Excitatory-inhibitory relationship in the fascia dentata in the Ts65Dn mouse model of down syndrome

Authors

  • Pavel V. Belichenko,

    Corresponding author
    1. Department of Neurology & Neurological Sciences, the Center for Research and Treatment of Down Syndrome and Neuroscience Institute at Stanford University, Stanford University Medical Center, Stanford, California 94305-5489
    • Neuroscience Institute at Stanford University, Stanford University Medical Center, 1201 Welch Road, Room P220, Stanford, CA 94305-5489
    Search for more papers by this author
  • Alexander M. Kleschevnikov,

    1. Department of Neurology & Neurological Sciences, the Center for Research and Treatment of Down Syndrome and Neuroscience Institute at Stanford University, Stanford University Medical Center, Stanford, California 94305-5489
    Search for more papers by this author
  • Eliezer Masliah,

    1. Department of Neurosciences, University of California San Diego, School of Medicine, La Jolla, California 92093-0624
    2. Department of Pathology, University of California San Diego, School of Medicine, La Jolla, California 92093-0624
    Search for more papers by this author
  • Chengbiao Wu,

    1. Department of Neurology & Neurological Sciences, the Center for Research and Treatment of Down Syndrome and Neuroscience Institute at Stanford University, Stanford University Medical Center, Stanford, California 94305-5489
    Search for more papers by this author
  • Ryoko Takimoto-Kimura,

    1. Department of Neurology & Neurological Sciences, the Center for Research and Treatment of Down Syndrome and Neuroscience Institute at Stanford University, Stanford University Medical Center, Stanford, California 94305-5489
    Search for more papers by this author
  • Ahmad Salehi,

    1. Department of Neurology & Neurological Sciences, the Center for Research and Treatment of Down Syndrome and Neuroscience Institute at Stanford University, Stanford University Medical Center, Stanford, California 94305-5489
    Search for more papers by this author
  • William C. Mobley

    1. Department of Neurology & Neurological Sciences, the Center for Research and Treatment of Down Syndrome and Neuroscience Institute at Stanford University, Stanford University Medical Center, Stanford, California 94305-5489
    Search for more papers by this author

Abstract

Down syndrome (DS) is a neurological disorder causing impaired learning and memory. Partial trisomy 16 mice (Ts65Dn) are a genetic model for DS. Previously, we demonstrated widespread alterations of pre- and postsynaptic elements and physiological abnormalities in Ts65Dn mice. The average diameter of presynaptic boutons and spines in the neocortex and hippocampus was enlarged. Failed induction of long-term potentiation (LTP) due to excessive inhibition was observed. In this paper we investigate the morphological substrate for excessive inhibition in Ts65Dn. We used electron microscopy (EM) to characterize synapses, confocal microscopy to analyze colocalization of the general marker for synaptic vesicle protein with specific protein markers for inhibitory and excitatory synapses, and densitometry to characterize the distribution of the receptor and several proteins essential for synaptic clustering of neurotransmitter receptors. EM analysis of synapses in the Ts65Dn vs. 2N showed that synaptic opposition lengths were significantly greater for symmetric synapses (∼18%), but not for asymmetric ones. Overall, a significant increase in colocalization coefficients of glutamic acid decarboxylase (GAD)65/p38 immunoreactivity (IR) (∼27%) and vesicular GABA transporter (VGAT)/p38 IR (∼41%) was found, but not in vesicular glutamate transporter 1 (VGLUT1)/p38 IR. A significant overall decrease of IR in the hippocampus of Ts65Dn mice compared with 2N mice for glutamate receptor 2 (GluR2; ∼13%) and anti-γ-aminobutyric acid (GABA)A receptor β2/3 subunit (∼20%) was also found. The study of proteins essential for synaptic clustering of receptors revealed a significant increase in puncta size for neuroligin 2 (∼13%) and GABAA receptor-associated protein (GABARAP; ∼13%), but not for neuroligin 1 and gephyrin. The results demonstrate a significant alteration of inhibitory synapses in the fascia dentata of Ts65Dn mice. J. Comp. Neurol. 512:453–466, 2009. © 2008 Wiley-Liss, Inc.

A balance between excitatory and inhibitory neurotransmission is essential for the normal function of neuronal circuits (Chen,2004; Hensch and Fagiolini,2005; Trevelyan and Watkinson,2005; Akerman and Cline,2007). A significant imbalance, whether of excitation or inhibition, manifests as circuit dysfunction in a variety of episodic disorders, and may feature prominently in disorders due to acute or chronic injury or neurodegeneration. Imbalance of excitatory/inhibitory neurotransmission is linked to epilepsy, pain, parkinsonism, schizophrenia, bipolar disorder, autism, etc. (Benes and Berretta,2001; Rubenstein and Merzenich,2003; Rippon et al.,2007; Zold et al.,2007). Down syndrome (DS), due to trisomy chromosome 21, manifests as a variety of neurological abnormalities in both children and adults. Earlier work suggested that DS was characterized by an impaired balance between excitatory and inhibitory systems (Reynolds and Warner,1988; Risser et al.,1997; but see Seidl et al.,2001). This has been supported by recent studies in mouse models of DS, in which increased inhibition in the hippocampus was shown to be responsible for failed induction of long-term potentiation (LTP) (Kleschevnikov et al.,2004). Measurements made in slice preparations of the hippocampus indicate reduced synaptic plasticity through a marked reduction in LTP in the CA1 area (Siarey et al.,1997,1999; Galdzicki et al.,2001) of Ts65Dn mice and fascia dentata (FD) of Ts65Dn and Ts1Cje mice (Kleschevnikov et al.,2004; Belichenko et al.,2007). Rescue of apparently normal LTP by treating slices with picrotoxin, an inhibitor of γ-aminobutyric acid (GABA)A receptors, pointed to an imbalance of neurotransmission manifested through increased inhibition. Raising the possibility that failed synaptic plasticity is linked to behavioral changes, Ts65Dn mice show spatial working and reference working memory impairments (Reeves et al.,1995; Demas et al.,1998; Escorihuela et al.,1998; Sago et al.,2000; Belichenko et al.,2007).

To explore further the molecular and cellular basis for increased inhibitory neurotransmission, we examined structural and biochemical features of synapses in the FD of Ts65Dn mice. We considered the following possibilities: that inhibition was increased to a greater extent than excitation, that inhibition was normal but excitation decreased, or that both were decreased but with more marked changes in excitation. In view of earlier findings in the FD, we favored the first alternative. Here we demonstrate that inhibition is selectively increased in Ts65Dn mice, as revealed by changes in ultrastructural, immunofluorescence, and biochemical markers.

MATERIALS AND METHODS

Mice husbandry

All experiments were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals, and with an approved animal protocol from the Stanford University Institutional Animal Care and Use Committee. The Ts65Dn mouse colony was maintained for more than 10 generations by crossing B6EiC3Sn-Ts(1716)65Dn females (Jackson Laboratory, Bar Harbor, ME) with B6EiC3Sn F1/J A/a males (Jackson Laboratory). This breeding scheme was used because trisomic mice breed very poorly or not at all when inbred; the B6C3 background has been the most successful. To distinguish 2N from Ts65Dn mice, genomic DNA was extracted from tail samples. A quantitative polymerase chain reaction (PCR) protocol (provided by the Jackson Laboratory) was used to measure Mx1 gene expression, which is present in three copies in Ts65Dn. Male mice were used in all studies.

