Excitatory synapses are stronger in the hippocampus of Rett syndrome mice due to altered synaptic trafficking of AMPA-type glutamate receptors

Significance

Rett syndrome is the most common intellectual disability in women after Down syndrome (1:10,000 incidence). Different mouse models of intellectual disability and autism exhibit deficits in synaptic plasticity, including impaired long-term potentiation (LTP) in mice lacking methyl-CpG-binding protein 2 (MeCP2); however, the bases of this deficit remain unclear. Using a combination of electrophysiology, time-lapse imaging, cell biology, and biochemistry, we provide direct evidence that naïve hippocampal synapses in Rett mice have all the hallmarks of potentiated synapses. Rett synapses also fail to insert and remove AMPA receptors properly to activated synapses, which freezes them in a nonplastic state. Our findings provide molecular, cellular, and network mechanisms underlying enhanced excitatory synaptic transmission and impaired LTP in Rett mice, identifying previously unidentified molecular targets for therapeutic intervention.

Abstract

Deficits in long-term potentiation (LTP) at central excitatory synapses are thought to contribute to cognitive impairments in neurodevelopmental disorders associated with intellectual disability and autism. Using the methyl-CpG-binding protein 2 (Mecp2) knockout (KO) mouse model of Rett syndrome, we show that naïve excitatory synapses onto hippocampal pyramidal neurons of symptomatic mice have all of the hallmarks of potentiated synapses. Stronger Mecp2 KO synapses failed to undergo LTP after either theta-burst afferent stimulation or pairing afferent stimulation with postsynaptic depolarization. On the other hand, basal synaptic strength and LTP were not affected in slices from younger presymptomatic Mecp2 KO mice. Furthermore, spine synapses in pyramidal neurons from symptomatic Mecp2 KO are larger and do not grow in size or incorporate GluA1 subunits after electrical or chemical LTP. Our data suggest that LTP is occluded in Mecp2 KO mice by already potentiated synapses. The higher surface levels of GluA1-containing receptors are consistent with altered expression levels of proteins involved in AMPA receptor trafficking, suggesting previously unidentified targets for therapeutic intervention for Rett syndrome and other MECP2-related disorders.

Rett syndrome (RTT) is a neurodevelopment disorder that affects girls, with an incidence of 1:10,000 (1). RTT individuals develop typically until 6–18 mo, when neurological symptoms begin, including intellectual disability, autistic features, deficits in motor control and sensory perception, breathing irregularities, and epilepsy disorders (2). Loss-of-function mutations in the transcriptional regulator methyl-CpG-binding protein 2 (MECP2) occur in 95% of RTT individuals (3), and Mecp2-deficient mice recapitulate several neurological features of RTT, including impaired hippocampal-dependent learning and memory (4).

Long-term potentiation (LTP), a cellular correlate of learning and memory (5), is impaired at hippocampal CA1 excitatory synapses of symptomatic Mecp2 knockout (KO) mice (6, 7), mice that express nonfunctional MeCP2 (Mecp2³⁰⁸) (8), and mice with a STOP codon before Mecp2 exon 3 (Mecp2^stop) (9, 10). LTP is also impaired at excitatory synapses of layers II/III and V in the primary somatosensory cortex of Mecp2 KO mice (11, 12) and Mecp2³⁰⁸ mice (8) and in cortico-lateral amygdala synapses of Mecp2 KO mice (13). Several mechanisms have been proposed to underlie these deficits, including altered composition of synaptic NMDA receptors, sparse connectivity through weak synapses, and LTP saturation (6, 10, 11).

Despite the consensus that LTP is impaired in the absence of MeCP2, the underlying cellular and synaptic mechanisms remain unclear. It is well established that long-term plasticity of excitatory synaptic transmission is critically determined by activity-dependent insertion and removal of AMPA receptors (AMPARs) to and from the postsynaptic membrane (14). Here, we assessed the synaptic strength of naïve hippocampal CA1 synapses in Mecp2 KO mice and revisited the issue of their LTP capability using a combination of electrophysiology and voltage-sensitive dye (VSD) imaging and tested whether activity-dependent synaptic AMPAR insertion and removal are affected by Mecp2 deletion. Our findings indicate that naïve hippocampal CA1 synapses are potentiated in Mecp2 KO mice due to altered activity-dependent synaptic trafficking of GluA1-containing receptors, which occludes LTP and spine plasticity.

Results

Excitatory Postsynaptic Currents Mediated by AMPARs Are Larger in Hippocampal Pyramidal Neurons of Symptomatic Mecp2 KO Mice.

We begin by confirming that the amplitude and spatiotemporal spread of neuronal depolarizations during single field excitatory postsynaptic potentials (fEPSPs) are enhanced in hippocampal slices from symptomatic male Mecp2 KO mice (P45–P65) (15). Hippocampal slices were stained with the fluorescent VSD RH-414 (30 µM) and imaged at 2,500 frames per second during single stimulation of Schaffer collateral (SC) afferents and recording of the evoked fEPSPs in CA1 stratum radiatum (SR) (Fig. 1A). The amplitude and spatiotemporal spread of VSD signals were larger in Mecp2 KO slices at four different afferent stimulus intensities (Fig. 1B, Top). The input–output (I–O) relationship of both fEPSP amplitudes (in mV) and VSD signal amplitudes (in ΔF/F) obtained at different stimulation strengths showed significantly larger responses in Mecp2 KO slices (WT n = 22 slices from seven mice and Mecp2^–/y n = 13/4; P < 0.0001; Fig. 1B, Bottom).

Fig. 1.

Enhanced excitatory transmission at hippocampal synapses in acute and cultured slices and primary neuronal cultures from Mecp2 KO mice. (A) Superimposed color-coded map of VSD signals on a bright field image of an acute hippocampal slice. Inset below shows the time course of ΔF/F from five individual pixels within CA1 SR evoked by a single afferent stimulation. (Scale bars, 10 ms, 0.1% ΔF/F.) (B) I–O relationship of VSD signals evoked by single afferent stimulation matches that of fEPSPs recorded simultaneously from CA1 SR. (Top) Representative color-coded maps of VSD signals at different stimulus intensities. Inset shows representative traces of VSD signals and fEPSPs from a WT (green line) and Mecp2 KO (red line) slice at 60 µA stimulus intensity. (Scale bars, 10 ms, 0.1% ΔF/F, 0.2 mV.) (C) I–O relationship between stimulus intensities and EPSC amplitudes. (Scale bars, 10 ms, 10 pA.) (D) Cumulative probability distribution of mEPSC amplitudes in CA1 pyramidal neurons in acute slices from symptomatic Mecp2 KO mice and age-matched WT mice. (Top) Representative traces of mEPSCs recorded in the presence of TTX (1 µM) and picrotoxin (50 µM). (Scale bars, 100 ms, 10 pA.) (E) Cumulative probability distribution of mEPSC amplitudes in CA1 pyramidal neurons in 10–12 DIV organotypic slice cultures prepared from either P5 Mecp2 KO mice or WT mice. Recording conditions, traces, and scale bars are as in D. (F) Cumulative probability distribution of mEPSC amplitudes in pyramidal-shaped hippocampal neurons in 9–11 DIV primary cultures prepared from either P1 Mecp2 KO mice or WT mice. Recording conditions, traces, and scale bars are as in D. Data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.01; Student’s t test in B, two-way ANOVA repeated measures in C, and K–S test in D–F. See also Fig. S1 and Table S1.

To directly measure the strength of excitatory synaptic transmission in CA1 pyramidal neurons, we performed whole-cell intracellular recordings. Evoked excitatory postsynaptic currents (EPSCs) were significantly larger in slices from symptomatic Mecp2 KO mice at all stimulation intensities (WT n = 18/9 and Mecp2^–/y n = 19/8; P < 0.0001; Fig. 1C). Measures of presynaptic function also indicate enhanced synaptic transmission: paired-pulse facilitation (PPF) of fEPSPs and EPSCs, and the coefficient of variation (CV) of EPSCs were smaller in Mecp2 KO neurons (PPF WT n = 11/5 and Mecp2^–/y n = 13/7; P = 0.029 for fEPSPs and P = 0.034 for EPSCs; CV WT n = 15/7 and Mecp2^–/y n = 13/9; P = 0.0467; Fig. S1 A–C). Miniature EPSCs (mEPSCs) mediated by AMPARs were also larger in CA1 pyramidal neurons of Mecp2 KO slices (1 µM TTX, 50 µM picrotoxin, at –60 mV; Fig. 1D, Top). The cumulative probability distribution of mEPSC amplitudes was shifted to the right in Mecp2 KO neurons (n = 14/8) compared with WT neurons (n = 9/5; P = 0.0063; Fig. 1D, Bottom). Albeit suggesting a higher mEPSC frequency in Mecp2 KO slices, the distribution of interevent intervals did not reach statistical significance (P = 0.067; Fig. S1D); the kinetics of mEPSCs were not affected by Mecp2 deletion (rise and decay times; Table S1). Consistent with larger mESPC amplitudes, total charge transfer and half-width were significantly larger in Mecp2 KO neurons (Table S1). Similar results were obtained in organotypic cultures of P5 hippocampal slices (recorded after 10–12 d in vitro) (Fig. 1E and Fig. S1E) and in dissociated cultures of P1 hippocampal neurons (recorded after 9–11 d in vitro) (Fig. 1F and Fig. S1F), where mEPSC amplitudes were significantly larger and mEPSC frequency significantly higher in Mecp2 KO neurons (P < 0.05). These results demonstrate that naïve excitatory synapses in hippocampal pyramidal neurons of Mecp2 KO mice are stronger than those in WT mice.

Fig. S1.

