Methyl-CpG binding protein 2 (MeCP2) is highly expressed in neurons in the vertebrate brain, and mutations of the gene encoding MeCP2 cause the neurodevelopmental disorder Rett syndrome. This study examines the role of MeCP2 in the development and function of thalamic GABAergic circuits. Whole cell recordings were carried out in excitatory neurons of the ventrobasal complex (VB) of the thalamus and in inhibitory neurons of the reticular thalamic nucleus (RTN) in acute brain slices from mice aged P6 through P23. At P14–P16, the number of quantal GABAergic events was decreased in VB neurons but increased in RTN neurons of Mecp2-null mice, without any change in the amplitude or kinetics of quantal events. There was no difference between mutant and wild-type mice in paired-pulse ratios of evoked GABAergic responses in the VB or the RTN. On the other hand, unitary responses evoked by minimal stimulation were decreased in the VB but increased in the RTN of mutants. Similar changes in the frequency of quantal events were observed at P21–P23 in both the VB and RTN. At P6, however, quantal GABAergic transmission was altered only in the VB not the RTN. Immunostaining of vesicular GABA transporter showed opposite changes in the number of GABAergic synaptic terminals in the VB and RTN of Mecp2-null mice at P18–P20. The loss of MeCP2 had no significant effect on intrinsic properties of RTN neurons recorded at P15–P17. Our findings suggest that MeCP2 differentially regulates the development of GABAergic synapses in excitatory and inhibitory neurons in the thalamus.
Rett syndrome (RTT) is a severe neurological disorder causing progressive loss of motor and language skills in girls, with disease onset taking place between 6 and 18 mo of age (Chahrour and Zoghbi 2007). Mutations in the X-linked gene encoding methyl-CpG binding protein 2 (MeCP2) are the primary causes of RTT (Amir et al. 1999). Mice with genetically altered Mecp2 provide powerful tools for investigation of RTT pathophysiology. Mecp2-null mice and those with truncated Mecp2 recapitulate many features of RTT including reduced brain size, reduced mobility, ataxic gait, seizures, irregular breathing, and cognitive impairment (Chen et al. 2001; Guy et al. 2001; Shahbazian et al. 2002). Mecp2-null male mice appear normal during the first 3 wk, but develop a mild ataxic gait and reduced spontaneous movement between 4 and 8 wk of age; they usually die before 10 wk of age (Chen et al. 2001; Guy et al. 2001). Although Mecp2 is widely expressed in the body, selective deletion of Mecp2 in the CNS using CamK-Cre transgenic mice results in a phenotype similar to that of Mecp2-null mice (Chen et al. 2001), indicating that MeCP2 deficiency in the CNS is the primary cause of RTT.
In RTT patients and Mecp2 mutant mice, brain cytoarchitecture appears normal, without detectable loss of neurons (Armstrong 2002; Chen et al. 2001), suggesting that changes in neuronal and synaptic functions are the primary causes of the neurological phenotype in RTT. Several recent studies have investigated the role of MeCP2 in excitatory synaptic transmission in the brain. Quantal excitatory transmission was significantly reduced in neocortical pyramidal neurons and hippocampal neurons of Mecp2-null mice (Chao et al. 2007; Dani et al. 2005; Nelson et al. 2006). Overexpression of Mecp2 increases quantal excitatory transmission in hippocampal neurons through an up-regulation of synapse number (Chao et al. 2007). MeCP2 deficiency also leads to reductions of long-term synaptic plasticity in the neocortex and hippocampus (Asaka et al. 2006; Moretti et al. 2006), and this defect can be rescued by restoring Mecp2 expression in mice (Guy et al. 2007). Together, these studies indicate that MeCP2 regulates the number and function of excitatory synapses in the brain.
MeCP2 is also abundant in inhibitory neurons in the brain of mice and primates (Akbarian et al. 2001). Recent studies have shown that MeCP2 deficiency resulted in an enhancement of GABAergic transmission in the neocortex (Dani et al. 2005) but a reduction of GABAergic transmission in the brain stem (Medrihan et al. 2008). The discrepancy between these findings indicates that the role of MeCP2 in GABAergic function is likely to be complex. In this study, we examined GABAergic circuits in the thalamus of Mecp2-null mice during early life. Our results showed that GABAergic transmission is altered in opposite directions in excitatory and inhibitory neurons in the thalamus of Mecp2-null mice. These findings suggest that MeCP2 differentially regulates GABAergic transmission in excitatory and inhibitory neurons.
Mecp2-null mice [B6.129P2(C)-Mecp2tm1.1Bird/J] (Guy et al. 2001) were obtained from The Jackson Laboratory's Mouse Repository (Stock 3890). These mice lack exons 3 and 4 of Mecp2, causing deletion of all but the amino-terminal eight amino acids of MeCP2. This line has been backcrossed to C57BL/6J for ≥15 generations and is maintained by mating heterozygote females with C57BL/6J males. Genotyping was done by PCR using four primers: 5′-GGTAAAGACCCATGTGACCC-3′ (exon 1 forward primer), 5′-TCCACCTAGCCTGCCTGTAC-3′ (reverse primer for the mutated allele), 5′-GCAAGCATGAGCCACTACAA-3′ (exon 3 forward primer), and 5′-GCAAGGTGGGGTCATCATAC-3′ (exon 3 reverse primer), which respectively amplify 400- and 180-bp fragments from the mutated and wild-type (WT) alleles. Only male mutants (Mecp2−/y) and WT littermates (Mecp2+/y) were used in this study. All procedures are in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and have been approved by The Jackson Laboratory Animal Care and Use Committee.
