Syndecan-3-Deficient Mice Exhibit Enhanced LTP and Impaired Hippocampus-Dependent Memory

MCN

Molecular and Cellular Neuroscience 21, 158 –172 (2002) doi:10.1006/mcne.2002.1167

Syndecan-3-Deficient Mice Exhibit Enhanced LTP and Impaired Hippocampus-Dependent Memory Marko Kaksonen,* ,1 Ivan Pavlov,* ,†,1 Vootele Vo˜ikar,* ,1 Sari E. Lauri,* ,†,2 Anni Hienola,* Ruusu Riekki,* ,† Merja Lakso,* ,3 Tomi Taira, † and Heikki Rauvala* ,4 *Laboratory of Molecular Neurobiology, Institute of Biotechnology and Department of Biosciences, P.O. Box 56, and †Division of Animal Physiology, Department of Biosciences, P.O. Box 65, University of Helsinki, 00014 Helsinki, Finland

Syndecan-3 (N-syndecan) is a transmembrane heparan sulfate proteoglycan expressed predominantly in the nervous system in a developmentally regulated manner. Syndecan-3 has been suggested to play a role in the development and plasticity of neuronal connections by linking extracellular signals to the regulation of the cytoskeleton. To study its physiological functions, we produced mice deficient in syndecan-3 by gene targeting. The mutant animals are healthy, are fertile, and have no apparent defects in the structure of the brain. We focused on characterizing the functions of the hippocampus, a brain area where expression of syndecan-3 is prominent in adults. Mice lacking syndecan-3 exhibited an enhanced level of long-term potentiation (LTP) in area CA1, while basal synaptic transmission and short-term plasticity were similar to those in wild-type animals. Further, the mutant mice were not responsive to the syndecan-3 ligand heparinbinding growth-associated molecule, which inhibits LTP in area CA1 in wild-type animals. Behavioral testing of the syndecan-3-deficient mice revealed impaired performance in tasks assessing hippocampal functioning. We suggest that syndecan-3 acts as an important modulator of synaptic plasticity that influences hippocampus-dependent memory.

1

These authors contributed equally to this work. Current address: Department of Anatomy, School of Medical Sciences, University of Bristol, University Walk, Bristol BS8 1TD, UK. 3 Current address: A. I. Virtanen Institute for Molecular Sciences, BioTeknia, Neulaniementie 2, 70210 Kuopio, Finland. 4 To whom correspondence should be addressed at the Laboratory of Molecular Neurobiology, Institute of Biotechnology and Department of Biosciences, P.O. Box 56 (Viikinkaari 5), FIN-00014 University of Helsinki, Finland. Fax: ⫹358 9 191 59068. E-mail: heikki.rauvala@ helsinki.fi. 2

158

INTRODUCTION Heparan sulfate proteoglycans (HSPG) are abundant on most cell surfaces and are involved in a wide range of cell– cell and cell–matrix interactions. They are known to regulate cell adhesion, cell migration, differentiation, and growth factor signaling (for reviews, see Bernfield, 1999; Sanderson, 2001; Yamaguchi, 2001). In the nervous system, HSPGs are involved in neurogenesis, neurite guidance, synaptogenesis, and synaptic plasticity (Yamaguchi, 2001). Syndecans together with glypicans form the two major families of cell-surface HSPGs (Bernfield et al., 1999). Syndecan-3 (N-syndecan) is one of the four mammalian syndecans and it is mainly expressed in the nervous system, especially during development (Carey et al., 1997). It has been suggested to function in cell adhesion, neurite guidance, and cell migration during development of the nervous system (Raulo et al., 1994). Syndecan-3 has also been implicated in the regulation of synaptic plasticity in the hippocampus (Lauri et al., 1999). Syndecan-3 is expressed in an activity-dependent manner in the CA1 pyramidal neurons, and application of exogenous syndecan inhibits the induction of longterm potentiation (LTP) (Lauri et al., 1999). Syndecan-3 has one transmembrane domain, a short cytoplasmic tail of 34 amino acids, and an extracellular domain that carries heparan sulfate chains (Carey et al., 1997). Syndecan-3 heparan sulfate chains bind to fibroblast growth factor-2 (FGF-2) (Chernousov and Carey, 1993) and heparin-binding growth-associated molecule (HB-GAM) (Raulo et al., 1994). Syndecans have been suggested to function as coreceptors with other signaling receptors, such as FGF receptors and integrins 1044-7431/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved.