Related to Fig. 1. PPR and CV of EPSCs and interevent interval of mEPSCs in acute slices, slice cultures, and cultured neurons. (A) PPR of fEPSP amplitudes demonstrates higher presynaptic release probability in acute slices from symptomatic Mecp2 KO mice (WT n = 11/5 and Mecp2^–/y n = 13/7; *P = 0.029). (B) PPR of EPSC amplitudes demonstrates higher presynaptic release probability in acute slices from symptomatic Mecp2 KO mice (WT n = 11/5 and Mecp2^–/y n = 13/7; *P = 0.034). (C) Coefficient of variance of EPSCs demonstrates higher presynaptic release probability in acute slices from symptomatic Mecp2 KO mice (WT n = 15/7 and Mecp2^–/y n = 13/9; *P = 0.047). (D) Cumulative probability distribution of interevent intervals of mEPSC demonstrates significantly higher mEPSC frequency in CA1 pyramidal neurons in acute slices from symptomatic Mecp2 KO mice (WT n = 9/5 and Mecp2^–/y n = 14/8; P = 0.067). (E) Cumulative probability distribution of interevent intervals of mEPSC demonstrates significantly higher mEPSC frequency in CA1 pyramidal neurons in organotypic slice cultures from P5 Mecp2 KO mice (WT n = 9/4 and Mecp2^–/y n = 11/5; P = 0.036). (F) Cumulative probability distribution of interevent intervals of mEPSC demonstrates significantly higher mEPSC frequency in pyramidal-shaped neurons in primary cultures from P1 Mecp2 KO mice (WT n = 9/5 and Mecp2^–/y n = 14/6; P = 0.032). Data are presented as mean ± SEM; Student’s t test in A–C and K–S test in D–F.

Table S1.

Related to Fig. 1: Kinetic properties of mEPSCs in hippocampal CA1 pyramidal neurons

Genotype	Rise time, r; ms	10–90% rise time, r; ms	Decay, tau, ms	Half-width, ms	Charge transfer, pC
WT	2.66 ± 0.18	1.99 ± 0.14	4.12 ± 0.38	2.19 ± 0.09	43.73 ± 1.59
Mecp2–/y	2.56 ± 0.15	1.96 ± 0.15	4.78 ± 0.36	2.84 ± 0.16*	56.76 ± 4.37*

Data are expressed as mean ± SEM; WT n = 9 cells/5 mice and Mecp2^-/y n = 14/8; *P < 0.05; Student’s t test.

AMPARs are heterotetromeric complexes composed of GluA1–4 subunits (16). AMPARs containing GluA2 subunits display a linear current–voltage (I–V) relationship, whereas those lacking GluA2 exhibit inward rectification, have a higher single-channel conductance, and are Ca²⁺ permeable (17). It is generally agreed that under basal conditions, most, if not all, synaptic AMPARs in CA1 pyramidal neurons contain GluA2 subunits. To test if Mecp2 deletion affects the synaptic composition of AMPARs, we examined the rectification properties of EPSCs in the presence of picrotoxin (50 µM), dl-2-amino-5-phosphonovaleric acid (APV) (100 µM), and intracellular spermine (100 µM). The I–V relationship of AMPAR-mediated EPSCs shows smaller outward currents at positive potentials in Mecp2 KO neurons (WT n = 9/6 and Mecp2^–/y n = 11/6; +40 mV P = 0.016 and +60 mV P = 0.015; Fig. 2A). The rectification index (EPSCs amplitude at +40 mV/–60 mV) was significantly smaller in Mecp2 KO neurons (WT n = 9/6 and Mecp2^–/y n = 11/6; P = 0.023; Fig. 2B), suggesting that naïve excitatory synapses in CA1 pyramidal neurons of Mecp2 KO have an atypically higher content of GluA2-lacking AMPARs. Indeed, the selective blocker of GluA2-lacking AMPAR 1-Naphthyl-acetyl-spermine (NASPM) (50 µM) (18) caused a significantly larger reduction in EPSC amplitude in Mecp2 KO than in WT controls (WT n = 5 slices/3 mice and Mecp2^–/y n = 6/3; P = 0.043; Fig. 2C).

Fig. 2.

Higher levels of GluA1 in Mecp2 KO synapses. (A) I–V relationships of AMPAR-mediated EPSCs in the presence of APV (100 µM) and picrotoxin (50 µM); spermine (100 µM) was added to intracellular solution. (B) Representative traces of EPSCs evoked at –60 mV and +40 mV (Top). Rectification index was calculated as EPSC_{+40 mV}/EPSC_{–60 mV} (Bottom). (Scale bars, 10 ms, 10 pA.) (C) Representative traces of AMPAR-mediated EPSCs before and after bath application of 50 µM NASPM (Top). Inhibition of AMPAR-mediated EPSCs was normalized to baseline level before NASPM (Bottom). (Scale bars, 10 ms, 10 pA.) (D) GluA1 protein levels in hippocampal homogenates (Left) and PSD fractions (Right). (Top) Representative examples of Western immunoblots. (Bottom) Quantification of GluA1 bands (normalized to β-actin). (E) Representative images of dual immunolabeling for MeCP2 (Left) and GluA1 (Middle) in the hippocampus of WT (Top) and Mecp2 KO (Bottom) mice. GluA1 intensity is color-coded. GluA1 intensity was measured across different CA1 layers (rulers, 250 µm length); regions (rectangles) are magnified in the Insets at Right. [Scale bars, 50 µm (low magnification) and 10 µm (high magnification).] (F) Average data of GluA1 expression from the regions as shown with the rulers in E; a. u., arbitrary unit. (G) Representative images of immunolabeling for GluA1 in area CA1. [Scale bars, 50 µm (low magnification) and 10 µm (high magnification).] SO, stratum oriens; SP, stratum pyramidale; SR, stratum radiatum. (H) Average data of integrated intensity of GluA1 puncta (Left) and numerical density of GluA1 puncta (Right) in different CA1 layers. (I) Representative images of immunolabeling for GluA1 in cultured hippocampal neurons transfected with EGFP. GluA1 intensity was measured only in puncta colocalized with dendritic spines. (Scale bars, 5 µm.) Data are presented as mean ± SEM; *P < 0.05, **P < 0.01; two-way repeated-measures ANOVA in A and F, Mann–Whitney test in B, and Student’s t test in C, D, H, and I. See also Figs. S2 and S3.

The altered rectification properties and larger amplitude of evoked and quantal AMPAR-mediated EPSCs in Mecp2 KO neurons could result from a higher expression of GluA1-containing receptors. Western blot analyses show that GluA1 protein levels in whole tissue homogenates (WT and Mecp2^–/y n = 7 mice; P = 0.0041) and in postsynaptic density fractions (WT and Mecp2^–/y n = 3/6; P = 0.013) were significantly higher in the hippocampus of Mecp2 KO mice (Fig. 2D). Consistently, GluA1 immunostaining was more intense in hippocampal areas CA1 and CA3 and the dentate gyrus of Mecp2 KO mice (WT n = 21 slices/7 mice and Mecp2^–/y n = 25/8; P < 0.0001; Fig. 2 E and F and Fig. S2 A and B). The intensity of individual GluA1 puncta in the dendritic and somatic layers of area CA1 and dentate gyrus was significantly higher in Mecp2 KO mice [WT n = 10/5 and Mecp2^–/y n = 16/8; stratum oriens (SO) P = 0.019, stratum pyramidale (SP) P = 0.014, and SR P = 0.01; Fig. 2 G and H and Fig. S3 A and 3B). Similar results were obtained in dissociated cultures of P1 hippocampal neurons, where the intensity of individual GluA1 puncta in dendritic spines of EGFP-expressing neurons was significantly higher in Mecp2 KO neurons (WT n = 13 neurons from three cultures and Mecp2^–/y n = 16/3; P = 0.0035; Fig. 2I). The numerical density of GluA1 puncta was not significantly different between Mecp2 KO and WT mice (Fig. 2H, Right; SO P = 0.33, SP P = 0.99, and SR P = 0.33), consistent with similar dendritic spine densities in hippocampal pyramidal neurons at this age (19).

Fig. S2.

Related to Fig. 2. GluA1 protein levels by Western immunoblots and immunohistochemistry, and extrasynaptic AMPAR-mediated currents. (A) Representative images of dual immunolabeling for MeCP2 and GluA1 in the hippocampus of WT and Mecp2 KO mice. Hippocampal area CA3 and DG with a defined area (rulers, 250 µm length) was selected for measurement of GluA1 intensity across different layers, and the regions (rectangles) were magnified (Far Right). The images were color-coded for visualizing the intensity of GluA1 expression. [Scale bars, 50 µm (low magnification) and 10 µm (high magnification).] (B) Quantitative analyses demonstrate higher GluA1 levels across different layers of CA3 (Left) and DG (Right) in Mecp2 KO mice (n = 16 sections/8 mice), compared with WT mice (n = 10/5). (C) Representative traces of membrane currents induced by two 10-min bath applications of 250 nM AMPA in the presence of 1 µM TTX. (Scale bar, 2 min, 20 pA.) (D) Gaussian-fitted all-point amplitude histograms of membrane currents before and during bath application of AMPA. (E) Average data of AMPA-induced currents. The AMPAR antagonist CNQX (20 µM) completely blocked AMPA-induced currents (shaded colors). Data are presented as mean ± SEM; two-way repeated-measures ANOVA in B and E.

Fig. S3.

Related to Fig. 2. GluA1 levels in the dentate gyrus by immunohistochemistry. (A) Representative images of immunolabeling for GluA1 in the DG of Mecp2 KO mice. Enlarged images from the regions (rectangles) show puncta expression of GluA1. GL, granule layer; ML, molecular layer; PL, polymorph layer. [Scale bars, 50 µm (low magnification) and 10 µm (high magnification).] (B) Quantitative analysis demonstrates increased GluA1 intensity (Top) in the molecular layer and granule layer of Mecp2 KO mice. However, the number of puncta (Bottom; per 100 µm²) measured in the randomly selected region is unaltered. a. u., arbitrary unit. Data are presented as mean ± SEM; *P < 0.05, **P < 0.01; Student’s t test in B.

