Development of NMDA NR2 Subunits and Their Roles in Critical Period Maturation of Neocortical GABAergic Interneurons

Abstract

The goals of this research are to 1) determine the changes in the composition of NMDA receptor (NMDAR) subunits in GABAergic interneurons during critical period (CP); and 2) test the effect of chronic blockage of specific NR2 subunits on the maturation of specific GABAergic interneurons. Our data demonstrate that: 1) The amplitude of NMDAR mediated EPSCs (EPSCs_NMDAR) was significantly larger in the postCP group. 2) The coefficient of variation (CV), τ_decay and half-width of EPSCs_NMDAR were significantly larger in the preCP group. 3) A leftward shift in the half-activation voltages in the postCP vs. preCP group. 4) Using subunit-specific antagonists, we found a postnatal shift in NR2 composition towards more NR2A mediated EPSCs_NMDAR. These changes occurred within a 2-day narrow window of CP and were similar between fast-spiking (FS) and regular spiking (RSNP) interneurons. 5) Chronic blockage of NR2A, but not NR2B, decreased the expression of parvalbumin (PV), but not other calcium binding proteins in layer 2/3 and 4 of barrel cortex. 6) Chronic blockage of NR2A selectively affected the maturation of IPSCs mediated by FS cells. In summary, we have reported, for the first time, developmental changes in the molecular composition of NMDA NR2 subunits in interneurons during CP, and the effects of chronic blockage of NR2A but not NR2B on PV expression and inhibitory synaptic transmission from FS cells. These results support an important role of NR2A subunits in developmental plasticity of fast-spiking GABAergic circuits during CP.

Introduction

Functional NMDARs are heteromers that contain multiple glycine-binding ζ1 (NR1) subunits and at least one type of ε1–4 (NR2 A–D) subunit that determines the receptor’s distinct physiological properties and functions (Monyer et al., 1994; Laube et al., 1998; Kohr et al., 2003; Cull-Candy and Leszkiewicz, 2004; Furukawa et al., 2005; Ulbrich and Isacoff, 2008). χ1 (NR3A) subunits alone cannot form functional receptors, but can coassemble with the NR1/NR2 complex (Das et al., 1998; Perez-Otano et al., 2001). The notion that there is a developmental increase in the NR2A/NR2B ratio of synaptic NMDARs in principal/excitatory neurons is supported by robust experimental evidence (Hestrin, 1992; Watanabe et al., 1992; Monyer et al., 1994; Sheng et al., 1994; Li et al., 1998; Stocca and Vicini, 1998; Nase et al., 1999; Liu et al., 2004b). The change in NR2A/2B ratio is concomitant with the development of neural circuits and is regulated by sensory experiences (Carmignoto and Vicini, 1992; Flint et al., 1997; Ramoa and Prusky, 1997; Quinlan et al., 1999a; Quinlan et al., 1999b; Roberts and Ramoa, 1999; Chen et al., 2000; Philpot et al., 2001a; Philpot et al., 2001b; Fagiolini et al., 2003; Franks and Isaacson, 2005). It has been postulated that the NMDAR mediated long-term plasticity may underlie the formation of the CP (Kirkwood and Bear, 1994; Feldman et al., 1999; Barth and Malenka, 2001; Philpot et al., 2001a; Erisir and Harris, 2003; Daw et al., 2007). However, changes in NMDAR subunit composition may not be necessary for the role of NMDARs during CP plasticity (Lu et al., 2001; Medina et al., 2001).

It has been postulated that cortical GABAergic cells are prime targets for NMDA hypofunction in schizophrenia (Olney and Farber, 1995; Grunze et al., 1996). Several lines of evidence also link NMDA function with GABAergic cells. First, the NMDA receptors located on GABAergic interneurons are more sensitive to NMDA receptor antagonists than the NMDA receptors on pyramidal neurons (Grunze et al., 1996; Li et al., 2002). Second, chronic administration of NMDA antagonists decrease GAD expression and GABAergic inhibition (Behrens et al., 2007; Zhang et al., 2008; Lodge et al., 2009). Genetic deletion of NMDAR from Ppp1r2-Cre positive neurons (which randomly labels corticolimbic GABAergic cells) from birth decreases the number of GABAergic cells accompanied by disinhibition in pyramidal neurons and produced schizophrenia-related behaviors in mice (Belforte et al., 2010). Despite the importance of NMDAR to GABAergic interneurons, little is known regarding 1) developmental regulation of the molecular composition of NMDAR subunits, 2) the contribution of specific NMDAR subunits to the development of interneurons. Therefore the main goal of this study is to experimentally examine these two questions, which will provide insights into the mechanisms underlying experience-dependent maturation of inhibition.

GABA-mediated inhibition plays an important role in CP plasticity in mouse visual cortex (Komatsu, 1994; Fagiolini and Hensch, 2000). In the barrel cortex, it has been demonstrated that GABA-immunopositive synaptic populations increase abruptly between P10 and P15 (Micheva and Beaulieu, 1996), which is coincident with the CP for the formation of receptive fields in layer 2/3 barrel cortex (Stern et al., 2001). In addition, the maturation of GABA synapses (Jiao et al., 2006) and intrinsic firing properties of fast-spiking interneurons are experience-dependent (Sun, 2009). Although the involvement of glutamatergic transmission has been demonstrated (Sun et al., 2009), it is unclear whether the change in the molecular composition of NMDARs plays any role during CP. The main obstacle in studying the GABAergic cells in cortical circuits is that they occupy 10–20% of total neurons within the neocortex (Defelipe, 1993; Cauli et al., 1997; Kawaguchi and Kubota, 1997; Gupta et al., 2000; Defelipe, 2002). Most widely used techniques (e.g. western blot, in situ hybridization) in analyzing receptor molecular composition could not provide quantitative information regarding receptors involved in synaptic transmission in this small group of cells. By using whole-cell patch clamp recording from GAD67 GFP-positive neurons, for the first time, we demonstrate the change in NR2A/NR2B ratio in cortical interneurons during CP development. We show that blocking NR2A from early postnatal period impaired the expression of PV but not other calcium binding proteins, and selectively impaired inhibitory transmission from FS cells. This evidence support the hypothesis that developmental shift in NMDARs NR2 subunits is involved in the developmental maturation of specific GABAergic circuits during CP.

Materials and Methods

The use of animals was based on protocols approved by Institutional Animal Use Comittee of Univ. Wyoming. Transgenic GAD67-GFP (Δneo) mice. The selective expression of GFP is under the control of endogenous GAD67 gene promoter (Tamamaki et al., 2003; Jiao et al., 2006). In this strain, GFP is expressed in vitually all (~95%) GABAergic interneurons. In layer 4 barrel cortex, ~80% of the GFP positive neurons are PV-positive fast-spiking (FS) basket cells (Sun, 2009). In this study, ~50% of the GFP positive neurons in layer 2/3 barrel cortex were FS cells, the rest of GFP positive neurons were predominatntly regular spiking non-pyramidal (RSNP) neurons (data not shown). We divided GAD67 GFP-positive mice into the following two age groups: P6–10 as a pre-CP group and P20–40 as a post-CP group. The rationale for choosing appropriate age groups are as follows: in the barrel cortex layer 2/3, the development of receptive field properties exhibits a very narrow CP between P12–P14 (Stern et al., 2001). The major ascending projection connecting L4 and L2/3 barrel cortex is not fully developed until P15 (Micheva and Beaulieu, 1996; Stern et al., 2001; Bender et al., 2003; Bureau et al., 2004). Meanwhile, the expression of NR2A subunits starts at P7 and reaches peak at P20 (Monyer et al., 1994). We further divided the preCP group into P6–7 and P8–10 subgroups, and postCP group into P20–30 and P31–40 subgroups (Fig. 1A1). In this study, the differences in the NMDAR compositions between the pre- and post-CP groups, as well as within the preCP and the postCP groups were analyzed. In addition, based on the firing patterns, the postCP interneurons can be further divided into FS and RSNP neurons (Kawaguchi and Kubota, 1997). However, the interneurons in the preCP group could not be separated into different subgroups based on the firing patterns (instead they were collectively called immature multiple-spiking neurons) (Massengill et al., 1997), morphological features (Fig. 1B1) or PV expression (Alcantara et al., 1993). Therefore, we only separated the interneurons in the postCP group and made comparisons between the preCP and postCP (FS+RSNP), the preCP and postCP (FS), as well as the preCP and postCP (RSNP).

Figure 1.

EPSCs_NMDAR in GABAergic interneurons of layer 2/3 barrel cortex. A1, Experimental designs. A2, Schematic of recording paradigms in layer 2/3 barrel cortex. Interneurons were represented with cells with roundish soma. Dashed lines demarcate different layers of barrel cortex. B1, B2, Neurobiotin filled GAD67-GFP positive neurons in a P6 (B1) and a P24 (B2) mouse, respectively. Scale bar, 20 μm. The stimulating electrodes were placed 20–50 μm from recorded cells. C, 5 consecutive EPSCs evoked at +40 mV holding potential at indicated stimulus intensities (in V) in a representative P9 (left) and a P24 (right) GABAergic interneuron, respectively. The threshold intensity was indicated by (*). D, The input (stimulus intensity) and output (response amplitude) relationship curve. Note the required minimal stimulus intensity for evoking reliable EPSCs were smaller in the P24 neuron; the amplitude of evoked EPSCs was larger in the P24 neuron.

Slice Preparation

GAD67-GFP-positive mice, ages ranging from P6 to P40, were deeply anesthetized with nembutal (40 mg/kg) and decapitated. Brains were quickly removed and transferred into cold (~4°C) oxygenated cutting solution containing the following (in mM): 2.5 KCl, 1.25 NaH₂PO₄, 10 MgSO₄, 0.5 CaCl₂, 11 glucose, and 23.4 sucrose. Thalamocortical (TC) slices were prepared according to methods described by Agmon and Connors (Agmon and Connors, 1991; Agmon et al., 1995). The reason why we use TC slices is because this preparation maximally preserves intracortical connections between layer 4 and layer 2/3 (Micheva and Beaulieu, 1996; Stern et al., 2001; Bender et al., 2003; Bureau et al., 2004). The slices (200 μm) were cut using a vibratome (TPI, St. Louis, MO) and incubated in a holding chamber at 35°C for 1 h and subsequently at room temperature before being transferred to a recording chamber. The slices were fixed to a modified microscope stage, and allowed to equilibrate for at least 30 min before recording. The slices were minimally submerged and continuously superfused with oxygenated (95% O₂-5% CO₂) artificial CSF (aCSF) containing the following (in mM):126 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 2 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 glucose, pH 7.4, at the rate of 4.0 ml/min.