EM processing and analysis

As previously described, ultrastructural analysis of synaptic architecture was performed on 6-month-old mice (Belichenko et al.,2004). In brief, mice were deeply anesthetized with sodium pentobarbital (200 mg/kg i.p.; Abbott Laboratories, North Chicago, IL) and transcardially perfused with ice-cold 0.9% sodium chloride for 1 minute, followed by ice-cold 4% paraformaldehyde and 1% glutaraldehyde in 0.1 cacodylate buffer, pH 7.4, for 10 minutes. After perfusion, the brains were immediately removed and coronally sectioned (300 μm) on a Vibratome and immersed in fixative. Sections were postfixed in 1% osmium tetroxide, stained with saturated uranyl acetate in 50% ethanol, and then dehydrated through a graded series of ethanols to 90% ethanol; 2-hydroxypropyl methacrylate was the intermediate solvent. All infiltrations of 2-hydroxypropyl methacrylate and Scipoxy 812 resin (Energy Beam Sciences, Agawam, MA) were carried out on a shaker at slow speed. After two changes of 100% resin, the plates were polymerized in a 65°C oven for 24 hours. The plastic was detached, and selected areas were cut and glued onto dummy blocks. Thin sections (80 nm) were cut on a Reichert Ultracut E Ultramicrotome (Leica, Vienna, Austria), picked up onto 200-mesh copper grids (Electron Microscopy Sciences, Fort Washington, PA), and poststained in ethanolic uranyl acetate followed by bismuth nitrite (Electron Microscopy Sciences).

Sections were analyzed with a Zeiss EM10 electron microscope (Zeiss, New York, NY). For morphometric analysis of synapses, five electron micrographs from each mouse were obtained at a final magnification of ×10,000. The area of each image was 61 μm2, and the total area studied for each animal was 305 μm2. Layer II–III of the motor cortex and the molecular layer of the superior blade of the FD were studied. The electron micrographs were digitized by using Duoscan F40 (Agfa, Mortsel, Belgium). By using well-established criteria for symmetric and asymmetric synapses (Gray,1959; Kurt et al.,2000), we counted the number of each type of synapse on electron microscopy (EM) images. Synapses with prominent postsynaptic density were identified as asymmetric, whereas synapses with pre- and postsynaptic densities of equal thickness were considered as symmetric. Synapse length was defined as the length of the active zone. The EM image was imported in LazerPix software (Bio-Rad, Hertfordshire, UK), a line was drawn along each active zone by using a mouse cursor, and its length was measured.

Indirect immunofluorescence

To study colocalization of synaptophysin (p38) with markers for inhibitory or excitatory synapses, slices were submitted to double-labeling immunohistochemistry. In brief, 2N and Ts65Dn mice were deeply anesthetized with sodium pentobarbital (200 mg/kg i.p.; Abbott Laboratories) and transcardially perfused for 1 minute with 0.9% sodium chloride (10 ml), and then for 10 minutes with 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS), pH 7.4 (100 ml) for immunofluorescent methods. After perfusion, the brains were removed immediately and placed in the same fixative for 2–3 days. The brain was then sectioned coronally at 100 μm with a Vibratome (series 1000, TPI, St. Louis, MO), and sections were placed in 0.1 M PBS. Free-floating coronal sections through the cortex and hippocampus from 3-month-old mice were preincubated in 5% nonfat milk in PBS and then incubated overnight at 4°C with mouse monoclonal anti-synaptophysin (i.e., p38; dilution 1:500) as the first primary antibody. The sections were then rinsed in PBS (20 minutes, three changes) and incubated for 1 hour at room temperature with biotinylated goat anti-mouse IgG (1:200; Jackson ImmunoResearch, West Grove, PA). After rinsing with PBS (20 minutes, three changes), sections were incubated with fluorescein isothiocyanate (FITC)-conjugated streptavidin (1:500; Jackson ImmunoResearch) for 1 hour at room temperature.

After careful rinsing, the sections were incubated overnight at 4°C with one of the following second primary antibodies: rabbit anti-glutamic acid decarboxylase 65 (GAD 65; 1:1,000); rabbit anti-vesicular glutamate transporter 1 (VGLUT1; Bellocchio et al.,2000; dilution 1:1,000); or rabbit anti-vesicular GABA transporter (VGAT; Chaudhry et al.,1998; dilution 1:1,000). Sections were then rinsed in PBS (20 minutes, three changes) and incubated for 1 hour at room temperature with Texas red-conjugated donkey anti-rabbit secondary antibody (1:200; Jackson ImmunoResearch). Following further careful rinsing, the sections were mounted on microscope glass slides and coverslipped with 90% glycerol in phosphate buffer (PB).

For single antibody labeling, sections from 2N and Ts65Dn mice were preincubated in 5% nonfat milk in PBS and then incubated overnight at 4°C with the primary antibody. The sections were rinsed in PBS (20 minutes, three changes) and incubated for 1 hour at room temperature with species-appropriate biotinylated secondary antibodies (1:200; Jackson ImmunoResearch). After rinsing with PBS (20 minutes, three changes), sections were incubated with FITC-conjugated streptavidin (1:500; Jackson ImmunoResearch) for 1 hour at room temperature. Following further careful rinsing, the sections were mounted on microscope glass slides and coverslipped with 90% glycerol in PB. Optimal antibody concentrations were determined for each antibody; in control experiments, omitting primary antibodies eliminated immunoreactivity.

Antibody characterization

The following text and Table 1 provide information on all antibodies used in this study. All antibodies were previously characterized by Western blot analysis and immunohistochemistry and are commercially available.