Because a significant portion of AMPARs is located in the extrasynaptic membrane, we next measured whole-cell currents evoked by a 10-min bath application of AMPA (250 nM, in the presence of 1 µM TTX; Fig. S2C). AMPA-induced currents calculated from all-point amplitude histograms (Fig. S2D) were significantly larger in Mecp2 KO neurons (WT n = 8/6 and Mecp2^–/y n = 9/6; P = 0.011; Fig. S2E). Intriguingly, a second AMPA application evoked significantly larger currents in WT neurons but not in KO neurons, which eliminated the statistical difference (P = 0.086; Fig. S2E). These observations suggest that Mecp2 KO neurons express higher levels of AMPARs in their plasma membrane.

Hippocampal CA1 Synapses in Slices from Symptomatic Mecp2 KO Mice Fail to Undergo LTP and Have AMPAR/NMDAR Ratios Similar to Those in Potentiated WT Synapses.

Simultaneous recordings of intracellular EPSCs and extracellular fEPSPs confirmed that Mecp2 KO slices failed to show LTP after theta-burst stimulation (TBS) of afferent SCs (WT n = 11/5 and Mecp2^–/y n = 13/7; Fig. 3 A–E). The failure of LTP in Mecp2 KO slices is not due to EPSC rundown because nonstimulated slices of both genotypes showed stable EPSC amplitudes throughout the entire recording time (Fig. 3 C and E, Bottom). Because naïve Mecp2 KO slices showed a steeper I–O of EPSC amplitude (Fig. 1C), we monitored the change of EPSC amplitude after LTP using a range of afferent stimulation intensities (measured at 45 min after TBS). The I–O relationship after LTP shows an upward shift in WT neurons, with larger EPSCs at all stimulation intensities (n = 10/5), whereas EPSC amplitude and their I–O relationship in Mecp2 KO neurons did not change after LTP-inducing stimulation (n = 6/4; Fig. 3F). On the other hand, CA1 pyramidal neurons in slices from younger (P20–P22) presymptomatic Mecp2 KO mice expressed LTP of EPSCs and an upward shift in the I–O relationship comparable to that observed in WT slices (WT n = 6 /3 and Mecp2^–/y n = 6/3; P > 0.05; Fig. S4).

Fig. 3.

Deficit in TBS-induced LTP and potentiated A/N ratio at CA3→CA1 synapses of Mecp2 KO mice. (A) Diagram of whole-cell recordings of EPSCs and simultaneous recordings of fEPSPs. (B) Representative traces of EPSCs before and 35 min after TBS–LTP induction. (Scale bars, 30 ms, 20 pA.) (C) Representative (Top) and average (Bottom) EPSC data show that TBS (arrows) evoked LTP of EPSCs in WT but not in Mecp2 KO neurons. In the representative recording (Top), note the larger baseline EPSC amplitudes in Mecp2 KO neurons. EPSC amplitude did not change in nonstimulated slices. (D) Representative traces of fEPSP before and 35 min after TBS–LTP induction. (Scale bars, 10 ms, 0.2 mV.) (E) Representative (Top) and average (Bottom) fEPSP data show that TBS (arrows) evoked LTP of fEPSPs in WT but not in Mecp2 KO slices. (F) Representative traces of EPSCs at different stimulation intensities before and after TBS–LTP induction. Shaded areas in the I–O plot represent the upward shift toward larger amplitude EPSCs after LTP induction. (Scale bars, 15 ms, 20 pA.) (G) Representative traces of EPSCs recorded at –60 mV and +40 mV before and after TBS–LTP induction. Bars show the time points where AMPAR-mediated and NMDAR-mediated EPSCs were measured. APV and picrotoxin blocked slower outward currents without affecting faster inward currents; addition of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) completely eliminated faster inward and outward currents. (Bottom) Average ratios of AMPAR to NMDAR amplitude in WT and Mecp2 KO slices, before and after TBS–LTP. (Scale bars, 20 ms, 50 pA.) Data are presented as mean ± SEM; *P < 0.05; two-way repeated-measures ANOVA in C, E, and F and Student’s t test in G. See also Figs. S4 and S5.

Fig. S4.

Related to Fig. 3. Basal synaptic transmission and TBS–LTP are unaffected in slices from presymptomatic Mecp2 KO mice. (A) Representative traces of EPSCs before and 20 min after TBS–LTP induction in slices from WT and presymptomatic Mecp2 KO mice (P20 and P22). (Scale bars, 30 ms, 20 pA.) (B) Representative (Top) and average (Bottom) EPSC data show that TBS (arrows) evoked LTP of EPSCs in both WT and presymptomatic Mecp2 KO neurons. In the representative recording (Top), note the similar baseline EPSC amplitudes in Mecp2 KO and WT neurons. (C) Representative traces of EPSCs at different stimulation intensities before and after TBS–LTP induction. Shaded areas in the I–O plot represent the upward shift toward larger amplitude EPSCs after LTP induction, which is observed in both WT and presymptomatic Mecp2 KO slices. (Scale bars, 15 ms, 20 pA.)

We next recorded AMPAR-mediated EPSCs at –60 mV and NMDAR-mediated EPSCs at +40 mV before and after LTP-inducing TBS stimulation (Fig. 3G). The amplitude of NMDAR-mediated EPSCs did not change after LTP-inducing stimulation either in WT or in Mecp2 KO neurons (WT n = 8/4 and Mecp2^–/y n = 11/6; P > 0.05; Fig. S5F). As expected from the synaptic insertion of AMPARs during LTP induction, the ratio of the amplitudes of AMPAR- and NMDAR-mediated EPSCs [AMPAR/NMDAR (A/N) ratio] was larger after LTP in WT neurons (n = 11/5), but it did not change in Mecp2 KO neurons (n = 13/7; Fig. 3G). Notably, the A/N ratio of naïve neurons before TBS was larger in Mecp2 KO neurons (WT n = 11/5 and Mecp2^–/y n = 13/7; P = 0.034).

Fig. S5.

Related to Figs. 3 and 4. Presynaptic function does not change after LTP induction in either WT or Mecp2 KO neurons. (A and B) Representative traces of EPSCs (A) and fEPSPs (B) evoked by paired-pulse stimulation before and after TBS-induced LTP. Mecp2 KO slices show lower PPR (n = 13/7) than WT slices (n = 11/5). PPR did not change after TBS–LTP. [Scale bar, 30 ms, 20 pA (A) and 10 ms, 0.2 mV (B).] (C and D) Representative traces of EPSCs (C) and fEPSPs (D) evoked by paired-pulse stimulation before and after pairing-induced LTP. Mecp2 KO slices show lower PPR (n = 9/4) than WT slices (n = 11/6). PPR did not change after pairing-induced LTP. [Scale bar, 30 ms, 20 pA (C) and 10 ms, 0.2 mV (D).] (E) The ratios of the inverse root of CV (1/CV²) of EPSC amplitudes before and 35 min after TBS–LTP induction were plotted against the normalized change in EPSC amplitude. Note that the 1/CV² in unstimulated naïve slices was also plotted as the controls. (F) Average amplitude of NMDAR-mediated EPSCs in WT and Mecp2 KO neurons before and after TBS-induced LTP. a, after; b, before. (Scale bars, 15 ms, 0.2 mV.) Data are presented as mean ± SEM; *P < 0.05, **P < 0.01; Student’s t test in A–F.

Because LTP induced by pairing postsynaptic depolarization and low-frequency presynaptic stimulation in synaptically connected layer 5 pyramidal neurons of primary somatosensory cortex is not affected in Mecp2 mutant mice (11), we tested a similar LTP-inducing protocol in hippocampal CA1 synapses (Fig. 4A). Depolarization to 0 mV during 3 Hz afferent stimulation for 50 s induced a robust LTP of EPSCs in WT neurons (n = 11/6), whereas it did not in Mecp2 KO neurons (n = 9/4; Fig. 4 B and C). As expected, the extracellularly recorded fEPSPs did not change after the pairing protocol because only the one neuron under recording was depolarized (Fig. 4 D and E). Neither PPF nor CV changed after LTP-inducing stimuli in WT slices (PPF TBS n = 11/5, pairing n = 10/5, and CV n = 9/4) and Mecp2 KO slices (TBS n = 16/8, pairing n = 5/3, and CV n = 9/4; Fig. S5 A–E).

Fig. 4.

Deficit of pairing-induced LTP in Mecp2 KO neurons. (A) Diagram of whole-cell recordings of EPSCs and simultaneous recordings of fEPSPs. (B) Representative traces of EPSCs before and 35 min after paring LTP induction. (Scale bars, 30 ms, 20 pA.) (C) Representative (Left) and average (Right) data show that pairing (arrows) induced LTP of EPSCs in WT but not in Mecp2 KO neurons. In the representative recording (Left), note the larger baseline EPSCs in Mecp2 KO neurons. (D) Representative traces of fEPSP before and 35 min after pairing LTP induction. (Scale bars, 10 ms, 0.2 mV.) (E) Representative (Left) and average (Right) fEPSP data were unaltered after pairing in both WT and Mecp2 KO slices. Data are presented as mean ± SEM; two-way repeated-measures ANOVA in C and E. See also Fig. S5.

LTP of VSD Signals Is Absent in Slices of Mecp2 KO Mice.

The induction of LTP at hippocampal CA1 synapses results in changes of the spatial spread of neuronal depolarizations across the hippocampal slice, which can be measured by VSD imaging (20). In WT slices, the amplitude of VSD signals (in ΔF/F) evoked in CA1 SR by a single pulse stimulation to SC afferents increased significantly after TBS-induced LTP, following a time course similar to that of the simultaneously recorded fEPSPs (n = 14/7; P < 0.0001 baseline vs. 40 min after TBS; Fig. 5 A–E and Movie S1). In addition, the area activated by a single SC stimulus increased significantly after LTP induction with the same time course of the fEPSP (P = 0.002; Fig. 5E); the duration of VSD signals did not change significantly after LTP (P = 0.14; Fig. S6). These changes in the amplitude and spatial spread of VSD signals after LTP induction were absent in Mecp2 KO slices (n = 8/5; amplitude P = 0.07 and spread P = 0.39; Fig. 5 A–E). In fact, evoked VSD signals in naïve Mecp2 KO slices were larger and spread more than those in WT slices (P < 0.0001; see also Fig. 1 A and B) and did not change after TBS.