We used a horizontal slice preparation that allows easy identification of the reticular thalamic nucleus (RTN) and ventral basal complex (VB) and also preserves the inhibitory connection from RTN neurons to VB neurons (Cox et al. 1997). Slices were prepared using methods that we described previously (Arsenault and Zhang 2006). Briefly, male mice between the ages of postnatal day 6 (P6, with the day of birth as P0) and P23 were anesthetized with tribromoethanol (250 mg/kg, ip) and decapitated. Brains were quickly removed and transferred into ice-cold solution containing (in mM) 210 sucrose, 3.0 KCl, 1.0 CaCl2, 3.0 MgSO4, 1.0 NaH2PO4, 26 NaHCO3, and 10 glucose, saturated with 95% O2-5% CO2. Horizontal slices were cut at 300 μm on a vibratome (VT 1000s, Leica) and kept in artificial cerebral spinal fluid (ACSF) containing (in mM) 124 NaCl, 3.0 KCl, 1.5 CaCl2, 1.3 MgSO4, 1.0 NaH2PO4, 26 NaHCO3, and 20 glucose, saturated with 95% O2-5% CO2 at room temperature (21–23°C). Slices were allowed to recover for ≥1 h before any recording.
Recordings were made at 32–34°C using whole cell patch-clamp techniques. Each slice was transferred to a submerge-type chamber where it was continuously exposed to ACSF heated to 32–34°C, saturated with 95% O2-5% CO2, and flowing at rate of 2.0 ± 0.2 ml/min. Slices were first viewed with a ×4 objective using a fixed-stage upright microscope (FN1, Nikon). Under polarized light illumination, the RTN and VB can be easily identified in horizontal slices; those sections containing both the RTN and VB were chosen for recording. Neurons in the RTN and VB were viewed under near infrared illumination with a ×40 water-immersion objective (×40/0.80 W, Nikon), Nomarski optics, and an infrared CCD camera. Patch pipettes were pulled from thick wall borosilicate glass (1.5/0.84 mm, WPI) on a horizontal puller (P-97, Sutter Instruments). Unless indicated otherwise, the pipette solution contained (in mM) 130 CsCl, 4 ATP-Mg, 0.3 GTP-Na, 0.5 EGTA, and 20 HEPES (pH 7.3, 270–280 mOsm with sucrose). Resistance of electrodes was between 2 and 4 MΩ. Liquid junction potential, usually <2 mV, was not corrected. Seal resistance was >2 GΩ. Whole cell recordings were made at the soma with a Multiclamp 700B amplifier (Molecular Devices). The series resistance (Rs), usually between 8 and 16 MΩ, was monitored throughout the recording, and data were discarded when Rs varied by 20% or more over the course of the recording. GABAergic inhibitory postsynaptic currents (IPSCs) were selectively recorded by blocking ionotropic glutamate receptors with 20 μM 6,7-dinitro-quinoxaline-2,3-dione (DNQX) and 1 mM kynurenic acid (KN). DNQX was obtained from Tocris; all other chemicals were obtained from Sigma-Aldrich.
To record evoked IPSCs in VB neurons, a concentric bipolar electrode (100 μm inner pole diameter, FHC) was placed in the RTN to deliver current pulses (50–500 μA, 100 μs in duration). Twisted nichrome microwires (38 μm in diameter with coating, A-M Systems) were used to evoke IPSCs in RTN neurons. The electrodes were placed ≥200 μm away from the recorded RTN neurons.
The intrinsic properties and firing patterns of RTN neurons were examined in current-clamp configuration with the pipette solution containing (in mM) 120 K-methylsulfate, 15 KCl, 4 ATP-Mg, 0.3 GTP-Na, 0.1 EGTA, and 10 HEPES (pH 7.3 with KOH, 280–285 mOsm with sucrose). The series resistance was fully compensated using the bridge circuit of the amplifier. Liquid junction potential, estimated to be +8 mV, was corrected post hoc.
Experiments were conducted using the AxoGraph X program (AxoGraph Scientific) with a PowerMac G5 connected to an ITC-18 interface. Data were filtered at 4 kHz and digitized at 16 kHz. Data were analyzed using AxoGraph X and IgorPro (WaveMetrics). Quantal IPSCs were detected using variable amplitude template functions with the rise time set at 1 ms and decay times set at 5 ms for VB neurons and 15–20 ms for RTN neurons. The template was slid along the data trace, one point at a time, and was scaled and offset to fit the data using a method that minimizes the sum of squared errors between the fitted template and the data region. The detection threshold was set at four times the SD of baseline noise. Our test results showed that a onefold change in the decay time of the detection template in either direction has little impact on event detection: neither the number nor the properties of the detected events were significantly altered. At least 100 isolated events for each cell were aligned and averaged to give the mean response. For VB neurons, the decay phase of the mean response was best fitted by a single exponential function. Some RTN neurons required the double exponential function: I = A1e–t/t1 + A2e–t/t2, where A1 and A2 are the maximum amplitudes of the fast and slow components, and t1 and t2 are the decay time constants of the two components. The weighted decay time constant tw was calculated as tw = (A1t1 + A2t2)/(A1 + A2).