Table 1. Primary Antibodies Used in This Study
AntibodyManufacturerCat. and lot no.SpeciesExact structure of immunizing antigen
Anti-synaptophysin (p38)Boehringer, Ingelheim, Germany902314, lot 12530025-03MouseVesicular fraction of bovine brain
Anti-glutamate decarboxylase 65 (GAD65)Chemicon, Temecula, CAAB5082, lot 24051184RabbitHuman GAD65 from baculovirus-infected cells
Anti-vesicular γ-aminobutyric acid (GABA) transporter (VGAT)Chaudhry et al.,1998 RabbitGST fusion protein containing N-terminal 99 amino acids of rat VGAT
Anti-vesicular glutamate transporter 1 (VGLUT1)Bellocchio et al., 1998 RabbitGST fusion protein containing last 68 amino acids (residues 493–560) of rat-specific Na+-dependent inorganic phosphate transporter
Anti-glutamate receptor 2 (GluR2), clone 6C4ChemiconMAB397, lot 22041088MouseRecombinant fusion protein TrpE-GluR2 (N-terminal portion, amino acids 175–430 of rat GluR2
Anti-glutamate receptor 1 (GluR1)ChemiconAB1504, lot 22090149RabbitCarboxy-terminus peptide of rat GluR1 conjugated to bovine serum albumin with glutaraldehyde (SHSSGMPLGATGL)
Anti-GABAA receptor β2/3, clone 62-3G1Upstate Biotechnology, Lake Placid, NY05-474, lot 17280MouseAffinity-purified GABAA receptor isolated from bovine brain
Anti-GABAA receptor α1Upstate Biotechnology06-868, lot 22589RabbitSynthetic peptide (QPSQDELKDNTTVFT-C) corresponding to amino acids 1–15 of the mature rat GABAA receptor α1 subunit containing a cysteine at the C-terminus for linkage to a carrier protein
Anti-GABAB receptor R1ChemiconAB5850, lot 23050396RabbitAmino acids 33–54 of rat GABAB R1 protein (PHLPRPHPRVPPHPSSERRAVY)
Anti-GABAB receptor R2, amino acids 42–54 ratChemiconAB5848, lot 23010140RabbitAmino acids 42–54 of rat 105-kDa GABAB R2 protein
Anti-neuroligin, clone 4F9Synaptic Systems, Gottingen, Germany129011, lot 129011/10MouseRecombinant protein containing the extracellular sequence (residues 1–695) of neuroligin 1
Anti-neuroligin 1, clone 4C12Synaptic Systems129111, lot 129111/1MouseRecombinant protein containing the extracellular sequence (residues 1–695) of neuroligin 1
Anti-neuroligin 2 (R-16)Santa Cruz Biotechnology, Santa Cruz, CAsc-14089, lot k1403GoatAmino acid peptide (KGGPLLPTAGRELPPEE) of neuroligin 2 of rat origin
Anti-GABARAPAbcam, Cambridge, MAab1398-50, lot 201420RabbitSynthetic peptide (RSEGEKIRKKYPDRVPV) corresponding to amino acids 15–31 of human gabarap
Anti-gephyrin, clone mAb7a (GlyR7a)Synaptic Systems147011, lot 147011/15MousePurified rat gephyrin
  • 1The synaptophysin (i.e., p38; dilution 1:500) antibody recognized a single band at ∼38 kDa on Western blot of rat brain lysate; immunoadsorption of this antibody with a 0.1% Triton X-100 extract from purified rat brain presynaptic vesicles gave negative results.
  • 2The glutamic acid decarboxylase 65 (GAD65; dilution 1:1,000) antibody was purchased from Chemicon (Temecula, CA). According to the manufacturer's product information, this antibody reacts strongly with GAD65-containing nerve terminals and recognizes a 65-kDa protein corresponding to GAD65 by Western blot of mouse brain extract.
  • 3The VGAT (Chaudhry et al.,1998; dilution 1:1,000) antibody was produced by immunizing rabbits with a GST fusion protein containing N-terminal 99 amino acids of rat VGAT (see more details in Chaudhry et al.,1998). On Western blots of brain extracts from rat brain, this antibody recognized a single 55–60-kDa, band in agreement with the molecular weight calculated from the amino acid composition. There was no staining of liver or control PC12 cells under the same conditions; however, VGAT-expressing PC12 cells show a band with similar molecular mass.
  • 4The VGLUT1 (Bellocchio et al.,2000; dilution 1:1,000) antibody was produced by immunizing rabbits with a GST fusion protein containing the last 68 amino acids (residues 493–560) of rat brain-specific Na+-dependent inorganic phosphate transporter (see more details in Bellocchio et al.,2000); preadsorption of this antibody with the GST fusion protein used as the immunogen completely abolishes immunoreactivity in brain sections.
  • 5The GluR2 (dilution 1:500) antibody on Western blots of brain extracts from mice recognizes a ∼102-kDa band corresponding to full-length GluR2; another ∼66-kDa band appears to be a breakdown product of GluR2. Preincubation of brain sections with the fusion protein GluR2 entirely blocked the immunoreactivity of this antibody.
  • 6The GluR1 (dilution 1:1,000) antibody had no cross-reaction with GluR2–4. According to the manufacturer's product information, Western blot analysis of rat brain extracts shows a single band comigrating with GluR1 (105–107 kDa) expressed in transfected cells. Our data on mouse brain extract also showed a single 105–107-kDa band (Supplementary Table 2). Specificity of this antibody was proved by using Western blot analysis of membranes from COS-7 cells transfected with the GluR1 cDNAs. Specific staining by this antibody on a Western blot of the soluble fraction from rat brain membranes was completely eliminated by preincubation of the antibody with the immunizing peptide.
  • 7The GABAA receptor β2/3 subunit (dilution 1:500) antibody was purchased from Upstate Biotechnology (Lake Placid, NY). According to the manufacturer's product information, this antibody was directed against affinity-purified GABAA receptor isolated from bovine brain. Western blot of rat brain microsomal preparations showed a 55-kDa band for β2 and a 57-kDa band for β3 subunits. Our data on mouse brain extract also showed identical bands (Supplementary Table 2).
  • 8The GABAA receptor α1 subunit (dilution 1:500) antibody recognized a single band of 51 kDa for the α1 subunit on Western blot of rat brain microsomal preparations (manufacturer's datasheet), and our data on mouse brain extract showed an identical band (Supplementary Table 2).
  • 9The GABAB receptor R1 subunit (dilution 1:100) antibody recognized a band of ∼130 kDa corresponding to the molecular weight for the biggest splice variant GABAB receptor R1a subunit on Western blot of rat brain membrane preparation. Mouse brain extract showed an identical band (Supplementary Table 2). As a control, primary antibody that was preabsorbed with their peptide antigen gave no specific staining.
  • 10The GABAB receptor R2 subunit (dilution 1:100) antibody recognized a single band of ∼105 kDa by Western blot of whole rat brain membrane preparations (manufacturer's datasheet), and our data on mouse brain extract showed an identical band (Supplementary Table 2). By using this antibody, we observed a staining pattern of immunofluorescent distribution in the mouse hippocampus that was identical with that of a previous report (Fritschy et al.,1999).
  • 11The neuroligin (clone 4F9, dilution 1:1,000) antibody recognizes mouse neuroligins 1, 2, and 3, as determined by Western blotting of cell lysates from HEK293 cells transfected with neuroligin 1, 2, or 3 cDNAs. Transfecting of hippocampal neurons in culture with lentiviruses delivering small-hairpin RNA (shRNA) for individual neuroligin isoforms reduced neuroligin expression in an isoform-specific manner; coinfection with shRNAs for all three neuroligin isoforms, completely abolished staining on Western blot.
  • 12The neuroligin 1 (clone 4C12, dilution 1:1,000) antibody recognizes mouse neuroligin 1; no cross reactivity was observed with neuroligin 2 or with neuroligin 3 (manufacture's datasheet). Transfection of hippocampal neurons in culture with lentiviruses delivering shRNA for neuroligin 1 reduced neuroligin 1 expression on Western blot.
  • 13The neuroligin 2 (dilution 1:1,000) antibody recognizes a 93-kDa band on Western blot of EOC 20 whole cell lysates (manufacturer's datasheet), and our data on mouse brain extract showed an identical band (Supplementary Table 2). Transfection of hippocampal neurons in culture with lentiviruses delivering shRNA for neuroligin 2 reduced neuroligin 2 expression with >90% efficiency in single cells and was confirmed by the absence of immunostaining when using this antibody.
  • 14The GABAA receptor-associated protein (GABARAP; dilution 1:500) antibody recognizes a single 14-kDa band on Western blot of human brain lysate (manufacturer's datasheet). Transfection of hippocampal neuronal cultures with small interfering RNA against GABARAP significantly reduced dendritic GABARAP fluorescence intensity (Marsden et al.,2007).
  • 15The gephyrin (clone mAb7a, dilution 1:500) antibody recognizes mouse, rat, human, pig, and goldfish gephyrin. This antibody detects only the brain-specific 93-kDa splice variant on crude synaptic membrane fractions of rat brain (manufacturer's datasheet) and displayed a staining pattern and distribution in the mouse hippocampus that was identical with those of a previous report (Kralic et al.,2006).

