## Abstract

Dystrophin is a cytoskeletal membrane-bound protein expressed in both muscle and brain. Brain dystrophin is thought to be involved in the stabilization of γ-aminobutyric acid (GABA)A-receptor (GABAA-R)clusters in postsynaptic densities (PSDs) at inhibitory synapses onto pyramidal cells, and its loss has been linked to cognitive impairments in Duchenne muscular dystrophy. Dystrophin-deficient mdx mice have learning deficits and altered synaptic plasticity in cornu ammonis (CA1) hippocampus, but the possibility that altered synapse morphology or distribution may underlie these alterations has not been examined. Here we used in vivo magnetic resonance imaging and histological analyses to assess brain volumetric and cytoarchitectonic abnormalities and quantitative electron microscopy to evaluate the density and ultrastructure of CA1 hippocampal synapses in mdx mice. We found that mdx mice have increased density of axodendritic symmetric inhibitory synapses and larger PSDs in perforated asymmetric excitatory synapses in the proximal, but not distal, CA1 apical dendrites that normally express dystrophin, in the absence of gross brain malformations. Data are discussed in light of the known molecular and neurophysiological alterations in mdx mice. We suggest that increased inhibitory synapse density reflects tenuous compensation of altered clustering of α2 subunit–containing GABAA-Rs in CA1 dendrites, whereas increased PSD length in perforated synapses suggests secondary alterations in excitatory synapse organization associated with enhanced synaptic excitation.

## Introduction

Dystrophin is a 427-KDa spectrin-like cytoskeletal protein that links a complex of intrinsic membrane proteins (dystrophin-associated glycoprotein complex) to the actin-based cytoskeleton in both muscle and brain cells and shows tissue-specific interactions with membrane receptors and/or ion channels (Blake and Kröger 2000). Dystrophin loss is responsible for Duchenne muscular dystrophy (DMD), an X-linked neuromuscular disease characterized by progressive muscle degeneration and nonprogressive cognitive deficits. DMD patients display variable degree of cognitive impairment ranging from mild deficits in verbal skills, selective attention, and memory performance to mental retardation with IQs below 70 (Billard et al. 1992; Hinton et al. 2001; Anderson et al. 2002). It has long been suggested that genetic loss of dystrophin is responsible for part of these deficits, as brain dystrophin is normally expressed in brain structures involved in diverse cognitive functions, such as the hippocampus, neocortex, and cerebellum (Gorecki et al. 1991; Lidov et al. 1993). Although macroscopic brain alterations such as ventricular enlargement and cortical atrophy have been reported in some DMD patients, brain autopsies and imaging studies have not led to consistent results, and a possible link between dystrophin loss and gross and/or ultrastructural brain abnormalities remains unclear (for reviews, see Septien et al. 1991; Anderson et al. 2002). However, the detection of glucose hypometabolism in hippocampus, cerebellum, and sensorimotor cortex, as well as dendritic abnormalities and of nonspecific electroencephalographic (EEG) disturbances, suggest altered synaptic function (Jagadha and Becker 1988; Anderson et al. 2002; Lee et al. 2002), an hypothesis consistent with the enrichment of dystrophin in postsynaptic densities (PSDs) in both human and mouse brains (Lidov et al. 1990; Kim et al. 1992, 1995).

Numerous studies in the dystrophin-deficient mdx mouse further support a role for dystrophin in brain and synapse function. Dystrophin-deficient mdx mice show specific learning and memory deficits (Muntoni et al. 1991; Vaillend et al. 1995, 1998, 2004), altered neuronal calcium homeostasis (Hopf and Steinhardt 1992), enhanced sensitivity of cornu ammonis (CA1) hippocampal neurons to hypoxia (Mehler et al. 1992), and morphological alterations of some cortical neurons (Sbriccoli et al. 1995). The absence of dystrophin in mdx mice leads to a marked reduction in the number and size of γ-aminobutyric acid (GABA)A receptor (GABAA-R) clusters in hippocampal area CA1 and cerebellum (Knuesel et al. 1999). Recent studies underscored that dystrophin is a PSD component of central inhibitory synapses involved in long-term stabilization of membrane GABAA-R clusters and the organization of GABAergic networks (Knuesel et al. 2001; Brünig et al. 2002; Lévi et al. 2002), suggesting that altered GABAergic function might play a key role in the cognitive deficits associated with dystrophin deficiency. We previously showed that mdx mice have abnormal enhancement of CA1 hippocampal long-term potentiation (LTP), and pharmacological analyses suggested that altered GABAergic function may be responsible for the altered threshold of N-methyl-D-aspartate receptor (NMDAr) activation and abnormal activity-dependent synaptic plasticity (Vaillend and Billard 2002; Vaillend et al. 2004). Modifications in the number, proportion, and morphology of synapses are thought to underlie changes in synaptic efficacy associated with normal development, lesion repair, synaptic plasticity, learning, and memory processes (Geinisman 2000; Marrone and Petit 2002), with major contribution to a variety of inherited forms of mental retardation and cognitive dysfunctions in mice (for review, see Vaillend et al. 2008). Although congenital alterations in GABAA-R clustering in dystrophin-deficient brains may have important consequences for synapse ultrastructure and integrity of hippocampal synaptic networks, this to date has not been studied in detail.

In the present study, we used quantitative serial electron microscopy (EM) and physical disector stereological methods to evaluate the density and PSD length of distinct synapse subtypes in the CA1 hippocampal area. Asymmetric excitatory, symmetric inhibitory, axospinous, axodendritic, axosomatic synapses, and multiple spine boutons (MSBs) were analyzed at 2 distinct levels of CA1 apical dendritic arbors in proximal stratum radiatum enriched in dystrophin and GABAA-R clusters and in distal radiatum, a region of wide interest to investigate the structural integrity of synaptic networks involved in synaptic plasticity. In the same tissue samples, we also estimated the relative density of CA1 pyramidal cells to identify putative neuronal loss and to determine the synapse-to-neuron ratio. Finally, volumetric in vivo magnetic resonance imaging (MRI) and histological volume estimation of distinct hippocampal regions by light microscopy was conducted to weight any ultrastructural abnormalities against possible macroscopic brain alterations.

## Materials and Methods

### Animals

Mice of the C57BL/10ScSn-Dmdmdx/J (mdx) mutant and C57BL/10J control strains were bred in our own colony as described (Vaillend et al. 2004). Heterozygous mdx females were mated with control males to provide the mdx and wild-type (WT) littermate male mice used in these experiments. A total of 31 male mice aged at least 6 months were used in this study (WT, n = 17; mdx, n = 14). Siblings were kept in the same cage (2–4 mice per cage) under a 12-h light–dark cycle (lights on: 7.00 AM) with food and water ad libitum. Mouse genotype was determined as previously described (Vaillend et al. 2004): Plasma analyses were undertaken to reveal high levels of serum creatine kinase in the mutants, which is a characteristic of the early phase of the disease in both DMD patients and mouse models of DMD, and additional postmortem histological analysis of quadriceps muscle confirmed that mutants had characteristic changes within the muscle fibers (i.e., centralization of the nuclei). All experiments were conducted blind to the genotype and in accordance with the European Communities Council Directive of 24 November 1986.

### Magnetic Resonance Imaging

Mice (>6 months old) were imaged under isoflurane anesthesia (induction 2%; flow rate 0.8% in 50% O2, 50% N2O) controlled on the basis of respiratory parameters. The body temperature was maintained at 37 °C using heated mattress. MRI measurements were performed on a 7-T horizontal bore magnet (Oxford, UK) driven by Paravision (Bruker®, Wissembourg, France) and equipped with a 300 mT/m actively shielded gradient device (internal diameter [ID] = 90 mm, Bruker). For MRI examination, the animal's head was introduced in a “bird-cage” 1-H coil (ID = 36 mm). Images were acquired perpendicular (axial) and parallel (coronal) to the brain anteroposterior axis. The mouse brain was positioned using scouting gradient-echo images in the 3 orthogonal directions. After the shimming process, Bruker sequences were acquired with fat suppression as 2D axial-Rapid acquisition with relaxation enhancement (RARE) (TR/TE eff = 4000/40 ms [repetition time/effective echo time]; rare factor = 8; 2 averages; 31 contiguous 0.5-mm-thick sections; 30 × 30 mm field of view; 256 × 256 matrix; pixel size = 117 μm2) and 3D coronal TurboRARE sequences (TR/TE eff = 2500/42.2 ms; rare factor = 16; 30 × 15 × 15 mm field of view; 256 × 128 × 128 matrix; voxel size = 117 μm3). The total time spent by the mouse in the magnet was around 50 min. After each experiment, mice were recovered from anesthesia and returned to their home cages with free access to food and water.

### Tissue Preparation

For light microscopy, mice were deeply anesthetized by intraperitoneal injection of sodium pentobarbital and then intracardially perfused with a 4% paraformaldehyde fixative solution in 0.1 M phosphate buffer, pH 7.3–7.4. Brains were cryoprotected in 30% sucrose and then quickly frozen in 2-methylbutane (Roth, Karlsruhe, Germany) cooled with dry ice to −40 °C. Coronal serial sections (30 μm thick) were collected and stored at −80 °C.

For EM, mice aged 6 months were perfused with a washing solution (0.1 M cacodylate buffer, pH 7.3–7.4, 1 min) followed by a fixative solution (2% paraformaldehyde, 2% glutaraldehyde, 0.002% CaCl2, 0.15% picric acid in 0.1 M cacodylate buffer, pH 7.3–7.4). Brains were quickly removed from the skull, and ∼3-mm-thick slabs, containing the whole hippocampus, were cut with the aid of a brain matrix (Harvard Apparatus, Les Ulis, France). Both hippocampi were postfixed in the same fixative solution overnight at 4 °C. An average of thirteen 250-μm-thick sections of the right and left hippocampi were made in the coronal plane with a vibrating microtome (VT 1000 S, Leica, Vienna, Austria). The anterodorsal hippocampus was dissected under stereomicroscope, and tissue samples were then osmicated in 1% buffered osmium tetroxide, soaked in 1% osmium tetroxide and potassium ferrocyanide (10 mg/mL) in 0.1 M cacodylate buffer, rinsed in buffer and water, contrasted in aqueous solution of uranyl acetate (1%) overnight at 4 °C, and dehydrated in graded ethanols. Finally, the tissue was flat embedded in Epon 812 (embed-812, EMS Co. Ltd, Fort Washington, PA) and polymerized at 60° for 48 h. Resin blocks were pseudorandomly selected within a given range of coordinates (bregma −1.70 to −2.04) (Paxinos and Franklin 2001). Serial semithin sections (2 μm thick) were then cut on a Leica UCT ultramicrotome (Leica), collected on superfrost plus (Roth) glass slides, and stained with toluidine blue for neuron number quantification. For synapse number estimation, the same blocks were then trimmed to a trapezoidal-like region (∼1 mm2) containing both the CA1 pyramidal cell bodies and the entire apical dendritic field of CA1. Ribbons of at least 4 ultrathin serial sections (70 nm thick) of silver gray interference color were collected on 2 × 1–mm–slot grids coated with a formvar–carbon support film and counterstained with uranyl acetate and lead citrate.