Whole-cell Patch Recording

Recordings were obtained at 35 ± 1°C from GFP positive neurons in layer 2/3 barrel cortex (Fig. 1A2). Capillary glass pipette recording electrodes (1.5–2 μm tip diameters, 3–6 MΩ) was filled with solution containing (in mM ): 120 cesium gluconate, 10 phosphocreatine-Tris, 3 MgCl₂, 0.07 CaCl₂, 4 EGTA, 10 HEPES, 4 Na₂-ATP, and 1 Na-GTP, (pH 7.4 adjusted with Cs-OH, 280 mOsm). Neurobiotin (0.5%; Vector Labs, Burlingame, CA) was regularly added to patch pipette solution for morphological reconstruction of neurons. A sharpened bipolar tungsten electrode, placed in close proximity to the recorded neurons, was used to deliver synaptic stimulation at low frequency (0.1 Hz) (Fig. 1A2). The intensity of the stimulus was maintained at ~ 15% over the threshold of the postsynaptic responses. The thresholds were defined as a large proportion of failures (Dobrunz and Stevens, 1997). For example, during five consecutive recordings, the stimulation could induce the postsynaptic response in only one or two of the recordings. In voltage clamp, the membrane potential was held at +40 mV to reveal EPSCs_NMDAR. EPSCs were recorded with a multiclamp 700B amplifier and pClamp software (Molecular Device, Sunnyvale, CA). Series resistance was continuously monitored. Experiments in which the resistance changed by >15% were rejected. EPSCs_NMDAR were evoked in the presence of 10 μM 2,3-dihydro-6-nitro-7-sulfamoyl-benzo(F) quinoxaline (NBQX) (Tocris Bioscience, Ellisville, MO) to block non-NMDA receptors, and 50 μM picrotoxin (PTX) (Tocris) to block GABA_A receptors. Monosynaptic inhibitory postsynaptic currents (IPSCs) were evoked in pyramidal neurons with the stimuli applied adjacently and was recorded at holding potential of 0 mV in the presence of glutamate antagonist NBQX (10 μM) and D(-)-2-amino-5-phosphonopentanotic acid (D-APV) (Tocris) (100 μM). Minimal stimulus intensity was defined based on methods described by Allen & Stevens (Allen and Stevens, 1994). The following chemicals were applied via a local perfusion system that allowed fast switching between media: 100 μM D-APV (Tocris), 3 μM ifenprodil (Tocris), 0.5 μM Ro25-6981 [(αR,βS)-α-(4-hydroxyphenyl)-β-methyl-4-(phenylmethyl)-1-piperidinepropanol] (Sigma), 0.5 μM (R)-[(S)-1-(4-bromo-phenyl)-ethylamino]-(2,3-dioxo-1,2,3,4-tetrahydroquinoxalin-5-yl)-methyl]-phosphonic acid (NVP-AAM077) (gift from Novartis, Switzerland), 0.5 μM (2S*,3R*)-1-(Phenanthren-2-carbonyl) piperazine-2,3-dicarboxylic acid (PPDA) (Ascent Scientific, Princeton, NJ).

Chronic blockage of NR2A or NR2B

GAD67-GFP-positive mice from the same littler were randomly divided into three groups: saline-injected (control), NVP-AAM077 injected (1.2 mg/kg, i.p.) (Fox et al., 2006) and Ro 25-6981-injected (6.0 mg/kg, i.p.) (Chaperon et al., 2003; Fox et al., 2006). The injections started at P7 for 11 or 18 days. The injection was given at a similar time daily. At P18 or P25, mice were given a lethal injection of Nembutal and perfused intracardially with 0.9% sodium chloride, followed by 4% paraformaldehyde. The brain was then removed, TC sections were prepared as previously described (Agmon and Connors, 1991).

Fluorescent labeling

TC sections (40 μm) were incubated in 0.6% H₂O₂ for 30 min, PBS washed, switched to 50% alcohol for 10 min, PBS washed, and then incubated in TBS with 0.5% Triton X-100, 2% BSA, and 10% normal goat serum for 2 h at room temperature, and incubated in primary antibodies directed against PV (1:1000; Calbiochem) and VGluT2 (1:250; Chemicon) at 4 oC overnight. The next day, after PBS washed, sections were incubated in Alexa Fluor 594 and goat anti-rabbit IgG (heavy and light chains; 1:1000; Invitrogen, Carlsbad, CA) and Alexa Fluor 350 and goat anti-mouse IgG (heavy and light chains; 1:1000; Invitrogen, Carlsbad, CA ) for PV and VGluT2 for 3 h and then washed, mounted, and coverslipped. The immunofluorescent specimens were examined using an epifluorescence microscope (Zeiss, Thornwood, NY) equipped with AxioCam digital color camera. Layer 2/3, layer 4 and layer 5 barrel cortex were identified by the VGluT2 staining. The outline areas of the layer 2/3, 4 and 5 were manually defined and cell numbers within the outline areas were counted by using interactive measurement function in Axiovision 4.6 (Zeiss, Thornwood, NY). Cell density was calculated as previously described (Jiao et al., 2006). The other primary antibodies used were a polyclonal rabbit anti-calbindin (1:1000; Sigma), a polyclonal rabbit anti-calretinin (1:500; Sigma), and a monoclonal mouse anti-GAD67 (1:1000; Chemicon).

Fixation, Immunochemistry and Histology

After recording, slices were fixed in 100 mM phosphate-buffered solution, pH 7.4, containing 4% paraformaldehyde at 4°C for at least 24 h. Endogenous peroxidase were blocked by incubating the slices in 1% H₂O₂ for 15–20 min. After several rinses in PB solution, the slices were transferred to 1% avidin-biotinylated horseradish peroxidase complex containing 0.1% Triton X-100 in PB (0.1 M, pH 7.4; ABC-Elite Camon, Wiesbaden, Germany) and left overnight at 4°C while being shaken lightly. Slices were then reacted using 3, 3-diaminobenzidine (DAB; Sigma) and 0.01% H₂O₂ until dendrites and axonal arbors were clearly visible (approximately 2–5 min). Slices were mounted on glass slides, embedded in DPX-mounting media (Aldrich, Milwaukee), and coverslipped for image analysis.

Data Analysis

Traces shown in the figures are the average of 10 to 20 consecutive EPSCs and all values are expressed as mean ± SEM. One-way ANOVA was used for multiple group comparisons and Bonferroni post-hoc test was used for the comparisons within groups. Two-tailed Student’s t test was performed for two group comparisons. Significance was placed at P <0.05. The rise time constants for EPSCs were calculated from a standard single-exponential fit of averaged recordings using Clampfit (Molecular Device, Sunnyvale, CA). The decay time constant was fitted by a standard double exponential function or a standard single-exponential function (Clampfit). The conductance-voltage (g-V) curve for each neuron was calculated as described by Kumar (Kumar and Huguenard, 2003) using the following equation:

$$g=I/(V-E_{\mathit{rev}·})$$ (1)

Where I is the averaged peak amplitude of 10 consecutive EPSCs while holding the membrane potential at a constant voltage. V is the holding potential. E_rev. is the reversal potential for each neuron. The maximum conductance (g_max) for each neuron was calculated by fitting the individual g-V curve with Boltzmann fit using Origin 6.1 (Microcal Software, Northampton, MA) with the following equation:

$$g={\{1+exp{[(v+v_{1/2})/k]}^{-1}\}}^{-1}.$$ (2)

Results

Isolation of EPSCs_NMDAR in preCP and postCP GFP-positive GABAergic interneurons in layer 2/3 barrel cortex

Whole-cell patch clamp recordings were obtained from visually identified GFP-positive GABAergic interneurons in layer 2/3 barrel cortex of GAD67-GFP mice. Interneurons were divided into two groups: preCP (postnatal days [P] 6–10, mean age=7.9 ± 0.2 day, n=65) and postCP (P20–40, mean age= 26.6 ± 0.7 day, n=65). EPSCs were evoked by tungsten bipolar electrode located in close proximity to the recorded cells (Fig. 1A2). Neurons in early postnatal period had fewer dendritic arbors and shorter dendritic processes (Fig. 1B1 vs. B2). The neurons in layer 2/3 barrel cortex receive sensory inputs from layer 4 ascending projections rather than thalamus (Micheva and Beaulieu, 1996; Stern et al., 2001; Bender et al., 2003; Bureau et al., 2004), thus the EPSCs were predominantly evoked intracortically from layer 4 to 2/3 projections, as well as excitatory synapses from the same layer pyramidal cells (Holmgren et al., 2003). In each recording, the stimulation intensity was adjusted to ~15% over the threshold to reliably evoke a single EPSC without failures (Fig. 1C). To isolate EPSCs_NMDAR, the neurons’ membrane potential was held at +40 mV in the presence of 50 μM PTX to block GABA_A receptors and 10 μM NBQX to block non-NMDA receptors. At P6, less than 15% (8/55) of the recorded interneurons had evoked EPSCs_NMDAR; at P7, ~40% (20/53); at P8–10, more than 85% (37/43); at P20–40, 94% (65/69). The synaptic EPSCs_NMDAR could not be evoked from P5 or younger neurons, and in rare cases when there were evoked EPSCs, they were very unstable and difficult to perform statistical analysis or pharmacological study upon. An example of the input/output relationship curve for EPSCs_NMDAR is shown in Fig. 1C&D. Compared with the postCP group, the preCP neurons usually required a higher stimulation intensity to evoke reliable EPSCs_NMDAR (Fig 1D) and the peak amplitude of the evoked EPSCs_NMDAR was generally smaller (e.g. Fig. 1D). Detailed comparisons between these two groups are shown in subsequent sections.

Differences in voltage-dependent properties of EPSCs_NMDAR in the preCP and postCP groups

Voltage-dependent properties of EPSCs_NMDAR in preCP and postCP GFP-positive GABAergic interneurons were examined by measuring the peak amplitude of EPSCs_NMDAR at different holding potentials from −80 to +40 mV. Representative traces for the two age groups were shown in Fig. 2A1& A2. For both preCP and postCP groups, the average reversal potentials (E_rev) were close to 0 mV (average E_rev was 1.3 ± 1.4 mV for preCP, n=6 and −2.6 ± 2.2 mV for postCP, n=13). The I-V curves showed prominent regions of inward rectification in I/V slopes in both age groups, however, the inward currents of the two groups peaked at slightly different holding potentials (−35 ± 3.1 mV in postCP and −30 ± 3.7 mV in preCP, p>0.05) (Fig. 2B). The conductance-voltage (g-V) relationship for each neuron was calculated from individual I-V curves for preCP (n=6) and postCP (n=13) groups. To quantify the voltage-dependent differences in the two groups, g-V relationships for each neuron were normalized to their respective maximum conductance (g_max) calculated from each individual Boltzmann fit, and the averaged g/g_max relationship was shown in Fig. 2C. The average half-maximal membrane potential (g/g_max=0.5) for the postCP group (−20.3 ± 2.0 mV) was significantly more hyperpolarized than preCP group (−11.1 ± 1.7 mV, p<0.05). This suggests that the voltage-dependent properties of the EPSCs_NMDAR are different between the two groups.