Confocal microscopy

Confocal imaging of brain slices labeled with one or two fluorophores was performed as previously described (Belichenko et al.,2004). Sections were scanned in a Radiance 2000 confocal microscope (Bio-Rad) by using a Nikon Eclipse E800 fluorescence microscope. The laser was an argon/krypton mixed gas laser with exciting wavelengths for FITC (488 λ) and Texas Red (568 λ). The optimal condition for confocal imaging of single immunoreactivity (IR) was the following: the lens was a ×20 objective (Nikon; Plan Apo x20/0.75); LaserSharp 2000 (Bio-Rad) software was used; laser power was 10%; the zoom factor was 3; and 500 fast speed scanning was employed. Each optical section was the result of three scans by Kalman filtering, the size of the image was 512 × 512 pixels (i.e., 246 × 246 μm). Dual-channel confocal microscopy was employed to study colocalization of p38 IR with markers for inhibitory or excitatory synapses. The optimal conditions for this method were: the lens was a ×60 objective (Plan Apo 60x/1.40 oil); LaserSharp 2000 (Bio-Rad) software was used; laser power was 10% for “green” and 50% for “red” channels; the zoom factor was 6; scanning was at 500 lps; a sequential mode of scanning was employed to reduced “bleed-through”; each optical section was the result of three scans with Kalman filtering; and image size was 512 × 512 pixels (i.e., 34 × 34 μm).

Image acquisition and analysis

For quantitative analysis of p38 IR and VGLUT, GAD65, or VGAD IR colocalization LazerPix colocalization software (Bio-Rad) was used. Two slices per mouse were used; 20 images from both the left and right side for each mouse (with n = 3, for a total 60 values) were analyzed. For each image, the intensity thresholds were estimated by analyzing the distribution of pixel intensities in the image areas that did not contain IR. This value, the background threshold, was then subtracted, and red-green colocalization coefficients were calculated.

Quantitative analysis of IR for neurotransmitter receptors and for synaptic proteins was carried out on single optical images by using Scion software (Scion, Frederick, MD). Two slices per mouse were used; 6 images from the left and right side from each mouse (with n = 6, for a total 36 values) were analyzed. For every area, background fluorescence was estimated on sections incubated without primary antibody (i.e., the optical density [OD] background). This value was then subtracted from the corresponding section stained with primary antibody to obtain the OD of immunofluorescence.

Gene expression and Western blot studies

Real-time PCR and Western blot were applied to measure gene expression and protein levels in the hippocampus (see details in Supplemental Material and Methods). For gene expression studies, five 2N and five Ts65Dn mice aged 3 months were used, as described in Supplemental Material and Methods. For Western blot, six 2N and five Ts65Dn mice aged 4 months were used, as described in Supplemental Material and Methods.

Statistical analyses

The data for synapse number and size, colocalization of p38 IR with markers for inhibitory or excitatory synapses, gene expression, protein measurement by Western blot, and receptor and synaptic proteins were exported to Excel (Microsoft, Redmond, WA), and statistical comparisons were performed by using two-way analysis of variance (ANOVA) and for two samples by using two-tailed Student's t-test. All results are expressed as mean ± SEM, and P values < 0.05 were considered significant.

RESULTS

We analyzed the morphology of excitatory and inhibitory synapses by using several approaches: 1) EM for morphometry of asymmetric and symmetric synapses; 2) double immunofluorescence for quantitative analysis of the coefficient of colocalization; and 3) densitometry of immunofluoresecence for quantitative analysis of receptor distributions, and the number and size of receptor clusters.

Selective changes in the ultrastructure of symmetric synapses in the Ts65Dn FD

By using established criteria for defining symmetric and asymmetric synapses (Gray,1959; Kurt et al.,2000), we counted the number of each type of synapse on EM images (Fig. 1a,b). In the FD, the total density of synapses was not significantly different between 2N and Ts65Dn mice (Fig. 1c; P = 0.77). Also, there was no significant difference in the ratio of asymmetric/symmetric synapses (Fig. 1d; P = 0.34). However, the average synaptic apposition length of synapses was significantly greater in Ts65Dn vs. 2N mice (Fig. 1e; P < 0.001. Most interesting was the finding that the synaptic apposition lengths were significantly greater for symmetric (P < 0.001), but not for asymmetric synapses (P = 0.99). The change appears to affect the entire population of symmetric synapses. The frequency distribution histogram shows a shift to larger sizes of the entire population of symmetric synapses, with no change in the distribution of asymetric synapses (Fig. 1f). These findings are evidence for neurotransmitter-selective changes in synaptic structure in the Ts65Dn FD.

Figure 1.

Quantitative analysis of EM images in molecular layer of fascia dentata. EM image of asymmetric (a) and symmetric (b) synapses; arrows denote the synaptic apposition measured. The area density of all synapses (c) and ratio of symmetric/asymmetric synapses (d) in 2N and Ts65Dn mice. The mean synaptic apposition length of all synapses combined and asymmetric or symmetric synapses separated (e) and their frequency distribution (f). Results are mean ± SEM. Number of mice examined: 2N = 4, Ts65Dn = 4. *, P < 0.05, significantly different from 2N mice. Scale bar = 0.5 μm in b (applies to a,b).

Increased colocalization of p38 IR with markers for inhibitory synapses in the Ts65Dn FD

To characterize further excitatory-inhibitory relationships in the FD, we used double-labeling immunofluorescence with anti-p38 antibody, to mark all synapses, and with markers for excitatory (anti-VGLUT1 antibody) or inhibitory synapses (anti-GAD-65 or anti-VGAT antibodies). Comparing 2N and Ts65Dn mice, and focusing first on p38 IR, the density of individual puncta of p38 IR was unchanged overall (2N: 2.01 ± 0.11 puncta/μm2, Ts65Dn: 2.04 ± 0.08 puncta/μm2, n = 3; P = 0.84). However, in the FD of Ts65Dn mice, the area occupied by p38 IR increased significantly by an average of 28% (P = 0.01), and the size of p38-IR puncta increased by an average of 21% (2N: 0.067 ± 0.007 μm2, Ts65Dn: 0.081 ± 0.002 μm2, n = 3; P = 0.04).

To test the idea that the change in p38 IR in the Ts65Dn FD was due to selective changes in inhibitory synapses, we extended the quantitative analysis to examine colocalization of p38 IR with markers for excitatory or inhibitory synapses. There was no difference in the average colocalization coefficient for VGLUT1 IR/p38 IR (2N: 0.72 ± 0.02, Ts65Dn: 0.79 ± 0.03, n = 3; P = 0.09). The density of individual puncta of VGLUT1 IR was overall the same in the FD (2N: 2.05 ± 0.08 puncta/μm2, Ts65Dn: 1.98 ± 0.07 puncta/μm2, n = 3; P = 0.55). These findings provided no evidence for changes in the size of excitatory synapses in the FD of Ts65Dn mice.

A different picture emerged in the quantitative analysis of p38 IR with inhibitory markers. The coefficient of colocalization in the molecular layer showed a significant overall increase in GAD65 IR/p38 IR (Fig. 2; P = 0.01). The density of individual GAD65-IR puncta did not differ (2N: 1.05 ± 0.08 puncta/μm2, Ts65Dn: 1.13 ± 0.06 puncta/μm2, n = 3; P = 0.44). Similarly, examining p38 IR and VGAT IR colocalization, the coefficient of colocalization in the molecular layer of the FD showed a significant overall increase in VGAT IR/p38 IR in Ts65Dn (2N: 0.22 ± 0.02, Ts65Dn: 0.31 ± 0.01, n = 3; P < 0.01). Finally, the density of individual VGAT-IR puncta did not differ (2N: 1.89 ± 0.07 puncta/μm2, Ts65Dn: 1.74 ± 0.14 puncta/μm2, n = 3; P = 0.36). We conclude that the increase in colocalization with p38 IR of GAD65 IR and VGAT IR, but not with VGLUT1 IR, reflects a change in inhibitory synapses in the FD of Ts65Dn mice. They are further evidence for a neurotransmitter system-selective change.

Figure 2.