### Brain Volumetric Analyses

MRI areas and volumes were estimated on coronal and axial brain sections using the AMIRA software (TGS Inc., San Diego, CA) (WT, n = 4; mdx, n = 4). Analyses were performed as previously described (Sebrié et al. 2008). Five main brain areas were delineated according to mouse brain atlas (Rosen et al. 2000; Paxinos and Franklin 2001): the cerebellum (including colliculus), olfactory bulbs, whole cortical area, ventricles, and hippocampus (Fig. 1). Brain volumes were extracted by multiplying areas by slice thickness.

Figure 1.

MRI evaluation of brain volumetric integrity in WT (top) and mdx mice (bottom). Sample MRI coronal images (a, WT; d, mdx), corresponding 2D cutting planes with brain regions delineated according to mouse brain atlas (b, WT; e, mdx) and 3D reconstructions of mouse brains (c, WT; f, mdx), show comparable brain gross anatomy in the 2 genotypes. Brain regions considered in the analysis are represented by distinct colors as follows: ventricles, red; cortex, yellow (b, e) or translucent golden gray (c, f); hippocampi, orange; other miscellaneous subcortical structures, white; olfactory bulbs, green; cerebellum/colliculus, purple.

Figure 1.

MRI evaluation of brain volumetric integrity in WT (top) and mdx mice (bottom). Sample MRI coronal images (a, WT; d, mdx), corresponding 2D cutting planes with brain regions delineated according to mouse brain atlas (b, WT; e, mdx) and 3D reconstructions of mouse brains (c, WT; f, mdx), show comparable brain gross anatomy in the 2 genotypes. Brain regions considered in the analysis are represented by distinct colors as follows: ventricles, red; cortex, yellow (b, e) or translucent golden gray (c, f); hippocampi, orange; other miscellaneous subcortical structures, white; olfactory bulbs, green; cerebellum/colliculus, purple.

The volume of the hippocampus was also estimated by histological analysis in formalin-fixed brain frozen sections (WT, n = 4; mdx, n = 4). The first and last hippocampal sections (30 μm thick) were identified according to mouse brain atlas (Paxinos and Franklin 2001). For systematic sampling, the first serial section containing hippocampus was randomly chosen within the 6 first sections. Then, one in every 6 sections were stained with cresyl violet and photographed at 1.25× with a Sony DFW-X700 digital camera (Sony Co., Tokyo, Japan) coupled to Olympus BX60 light microscope (Olympus Optical Co., Hamburg, Germany). The volumes of the whole hippocampus (bregma −0.94 to −4.04) and of anterodorsal hippocampus (bregma −0.94 to −2.30) were determined by boundary tracing of cross-sectional areas. Similarly, serial sections photographed with a 4× objective were used to estimate unilateral volumes of specific regions of interest for both the synapse and neuron counts in our study, that is, the anterodorsal CA1 subfield, stratum radiatum, and stratum pyramidale (Fig. 2a). Calibrated digital images were imported into the ImageJ software (http://rsb.info.nih.gov/ij/), and cross-sectional areas were outlined using a graphic tablet (Intuos3 Pen Tablet, Wacom Co. Ltd, Saitama, Japan). The volume was determined as the sum of the traced areas multiplied by the distance between sampled sections (180 μm) (Gundersen and Jensen 1987).

Figure 2.

Figure 2.

To evaluate the integrity of hippocampal regions in the same conditions as for ultrastructural study of CA1 hippocampal synapses, the volume of hippocampus was also estimated from thick (250 μm) vibratome sections in 8 mouse brains processed as for EM (WT, n = 4; mdx, n = 4). Serial sections were floated on a bath filled with 0.1 M cacodylate buffer (pH 7.3–7.4) at room temperature and then placed onto clean slides to take pictures. Cross-sectional areas were outlined for every section containing the hippocampus and the values converted into volume estimation of the whole and anterodorsal hippocampus as above.

### Quantification of CA1 Hippocampal Neurons

The numerical density (Nv) of pyramidal cells in anterodorsal CA1 was estimated by applying the disector method (Sterio 1984) to pairs of adjacent serial semithin sections. On average, 7 pairs of sections per animal with a 6-μm interpair separation were photographed with a 40× objective and aligned using the IGL Align software (http://synapse-web.org/tools/index.stm). Quantification was performed with the aid of several macros adapted for stereology estimations in ImageJ software. Briefly, the CA1 pyramidal layer was delimitated, and a test grid (rectangular unbiased counting frames of 350 μm2) was randomly overlaid on each pair of sections (Fig. 2b) to perform a uniform sampling of equally spaced counting frames. Frames with the upper left corner falling outside the pyramidal layer were discarded. On average, 102 neurons per animal were counted in 455 frames using the top of the nucleus as the counting unit. As a general rule, neurons touching the left or bottom sides of the counting frame were discarded. Only neurons seen in the first section (reference section), but not in the adjacent one (lookup section), were counted (Fig. 2b). The process was then repeated inverting the reference and lookup sections to increase the efficiency of the quantification. The relative CA1 pyramidal neuron density (Nv) was finally calculated as: $Nv = ΣQ−/a × h, where ΣQ−$ is the particle count, a is the sampled area, and h the thickness of a semithin section (2 μm).

Additional morphometric measures were taken using a randomly chosen single section from each pair of serial sections. As for neuron counting, the boundaries of the stratum pyramidale were delimitated and a test grid randomly overlaid. The neurons’ profiles were outlined and the following parameters derived: 1) The number of neurons per surface area, that is, the ratio of the counted neurons over the total area covered by the counting frames (13 650–18 550 μm2); 2) the nuclear area; 3) the relative neuron packing, that is, percentage of area covered by neurons in sample area; and 4) the nucleus circularity defined as (4 × π × area)/(perimeter)2, which provided values ranging from 1.0 (perfect circle) to 0.1 (flat-shaped nucleus) (ImageJ, 1.37v; http://rsb.info.nih.gov/ij/) (DeCoster 2007).

### Quantification of CA1 Hippocampal Synapses

#### EM Inspection and Regions of Interest

Ultrathin sections placed on slot grids were visualized at 14 000× using a Philips EM208 transmission electron microscope (Philips Electron Optics, Eindhoven, NL) and photographed with a CCD AMT digital camera (Hamamatsu City, Japan). Identification and quantification of synaptic profiles was undertaken in CA1 stratum radiatum of the anterodorsal hippocampus, in which the association of brain dystrophin with GABAA-Rs and its role in synaptic plasticity have been previously investigated. Stratum radiatum is formed by apical dendrites of CA1 pyramidal cells that receive synaptic inputs from the commissural and schaffer collateral fibers (Amaral and Witter 1989). We ultrastructurally identified this region by the striated appearance of dendritic shafts running in parallel from apices of pyramidal cells perpendicular to the hippocampal fissure (Megías et al. 2001). Synapses were examined in both proximal (from 0 to 30 μm from CA1 pyramidal cell bodies) and distal (∼200 μm from the CA1 pyramidal cell bodies) radiatum. Low-power micrographs (180×) taken right after EM inspection were used to verify that the proximal and distal pictures were taken at comparable levels among animals and genotypes (Fig. 2c).

#### Synapse Nomenclature

Synapses were first classified as asymmetric (type I) and symmetric (type II) according to standard anatomical criteria (Colonnier 1968). Asymmetric synapses are non-GABA immunoreactive, usually contain glutamate and are therefore considered excitatory, whereas symmetric synapses contain GABA and are considered inhibitory (DeFelipe et al. 1997; Fiala and Harris 1999; Megías et al. 2001). Asymmetric (excitatory) synapses were identified by a thick PSD and presence of round-shaped vesicles in presynaptic terminals (e.g., Fig. 2e and Fig. 4c,d), whereas symmetric (inhibitory) synapses were characterized by a thin PSD and flattened presynaptic vesicles in presynaptic terminal (Fig. 3c,d). Asymmetric synapses with one or more perforations in the PSD were further classified as perforated (Fig. 4d) and those showing a continuous PSD as nonperforated (Fig. 4c). According to their site of contact, synapses were further categorized as axospinous, axodendritic, and axosomatic. Dendritic shafts were identified by the presence of mitochondria and arrays of parallel microtubules. Postsynaptic spines were identified based on the absence of mitochondria and microtubules and the presence of the spine apparatus commonly located at their neck. Finally, somata portions were identified by the presence of a cell nucleus, rough endoplasmic reticulum, and/or Golgi apparatus (Fiala and Harris 1999). Synapses could be thus classified in the following categories: axospinous asymmetric nonperforated, axospinous asymmetric perforated, axodendritic asymmetric, axodendritic symmetric, and axosomatic symmetric. Finally, single presynaptic terminals contacting more than one postsynaptic spine were defined as MSBs. Rigorous identification of perforated synapses and MSBs was facilitated by the use of several guard sections (see Fig. 2d,e).

Figure 3.

Genotype effect on the relative density of axodendritic symmetric inhibitory synapses. (a) Synapse density expressed as mean ± SEM per 100 μm2 in proximal and distal stratum radiatum of hippocampal CA1. Mdx mice (black bars) show a significant increase in the number of inhibitory synapses in the proximal but not distal radiatum as compared with WT mice (white bars). *P < 0.05. (b) Density of axodendritic symmetric synapses within the proximal radiatum in mdx (black symbol) and WT littermate mice (white symbols). The distribution of individual values revealed no overlap between the 2 groups of mice. Horizontal lines indicate the group means. (c) Micrograph showing an axodendritic symmetric inhibitory synapse (mdx mouse) in proximal radiatum. Arrow indicates the presence of pleomorphic (flattened) vesicles in presynaptic terminal. Arrowheads show the length of the typically thin PSD of symmetric synapses. Postsynaptic dendritic shaft identified by the presence of mitochondria (m). Scale bar, 0.5 μm. (d) Different examples of axodendritic symmetric inhibitory synapses identified in the proximal radiatum by the presence of symmetrical presynaptic and postsynaptic densities and pleomorphic vesicles in the presynaptic terminal. Scale bar, 0.5 μm.