159

Syndecan-3 in Hippocampal LTP and Memory

(Woods and Couchman, 2000). Syndecan-3 may also have a signaling role of its own, as it is known to activate the cortactin/src kinase pathway when bound to HB-GAM (Kinnunen et al., 1998). Signaling may involve oligomerization of syndecan-3 core proteins (Asundi and Carey, 1995). Syndecan-3 is known to interact with the PDZ (PSD-95/discs large/ZO-1)-domain-containing proteins CASK and syntenin via the extreme C-terminus of its cytoplasmic portion (Hsueh et al., 1998; Cohen et al., 1998; Grootjans et al., 1997). To study syndecan-3 function in vivo we created a targeted mutation in the mouse syndecan-3 gene. The mutant animals are viable and do not exhibit any apparent developmental abnormalities, but we found specific changes in hippocampal synaptic plasticity and in hippocampus-dependent memory. This same targeted mutation was recently used to demonstrate the role of syndecan-3 in feeding behavior but has not been otherwise characterized (Reizes et al., 2001).

RESULTS Generation of Syndecan-3-Deficient Mice We used gene targeting by homologous recombination in mouse embryonal stem (ES) cells to create a mutated syndecan-3 allele. The targeting construct was electroporated into ES cells and neomycin– gancyclovirresistant colonies were selected. Cell clones were screened by Southern blotting using a flanking probe (Fig. 1). Clones that carried the targeted mutation were expanded, reanalyzed by Southern hybridization with both 5⬘ and 3⬘ flanking probes, and then injected into mouse blastocysts. Two of the cell clones transmitted the targeted mutation into the germ line. To verify the mutation as a null we extracted RNA from the brains of 2-week-old syndecan-3 ⫹/⫹, ⫹/⫺, and ⫺/⫺ mice. Syndecan-3 expression peaks in the brain at the age of about 2 weeks (Nolo et al., 1995). The RNA was then probed with a mouse syndecan-3 cDNA probe that contains the full sequence coding for the mature protein. No signal was detected in the ⫺/⫺ samples, and in the ⫹/⫺ samples the signal was about half of the ⫹/⫹ signal (Fig. 1C). To study the levels of syndecan-3 protein we prepared hippocampal neurons from newborn animals and cultured them for 3 weeks in vitro and then immunostained them using syndecan-3 antibody. Syndecan-3 immunostaining was seen in the ⫹/⫹ neurons at the cell surface in the cell soma and in the processes. Most if not all processes were stained. Heterozygous neurons had reduced staining

intensity and the ⫺/⫺ neurons did not show any specific staining. The targeted mutation thus appears to be syndecan-3 null. Mice carrying the syndecan-3 null mutation were born in ratios that did not significantly differ from expected Mendelian ratios (of 194 pups 41 were ⫹/⫹, 107 ⫹/⫺, and 46 ⫺/⫺). Homo- and heterozygous mutant mice have a normal life span and they are fertile, apparently healthy, and morphologically normal. Brain Structure of Syndecan-3-Deficient Mice Histological sections of brains were made from the ⫹/⫹, ⫹/⫺, and ⫺/⫺ mice and stained by either hematoxylin– eosin to show the general morphology or silver staining to reveal the fiber pathways (Fig. 2A). No differences in overall morphology were found and the major axonal pathways appeared similar in all genotypes (data not shown). Cell numbers in layers II–III/IV of the cerebral cortex were measured from the hematoxylin– eosin-stained sections. No statistically significant differences in the numbers of neurons were found between the different genotypes (64,594.3 ⫾ 14,836.7 in ⫹/⫹ animals, n ⫽ 3, and 67,903.2 ⫾ 3346.7 in ⫺/⫺ animals, n ⫽ 4). As syndecan-3 has been implicated in hippocampal plasticity (Lauri et al., 1999) and syndecan-2 has been shown to regulate maturation of dendritic spines in hippocampal neurons in vitro (Ethell and Yamaguchi, 1999), we focused on the structure of the hippocampus. We used Golgi impregnation staining to visualize neuronal morphology in the hippocampus and to measure the density of dendritic spines on apical dendrites of the CA1 pyramidal neurons. No differences in overall appearance of the hippocampal pyramidal neurons or morphology of the dendritic spines were seen (Fig. 2B). The spine densities were similar in all genotypes (Fig. 2C). We also used antibodies against the presynaptic marker synaptophysin to estimate the number of synapses in hippocampal sections from the syndecan-3deficient and wild-type animals. No differences between the genotypes were found in the level of synaptophysin immunofluorescence in the area CA1 of hippocampus (Fig. 2D). The hippocampal neurons cultured in vitro appeared normal in growth and morphology (Fig. 1). The distribution of spine lengths was measured in DiI-stained cultured neurons. No significant differences were seen in mean spine lengths in the wild-type and null mutant neurons (1.93 ⫾ 0.09 ␮m in ⫹/⫹ neurons and 2.02 ⫾ 0.09 ␮m in ⫺/⫺ neurons; total of 220 spines).

160

Kaksonen et al.