Colocalization of synaptophysin (p38, green) and GAD65 (magenta) in the middle molecular layer of fascia dentata in 3-month-old 2N (a,b) and Ts65Dn (c,d) mice. b and d represent only colocalization (white) between these two markers. Note increased GAD65 to p38 colocalization in Ts65Dn (d) vs. 2N (b) mice. On average, the coefficient of colocalization of GAD65/p38 IR was ∼30% higher in the molecular layer of the fascia dentata in Ts65Dn vs. 2N mice (e). Results are mean ± SEM. Number of mice examined: 2N = 3, Ts65Dn = 3. *, P < 0.05, significantly different from 2N mice. Scale bar = 5 μm in d (applies to a–d).

Gene expression and immunoblotting studies of receptors and synaptic proteins

Changes in the structure of inhibitory synapses predicted changes in the expression of genes for neurotransmitter receptors and synaptic proteins that mediate inhibitory neurotransmission. We examined inhibitory and excitatory receptors and functionally related peptides in sections from the hippocampus in 2N and Ts65Dn mice. As measured at the mRNA level, the expression of the genes for GABAA receptors (subunits β2, β/3, and α1) and GABAB receptors (subunits R2 and R1) did not differ (Supplementary Table 1). We also detected no changes in the mRNA levels for the excitatory neurotransmitter receptors GluR1 and GluR2 (Supplementary Table 1). Four proteins essential for synaptic clustering of neurotransmitter receptors were also studied. Neuroligin 1 is present at excitatory synapses where, as a cell adhesion protein, it plays a role in postsynaptic formation and remodeling (Chih et al.,2005). At inhibitory synapses, neuroligin 2 appears to play the same or a similar role (Chih et al.,2005). Gephyrin is a microtubule-associated bridging protein for inhibitory glycine receptors (Meier and Grantyn,2004). GABARAP is GABAA receptor-associated protein that interacts with the γ2 subunit of GABAA receptors (Wang and Olsen,2000). We detected no change in the expression of any of these genes at the mRNA level (Supplementary Table 1).

Next, we asked whether changes could be detected at the protein level. Hippocampal lysates from 2N and Ts65Dn mice were examined by immunoblotting. When we loaded equal amounts of total protein, no significant differences were detected (Supplementary Table 2). These findings provided no evidence for marked changes in overall levels of several gene products that mediate excitatory and inhibitory neurotransmission in Ts65Dn hippocampus.

Immunolocalization of neurotransmitter receptors and synaptic proteins in the hippocampus

Because studies of gene expression examined the entire hippocampus, we reasoned that changes in specific regions of the hippocampus may have gone undetected. To study differences in the regional distribution of receptors, we used single-labeling immunofluorescence for markers of excitatory glutamate receptors (anti-GluR1 and anti-GluR2 antibody) or inhibitory synapses (anti-GABAA receptors subunits β2/3 or α1 and GABAB receptors subunits R2 or R1 antibodies) and quantified the IR by using confocal microscopy. The same strategy was used to examine the distribution of the synaptic proteins gephyrin, GABARAP, and neuroligins 1 and 2 in the molecular layer of the FD at age 6–7 months (Figs. 4, 5).

Glutamate receptors.

We investigated the distribution of glutamate receptors (GluR1 and GluR2) in both blades of the dentate gyrus and in the CA1 of Ts65Dn and 2N mice. There was a significant overall reduction of GluR2 immunostaining (Fig. 3a–c) in Ts65Dn mice aged 3 months (Table 2). Reductions were found in the FD (the granular cell layer, the inferior blade, and the inner and middle molecular layers of the superior blade), hilus, and stratum lacunosum moleculare (Table 2). In Ts65Dn mice at age 8 months, the differences in GluR2 receptor IR reached significance in the superior blade of the FD and in the stratum lacunosum moleculare layer of the CA1 (Table 2). Comparing the two ages, the changes in the superior blade and CA1 were more marked in older mice.

Figure 3.

Confocal images of GluR2-IR (a,b) and GABAA subunit β2/3 IR (d,e) from the hippocampus in 3-month-old 2N and Ts65Dn mice. Note decreased optical density of GluR2-IR and GABAA subunit β2/3 IR in Ts65Dn compared with 2N mice. On average, optical densities were ∼13% lower for GluR2 IR (c) and ∼20% lower for GABAA subunit β2/3-IR (f) in the fascia dentata in Ts65Dn vs. 2N mice. Results are mean ± SEM. Number of mice examined: 2N = 6, Ts65Dn = 5. *, P < 0.05, significantly different from 2N mice. GCL, granular cell layer; ML, molecular layer. Scale bar = 100 μm in e (applies to a,b,d,e).

Table 2. Quantitative Analysis of GluR2 and GluR1 Receptors in 3- and 8-Month-Old Ts65Dn and 2N Mice Hippocampus1
AreaGluR2GluR1GluR2/GluR1 ratio at 3 monthsGluR2/GluR1 ratio at 8 months
3 months old8 months old3:8- Month ratio3 months old8 months old3:8- Month ratio
% from 2NP value% from 2NP value% from 2NP value% from 2NP value
  • 1

    The number of mice used was as follows: 2N/Ts65Dn = 3/3 for 3-month-old mice and 3/4 for 8-month-old mice. ML, molecular layer. Bold-face type represents significant difference in the parameters studied.

Fascia dentata, inferior blade            
 Inner ML82.4 ± 3.60.00481.3 ± 8.20.0891.0189.1 ± 5.50.16782.6 ± 5.70.0541.080.920.98
 Middle ML80.4 ± 4.80.01392.0 ± 17.10.3370.8885.7 ± 5.40.10785.9 ± 5.50.1151.000.941.07
 Outer ML72.6 ± 5.10.00586.7 ± 18.00.2560.8483.4 ± 4.20.04483.5 ± 6.60.0651.000.871.04
 Granular layer84.1 ± 4.00.00591.8 ± 8.80.2570.9286.4 ± 6.90.13684.3 ± 10.50.1481.020.971.09
Fascia dentata, superior blade            
 Inner ML82.6 ± 3.50.00173.8 ± 4.30.0041.1287.1 ± 6.10.12485.0 ± 2.70.0741.030.950.87
 Middle ML91.1 ± 4.50.02475.1 ± 3.20.0091.2188.9 ± 8.40.19686.8 ± 4.20.1471.021.030.87
 Outer ML90.9 ± 9.30.09875.9 ± 4.70.0091.2090.9 ± 9.20.267103.0 ± 7.50.4300.881.000.74
 Granular layer86.4 ± 2.80.03490.8 ± 10.90.2460.9588.0 ± 9.40.13791.8 ± 8.70.2650.960.980.99
Hilus, neuropil87.0 ± 3.00.00586.2 ± 3.00.0901.0187.5 ± 5.60.05376.5 ± 5.20.0401.140.991.13
CA1 area of hippocampus            
 Stratum oriens93.1 ± 4.80.07479.0 ± 6.10.0351.1890.1 ± 2.60.09790.0 ± 6.20.1821.001.030.88
 Stratum pyramidale96.0 ± 5.70.23380.5 ± 11.00.0401.1991.7 ± 6.40.22182.4 ± 5.80.0721.111.050.98
 Stratum radiatum93.3 ± 5.80.14076.1 ± 3.80.0451.2387.0 ± 3.30.07387.0 ± 6.30.1001.001.070.87
 Stratum lacunosum moleculare85.3 ± 2.60.02070.7 ± 5.60.0071.2182.4 ± 3.70.02979.9 ± 4.80.0181.031.040.88

Significant reduction of GluR1 receptor IR was detected in the outer molecular layer of the inferior blade and stratum lacunosum molecular of Ts65Dn mice aged 3 months, and in the hilus and stratum lacunosum molecular of these mice aged 8 months (Table 2). However, most values for GluR1 receptor IR in Ts65Dn mice were not significantly different from 2N mice at either 3 or 8 months (Table 2), nor could we detect an apparent change in the intensity of IR between 3 and 8 months.