For recordings using minimal stimulation protocol, the amplitude of IPSC was estimated as the difference between the average of a 1-ms segment at the peak of IPSCs and that just before stimulation (baseline). Failures were scored when the difference was <4 times the SD of the baseline. Episodes where spontaneous synaptic events occur around the peak of evoked IPSCs were excluded from analysis. CV was calculated with 30 consecutive IPSCs.
Intrinsic properties of RTN neurons were analyzed using methods that we described previously (Zhang 2004; Zhang and Arsenault 2005). Briefly, resting potentials of RTN neurons were measured within 20 s of break-in. Input resistance (RN) and membrane time constant (tm) were estimated from voltage responses to 400 ms current pulses at −100 pA. Cell membrane capacitance (Cm) was calculated as Cm = tm/RN. Action potentials were evoked using positive current pulses. Action potential threshold was estimated as the point when the slope of rising membrane potential exceeds 50 mV/ms, and each data point is the mean value of five action potentials from a single neuron.
Recordings of quantal IPSCs at P6 and P14–P16 and of minimal stimulation were done with the experimenter blind to the genotype. Recordings of quantal IPSCs at P21–P23, intrinsic properties of RTN neurons, and some of the recordings of pair-pulse responses were not performed blind.
Mice were anesthetized with tribromoethanol, perfused with 4% paraformaldehyde in 0.1 M PBS, and postfixed overnight. Coronal sections 50 μm thick were cut on a vibratome. Sections corresponding to the coronal plane at 1.7 mm behind the bregma (Paxinos and Franklin 2001) were incubated with 1% BSA, 0.3% Triton X-100, and 3% normal goat serum for 2 h at room temperature. Sections were incubated with polyclonal antivesicular GABA transporter (VGAT; 1:2,000, Synaptic Systems), a monoclonal anti-NeuN (neuronal nuclear antigen A60) (1:1,000, Chemicon), or a monoclonal anti-GAD67 (1:1,500, Chemicon), 1% BSA, 0.3% Triton X-100, and 3% normal goat serum for 48 h at 4°C. After wash, sections were incubated with goat Alexa-fluor–conjugated secondary antibodies (1:250, Invitrogen) for 2 h at room temperature. Sections were mounted, dehydrated, and coverslipped in DPX medium (Sigma).
Imaging and image analysis were conducted with the experimenter blind to the genotype. Confocal images (1,024 × 1,024 pixels) were captured with a Leica SP5 microscope using a ×63 objective (1.4 NA) and ×2.0 optic zoom. For each pair of mice (1 mutant with 1 WT littermate), the exact same laser and microscope settings were used for all the samples of either genotype. Two adjacent sections from each mouse were analyzed. For each section, three images were taken at 5 μm below the surface from the middle portion of the VB or RTN. Confocal images were analyzed with ImageJ. VGAT-positive puncta were outlined by adjusting image contrast and threshold; the number of puncta was measured automatically using particle analysis. Because cell bodies of RTN neurons were often VGAT positive, we excluded structures with sizes >4 μm. For each mouse, the results from six images were averaged. To estimate the density of RTN and VB neurons, four images were taken from the middle portion of the RTN or VB from two sections (2 images per section). The number of NeuN-positive cells was counted manually for each image. The results from the four images were averaged.
All data were presented as means ± SE. Unless indicated otherwise, statistical differences were compared using Mann-Whitney test.
Opposite changes in spontaneous GABAergic transmission in VB and RTN neurons of Mecp2-null mice
In addition to motor and cognitive deficits, RTT patients have significant sensory impairment that affects auditory, visual, and somatosensory systems (Glaze 2005; Nomura 2001). For example, RTT patients showed abnormally large somatosensory responses (Kimura et al. 1992; Yoshikawa et al. 1991). As the gateway to the neocortex, the thalamus plays a key role in sensory functions and sensorimotor integration. Although alternations in synaptic transmission have been reported in the somatosensory cortex of Mecp2-null mice (Dani et al. 2005), nothing is known about synaptic defect in the thalamus.
We focused on the inhibitory circuits in the RTN and VB of the thalamus. GABAergic neurons in the RTN provide the main inhibitory input to excitatory somatosensory relay neurons in the VB (Cox et al. 1996, 1997; Ohara and Lieberman 1985; Pinault et al. 1995). Neurons in the RTN are also interconnected via GABAergic synapses (Cox et al. 1996; Pinault et al. 1997; Shu and McCormick 2002). We decided to examine mice at 2 wk of age because it is before the onset of the symptoms in Mecp2-null mice (Chen et al. 2001; Guy et al. 2001), yet at this age, GABAergic synapses in the thalamus are close to the mature form (Huntsman and Huguenard 2000). Spontaneous synaptic currents were recorded in VB and RTN neurons from P14–P16 mice at the holding potential of −70 mV in the presence of DNQX (20 μM) and KN (1 mM); all spontaneous events were abolished by the selective GABAA receptor antagonist SR95531 (5 μM; Fig. 1, B and C), indicating that they were GABAergic synaptic events mediated by GABAA receptors.