Figure 3.

Genotype effect on the relative density of axodendritic symmetric inhibitory synapses. (a) Synapse density expressed as mean ± SEM per 100 μm2 in proximal and distal stratum radiatum of hippocampal CA1. Mdx mice (black bars) show a significant increase in the number of inhibitory synapses in the proximal but not distal radiatum as compared with WT mice (white bars). *P < 0.05. (b) Density of axodendritic symmetric synapses within the proximal radiatum in mdx (black symbol) and WT littermate mice (white symbols). The distribution of individual values revealed no overlap between the 2 groups of mice. Horizontal lines indicate the group means. (c) Micrograph showing an axodendritic symmetric inhibitory synapse (mdx mouse) in proximal radiatum. Arrow indicates the presence of pleomorphic (flattened) vesicles in presynaptic terminal. Arrowheads show the length of the typically thin PSD of symmetric synapses. Postsynaptic dendritic shaft identified by the presence of mitochondria (m). Scale bar, 0.5 μm. (d) Different examples of axodendritic symmetric inhibitory synapses identified in the proximal radiatum by the presence of symmetrical presynaptic and postsynaptic densities and pleomorphic vesicles in the presynaptic terminal. Scale bar, 0.5 μm.

Figure 4.

Comparison of the PSD length of perforated and nonperforated excitatory synapses quantified in the proximal and distal radiatum of mdx and WT mice. (a) The length of perforated PSDs was significantly larger in distal than in proximal radiatum and increased in mdx as compared with WT mice. *P < 0.05, 2-way ANOVA. The length of nonperforated PSDs was comparable between regions and genotypes (P > 0.05). (b) Genotype difference in perforated synapse PSD length was particularly evident in proximal radiatum. (c) Micrograph showing the PSD length in an axospinous asymmetric nonperforated synapse (WT mouse, proximal radiatum). Scale bar, 0.5 μm. (d) Example micrograph showing how the total length of a perforated PSD was estimated as the sum of the individual electron-dense traces (mdx mouse, distal radiatum). Scale bar, 0.5 μm.

Figure 4.

Comparison of the PSD length of perforated and nonperforated excitatory synapses quantified in the proximal and distal radiatum of mdx and WT mice. (a) The length of perforated PSDs was significantly larger in distal than in proximal radiatum and increased in mdx as compared with WT mice. *P < 0.05, 2-way ANOVA. The length of nonperforated PSDs was comparable between regions and genotypes (P > 0.05). (b) Genotype difference in perforated synapse PSD length was particularly evident in proximal radiatum. (c) Micrograph showing the PSD length in an axospinous asymmetric nonperforated synapse (WT mouse, proximal radiatum). Scale bar, 0.5 μm. (d) Example micrograph showing how the total length of a perforated PSD was estimated as the sum of the individual electron-dense traces (mdx mouse, distal radiatum). Scale bar, 0.5 μm.

#### Synapse Quantification

The numerical density (Nv) of synapses was estimated from serial ultrathin sections. Following the procedure described by Eyre et al. (2003), each serial section of the ribbons mounted on slot grids was labeled with a letter (A, B, C, and D) as shown in Figure 2d. Two distal series mounted on different grids were photographed taking at least 30 images from each section. For uniform sampling, an object or structure (e.g., blood vessel, dendritic segment, or cross-sectioned myelinized axon) present in all sections was taken as the first reference point. Images were converted to BMP 8 bits gray scale files, aligned with the IGL software, and quantified with ImageJ. The physical disector method (described above) was applied to all pairs of images labeled B and C. The first and last images of a ribbon (i.e., A and D) were used as additional guard sections to facilitate the identification of synapses in the z axis (see example in Fig. 2e). For each mouse, stereological quantification of synapses was performed on an average of 33 disector pairs for the proximal radiatum and 26 disector pairs for distal radiatum. The reference and lookup sections were designated, and a counting frame (29.75 μm2) was randomly superimposed. The counting unit (Q) corresponded to the synaptic profiles with clearly identified PSD, synaptic cleft, and presynaptic component with synaptic vesicles (Mayhew 1996). The height of the disector h corresponded to the mean thickness of ultrathin sections calculated by the minimum fold method (Weibel 1979) and resulted in a mean value of 60 nm. The sampled area a corresponded to the total surface of the neuropil covered by the counting frames (1990 μm2 in proximal radiatum and 1503 μm2 in distal radiatum). Systematic sampling may introduce some bias when counts and reference area estimations are made in regions that contain large dendritic profiles and cell somata portions (DeFelipe et al. 1999). Because this was typically observed in the proximal radiatum, the dendritic and somata portions in this area were outlined and the resulting area (∼408 μm2) subtracted from the total sampled area. In accordance with the disector method, the number of synaptic profiles that were present in the reference section but disappeared in the lookup section was divided by the disector volume (i.e., a × h, see above) to obtain the total number of synapses (i.e., asymmetric + symmetric) per cubic microns. For large, scarce, or complex sets of synaptic profiles (i.e., perforated, axodendritic, axosomatic synapses, and MSBs), the number of which is underestimated by the disector method when h ranges 60–70 nm, single-section quantification (i.e., 2D) was used to derivate the number of synapse types per square microns and optimize quantification and statistical comparisons (DeFelipe et al. 1999; Shi et al. 2005). To take into account a possible bias due to differential tissue shrinkage, the number of synapses was also normalized to the number of CA1 pyramidal neurons counted in the same resin blocks and was expressed as synapse-to-neuron ratio.

#### PSD Length

For all synapses identified in the proximal and distal radiatum, the PSD length was measured from single ultrathin sections. Measures were taken with ImageJ software in all frames of sections B because this second section of the ribbon was never distorted during digital aligning of serial images. As in Desmond and Levy (1986a) when the PSD was perforated, the sum of the individual trace lengths was considered (e.g., see Fig. 4d).

### Statistical Analysis

Data are presented as means ± standard errors of the mean (SEMs). All data successfully passed normality and equal variance tests (SigmaStat 2.0, SPSS Inc., Chicago, IL; P > 0.05) and were therefore analyzed with parametric tests. Volumes and neuron density were analyzed with 1-way analysis of variance (ANOVA) with genotype (WT and mdx) as the between-subject factor. Following the recommendations of DeFelipe et al. (1999) for the analyses of synapse density in different neuropil regions, the synapse-to-neuron ratio and the number of synapses per surface area recorded in proximal and distal radiatum were analyzed separately. For PSD length, a parameter that is not affected by differences between sampled areas, the region of interest (i.e., proximal vs. distal radiatum) was used in 2-way ANOVAs as a within-subject factor. P values <0.05 were considered statistically significant.

## Results

### In Vivo MRI and Histological Evaluation of Brain Integrity

We first explored possible changes in the volume of brain ventricles and of various brain structures in mdx mice to determine whether dystrophin loss results in macroscopic brain alterations as reported in a subset of DMD patients (Anderson et al. 2002). Table 1 shows volumes of different brain structures derived from in vivo MRI and “postmortem” histological analyses. MRI was performed on 4 mdx and 4 WT littermate mice of comparable age (>6 months). The volumes of the whole brain and ventricles were comparable between genotypes (P > 0.05). Similarly, there were no significant changes in the volumes of hippocampus and whole cortical areas. Finally, no significant differences were detected for the volume of the olfactory bulbs and cerebellum (P > 0.05). MRI analysis therefore revealed no major alteration of brain anatomy in mdx mice and no significant changes in brain structures that normally express dystrophin (cortex, hippocampus, and cerebellum) (Fig. 1).

Table 1

Volumes of different brain structures estimated from in vivo MRI and histological analyses

 MRI Vibratome sections Frozen sections WT mdx WT mdx WT mdx Whole brain 465.87 ± 6.82 473.19 ± 9.23 — — — — Lateral ventricles 24.93 ± 4.17 33.08 ± 10.30 — — — — Olfactory bulbs 22.60 ± 1.25 23.30 ± 1.06 — — — — Cerebellum 63.99 ± 0.75 66.51 ± 1.74 — — — — Cortices 158.76 ± 2.58 158.79 ± 3.22 — — — — Hippocampus 25.78 ± 1.26 26.04 ± 0.96 19.80 ± 0.49 19.99 ± 0.52 18.48 ± 0.46 19.56 ± 0.43 AD hippocampus — — 5.96 ± 0.47 5.69 ± 0.56 7.68 ± 0.25 7.57 ± 0.45 CA1 field — — — — 0.946 ± 0.060 0.912 ± 0.067 CA1 pyramidal layer — — — — 0.088 ± 0.004 0.086 ± 0.006 CA1 stratum radiatum — — — — 0.373 ± 0.028 0.365 ± 0.031
 MRI Vibratome sections Frozen sections WT mdx WT mdx WT mdx Whole brain 465.87 ± 6.82 473.19 ± 9.23 — — — — Lateral ventricles 24.93 ± 4.17 33.08 ± 10.30 — — — — Olfactory bulbs 22.60 ± 1.25 23.30 ± 1.06 — — — — Cerebellum 63.99 ± 0.75 66.51 ± 1.74 — — — — Cortices 158.76 ± 2.58 158.79 ± 3.22 — — — — Hippocampus 25.78 ± 1.26 26.04 ± 0.96 19.80 ± 0.49 19.99 ± 0.52 18.48 ± 0.46 19.56 ± 0.43 AD hippocampus — — 5.96 ± 0.47 5.69 ± 0.56 7.68 ± 0.25 7.57 ± 0.45 CA1 field — — — — 0.946 ± 0.060 0.912 ± 0.067 CA1 pyramidal layer — — — — 0.088 ± 0.004 0.086 ± 0.006 CA1 stratum radiatum — — — — 0.373 ± 0.028 0.365 ± 0.031

Note: Estimated volumes in cubic millimeters are expressed as mean ± SEM (mdx, n = 4; WT, n = 4; for each condition). For anterodorsal (AD) CA1 hippocampal field, CA1 pyramidal cell and apical dendritic (stratum radiatum) layers, the volumes were estimated unilaterally.