When we examined the ratio of immunostaining for GluR2/GluR1, significant negative deviations from 1.0 were detected in the outer molecular layer of the inferior blade at 3 months, in the molecular layer of the superior blade, and in the stratum radiatum of the CA1 at 3 month; a significantly increased ratio was detected for the hilus at 8 months.

GABAA receptors.

We investigated the distribution of GABAA receptors (subunits β2/3 and α1) in Ts65Dn and 2N mice aged 3 and 8 months (Table 3). At age 3 months, the overall IR of GABAA subunits β2/3 was reduced in Ts65Dn mice (Fig. 3d–f); decreases reached significance in all regions of the inferior blade of the FD, in the inner molecular layer of the superior blade of the FD, in the hilus, and in all strata of the CA1 (Table 3). In contrast, we detected no significant changes in the amount of β2/3 at age 8 months. Although the lack of significant changes at this age relative to 2N mice can be ascribed in part to increased variability of the measurements, the most salient change was a marked increase in the overall level of IR (Table 3). Indeed, the overall IR of GABAA β2/3 in Ts65Dn mice was elevated from 80% at 3 months to 99% at 8 months vs. 2N (P < 0.01). Thus, IR for β2/3 subunits appears to have increased considerably between 3 and 8 months (Table 3).

Table 3. Quantitative Analysis of GABAA Receptor Subunits β2/3 and α1 in 3- and 8-Month-Old Ts65Dn and 2N Mice Hippocampus1
Areaβ2/3α1β2/3/α1 ratio at 3 monthsβ2/3/α1 ratio at 8 months
3 months old8 months old3:8- Month ratio3 months old8 months old3:8- Month ratio
% from 2NP value% from 2NP value% from 2NP value% from 2NP value
  • 1

    The number of mice used was as follows: 2N/Ts65Dn = 3/3 for 3 month-old mice and 3/4 for 8-month-old mice. ML, molecular layer. Bold-face type represents significant difference of parameters studied.

Fascia dentata, inferior blade            
 Inner ML75.1 ± 3.50.003106.0 ± 14.20.4580.7193.9 ± 4.70.165109.1 ± 6.90.3560.860.800.97
 Middle ML75.8 ± 2.70.001105.4 ± 15.90.4440.7296.5 ± 4.20.294112.6 ± 5.20.2650.860.790.94
 Outer ML67.8 ± 2.40.001100.7 ± 17.10.2790.6797.3 ± 4.60.314110.1 ± 5.60.3120.880.700.91
 Granular layer75.9 ± 4.60.040116.2 ± 8.90.2360.65106.2 ± 10.30.111104.6 ± 14.10.3991.020.711.11
Fascia dentata, superior blade            
 Inner ML82.4 ± 2.10.03488.1 ± 10.50.1600.9495.6 ± 4.50.265104.6 ± 16.20.3820.910.860.84
 Middle ML84.7 ± 3.70.05497.9 ± 13.20.2950.8694.3 ± 4.10.22097.7 ± 15.80.2800.970.901.00
 Outer ML82.1 ± 3.30.05995.6 ± 12.10.2800.8695.3 ± 5.10.21498.8 ± 12.50.3240.960.860.97
 Granular layer81.8 ± 4.30.082113.4 ± 13.10.3790.72114.1 ± 7.60.068101.7 ± 19.50.2691.120.721.11
Hilus, neuropil86.7 ± 1.50.00190.1 ± 8.30.1170.9693.7 ± 5.90.12197.5 ± 11.00.3390.960.930.92
CA1 area of hippocampus            
 Stratum oriens87.2 ± 1.40.00286.6 ± 6.50.0541.0192.2 ± 3.30.10194.4 ± 8.40.2820.980.950.92
 Stratum pyramidale78.2 ± 3.10.001109.5 ± 13.50.4720.71102.6 ± 9.60.40488.2 ± 16.30.1731.160.761.24
 Stratum radiatum85.9 ± 1.80.00694.1 ± 11.50.2110.9192.5 ± 6.10.10184.9 ± 7.10.1791.090.931.11
 Stratum lacunosum moleculare74.7 ± 2.60.00585.0 ± 10.60.1180.8885.0 ± 4.20.02182.1 ± 7.50.1251.040.881.04

IR for GABAA α1 subunits showed little change relative to 2N mice; it was significantly reduced only in the stratum lacunosum moleculare and only at 3 months (Table 3). The ratio of GABAA β2/3 to α1 was calculated in an attempt to evaluate subunit composition (Table 3). The values deviated significantly from 1.0 as follows: at 3 months of age—20–30% less in the molecular layer of the inferior blade FD, ∼28% less in granular layers, 14% less in the inner and outer molecular layer of the superior blade FD, and 24% less in the pyramidal layer of the CA1; at age 8 months—16% less in the outer molecular layer of the superior blade FD and 24% more in the pyramidal layer of the CA1. More interesting was the finding that, because of the relative increase in IR for β2/3 subunits, the overall average ratio was increased from 0.83 at 3 months to 1.01 at 8 months, a difference that was significant (P < 0.01). We conclude that there are age-related changes in immunostaining for GABAA receptors in the hippocampus of Ts65Dn mice.

GABAB receptors.

We also investigated the distribution of IR for GABAB receptors (subunits R2 and R1) in Ts65Dn and 2N mice aged 3 and 11.5 months (Table 4). At age 3 months, the overall IR for R1 or R2 subunits was unchanged in Ts65Dn mice, except in the granular layers of the FD (14% less GABAB R1; P = 0.02) (Table 4). In contrast, significantly less GABAB R1 immunostaining was found in Ts65Dn mice aged 11.5 months in the superior blade of the FD, in the hilus, and in the CA1 (Table 4). GABAB receptor R2 subunit was unchanged at 11.5 months of age (Table 4). The ratio of GABAB R1 to R2 was not significantly different from 1.0 at 3 months, but at 11.5 months was decreased by ∼ 14% in the FD, 30% in the hilus, and 20% in the CA1. As for GABAA receptors, these findings are evidence for changes in GABAB receptors during aging, but with an overall decrease rather than the increase documented for GABAA receptors.

Table 4. Quantitative Analysis of GABAB R1 and R2 Receptors in 3- and 11.5-Month-Old Ts65Dn and 2N Mice Hippocampus1
AreaR1R2R1/R2 ratio at 3 monthsR1/R2 ratio at 11.5 months
3 months old11.5 months old3:11.5- Month ratio3 months old11.5 months old3:11.5- Month ratio
% from 2NP value% from 2NP value% from 2NP value% from 2NP value
  • 1

    The number of mice used was as follows: 2N/Ts65Dn = 5/5 for each age. ML, molecular layer. Bold-face type represents significant difference of parameters studied.