Fig. 5.

Deficit of LTP of VSD signals in Mecp2 KO slices. (A) Representative time-lapse images showing VSD signals imaged at 2,500 frames per second and evoked by a single stimulation pulse before and after induction of TBS–LTP. (Scale bar, 200 µm.) (B) Representative traces showing simultaneously recorded fEPSPs and VSD signals (from A) before and 50 min after TBS–LTP induction in WT (blue line) and KO (red line) slices. (Scale bars, 10 ms, 0.1% ΔF/F.) (C–E) TBS of afferent fibers induced LTP of fEPSPs (C) and of the amplitude (D) and spatial spread (E) of VSD signals in WT slices but not in Mecp2 KO slices. Arrows show the time of TBS. Data are presented as mean ± SEM; two-way repeated-measures ANOVA in C–E. See also Fig. S6 and Movie S1.

Fig. S6.

Related to Fig. 5. The duration of VSD signals does not change after LTP induction. The duration of optical signals does not change after TBS–LTP induction either in WT or in Mecp2 KO slices. Arrow indicates the time of TBS. Data are presented as mean ± SEM; two-way repeated-measures ANOVA.

Higher Synaptic GluA1 Levels and Failure of Activity-Dependent Synaptic Trafficking of GluA1 in Cultured Mecp2 KO Hippocampal Neurons.

Incorporation of AMPARs into synapses is a well-established mechanism for the expression of LTP (21). Because naïve Mecp2 KO neurons show larger AMPAR-mediated synaptic currents and fail to express LTP, we tested whether activity-dependent synaptic trafficking of GluA1 is impaired in Mecp2 KO hippocampal neurons. Following established protocols (22), live neurons were stimulated with glycine (200 µM for 3 min) to induce chemical LTP (chem-LTP) and stained with an anti–N-terminal GluA1 antibody to label surface receptors, followed by fixation and permeabilization for subsequent staining of the total cellular GluA1 content. VGLUT1 was used to identify excitatory presynaptic terminals and allowed the quantification of the surface levels of GluA1 at synapses. In addition, expression of EGFP allowed aligning presynaptic VGLUT1 and postsynaptic GluA1 puncta with dendritic spines (Fig. 6A). As expected, the surface/total ratio of the intensity of synaptic GluA1 puncta was significantly higher in WT neurons after chem-LTP (WT Ctl n = 23 cells/8 cultures and WT Gly n = 12/5; P = 0.016; Fig. 6 B and C and Fig. S7). On the other hand, glycine stimulation failed to increase the synaptic GluA1 surface/total ratio in Mecp2 KO neurons (Mecp2^–/y Ctl n = 10/5 and Mecp2^–/y Gly n = 27/10; P = 0.48; Fig. 6 B and C). Interestingly, the GluA1 surface/total ratio was significantly higher in naïve Mecp2 KO synapses compared with naïve WT synapses (P = 0.0087).

Fig. 6.

Deficit of activity-dependent GluA1 insertion to synapses in Mecp2 KO neurons. (A) Representative images of labeling for VGLUT1 and total GluA1 in EGFP-expressing hippocampal neurons. (Scale bar, 2 µm.) (B) Representative images of triple labeling for VGLUT1, surface GluA1, and total GluA1 in hippocampal neurons. Enlarged images from the regions (rectangles) show puncta expression of these proteins, and arrows indicate their colocalization. [Scale bars, 10 µm (low magnification) and 5 µm (high).] (C) Quantitative analysis of GluA1 puncta intensity as the ratio of surface to total GluA1 in WT and Mecp2 KO neurons treated with either glycine or control buffer for 3 min. (D) Representative images of hippocampal neurons transfected with pH-sensitive SEP–GluA1 before and after glycine-induced chem-LTP. Arrows indicate appearance of new SEP–GluA1 puncta. Color code represents puncta intensity. (Scale bar, 1 µm.) (E) Representative line profiles of fluorescent intensity across SEP–GluA1 puncta at different time points. (F) Average data of the SEP–GluA1 puncta intensity before and after chem-LTP. (G) Quantitative analysis of GluA1 puncta intensity as the ratio of surface to total GluA1 in WT and Mecp2 KO neurons treated with either NMDA or control buffer for 3 min (Left). Representative images of labeling for surface GluA1 and total GluA1 in hippocampal neurons, treated with either NMDA or control buffer (Right). (Scale bar, 5 µm.) Data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001; Student’s t test in C and G and two-way repeated-measures ANOVA in F. See also Fig. S7.

Fig. S7.

Related to Fig. 6. Synaptic insertion of GluA1 does not occur after chem-LTP in cultured Mecp2 KO hippocampal neurons. Also shown is the average numerical density of SEP–GluA1 puncta before and after glycine-induced chem-LTP. Application of glycine (200 µM) for 3 min increased the number of SEP–GluA1 puncta in WT neurons (n = 12 cells/8 cultures) but not in Mecp2 KO neurons (n = 19/10). Data are presented as mean ± SEM; *P < 0.05; two-way repeated-measures ANOVA.

To directly monitor activity-dependent membrane insertion of GluA1, we transfected hippocampal neurons with plasmids containing superecliptic pHluorin (SEP)–GluA1 and performed time-lapse imaging during glycine stimulation (Fig. 6D). SEP is quenched by the acidic lumen of intracellular vesicles and recovers its fluorescence when exposed to the normal pH of the extracellular media upon insertion of SEP–GluA1 into the plasma membrane (23). As expected from activity-dependent insertion of GluA1, glycine significantly increased the number of SEP–GluA1 in WT neurons (Fig. S7). The fluorescence intensity of already existing SEP–GluA1 puncta became significantly higher after glycine stimulation (n = 12/8; P = 0.025, baseline vs. 30 min after treatment; Fig. 6 E and F). On the other hand, glycine did not affect the number or the intensity of SEP–GluA1 puncta in Mecp2 KO neurons (n = 19/10; P = 0.71; Fig. 6 D–F and Fig. S7).

To test activity-dependent endocytosis of GluA1 at synapses, we used an established protocol of chemically induced long-term depression (chem-LTD) by exposing cultured hippocampal neurons to NMDA (20 µM) for 3 min (24, 25), followed by estimation of surface GluA1 levels at synapses as described above. As expected, the surface/total ratio of the intensity of synaptic GluA1 puncta was significantly lower in WT neurons after chem-LTD (WT Ctl n = 16/5 and WT NMDA 22/5; P = 0.0035; Fig. 6G). On the other hand, NMDA stimulation failed to reduce the synaptic GluA1 surface/total ratio in Mecp2 KO neurons (Mecp2^–/y Ctl n = 13/4 and Mecp2^–/y NMDA n = 21/5; P = 0.69; Fig. 6G).

Altogether, these data indicate that the lack of activity-dependent synaptic trafficking of GluA1 is a plausible mechanism for the impaired expression of LTP in Mecp2 KO mice.

Lack of Structural Plasticity of Dendritic Spines in Mecp2 KO Mice.

The content of AMPARs in the postsynaptic density of spine synapses correlates positively with the size of the dendritic spine (26). Consistent with larger synaptic currents and higher synaptic GluA1 levels, the volume of dendritic spines was larger in EGFP-expressing Mecp2 KO hippocampal neurons in primary culture (P < 0.0001; Fig. 7A).

Fig. 7.

Larger spines and lack of activity-dependent spine plasticity in Mecp2 KO neurons. (A) Representative images of dendritic segments from different EGFP-expressing hippocampal neurons from WT (Top) and Mecp2 KO (Bottom) mice after 9–10 DIV. (Scale bar, 2 µm.) Cumulative probability distribution of spine volumes in WT and Mecp2 KO neurons (Right). (B) Representative image of a WT CA1 pyramidal neuron filled with biocytin during whole-cell recording and stained with Alexa 488-conjugated streptavidin after fixation (Top). Representative dendritic segment (rectangle) imaged at higher resolution (in the middle) and reconstructed for semiautomated spine detection, classification, and measurements (Bottom). [Scale bar, 50 µm (low magnification) and 10 µm (high magnification).] (C) Representative examples of dendritic segments from WT and Mecp2 KO CA1 pyramidal neurons with or without induction of TBS–LTP (Top). Average dendritic spine volume in WT and Mecp2 KO CA1 pyramidal neurons before and after induction of TBS–LTP (Bottom). (Scale bar, 2 µm.) (D) Cumulative probability distribution of spine volumes in WT and Mecp2 KO CA1 pyramidal neurons with or without induction of TBS–LTP. Data are presented as mean ± SEM; **P < 0.01, ***P < 0.001; Student’s t test in C and K–S test in A and D.

Different manipulations that induce LTP in CA1 pyramidal neurons also cause an enlargement of their dendritic spines (23, 27, 28). To characterize structural plasticity of dendritic spines, we performed measurements of spine volumes in CA1 pyramidal neurons filled with biocytin during whole-cell recordings of LTP. Forty-five minutes after TBS (or sham) stimulation, slices were fixed and neurons were stained with Alexa 488-conjugated streptavidin for subsequent confocal microscopy and three-dimensional reconstructions (Fig. 7B). Dendritic spine volumes in naïve Mecp2 KO neurons were larger than that of unstimulated WT neurons (P = 0.0021; Fig. 7C). As expected from activity-dependent spine plasticity, the volume of dendritic spines in WT CA1 pyramidal neurons that received TBS and showed LTP was significantly larger than those of unstimulated WT neurons (WT Ctl n = 962 spines/34 dendritic segments/8 neurons/4 mice and WT TBS n = 1,365/50/11/5; P < 0.0001; Fig. 7C). On the other hand, spine volume did not change after TBS in Mecp2 KO neurons (Mecp2^–/y Ctl n = 862/32/7/3 and Mecp2^–/y TBS n = 1,683/66/12/5; P = 0.33; Fig. 7C). Cumulative probability distributions of spine volumes show a statistically significant shift toward larger volumes in WT-TBS, Mecp2 KO-Ctl, and Mecp2 KO-TBS neurons compared with WT-Ctl cells (P < 0.0001; Fig. 7D). These results confirm that naïve excitatory CA3→CA1 spine synapses of Mecp2 KO neurons are stronger than in WT neurons and demonstrate that they do not express activity-dependent structural plasticity.