Figure 2.

Voltage-dependent properties of pharmacologically isolated EPSCs_NMDAR. A1, A2, The averaged traces of EPSCs_NMDAR from a representative P8 (A1) and P28(A2) neuron, respectively, recorded at indicated holding potentials ranged from −80 to +40 mV. Note that the amplitudes of inward and outward currents were larger in the P28 neuron. B, The average I-V curve for the neurons from preCP (n=6) and postCP (n=13). C, Normalized g-V data showing a leftward shift in V_0.5 for the postCP group (V_{0.5 postCP} = −13.8 ± 2.0 mV vs. V_{0.5 preCP} = −6.3 ± 5.7 mV, p<0.01). The solid line is the best-fitting Boltzmann equation, I/I_max = {1 + exp [(V + V_1/2)/K]}⁻¹, where V_1/2=−5.1 ± 3.9 mV, K= 0.015 in preCP and V_1/2=−6.7 ± 2.1 mV, K= 0.004 in postCP, respectively.

Difference in other properties of EPSCs_NMDAR

The postCP interneuron subtypes were further separated into FS and RSNP groups. FS and RSNP cells exhibit clear differences in expression of GABA_A receptors (Bacci et al., 2003), gap junctions (Beierlein et al., 2003), metabotropic glutamate receptor properties (Sun et al., 2009) and sensitivity to sensory deprivation (Sun, 2009). Under voltage-clamp mode, prior to break-in, we applied small amount of current to gain electrical access to recorded cells. We adjusted the amount of injected currents to avoid break-in and formation of whole-cell mode. Under this condition, the access resistance ranged from 400–700MΩ, which was sufficient to record truncated inward sodium currents underlying action potentials (APs) from interneuron (e.g. Fig 3A). The holding potentials were kept at a negative value of −60 mV to prevent Cs⁺ from diffusing into the cytoplasm. Under these conditions, the sodium currents underlying AP firing patterns from both the preCP and postCP interneurons were recorded and analyzed. In the postCP group, the AP firing patterns were different (Fig 3A) and clearly consistent with those of FS and RSNP groups as previously defined (Kawaguchi et al., 1995; Sun, 2009). In this study, cells were defined as FS neurons only when their AP properties met the following three criteria: the τ_decay<2.5 ms, half-width <2.0 ms, firing frequency >100 Hz (Fig. 3B1&B2). The interneurons were defined as RSNP neurons only when their AP properties met the three criteria: the τ_decay >3.5 ms, half-width >2.5 ms, firing frequency <100 Hz (Fig. 3B1&B2). The neurons that did not meet the above criteria were discarded from RSNP or FS groups (but were kept as part of the postCP group). In addition, the reconstruction of neurobiotin-filled neurons could also provide useful morphological information for the postCP group (supplemental Fig. 1A&B). Based on the electrophysiological analysis, we divided the 65 recorded postCP interneurons into FS (n=22) and RSNP (n=20) groups. Because the preCP interneurons were not fully differentiated, they exhibited similar firing patterns (Massengill et al., 1997) and morphological features. The reconstruction of neurobiotin-filled neurons could not provide useful morphological information for the preCP group (Fig. 1B1).

Figure 3.

Separation of FS and RSNP neurons in postCP group. A, Representative action potential (AP) traces from a FS (top) and a RSNP (bottom) neuron. Inset: the normalized traces of APs from the FS and RSNP neurons indicated by the squares. B1, B2, Separation of FS and RSNP neurons based on the firing frequency, half-width and weighted τ_decay of APs. C1, 10 consecutive EPSCs_NMDAR in a representative P7 (top), P23 RSNP (middle) and P23 FS (bottom) neuron, respectively. C2, The coefficient of variation (CV) in preCP and postCP RSNP and postCP FS neurons. * p<0.05, ** p<0.01, ***p<0.001, ns: no significance, as in this and the following graphs.

The properties of EPSCs_NMDAR in preCP and postCP GFP-positive GABAergic interneurons were tested at holding potential +40 mV and characterized by 5 parameters: the peak amplitude, the mean coefficient of variation (CV) of the peak amplitude, the half-width (widths at half-maximum amplitude, HWs), rise time constant (τ_rise) and decay time constant (τ_decay). The mean CV of the EPSCs_NMDAR in the preCP group neurons was significantly larger than the postCP RSNP (p<0.01) and postCP FS groups (p<0.001; Fig. 3C1&C2), respectively. The peak amplitude of EPSCs_NMDAR of the preCP group neurons was significantly smaller than the postCP RSNP (p<0.001) and postCP FS groups (p<0.001, Fig. 4A&B), respectively. HWs of the EPSCs_NMDAR in the preCP group neurons were significantly larger than the postCP RSNP (p<0.001) and postCP FS groups (p<0.001, Fig. 4C), respectively. The τ_rise of EPSCs_NMDAR in preCP group was significantly slower than postCP RSNP (p<0.01) and postCP FS groups (p<0.01, Fig. 4D), respectively. Based on the results from double exponential fit (τ_decay-Fast and τ_decay-Slow), the EPSCs_NMDAR in preCP group decayed at a significantly slower rate than the postCP RSNP and postCP FS groups (Fig. 4A1, A2&E), respectively. Using single exponential fit, τ_decay of preCP group (145.4 ± 6.2 ms) was also significantly slower than the postCP group (63.0 ± 2.9 ms, p<0.001, supplemental Fig. 2A1–A3). Within the postCP group, there were no significant differences between P20–30 and P31–40 subgroups in both decay time and half-width in the same cell type (FS or RSNP, see supplemental Fig 2). This suggests that developmental changes in these properties occurred during the CP.

Figure 4.

Developmental changes in the properties of EPSCs_NMDAR. A1, The averaged traces of EPSCs_NMDAR in a representative P7 (left) P23 RSNP (middle) and P23 FS (right) neuron, respectively. A2, The normalized traces of A1. The arrows indicated the values of fast and slow τ_decay for each traces. B, The comparison of the amplitude of EPSCs_NMDAR in preCP (white bar), postCP RSNP (grey bar) and postCP FS (black bar) neurons. Both RSNP and FS neurons of the postCP group had larger peak amplitude than preCP neurons (***p<0.001). No significant difference in the amplitude of EPSCs_NMDAR between RSNP and FS neurons. C, The comparison of the half-width of EPSCs_NMDAR in preCP (white bar), postCP RSNP (grey bar) and postCP FS (black bar) neurons (***p<0.001). D, The comparison τ_rise of the EPSCs_NMDAR in preCP (white bar), postCP RSNP (grey bar) and postCP FS (black bar) neurons (**p<0.01, ***p<0.001). E, The comparison of the τ_decay of EPSCs_NMDAR in preCP (white bar), postCP RSNP (grey bar) and postCP FS (black bar) neurons. (***p<0.001).

No significant differences in the CV (p=0.7), peak amplitude (p=0.2), τ_rise (p=0.7) and τ_decay-Slow (p=0.2) between RSNP and FS cells of the postCP group (Fig. 4B, D&E). However, the RSNP neurons had significant longer half-width (p<0.001) and τ_decay-Fast (p<0.001) than the FS (Fig. 4C&E). These data showed that there were a number of differences in the EPSCs_NMDAR properties, which may be due, at least in part, to the differences in the subunit composition of the NMDARs.

Since the NR2A and NR2B mediate different phase of EPSCs_NMDAR (Monyer et al., 1994; Sheng et al., 1994), we also calculated the area of the EPSCs_NMDAR. We compared the areas between control groups (preCP, postCP RSNP and postCP FS) and found that there was no significant difference (one-way ANOVA, p>0.05; supplemental Fig. 3A). This lack of differences in area of EPSCs_NMDAR can be explained by the opposite developmental changes in the peak amplitude and τ_decay. The neurons in preCP group had the smallest peak amplitude and the longest τ_decay; the postCP FS neurons had the largest peak amplitude and shortest τ_decay (Fig. 4). As a result, the areas (amplitude*duration) measured in the preCP and postCP FS were similar.

To examine whether different stimulation intensities could change the outcome of our results, we analyzed the responses of neurons under minimal and 15% over minimal stimulation conditions (supplemental Fig. 1C&D). Although the evoked responses were smaller under minimal stimulation vs. 15% over minimal stimulation, the trends of the developmental changes were the same. This suggests that the developmental differences in NMDAR properties demonstrated here do not depend on the stimulation paradigms used.

Developmental shift in NR2A/NR2B ratio

We next examined if there were any changes in pharmacological properties of EPSCs_NMDAR in GABAergic interneurons. Local perfusion of 0.5 μM NVP-AAM077 (a selective NR2A subunit antagonist, see next section for detailed pharmacological characterizations) (Auberson et al., 2002; Liu et al., 2004a; Gerkin et al., 2007) or 0.5 μM Ro25-6981 (an ifenprodil derivative that has high affinity and selectivity as an activity-dependent, voltage-independent, non-competitive NR2B subunit antagonist) (Mutel et al., 1998; Lynch and Guttmann, 2001), followed by the mixture (i.e. NVP-AAM077 plus Ro 25-6981) were used to examine the subunit composition for NMDARs. If the EPSCs_NMDAR were mediated solely by NMDARs composed of NR1/NR2A or NR1/NR2B subunits, the application of the mixture (NVP-AAM077 plus Ro25-6981) would eliminate the EPSCs_NMDAR. In cells where EPSCs_NMDAR were not completely blocked by the mixture, PPDA (selective NR2C/D antagonist, 0.5 μM) (Feng et al., 2004; Morley et al., 2005; Harney et al., 2008) were used to test whether the EPSCs were mediated by NR2C/NR2D. APV (100 μM) was used to confirm whether the EPSCs were mediated by NMDARs. The representative traces recorded in the absence and presence of each antagonist, or the mixture were shown in Fig 5A &B. The typical time courses of all antagonists’ actions were shown in Fig. 6A1–A3 & B1–B3.

Figure 5.