Fascia dentata, inferior blade            
 Inner ML90.9 ± 5.50.18686.5 ± 4.20.11781.05102.0 ± 2.90.36100.4 ± 7.10.451.020.890.86
 Middle ML95.5 ± 4.70.31086.8 ± 4.30.18231.10104.8 ± 3.00.2199.0 ± 5.60.401.060.910.88
 Outer ML87.2 ± 4.50.13696.5 ± 5.90.38160.9097.9 ± 4.00.3898.1 ± 3.40.431.000.890.98
 Granular layer86.3 ± 4.00.02186.9 ± 3.90.07210.9997.5 ± 3.70.26100.5 ± 4.10.490.970.890.86
Fascia dentata, superior blade            
 Inner ML95.8 ± 5.80.28478.6 ± 2.00.00071.22101.3 ± 3.10.39101.2 ± 4.50.421.000.950.78
 Middle ML95.4 ± 3.70.17585.3 ± 5.80.05691.12107.5 ± 2.90.09105.8 ± 3.70.251.020.890.81
 Outer ML94.3 ± 3.30.19281.1 ± 2.10.00151.16105.3 ± 4.10.2196.7 ± 1.90.331.090.900.84
 Granular layer93.4 ± 3.50.21675.6 ± 2.80.00111.2499.8 ± 2.40.4890.8 ± 5.80.171.100.940.83
Hilus, neuropil79.9 ± 6.30.06862.6 ± 2.40.00021.2886.4 ± 10.40.1189.5 ± 9.20.190.960.930.70
CA1 area of hippocampus            
 Stratum oriens95.6 ± 3.20.28581.3 ± 3.70.01411.17101.5 ± 2.50.36106.9 ± 6.20.190.950.940.76
 Stratum pyramidale92.6 ± 3.20.11977.4 ± 6.50.01091.2099.4 ± 1.00.4197.7 ± 4.30.391.020.930.79
 Stratum radiatum100.9 ± 2.80.45382.0 ± 4.30.03631.23101.5 ± 2.50.4091.7 ± 3.50.091.111.000.89
 Stratum lacunosum moleculare94.5 ± 3.20.20266.4 ± 6.40.00531.4295.1 ± 5.60.3089.1 ± 5.70.131.070.990.74

Next we studied the distribution of proteins essential for synaptic clustering of neurotransmitter receptors: gephyrin, GABARAP, and neuroligins in 2N and Ts65Dn mice aged 6 to 7 months (Figs. 4, 5).

Figure 4.

Quantitative analysis of the distribution of gephyrin-IR (ad) and GABARAP-IR (eh) in the fascia dentata in 2N and Ts65Dn mice at 3 months old. The density (c,g) and area of individual puncta (d,h) with intensity value above a background threshold were calculated. Results are mean ± SEM. Number of mice examined: 2N = 6, Ts65Dn = 6. *, P < 0.05, significantly different from 2N mice. Scale bar = 10 μm in f (applies to a,b,e,f).

Figure 5.

Quantitative analysis of the distribution of neuroligin IR (ad), neuroligin 1 IR (eh), and neuroligin 2 IR (il) in the fascia dentata in 2N and Ts65Dn mice at 3 months old. The density (c,g,k) and area of individual puncta (d,h,l) with intensity value above a background threshold were calculated. Results are mean ± SEM. Number of mice examined: 2N = 6, Ts65Dn = 6. *, P < 0.05, significantly different from 2N mice. Scale bar = 10 μm in j (applies to a,b,e,f,i,j).

Gephyrin.

When we examined immunostaining for this protein, we evaluated the density of the puncta, the size of the puncta, and the average OD of the puncta. There was no difference in the average density of gephyrin-IR puncta in Ts65Dn vs. 2N mice (Fig. 4a–c; P = 0.94). The average size of gephyrin-IR puncta in Ts65Dn vs. 2N mice was also the same (Fig. 4d; P = 0.22). The average OD of the puncta was correspondingly unchanged (2N: 123.58 ± 0.42; Ts65Dn: 123.00 ± 0.43, n = 6; P = 0.34).

GABARAP.

There was significant difference (increase of 33%) in the average density of GABARAP-IR puncta in Ts65Dn vs. 2N mice (Fig. 4e–g; P = 0.01). The average size of individual GABARAP-IR puncta in Ts65Dn vs. 2N mice was also significantly greater (increase of 15%; Fig. 4h; P = 0.01).

Neuroligins.

By using an antibody (clone 4F9) that marks both neuroligin 1 and 2, we detected no difference in the average density of neuroligin IR of individual puncta in Ts65Dn vs. 2N mice (Fig. 5a–c; P = 0.06). However, the average size of individual neuroligin-IR puncta was significantly larger in the Ts65Dn mice by 14% (Fig. 5d; P < 0.01).

There was no difference in the average density of neuroligin 1-IR (clone 4C12) individual puncta numbers in Ts65Dn vs. 2N mice (Fig. 5e–g; P = 0.25). The average size of individual neuroligin 1-IR puncta in Ts65Dn vs. 2N mice was also the same (Fig. 5h; P = 0.17).

However, there was a significant difference (55% increase) in the average density of neuroligin 2-IR individual puncta numbers in Ts65Dn mice (Fig. 5i–k; P = 0.02). In addition, the average size of individual neuroligin 2-IR puncta in Ts65Dn mice was significantly greater (13%; Fig. 5l; P = 0.03). These data are further evidence for changes in inhibitory synapses in the Ts65Dn hippocampus.

Ultrastructural studies of motor cortex

In earlier studies, we noted that changes in the size of presynaptic boutons and spines involved not just the hippocampus but other regions as well, including the motor cortex (Belichenko et al.,2004). To discern whether the ultrastructural changes in the hippocampus extended to other affected regions, we examined the motor cortex of Ts65Dn and 2N mice by EM. As for the hippocampus, the total density of synapses was not significantly different between 2N and Ts65Dn mice (Supplementary Fig. 1; P = 0.83). There was also no significant difference in the ratio of asymmetric/symmetric synapses in the motor cortex (Supplementary Fig. 1; P = 0.77). The average length of synaptic appositions was increased in the motor cortex (Supplementary Fig. 1; P < 0.01). In contrast to the findings in hippocampus, synaptic apposition lengths in the motor cortex were significantly greater for asymmetric synapses (Supplementary Fig. 1; P < 0.01), but not for symmetric synapses (Supplementary Fig. 1; P = 0.39). Thus, although synaptic structures show widespread changes, detailed examination shows that brain regions differ in the type of changes detected.

DISCUSSION

Our study provides evidence for selective involvement of inhibitory neurotransmission in the FD of the Ts65Dn mouse. Extending recent physiological studies documenting increased inhibition, the current report shows that these changes are linked to marked changes in the structure of synapses. Comparing Ts65Dn mice with their 2N controls, we detected: a selective enlargement of the active zones of asymmetric synapses, increased colocalization of p38 with markers for inhibitory neurotransmission, and increased immunostaining for synaptic proteins that mark inhibitory synapses. Remarkably, the changes in the FD appear to be region specific, because a very different pattern of changes was seen in the motor cortex. The findings for the hippocampus suggest that changes in inhibitory neurotransmission are progressive, raising the possibility that they are compensatory to undefined cellular events. In providing new insight into the neurobiology of DS, this work points to the importance of defining more fully the excitatory and inhibitory circuits of the DS hippocampus and of continuing studies exploring restoration of the balance of excitatory and inhibitory neurotransmission.

Selective changes in inhibitory synapses in FD of Ts65Dn mice

Quantitative analysis of EM data revealed no significant difference in the area density of synapses of all types combined and the asymmetric/symmetric ratio of synapses in the FD and motor cortex of Ts65Dn vs. 2N mice. The findings are consistent with those of Kurt and colleagues (2000), who found no significant difference in the area density when all synapses were included. Also, Kurt and colleagues (2000), as well as an earlier (Belichenko et al.,2004) and the current study from this laboratory, showed a significantly greater synaptic apposition zone length in Ts65Dn vs. 2N mice that extended to both symmetric and asymmetric synapses. In contrast to our findings, although Kurt and colleagues (2000) detected a significant reduction in asymmetric synapses in the temporal cortex of aged mice, we found no evidence for synaptic loss in young mice. Furthermore, we found that synaptic length of symmetric synapses was selectively increased in the FD and that the reverse was true in the motor cortex. It is now apparent that region-specific differences exist with respect to the type of synapses whose size is increased: symmetric synapses in the FD, asymmetric synapses in the motor cortex, and both in the temporal cortex (Kurt et al.,2000).