Altered Levels of GluA1, Phospho-845 GluA1, SAP97, and EEA1 in Mecp2 KO Mice.

The phosphorylation of GluA1 subunits at Ser831 or Ser845 mediates its trafficking into synapses during LTP (29). To test the status of this posttranslational modification in Mecp2 KO mice, we measured the levels of GluA1 phospho-Ser831 and phospho-Ser845 GluA1 in acute hippocampal slices after inducing chem-LTP with forskolin (50 µM) and the phosphodiesterase inhibitor rolipram (0.1 µM) (30). Similar to the deficits in TBS–LTP and pairing-induced LTP, Mecp2 KO slices failed to express chem-LTP (WT n = 8 slices/5 mice and Mecp2^–/y n = 8/5; Fig. 8A). WT slices homogenized at the end of chem-LTP recordings show significantly higher levels of phospho-Ser845 GluA1, without changes in phospho-Ser831 (WT Ctl n = 13 slices/6 mice and WT forskolin n = 14/7; P < 0.0001; Fig. 8 B and C). The levels of phospho-Ser845 GluA1 increased after forskolin stimulation in Mecp2 KO slices but significantly less (∼2.8-fold) than in WT slices (∼4.3-fold; Mecp2^–/y Ctl n = 13/6 and Mecp2^–/y Fors n = 14/7; P < 0.0001; Fig. 8B); levels of phospho-Ser831 GluA1 did not change after forskolin (P = 0.75; Fig. 8C). Consistent with the observations that naïve Mecp2 KO synapses have features of potentiated synapses, the levels of phospho-Ser845 GluA1 were significantly higher in untreated Mecp2 KO slices compared with WT slices (P = 0.0037; Fig. 8B), without differences in phospho-Ser831 (P = 0.82; Fig. 8C).

Fig. 8.

GluA1 phospho-Ser845 levels are higher and do not change after chem-LTP in Mecp2 KO mice. Shown is the altered expression of SAP97 and EEA1 in Mecp2 KO mice. (A) Representative traces of fEPSPs before and after induction of forskolin-induced chem-LTP. Also shown is the time course of average fEPSPs during forskolin-induced chem-LTP. (Scale bar, 10 ms, 0.2 mV.) (B and C) Quantitative analysis of GluA1 phosopho-Ser845 (B) and phospho-Ser831 (C) in homogenates of hippocampal slices with or without forskolin/rolipram exposure, with (Fors/s) or without (Fors) afferent stimulation. (Top) Representative Western immunoblots. (D) Quantitative analysis of SAP97 protein (Left) and mRNA (Right) in homogenates of hippocampal samples. (Top) Representative Western immunoblots. (E) Quantitative analysis of EEA1 protein (Left) and mRNA (Right) in homogenates of hippocampal samples. (Top) Representative Western immunoblots. Data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001; two-way repeated-measures ANOVA in A and Student’s t test in B–E.

Phosphorylation of GluA1 at serine 845 by protein kinase A (PKA) requires its association to the GluA1 C terminus via synapse-associated protein 97 (SAP97) (31). SAP97 is a major component of the postsynaptic density and plays a critical role in activity-dependent synaptic trafficking of GluA1 (32, 33). Consistent with a model of deregulated synaptic trafficking of AMPARs, mRNA and protein levels of SAP97 in hippocampal samples from Mecp2 KO mice were significantly higher than in WT (WT n = 5 mice and Mecp2^–/y n = 5; mRNA P = 0.008 and protein P = 0.02; Fig. 8D). Because surface-level AMPARs at synapses depend on the relative rates of receptor exocytosis and endocytosis, higher levels of GluA1 may also result from impaired endocytosis of AMPARs. EEA1, a membrane protein associated with the cytoplasmic face of early endosomes, participates in the endocytosis of neurotransmitter receptors by regulating the fusion of vesicle-carrying receptors to the recycling endosome at hippocampal synapses (34). mRNA and protein levels of EEA1 in hippocampal samples from Mecp2 KO mice were significantly lower than in WT (WT n = 5 and Mecp2^–/y n = 5; mRNA P = 0.047 and protein P = 0.044; Fig. 8E).

Taken altogether, these observations provide a potential molecular mechanism for the impairment of LTP in Mecp2 KO mice—namely, the unbalanced synaptic trafficking of GluA1 caused by increased insertion into and impaired endocytosis from the postsynaptic density.

Discussion

Our results indicate that naïve excitatory synapses on hippocampal pyramidal neurons from symptomatic Mecp2 KO mice have all of the features of potentiated connections: larger AMPAR-mediated evoked EPSCs and mEPSCs, larger AMPAR/NMDAR ratio, reduced inward rectification of AMPAR-mediated EPSCs, higher surface levels of GluA1 at synapses, and larger dendritic spines. Furthermore, manipulations that normally increase synaptic GluA1 surface levels and dendritic spine volume and reduce synaptic GluA1 surface levels are ineffective in Mecp2 KO hippocampal neurons. On the other hand, basal synaptic strength and LTP were not affected in slices from younger presymptomatic Mecp2 KO mice. Our findings suggest that LTP is occluded in Mecp2 KO hippocampal synapses due to altered levels of proteins involved in synaptic trafficking of GluA1-containing AMPARs.

Enhanced AMPAR-Mediated Transmission at Hippocampal Synapses.

The consequences of MeCP2 dysfunction on neuronal network function vary in different brain regions (35). Principal neurons in the somatosensory and the medial prefrontal cortices of Mecp2 KO mice have reduced activity caused by a selective impairment in excitatory synaptic transmission and by hypo-connectivity between excitatory neurons (11, 36, 37). The visual cortex of Mecp2 KO mice shows a similar reduction of pyramidal neuron and network activity but caused by stronger inhibition due to hyperinnervation of parvalbumin interneurons (38). Neuronal activity is elevated in the hippocampus (7, 8, 15, 39–41) and brainstem (42–44) of Mecp2 KO mice. Enhanced synaptic transmission and spatiotemporal spread of neuronal depolarizations in CA1 is due to higher spontaneous activity of CA3 pyramidal neurons in Mecp2 KO hippocampal slices (15). Because the intrinsic membrane properties of CA3 and CA1 pyramidal neurons as well as of four different classes of GABAergic interneurons in area CA3 were not affected in Mecp2 KO mice, the hyperactivity of the hippocampal network is due to an imbalance of synaptic inputs favoring excitation in CA3 (41). Here, we addressed the consequences of this excitatory/inhibitory (E/I) imbalance in area CA3 on the features and plasticity of downstream CA1 synapses due to their role in learning and memory.

Structural correlates of synaptic strength and AMPAR content at CA3→CA1 synapses are consistent with electrophysiological measures: the intensity of synaptic GluA1 puncta is higher, and the volume of dendritic spines is larger in Mecp2 KO neurons. The larger effect of the selective inhibitor NASPM on AMPAR-mediated EPSCs and their smaller inward rectification in Mecp2 KO neurons indicate a preponderance of GluA2-lacking receptors (17, 18), which is consistent with higher total and surface GluA1 levels in acute hippocampal slices and cultured neurons. In addition, smaller PPF and CV of EPSCs and higher frequency of mEPSC indicate enhanced presynaptic release probability in Mecp2 KO neurons, as described in Mecp2 KO cultured neurons (45).

Levels of glutamate receptor subunits in RTT individuals and MeCP2-deficient mice vary largely depending on brain region, age, and mouse genetic background. Autoradiography in RTT autopsy samples showed higher AMPAR levels in the superior frontal gyrus at young ages but lower at older ages (46); in contrast, AMPAR levels are typical in the caudate and the putamen of younger RTT cases (47). Protein levels of GluA1 and GluA2 are unaffected in hippocampal homogenates from two different Mecp2 mutant mouse lines kept in mixed genetic backgrounds (6), whereas GluA2 protein levels are lower in cultured cortical neurons of Mecp2 KO neurons (48). Here, a combination of electrophysiological, biochemical, and immunostaining approaches demonstrates that synaptic GluA1 levels are higher in Mecp2 KO neurons. Consistent with direct GluA1 transcriptional repression by MeCP2, chromatin immunoprecipitation revealed MeCP2 enrichment in the Gria1 promoter and overexpression of MeCP2 in the rat central amygdala reduced GluA1 levels, eliminating morphine seeking after withdrawal in a model of persistent pain (49).

Impaired LTP of EPSCs, Spatial Spread of Neuronal Depolarizations, and Spine Volume.

Neither TBS nor pairing low-frequency stimulation with postsynaptic depolarization were able to induce LTP at hippocampal CA1 synapses of Mecp2 KO mice, as shown previously with tetanic high-frequency stimulation in different MeCP2-deficient models (6–10). The potentiation of the amplitude and spatial spread of VSD signals observed after TBS in WT slices is absent in Mecp2 KO slices. In addition, forskolin-induced chem-LTP is impaired in Mecp2 KO slices. Lastly, the enduring enlargement of dendritic spines observed after LTP induction (27, 28) is absent in Mecp2 KO neurons.