Because preserved hippocampal volume is an important factor for reliable estimation of synapse density in EM studies, we further analyzed the volumes of distinct hippocampal subregions in postmortem histological sections from 2 additional groups of mdx and WT mice. In the first group, brains were fixed with 4% paraformaldehyde (frozen-section group; mdx, n = 4; WT, n = 4), whereas in the second group, they were processed as for EM (vibratome-section group; mdx, n = 4; WT, n = 4). As shown in Table 1, hippocampal volumes estimated under such experimental conditions were smaller compared with the in vivo MRI analysis, which likely reflects tissue shrinkage due to the fixation and dehydration processes. However, hippocampal volumes estimated from both frozen and vibratome sections were comparable between genotypes (P > 0.05). The volume of lateral ventricles was not assessed in histological sections, as ventricular spaces are prone to mechanical distortion after sectioning, handling, and coverslipping. The volume of the anterodorsal hippocampus was then determined in frozen sections, as it specifically corresponds to the portion of the hippocampus used for EM studies below. The spatial resolution of MRI measurements was not optimal for this purpose. As shown in Table 1, the volume of the anterodorsal hippocampus was comparable between genotypes in frozen sections (P > 0.05). The volume of the anterodorsal hippocampus was smaller when estimated from vibratome compared with frozen sections, which was likely due to the greater thickness and reduced number of sections containing this hippocampal portion in vibratome sections. However, the lack of difference between genotypes was confirmed in vibratome sections (P > 0.05), confirming with different evaluation methods that hippocampal volume is not altered in mdx mice. Frozen sections were further used to specifically evaluate the volumes of the CA1 subfield, CA1 radiatum, and pyramidal layers where synapse and neuron densities were subsequently investigated. Consistent with the results above, no group differences were detected in any of these hippocampal subregions (P > 0.05; Table 1).

### CA1 Pyramidal Cell Density and Nuclear Morphology

Previous studies in mdx mice suggested significant cell loss and altered shape of pyramidal cells in some brain structures (Sbriccoli et al. 1995; Carretta et al. 2001), but this had not been yet investigated in the hippocampus. Here we compared the numerical density of CA1 pyramidal cells using the disector method in semithin sections of brain tissue prepared for EM analysis (Fig. 2b) in 4 mdx and 4 WT littermate mice. As shown in Table 2, the number of neurons per cubic millimeters was comparable in the 2 groups of mice. In addition to neuron density, we analyzed the mean nuclear area and circularity of pyramidal cells in both genotypes because these parameters have been shown to be altered in apoptotic neurons that can be morphologically identified by neurite retraction, rounder nuclei, cell body, and nucleus shrinkage (Kasischke et al. 2001; Daniel and DeCoster 2004). However, as shown in Table 2, the nuclear area and circularity were not significantly different between genotypes (P > 0.05). The relative neuron packing was also calculated because this parameter, defined as the percentage of area covered by cell nuclei over the sampled area, could reveal alterations in cellular organization independently of neuronal loss (Carretta et al. 2004). No difference was found between mdx and control mice (P > 0.05; Table 2). These results, consistent with the unaltered volume of CA1 pyramidal cell layer, suggest that the lack of dystrophin does not induce neuron loss, changes in cellular organization, or apoptosis-related structural modifications in the anterodorsal hippocampal CA1 region.

Table 2

Pyramidal neuron density and nuclear parameters in CA1 anterodorsal hippocampus

 WT mdx P value Neurons/mm3 165742.45 ± 27264.84 176361.01 ± 7669.50 0.721 Nuclear area (μm2) 63.36 ± 1.86 67.62 ± 4.27 0.395 Area covered by nuclei (%) 26.559 ± 1.269 24.649 ± 1.522 0.372 Nucleus circularity 0.903 ± 0.008 0.908 ± 0.005 0.611
 WT mdx P value Neurons/mm3 165742.45 ± 27264.84 176361.01 ± 7669.50 0.721 Nuclear area (μm2) 63.36 ± 1.86 67.62 ± 4.27 0.395 Area covered by nuclei (%) 26.559 ± 1.269 24.649 ± 1.522 0.372 Nucleus circularity 0.903 ± 0.008 0.908 ± 0.005 0.611

Note: There was no difference between genotypes (WT, n = 4; mdx, n = 4) for these parameters (P values from 1-way ANOVAs).

### Synaptology

#### Tissue Ultrastructure and Qualitative EM Inspection

Tissue was considered perfectly preserved when no signs of extended extracellular spaces, swelling of mitochondria, or selective swelling of dendrites were observed (Tao-Cheng et al. 2007). Qualitative inspection was also performed to evaluate typical signs of synapse pathology associated with neurocognitive disorders, such as distorted spine shape, electron-dense spines, varicosity formation, or hypertrophied spine apparatus (see Fiala et al. 2002). This analysis revealed no sign of synapse pathology in mdx mice. We also used low-power micrographs taken from ultrathin sections after synapse quantification to confirm that the CA1 stratum radiatum did not undergo differential shrinkage between genotypes. Figure 2c shows the electron-beam traces that were used as references to measure the distance from the CA1 pyramidal cell bodies to the limits of the proximal and distal regions. The radiatum length (microns) was comparable in WT (248.23 ± 19.04) and mdx mice (260.11 ± 13.84) (P > 0.05), thus confirming our above histological analysis. The distance (microns) from cell bodies to the limits of the proximal radiatum did not differ between WT (26.39 ± 1.39) and mdx mice (31.32 ± 4.66) (P > 0.05), as was the distance (microns) to the distal limit in WT (207.96 ± 16.33) and mdx mice (199.34 ± 16.73) (P > 0.05). This ensures that quantifications in the proximal and distal radiatum can be performed in comparable conditions among animals and genotypes.

At the ultrastructural level, a range of functionally different synapse subtypes were characterized according to standard morphological features (see “synapse nomenclature” in the Materials and methods). As shown in Table 3, asymmetric excitatory synapses mainly targeted dendritic spines (axospinous synapses) and to a much lesser extent dendritic shafts (axodendritic asymmetric). Symmetric inhibitory synapses were mainly located on dendritic shafts or somata portions and were in general larger (Table 4) and less numerous than asymmetric synapses (Table 3).

Table 3

Relative density and distribution of distinct synapse subtypes in CA1 proximal and distal stratum radiatum of anterodorsal hippocampus

 Proximal Distal WT mdx WT mdx Synapse-to-neuron Asymmetric + symmetric 6919.94 ± 1234.20 6616.39 ± 749.73 8501.60 ± 1365.23 7036.06 ± 509.46 Synapses per 100 μm2 Axospinous nonperforated 18.76 ± 2.12 20.25 ± 1.56 19.64 ± 2.68 18.55 ± 1.06 Axospinous perforated 1.85 ± 0.48 1.72 ± 0.21 3.17 ± 0.19 2.40 ± 0.43 Axodendritic asymmetric 0.48 ± 0.29 0.64 ± 0.15 0.64 ± 0.23 0.77 ± 0.27 Axodendritic symmetric (*) 0.72 ± 0.08 1.03 ± 0.03 0.32 ± 0.08 0.28 ± 0.08 Axosomatic symmetric 0.33 ± 0.16 0.19 ± 0.12 — — MSBs 1.20 ± 0.26 0.91 ± 0.14 1.28 ± 0.49 1.89 ± 0.70 Synapse distribution (%) Asymmetric 94.96 ± 1.08 94.20 ± 0.88 98.31 ± 0.21 98.08 ± 0.46 Symmetric 4.910 ± 1.16 5.190 ± 0.70 1.460 ± 0.09 1.460 ± 0.51 Axospinous 93.08 ± 1.18 91.52 ± 1.10 95.59 ± 1.03 94.45 ± 1.68 Axodendritic 5.290 ± 0.97 7.030 ± 0.77 4.020 ± 0.82 4.970 ± 1.59 Axosomatic 1.490 ± 0.75 0.840 ± 0.51 — — Perforated 8.210 ± 1.64 7.190 ± 0.71 13.66 ± 1.25 10.74 ± 1.59 Nonperforated 86.74 ± 1.72 87.01 ± 1.11 84.65 ± 1.21 87.34 ± 1.25
 Proximal Distal WT mdx WT mdx Synapse-to-neuron Asymmetric + symmetric 6919.94 ± 1234.20 6616.39 ± 749.73 8501.60 ± 1365.23 7036.06 ± 509.46 Synapses per 100 μm2 Axospinous nonperforated 18.76 ± 2.12 20.25 ± 1.56 19.64 ± 2.68 18.55 ± 1.06 Axospinous perforated 1.85 ± 0.48 1.72 ± 0.21 3.17 ± 0.19 2.40 ± 0.43 Axodendritic asymmetric 0.48 ± 0.29 0.64 ± 0.15 0.64 ± 0.23 0.77 ± 0.27 Axodendritic symmetric (*) 0.72 ± 0.08 1.03 ± 0.03 0.32 ± 0.08 0.28 ± 0.08 Axosomatic symmetric 0.33 ± 0.16 0.19 ± 0.12 — — MSBs 1.20 ± 0.26 0.91 ± 0.14 1.28 ± 0.49 1.89 ± 0.70 Synapse distribution (%) Asymmetric 94.96 ± 1.08 94.20 ± 0.88 98.31 ± 0.21 98.08 ± 0.46 Symmetric 4.910 ± 1.16 5.190 ± 0.70 1.460 ± 0.09 1.460 ± 0.51 Axospinous 93.08 ± 1.18 91.52 ± 1.10 95.59 ± 1.03 94.45 ± 1.68 Axodendritic 5.290 ± 0.97 7.030 ± 0.77 4.020 ± 0.82 4.970 ± 1.59 Axosomatic 1.490 ± 0.75 0.840 ± 0.51 — — Perforated 8.210 ± 1.64 7.190 ± 0.71 13.66 ± 1.25 10.74 ± 1.59 Nonperforated 86.74 ± 1.72 87.01 ± 1.11 84.65 ± 1.21 87.34 ± 1.25

Note: Data are expressed as mean ± SEM in the mdx (n = 4) and WT (n = 4) groups.

*P < 0.05 in proximal radiatum.

