boutons (96 ± 4%; n = 242). In contrast, ZnT-3 was abundant in ZN+ boutons (92 ± 6%) but scarce among ZN− boutons (7 ± 2%; n = 55; P < 0.01), the latter likely corresponding to misidentified ZN+ boutons. (J,K) Immunoreactivity for GABA (arrowheads) was absent from synaptic vesicles of ZN+ boutons in CA1 (J) and MFs in CA3 (K), but present in boutons (b−) making symmetric synapses on shafts (d) and in mitochondria (m). s, spine; t, thorny excrescence. Scale bars = 0.2 μm.

Figure 2.

Expression of AMPA- and NMDA-receptor subunits at ZN+ synapses in CA1 and CA3 hippocampal regions. Gold particles (10 nm) coding for NR1 + NR2A/B (AE) and GluR1–3 (FL) subunits localized to asymmetric synapses made by ZN+ (b+) or ZN− (b−) boutons in CA1 stratum radiatum (AD,FK) and CA3 stratum lucidum (E,L). Particles located as far as 45 nm away from the inner face of the postsynaptic membrane (2G, dashed lines) were considered ‘synaptic’ (black arrowheads) and beyond, ‘extrasynaptic’ (white arrowheads). (AD) Co-localization of postsynaptic NMDAR label and vesicular zinc at single spine (s) synapses. (E) A large MF terminal heavily stained for vesicular zinc and contacting two thorny excrescences (t), of which only one is NMDAR-immunoreactive. (FI) Co-localization of postsynaptic AMPAR label and vesicular zinc at single synapses. Note in (G) a silver grain (solid arrow) by the presynaptic membrane and opposite two AMPAR particles. Scale bar = 50 nm. m, pre- and postsynaptic membranes; b, bouton; c, synaptic cleft; d, postsynaptic density. (J,K) Synapses in CA1 that were cryoembedded in Lowicryl and immunoreacted for GluR2/3: (J) samples not developed for zinc; (K) double staining for zinc and GluR2/3. Note the poor structural preservation and difficulty of identifying synaptic vesicles in both cases. (L) A giant MF terminal is stained for vesicular zinc and contacts several AMPAR-immunoreactive thorny excrescences. Scale bars = 0.2 μm.

Figure 2.

Expression of AMPA- and NMDA-receptor subunits at ZN+ synapses in CA1 and CA3 hippocampal regions. Gold particles (10 nm) coding for NR1 + NR2A/B (AE) and GluR1–3 (FL) subunits localized to asymmetric synapses made by ZN+ (b+) or ZN− (b−) boutons in CA1 stratum radiatum (AD,FK) and CA3 stratum lucidum (E,L). Particles located as far as 45 nm away from the inner face of the postsynaptic membrane (2G, dashed lines) were considered ‘synaptic’ (black arrowheads) and beyond, ‘extrasynaptic’ (white arrowheads). (AD) Co-localization of postsynaptic NMDAR label and vesicular zinc at single spine (s) synapses. (E) A large MF terminal heavily stained for vesicular zinc and contacting two thorny excrescences (t), of which only one is NMDAR-immunoreactive. (FI) Co-localization of postsynaptic AMPAR label and vesicular zinc at single synapses. Note in (G) a silver grain (solid arrow) by the presynaptic membrane and opposite two AMPAR particles. Scale bar = 50 nm. m, pre- and postsynaptic membranes; b, bouton; c, synaptic cleft; d, postsynaptic density. (J,K) Synapses in CA1 that were cryoembedded in Lowicryl and immunoreacted for GluR2/3: (J) samples not developed for zinc; (K) double staining for zinc and GluR2/3. Note the poor structural preservation and difficulty of identifying synaptic vesicles in both cases. (L) A giant MF terminal is stained for vesicular zinc and contacts several AMPAR-immunoreactive thorny excrescences. Scale bars = 0.2 μm.

Figure 3.

Frequency distribution of NMDAR and AMPAR particles obtained from single section analysis of ZN+ and ZN− synapses in CA1. The frequency (%) of synapses labelled at the postsynaptic membrane or density (PSD) is plotted against the number of gold particles. (A) Immunolabelling for NR1 + NR2A/B is similarly distributed in ZN+ and ZN− synapses. (B) Immunolabelling for GluR1–3 at ZN+ synapses is skewed to lower values than at ZN− synapses. There was a higher fraction of ZN+ synapses stained with only one particle (P < 0.02) and a lower fraction with three (P < 0.05) and four particles (P < 0.02). (C) Observed distributions for AMPAR labelling at both ZN+ and ZN− synapses fit well the Poisson distributions expected for the corresponding mean (particle per synapse) values (ZN+ = 1.74 ± 0.3; ZN− = 1.05 ± 0.2). Data are expressed as mean ± SEM.