Because AMPAR-mediated synaptic transmission is stronger in naïve hippocampal Mecp2 KO neurons, the parsimonious interpretation is that LTP is occluded because hippocampal synapses are saturated—that is, at the ceiling of their modification range. Neurons in the CNS are known to have a finite capacity to modulate synaptic efficacy, termed the synaptic modification range (50). Approaching the upper or lower limits of this range can occlude LTP or LTD, respectively, even though the range itself can also be modifiable (i.e., metaplasticity). Saturation of the synaptic modification range has serious behavioral consequences: hippocampal-dependent spatial learning is occluded following delivery of multiple tetani to the hippocampus in vivo (51), and repeated seizures saturate synaptic responses, preventing LTP and impairing memory function (52). Saturation of synaptic plasticity is expressed by synaptic insertion of GluA1 subunits after visual deprivation (53), knockout of the transcriptional repressor eukaryotic initiation factor 4E-binding protein-2 (54), or exposure to the general anesthetic isoflurane (55). An earlier study reported that repeated tetani failed to progressively increase LTP in hippocampal slices from Mecp2^stop mice but curiously did not find evidence of enhanced synaptic transmission in naïve slices and did not explore the cellular bases of the proposed LTP saturation (10). Two reports described steeper I–O curves of fEPSPs and impaired LTP at CA3→CA1 synapses in MeCP2-deficient mice, but they did not elaborate on the bases of that enhancement (7, 8). Here, we show that the A/N ratio, the I–O relationship, and rectification index of AMPAR-mediated EPSCs in naïve Mecp2 KO neurons resemble those of potentiated WT neurons and do not change after LTP induction as they normally do in WT neurons, which demonstrates that LTP is occluded by strong naïve synapses.

Impaired Activity-Dependent Insertion and Removal of AMPARs at Synapses.

The majority of AMPARs in the hippocampus and cortex are composed of GluA1/2 and GluA2/3 combinations (56, 57), and the regulated trafficking of GluA1 and GluA2 into and out of synapses has been well documented during Hebbian and homeostatic synaptic plasticity (21). We show that neither activity-dependent insertion nor removal of GluA1 to synapses occurs in Mecp2 KO neurons, which have a higher GluA1 content than WT synapses. Furthermore, Mecp2 KO neurons failed to show the increase in the synaptic insertion of the pH-sensitive SEP–GluA1 reporter after chemically induced LTP.

We propose a model in which higher GluA1 levels at hippocampal synapses result from a gradual accumulation and impaired removal after initial synapse formation, ultimately manifesting as enhanced synaptic transmission at symptomatic ages, which cause neurological symptoms such as seizures and impaired learning and memory. Consistent with this model, the amplitude and spatiotemporal spread of VSD signals, basal synaptic strength, and LTP in slices from younger presymptomatic Mecp2 KO mice were all comparable to those observed in age-matched controls (15) (Fig. S4). Collectively, our findings provide molecular, cellular, and network mechanisms underlying enhanced excitatory synaptic transmission and impaired long-term synaptic plasticity in the hippocampus of Mecp2 KO mice, identifying previously unidentified molecular targets for intervention in RTT and other MECP2-related disorders.

Materials and Methods

Ethical Approval and Experimental Animals.

All procedures were in accordance with the Office of Laboratory Animal Welfare of the NIH and were approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham (UAB). Experimental subjects were hemizygous Mecp2 KO male mice between postnatal day 20 and 22 (P20 and P22; i.e., presymptomatic) and between P45 and P65, when they exhibit RTT-like motor symptoms (hypoactivity, hind-limb clasping, reflex impairments); all controls were age-matched WT littermates.

Acute Hippocampal Slices.

Mice were anesthetized and perfused with ice-cold cutting artificial cerebrospinal fluid (aCSF), brains cut transversely at 300 µm using a vibratome, and slices transferred to normal aCSF at 32 °C for 30 min and then to room temperature for at least 1 h before recordings.

Hippocampal Neuron Cultures.

Primary hippocampal neurons were prepared from P1 mice. Neurons at 15–17 days in vitro (DIV) were transfected with EGFP or pCl–SEP–GluR1 plasmids using Lipofectamine 2000.

Electrophysiology.

Whole-cell recordings were performed alone or simultaneously with extracellular field recordings in hippocampal area CA1 or in 9–11 DIV pyramidal-shaped hippocampal neurons. mEPSCs were recorded at –60 mV in TTX and picrotoxin. For EPSC I–V curves, aCSF had picrotoxin and APV, and the patch solution contained spermine. CA1 SCs were stimulated using a theta glass pipette, with a stimulus intensity that evoked 30–40% maximal EPSCs. Before and after LTP induction, I–O curves of AMPAR- and NMDAR-mediated EPSCs were obtained at –60 mV and +40 mV, respectively. TBS consisted of four trains of 10 bursts, with each burst having five pulses at 100 Hz, 200 ms between bursts, and 5 s between trains. Pairing-induced LTP consisted of afferent stimulation (150 pulses at 3 Hz) during postsynaptic depolarization to 0 mV. Chem-LTP in slices was induced with forskolin and rolipram. Chem-LTP in cultures was induced with glycine, strychnine, and bicuculline in Mg²⁺-free aCSF; chem-LTD in cultures was induced with NMDA.

VSD Imaging.

Acute slices were stained with the fluorescent VSD RH-414. VSD signals and extracellular fEPSPs were evoked in CA1 SR by stimulation of SCs. RH-414 was excited at 535 ± 50 nm, and its emission at >594 nm was imaged at 2,500 frames per second with a scientific complementary metal-oxide semiconductor (CMOS) camera (128 × 128) using a 10× 0.5 N.A. objective. Fluorescence intensity during image sequences was normalized to resting light intensity (ΔF/F); spatial spread is the area showing ΔF/F levels 2× the baseline noise; duration is the interval between the first and last frames in a sequence that showed ΔF/F levels 2× the baseline noise.

Time-Lapse Cell Imaging of GluA1 Insertion.

SEP–GluA1-expressing neurons were imaged live in a confocal microscope before and after a 3-min exposure to glycine; images of dendritic segments with clear spines were acquired at 0.1-µm intervals in the z-plane.

Immunostaining.

Cryostat sections of perfusion-fixed brains were incubated with primary antibodies, followed by Alexa Fluor-594 or Alexa Fluor-488 secondary antibodies. Nonpermeabilized hippocampal cultures were incubated with anti-GluA1, followed by Alexa Fluor-594 secondary antibodies. After permeabilization, cultures were incubated with anti-GluA1 to detect total GluA1 levels and anti-VGLUT1 to label presynaptic terminals, followed with Alexa Fluor-488 and Alexa Fluor-647 secondary antibodies. Coverslips were imaged in a confocal microscope, and puncta were counted by unbiased ImageJ particle analysis.

Dendritic Spine Analyses.

CA1 pyramidal neurons in hippocampal slices were filled with biocytin during whole-cell recordings, fixed, and stained with streptavidin-conjugated Alexa Fluor-488. Primary hippocampal cultures were transfected with EGFP with Lipofectamine-2000, fixed, and immunostained with anti-GFP. The z-stack confocal images of dendritic spines were acquired, and Imaris software was used for dendritic reconstruction, semiautomated spine detection, and measurements of individual spines.

Western Immunoblotting.

Brain samples and postsynaptic density (PSD) fractions were homogenized in lysis buffer, and proteins were transferred to PVDF membranes and incubated with primary antibodies against GluA1, GluA1 phospho-Ser831, GluA1 phospho-Ser845, SAP97, or EEA1 followed by HRP-conjugated secondary antibodies. Proteins were detected using ECL substrate.

Statistics.

All data were analyzed using Prism (GraphPad). Comparisons were analyzed by unpaired Student’s t test, nonparametric Mann–Whitney test, Kolmogorov–Smirnov (K–S) test, or two-way ANOVA repeated measures.

SI Materials and Methods

Ethical Approval and Experimental Animals.

Mice were housed and handled and all experimental procedures were in accordance with the Office of Laboratory Animal Welfare of the National Institutes of Health (58). All experimental protocols were annually reviewed and approved by the Institutional Animal Care and Use Committee of UAB. Breeding pairs of mice lacking exon 3 of Mecp2 (59) were obtained from the Mutant Mouse Regional Resource Center (University of California, Davis) and maintained on a C57BL/6 background. A colony was established at UAB by mating WT C57BL/6 males with heterozygous Mecp2 mutant females. Experimental subjects were hemizygous Mecp2 KO male mice between postnatal day 20 and 22 (P20 and P22—i.e., presymptomatic) and between P45 and P65, when they exhibit RTT-like motor symptoms (hypoactivity, hind-limb clasping, reflex impairments); all controls were age-matched WT littermates.

Acute Hippocampal Slices.

Mice were deeply anesthetized with a mixture of 100 mg/kg ketamine and 10 mg/kg xylazine and transcardially perfused with ice-cold cutting aCSF containing (in mM) 87 NaCl, 2.5 KCl, 0.5 CaCl₂, 7 MgCl₂, 1.25 NaH₂PO₄, 25 NaHCO₃, 25 glucose, and 75 sucrose, bubbled with 95% O₂/5% CO₂. The brain was rapidly removed and cut transversely at 300 µm using a vibrating blade microtome (VT1200S, Leica Microsystems). Slices were transferred to normal aCSF containing (in mM) 119 NaCl, 2.5 KCl, 2.5 CaCl₂, 1.3 MgCl₂, 1.3 NaH₂PO4, 26 NaHCO₃, and 20 glucose at 32 °C for 30 min and then allowed to recover for 1 h at room temperature before recordings.

Hippocampal Neuron Cultures and Transfection.

Primary hippocampal neurons were prepared from P1 Mecp2 KO mice and their WT littermates. Both hippocampi were dissociated in papain (20 U/mL) plus DNase I (Worthington) for 20–30 min at 37 °C, as described (60). The tissue was then triturated to obtain a single-cell suspension, and neurons were seeded at a density of 50,000 cells/cm² on 18-mm coverslips coated with poly–d-lysine/laminin. Neurons were cultured with Neurobasal medium supplemented with B27 and 2 mM glutamine (Life Technologies), with half of the fresh medium changed every 3–4 d. For some experiments, neurons were transfected with pCl-SEP-GluR1 (Addgene, DIV 15–17) or EGFP plasmids (DIV 7–8) using Lipofectamine 2000 (Life Technologies) (1.6 µg DNA), as described (60).