Quantal GABAergic events [miniature IPSCs (mIPSCs)] were selectively recorded in the presence of TTX (0.4 μM), DNQX (20 μM), and KN (1 mM) in Mecp2-null and WT mice aged P14–P16. TTX at 0.4 μM completely blocked voltage-dependent Na+ spikes in both VB and RTN neurons (data not shown). Mutant VB neurons showed significantly fewer mIPSCs than those in aged-matched WT mice (Fig. 2A). We recorded mIPSCs in 21 VB neurons from six mutant mice and 20 VB neurons from seven WT mice. Cumulative distributions of interevent interval and peak amplitude were obtained using 200 consecutive events from each of 20 WT cells and 21 mutant cells. The distribution of interevent interval shifted to the right in mutant VB neurons (Fig. 2B), whereas that of peak amplitude was unchanged (Fig. 2C). The mean frequency of mIPSCs was reduced by 39% in Mecp2-null mice (Fig. 2D; 5.7 ± 0.7 Hz for WT, 3.5 ± 0.3 Hz for mutant; P < 0.01). On the other hand, there was no difference between mutant and WT mice in the peak amplitude (Fig. 2E; 23.6 ± 2.1 pA for WT, 22.3 ± 1.6 pA for mutant; P > 0.5), the rise time (0.78 ± 0.04 ms for WT, 0.81 ± 0.05 ms for mutant; P > 0.3), or the decay time constant (4.4 ± 0.1 ms for WT, 4.4 ± 0.1 ms for mutant; P > 0.5) of mIPSCs. These results indicate that, at P14–P16, the reduction of GABAergic transmission in VB neurons of Mecp2-null mice is caused by changes in either presynaptic function or the number of synapses. It is noted that the decay time constant of mIPSCs in WT or mutant VB neurons was significantly shorter than those previously reported (Mozrzymas et al. 2007; Schofield and Huguenard 2007), presumably because we recorded at a higher temperature (32°C instead of room temperature).
In contrast to what we observed in the VB, RTN neurons from Mecp2-null mice at P14–P16 exhibited significantly more mIPSCs than those from aged-matched WT mice (Fig. 2F). We recorded mIPSCs in 14 neurons from four Mecp2-null mice and 15 neurons from four WT mice aged P14–P16 in the presence of TTX (0.4 μM). The cumulative distribution of the interevent interval shifted to the left in mutant RTN neurons (Fig. 2G), whereas that of the peak amplitude was unchanged (Fig. 2H). The mean frequency of mIPSCs was 60% greater in Mecp2-null mice than in WT mice (Fig. 2I; 3.4 ± 0.4 Hz for WT; 5.4 ± 0.3 Hz for mutants; P < 0.01). There was no significant difference between WT and mutant mice in the peak amplitude (Fig. 2J; 35.4 ± 2.5 pA for WT; 36.0 ± 2.9 pA for mutants; P > 0.5), rise time (0.75 ± 0.02 ms for WT; 0.78 ± 0.02 ms for mutants; P > 0.5), and decay constant (19.6 ± 0.8 ms for WT; 18.8 ± 0.4 ms for mutants; P > 0.15) of mIPSCs. These findings suggest that changes in either presynaptic function or the number of synapses may contribute to the increase of GABAergic transmission in the RTN of mutant mice.
Evoked IPSCs in VB or RTN neurons of Mecp2-null mice
To examine whether MeCP2 deficiency leads to changes in presynaptic function, we analyzed paired-pulse ratio (PPR) of evoked IPSCs in VB and RTN neurons at P14–P16. For VB neurons, monosynaptic IPSCs were evoked by current pulses delivered to the RTN using a concentric bipolar electrode. As shown in Fig. 3A, evoked IPSCs in VB neurons show little trial-to-trial variation. We tested two interstimulus intervals, 50 and 100 ms, at stimulus intensities that produced 60–80% of the maximal responses. There was no difference between mutant and WT mice in the PPR of evoked IPSCs (Fig. 3B). At a 50 ms interval, the PPR was 0.65 ± 0.02 (n = 9 from 3 mice) for mutants and 0.62 ± 0.02 (n = 9 from 3 mice) for WT (P > 0.2); at a 100 ms interval, the PPR was 0.65 ± 0.02 (n = 9) for mutants and 0.63 ± 0.01 (n = 11) for WT (P > 0.4). CV was calculated using 10 consecutive responses evoked at 0.1 Hz at intensities that produced the maximal responses. There was a small but significant increase in CV in mutants (0.092 ± 0.008, n = 21 from 5 mutant mice; 0.069 ± 0.007, n = 17 from 4 WT mice; P < 0.05). Cumulative distributions of the maximal responses were established using 10 consecutive responses for each of the 21 mutant cells and 17 WT cells. The distribution of the maximal response shifted to the left in mutant VB neurons (Fig. 3C). The mean maximal response was reduced by 40% in mutant neurons (660 ± 77 pA for mutants; 1,093 ± 137 pA for WT; P < 0.01).