FIG. 1. Targeted disruption of the syndecan-3 gene. (A) Targeting strategy. Syndecan-3 gene organization (exons 3–5 shown as filled boxes), the targeting vector, and the mutant syndecan-3 locus expected after homologous recombination. BS, pBluescript plasmid; neo, neomycin phosphotransferase gene; TK, thymidine kinase gene. (B) PCR analysis of syndecan-3 alleles from mouse tail DNA. (C) Northern blot of postnatal mouse brain RNA with syndecan-3 cDNA probe. (D) Hippocampal neurons grown in vitro and stained with anti-syndecan-3 antibodies. The top shows the phase-contrast image and the bottom the immunofluorescence images of the same fields.

The distribution of AMPA and NMDA receptors in cultured hippocampal neurons was studied by immunostaining and found to be unchanged in the mutants.

The proportion of “morphologically silent synapses” (Liao et al., 1999) that contain NMDA but not AMPA receptors was similar in all genotypes (31.7 ⫾ 3.4% in

161

Syndecan-3 in Hippocampal LTP and Memory

FIG. 2. Hippocampal morphology and synapse density in the CA1 region of the hippocampus. (A) Hematoxylin-stained hippocampal sections of the ⫹/⫹ and ⫺/⫺ animals. (B) Golgi–Cox-stained apical dendrites of the CA1 pyramidal neurons in the ⫹/⫹ and ⫺/⫺ animals. (C) Quantitation of the density of the dendritic spines on apical dendrites of the CA1 pyramidal neurons (ⱖ400 spines measured/genotype, 3 animals/genotype). (D) The level of synaptophysin immunofluorescence in the stratum radiatum of the CA1 region in the ⫹/⫹ (n ⫽ 7) and ⫺/⫺ animals (n ⫽ 8).

⫹/⫹, 36.5 ⫾ 2.2% in ⫹/⫺, and 34.4 ⫾ 4.3% in ⫺/⫺ neurons; 3 embryos/genotype). Basal Synaptic Transmission Is Normal in Syndecan-3-Deficient Mice To study the effect of the mutation on synaptic function in hippocampus we studied field excitatory postsynaptic potentials (fEPSPs) from the CA1 region evoked by Schaffer collateral stimulation. The relationships between the presynaptic fiber volley amplitude and the slope of fEPSP were indistinguishable between the genotypes (Fig. 3A), indicating that single-pulseevoked responses are not affected by the mutation. Further, we examined paired-pulse facilitation (PPF).

The facilitation of fEPSP obtained by paired-pulse stimulation is a form of short-term plasticity known to inversely correlate with the neurotransmitter release probability (Zucker, 1989; Manabe et al., 1993; Dobrunz and Stevens, 1997). The ⫺/⫺ mice did not show any significant difference compared to the wild-type controls in the PPF over an interpulse interval range of 20 to 200 ms (Fig. 3B). These results suggest that basal characteristics of synaptic transmission are not altered in the syndecan-3 knockout mice. LTP Is Enhanced in Syndecan-3-Deficient Mice We next studied the role of syndecan-3 in hippocampal synaptic plasticity. LTP was induced in the CA1

162

Kaksonen et al.

(HFS) trains were given 10 min apart. Both in the knockout and in the wild-type mice LTP was saturated after the third train stimulus. However, in the mice lacking syndecan-3 the saturation level was significantly higher (245 ⫾ 10.8% of the baseline) than in the control animals (210 ⫾ 7.7% of the baseline; P ⬍ 0.05; Fig. 4C). In contrast, the low-frequency stimulation protocol (1 Hz/900 pulses) did not produce any significant changes in the slope of fEPSPs in either the syndecan-3-knockout or the control mice under the present experimental conditions (Fig. 4D). Recombinant HB-GAM Does Not Affect LTP in Syndecan-3-Deficient Mice

FIG. 3. Normal baseline synaptic transmission in the hippocampal slices from mice lacking syndecan-3. (A) Slopes of fEPSPs plotted as a function of presynaptic fiber volley (PSFV) amplitude in slices from the homozygous (n ⫽ 8 slices/4 mice) and the heterozygous (n ⫽ 7 slices/4 mice) syndecan-3 mutant mice and from the wild-type control animals (n ⫽ 7 slices/4 mice). (B) Facilitation induced by paired stimuli with interpulse intervals (IPI) between 20 and 200 ms was not different between the genotypes (⫺/⫺, n ⫽ 13 slices/6 mice; ⫹/⫹, n ⫽ 17 slices/6 animals; ⫾, n ⫽ 8 slices/4 mice). All data represent means ⫾ SEM.