Electrophysiology.

Individual slices were transferred to a submerged chamber mounted on a fixed-stage upright microscope (Zeiss Axioskop FS) and continuously perfused at room temperature with normal oxygenated aCSF containing (in mM) 119 NaCl, 2.5 KCl, 2.5 CaCl₂, 1.3 MgCl₂, 1.3 NaH₂PO4, 26 NaHCO₃, and 20 glucose. Pyramidal neurons in CA1 SP were visualized by infrared differential interference contrast microscopy with a water-immersion 63× objective (0.9 N.A., Zeiss). Individual coverslips with cultured neurons were transferred to a submerged chamber mounted on a fixed-stage upright microscope (Leica-DMLFS), and pyramidal-shaped neurons were visualized with a water-immersion 63× objective (1.0 N.A., Zeiss). Whole-cell voltage-clamp recordings were performed with unpolished pipettes containing (in mM) 120 Cs-gluconate, 17.5 CsCl, 10 Na-Hepes, 4 Mg-ATP, 0.4 Na-GTP, 10 Na₂-creatine phosphate, 0.2 Na-EGTA, and 290–300 mOsm, pH 7.3 (final resistance, 3–4 MΩ). Membrane currents were acquired with Axopatch-200B amplifiers (Molecular Devices), filtered at 2 kHz, and digitized at 10 kHz with ITC-18 A/D–D/A interfaces (Instrutech) controlled by custom-written software in G5 PowerMac or Intel MacPro computers (TI-WorkBench, provided by Dr. Takafumi Inoue, Waseda University, Tokyo). Input resistance was measured with hyperpolarizing voltage pulses (50 ms, 20 mV). Cells with series resistances above 15 MΩ were discarded, and cells were also excluded if any whole-cell parameter (i.e., Cm, Ri, Rs) changed by ≥20% during the recordings. Extracellular field recordings in slices were performed alone or simultaneously with whole-cell recordings and were acquired with an Axoclamp-2A amplifier (Molecular Devices) in Bridge current-clamp mode and Model 210 preamplifier (Brownlee Precision). fEPSPs were recorded from CA1 SR with a glass pipette filled with normal aCSF (final resistance, 2–3 MΩ).

mEPSCs were recorded with membrane voltage held at –60 mV in the presence of the voltage-gated sodium channel blocker TTX (1 µM) and the GABA_AR antagonist picrotoxin (50 µM). mEPSCs were analyzed using the MiniAnalysis program (Synaptosoft), with a detection threshold of 8 pA. The kinetics properties of mEPSCs analyzed included rise time, 10–90% rise time, decay, half-width, and charge transfer. Rise time and 10–90% rise time were defined as the time from the baseline to the peak and from the corresponding percentage, respectively. Half-width was defined as the time between the rising and decay phases of mEPSCs at 50% of the peak amplitude. Charge transfer was determined by integrating the area of mEPSC waveform from baseline. To obtain AMPAR I–V curves, brain slices were perfused with aCSF containing picrotoxin (50 µM) and the NMDAR antagonist APV (100 µM), and the internal solution was supplemented with spermine (100 µM). The rectification index was determined as the ratio of the amplitude of EPSCs evoked at +40 mV to the amplitude of those evoked at –60 mV. NASPM (50 µM) was used to selectively inhibit GluA2-lacking AMPARs. To characterize presynaptic features, the paired-pulse ratio (PPR) of fEPSPs and EPSCs was determined as the ratio of the second amplitude to the first one in response to two consecutive stimuli at 50-ms intervals. AMPA-induced currents were obtained by bath application of S-AMPA (250 nM) for 10 min in the presence of TTX (1 µM). After complete recovery to baseline, a second AMPA application was performed. In some experiments, the AMPAR-selective antagonist CNQX (20 µM) was included in the aCSF before the second AMPA application. All-point amplitude histograms with Gaussian fits were generated using the Noise Analysis Function of the MiniAnalysis program.

Electrical LTP.

CA1 SCs were stimulated using an aCSF-filled theta glass patch pipette connected to an isolated stimulator (ISO-Flex, AMPI). The stimulus intensity was adjusted to elicit 30–40% of maximal postsynaptic responses. A stable baseline of 5–10 min was recorded before a train of stimuli was given. LTP was induced with either TBS or pairing low-frequency presynaptic stimulation with postsynaptic depolarization. TBS consisted of four trains of 10 bursts, each burst having five pulses at 100 Hz, with 200 ms between bursts and 5 s between trains; TBS was delivered under current clamp. The pairing protocol consisted of 150 pulses of afferent presynaptic stimulation at 3 Hz during a 50-s step depolarization of the whole-cell electrode to 0 mV.

Before and following LTP recording, I–O curves of AMPAR- and NMDAR-mediated EPSCs were performed at –60 mV and +40 mV, respectively, and the AMPAR/NMDAR ratios were calculated. The AMPAR/NMDAR ratios were calculated as the ratio of the amplitude of the AMPAR EPSC peak current to that of the NMDAR 100 ms after the peak current. Application of the NMDAR antagonist APV (100 µM) and the GABA_AR antagonist picrotoxin (50 µM) blocked slower outward currents without apparent influence on faster inward currents; addition of the AMPAR antagonist CNQX (20 µM) completely eliminated faster inward and outward currents. Changes in presynaptic function are usually accompanied by changes in the CV (CV²) of EPSC amplitudes (61). The inverse of CV² of AMPAR EPSC amplitudes was compared between genotypes as well as 5 min before and 30 min after LTP induction.

Chem-LTP and LTD.

Glycine-induced chem-LTP (22, 62) and NMDA-induced chem-LTD (24, 25) were induced in cultured hippocampal neurons as described previously. Briefly, neurons were maintained for 5 min in normal extracellular solution containing (in mM) 5 Hepes (pH 7.3), 125 NaCl, 2 CaCl₂, 2.5 KCl, 1 MgCl₂, and 33 glucose. Chem-LTP was induced by changing the extracellular solution for 3 min to a Mg²⁺-free solution containing 200 µM glycine, 20 µM bicuculline, and 3 µM strychnine (bubbled with 95% O₂/5% CO₂ and perfused at 2 mL/min). Chem-LTD was induced by changing the extracellular solution for 3 min to a solution containing 20 µM NMDA (bubbled with 95% O₂/5% CO₂ and perfused at 2 mL/min). Forskolin-induced chem-LTP in acute slices was induced by exposing acute slices to aCSF containing the adenylate cyclase activator forskolin (50 µM) and the phosphodiesterase-4 inhibitor rolipram (0.1 µM) for 10 min (30); slices were then maintained in normal aCSF for 1 h. When testing the role of basal synaptic activity in forskolin-induced chem-LTP, afferent stimulation to SCs was given every 6 s (30). At the end of the experiments, brain slices were dissected and the hippocampi were snap frozen for subsequent homogenization for Western immunoblotting.

VSD Imaging.

After recovery, individual acute slices were stained with the voltage-sensitive fluorescent dye N-(3 triethylammoniumpropyl)-4-{4-[4-(diethylamino)phenyl]butadienyl}pyridinium dibromide (RH-414; 30 µM) for 50 min at room temperature in normal aCSF (bubbled with 95% O₂–5% CO₂). Slices were transferred to the submerged chamber and washed for at least 20 min to eliminate excess dye before recordings at 32 °C. RH-414 was excited at 535 ± 50 nm with a phosphor-pumped LED (89 North Heliophor), and its emission at >594 nm was imaged at 2,500 frames per second with a scientific CMOS camera (128 × 128; NeuroCMOS-SM128, RedShirt Imaging) using a 10× 0.5 N.A. objective (Zeiss). VSD signals and extracellular fEPSPs were evoked in CA1 SR by stimulation of SCs. For I–O of optical signals and fEPSPs, four different stimulus intensities with 30-µA increments were delivered through a theta glass pipette filled with aCSF and connected to an isolated stimulator (ISO-Flex, AMPI). TBS–LTP was induced with stimulus intensities set at 30–40% of maximal responses; TBS consisted of four trains of 10 bursts, each burst having five pulses at 100 Hz, with 200 ms between bursts and 5 s between trains. Fluorescence intensity during image sequences was normalized to resting light intensity (ΔF/F); RH-414 bleaching was corrected by exponential fit subtraction. Membrane depolarization causes a decrease in RH-414 fluorescence intensity, but for illustration purposes and following an established convention (63), we inverted the RH-414 ΔF/F ratios and used a rainbow pseudocolor scale (warmer colors correspond to larger depolarization). Regions of interest (3 × 3 pixels) were used to obtain the peak amplitude of ΔF/F signals. Spatial spread of optical signals was obtained by measuring the area showing ΔF/F levels twice the baseline noise. Duration of optical signals was calculated from the interval between the first and the last frames in a sequence that showed ΔF/F levels twice the baseline noise.

Time-Lapse Cell Imaging.

Twenty-four hours following transfection, coverslips were mounted for live imaging in a chamber perfused with normal extracellular solution. After the first image was taken, the solution was replaced with the LTP induction medium. Following glycine application for 3 min, the solution was switched back to normal extracellular solution, and the images were taken immediately, as well as 15 and 30 min later. Confocal images of SEP–GluA1-transfected spines were acquired in a laser-scanning confocal microscope (Fluoview FV300, Olympus) using an oil immersion 60× objective (1.45 N.A.), with additional digital zoom. Z-stack images with 0.1-µm intervals were obtained for dendritic segments and collapsed as maximum intensity projection for subsequent analysis with Imaris software.

Immunostaining.