For RTN neurons, we used bipolar electrodes made from twisted microwires to minimize direct depolarization of the recorded neurons by the stimuli. Electrodes were placed in the RTN ≥200 μm away from the recorded neuron. Even with these precautions, direct depolarization was frequently observed at intensities >200 μA, and we were unable to obtain the maximal responses in RTN neurons. At intensities that did not produce direct depolarization of recorded neurons, evoked IPSCs show large trial-to-trial variations (Fig. 3D). PPR was calculated using averaged responses. No difference was found between WT and mutant mice in the PPR of evoked IPSCs (Fig. 3E). At a 50 ms interval, the PPR was 1.22 ± 0.08 (n = 11 from 4 mice) for mutants and 1.29 ± 0.08 (n = 10 from 4 mice) for WT (P > 0.5); at a 100 ms interval, the PPR was 0.97 ± 0.06 (n = 11 from 4 mice) for mutants and 1.06 ± 0.07 (n = 10 from 4 mice) for WT (P > 0.3).
We also examined unitary IPSCs in VB and RTN neurons using minimal stimulation. Evoked IPSCs were obtained as described above for paired-pulse experiments. Stimulation intensity was reduced to a level where ≥20% of stimuli failed to evoke an IPSC; examples are shown in Fig. 4,A (VB) and C (RTN). For VB neurons, cumulative distributions of IPSC amplitude were obtained using 30 consecutive responses from each of the recorded cells (n = 11 from 4 mutant mice; n = 11 from 4 WT mice). The distribution of IPSC amplitude shifted to the left in Mecp2-null mice (Fig. 4B). The mean amplitude of unitary IPSC was significantly smaller in mutant VB neurons (198.4 ± 34.9 pA, n = 11 from 4 mutant mice vs. 394.2 ± 61.6 pA, n = 11 from 4 WT mice; P < 0.02). CV of IPSC amplitude, calculated with 30 consecutive responses, showed a small increase in mutant VB neurons (0.19 ± 0.02 for mutant vs. 0.14 ± 0.02 for WT; P < 0.05). It is noted that in either WT or mutant VB neurons, the average amplitudes of unitary IPSCs were several-fold larger that those of mIPSCs, indicating that each RTN neuron makes multiple synapses with a single VB neuron.
For RTN neurons, cumulative distributions of IPSC amplitude were obtained using 30 consecutive responses from each of the 13 WT and 13 mutant cells recorded. The distribution of unitary IPSC amplitude shifted to the right in Mecp2-null mice (Fig. 4D). The mean amplitude of IPSCs was 71.4 ± 6.3 pA (n = 13 from 3 mice) in mutant RTN neurons and 52.3 ± 5.1 pA (n = 13 from 3 mice) for WT RTN neurons (P < 0.03). CV of IPSC amplitude was not significantly different between mutant and WT RTN neurons (0.50 ± 0.03 for mutant vs. 0.42 ± 0.03 for WT; P > 0.1).
Developmental regulation of GABAergic defects in the thalamus of Mecp2-null mice
Thus far we have shown that MeCP2 deficiency leads to changes in GABAergic transmission at P14–P16. Such defects could be caused by disruptions of synapse formation, refinement, or maintenance. To distinguish these possibilities, we first recorded mIPSCs in VB and RTN neurons at P6. We observed a significant reduction in the number of mIPSC events in mutant VB neurons at P6 (Fig. 5A). The distribution of interevent interval shifted to the right in mutant neurons (Fig. 5B), whereas the distribution of mIPSC amplitude showed no change (Fig. 5C). The mean mIPSCs frequency was 2.2 ± 0.2 Hz (n = 21 from 4 mice) for mutant and 3.6 ± 0.3 Hz (n = 18 from 4 mice) for WT (Fig. 5D; P < 0.001). There was no difference between mutant and WT mice in the mean peak amplitude (Fig. 5E; 26.9 ± 1.9 pA for mutant; 26.0 ± 2.0 pA for WT; P > 0.7), the rise time (0.58 ± 0.03 ms for mutant; 0.57 ± 0.02 ms for WT; P > 0.7), and the decay constant (8.8 ± 0.6 ms for mutant; 9.6 ± 0.7 ms for WT; P > 0.4) of mIPSCs in VB neurons. These findings suggest that MeCP2 is involved in early phases of GABAergic innervation in the VB.
In contrast to the VB, we did not observe any significant differences in the RTN at P6 between mutant and WT mice (Fig. 5F). The distributions of interevent intervals and peak amplitude of mIPSCs were unchanged in mutant RTN neurons (Fig. 5, G and H). The mean frequency of mIPSCs was 1.9 ± 0.1 Hz (n = 22 from 4 mice) for mutant and 2.0 ± 0.2 Hz (n = 17 from 4 mice) for WT (P > 0.9; Fig. 5I). The peak amplitude of mIPSCs was 42.8 ± 2.0 pA for mutant and 41.2 ± 3.8 pA for WT (P > 0.6; Fig. 5J). The decay constant of mIPSCs was 28.6 ± 1.1 ms for mutant and 26.1 ± 0.9 ms for WT (P > 0.08).