area of the hippocampus by high-frequency stimulation protocol (100 Hz/1 s). A repeated-measures ANOVA revealed significantly larger potentiation of the fEPSP slope produced 1 h after the LTP induction in the homozygous (163.12 ⫾ 9.55% of baseline, n ⫽ 18 slices/9 animals) and heterozygous (167.35 ⫾ 14.35% of baseline, n ⫽ 7 slices/4 animals) syndecan-3 mutant mice compared to the wild-type mice (137.65 ⫾ 5.54% of baseline, n ⫽ 18 slices/8 animals; F(2,18) ⫽ 5.35, P ⬍ 0.02; Fig. 4A). No changes between the genotypes were seen in PPF measured 60 min after the LTP induction (Fig. 4B). To find the maximal LTP that could be elicited in slices from the syndecan-3-deficient mice, we examined the saturation of LTP measured after repetitive application of tetanic stimulation until no further increase in the fEPSP slope occurred. High-frequency stimulation

Syndecan-3 is a receptor for HB-GAM, an extracellular matrix protein implicated in the regulation of synaptic plasticity in hippocampus (Raulo et al., 1994; Lauri et al., 1996; Amet et al., 2001). Further experiments were carried out to directly test whether syndecan-3 is involved in modulation of LTP by HB-GAM. Pressure injection of HB-GAM into the CA1 dendritic area close to the recording site has been shown to inhibit induction of LTP in rat slices (Lauri et al., 1998). Here we find that in hippocampal slices prepared from the wild-type control mice HB-GAM suppressed HFS-induced LTP (P ⬍ 0.02, t test) without affecting posttetanic potentiation or baseline synaptic transmission (Fig. 5A). However, injection of recombinant HB-GAM into slices from the syndecan-3-knockout animals did not have any detectable effect on LTP (Fig. 5B). Control injections with phosphate-buffered saline (PBS) did not influence LTP in either genotype (Figs. 5A and 5B). The above results strongly support the idea that syndecan-3 plays a crucial role in modulatory effects of HB-GAM on hippocampal plasticity. Hippocampus-Dependent Memory Is Impaired in Syndecan-3-Deficient Mice We carried out a set of behavioral tests (Vo˜ ikar et al., 2001) with the mutant mice and their wild-type littermates. We did not establish any difference between these groups in basic neurological functions (postural, righting, visual placing reflexes), nociception (hot plate), motor coordination (rotarod), locomotor activity (open field), or anxiety-like behavior (elevated plus maze and light– dark exploration tests). The results from these tests are summarized in Table 1. In order to study the hippocampus-dependent forms of memory in the syndecan-3-deficient mice we carried

Syndecan-3 in Hippocampal LTP and Memory

163

FIG. 4. Enhanced LTP in the syndecan-3-mutant mice. (A) LTP induced by HFS (100 Hz/1 s) is significantly enhanced in the homozygous (n ⫽ 18 slices/9 mice; P ⬍ 0.01, LSD post hoc comparison) and the heterozygous (n ⫽ 7 slices/4 mice; P ⬍ 0.03, LSD post hoc comparison) mutant mice compared to the wild-type controls (n ⫽ 18 slices/8 mice). Superimposed single fEPSPs from knockout mice and wild-type animals before and 1 h after LTP induction are shown on the top. (B) PPF 60 min after LTP induction is similar in the wild-type mice (n ⫽ 14 slices/6 mice) and in the mutants (⫺/⫺, n ⫽ 14 slices/6 mice; ⫾, n ⫽ 5 slice/3 mice). (C) LTP saturation level is higher in the syndecan-3-knockout mice (n ⫽ 8 slices/4 mice) than in the wild-type animals (n ⫽ 10 slices/5 mice; P ⬍ 0.05). (D) Similar response to the low-frequency stimulation (1 Hz/900 pulses) in the syndecan-3-knockout (n ⫽ 9 slices/6 mice) and the wild-type mice (n ⫽ 9 slices/7 mice). All data represent means ⫾ SEM.

out a spatial learning task in the water maze. The escape latency across the training trials decreased significantly and similarly [repeated-measures ANOVA: effect of training blocks F(6,108) ⫽ 60.8, P ⬍ 0.01 and genotype ⫻ training block F(6,108) ⫽ 0.5, P ⫽ 0.79]. The effect of the genotype was not significant [F(1,18) ⫽ 3.4, P ⫽ 0.08] although post hoc Newman–Keuls test re-

vealed a significant difference (P ⬍ 0.05) in the escape latency in the third training block (Fig. 6A). In order to estimate the spatial strategy and selectivity of navigation we performed two transfer tests. In the first, after four blocks of training (16 trials), the wildtype animals displayed significant spatial memory as suggested by the number of counter crossings

164

FIG. 5. Recombinant HB-GAM pressure-injected into the stratum radiatum of the CA1 region in slices from the wild-type mice (A) inhibits LTP induction (n ⫽ 6; P ⬍ 0.02, Student’s t test). However, when injected into the slices from the syndecan-3-knockout mice (B), HB-GAM does not affect LTP (n ⫽ 7). Control injections performed with PBS (A, B) have no detectable effect on LTP (n ⫽ 5). The data represent means ⫾ SEM.