The pharmacological experiments showing developmental shift in NR2B/NR2A ratio. A1, B1, Representative EPSCs_NMDAR traces in the absence (black) and presence of NVP-AAM077 (0.5 μM, dark gray) and NVP-AAM077 (0.5 μM) plus Ro 25-6981 (0.5 μM, light gay) in a P8 neuron (A1) and P25 neuron (B1), respectively. A2, B2, Representative EPSCs_NMDAR traces in the absence (black) and presence of Ro 25-6981 (0.5 μM, dark gray) and the mixture of NVP-AAM077 (0.5 μM) plus Ro 25-6981 (0.5 μM, light gay) in a P8 neuron (A2) and P 28 neuron (B2), respectively. A3, B3, Representative EPSCs_NMDAR traces in the absence (black) and presence of ifenprodil (3 μM, dark gray) and NVP-AAM077 (0.5 μM) plus ifenprodil (3 μM, light gay), and PPDA (0.5 μM, light black) in a P7 neuron (A3) and P24 neuron (B3), respectively. C, The effects of NVP-AAM077, Ro 25-6981 and ifenprodil on the amplitudes of EPSCs_NMDAR. D, The effects of NVP-AAM077 and ifenprodil on the amplitude of EPSCs_NMDAR in postCP RSNP and postCP FS neurons.

Figure 6.

A1–A3, B1–B3, Representative traces from preCP (A1–A3) and postCP (B1–B3) groups, showing the effects of NR2A and NR2B subunit-specific antagonists on the amplitude of EPSCs_NMDAR. C, The dose-response curve of NVP-AAM077 fitted with Hill equation. The arrows indicated the IC₅₀ values of the fitted curves with (black dots) and without (grey dots) ifenprodil (3 μM) in the bath perfusion solution. D, E, The effect of NR2A and NR2B subunit-specific antagonists in all GABAergic cells regardless their age group. D, The effects of NVP-AAM077, Ro 25-6981 and ifenprodil on the amplitudes of EPSCs_NMDAR. E, Plot showing ‘NR2B remaining EPSCs’ (n=48, light gray bar), comparing with Ro 25-6981 (n=28, dark grey bar) and ifenprodil (n=51, black bar). ‘NR2A remaining EPSCs 1’ (from Ro 25-6981 experiments, n=28, light gray bar), and ‘NR2A remaining EPSCs 2’ (from ifenprodil experiments, n=51, dark gray bar) comparing with NVP-AAM077 (n=48, black bar) in all neurons examined, regardless their age group.

Local perfusion of NVP-AAM077 (0.5 μM) blocked 54.2 ± 13.4 % of EPSCs_NMDAR at P6–7; 62.8 ± 5.9 % at P8–10; 81.1 ± 6.1 % at P20–30; 73.5 ± 7.4 % at P31–40. Application of Ro25–6981 (0.5 μM) blocked 39.1 ± 8.7 % of EPSCs_NMDAR at P6–7; 37.9 ± 7.7 % at P8–10; 24.9 ± 9.9% at P20–30; 21.6 ± 7.2 % at P31–40 (Fig 5C). Next, we examined the effects of another NR2B antagonist, ifenprodil (Tovar and Westbrook, 1999; Cull-Candy et al., 2001). The results obtained from ifenprodil (3 μM) application were very similar to the effects of Ro25–6981. Ifenprodil blocked 43.2 ± 12.9 % of EPSCs_NMDAR at P6–7; 30.4 ± 8.1 % at P8–10; 21.2 ± 4.0 % at P20–30; 10.4 ± 2.6 % at P31–40 (Fig. 5C). The typical traces representing the effects of NVP-AAM077, Ro 25–6981 and ifenprodil on the amplitudes and τ_decay of EPSCs for preCP and postCP groups were shown in Fig 5A1–A3& B1–B3. If these pharmacological agents were receptor subunit selective as previously demonstrated, these pharmacological data demonstrated that there was a developmental shift in pharmacological sensitivities of the EPSCs_NMDAR during development: while postCP EPSCs_NMDAR became less sensitive to Ro 25–6981 or ifenprodil, they became more sensitive to NVP-AAM077. In addition, we analyzed the effects of NVP-AAM077 (0.5 μM) and ifenprodil (3 μM) on the peak amplitude of RSNP and FS neurons at P20–30 and P31–40. NVP-AAM077 blocked 67.5 ± 6.6 % of EPSCs_NMDAR in RSNP and 79.1 ± 4.8 % in FS at P20–30; 64.4 ± 11.3 % in RSNP and 80.1 ± 8.5 % in FS at P31–40. No significant difference between RSNP and FS neurons (Fig. 5D). Ifenprodil blocked 20.4 ± 5.2 % of EPSCs_NMDAR in RSNP and 20.4 ± 4.4 % in FS at P20–30; 21.4 ± 13.1 % in RSNP and 16.1 ± 6.6 % in FS at P31–40. No significant difference between RSNP and FS neurons (Fig. 5D). There were no significant differences in NR2A (NVP-AAM077) or NR2B (ifenprodil) antagonists mediated effects between P6–7 and P8–10 subgroups (Fig. 5C), or between P20–30 and P31–40 subgroups in the same cell types (FS or RSNP, Fig. 5D). This suggests that developmental changes in the pharmacological properties occurred during the CP.

The decay time is tens of milliseconds for the NR1/NR2A combination and hundreds of milliseconds for the NR1/NR2B and NR1/NR2C combination (Vicini et al., 1998), and seconds for NR1/NR2D channels (Monyer et al., 1994). Our earlier results indicated that young EPSCs_NMDAR had a much slower τ_decay (Fig. 4E). If the pharmacological agents are as selective as suggested, the application of these antagonists would be expected to have different effects on τ_decay. This was indeed the case. The τ_decay not only increased after NVP-AAM077 application, but the increase in τ_decay by NVP-AAM077 was also different in the two age groups. In the preCP group (n=16), the τ_decay-Fast increased 10.9 ± 17.3 ms (p= 0.3) and the τ_decay-Slow increased 298.9 ± 141.6 ms (p<0.05). In the postCP group (n=11), the τ_decay-Fast increased 35.6 ± 18.7 ms (p<0.05) and the τ_decay-Slow increased 168.3 ± 81.0 ms (p<0.05) (Table 1). Ro25–6981 and ifenprodil had the opposite effect on τ_decay: i.e., induced a decrease in τ_decay. The decrease in τ_decay was also different between the two age groups. In the preCP group (n=14), the τ_decay-Fast decreased 63.4 ± 20.1 ms (p<0.01) and the τ_decay-Slow decreased 289.6 ± 65.3 ms ₍p<0.0001). In the postCP group (n=12), the τ_decay-Fast decreased 6.5 ± 4.1 ms (p=0.1) and the τ_decay-Slow decreased 33.5 ± 72.5 ms (p=0.7) (Table 1). The effect of ifenprodil was similar to Ro 25–6981. In the preCP group (n=20), the τ_decay-Fast decreased 37.1 ± 8.8 ms (p<0.001) and the τ_decay-Slow decreased 173.6 ± 51.8 ms (p<0.001). In the postCP group (n=35), the τ_decay-Fast decreased 0.1 ± 1.9 ms (p=1.0) and the τ_decay-Slow decreased 72.2 ± 29.6 ms (p<0.05) (Table 1). The changes in the fast τ_decay did not reach statistical significance in the postCP group after Ro 25–6981 or ifenprodil application. These results suggest that the τ_decay-Fast was less likely mediated by NR2B in the postCP group. In contrast, the change in the τ_decay-Fast did not reach statistical significance in the preCP group after NVP-AAM077 application. It may indicate that the τ_decay-Fast was less likely mediated by NR2A in the preCP group. Thus, our data with either ifenprodil (or Ro25–6981) alone, or NVP-AAM077 alone, showed significant differences between preCP vs. postCP group. The effects of pharmacological agents are thus consistent with the EPSCs_NMDAR τ_decay data, which suggest that there was a shift in the contribution of NR2B to NR2A subunits to EPSCs_NMDAR in GABAergicinterneurons during development.

Table 1.

The effects of NR2A and NR2B antagonists on the fast and slow τ_decay of EPSC_NMDAR between the pre- and postCP interneurons.

	PreCP					PostCP
	Fast decay tau (ms)		Slow decay tau (ms)			Fast decay tau (ms)		Slow decay tau (ms)
	Before	After	Before	After	n	Before	After	Before	After	n
NVP-AAM077	124.4 ± 12.1	135.3 ± 14.0	491.1 ± 60.80	790.0 ± 143.1*	n=16	52.7 ± 8.5	88.2 ± 26.0*	287.4 ± 65.2	455.7 ± 90.6*	n=11
Ifenprodil	101.5 ± 4.5	64.4 ± 6.5 ***	455.1 ± 50.4	281.5 ± 44.5**	n=20	40.1 ± 2.8	39.9 ± 2.6	286.2 ± 28.1	214.0 ± 22.3*	n=35
Ro 25-6981	110.9 ± 15.7	47.4 ± 10.5**	543.7 ± 72.3	254.2 ± 40.7***	n=14	45.0 ± 4.5	38.5 ± 3.9	250.3 ± 49.8	216.8 ± 47.8	n=12

Data expressed in mean ± S.E.M. Paired Student’s t test was performed for comparisons of before and after antagonists’ applications. Significance was placed at P <0.05.

^*p<0.05,

^**p<0.01 and

^***p<0.001

We also compared the inhibition levels calculated by area following the application of antagonists. The area is determined by both peak amplitude and duration of EPSCs_NMDAR. The NR1/NR2A combination has larger peak amplitude (Monyer et al., 1992) and shorter decay time course than NR1/NR2B (Vicini et al., 1998). In theory, when NMDARs are mainly composed of NR1/NR2A and/or NR1/NR2B, blocking of NR2A subunits will decrease the peak amplitude of EPSCs_NMDAR and increase the decay time of NMDAR_EPSCs. In contract, blocking of NR2B subunits will decrease both the amplitude and decay time of EPSCs_NMDAR. To see whether this was the case, we compared the results calculated by peak amplitude with the results calculated by area. We found that the percentage of EPSCs_NMDAR blocked by NVP-AAM077 (NR2A antagonist) calculated by area was smaller than that by peak amplitude. In contrast, the percentage of EPSCs_NMDAR blocked by ifenprodil or Ro25–6981 (NR2B antagonists) calculated by area was bigger than that by peak amplitude (Fig. 5C vs. supplemental Fig. 3B; Fig. 5D vs. supplemental Fig. 3C). These results demonstrate that NR2A and NR2B subunit-ediated EPSCs_NMDAR exhibit distinct characteristics.