For GluA1 and MeCP2 immunostaining, unfixed brains were dissected and sliced into 100-µm sections using a vibrotome. Tissue sections were fixed in 4% (wt/vol) paraformaldehyde in PBS overnight at 4 °C. Sections were cryprotected with 30% (wt/vol) sucrose overnight, permeabilized with 0.25% Triton X-100 for 2 h, and blocked with 10% (vol/vol) normal goat serum for 1 h. Sections were incubated with polyclonal rabbit anti-GluA1 (AB1504, Millipore) and/or mouse anti-MeCP2 (Sigma-Aldrich) primary antibodies, followed by incubation with corresponding anti-rabbit Alexa Fluor-594 and/or anti-mouse Alexa Fluor-488 secondary antibodies, respectively. Confocal microscopy was performed with a Fluoview FV300 microscope (Olympus) equipped with argon (488 nm) and HeNe green (543nm) lasers for excitation. The z-stack images were acquired with the same settings of laser power and photomultiplier tube (PMT) voltage and gain for semiquantitative comparison using ImageJ (NIH). Images with background subtracted were threshold for measurement of fluorescent intensity in different layers of hippocampal CA1, CA3, or dentate gyrus (DG). The number of cells or puncta in these regions was counted by unbiased ImageJ particle analysis with defined puncta size range.

For immunostaining in neuronal cultures, 48 h after GFP transfection, neurons were fixed with 4% paraformaldehyde/sucrose in PBS for 10 min and permeablized with 0.25% Triton X-100 for 15 min. After blocking with 10% goat serum in PBS, cells were incubated with primary antibodies GluA1 (Millipore) and GFP (Abcam) overnight at 4 °C, rinsed in PBS, and incubated for 1 h at room temperature with secondary antibodies conjugated to Alexa 594 and Alexa 488 (Jackson ImmunoResearch), respectively. Coverslips were then mounted with Vectashield (Vector Laboratories) and imaged in a confocal microscope (Zeiss LSM510) using a 63 × 1.4 N.A. oil immersion objective. GluA1 intensity was quantified by measuring the integrated intensity of colocalized puncta using ImageJ (NIH).

To study GluA1 trafficking during chem-LTP or chem-LTD, hippocampal neurons after treatment with glycine or NMDA were blocked in 10% goat serum for 10 min at 37 °C in an incubation chamber and incubated with anti-mouse GluA1 (N terminus, MAB2263, Millipore) for 30 min at 37 °C under nonpermeabilizing conditions. Neurons were then fixed in 4% paraformaldehyde at room temperature and washed in PBS three times for 5 min each. Secondary anti-mouse Alexa Fluor-594 antibody was used to detect the primary antibody, followed by washing in PBS. Neurons were then permeabilized for 15 min in ice-cold 0.1% Triton X-100 in PBS and blocked again in 10% goat serum at room temperature. After blocking for 30 min, neurons were incubated overnight at 4 °C with primary anti-GluA1 antibody to detect total GluA1 levels and anti-guinea pig VGLUT1 (AB5905, Millipore) antibody to label presynaptic terminals. After three washes, neurons were incubated with Alexa Fluor-488 and anti-guinea pig Alexa Fluor-647 antibody. Slides were covered with VectaShield (Vector), coverslipped, and stored at 4 °C. Confocal microscopy was performed with a Zeiss 510-META microscope equipped with a 63× oil-immersion objective (1.45 N.A.), an argon laser (488 nm), an HeNe green laser (543 nm), and an HeNe red laser (633 nm). Image acquisition and analysis were performed as described above. The ratio of surface/total GluA1 puncta intensity was used to determine the insertion or endocytosis of GluA1 AMPARs from synaptic sites during chem-LTP and chem-LTD, respectively.

Dendritic Spine Analyses.

Pyramidal neurons in the CA1 of hippocampal slices were filled with biocytin (8 mM, Sigma-Aldrich) during whole-cell recordings and stained with streptavidin-conjugated Alexa Fluor-488 (Life Technologies) after fixation. For cultured hippocampal neurons, dendritic spines were visualized after immunostaining with anti-GFP antibody 48 h after transfection. The z-stack images of dendritic spines were acquired using a Fluoview FV300 microscope (Olympus). Imaris 7.6.5 software (Bitplane) was used for dendritic spine analyses in 3D z-projection stacks, as described previously (64). Briefly, regions of interest (ROIs) encompassing spiny dendrites were selected for creating a filament using the Autopath mode. Automatic thresholds were used for assigning the starting and ending points of each dendritic segment as well as for building surface renderings. For semiautomated detection of dendritic spines, maximum spine length and minimum spine end diameter were set at 5 µm and 0.215 µm, respectively, to generate spine seeds. Spine seeds were carefully examined for their localization, and those that were not closely associated with dendrites were discarded from the dendritic surface rendering. Spine parameters including density and volume were automatically calculated. The average results of spine volume were calculated based on the dendritic segments instead of individual spines. Maximum-intensity projections of z-stacks were only used for display purposes.

RT-qPCR and Western Immunoblots.

For whole hippocampus real-time RT-PCR, total RNA was extracted using RNeasy Plus Mini Kit (Qiagen). RNA concentrations were determined in a NanoDrop Spectrophotometer (Thermo Scientific). mRNAs were reverse-transcribed using the iScript cDNA Synthesis Kit (Bio-Rad). PCR amplifications were performed using iQ SYBR Green Supermix (Bio-Rad) at 95 °C for 3min, followed by 50 cycles of 95 °C for 10 s, 60 °C for 30 s, 72 °C for 45 s, and incubation at 70 °C for 10 min. The following primers were used: SAP97, 5ʹ-ATGCTTCTGACGACGAGTG-3′ (forward), 5ʹ- GACCACGGTAACTACTTTCA-3ʹ (reverse); EEA1, 5ʹ-CACGTCCTCCCAAGATAGC-3ʹ (forward), 5ʹ-TGGGCATGTACTCTAGACTC-3′ (reverse); and GADPH, 5′-AAGGGCTCATGACCACAGTC-3′ (forward), 5′-ACACATTGGGGGTAGGAACA′ (reverse).

For whole hippocampus Western immunoblots, brain samples were dissected and homogenized in Nonidet P-40 buffer (20 mM Tris, pH 8.0, 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, 2 mM EDTA) containing protease and phosphatase inhibitors. The homogenates were maintained with constant agitation for 2 h at 4 °C and centrifuged at 10,000 × g for 20 min. The supernatants were aspirated and protein concentrations determined by Lowry method. Equal amounts of protein sample were denatured in loading buffer (125 mM Tris at pH 6.8, 20% glycerol, 6% SDS, and 5% 2-mercaptoethanol), boiled for 3 min, and subjected to SDS/PAGE. Proteins were transferred to PVDF membrane and blocked with 5% nonfat milk in Tris-buffered saline with Tween 20 (TBST) (20 mM Tris, pH 7.6, 150 mM NaCl, and 0.1% Tween-20) for 1 h. Membranes were incubated with primary antibodies against MeCP2 (07-013, Millipore; M7443, Sigma-Aldrich), GluA1 (N terminus, MAB2263, Millipore), GluA1 (cytoplasmic domain, AB1504, Millipore), GluA1 phospho-Ser831 (AB5847, Millipore), GluA1 phospho-Ser845 (04-1073, Millipore), SAP97 (75-030, Antibodies Incorporated), or EEA1 (Cell Signaling), followed by corresponding HRP-conjugated secondary antibodies (Santa Cruz). The proteins were detected using the Pierce ECL Substrate (Thermo Fisher Scientific), and signals were captured on autoradiography film and quantified by computer-assisted densitometry. Membranes were reprobed for the loading control with β-actin (MA5-15739, Thermo Fisher Scientific) or GADPH (Cell Signaling) and detected using the Odyssey infrared imaging system (Li-Cor Bioscience).

PSD Fractionation.

Hippocampal tissues pooled from two mice were homogenized in ice-cold lysate buffer (0.32 M sucrose, 10 mM Hepes, pH 7.4), containing protease inhibitors. The homogenate was centrifuged at 900 × g for 15 min, and the supernatant was centrifuged at 18,000 × g for 15 min to yield crude synaptosomal pellet (P2). The P2 pellet was resuspended in Hepes-buffered sucrose and centrifuged at 18,000 × g for 15 min to obtain washed crude synaptosomal fraction. The pellet was resuspended in hypo-osmotic Hepes buffer (4 mM Hepes, pH 7.4) and incubated for 2 h at 4 °C. The lysate was centrifuged at 25,000 × g for 20 min to yield the synaptosomal membrane fraction. The pellet was resuspended in Hepes-buffered sucrose, layered over a discontinuous sucrose gradient (0.8 M, 1 M, 1.2 M), and centrifuged at 150,000 × g for 2 h. Synaptic plasma membrane was recovered between the 1.0 M and 1.2 M layers, diluted in Hepes-buffered sucrose, and centrifuged at 150,000 × g for 30 min. The pellet was resuspended in Hepes/EDTA buffer (50 mM Hepes, 2 mM EDTA, pH 7.4), followed by addition of 0.5% Triton X-100, and incubated for 15 min. The suspension was centrifuged at 200,000 × g for 20 min to obtain PSD fractions, and the PSDs were resuspended in Hepes/EDTA buffer containing 0.5% SDS. Western immunoblots were performed as described above. Primary antibodies against PSD-95 (36233, Cell Signaling) and GluA1 (MAB2263, Millipore) were used.

Statistics.

All data were analyzed using Prism (GraphPad). Comparisons between groups were analyzed by two-tailed unpaired Student’s t test or nonparametric Mann–Whitney test. Comparisons for cumulative probability distributions were analyzed with K–S test. Two-way ANOVA repeated measures with post hoc analysis were used for data of LTP, EPSC I–O, GluA1 immunostaining, and SEP–GluA1 puncta density and intensity. All data are shown as the mean ± SEM. Statistical differences were considered significant at P < 0.05.