[F(2,33) ⫽ 6.3, P ⬍ 0.01; Fig. 6B] as well as the time spent searching in the correct quadrant of the water maze [F(2,33) ⫽ 7.9, P ⬍ 0.01; Fig. 6C]. In contrast, the knockout mice did not show any spatial preference [counter crossings F(2,21) ⫽ 0.2, P ⫽ 0.8; quadrant time F(2,21) ⫽ 0.5, P ⫽ 0.6]. After seven training blocks (28 trials) both groups crossed the counter more often in the trained quadrant [wild type F(2,33) ⫽ 16.1, P ⬍ 0.01; knockout F(2,21) ⫽ 8.0, P ⬍ 0.01; Fig. 6D]. However, analysis of swimming time in quadrants (Fig. 6E) revealed better spatial preference in the wild-type group [F(2,33) ⫽ 19.0, P ⬍ 0.01] compared to the knockout mice [F(2,21) ⫽ 3.1, P ⫽ 0.07]. The knockout mice spent significantly more time swimming in the opposite quadrant than the wild-type animals [F(1,18) ⫽ 5.9, P ⬍ 0.05]. It is important to note that the swimming distances and the time in thigmotaxis did not differ between the groups in either transfer test. Moreover,

Kaksonen et al.

training in a cued version of water maze did not reveal any difference between the genotypes [genotype F(1,18) ⫽ 1.0, P ⫽ 0.32; trials F(7,126) ⫽ 14.4, P ⬍ 0.01; genotype ⫻ trial F(7,126) ⫽ 0.3, P ⫽ 0.93]. To further investigate learning and memory in the syndecan-3-knockout mice we applied the cued fear conditioning test. There was no difference in the locomotor activity or freezing behavior in the preconditioning phase (Figs. 7A and 7B). However, significant differences were established in contextual freezing. Mice were returned twice to the conditioning box (24 and 72 h after conditioning). Post hoc Newman–Keuls test after repeated-measures ANOVA revealed significantly (P ⬍ 0.05) reduced freezing for the knockout mice in both phases compared to the wild-type mice. Simultaneously we registered locomotor activity of the animals, and repeated-measures ANOVA established significant effects of the genotype [F(1,18) ⫽ 4.2, P ⫽ 0.05] and the phase [F(1,18) ⫽ 13.1, P ⬍ 0.01]. The genotype ⫻ phase interaction was not significant [F(1,18) ⫽ 3.6, P ⫽ 0.08]. Post hoc Newman–Keuls test confirmed that decreased freezing in the knockout group in the second test was accompanied by significantly enhanced locomotor activity (P ⬍ 0.01; Fig. 7B). Both groups showed similar freezing in novel context when the conditioning stimulus was applied [F(1,18) ⫽ 0.1, P ⫽ 0.7]. The injection of LiCl after drinking of 15% sucrose solution induced significant aversion to sucrose in both groups as measured 24 and 72 h after conditioning (Fig. 7C). Furthermore, the groups did not differ significantly in this paradigm [24 h F(1,18) ⫽ 3.4, P ⫽ 0.08; 72 h

TABLE 1 Behavioral Screening of the Syndecan-3-Deficient Mice

EPM: Latency to enter open arm (s) EPM: Number of open arm entries EPM: Number of closed arm entries EPM: % of open arm entries EPM: % of time on open arms LD: Number of light/dark transitions LD: % of time in light OF: Distance traveled in 30 min (cm) RR: Retention time on RR (s) HP: Latency to lick hind paw (s)

⫹/⫹ mice (n ⫽ 8)

⫺/⫺ mice (n ⫽ 12)

151.3 ⫾ 24.2 1.8 ⫾ 0.4 8.6 ⫾ 0.6 15.0 ⫾ 3.0 5.7 ⫾ 1.7 11.3 ⫾ 1.3 36.0 ⫾ 5.6 1081 ⫾ 194 82.0 ⫾ 11.7 14.4 ⫾ 0.6

149.1 ⫾ 21.4 2.2 ⫾ 0.5 7.2 ⫾ 0.7 18.1 ⫾ 3.0 8.1 ⫾ 1.7 12.5 ⫾ 1.5 29.3 ⫾ 3.9 1080 ⫾ 254 72.6 ⫾ 12.3 14.9 ⫾ 1.0

Note. EPM, elevated plus maze (5 min); LD, light– dark exploration (6 min); OF, open-field activity (30 min); RR, rotarod (12 rpm, two trials, the best result was used for calculating the group mean); HP, hot plate at 52°C. The EPM, LD, OF, and RR experiments were carried out as described in Vo˜ ikar et al. (2001).

Syndecan-3 in Hippocampal LTP and Memory

165

FIG. 6. Impaired memory of the syndecan-3-null mice in water maze. (A) Escape latency to the hidden platform across seven training blocks (mean of four trials for each block). (B) Number of annulus crossings in the first transfer test (TT1). (C) Time spent swimming in the different quadrants in TT1. (D) Number of annulus crossings in the second transfer test (TT2). (E) Time spent swimming in the different quadrants in TT2. Chance level on (C) and (E) is shown with horizontal line. (*P ⬍ 0.05 between the wild-type and the knockout mice.)