In addition, within the same age group (such as preCP and postCP), there was variability in the amplitude of evoked EPSCs_NMDAR under 15% over minimal stimulation intensity. To examine whether the variability in the amplitude of EPSCs_NMDAR affected the effect of antagonists, we also analyzed the relationship between the effect of ifenprodil (NR2B antagonist) and the amplitude of EPSCs_NMDAR. For each age group, the neurons were further divided into two subgroups based on their peak amplitude. Although there was a significant difference in the evoked amplitude of EPSCs_NMDAR between the groups (supplemental Fig. 1E), there was no significant difference in the effect of ifenprodil (supplemental Fig. 1F). These results indicate that the evoked amplitude of EPSCs_NMDAR has no significant impact on the effect of antagonists.

The specificity of NMDAR antagonists

Although NVP-AAM077 is reported to have very high affinity for NR2A subunits (Auberson et al., 2002; Liu et al., 2004a; Gerkin et al., 2007), it is also reported that NVP-AAM077 discriminates poorly (~10 fold) between NR2A- and NR2B-containing receptors in rodent NMDARs expressed in Xenopus oocytes examined by the ability of NVP-AAM077 to inhibit current responses induced by application of glutamate plus glycine (Neyton and Paoletti, 2006; de Marchena et al., 2008). To examine whether the concentrations of the antagonists used in this experiment were sufficient to distinguish different receptor subtypes underlying synaptically activated EPSCs_NMDAR, we next tested the specificity of NMDAR antagonists. In our dose-response experiments, we tested different concentrations of NVP-AAM077 (0.01, 0.05, 0.1, 0.5, and 5 μM) with or without ifenprodil (3 μM) in the bath perfusion solution (Fig. 6C) in P8–10 (n=6) and P20–21 (n=6) interneurons. As indicated in the dose-response curve, 0.5 μM NVP-AAM077 blocked 95.7± 3% of the NMDAR EPSCs with 3 μM ifenprodil in the bath perfusion, and 72.9 ± 5% of EPSCs_NMDAR without ifenprodil. 3 μM ifenprodil alone blocked 28.6± 3.4% (n=55) of EPSCs_NMDAR, which was very similar to the differences between the mixture (i.e. NVP-AAM077 plus ifenprodil) and NVP-AAM077 alone (95.7% −72.9% = 24%). In addition, there were no significant differences in the IC₅₀ values for NVP-AAM077 in the presence or absence of ifenprodil (Fig. 6C). These results suggested that 0.5 μM NVP-AAM077 was an appropriate concentration to be used for pharmacological separation of NR2A vs. NR2B mediated EPSCs_NMDAR in brain slices used in similar experiment conditions.

The specificity of antagonists to different subunits should be independent of animals’ age. We found that this was indeed true. When we combined the data from both age groups, NVP-AAM077 (0.5 μM) alone blocked 70.3 ± 3.7% of EPSCs_NMDAR (n=27); Ro25-6981(0.5 μM) alone blocked 31.5 ± 4.3% (n=26), and ifenprodil (3 μM) alone blocked 28.6 ± 3.4% (n=55) (Fig. 6D). If the concentrations we used were too high or too low, when we added the effects of NVP-AAM077 alone plus Ro25-6981alone (or NVP-AAM077 alone plus ifenprodi1alone), the percentage would be much larger or lower than 100%, respectively. We found that none of the above scenario appeared to be true. When the effects of NVP-AAM077 alone and Ro 25-6981alone, or NVP-AAM077 alone and ifenprodil alone were added together, they were very close to 100% (see Fig. 6D & E). In addition, the application of mixtures (NVP-AAM077 plus Ro 25-8198 or ifenprodil) blocked 98.4 ± 0.3% of EPSCs_NMDAR when NVP-AAM077 was applied first, 99.8 ± 0.1% when Ro 25-6981 was applied first, and 92.7 ± 2.1% when ifenprodil was applied first (Fig. 6D). After the mixtures were applied, the application of APV (100 μM) completely eliminated the residual of EPSCs_NMDAR in NVP-AAM077 (applied first) or Ro 25-6981 (applied first) groups, and inhibited 99.8 ± 0.2% of the residual of EPSCs_NMDAR in ifenprodil (applied first) group (Fig. 6D). These data suggested that at the current concentration used in our experiments, these antagonists (NVP-AAM077, Ro 25-8198, and ifenprodil) selectively blocked distinct groups of NMDARs. Together these data suggested that the NMDARs in GABAergic interneurons are mainly composed of NR1, NR2A and NR2B subunits.

If the NMDARs were mainly composed of NR1, NR2A and NR2B subunits, after the application of NR2A specific antagonist, the remaining EPSCs_NMDAR would be mediated by NR2B subunits, and vice versa. For additional confirmation, we calculated the remaining NR2A-and NR2B-mediated EPSCs by the following equations:

$$NR2B{\mathit{EPSCs}}_{\mathit{remaining}}=(\mathit{NVP}-\mathit{AAM}077+Ro25-6981)-\mathit{NVP}-\mathit{AAM}077.$$ (3)

$$NR2A{\mathit{EPSCs}}_{\mathit{remaining}1}=(\mathit{NVP}-\mathit{AAM}077+Ro25-6981)-Ro25-6981$$ (4)

and

$$NR2A{\mathit{EPSCS}}_{\mathit{remaining}2}=(\mathit{NVP}-\mathit{AAM}077+\mathit{ifenprodil})-\mathit{ifenprodil}$$ (5)

The average remaining NR2B EPSCs calculated from the equation (3) was 29.4 ± 3.7%, (n=27), which was not statistically different (p>0.05) from the effects produced by Ro 25-6981 alone (31.5 ± 4.3%, n=26) or by ifenprodil alone (28.6± 3.4%, n=55). The average remaining NR2A-mediated EPSCs calculated from equations (4) and (5) were 68.4 ± 4.3% (n=26) and 64.2 ± 4.2% (n=55), respectively, which were not statistically different (p>0.05) from the percentage of inhibition produced by NVP-AAM077 alone (70.3 ± 3.7%, n=28). In addition, there was no statistical significance between NR2A EPSCs _remaining 1 and 2 (p>0.05) (Fig. 6E). The results calculated by area were similar to that calculated by the peak amplitude (supplemental Fig. 3D&E vs. Fig. 6D&E). The difference in the selectivity of NVP-AAM077 between pyramidal neurons and interneurons may be due to the difference in the NMDAR components. Our results suggested that in interneurons NR2A and NR2B subunit form distinct receptors (see further details below). This may not be the case in principal neurons.

NR1/NR2A/NR2B and/or NR1/NR2C/NR2D mediated currents in preCP and postCP groups

In the preCP group (n=65), local perfusion of the mixture (NVP-AAM077 plus ifenprodil, or NVP-AAM077 plus Ro 25-6981) eliminated 99.7 ± 0.2% of the EPSCs_NMDAR in 64 out of 65 neurons, and local perfusion of APV (100 μM) completely eliminated the residual EPSCs_NMDAR. There was one neuron (P9), in which the mixture only eliminated 68.5% of EPSCs_NMDAR. The residual EPSCs_NMDAR was completely blocked by NR2C/D antagonist PPDA (0.5 μM).

In the postCP group (n=65), the mixture (NVP-AAM077 plus ifenprodil, or NVP-AAM077 plus Ro 25-6981) blocked 99.7 ± 0.1% of EPSCs_NMDAR in 56 out of 65 neurons. The residual EPSCs_NMDAR were completely blocked by APV. There were 9 out of 65 neurons, in which the mixture only blocked 69.3 ± 4.7% of EPSCs_NMDAR, and in 2 (a P20 and a P21) out of the 9 neurons, the residual EPSCs_NMDAR were completely blocked by PPDA (21% and 53.2%, respectively). However, the PPDA had no effect on the residual EPSCs_NMDAR of the rest of 7 neurons. The residual EPSCs of the 7 neurons were blocked 98.5 ± 1.2 % by APV.

These results indicate that in both the preCP and postCP groups, very small population of cells (1/65 in preCP and 2/65 cells in postCP) have NR2C/D-containing NMDARs, similar to spiny stellate neurons in layer 4 barrel cortex (Binshtok et al., 2006). In the postCP group (7/65) but not in the preCP group (0/65), NR1/NR2A/NR2B-containing NMDARs may contribute to EPSCs_NMDAR (there was a significant difference between the preCP and postCP groups, p<0.01). This was because the residual EPSCs_NMDAR were not blocked by NR2A, NR2B or NR2C/D antagonists in 7/65 cells tested in the postCP group, but were completely eliminated by APV, which indicates that the residual EPSCs_NMDAR may be mediated by NR1/NR2A/NR2B (Sheng et al., 1994; Kumar and Huguenard, 2003) which may have a reduced sensitivity to the antagonists (Hatton and Paoletti, 2005).

The inhibition of plasmalemmal glutamate transporter activity increased decay time constant of EPSCs_NMDAR

Although the τ_decay of EPSCs_NMDAR mainly depends on the NMDAR subunit composition (Monyer et al., 1992), changes in transmitter clearance can also cause changes in synaptic properties (Clements et al., 1992). Glutamate is not degraded in the synaptic cleft. The uptake of glutamate through glutamate transporters [GTs; (Gegelashvili and Schousboe, 1997), for a review, see (Danbolt, 2001)] is the most efficient way for removing glutamate from the extracellular space andmaintaining low synaptic glutamate levels. Here we tested the effects of glutamate spillover on the τ_decay of NMDA channels. TBOA is a nonsubstrate glutamate transporter inhibitor (Shimamoto et al., 1998) and can prolong the decay time course of NMDAR channels (Diamond, 2001; Arnth-Jensen et al., 2002; Tsukada et al., 2005). The slowing of the decay of EPSCs _NMDAR would reflect the extent of glutamate spillover, which causes cooperation in the activation of neighboring postsynaptic or peri-synaptic NMDARs (Carter and Regehr, 2000; Diamond, 2001; Arnth-Jensen et al., 2002; Lozovaya et al., 2004; Scimemi et al., 2004). To examine the contribution of plasmalemmal glutamate transporter to the τ_decay, we recorded the EPSCs_NMDAR in the absence and presence of 30 μM TBOA (Tocris) in the bath perfusion. The brain slices were allowed to equilibrate in bath solution (absence or presence of 30 μM TBOA) for 30 min before recording. Representative traces obtained from averaging 10 consecutive EPSCs were shown in Fig. 7A1& B1. The application of TBOA had no significant effect on the τ_rise in both groups (Table 2). However, the τ_decay and HWs increased significantly in both groups after TBOA application (Table 2). Interestingly, the amplitude of EPSCs_NMDAR decreased significantly after TBOA application in both pre- and postCP groups (Table 2). The decrease in the amplitude could be due to the postsynaptic NMDAR desensitization, which may be resulted from calcium dependent (Legendre et al., 1993; Medina et al., 1995; Krupp et al., 1996; Krupp et al., 1999) and calcium independent (Sather et al., 1990; Tong and Jahr, 1994) weakening of glycine binding site affinity (Benveniste et al., 1990; Vyklicky, Jr. et al., 1990). The magnitude of desensitization is controlled by the key residues in the NR2 subunit (Krupp et al., 1998) and actively modulated by second messengers in the postsynaptic cell (Rosenmund et al., 1995; Tong et al., 1995; Raman et al., 1996). In addition, at holding potential +40 mV, the holding current slightly increased from 370.7 ± 20.2 pA to 453.3 ± 55.2 pA after TBOA application in the postCP group (p=0.2), and significantly increased in preCP group (132.9 ± 9.3 pA in control vs. 210.0 ± 25.9 pA in TBOA, p=0.01). The effects of TBOA on both HWs and τ_decay were much larger in the preCP group vs. post CP group, thus suggesting that TBOA-sensitive plasmalemmal glutamate transporter play important role in the clearance of glutamate and that their role are presumably taken over by TBOA-insensitive transporters as interneurons develop. However, it was also possible that the difference in the change in time constants could be due to a developmental difference in presynaptic release between preCP and postCP groups. To test whether it was the case, we compared the change in the CV before and after TBOA application in both preCP and postCP groups. The CV significantly increased in the preCP group after TBOA application, which indicates the decrease in presynaptic release probability in preCP neurons (Schulz et al., 1994; Choi and Lovinger, 1997; Kirischuk et al., 2002). There was no significant change in the CV in postCP neurons after TBOA application (Table 2). However, the τ_decay significantly changed in both preCP and postCP groups after TBOA admission, which implies that the change in the time constants is unlikely due to the developmental difference in presynaptic release probabilities.