Figure 3.

Frequency distribution of NMDAR and AMPAR particles obtained from single section analysis of ZN+ and ZN− synapses in CA1. The frequency (%) of synapses labelled at the postsynaptic membrane or density (PSD) is plotted against the number of gold particles. (A) Immunolabelling for NR1 + NR2A/B is similarly distributed in ZN+ and ZN− synapses. (B) Immunolabelling for GluR1–3 at ZN+ synapses is skewed to lower values than at ZN− synapses. There was a higher fraction of ZN+ synapses stained with only one particle (P < 0.02) and a lower fraction with three (P < 0.05) and four particles (P < 0.02). (C) Observed distributions for AMPAR labelling at both ZN+ and ZN− synapses fit well the Poisson distributions expected for the corresponding mean (particle per synapse) values (ZN+ = 1.74 ± 0.3; ZN− = 1.05 ± 0.2). Data are expressed as mean ± SEM.

Figure 4.

Immunolabelling for GluR1 and GluR2 subunits in epon-embedded, stratum radiatum in CA1. Examples of ZN+ and ZN− synapses labelled with gold particles coding for GluR1 (15 nm) in (AC) or for GluR2 in (DF) (D and F, 10 nm; E, 15 nm). Black arrowheads, synaptic particles; white arrowheads, extrasynaptic particles. Scale bar = 0.2 mm.

Figure 4.

Immunolabelling for GluR1 and GluR2 subunits in epon-embedded, stratum radiatum in CA1. Examples of ZN+ and ZN− synapses labelled with gold particles coding for GluR1 (15 nm) in (AC) or for GluR2 in (DF) (D and F, 10 nm; E, 15 nm). Black arrowheads, synaptic particles; white arrowheads, extrasynaptic particles. Scale bar = 0.2 mm.

References

Andrasfalvy BK, Magee JC (
2001
) Distance-dependent increase in AMPA receptor number in the dendrites of adult hippocampal CA1 pyramidal neurons.
J Neurosci
 
21
:
9151
–9159.
Assaf SY, Chung SH (
1984
) Release of endogenous Zn2+ from brain tissue during activity.
Nature
 
308
:
734
–736.
Beaulieu C, Dyck R, Cynader M (
1992
) Enrichment of glutamate in zinc-containing terminals of the cat visual cortex.
Neuroreport
 
3
:
861
–864.
Bellocchio EE, Hu H, Pohorille A, Chan J, Pickel VM, Edwards RH (
1998
) The localization of the brain-specific inorganic phosphate transporter suggests a specific presynaptic role in glutamatergic transmission.
J Neurosci
 
18
:
8648
–8659.
Chen N, Moshaver A, Raymond LA (
1997
) Differential sensitivity of recombinant N-methyl-d-aspartate receptor subtypes to zinc inhibition.
Mol Pharmacol
 
51
:
1015
–1023.
Cole TB, Robbins CA, Wenzel HJ, Schwartzkroin PA, Palmiter RD (
2000
) Seizures and neuronal damage in mice lacking vesicular zinc.
Epilepsy Res
 
39
:
153
–169.
Danscher G (
1981
) Histochemical demonstration of heavy metals. A revised version of the sulphide silver method suitable for both light and electron microscopy.
Histochemistry
 
71
:
1
–16.
Debanne D, Gahwiler BH, Thompson SM (
1999
) Heterogeneity of synaptic plasticity at unitary CA3–CA1 and CA3–CA3 connections in rat hippocampal slice cultures.
J Neurosci
 
19
:
10664
–10671.
Dobrunz LE, Stevens CF (
1997
) Heterogeneity of release probability, facilitation, and depletion at central synapses.
Neuron
 
18
:
995
–1008.
Fabian-Fine R, Volknandt W, Fine A, Stewart MG (
2000
) Age-dependent pre- and postsynaptic distribution of AMPA receptors at synapses in CA3 stratum radiatum of hippocampal slice cultures compared with intact brain.
Eur J Neurosci
 
12
:
3687
–3700.
Golshani P, Liu XB, Jones EG (
2001
) Differences in quantal amplitude reflect GluR4-subunit number at corticothalamic synapses on two populations of thalamic neurons.
Proc Natl Acad Sci USA
 