To confirm that inhibitory synapses were those identified in the EM studies, we employed double-labeling immunohistochemistry to study colocalization of p38 with markers for inhibitory or excitatory synapses. The antibody against p38 labeled all synapses, whereas antibodies against GAD65 or VGAT labeled only inhibitory synapses, and the antibody against VGLUT1 marked only excitatory synapses. In studies of colocalization of these markers, we detected an overall significant increase in colocalization coefficients for GAD65 with p38 IR (∼27%) and for VGAT with p38 IR (∼41%); in contrast, there was no increase in the colocalization for VGLUT1 with p38 IR. Taken together, the morphological studies point compellingly to changes in inhibitory synapses that are significant and selective in the FD of Ts65Dn mice.

Changes in inhibitory synaptic proteins in FD of Ts65Dn mice

The results of morphological studies suggested that selective changes might also be detected in the synaptic proteins and receptors that mediate inhibitory neurotransmission. By using confocal microscopy, we noted changes in the distributions of proteins essential for receptor clustering in the Ts65Dn hippocampus. There was a significant increase in the size and number of IR puncta for neuroligin 2 and GABARAP (both by ∼13%), proteins that mark inhibitory synapses (Wang and Olsen,2000; Chih et al.,2005), but not for neuroligin 1, which marks excitatory synapses (Chih et al.,2005). Although the changes in immunostaining for neuroligin 2 and GABARAP were robust, they were not detected by immunoblotting. The most plausible explanation is that the changes were diluted when we examined the entire hippocampus. Of note, there was no change in gephyrin, a microtubule-associated bridging protein of glycine receptors (Meier and Grantyn,2004) in Ts65Dn mice, suggesting that changes in inhibitory synapses select a subset of proteins and receptors. These data are further evidence for significant changes in inhibitory synapses.

Changes in both excitatory and inhibitory receptors in Ts65Dn mice

Keys to the operation of synapses are the receptors for glutamate and GABA. Therefore, we examined next receptors that mediate excitatory and inhibitory neurotransmitter systems at the protein level, by using immunoblotting and confocal immunofluorescence microscopy. The latter studies contributed the most salient findings. We examined the hippocampal region-specific distributions of receptor IR by using antibodies against GluR1, GluR2, GABAA receptor α1 and β2/3 subunits, and GABAB receptors R1 or R2 subunits. Significant changes were detected for GluR2, GABAA β2/3, and GABAB R1, with differences noted between the ages examined (Tables 2–4). There was an overall decrease of IR in the hippocampus in young Ts65Dn mice for GluR2 (∼13%) and for GABAA receptor β2/3 subunit (∼20%). Older Ts65Dn mice showed an overall decrease in IR for GluR2 (∼18%) and GABAB receptor R1 (∼19%). Perhaps the most interesting findings were for changes in the relative amounts with respect to 2N mice between the ages examined. The pattern was for decreases in GluR2 and GABAB R1 subunits inversely correlated with increases in GABAA β2/3. These data are evidence against the view that changes in inhibitory neurotransmission, or indeed of inhibitory synapses, can be explained simply by an increase in the number of inhibitory receptors or a decrease in excitatory receptors. Instead, it will be necessary to explore in detail and quantitatively, including at the protein level, the precise loci of the changes detected within hippocampal networks in which increased inhibition has been demonstrated.

Understanding the neurobiology of synaptic changes in the Ts65Dn hippocampus

The balance between excitation and inhibition, and the plasticity of circuits, is partly a function of the number of such receptors, their subunit composition, and the cellular elements in which they are expressed. Deviations from normal can readily be envisioned and have been shown to contribute to neurological disorders (Luscher and Keller,2004; Greger et al.,2007). In Ts65Dn mice, on average the GluR2/GluR1 ratio was 0.99 ± 0.02 in the hippocampus for young and 0.95 ± 0.03 for old mice. However, there was a 13% decrease in the outer molecular layer of the inferior blade for young Ts65Dn mice. As mice aged, the GluR2/GluR1 ratio decreased by 13–26% in all three sublayers of the molecular layer in the superior blade, as well as in the stratum radiatum of the CA1. In contrast, a 13% increase was noted in the hilus. The physiological consequence of such changes is as yet obscure, but the speculation can be offered that synaptic plasticity may be decreased to a greater extent with aging in the Ts65Dn hippocampus. The decrease in the levels of GluR2 receptors, whose presence within AMPA receptors regulates calcium flux (Isaac et al.,2007), may contribute to the demonstrated alterations in synaptic plasticity in Ts65Dn mice (Siarey et al.,1997; Galdzicki et al.,2001; Kleschevnikov et al.,2004).

Perhaps consistent with a decrease in synaptic plasticity are the changes detected in GABAA and GABAB receptors during aging. Here there was a relative increase in GABAA receptor β2/3 subunits with aging and an increase in the ratio of β2/3 to α1 subunit of 18%. Although the physiological consequences of such alterations are uncertain, these changes may reflect a compensatory adjustment of the inhibitory circuits during the lifespan. If so, the question arises as to what stimulus could be responsible. It will be important to consider in future studies the possibility of a preexisting and persistent increase in excitatory neurotransmission that induces a compensatory change in inhibitory neurotransmission and to test the idea rigorously through electrophysiological studies in very young mice. Studies examining in detail the structure and function of inhibitory synapses will also be essential. By using this approach it will be possible to understand the underlying pathophysiology of circuits and the methods that might be used to enhance hippocampal function.

Similarity in alterations of excitatory-inhibitory relationship in subjects with DS and Ts65Dn mice

Most of the neuropathological literature for DS subjects has focused on cholinergic neurotransmitter systems (Casanova et al.,1985; Mann et al.,1985; see also review in Salehi et al.,2007). Few studies address amino acid neurotransmitters and the circuits in which they participate, and the findings are as yet inconclusive. Reynolds and Warner (1988) showed significant reductions in glutamate in the hippocampus of DS subjects, with a tendency to lower levels in the amygdala and caudate. The levels of GABA were diminished by 27% and 19% in the hippocampus and temporal cortex, respectively, but did not reach significance (Reynolds and Warner,1988). Risser et al. (1997) found significantly reduced glutamate in the temporal cortex, but not in the frontal cortex. No regional differences were apparent in the control group, but a significantly lower level of glutamate was found in the parahippocampal gyrus than in the frontal cortex of DS subjects (Risser et al.,1997). In contrast, Seidl et al. (2001) found no significant differences in amino acid concentrations in adult DS subjects relative to controls. Recently, Whittle et al. (2007) showed a significant reduction in GABA, but not glutamate, in the frontal cortex of fetal subjects with DS. Additional studies of humans and human tissue will be needed to document changes in neurotransmission and to examine them in the context of mouse model studies of the type reported here. In providing evidence in a mouse model for region-specific changes in synaptic structure and composition, we suggest that changes in neuronal circuits in people with DS will also be region specific and that the pattern of changes can be used to elucidate important features of the underlying changes in the biology of cognitive and other circuits.

Acknowledgements

We thank the members of the Mobley Laboratory for critically reading the manuscript and for many helpful discussions.

Ancillary