F(1,18) ⫽ 1.2, P ⫽ 0.28]. In addition, both the knockout and the wild-type mice displayed similarly [F(1,18) ⫽ 1.2, P ⫽ 0.28] strong aversion to 0.1% quinine, suggesting normal gustatory function.

DISCUSSION In Vivo Roles of Syndecans Our results show that syndecan-3 in mouse is not essential for basic processes of development. Syndecan-3 is, however, involved in the activity-dependent regulation of synaptic strength in hippocampus. Synde-

can-3 deficiency also impairs hippocampus-dependent memory. Targeted mutations in the other syndecans have revealed developmentally mild phenotypes. Syndecan-1deficient mice are apparently normal and fertile (Alexander et al., 2000). The lack of syndecan-1 reduced mammary epithelia hyperplasia and tumorigenesis induced by transgenic Wnt1 overexpression (Alexander et al., 2000). Syndecan-4-null mice are also viable, fertile, and macroscopically normal (Echtermeyer et al., 2001; Ishiguro et al., 2000). They, however, exhibit reduced wound healing and angiogenesis (Echtermeyer et al., 2001). Null mutations in syndecan-1, -3, and -4 show

166

FIG. 7. Conditioning experiments. (A) Percentage of time in freezing (absence of any movement for more than 5 s) in the preconditioning phase (2 min) and two context tests (3 min) 24 and 72 h after the conditioning, respectively. (B) Locomotor activity expressed as distance traveled (cm) in the preconditioning phase and the context tests. (C) Sucrose aversion 24 and 72 h after conditioning and quinine aversion. (*P ⬍ 0.05, **P ⬍ 0.01 between the wild-type and the knockout mice.)

that individual mammalian syndecans are not essential for development but may be involved in the regulation of homeostasis, regeneration, synaptic plasticity, and tumorigenesis. Syndecan-3 in the Regulation of LTP Here we find that syndecan-3 ablation does not affect basal synaptic transmission in hippocampus. However, mice hetero- or homozygous for the null allele of syndecan-3 show enhanced LTP in the CA1 area of hippocampus compared to the wild-type mice. The mutant animals not only display a larger increase in the slope of

Kaksonen et al.

fEPSP following the first tetanization, but also have a higher saturation level of LTP. It has been shown previously that syndecan-3 is expressed in an activity-dependent manner in hippocampal pyramidal neurons so that the expression level of its mRNA is enhanced after induction of LTP by highfrequency stimulation. In addition, application of soluble syndecan-3 blocks HFS-induced LTP (Lauri et al., 1999). Thus, the data presented here together with earlier results suggest that syndecan-3 suppresses LTP in the CA1 area. Most of the null mutations expressing an LTP phenotype described to date display either reduced or impaired LTP, thus revealing molecular mechanisms that are necessary for or promote LTP. However, it is evident that stabilizing or suppressory mechanisms are also important for the regulation of synaptic plasticity (Abel et al., 1998). Syndecan-2 overexpression has been shown to enhance the rate of maturation of synaptic spines in cultured hippocampal neurons (Ethell and Yamaguchi, 1999). In the case of syndecan-3 mutant mice we did not, however, see any changes in the hippocampal histology, synaptic density, or morphology of the dendritic spines, which could explain the abnormal LTP. The effect of both homo- and heterozygous syndecan-3 mutation on LTP is equally strong. This haploinsufficiency is similar to that reported for syndecan-4null mutants in wound healing and angiogenesis (Echtermeyer et al., 2001). As reduction of syndecan-3 levels to about 50% (Fig. 1D) leads to full phenotypic effect it is conceivable that small physiological changes in the syndecan-3 expression level could affect synaptic efficacy. In addition to the rapid, activity-induced changes in the syndecan-3 expression in hippocampus in vivo (Lauri et al., 1999), syndecan-3 expression levels oscillate as a function of the feeding status in the hypothalamic neurons that control feeding behavior (Reizes et al., 2001). Syndecan-3 expression may thus act as an inducible mechanism used for functional adaptation of neuronal circuitry. Syndecan-3 in the Regulation of HippocampusDependent Learning and Memory Behavioral testing of the syndecan-3-deficient mice confirmed that they do not exhibit any impairment in development or general appearance as suggested by normal motor activity, coordination, and pain sensitivity. Furthermore, their emotional behavior in anxiety models (elevated plus maze, light– dark exploration) was similar to that of the wild-type animals. We continued behavioral analysis with learning and memory