Table 4

Length of synapse PSDs in proximal and distal radiatum

 Proximal Distal WT mdx WT mdx Axospinous nonperforated 0.242 ± 0.003 0.244 ± 0.003 0.245 ± 0.003 0.250 ± 0.006 Axospinous perforated (*) 0.293 ± 0.017 0.332 ± 0.009 0.358 ± 0.013 0.372 ± 0.004 Axodendritic asymmetric 0.357 ± 0.039 0.328 ± 0.026 0.261 ± 0.042 0.221 ± 0.077 Axodendritic symmetric 0.335 ± 0.030 0.355 ± 0.033 0.429 ± 0.038 0.331 ± 0.029 Axosomatic 0.322 ± 0.040 0.348 ± 0.011 — —
 Proximal Distal WT mdx WT mdx Axospinous nonperforated 0.242 ± 0.003 0.244 ± 0.003 0.245 ± 0.003 0.250 ± 0.006 Axospinous perforated (*) 0.293 ± 0.017 0.332 ± 0.009 0.358 ± 0.013 0.372 ± 0.004 Axodendritic asymmetric 0.357 ± 0.039 0.328 ± 0.026 0.261 ± 0.042 0.221 ± 0.077 Axodendritic symmetric 0.335 ± 0.030 0.355 ± 0.033 0.429 ± 0.038 0.331 ± 0.029 Axosomatic 0.322 ± 0.040 0.348 ± 0.011 — —

Note: PSD length is expressed in microns as mean ± SEM.

*P < 0.05, significant genotype and region effects in 2-way ANOVA.

#### Synapse Density

Synapse density was estimated in both the proximal and distal radiatum of CA1 (mdx, n = 4; WT, n = 4). As shown in Table 3, in the proximal radiatum, the total number of synapses (i.e., asymmetric + symmetric) expressed as synapse-to-neuron ratio was similar in mdx and WT mice (P > 0.05). The density of each synaptic subtype expressed as synapses per 100 μm2 was then compared between genotypes. The density of axosomatic synapses, only present in proximal radiatum, that of axospinous asymmetric excitatory synapses, perforated and nonperforated excitatory synapses, and MSBs were not statistically different between genotypes (P > 0.05). In contrast, mdx mice showed a large (42.85%) and statistically significant increase in the density of axodendritic symmetric inhibitory synapses as compared with the WT controls (P < 0.02) (Fig. 3). This selective change in the number of inhibitory synapses was the only significant change in synapse density detected in mdx mice in the proximal radiatum. The comparable density of perforated synapses and MSBs in the 2 groups of mice (P > 0.05) suggests that the number of excitatory synapses typically associated with activity-dependent plasticity (Geinisman 2000) was not altered by the lack of brain dystrophin in mdx mice.

In distal radiatum, the synapse-to-neuron ratio for total synapses was comparable in the 2 genotypes (P > 0.05), suggesting that dystrophin loss is not associated with a net change in synapse density in either distal or proximal radiatum areas (Table 3). As compared with proximal radiatum, both genotypes showed a reduced percentage of symmetric inhibitory synapses in distal radiatum and a consequent increase in the percentage of asymmetric excitatory synapses. This is in line with the common observation that asymmetric synapses are enriched in the distal stratum radiatum. In this specific region, the density of axodendritic symmetric synapses was comparable between genotypes (P > 0.05). This suggests that the increased density of inhibitory synapses in mdx mice is restricted to the proximal dendrites of CA1 pyramidal cells where dystrophin is normally expressed. As shown in Table 3, the density of axodendritic asymmetric synapses and axospinous nonperforated and perforated synapses did not differ between genotypes (all P >0.05). Finally, although there was a tendency in mdx mice to present an increased density of MSBs, this did not reach statistical significance (P > 0.05).

In all, these results indicate that dystrophin loss in mdx mice is not associated with a net change in synapse density but with a relative and selective increase in the density of axodendritic symmetric inhibitory synapses, which is restricted to the hippocampal CA1 proximal radiatum that normally expresses dystrophin.

#### PSD Length

Alterations in the size of synapses have been linked to different neurodevelopmental disorders (Fiala et al. 2002). We indirectly estimated the size of the different types of synapses by measuring the length of the PSD, a parameter that can be affected by changes in synaptic efficacy, altered plasticity, and mental retardation (Desmond and Levy 1986b; Toni et al. 2001; Boda et al. 2004; Moretti et al. 2006). Table 4 shows the estimated length of each synapse subtype in WT (n = 4) and mdx (n = 4) mice in both the proximal and distal radiatum. On the whole, axodendritic, axosomatic, and perforated synapses presented the largest PSD lengths. The PSD lengths of axosomatic, axodendritic symmetric and asymmetric, and axospinous nonperforated synapses were statistically comparable between genotypes in both proximal and distal radiatum (P > 0.05). However, the PSD length of axospinous perforated synapses was significantly larger in mdx mice compared with WT controls (Fig. 4a). Statistical comparison revealed significant effects of both genotype (P < 0.05) and region (P < 0.001). As shown in Figure 4b, the difference between genotypes was more pronounced in the proximal region although the genotype × region interaction was not significant (P > 0.1). The marked difference between genotypes in proximal radiatum suggests that this change is more closely associated with the region where dystrophin is normally present, as for the change in inhibitory synapse density described above. Interestingly, this enhanced size of PSDs in mdx mice appears to be restricted to the perforated subtype of excitatory synapses that has been consistently associated with activity-dependent changes in synaptic efficacy, suggesting a possible link with the enhanced CA1 hippocampal synaptic plasticity reported in mdx mice (Vaillend et al. 1998, 1999, 2004).

## Discussion

Previous biochemical and electrophysiological studies in mdx mice have highlighted the functional importance of brain dystrophin for the stabilization of postsynaptic membrane GABAA-R clusters containing the α2 subunit in hippocampus (Knuesel et al. 1999) and for the expression of CA1 hippocampal LTP (Vaillend et al. 2004), suggesting a possible role for GABAergic inhibitory processes in the genesis of cognitive deficits in DMD patients. Here we used a combination of in vivo MRI, light, and EM approaches to determine whether and how congenital loss of dystrophin may affect the cellular and synaptic organization of hippocampal networks in mdx mice. We demonstrate that dystrophin deficiency in hippocampus is associated with an increased number of inhibitory synaptic contacts and ultrastructural changes in the PSD of excitatory synapses.

### Dystrophin Loss Does Not Alter Gross Brain Anatomy

The presence of enlarged brain ventricles and slight cortical atrophy has been reported in a subset of DMD patients (Septien et al. 1991), suggesting white matter atrophy or neuronal cell death. Brain imaging and autopsy studies have shown inconsistent structural or volumetric differences in DMD patients compared with healthy controls, and no correlation between ventricle dilation and verbal intelligence could be demonstrated (Al-Qudah et al. 1990), which led to the general conclusion that DMD is not associated with gross anatomical brain abnormalities (for review, see Dubowitz and Crome 1969; Anderson et al. 2002). However, several methodological factors that could be sources of variability or misleading results have been pointed out, such as the small size of experimental groups, the patients’ young age, and the heterogeneity of the mutations that may differentially affect expression of other dystrophin gene products resulting from distinct internal promoters (Lidov 1996). A reduced number of neurons has been reported in the motor cortex of mdx mice (Sbriccoli et al. 1995), which is potentially compatible with cortical atrophy and/or ventricle dilation. However, no systematic study of mdx brain anatomy has been undertaken to confirm this hypothesis. Our present in vivo structural MRI investigation provides the first evidence that dystrophin loss is not associated with significant volumetric changes in the mdx mouse brain. Volumes of the whole brain, ventricles, and selected brain regions that normally express dystrophin, that is, neocortex, hippocampus, and cerebellum, were not significantly modified in mdx compared with WT littermate mice. Detection of subtle alterations in cortical cell number or distribution in specific cortices (Carretta et al. 2003) was beyond the resolution of this MRI analysis. Nevertheless, our results suggest that such alterations linked to the loss of brain dystrophin in mdx mice, although they might contribute to cognitive dysfunction, are not associated with gross brain volumetric abnormalities. Demonstrating the absence of volumetric changes in the hippocampus of this mutant was particularly important as it is a key structure in learning and memory processes (Squire 1992) and as abnormal synaptic plasticity has been reported in this brain area in mdx mice (Vaillend et al. 1998, 1999, 2004). Moreover, it is a prerequisite for reliable estimations of hippocampal neuron and synapse densities. Our histological analyses in both standard formalin-fixed frozen sections and in tissue sections processed as for EM confirmed the results of the MRI study. In addition, detailed volumetric analysis of the anterodorsal hippocampus, CA1 subfield, CA1 pyramidal cell, and dendritic layers did not reveal any difference between genotypes, thus confirming that the gross anatomy of the hippocampus is preserved in mdx mice.

Because previous studies in mdx mice revealed significant cell loss and altered pyramidal cell morphology in the corticospinal tract (Sbriccoli et al. 1995), the question arose as to whether the number of pyramidal cells could also be affected in the hippocampus of this mouse model. To address this question, we first estimated the relative density of CA1 pyramidal cells by histological examination on semithin sections and stereological methods (physical disector) and found no difference between mutant and control mice. This was further confirmed by single-plane quantification performed to provide a relative measure of cell packing (Table 2). Consistent with the unchanged volume of the CA1 pyramidal cell layer in mdx mice, this analysis revealed no differences in neuron density, packing, or cell distribution between genotypes. Moreover, mutant mice displayed no change in neuronal nuclear size and circularity. We therefore conclude that there are no obvious signs of neuronal loss or cellular pathology in the CA1 hippocampal subfield of mdx mice. The results of our MRI study further support the idea that neuron loss in mdx mice cannot be generalized to all cortices. Moreover, such changes do not seem to occur in the anterodorsal part of the hippocampus investigated here, suggesting that the neuronal substrate of altered synaptic plasticity in mdx mice (Vaillend et al. 2004) is not at the cellular level but likely relies on altered synaptic or subcellular mechanisms.

### Increased Density of Axodendritic Inhibitory Synapses onto Dystrophin-Deficient Pyramidal Neurons: A Compensatory Mechanism?

How a congenital loss of dystrophin and the resulting reduction in number and size of GABAA-R clusters in mdx mice (Knuesel et al. 1999) may impact on the organization of hippocampal synaptic networks and possibly mediate behavioral and cognitive dysfunctions? Cognitive function depends on the plastic nature of the nervous system expressed through its capacity to modify the number, proportion, shape, and size of synapses in an activity-dependent manner. Regulation of cytoskeleton dynamics and membrane receptor trafficking are largely involved in such processes (Matus 1999; Marrone and Petit 2002), and dysfunctions of these mechanisms underlie various forms of inherited mental retardation associated with changes in dendritic arbor complexity, synapse number, and ultrastructure (for reviews, see Fiala et al. 2002; Vaillend et al. 2008). We therefore investigated the ultrastructure and density of a variety of synapse subtypes in the hippocampus of mdx and WT littermate mice.