Figure 7.

Effects of TBOA on EPSCs_NMDAR. A1, B1, The representative traces from preCP (A1) and postCP (B1) in the absence (black) and presence of TBOA (30 μm, gray). The traces were the average of 10 consecutive EPSCs. The insets show the normalized traces. Note the change in the half-width and τ_decay in both groups. Insets: normalized traces of A1 & B1.

Table 2.

The effects of Inhibition of plasmalemmal glutamate transporter activity on the properties of EPSC_NMDAR.

		CV	Amplitude (pA)	Area (pA* ms)	τrise (ms)	HWs (ms)	τdecay-Fast (ms)	τdecay-Slow (ms)	n
Pre-CP	Control	0.3 ± 0.0	36.5 ± 2.3	5588.1 ± 551.5	5.1 ± 2.0	99.1 ± 32.8	106.9 ± 6.9	472.7 ± 29.7	65
	TBOA	0.4 ± 0.0 **	24.5 ± 2.5 **	9119.1 ± 1382.6	5.7 ± 0.8	269.5 ± 35.4 ***	390.8 ± 107.9 **	2228.6 ± 672.2 **	18
Post-CP	Control	0.2 ± 0.0	81.0 ± 5.5	6604.7 ± 759.7	4.1 ± 0.3	49.5 ± 2.7	47.7 ± 3.5	281.0 ± 21.2	65
	TBOA	0.2 ± 0.0	46.2 ± 1.5 ***	3061.2 ± 317.5 ***	4.9 ± 0.4	68.1 ± 9.8 *	119.7 ± 16.0 ***	651.5 ± 165.2 *	12

Data expressed in mean ± S.E.M. Paired Student’s t test was performed for comparisons of before and after antagonists’ applications. Significance was placed at P <0.05.

^*p<0.05,

^**p<0.01 and

^***p<0.001

Chronic blockage of NR2A from early postnatal age decrease the expression of PV in layer 2/3 and 4 barrel cortex in GAD67 GFP interneurons

If there were developmental increases in NR2A and decreases in NR2B mediated-EPSCs_NMDAR in interneurons, as we have demonstrated in this study, does this play any functional role in the maturation of interneurons during the CP of development? PV expression occurs at the CP in barrel and visual cortex (Alcantara and Ferrer, 1994; Yan et al., 1996; Czeiger and White, 1997; Maier and McCasland, 1997; Letinic and Kostovic, 1998; Hada et al., 1999; Moon et al., 2002). PV-cell network controls ocular dominance plasticity onset (Hensch, 2005; Sugiyama et al., 2008). Several groups (Itami et al., 2007; Liguz-Lecznar et al., 2009; Belforte et al., 2010), including ours (Jiao et al., 2006; Sun, 2009), have shown that the amount of PV expression is regulated by sensory activities during the CP. Thus, we examined the effects of the chronic blockage of NR2A or NR2B on PV expression in GAD67–GFP-positive interneurons. The density of PV-positiveneurons significantly decreased in layer 2/3 (170 ± 10/cm² in NVP-injected vs. 230 ±10/cm² in control, p<0.01) and layer 4 (310 ±20/cm² in NVP-injected vs. 460 ±40/cm² in control, p<0.001) barrel cortex in NVP-injected brain (n=6 mice) compared with control (n=6 mice, Fig. 8A1, B1&D1). We next tested whether the reduction in PV-positive cells is due to the reduced proliferation/increased apoptosisof PV-type GABAergic cells or the down-regulation of PV expressionin a subpopulation of GABAergic cells. By using the GAD67–GFPmouse, in which >95% of GABAergic cells express GFP under the promoter of GAD67 in motor cortex (Tamamaki et al., 2003) and in somatosensory cortex (Jiao et al., 2006), we counted the GFP-positive cells. The densities of GAD67–GFP-positive neuronsin the NVP-AAM077-injected barrels were similar to control brains (Fig. 8A2, B2 &E1). In layer 2/3, 510 ± 20/cm² in NVP-injected vs. 510 ±20/cm² in control (p=0.9); in layer 4, 610 ±40/cm² in NVP-injected vs. 600 ±40/cm² in control (p=0.8). The PV/GFP ratio showed a significant reduction in layer 2/3 and layer 4 in NVP-injected brains (Fig. 8A3, B3&F1). In layer 2/3, 0.45 ± 0.02 in control vs. 0.34 ±0.02 in NVP-injected (p<0.01); in layer 4, 0.76 ±0.03 in control vs. 0.51 ±0.03 in NVP-injected (p<0.001). However, the total numberof GAD67–GFP neurons, which is similar to the total numberof GABAergic neurons, remained overall unchanged. We concluded that chronic blockage of NR2A from early postnatal age(P7) can induce profound down-regulation of PV expressionin layer 2/3 and 4 GABAergic interneurons. We also examinedwhether there was a sensitive period for the down-regulation of PV expression by the chronic blockage of NR2A. In the GAD67-GFP mice (n=4 mice for control and NVP-injected, respectively), inwhich the NVP-AAM077 injection stopped at P18, we did not find significant changes in the density of PV cells. In layer 2/3,170 ± 2/cm² in control vs. 150 ± 9/cm² in NVP-AAM077-injected (p=0.8); in layer 4, 310 ± 40/cm² in control vs. 360 ± 20/cm² in NVP-AAM077-injected, (p=0.3, n = 12 slices, respectively; Fig. 8D2–F2). Thus our results suggest that chronic blockage of NR2A from P7 to P25, but not from P7 to P18, impaired developmental expression of PV. Finally, we examined the effects of chronic blockage of NR2A on the expression of calbindin and calretinin in GAD67–GFP-interneuons. No significant difference was found in the expression of calbindin and calretinin in control (n=4 mice) vs. NVP-AAM077-injected mice (n=4 mice, Table 3). In contrast, chronic blockage of NR2B via injection of Ro 25-6981 from P7 for the same period (i.e. to P25 or P18), did not induce significant changes in PV expression in GAD-67 GFP interneurons (Fig. 8), suggesting that NR2A, but not NR2B, is important for the developmental expression of PV.

Figure 8.

Effects of NVP-AAM077 and Ro25-6981 injection on PV expression in barrel cortex of GAD67–GFP mouse (TC sections). A, B&C, Photomicrographs of double-immunofluorescence-stained TC section (40 μm thickness) from a control mouse (A), a NVP-AAM077 injected (from P7–P25) mouse (B) and a Ro25-6981 injected (from P7–P25) mouse (C). Photomicrographs of PV (A1–C1), GFP (A2–C2), and merged images of PV+GFP (A3–C3). Scale bars, 50 μm. Dashed white lines outline different layers throughout the barrel cortex. D1–F1, Statistical comparison of cell density of PV-positive (D1), GFP-positive (E1) cells, and the ratio of PV/GFP-positive (F1) neurons in control (white bars, n=6 mice), NVP-AAM077 treated (grey bars, n=6 mice) and Ro25-6981 treated (black bars, n=6 mice) mice at age P25 (injection from P7 to P25). D2–F2, Statistical comparison of cell density of PV-positive and GFP-positive cells in control (white bars, n=3), NVP-AAM077 treated (grey bars, n=3) and Ro25-6981 treated (black bars, n=3) mice at age P18. No significant difference in the density of PV-positive (D2), GFP-positive (E2) and the ratio of PV/GFP-positive (F2) neurons between the control and NVP-AAM077 treated groups.

Table 3.

The calbindin and calretinin antibody staining in control and NVP-AAM077-injected groups.

		Calbindin			Calretinin
		Calbindin (cell/cm2)	GFP (cell/cm2)	Calbindin/GFP	Calretinin (cell/cm2)	GFP (cell/cm2)	Calretinin/GFP
Control	Layer 2/3	3000 ± 500	600 ± 60	5.5 ± 0.9	100 ± 20	600 ± 50	0.2 ± 0.03
	Layer4	3000 ± 500	700 ± 40	4.2 ± 0.5	60 ± 1	700 ± 50	0.09 ± 0.01
	Layer5	1000 ± 200	600 ± 40	1.9 ± 0.2	80 ± 4	600 ± 50	0.1 ± 0.05
NVP-injected	Layer 2/3	3000 ± 900	600 ± 40	5.1 ± 0.8	200 ± 50	600 ± 40	0.3 ± 0.02
	Layer4	3000 ± 400	700 ± 70	4.7 ± 0.7	60 ± 8	700 ± 50	0.1 ± 0.007
	Layer5	1000 ± 200	600 ± 50	1.8 ± 0.3	80 ± 3	600 ± 80	0.1 ± 0.03

No significant difference in the density of calbindin and calretinin expression in the layer II/III, IV and V barrel cortex between control and NVP-AAM07 treated groups.