To determine whether the GABAergic defects in Mecp2-null mice are confined to the first 2 wk, we recorded from mutant and WT mice aged P21–P23. For VB neurons, there was a decrease in the number of mIPSC events in mutant mice (Fig. 6, A and B). The mean mIPSC frequency was reduced by 32% in mutant VB neurons (Fig. 6D; 3.2 ± 0.3 Hz, n = 13 from 3 mutant mice; 4.7 ± 0.4 Hz, n = 14 from 3 WT mice; P < 0.003). There was no difference between mutant and WT in the peak amplitude (Fig. 6, C and E, 39.7 ± 4.8 vs. 39.2 ± 3.7 pA, P > 0.7), rise time (0.59 ± 0.03 vs. 0.55 ± 0.02 ms, P > 0.2), and decay constant of mIPSCs (4.5 ± 0.1 vs. 4.4 ± 0.1 ms, P > 0.7). For RTN neurons, we observed a small but significant increase of mIPSC events in mutant mice at P21–P23 (Fig. 6, F and G). The mean mIPSC frequency was 1.8 ± 0.2 Hz (n = 13 from 3 mice) for mutant and 1.1 ± 0.2 Hz (n = 14 from 3 mice) for WT (Fig. 6I; P < 0.008). On the other hand, there was no difference between mutant and WT mice in the peak amplitude (Fig. 6, H and J, 42.4 ± 3.7 vs. 44.1 ± 3.4 pA, P > 0.7), rise time (0.64 ± 0.02 vs. 0.65 ± 0.01 ms, P > 0.7), and decay constant of mIPSCs (20.3 ± 1.1 vs. 20.4 ± 0.5 ms, P > 0.8). Together the results obtained at P21–P23 were comparable with those obtained at P14–P16, suggesting that GABAergic defects in the VB and RTN of Mecp2-null mice persist throughout early life.
Lack of change in intrinsic properties of RTN neurons of Mecp2-null mice
In addition to GABAergic transmission, MeCP2 may regulate membrane properties of GABAergic neurons. This possibility was examined using current-clamp recordings in slices obtained from mutant and WT mice at P15–P17. Potassium methylsulfate was used to replace CsCl in the pipette solution, and RTN cells were recorded in the presence of kynurenic acid (1 mM), DNQX (20 μM), and picrotoxin (100 μM). Few RTN neurons, WT or mutant, showed spontaneous firing (2/18 WT cells; 2/20 mutant cells). There was no difference between WT and mutant RTN neurons in firing pattern. When initially held at −60 to −65 mV and tested with suprathreshold depolarizing current pulses (400 ms), both WT and mutant cells showed spike trains with little frequency adaptation (data not shown). When initially held at −75 to −80 mV, both WT and mutant cells showed one burst consisting of two to three fast spikes in response to suprathreshold current pulses. In addition, there was no difference between WT (n = 18 from 5 mice) and mutant (n = 20 from 5 mice) cells in the resting potential (−56.7 ± 1.4 vs. −57.1 ± 1.5 mV, P > 0.5), input resistance (155 ± 11 vs. 159 ± 13 MΩ, P > 0.7), membrane capacitance (107 ± 6 vs. 103 ± 6 pF, P > 0.5), or action potential threshold (−39.8 ± 0.6 vs. −40.5 ± 0.5 mV; P > 0.4). We also measured the rebound depolarization evoked by hyperpolarization pulses (−100 pA for 400 ms) in cells held initially at −65 mV. There was no difference between WT and mutant cells in the amplitude of the rebound depolarization (8.6 ± 2.1 mV for WT; 8.2 ± 2.3 mV for mutant; P > 0.5). Together, these results indicate that MeCP2 is not needed for functional maturation of RTN cells.
Altered GABAergic innervation in the thalamus of Mecp2-null mice
To determine whether MeCP2 deficiency leads to changes in the number of GABAergic synapses in the thalamus, we performed immunofluorescence staining using antibodies against VGAT in brain sections obtained from five pairs of mice at P18–P20. Each pair consists of one Mecp2-null and its WT littermate. As shown in Fig. 7A, Mecp2-null mice showed significantly fewer VGAT-positive puncta in the VB than WT littermates. The density of VGAT-positive puncta was reduced by 35% in the VB of Mecp2-null mice (Fig. 7C; P < 0.01). In contrast, the density of puncta in mutant RTN was 48% higher than that of WT mice (Fig. 7, B and C; P < 0.01).
We performed double immunostaining of VGAT and GAD67 (glutamate decarboxylase with molecular weight of 67 kDa) in two pairs of mice at P18–P20. As shown in Fig. 7D, all VGAT-positive puncta were also GAD67 positive. However, VGAT was also found in cell body of RTN neurons. Figure 7E showed that VGAT immunoreactivity was present in the cytoplasm of RTN neurons, but the intensity of VGAT signals in the cytoplasm was significantly weaker than those in the puncta located around the cell bodies. Although GAD67 staining reliably labeled GABAergic terminals, axons and cell bodies of RTN neurons also showed strong GAD67 immunoreactivity (data not shown).
The reduction of GABAergic innervation in the VB may be caused by a loss of RTN neurons in mutant mice. To examine this possibility, we analyzed the density of neurons in the RTN by counting NeuN-positive cells in the RTN of four pairs of mice (mutant and WT littermate) at P18–P20. Consistent with previous findings in the hippocampus and cortex (Chen et al. 2001), the density of RTN neurons in mutant mice was 12% higher than that in WT mice (18.7 ± 0.8 cells/104 μm2 for mutants vs. 16.7 ± 0.6 cells/104 μm2 for WT; P < 0.03, paired 2-tailed t-test). A similar increase (12%) in neuronal density was also observed in the VB of mutant mice (22.1 ± 1.5 cells/104 μm2 for mutants vs. 19.8 ± 1.5 cells/104 μm2 for WT; P < 0.01, paired 2-tailed t-test). These results indicate that MeCP2 is not needed for the survival of RTN and VB neurons.