An initial qualitative inspection of hippocampal tissue did not reveal clear signs of synapse pathology in mdx mice, in contrast to other mouse models of neurodevelopmental disorders (Fiala et al. 2002). We thus performed a thorough quantitative EM study using serial ultrathin sections and stereological methods to estimate the relative density of CA1 hippocampal synapses. The volume of CA1 stratum radiatum, the global synapse density, and the synapse-to-neuron ratio were unaffected in mdx mice, suggesting that dystrophin loss does not alter the basic processes of early synaptogenesis. Examination of the relative distribution of distinct synapse subtypes, however, revealed a selective and significant increase in the number of axodendritic symmetric inhibitory synapses in mdx mice, suggesting a major reorganization of inhibitory circuits as a consequence of impaired GABAA-R clustering.

Reorganization of GABAergic synapses has been observed in several animal models characterized by a compromised balance between excitation and inhibition, such as in models of temporal lobe epilepsy (Knuesel et al. 2001) and in the mouse barrel cortex after increasing excitatory transmission by whisker stimulation (Knott et al. 2002). This indicates that structural plasticity at GABAergic synapses can occur to maintain excitatory/inhibitory balance (Liu 2004). The mechanisms underlying these compensatory responses are not fully understood. They may involve changes in the distribution of interneurons targeting principal cells and/or upregulation of different GABAA-R subunits and of the scaffolding protein gephyrin, as shown in the dentate gyrus after medial entorhinal cortex lesion (Simbürger et al. 2001) and in specific cerebellar and hippocampal interneurons of the α1 subunit–null mouse (Fritschy et al. 2006; Schneider Gasser et al. 2007). In the hippocampus of mdx mice, increased inhibitory synapse density was selectively found in proximal radiatum in which dystrophin is enriched and colocalizes with α2 GABAA-R subunits, but not in the distal radiatum where dystrophin and GABAA-R subunits are not detected (Knuesel et al. 1999, 2000). This high selectivity toward the proximal dendritic layer of CA1 suggests that it could rely on the reorganization of specific subpopulations of interneurons targeting proximal dendrites of pyramidal cells (Freund 2003), which would be reminiscent of the altered distribution of neurons expressing calcium-binding proteins in some cortical areas in the mdx mouse (Carretta et al. 2003). This also strengthens our conclusion of a rearrangement of inhibitory circuits that could correspond to a homeostatic response in an attempt to maintain inhibitory drive onto dystrophin-deficient principal neurons.

Whereas dystrophin seems dispensable for initiation of GABAA-R clustering, it is believed to be involved in the stabilization of postsynaptic GABAA-R complexes through interaction of its N-terminal domain with actin cytoskeleton, probably by limiting the lateral diffusion of GABAA-Rs outside the synapse (Brünig et al. 2002; Lévi et al. 2002). The functional role of such a mechanism, whether it is important for basal inhibitory transmission or for some form of activity-dependent plasticity at GABAergic synapses remains to be elucidated (Fritschy and Brünig 2003). In hippocampus of mdx mice, the reduced clustering of α2 subunits associated with dystrophin loss does not affect inhibitory postsynaptic currents (IPSCs) evoked in the pyramidal cell layer (Graciotti et al. 2008), perhaps due to a preserved perisomatic inhibition mediated by α1-containing receptors (Freund 2003). In contrast, we previously showed that field excitatory postsynaptic potentials (EPSPs) elicited and recorded in CA1 stratum radiatum are less sensitive to the GABAA-R antagonist bicuculline in mdx compared with control mice (Vaillend and Billard 2002), which suggests a reduction of GABAA-R–mediated transmission in dendrites. Moreover, we found facilitated NMDAr activation and abnormal enhancement of CA1 hippocampal LTP in mdx mice (Vaillend et al. 1999, 2004) and showed that these alterations are sensitive to bicuculline, thus suggesting a link with reduced dendritic inhibition. Strikingly, our present data demonstrate an increased density of axodendritic, not axosomatic, inhibitory synapses in CA1 hippocampal area of mdx mice. Such a specific rearrangement of inhibitory contacts may therefore reflect a compensatory mechanism against these alterations and also points to a major role of dystrophin and α2 subunit clusters at the dendritic rather than the somatic level in hippocampal pyramidal cells. In any event, if in the absence of dystrophin the increase in axodendritic inhibitory synapse density represents an attempt to compensate an altered inhibitory drive in dendrites, the electrophysiological data suggest that this process may not lead to full compensation.

### Altered GABA Function and Enhanced Hippocampal LTP in mdx Mice: What Is the Missing Link?

How can GABAergic alterations reported in mdx mice lead to abnormally enhanced CA1 hippocampal synaptic plasticity? First, reports have shown that GABAA-R antagonists applied on hippocampal (Vaillend and Billard 2002) and cerebellar slices (Anderson et al. 2003) are less potent to modulate the strength of excitatory synaptic responses in mdx than in control mice. Second, mdx mice show impairments in paired-pulse facilitation of CA1 IPSCs (Graciotti et al. 2008), suggesting impaired adaptability and plasticity of inhibitory responses to repetitive stimulations. Moreover, the facilitated induction of CA1 hippocampal LTP in mdx mice can be prevented by a GABAA-R antagonist (Vaillend and Billard 2002). In all, these converging data suggest that neuronal disinhibition plays a key role in the regulation of synaptic plasticity in dystrophin-deficient neurons. This may involve reduced high-frequency stimulation (HFS)–induced feedforward inhibition, which can normally suppress LTP in the mature brain by reducing the time window for EPSPs summation and synaptic integration (Mittmann et al. 2004; Akerman and Cline 2007), although the contribution of nonsynaptic components of the GABA-mediated response evoked by HFS cannot be ruled out (Smirnov et al. 1999). The distribution of dystrophin along proximal dendrites, the dendritic reorganization of inhibitory synapses, and apparent reduction of dendritic GABAA-R–mediated transmission in mdx mice suggest a specific role of α2-containing GABAA-R clusters in signal integration in dendrites. Interestingly, spike backpropagation from axon to dendrites has been proposed to enhance dendritic depolarization and thereby facilitate synaptic plasticity (Mehta 2004). This may be normally modulated by various processes including local inhibition mediated by GABA (Tsubokawa and Ross 1996). Thus, it seems conceivable that reduced dendritic GABAergic inhibition in mdx mice may facilitate dendritic depolarization and thereby enhance synaptic plasticity.

If a deficit in dendritic inhibitory control can facilitate strengthening of excitatory synapses in mdx mice, could this find a morphological correlate at the synapse level? Changes in the efficacy of synaptic transmission are commonly associated with modifications of several synaptic-contact parameters including the size of the pre- and postsynaptic elements and the density or proportion of different synapse populations (Marrone and Petit 2002). In the present study, we quantified the incidence of perforated excitatory synapses and MSBs, which represent morphological states associated with adult synaptogenesis and enhanced synaptic efficacy (for reviews, see Geinisman 2000; Marrone and Petit 2002; Nikonenko et al. 2002). These 2 parameters were not affected in mdx mice. However, we found a significant and selective increase in PSD length of axospinous perforated excitatory synapses in mdx mice. Again, this difference was more pronounced in the proximal radiatum normally enriched in dystrophin than in the distal radiatum, thus supporting a causal role for dystrophin loss. Because perforated synapses are hallmarks of activity-dependent synaptic plasticity, this may represent a morphological correlate of the enhanced LTP and facilitated activation of NMDAr reported in mdx mice (Vaillend et al. 1999). Indeed, it is believed that modifications in the size of the PSD occur as a result of neurotransmitter receptors aggregation and changes in the density of various membrane proteins involved in signal transduction. For instance, NMDAr-dependent synaptic facilitation in CA1 pyramidal cells after ischemic episodes is associated with increased phosphorylation of NMDAr subunits 2A and 2B and increased PSD size due to recruitment of various signal transduction molecules to the PSD (Cheung et al. 2000; Martone et al. 2000). Moreover, proportional augmentation in PSD size and synaptic strength has been associated with LTP (Desmond and Levy 1986b; Toni et al. 2001), and conversely, a reduction in the size of perforated PSDs has been reported in the hippocampus of aged, memory-deficient rats (Nicholson et al. 2004), suggesting a positive correlation between PSD length and cognitive function. This assumption is further supported by the report of a decrease in PSD size along with impaired cognitive and synaptic functions in several animal models of mental retardation (Boda et al. 2004; Moretti et al. 2006). In mdx mice, the change in PSD length was specific to perforated synapses, suggesting a selective, and perhaps transitory, role for dystrophin and associated proteins in the formation of new excitatory synapses in adult brain. This is consistent with a model in which dystrophin, because of its late expression during development, may not be involved during the early stages of synaptogenesis, as suggested elsewhere (Fritschy and Brünig 2003). How the stabilization of GABAergic PSDs by dystrophin complexes may influence processes of activity-dependent remodeling of excitatory synapses, however, remains an open question for future studies.

### Concluding Remarks

One main finding in the present study is the demonstration that dystrophin loss not only impairs clustering of specific GABAA-R subunits but also impacts on the general organization of inhibitory synapses and on the ultrastructure of a specific excitatory synapse subtype. Our results suggest that the lack of dystrophin in the CA1 hippocampal subfield is associated with a reorganization of dendritic inhibitory contacts reflecting tenuous compensation of altered GABAA-R clustering. Increase in the PSD length of excitatory perforated synapses suggests that dystrophin deficiency also affects the functional organization and/or remodeling of glutamatergic synapses. We believe that these morphological changes may explain the abnormal electrophysiological properties of hippocampal CA1 neurons found in mdx mice. Functional brain defects resulting from dystrophin loss therefore appear to be mainly expressed at the level of synaptic connectivity in the absence of gross anatomical abnormalities.

Many lines of electrophysiological data led to a functional model according to which dendritic inhibition may control the efficacy of excitatory synaptic inputs onto principal cells and their plasticity (Freund 2003). Recent work in the collybistin-deficient mouse, a genetic model displaying a specific loss of GABAA-R clusters in the hippocampus, has revealed reduced dendritic GABAergic transmission and enhanced synaptic plasticity associated with impaired spatial learning (Papadopoulos et al. 2007). Although mdx mice only display a partial loss (about 50%) of α2 subunit clusters (Knuesel et al. 1999) and mild cognitive impairments (Vaillend et al. 1995), both models support a role for dendritic GABAA-R clustering in regulating activity-dependent synaptic plasticity and cognitive functions. Our current knowledge of DMD physiopathology (for reviews, see Lidov 1996; Blake and Kröger 2000; Anderson et al. 2002) suggests that dystrophin loss may not account for the most severe cases of mental retardation, which can be due to further brain anomalies resulting from the additional loss of other short dystrophin gene products in patients carrying C-terminal mutations (Moizard et al. 1998). However, because all patients lack the full-length dystrophin as is the case of the mdx mice, defective GABAergic function secondarily affecting synaptic plasticity at pyramidal cell glutamatergic synapses may be considered a common phenotypic feature associated with the DMD syndrome. This may open the search for effective treatments with therapeutic agents modulating dendritic GABAA-R function.