Syndecan-3 in Hippocampal LTP and Memory

tests in order to study whether the alterations in hippocampal synaptic plasticity affected the hippocampusdependent forms of learning and memory. We found spatial memory in the syndecan-3-deficient mice to be significantly impaired in the water maze test. Furthermore, the hippocampal component of fear conditioning, context-dependent freezing, was reduced in the mutant mice compared to the wild-type animals. However, memory about conditioning tone (mainly dependent on amygdala function) was intact. Long-term potentiation is the major cellular model of the mechanisms underlying activity-dependent plasticity in the hippocampus (Bliss and Collingridge, 1993; Malenka and Nicoll, 1999). Spatial learning and memory are often correlated with LTP in hippocampus (Tang et al., 1999; Malleret et al., 2001; Tsien et al., 1996; Minichiello et al., 1999). However, several reports have shown increased LTP in mutant mice to be associated with impaired learning or memory (Gerlai et al., 1998; Migaud et al., 1998; Uetani et al., 2000). Moreover, it has been shown that saturation of LTP results in memory impairment (Moser et al., 1998) and induction of LTP after learning can disrupt retention of spatial information (Brun et al., 2001). Our recent observations (Pavlov et al., 2002) with HB-GAM knockout mice have revealed a similar phenotype with enhanced LTP and reduced spatial learning and memory. Therefore it is tempting to speculate that deficiency of the HB-GAM/syndecan-3 signaling leads to enhanced LTP and thereby impairs hippocampus-dependent forms of memory. Interestingly, it has recently been reported that changes in syndecan-3 level in hypothalamus modulate feeding behavior of mice (Reizes et al., 2001). Hypothalamus and amygdala receive gustatory information through the gustatory zone of the nucleus of the solitary track (reviewed in Houpt, 2000) and have been shown to be involved in mediating conditioned taste aversion. In this test animals learn to associate a sweet taste of sucrose solution with malaise (induced by injection of LiCl). This type of learning is ethologically well conserved, being important in remembering safe or unsafe food. The syndecan-3 knockout mice did not exhibit any impairment in this task. In summary, we suggest that syndecan-3 deficiency affects specifically hippocampus-dependent forms of memory. Molecular Mechanisms of Syndecan-3 in Neuronal Plasticity HB-GAM is a good candidate for a ligand of syndecan-3 in hippocampal plasticity, in agreement with the receptor function of syndecan-3 in HB-GAM-induced

167 neurite outgrowth in brain neurons in vitro (Raulo et al., 1994). HB-GAM and syndecan-3 are coexpressed in the hippocampus (Nolo et al., 1995). HB-GAM was shown to regulate hippocampal LTP via binding to heparan sulfates (Lauri et al., 1998) and HB-GAM-deficient mice display a decreased threshold for LTP induction (Amet et al., 2001). In addition, expression levels of both HBGAM and syndecan-3 in hippocampus are increased after high-frequency stimulation, leading to LTP (Lauri et al., 1996, 1999). Coexpression and coregulation of these two molecules in the hippocampus suggest their common function in the modulation of synaptic plasticity. Remarkably, HB-GAM application into the CA1 area of hippocampus, which strongly attenuates LTP in wild-type mice, had no effect in syndecan-3 knockouts. Thus, the experiments presented here support the idea that syndecan-3 acts as a functional receptor for HBGAM in plasticity regulation. Another candidate for mediating the effects of HB-GAM in synaptic plasticity is the receptor-type protein tyrosine phosphatase ␨/␤ (RTPTP␨/␤) (Maeda and Noda, 1998) and interestingly, deletion of RTPTP␦ (belonging to the same receptor family) in mice leads to enhanced LTP and impaired spatial learning (Uetani et al., 2000). Binding of HB-GAM to syndecan-3 leads to association of src-kinase and actin-binding protein cortactin to the cytoplasmic tail of syndecan-3 and to activation of src-kinase and phosphorylation of cortactin (Kinnunen et al., 1998). The association of src-family kinases and cortactin with the syndecan-3 cytoplasmic domain was also shown to take place in hippocampal neurons and to be enhanced by electrical stimulation leading to LTP (Lauri et al., 1999). Cortactin can induce actin polymerization by activating the arp2/3 complex (Uruno et al., 2001). With its SH3 domain it can interact with PDZ proteins, such as SHANK, which can link it to the membrane proteins (Naisbitt et al., 1999). Syndecan-3 interactions with PDZ-domain-containing proteins, such as CASK (Hsueh et al., 1998) and syntenin (Grootjans et al., 1997), may also be important in the regulation of LTP. PDZ-domain proteins have been implicated in the regulation of synaptic plasticity (Tomita et al., 2001). A phenotype resembling that of syndecan-3-deficient mice with enhanced LTP and impaired memory has been reported in mice lacking the PDZ-domain protein PSD-95 (Migaud et al., 1998). Actin binding band 4.1 proteins provide an interesting putative link between syndecans and synaptic plasticity. Band 4.1 protein binds in complex with CASK to the syndecan-2 cytoplasmic domain (Cohen et al., 1998). This interaction probably takes place also with syndecan-3 as it binds CASK, and the region involved in band