We examined the role of MeCP2 in the function of GABAergic circuits in the thalamus. Using patch-clamp recording in acute brain slices, we showed that MeCP2 deficiency leads to opposite changes in the frequency of GABAergic events in excitatory and inhibitory neurons of the thalamus during postnatal development. Quantal GABAergic transmission is reduced in VB (excitatory) neurons but enhanced in RTN (inhibitory) neurons of Mecp2-null mice at 2–3 wk. These studies suggest that MeCP2 regulates GABAergic synapses differentially in excitatory and inhibitory neurons in the thalamus and that the GABAergic system may play a key role in the pathogenesis of Rett syndrome.
GABAergic defects in the brain of Mecp2-null mice
MeCP2 is highly expressed in GABAergic neurons in the brain (Akbarian et al. 2001). Little is known, however, about the role of MeCP2 in the development and function of GABAergic system. Previous studies in acute slices have shown an increase of spontaneous GABAergic transmission in layer 5 pyramidal neurons of Mecp2-null mice, without any significant changes in the mean frequency or peak amplitude of mIPSCs (Dani et al. 2005). In contrast, a significant reduction in spontaneous GABAergic transmission has been observed in the ventrolateral medulla of Mecp2-null mice (Medrihan et al. 2008). The reason behind this discrepancy is unknown. By focusing on GABAergic circuits in the thalamus, we provided the first evidence that quantal GABAergic transmission is altered in opposite directions in interconnected GABAergic and glutamatergic neurons in Mecp2-null mice.
In excitatory neurons in the VB of Mecp2-null mice, there was a significant reduction in the frequency of mIPSCs. On the other hand, the peak amplitude, rise time, and decay time constant of mIPSCs were not altered in mutant VB neurons, indicating that MeCP2 deficiency does not lead to postsynaptic modification. This latter finding also argues against the possibility that the reduction in mIPSC frequency is caused by altered cable properties or changes in the location of GABAergic synapses in mutant neurons. The lack of change in PPR of evoked IPSCs in VB neurons suggests that the release probability at the RTN–VB synapse was not significantly altered in Mecp2-null mice. On the other hand, the reduction in the amplitude of unitary IPSCs in mutants indicates that VB neurons in Mecp2-null mice receive fewer inputs from single RTN neuron. Furthermore, our immunostaining data showed a significant reduction in the number of GABAergic terminals in the VB of mutants without loss of RTN neurons. Together, these findings suggest that each RTN neuron in Mecp2-null mice makes fewer synapses in the VB. A direct examination of this hypothesis, however, is nontrivial. Previous studies in the rat indicate that each RTN neuron gives raise to a large number of branches and several thousand of synaptic boutons in the relay thalamus (Pinault 1996; Pinault and Deschenes 1998; Pinault et al. 1995). Although it is likely that in mutant mice each RTN neuron has significantly fewer axon branches or fewer synaptic boutons in the relay thalamus, quantification of these changes will require labeling in vivo, using either chemical or genetic methods, of a single RTN neuron with all its axon branches and synaptic terminals.
In contrast to the VB, there was a significant increase in the number of quantal GABAergic events in RTN neurons of Mecp2-null mice. What caused this increase remains unclear. Neither the peak amplitude nor the decay time constant of mIPSCs was altered in mutant RTN neurons, indicating that postsynaptic modifications were not involved. However, a number of changes at the presynaptic level can contribute to the change in mIPSC events. First, GABAergic synapses in mutant RTN neurons may have higher levels of spontaneous vesicular release. Because of the complexity in GABAergic connections in the RTN, PPR may not be a reliable indicator of release probability at these synapses. Even if one could reliably compare the release probability for evoked responses, it is possible that spontaneous release behaves very differently (Neher and Sakaba 2008). Second, each RTN neuron in mutants may be innervated by more RTN neurons. Previous studies have shown that dendrodendritic contacts constitute a major portion of local connections in the RTN (Pinault et al. 1997). The higher density of neurons seen in the mutant RTN could lead to an increase in the number of target neurons for each RTN neuron. Last, the increase in the minimal stimulation response suggests that each RTN neuron in mutants may make more synapses with other RTN neurons. These three models are not mutually exclusive, and thus far, we do not have evidence to exclude any of them.
Similar to the results obtained in pyramidal neurons of the neocortex (Dani et al. 2005), the loss of MeCP2 had no effect on the membrane properties and excitability of RTN neurons. There was no difference between WT and mutant RTN cells in the input resistance, membrane capacitance, resting potential, spike threshold, or firing pattern. Recent studies in the rat described several subtypes of RTN neurons with regard to excitability. Many neurons in the dorsal part of the RTN lack burst discharge, whereas most cells in the ventral RTN show typical burst discharge (Lee et al. 2007). In our studies, both WT and mutant RTN cells displayed burst discharge when initially held at hyperpolarized membrane potential. The lack of heterogeneity may be because of the fact that our recordings were done in the ventral part of the RTN. Despite this caveat, our results indicate that MeCP2 is not needed for the functional maturation of neurons but plays an important role in synapse development.
Our results in the thalamus are significantly different from those observed in the medulla. The frequency, peak amplitude, and decay time constant of mIPSCs were all reduced in neurons in the ventrolateral medulla of Mecp2-null mice at P7 (Medrihan et al. 2008), indicating both pre- and postsynaptic changes. The differences between the results obtained in the medulla and our findings in the thalamus imply that cellular and molecular mechanisms underlying MeCP2-mediated regulation of GABAergic transmission are likely to be regional and cell type specific.