## Funding

AFM (Association Française contre les Myopathies, France, Grant # DdT1 2006) to C.V., from OSR (Oficina de Suport a la Recerca, Universitat de les Illes Balears, Spain) to R.M. and by CNRS (Centre National de la Recherche Scientifique, France).

The authors are grateful to S. Vandergeenst, N. Samson, and P. Veyrac for animal care. We also thank C. Rampon for fruitful advice, M. Guegan for her collaboration in the histology, and L. Damas for her participation to preliminary EM experiments. Conflict of Interest: None declared.

## References

Akerman
CJ
Cline
HT
Refining the roles of GABAergic signaling during neural circuit formation
Trends Neurosci
,
2007
, vol.
30
(pg.
382
-
389
)
Al-Qudah
AA
Kobayashi
J
Chuang
S
Dennis
M
Ray
P
Etiology of intellectual impairment in Duchenne muscular dystrophy
Pediatr Neurol
,
1990
, vol.
6
(pg.
57
-
59
)
Amaral
DG
Witter
MP
The three-dimensional organization of the hippocampal formation: a review of anatomical data
Neuroscience
,
1989
, vol.
31
(pg.
571
-
591
)
Anderson
JL
SI
Morley
JW
Altered inhibitory input to Purkinje cells of dystrophin-deficient mice
Brain Res
,
2003
, vol.
982

2
(pg.
280
-
283
)
Anderson
JL
SI
Rae
C
Morley
JW
Brain function in Duchenne muscular dystrophy
Brain
,
2002
, vol.
125
(pg.
4
-
13
)
Billard
C
Gillet
P
Signoret
JL
Uicaut
E
Bertrand
P
Fardeau
M
Barthez-Carpentier
MA
Santini
JJ
Cognitive functions in Duchenne muscular dystrophy: a reappraisal and comparison with spinal muscular atrophy
Neuromuscul Disord
,
1992
, vol.
2
(pg.
371
-
378
)
Blake
DJ
Kröger
S
The neurobiology of duchenne muscular dystrophy: learning lessons from muscle?
Trends Neurosci
,
2000
, vol.
23
(pg.
92
-
99
)
Boda
B
Alberi
S
Nikonenko
I
Node-Langlois
R
Jourdain
P
Moosmayer
M
Parisi-Jourdain
L
Muller
D
The mental retardation protein PAK3 contributes to synapse formation and plasticity in hippocampus
J Neurosci
,
2004
, vol.
24
(pg.
10816
-
10825
)
Brünig
I
Suter
A
Knuesel
I
Lüscher
B
Fritschy
JM
GABAergic terminals are required for postsynaptic clustering of dystrophin but not of GABA(A) receptors and gephyrin
J Neurosci
,
2002
, vol.
22
(pg.
4805
-
4813
)
Carretta
D
Santarelli
M
Sbriccoli
A
Pinto
F
Catini
C
Minciacchi
D
Spatial analysis reveals alterations of parvalbumin- and calbindin-positive local circuit neurons in the cerebral cortex of mutant mdx mice
Brain Res
,
2004
, vol.
1016
(pg.
1
-
11
)
Carretta
D
Santarelli
M
Vanni
D
Carrai
R
Sbriccoli
A
Pinto
F
Minciacchi
D
The organisation of spinal projecting brainstem neurons in an animal model of muscular dystrophy. A retrograde tracing study on mdx mutant mice
Brain Res
,
2001
, vol.
895
(pg.
213
-
222
)
Carretta
D
Santarelli
M
Vanni
D
Ciabatti
S
Sbriccoli
A
Pinto
F
Minciacchi
D
Cortical and brainstem neurons containing calcium-binding proteins in a murine model of Duchenne's muscular dystrophy: selective changes in the sensorimotor cortex
J Comp Neurol
,
2003
, vol.
456
(pg.
48
-
59
)
Cheung
HH
Takagi
N
Teves
L
Logan
R
Wallace
MC
Gurd
JW
Altered association of protein tyrosine kinases with postsynaptic densities after transient cerebral ischemia in the rat brain
J Cereb Blood Flow Metab
,
2000
, vol.
20
(pg.
505
-
512
)
Colonnier
M
Synaptic pattern on different cell types in the different laminae of the cat visual cortex. An electron microscope study
Brain Res
,
1968
, vol.
9
(pg.
268
-
287
)
Daniel
B
DeCoster
MA
Quantification of sPLA2-induced early and late apoptosis changes in neuronal cell cultures using combined TUNEL and DAPI staining
Brain Res Brain Res Protoc
,
2004
, vol.
13
(pg.
144
-
150
)
DeCoster
M
Méndez-Vilas
A
Díaz
J
The nuclear area factor (NAF): a measure for cell apoptosis using microscopy and image analysis
Modern research and educational topics in microscopy
,
2007
Formatex
(pg.
378
-
384
)
DeFelipe
J
Marco
P
Busturia
I
Merchán-Pérez
A
Estimation of the number of synapses in the cerebral cortex: methodological considerations
Cereb Cortex
,
1999
, vol.
9
(pg.
722
-
732
)
DeFelipe
J
Marco
P
Fairén
A
Jones
EG
Inhibitory synaptogenesis in mouse somatosensory cortex
Cereb Cortex
,
1997
, vol.
7
(pg.
619
-
634
)
Desmond
NL
Levy
WB
Changes in the numerical density of synaptic contacts with long-term potentiation in the hippocampal dentate gyrus
J Comp Neurol
,
1986
, vol.
253
(pg.
466
-
475
)
Desmond
NL
Levy
WB
Changes in the postsynaptic density with long-term potentiation in the dentate gyrus
J Comp Neurol
,
1986
, vol.
253
(pg.
476
-
482
)
Dubowitz
V
Crome
L
The central nervous system in Duchenne muscular dystrophy
Brain
,
1969
, vol.
92
(pg.
805
-
808
)
Eyre
MD
Richter-Levin
G
Avital
A
Stewart
MG
Morphological changes in hippocampal dentate gyrus synapses following spatial learning in rats are transient
Eur J Neurosci
,
2003
, vol.
17
(pg.
1973
-
1980
)
Fiala
JC
Harris
KM
Stuart
G
Spruston
N
Häusser
M
Dendrite structure
Dendrites
,
1999
New York
Oxford University Press
(pg.
1
-
34
)
Fiala
JC
Spacek
J
Harris
KM
Dendritic spine pathology: cause or consequence of neurological disorders?
Brain Res Rev
,
2002
, vol.
39
(pg.
29
-
54
)
Freund
TF
Interneuron diversity series: rhythm and mood in perisomatic inhibition
Trends Neurosci
,
2003
, vol.
26
(pg.
489
-
495
)
Fritschy
JM
Brünig
I
Formation and plasticity of GABAergic synapses: physiological mechanisms and pathophysiological implications
Pharmacol Ther
,
2003
, vol.
98
(pg.
299
-
323
)
Fritschy
JM
Panzanelli
P
Kralic
JE
Vogt
KE
Sassoè-Pognetto
M
Differential dependence of axo-dendritic and axo-somatic GABAergic synapses on GABAA receptors containing the alpha1 subunit in Purkinje cells
J Neurosci
,
2006
, vol.
26
(pg.
3245
-
3255
)
Geinisman
Y
Structural synaptic modifications associated with hippocampal LTP and behavioral learning
Cereb Cortex
,
2000
, vol.
10
(pg.
952
-
962
)
Gorecki
D
Geng
Y
Thomas
K
Hunt
SP
Barnard
EA
Barnard
PJ
Expression of the dystrophin gene in mouse and rat brain
Neuroreport
,
1991
, vol.
2
(pg.
773
-
776
)
Graciotti
L
Minelli
A
Minciacchi
D
Procopio
A
Fulgenzi
G
GABAergic miniature spontaneous activity is increased in the CA1 hippocampal region of dystrophic mdx mice
Neuromuscul Disord
,
2008
, vol.
18
(pg.
220
-
226
)
Gundersen
HJ
Jensen
EB
The efficiency of systematic sampling in stereology and its prediction
J Microsc
,
1987
, vol.
147
(pg.
229
-
263
)
Hinton
VJ
De Vivo
DC
Nereo
NE
Goldstein
E
Stern
Y
Selective deficits in verbal working memory associated with a known genetic etiology: the neuropsychological profile of duchenne muscular dystrophy
J Int Neuropsychol Soc
,
2001
, vol.
7
(pg.
45
-
54
)
Hopf
FW
Steinhardt
RA
Regulation of intracellular free calcium in normal and dystrophic mouse cerebellar neurons
Brain Res
,
1992
, vol.
578
(pg.
49
-
54
)
V
Becker
LE
Brain morphology in Duchenne muscular dystrophy: a Golgi study
Pediatr Neurol
,
1988
, vol.
4
(pg.
87
-
92
)
Kasischke
K
Büchner
M
Ludolph
AC
Riepe
MW
Nuclear shrinkage in live mouse hippocampal slices
Acta Neuropathol
,
2001
, vol.
101
(pg.
483
-
490
)
Kim
TW
Wu
K
Black
IB
Deficiency of brain synaptic dystrophin in human Duchenne muscular dystrophy
Ann Neurol
,
1995
, vol.
38
(pg.
446
-
449
)
Kim
TW
Wu
K
Xu
JL
Black
IB
Detection of dystrophin in the postsynaptic density of rat brain and deficiency in a mouse model of Duchenne muscular dystrophy
,
1992
, vol.
89
(pg.
11642
-
11644
)
Knott
GW
Quairiaux
C
Genoud
C
Welker
E
Formation of dendritic spines with GABAergic synapses induced by whisker stimulation in adult mice
Neuron
,
2002
, vol.
34
(pg.
265
-
273
)
Knuesel
I
Bornhauser
BC
Zuellig
RA
Heller
F
Schaub
MC
Fritschy
JM
Differential expression of utrophin and dystrophin in CNS neurons: an in situ hybridization and immunohistochemical study
J Comp Neurol
,
2000
, vol.
422
(pg.
594
-
611
)
Knuesel
I
Mastrocola
M
Zuellig
RA
Bornhauser
B
Schaub
MC
Fritschy
JM
Altered synaptic clustering of GABAA receptors in mice lacking dystrophin (mdx mice)
Eur J Neurosci
,
1999
, vol.
11
(pg.
4457
-
4462
)
Knuesel
I
Zuellig
RA
Schaub
MC
Fritschy
JM
Alterations in dystrophin and utrophin expression parallel the reorganization of GABAergic synapses in a mouse model of temporal lobe epilepsy
Eur J Neurosci
,
2001
, vol.
13
(pg.
1113
-
1124
)
Lee
JS
Pfund
Z
Juhász
C
Behen
ME
Muzik
O
Chugani
DC
Nigro
MA
Chugani
HT
Altered regional brain glucose metabolism in Duchenne muscular dystrophy: a pet study
Muscle Nerve
,
2002
, vol.
26
(pg.
506
-
512
)
Lévi
S
RM
Henry
MD
Campbell
KP
Sanes
JR
Craig
AM
Dystroglycan is selectively associated with inhibitory GABAergic synapses but is dispensable for their differentiation
J Neurosci
,
2002
, vol.
22
(pg.
4274
-
4285
)
Lidov
HG
Dystrophin in the nervous system
Brain Pathol
,
1996
, vol.
6
(pg.
63
-
77
)
Lidov
HG
Byers
TJ
Kunkel
LM
The distribution of dystrophin in the murine central nervous system: an immunocytochemical study
Neuroscience
,
1993
, vol.
54
(pg.
167
-
187
)
Lidov
HG
Byers
TJ
Watkins
SC
Kunkel
LM
Localization of dystrophin to postsynaptic regions of central nervous system cortical neurons
Nature
,
1990
, vol.
348
(pg.
725
-
728
)
Liu
G
Local structural balance and functional interaction of excitatory and inhibitory synapses in hippocampal dendrites
Nat Neurosci
,
2004
, vol.
7
(pg.
373
-
379
)
Marrone
DF
Petit
TL
The role of synaptic morphology in neural plasticity: structural interactions underlying synaptic power
Brain Res Rev
,
2002
, vol.
38
(pg.
291
-
308
)
Martone
ME
Hu
BR
Ellisman
MH
Alterations of hippocampal postsynaptic densities following transient ischemia
Hippocampus
,
2000
, vol.
10
(pg.
610
-
616
)
Matus
A
Postsynaptic actin and neuronal plasticity
Curr Opin Neurobiol
,
1999
, vol.
9
(pg.
561
-
565
)
Mayhew
TM
How to count synapses unbiasedly and efficiently at the ultrastructural level: proposal for a standard sampling and counting protocol
J Neurocytol
,
1996
, vol.
25
(pg.
793
-
804
)
Megías
M
Emri
Z
Freund
TF
Gulyás
AI
Total number and distribution of inhibitory and excitatory synapses on hippocampal CA1 pyramidal cells
Neuroscience
,
2001
, vol.
102
(pg.
527
-
540
)
Mehler
MF
Haas
KZ
Kessler
JA
Stanton
PK
Enhanced sensitivity of hippocampal pyramidal neurons from mdx mice to hypoxia-induced loss of synaptic transmission
,
1992
, vol.
89
(pg.
2461
-
2465
)
Mehta
MR
Cooperative LTP can map memory sequences on dendritic branches
Trends Neurosci
,
2004
, vol.
27
(pg.
69
-
72
)
Mittmann
W
P
Häusser
M
Neuronal microcircuits: frequency-dependent flow of inhibition
Curr Biol
,
2004
, vol.
14
(pg.
R837
-
R839
)
Moizard
MP
Billard
C
Toutain
A
Berret
F
Marmin
N
Moraine
C
Are Dp71 and Dp140 brain dystrophin isoforms related to cognitive impairment in Duchenne muscular dystrophy?
Am J Med Genet
,
1998
, vol.
80
(pg.
32
-
41
)
Moretti
P
Levenson
JM
Battaglia
F
Atkinson
R
Teague
R
Antalffy
B
Armstrong
D
Arancio
O
Sweatt
JD
Zoghbi
HY
Learning and memory and synaptic plasticity are impaired in a mouse model of Rett syndrome
J Neurosci
,
2006
, vol.
26
(pg.
319
-
327
)
Muntoni
F
Mateddu
A
Serra
G
Passive avoidance behaviour deficit in the mdx mouse
Neuromuscul Disord
,
1991
, vol.
1
(pg.
121
-
123
)
Nicholson
DA
Yoshida
R
Berry
RW
Gallagher
M
Geinisman
Y
Reduction in size of perforated postsynaptic densities in hippocampal axospinous synapses and age-related spatial learning impairments
J Neurosci
,
2004
, vol.
24
(pg.
7648
-
7653
)
Nikonenko
I
Jourdain
P
Alberi
S
Toni
N
Muller
D
Activity-induced changes of spine morphology
Hippocampus
,
2002
, vol.
12
(pg.
585
-
591
)
T
Korte
M
Eulenburg
V
Kubota
H
Retiounskaia
M
Harvey
RJ
Harvey
K
O'Sullivan
GA
Laube
B
Hülsmann
S
, et al.  .
Impaired GABAergic transmission and altered hippocampal synaptic plasticity in collybistin-deficient mice
EMBO J
,
2007
, vol.
26
(pg.
3888
-
3899
)
Paxinos
G
Franklin
KBJ
The mouse brain in stereotaxic coordinates
,
2001
2nd ed.
San Diego (CA)
Rosen
GD
William
AG
Capra
A
Connoly
MT
Cruz
B
2000
, vol.
14
pg.
166