98
:
4172
–4177.
Groc L, Gustafsson B, Hanse E (
2002
) Spontaneous unitary synaptic activity in CA1 pyramidal neurons during early postnatal development: constant contribution of AMPA and NMDA receptors.
J Neurosci
 
22
:
5552
–5562.
Haug FM (
1967
) Electron microscopical localization of the zinc in hippocampal mossy fibre synapses by a modified sulfide silver procedure.
Histochemie
 
8
:
355
–368.
He Y, Janssen WG, Rothstein JD, Morrison JH (
2000
) Differential synaptic localization of the glutamate transporter EAAC1 and glutamate receptor subunit GluR2 in the rat hippocampus.
J Comp Neurol
 
418
:
255
–269.
Heynen AJ, Quinlan EM, Bae DC, Bear MF (
2000
) Bidirectional, activity-dependent regulation of glutamate receptors in the adult hippocampus in vivo.
Neuron
 
28
:
527
–536.
Howell GA, Welch MG, Frederickson CJ (
1984
) Stimulation-induced up-take and release of zinc in hippocampal slices.
Nature
 
308
:
736
–738.
Kharazia VN, Weinberg RJ (
1999
) Immunogold localization of AMPA and NMDA receptors in somatic sensory cortex of albino rat.
J Comp Neurol
 
412
:
292
–302.
Koh JY, Suh SW, Gwag BJ, He YY, Hsu CY, Choi DW (
1996
) The role of zinc in selective neuronal death after transient global cerebral ischemia.
Science
 
272
:
1013
–1016.
Li Y, Hough CJ, Suh SW, Sarvey JM, Frederickson CJ (
2001
) Rapid translocation of Zn(2+) from presynaptic terminals into postsynaptic hippocampal neurons after physiological stimulation.
J Neurophysiol
 
86
:
2597
–2604.
Li Y, Hough CJ, Frederickson CJ, Sarvey JM (
2001
) Induction of mossy fiber — Ca3 long-term potentiation requires translocation of synaptically released Zn2+.
J Neurosci
 
21
:
8015
–8025.
Lu YM, Taverna FA, Tu R, Ackerley CA, Wang YT, Roder J (
2000
) Endogenous Zn(2+) is required for the induction of long-term potentiation at rat hippocampal mossy fiber-CA3 synapses.
Synapse
 
38
:
187
–197.
Luscher C, Nicoll RA, Malenka RC, Muller D (
2000
) Synaptic plasticity and dynamic modulation of the postsynaptic membrane.
Nat Neurosci
 
3
:
545
–550.
Martínez-Guijarro FJ, Soriano E, Del Rio JA, Lopez-Garcia C (
1991
) Zinc-positive boutons in the cerebral cortex of lizards show glutamate immunoreactivity.
J Neurocytol
 
20
:
834
–843.
Matsubara A, Laake JH, Davanger S, Usami S, Ottersen OP (
1996
) Organization of AMPA receptor subunits at a glutamate synapse: a quantitative immunogold analysis of hair cell synapses in the rat organ of Corti.
J Neurosci
 
16
:
4457
–4467.
Nusser Z, Cull-Candy S, Farrant M (
1997
) Differences in synaptic GABA(A) receptor number underlie variation in GABA mini amplitude.
Neuron
 
19
:
697
–709.
Nusser Z, Lujan R, Laube G, Roberts JD, Molnar E, Somogyi P (
1998
) Cell type and pathway dependence of synaptic AMPA receptor number and variability in the hippocampus.
Neuron
 
21
:
545
–559.
Palmiter RD, Cole TB, Quaife CJ, Findley SD (
1996
) ZnT-3, a putative transporter of zinc into synaptic vesicles.
Proc Natl Acad Sci USA
 
93
:
14934
–14939.
Paoletti P, Ascher P, Neyton J (
1997
) High-affinity zinc inhibition of NMDA NR1–NR2A receptors.
J Neurosci
 
17
:
5711
–5725.
Petralia RS, Esteban JA, Wang YX, Partridge JG, Zhao HM, Wenthold RJ, Malinow R (
1999
) Selective acquisition of AMPA receptors over postnatal development suggests a molecular basis for silent synapses.
Nat Neurosci
 
2
:
31
–36.
Pérez-Clausell J, Danscher G (
1985
) Intravesicular localization of zinc in rat telencephalic boutons. A histochemical study.
Brain Res
 