Perisomatic inhibition is impaired after the chronic blockage of NR2A subunits

The properties of inhibitory synaptic transmission are associated with specific calcium channels located at the nerve terminals (Poncer et al., 1997; Poncer et al., 2000). In hippocampus or neocortex P/Q channels are predominantly expressed at GABAergic terminals of FS cells, whereas N- type calcium channels were predominantly expressed at GABAergic terminals of non-FS GABAergic cells (Poncer et al., 1997; Poncer et al., 2000; Ali and Nelson, 2006; Zaitsev et al., 2007). To further study the functional maturation of GABA transmission associated with FS and RSNP cells, we applied ω-agatoxin-IVA (2 μM) (Bachem Bioscience Inc., King of Prussia, PA ), a selectively P/Q calcium channels antagonist (Poncer et al., 1997; Poncer et al., 2000; Sun and Dale, 1998), and analyzed both evoked IPSCs (eIPSCs) and spontaneous IPSCs (sIPSCs) in spiny neurons in layer 4 barrel cortex, because layer 4 PV+ cells were also significantly decreased in NVP-AAM077 injected mice (Fig. 8D1 &F1). As shown in Fig. 9, the amplitude of eIPSCs was significant smaller in NVP-AAM077 injected neurons (Fig. 9A&B) and the half-width and τ_decay were significantly larger in NVP-AAM077 injected neurons (Fig. 9C&D). In addition, CV and paired-pulse ratio (IPSC2/IPSC1) were significantly larger in NVP-AAM077 injected neurons (Fig. 9E&F), indicating that the presynaptic release probability was lower in NVP-injected neurons. eIPSCs in both control and NVP-AAM077 injected neurons were almost entirely eliminated by bath application of ω-agatoxin-IVA (2 μM) (Fig. 9A&B), suggesting that eIPSCs onto the spiny neurons is produced almost exclusively by FS cells, as shown in previous studies (Poncer et al., 1997; Poncer et al., 2000; Ali and Nelson, 2006; Zaitsev et al., 2007). Therefore the decrease in eIPSCs in NVP-AAM077 injected neurons suggests a down-regulation of inhibitory transmission from FS cells.

Figure 9.

Characterization of evoked IPSCs from layer 4 pyramidal neurons in control and NVP-AAM077 injected mice. A, Representative traces of evoked IPSCs under minimal stimulation intensity before and after the application of agatoxin in both control and NVP-AAM077 injected neurons. B–F, The characterization of evoked IPSCs. B, The amplitude (measured by the first IPSC); C, HWs; D, τ_decay; E, CV of the amplitude; F, The paired-pulse ratio (IPSC2/IPSC1).

Next, we tested the effects of ω-agatoxin-IVA on sIPSCs. In control condition, the amplitude and frequency of IPSCs were reduced by agatoxin (Fig 10A1, B&C), suggesting that sIPSCs were mainly mediated via P/Q type calcium-dependent release from FS cells. There was no significant change in half-width and τ_decay (Fig. 10D&E). In NVP-AAM077 injected neurons, both the amplitude and frequency of sIPSCs were smaller than control (Fig 10A1vs. A2, B&C). Agatoxin decreased the amplitude of sIPSCs (Fig. 10A2&B) and had no significant effect on the frequency, half-width and τ_decay of sIPSCs (Fig 10C, D&E), supporting results with eIPSCs.

Figure 10.

Characterization of spontaneous IPSCs from layer 4 pyramidal neurons in control and NVP-AAM077 injected mice. A1, A2, Representative traces of spontaneous IPSCs before and after the application of agatoxin in both control (A1) and NVP-AAM077 injected (A2) neurons. B–E, The characterization of spontaneous IPSCs. B, The amplitude; C, The instant frequency; D, HWs and E, τ_decay.

Chronic blockage of NR2A from P7 to P25 increased the NR2B-mediacated EPSCs_NMDAR

To further test whether the chronic blockade of NR2A activity during P7–25 affects the development changes of NMDA receptor subunit compositions in the interneuron, we analyzed the proportion of NR2B-mediated EPSCs_NMDAR in layer 2/3 interneurons from both control and NVP-AAM077-injected mice. Because RSNP and FS showed similar developmental changes in NR2A/NR2B ratio in naïve animals (Fig 4), we therefore treated the interneurons as one group and compared the properties of EPSCs_NMDAR and the percentage of NR2B-mediated currents in control and NVP-AAM077 injected mice. We found that there were a decrease in both the amplitude and area of EPSCs_NMDAR and an increase in CV in NVP-AAM077 injected neurons. This result suggested that there were reductions in presynaptic release probability and changes in postsynaptic properties induced by NVP-AAM077. The half-width and τ_decay-fast and τ_decay-slow increased in NVP-AAM077 injected neurons, which suggested an increase of NR2B-mediated currents (Table 4). The application ifenprodil (3 μM) blocked 44.0 ± 7.9% of the peak amplitude of EPSCs_NMDAR in NVP-injected neurons, which was significantly larger than control (20.2 ± 3.4%, p<0.05) (Fig. 11A&B). In addition, the application ifenprodil (3 μM) blocked 56.8 ± 8.4% of the area of EPSCs_NMDAR in NVP-injected neurons, which was significantly larger than control (27.2 ± 4.2%, p<0.01) (Fig. 11C). The increases in the percentage of EPSCs_NMDAR blocked by ifenprodil, as well as the increase in half-width and τdecay, together indicated that the NR2B-mediated EPSCs_NMDAR increased after the chronic blockage of NR2A from P7–25. These results also support the notion that NR2A subunits, not the amount of NMDA currents carried by NR2A subunits per se, play an important role in the maturation of FS GABAergic cells.

Table 4.

The effects of chronic NVP-AAM077 injection on the properties of EPSC_NMDAR.

	CV	Amplitude (pA)	Area (pA* ms)	τrise (ms)	HWs (ms)	τdecay-Fast (ms)	τdecay-Slow (ms)	n
Control	0.2 ± 0.0	92.6 ± 9.4	6506.9 ± 948.9	3.6 ± 0.2	48.8 ± 0.8	44.8 ± 3.9	274.9 ± 31.6	29
NVP-injected	0.4 ± 0.0 **	59.6 ± 6.4 **	4987.4 ± 343.9 ***	3.5 ± 0.3	57.8 ± 3.4 *	55.8 ± 3.8 *	402.5 ± 47.0 *	22

Data expressed in mean ± S.E.M. Paired Student’s t test was performed for comparisons of before and after antagonists’ applications. Significance was placed at P <0.05.

^*p<0.05,

^**p<0.01 and

^***p<0.001

Figure 11.

Effects of chronic blockage of NR2A on EPSCs_NMDAR. A1, A2, The representative traces from control (A1) and NVP-injected (A2) in the absence (black) and presence of ifenprodil (3μm, gray). The traces were the average of 10 consecutive EPSCs. B, C. The percentage of inhibition of the peak amplitude of EPSCs_NMDAR (B) and the area of EPSCs_NMDAR (C) provided by ifenprodil (3 μm) in control and NVP-AAM077 injected mice.

Discussion

Developmental shift in the contribution of NR2A vs. NR2B subunits to EPSCs_NMDAR

In this study, by using whole cell patch clamp recording with application of subunit-specific antagonists, we demonstrated that there were developmental changes in the molecular composition of NMDAR in GABAergic interneurons in layer 2/3 barrel cortex. We found that NR2B-subunit-specific antagonists had larger effects on EPSCs_NMDAR in the preCP vs. postCP group. An opposite effect was obtained with NR2A-subunit-specific antagonist, NVP-AAM077, which had larger effect on EPSCs_NMDAR in the postCP vs. preCP group. The effects of NVP-AAM077 and ifenprodil (or Ro25-6981) were complimentary: while there was roughly 20% increase in NVP-AAM077-sensitive currents (i.e. NR2A), there was a similar amount of decrease in ifenprodil-sensitive currents (i.e. NR2B) in the postCP vs. preCP group (Fig. 5). A similar decrease in the sensitivity of EPSCs_NMDAR to NR2B antagonist during postnatal development has also been found in the hippocampal pyramidal neurons (Kirson and Yaari, 1996).

Compared with NR2B-containg receptor channels, NR2A-containing receptor channels have considerably faster rising and decaying currents and larger peak amplitude of EPSCs (Carmignoto and Vicini, 1992; Hestrin, 1992; Monyer et al., 1992; Kutsuwada et al., 1992; Monyer et al., 1994; Kirson and Yaari, 1996; Flint et al., 1997; Stocca and Vicini, 1998; Chen et al., 1999; Roberts and Ramoa, 1999; Cathala et al., 2000; Lei and McBain, 2002; Prybylowski et al., 2002), as well as much lower affinity for ifenprodil (Williams, 1993). In this study, the postCP (RSNP and FS) interneurons have larger peak amplitude, faster rising and decaying currents (Fig 4 & Table 1), which are consistent with a larger contribution from NR2A.

The NMDAR components are highly variable among interneurons (Sah et al., 1990; Lei and McBain, 2002; Maccaferri and Dingledine, 2002). Our results are consistent with these earlier findings. We found that the preCP neurons have the longest half-width and τ_decay; the postCP FS neurons have the shortest half-width and τ_decay; the postCP RSNP neurons have the intermediate half-width and τ_decay (Fig. 4). These data not only indicate that there are differences in the molecular composition of NMDARs between the pre- and postCP (RSNP+FS), but also imply that the postCP FS neurons may have the most NR2A component. This result is consistent with an early study from cultured interneuron (Kinney et al., 2006). It is also possible that RSNP neurons have more NR1/NR2A/NR2B tri-heteromeric channels, which exhibit intermediate decay time courses between the NR1/NR2A and NR1/NR2B channel types (Vicini et al., 1998). Although the overwhelming evidence that shows developmental switch in NR2A over NR2B in pyramidal neurons (Hestrin, 1992; Watanabe et al., 1992; Monyer et al., 1994; Sheng et al., 1994; Li et al., 1998; Stocca and Vicini, 1998; Nase et al., 1999; Liu et al., 2004b), here we provide the first evidence (to the best of our knowledge), to demonstrate that there is a similar developmental shift in the NR2A/NR2B ratio in GABAergic interneurons and this shift occurs exclusively during the 2 day CP (Fig 5 & Supplemental Fig 2).

In addition to the differences in the contribution of NR2A vs. NR2B to EPSCs_NMDAR, we also demonstrate several differences in glutamatergic synapses that are not related to postsynaptic expression of NMDARs. For example, the synaptic densities are low at early postnatal age (Micheva and Beaulieu, 1996; Washbourne et al., 2004), NMDARs are expressed on the surface of the dendrites and actively exo/endocytosed through the dendritic membrane (Washbourne et al., 2004), which could cause the smaller amplitude of EPSCs_NMDAR in preCP. The CV reflects both the number of release sites (N) and the release probability (p). Lower neurotransmitter release probability (Chuhma and Ohmori, 1998) and fewer release site (Iwasaki and Takahashi, 2001) in the preCP group may also be responsible for the larger CV in the preCP group. Interestingly, some of these features remained immature in chronic NVP-AAM077 treated animals (table 4).