The mouse brain library. Available at: www.mbl.org. Int Mouse Genome Conf
Sbriccoli
A
Santarelli
M
Carretta
D
Pinto
F
Granato
A
Minciacchi
D
Architectural changes of the cortico-spinal system in the dystrophin defective mdx mouse
Neurosci Lett
,
1995
, vol.
200
(pg.
53
-
56
)
Schneider Gasser
EM
Duveau
V
Prenosil
GA
Fritschy
JM
Reorganization of GABAergic circuits maintains GABAA receptor-mediated transmission onto CA1 interneurons in alpha1-subunit-null mice
Eur J Neurosci
,
2007
, vol.
25
(pg.
3287
-
3304
)
Sebrié
C
Chabert
C
Ledru
A
Guedj
F
Po
C
Smith
DJ
Rubin
E
Rivals
I
Beloeil
JC
Gillet
B
, et al.  .
Increased dosage of DYRK1A and brain volumetric alterations in a YAC model of partial trisomy 21
Anat Rec (Hoboken)
,
2008
, vol.
291
(pg.
254
-
262
)
Septien
L
Gras
P
Borsotti
JP
Giroud
M
Nivelon
JL
Dumas
R
Mental development in Duchenne muscular dystrophy. Correlation of data of the brain scanner
Pediatrie
,
1991
, vol.
46
(pg.
817
-
819
)
Shi
L
Linville
MC
Tucker
EW
Sonntag
WE
Brunso-Bechtold
JK
Differential effects of aging and insulin-like growth factor-1 on synapses in CA1 of rat hippocampus
Cereb Cortex
,
2005
, vol.
15
(pg.
571
-
577
)
Simbürger
E
Plaschke
M
Fritschy
JM
Nitsch
R
Localization of two major GABA(A) receptor subunits in the dentate gyrus of the rat and cell type-specific up-regulation following entorhinal cortex lesion
Neuroscience
,
2001
, vol.
102
(pg.
789
-
803
)
Smirnov
S
Paalasmaa
P
Uusisaari
M
Voipio
J
Kaila
K
Pharmacological isolation of the synaptic and nonsynaptic components of the GABA-mediated biphasic response in rat CA1 hippocampal pyramidal cells
J Neurosci
,
1999
, vol.
19
(pg.
9252
-
9260
)
Squire
LR
Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans
Psychol Rev
,
1992
, vol.
99
(pg.
195
-
231
)
Sterio
DC
The unbiased estimation of number and sizes of arbitrary particles using the disector
J Microsc
,
1984
, vol.
134
(pg.
127
-
136
)
Tao-Cheng
JH
Gallant
PE
Brightman
MW
Dosemeci
A
Reese
TS
Structural changes at synapses after delayed perfusion fixation in different regions of the mouse brain
J Comp Neurol
,
2007
, vol.
501
(pg.
731
-
740
)
Toni
N
Buchs
PA
Nikonenko
I
Povilaitite
P
Parisi
L
Muller
D
Remodeling of synaptic membranes after induction of long-term potentiation
J Neurosci
,
2001
, vol.
21
(pg.
6245
-
6251
)
Tsubokawa
H
Ross
WN
IPSPs modulate spike backpropagation and associated [Ca2+]i changes in the dendrites of hippocampal CA1 pyramidal neurons
J Neurophysiol
,
1996
, vol.
76
(pg.
2896
-
2906
)
Vaillend
C
Billard
JM
Facilitated CA1 hippocampal synaptic plasticity in dystrophin-deficient mice: role for GABAA receptors?
Hippocampus
,
2002
, vol.
12
(pg.
713
-
717
)
Vaillend
C
Billard
JM
Claudepierre
T
Rendon
A
Dutar
P
Ungerer
A
Spatial discrimination learning and CA1 hippocampal synaptic plasticity in mdx and mdx3cv mice lacking dystrophin gene products
Neuroscience
,
1998
, vol.
86
(pg.
53
-
66
)
Vaillend
C
Billard
JM
Laroche
S
Impaired long-term spatial and recognition memory and enhanced CA1 hippocampal LTP in the dystrophin-deficient mdx mouse
Neurobiol Dis
,
2004
, vol.
17
(pg.
10
-
20
)
Vaillend
C
Poirier
R
Laroche
S
Genes, plasticity and mental retardation
Behav Brain Res
,
2008
, vol.
192

1
(pg.
88
-
105
)
Vaillend
C
Rendon
A
Misslin
R
Ungerer
A
Influence of dystrophin-gene mutation on mdx mouse behavior. I. Retention deficits at long delays in spontaneous alternation and bar-pressing tasks
Behav Genet
,
1995
, vol.
25
(pg.
569
-
579
)
Vaillend
C
Ungerer
A
Billard
JM
Facilitated NMDA receptor-mediated synaptic plasticity in the hippocampal CA1 area of dystrophin-deficient mice
Synapse
,
1999
, vol.
33
(pg.
59
-
70
)
Weibel
ER
Stereological methods. Practical methods for biological morphometry
,
1979
New York