337
:
91
–98.
Phend KD, Rustioni A, Weinberg RJ (
1995
) An osmium-free method of epon embedment that preserves both ultrastructure and antigenicity for post-embedding immunocytochemistry.
J Histochem Cytochem
 
43
:
283
–292.
Racca C, Stephenson FA, Streit P, Roberts JD, Somogyi P (
2000
) NMDA receptor content of synapses in stratum radiatum of the hippocampal CA1 area.
J Neurosci
 
20
:
2512
–2522.
Rubio ME, Wenthold RJ (
1997
) Glutamate receptors are selectively targeted to postsynaptic sites in neurons.
Neuron
 
18
:
939
–950.
Sanchez-Andres JV, Palop JJ, Ramirez C, Nacher J, Molowny A, Lopez-Gracia C (
1997
) Zinc-positive presynaptic boutons of the rabbit hippocampus during early postnatal development.
Brain Res Dev Brain Res
 
103
:
171
–183.
Scanziani M, Gahwiler BH, Charpak S (
1998
) Target cell-specific modulation of transmitter release at terminals from a single axon.
Proc Natl Acad Sci USA
 
95
:
12004
–12009.
Shigemoto R, Kulik A, Roberts JD, Ohishi H, Nusser Z, Kaneko T, Somogyi P (
1996
) Target-cell-specific concentration of a metabotropic glutamate receptor in the presynaptic active zone.
Nature
 
381
:
523
–525.
Slomianka L (
1992
) Neurons of origin of zinc-containing pathways and the distribution of zinc-containing boutons in the hippocampal region of the rat.
Neuroscience
 
48
:
325
–352.
Soriano E, Frotscher M (
1993
) Spiny nonpyramidal neurons in the CA3 region of the rat hippocampus are glutamate-like immunoreactive and receive convergent mossy fiber input.
J Comp Neurol
 
333
:
435
–448.
Suh SW, Danscher G, Jensen MS, Thompson R, Motamedi M, Frederickson CJ (
2000
) Release of synaptic zinc is substantially depressed by conventional brain slice preparations.
Brain Res
 
879
:
7
–12.
Takumi Y, Ramirez-Leon V, Laake P, Rinvik E, Ottersen OP (
1999
) Different modes of expression of AMPA and NMDA receptors in hippocampal synapses.
Nat Neurosci
 
2
:
618
–624.
Ueno S, Tsukamoto M, Hirano T, Kikuchi K, Yamada MK, Nishiyama N, Nagano T, Matsuki N, Ikegaya Y (
2002
) Mossy fiber Zn2+ spillover modulates heterosynaptic N-methyl-d-aspartate receptor activity in hippocampal CA3 circuits.
J Cell Biol
 
158
:
215
–220.
van Lookeren CM, Oestreicher AB, van der Krift TP, Gispen WH, Verkleij AJ (
1991
) Freeze-substitution and Lowicryl HM20 embedding of fixed rat brain: suitability for immunogold ultrastructural localization of neural antigens.
J Histochem Cytochem
 
39
:
1267
–1279.
Varoqui H, Schafer MK, Zhu H, Weihe E, Erickson JD (
2002
) Identification of the differentiation-associated Na+/PI transporter as a novel vesicular glutamate transporter expressed in a distinct set of glutamatergic synapses.
J Neurosci
 
22
:
142
–155.
Vogt K, Mellor J, Tong G, Nicoll R (
2000
) The actions of synaptically released zinc at hippocampal mossy fiber synapses.
Neuron
 
26
:
187
–196.
Weisskopf MG, Nicoll RA (
1995
) Presynaptic changes during mossy fibre LTP revealed by NMDA receptor-mediated synaptic responses.
Nature
 
376
:
256
–259.
Wenthold RJ, Petralia RS, Blahos J II, Niedzielski AS (
1996
) Evidence for multiple AMPA receptor complexes in hippocampal CA1/CA2 neurons.
J Neurosci
 
16
:
1982
–1989.
Westbrook GL, Mayer ML (
1987
) Micromolar concentrations of Zn2+ antagonize NMDA and GABA responses of hippocampal neurons.
Nature
 
328
:
640
–643.
Yamada Y, Iwamoto T, Watanabe Y, Sobue K, Inui M (
2002
) PSD-95 eliminates Src-induced potentiation of NR1/NR2A-subtype NMDA receptor channels and reduces high-affinity zinc inhibition.
J Neurochem
 
81
:
758
–764.