MeCP2 in the development of GABAergic circuits
The fact that GABAergic transmission is defective in neonatal Mecp2-null mice suggests a role for MeCP2 in the development of GABAergic circuits in the brain. In the mutant VB, our results showed a persistent reduction in mIPSC frequency throughout early life (P6–P23), suggesting that MeCP2 is needed for the development of GABAergic innervation in the VB. In contrast, mutant RTN neurons did not show any difference in quantal GABAergic transmission at P6, but an increase in mIPSC frequency at P14–P16 and P21–P23. It is noted that there was a significant decrease in mIPSC frequency between P14–P16 and P21–P23 in both WT (3.4 ± 0.4 vs. 1.1 ± 0.2 Hz; P < 0.001) and mutant RTN neurons (5.4 ± 0.3 vs. 1.8 ± 0.2 Hz; P < 0.001). There are two caveats about these results. First, there was a significant increase in baseline noise in RTN neurons between P14 and P21; this increase in baseline noise may reduce the success rate of minidetection at older ages. Second, RTN neurons at P21–P23 may have larger dendritic trees and thus are more likely to be truncated in slice preparation. Despite these caveats, these results raised the possibility that, during normal development, there is a transient overproduction of local connections in the RTN, and this developmental process is regulated by MeCP2.
Similar to its action on glutamatergic synapses, MeCP2 seems to regulate the number of GABAergic synapses in the thalamus and medulla. As a chromatin-associated protein, MeCP2 regulates the expression of a large number of genes in the brain (Chahrour et al. 2008). One of the targets, the brain-derived neurotrophic factor (BDNF), is likely to mediate the action of MeCP2 on GABAergic circuits. BDNF levels are reduced in immature brains of Mecp2-null mice (Chang et al. 2006; Wang et al. 2006). Many studies have shown that BDNF promotes the formation and maturation of GABAergic synapses (Huang et al. 1999; Marty et al. 2000). For example, deletion of the BDNF gene in single cortical neurons leads to a reduction in the frequency of mIPSCs without any change in the peak amplitude (Kohara et al. 2007). Recent studies have shown that activity-dependent regulation of BDNF is needed for the normal development of GABAergic synapses (Hong et al. 2008) and that MeCP2 promotes activity-dependent regulation of BDNF (Zhou et al. 2006). Therefore MeCP2 deficiency may lead to a reduction of GABAergic synapses through a down-regulation of BDNF expression.
Although the BDNF hypothesis can explain the changes in the VB and medulla of Mecp2-null mice, it is unclear whether this hypothesis can also explain the findings in the RTN. In culture, BDNF promotes the differentiation of GABAergic neurons and the formation of synapses between GABAergic neurons (Berghuis et al. 2004; Marty et al. 1996). Thus the reduction of BDNF levels in Mecp2-null mice should result in a decrease of GABAergic transmission in the RTN. Instead, we found an enhancement of GABAergic transmission in RTN neurons of mutant mice. It is likely that mechanisms other than BDNF signaling play major roles in MeCP2-mediated regulation of GABAergic transmission in the RTN.
GABAergic systems in the pathogenesis of Rett syndrome
Several lines of evidence suggest that GABAergic defects are involved in the pathophysiology of Rett syndrome. Expression analyses showed a reduction of the β3 subunit of GABAA receptor in cortical samples of Rett syndrome patients and in the neocortex of MeCP2-deficient mice (Samaco et al. 2005). Extracellular recordings from CA3 region of the hippocampus indicated a reduction in basal inhibitory activity in symptomatic Mecp2-null mice (Zhang et al. 2008). A recent study reported that selective deletion of Mecp2 in GABAergic neurons in the forebrain led to impaired motor coordination in the mouse, suggesting that the motor impairment in Rett syndrome patients may be caused by GABAergic defects in the forebrain (Chao et al. 2008). Our results in the thalamus, together with those obtained in the medulla (Medrihan et al. 2008), indicate that GABAergic defects develop in Mecp2-null mice before the onset of disease phenotypes. These findings raise the possibility that defects of GABAergic systems play an important role in the pathogenesis of Rett syndrome.
A functional implication of our findings in the thalamus is that opposite changes in GABAergic transmission in excitatory and inhibitory neurons will dramatically alter the dynamics of thalamocortical circuits. Such changes may be a neuronal basis for abnormal somatosensory responses, disrupted sleep pattern, and seizures observed in Rett syndrome patients (Glaze 2005; Yoshikawa et al. 1991; Young et al. 2007). This possibility will be examined in future studies combining in vivo and in vitro recordings. Furthermore, it may be important to determine whether cell type–specific effects of MeCP2 on GABAergic transmission are also present in other brain regions. Elucidating the role MeCP2 in the development and function of GABAergic systems will significantly contribute to our understanding and treatments of Rett syndrome.
This work was supported by a National Institute of Neurological Disorders and Stroke Grant NS063071 to Z.-W. Zhang.
No conflicts of interest are declared by the authors.
We thank M. Deschenes, W. Frankel, and R. Burgess for helpful comments on a previous version of the manuscript.
- Copyright © 2010 the American Physiological Society