The pharmacological data indicate that the τ_decay was increased by acute administering NVP-AAM077 and decreased by Ro25-6981 or ifenprodil in both preCP and post CP (RSNP and FS) groups. Interestingly, the changes in the τ_decay observed in the presence of NR2A or NR2B antagonists were different between the preCP vs. postCP neurons (Table 1), which may be due to the differences in phosphorylation state (Lieberman and Mody, 1994; Tong et al., 1995) and glutamate uptake (Carter and Regehr, 2000; Diamond, 2001; Arnth-Jensen et al., 2002; Lozovaya et al., 2004; Scimemi et al., 2004).

The contribution of NR1/NR2A/NR2B to the EPSCs_NMDAR

The g/g_max curve was leftward shifted in the postCP group, which might have resulted from the different affinity of NMDAR channels for Mg²⁺ (Chen and Huang, 1992; Kato and Yoshimura, 1993; Mayer et al., 1989). The NMDARs activated at more hyperpolarized holding potential would be expected to have a lower affinity for Mg²⁺ and less voltage-dependent Mg²⁺ blockage (Kumar and Huguenard, 2003). However, difference in affinity for Mg²⁺ has not been related to NR2A and NR2B receptors (Mayer et al., 1984; Nowak et al., 1984), but may be related to NR2C and NR2D (Kutsuwada et al., 1992; Binshtok et al., 2006). NR2C subunits have similar developmental patterns as NR2A subunits and are expressed mainly in cerebellum. NR2D subunits already present at birth and are located predominantly at extrasynaptic area at midbrain structures (Cull-Candy et al., 1998; Misra et al., 2000; Momiyama, 2000). NR2C and NR2D subunits are also expressed in dispersed interneurons at hippocampus (Monyer et al., 1994) and neocortex (Cauli et al., 1997) and in spiny stellate cells in layer 4 barrel cortex (Binshtok et al., 2006). However, the proportions of NR2C/2D-containing receptors are similar between pre- and postCP groups (1/65 vs. 2/65), in this case, it should not be responsible for the left-shift of g/g_max curve. Triheteromeric NMDARs, containing both NR2A and NR2B subunits have also been identified (Sheng et al., 1994; Mirshahi and Woodward, 1995), but their kinetic properties have not been determined due to lack of selective antagonists (Hatton and Paoletti, 2005). Our results suggest that an increase in NR1/NR2A/NR2B-containing NMDARs in 10% of postCP interneurons may be responsible for the difference in the conductance between pre- and postCP groups.

Postnatal chronic blockage of NR2A, but not NR2B, down-regulated the density of PV-positive interneurons and inhibitory synaptic transmission from FS cells

GABAergic interneurons not only participate in broad range of physiologically relevant process in the mammalian CNS (Toth and McBain, 2000; McBain and Fisahn, 2001; Klausberger and Somogyi, 2008; Gao and Strowbridge, 2009; McBain and Kauer, 2009; Wulff et al., 2009), but are also involved in neurological diseases. PV is a calcium binding protein that can bind the cytoplasmic Ca²⁺ and prevent the activation of Ca²⁺ activated K⁺ current, thus shorten the refractory period of action potential (Celio, 1986). PV-positive interneurons fire high-frequency action potential and synapse on the soma or axon initial segment of glutamatergic neurons, which enable them to potently regulate pyramidal neuron output (van Brederode et al., 1991). The loss of PV-positive interneurons is functionally associated with epilepsy (Marco et al., 1997; Arellano et al., 2004), schizophrenia (Woo et al., 1998; Zhang and Reynolds, 2002; Lewis et al., 2005; Lodge et al., 2009) and autism (Selby et al., 2007). Intriguingly, the treatment of NMDAR antagonists can produce a similar decrease in PV expression (Keilhoff et al., 2004; Kinney et al., 2006; Abekawa et al., 2007; Behrens et al., 2007). Genetic deletion of NR1 from interneurons produced similar results (Belforte et al., 2010). However, evidence linking the function of NR2A and NR2B receptor with PV neuron development in vivo is lacking.

In this experiment, chronic blockage of NR2A, but not NR2B, from P7 to P25, reduced the density of PV-positive interneurons in layer 2/3 and layer 4 barrel cortex (Fig. 8). The chronic administration of NR2A and NR2B antagonists will affect both pyramidal neurons and interneurons. NR2A-containingNMDARs increase the surface expression of GluR1 subunitof AMPA receptors in the pyramidal neurons (Kim et al., 2005). The decrease in the PV expression by chronic blockage of NR2A could be caused by a decrease in the glutamatergic transmission from the pyramidal neurons. However, because PV interneurons exhibit a higherNR2A/NR2B ratio than pyramidal neurons (Kinney et al., 2006), it is more likely that the decrease in the PV expression was caused by direct effects in interneurons. By measuring the eIPSCs and sIPSCs in layer 4 spiny cells, we found that the perisomatic inhibition, which was mainly mediated by PV-positive FS interneuron, was impaired after blocking NR2A activity (Fig. 9&10). Also, the NR2B-mediated EPSCs_NMDAR in layer 2/3 interneurons increased after the chromic blockage of NR2A from P7 to P25 (Fig. 11 and table 4). For the first time, we demonstrated that the NR2A receptor can affect the expression of calcium binding proteins and synaptic transmission from a specific group of interneurons (FS cells). However, little is known about the mechanisms underlying the role of NR2A in the maturation of PV-positive cells, which warrants future more detailed study of the NR2A-mediated signaling in PV interneurons in vivo.

The chronic blockage of NR2B had no effect on the expression of PV in the barrel cortex (Fig. 8). Our results obtained in vivo are similar to an earlier study, where NR2A, but not NR2B receptors, are shown to play a pivotal role in the maintenance of the GABAergic function of PV interneurons in culture (Kinney et al., 2006). In both studies, the application of NVP-AAM077, but not Ro25-6981, affected PV expressions. On surface, this result appears to link PV expression with NR2A receptor only. However, the interpretation of this result needs some caution. For example, NVP-AAM077 provides a much stronger inhibition of EPSCs_NMDAR than Ro 25-6981 does. The reduction in PV-containing interneurons after NVP-AAM077 injection could simply be due to the stronger inhibition provided by NVP-AAM077, rather than its subunit specificity. However, our results regarding increase in NR2B mediated currents in chronic NVP–AAM077 treated mice (see Fig. 11) suggest that it is NR2A subunits, but not the amount of NMDA currents carried by NR2A per se, appears to be responsible for developmental maturation of GABAergic neurons. Additionally, chronic blockage of NR2A from P7 to P18 did not induce significant change in PV-expression; this could have been due to the difference in the length of treatments, as opposed to a sensitive period effects. We do not know the expected in vivo half-life for NVP-AAM077 and Ro25-6981 in the brain. Although Ro 25-6981 has a very slow half-life of dissociation (>5 h) at 4°C in vitro (Mutel et al., 1998), it is difficult to predict how long the effectiveness of NVP-AAM077 or Ro25-6981can last in the brain in vivo. Under this circumstance, there are two factors that will determine whether the treatment will be effective or not. One is the cumulative effect of the chemicals (NVP-AAM077 or Ro25-6981) in vivo and the other is the length of the treatment.

In summary, we recorded synaptically evoked EPSCs_NMDAR from an early postnatal group (preCP) and a juvenile group (postCP) of GABAergic interneurons in layer2/3 barrel cortex. The differences in EPSCs_NMDAR properties and sensitivity to NR2 subunit-specific antagonists indicate that the molecular composition of NMDARs in GABAergic interneurons alters during critical developmental stages. The results from this research suggest that NR2A containing NMDARs play a pivotal role in the maturation of PV-positive cells in vivo. These results raise possibilities regarding involvement of specific NR2 subunits in experience-dependent plasticity of cortical inhibitory networks.

Supplementary Material

Supp Table S1

Supplemental Figure 1. A&B, Reconstruction of neurobiotin-filled neurons. A, Fast spiking interneurons. B, Regular spiking non-pyramidal neurons. Scale bar, 20 μm. C&D, The peak amplitude (C) and area (D) of evoked EPSCs_NMDAR under minimal vs. 15% over minimal stimulation conditions. E&F, Comparisons between neurons with peak amplitude smaller or lager than mean amplitude in preCP and postCP groups. E, In both preCP and postCP groups, the peak amplitude was much smaller in the neurons that belong to < mean amplitude group. F, In both preCP and postCP groups, there was no significant difference in the effect of ifenprodil.

Supp Table S2

Supplemental Figure 2. A1–A3. The single exponential fit in a preCP (A1), a postCP (RSNP) (A2) and a postCP FS (A3) neuron. Dashed lines were single exponential fitted τ_decay. Arrows indicated the values of the τ_decay B1–B3. The double exponential fit in the same neurons as in A1–A3. Dashed lines were double exponential fitted τ_decay. Arrows indicated the values of the τ_decay. C1–C3. The manually double exponential fit in the same neurons as in A1–A3. Dashed lines were double exponential fitted τ_decay. Arrows indicated the values of the τ_decay D-H2, Differences in the properties of EPSCs_NMDAR between postCP RSNP and FS neurons at P20–30 vs. P31–40. No significant differences were found in CV (D), peak amplitude (E), τ_rise (G) and slow τ_decay (H2) either within group (P20–30 or P31–40) or between groups (P20–30 vs. P31–40). There were significant differences in the HWs (F) and fast τ_decay (H1) within group, no significant difference between groups.

Supp Table S3

Supplemental Figure 3. A, The comparison of the area of EPSCs_NMDAR in preCP (white bar), postCP RSNP (grey bar) and postCP FS (black bar) neurons. No significant differences among groups (p>0.05). B, The effects of NVP-AAM077, Ro 25-6981 and ifenprodil on the area of EPSCs_NMDAR. C, The effects of NVP-AAM077 and ifenprodil on the area of EPSCs_NMDAR in postCP RSNP and postCP FS neurons. D, The total effects of NVP-AAM077, Ro 25-6981 and ifenprodil on the area of EPSCs_NMDAR. E, Plot of data calculated by area of EPSCs_NMDAR showing ‘NR2B remaining EPSCs’ (n=48, light gray bar), comparing with Ro 25-6981 (n=28, dark grey bar) and ifenprodil (n=51, black bar). ‘NR2A remaining EPSCs 1’ (from Ro 25-6981 experiments, n=28, light gray bar), and ‘NR2A remaining EPSCs 2’ (from ifenprodil experiments, n=51, dark gray bar) comparing with NVP-AAM077 (n=48, black bar) in all neurons examined, regardless their age group.