168 4.1 protein binding is highly conserved across syndecan-2 and -3 (Hsueh and Sheng, 1999). Neural forms of band 4.1 proteins were recently shown to bind and cluster AMPA receptors (Shen et al., 2000). Interestingly, targeted mutation of band 4.1R protein impairs memory (Walensky et al., 1998) and the mutation is haploinsufficient, as is the case with the syndecan-3-null mutation. Though, as shown here, the distribution of AMPA and NMDA receptors was similar in cultured neurons from knockout and wild-type mice the possibility still remains that syndecan-3 could be linked to AMPA receptors via band 4.1 proteins and thus regulate AMPA receptor localization or activity in vivo. We suggest that syndecan-3 functions as a negative regulator of hippocampal LTP and as a modulator of hippocampus-dependent memory. The mechanism of suppression by syndecan-3 is likely to involve interactions with extracellular HB-GAM and with intracellular linker molecules that bind to the cytoplasmic domain of syndecan-3.

EXPERIMENTAL METHODS Generation of Syndecan-3-Deficient Mice Syndecan-3 genomic sequence was cloned from the 129SV mouse genomic library (Stratagene) using rat syndecan-3 cDNA as a probe (Carey et al., 1997). The identity of the clone was verified by partial sequencing. The homologous arms of targeting vector were made of 3.0- and 3.2-kb fragments from the genomic clone. Positive–negative targeting strategy was used (Mansour et al., 1988). A neo-resistance cassette was used to interrupt the syndecan-3 gene and a thymidine kinase cassette was used at the end of the targeting construct for negative selection. Syndecan-3 is composed of five exons. The targeting construct replaced the 3⬘ half of exon 3 and the whole exon 4 with the neo cassette. The targeting vector was linearized and electroporated into CJ7 mouse ES cells (Swiatek and Gridley, 1993). The cells were plated on neo-resistant feeder fibroblasts and grown in the presence of 300 ␮g/ml G418 and 2 ␮M gancyclovir for 7 days. Resistant colonies were picked and grown for DNA analysis by Southern hybridization with flanking 5⬘ and 3⬘ probes (Fig. 1). Three clones carried a targeted allele. These cell lines were grown and injected into C57BL/6J blastocysts that were then transferred into pseudopregnant mice. Chimeric animals were mated with C57BL/6J mice. Two of the ES cell lines were transmitted into the germ line. The genotype of the animals was analyzed

Kaksonen et al.

initially with Southern blotting and later routinely with PCR from tail DNA. Animals used in the studies were 129SV ⫻ C57BL/6J hybrids. Northern hybridization was done according to Nolo et al. (1995). The probe was made by RT-PCR from mouse brain RNA and contained the sequence coding for the full-length mature protein. Histological Methods Brains were dissected out from sacrificed animals and rinsed in PBS. They were then cut into two parts from the level of bregma. Pieces were then fixed overnight in 4% paraformaldehyde and 0.5% glutaraldehyde (in PBS) and embedded in paraffin after being dehydrated in increasing alcohol concentration and cleared in xylene. The whole samples were cut in 10-␮mthick serial sections. Every other section from one series was stained with hematoxylin and eosin (HE) and every other with the Bielschowsky–silver staining method. To approximate possible cell density differences between mutant and normal brains, a random sample from HE-stained sections was photographed and the cell density was estimated with the selector method (McMillan et al., 1992; Everall et al., 1997). For Golgi staining the dissected brains were immersed in 10 ml of Golgi–Cox solution (10 mg/ml potassium dichromate, 10 mg/ml mercuric chloride, and 8 mg/ml potassium chromate) and incubated in the dark at room temperature. After 2 days the Golgi–Cox solution was replaced with fresh solution and the brains were further incubated for 4 weeks. The brains were then washed in H 2O and embedded in polyethylene glycol (Mazurkiewicz and Nakane, 1972). Sections (100 ␮m) were cut with a sliding microtome, rehydrated in H 2O, developed in 25% ammonia for 15 min, washed in H 2O, dehydrated in an alcohol series (70, 95, and 100%), cleared in xylene, and mounted in Permount. Randomly chosen dendrite segments from the hippocampal CA1 area of blind-coded samples were photographed with a CCD camera and a microscope using a 100x/NA 1.3 objective. Spine density was measured using the ImageJ program (National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij/). Immunofluorescence and Confocal Microscopy Animals were deeply anesthetized with intraperitoneal injection of sodium pentobarbital and were perfused transcardially with freshly depolymerized icecold 4% paraformaldehyde– 0.15% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4). Brains were removed, and tissue blocks were prepared and postfixed in the

169

Syndecan-3 in Hippocampal LTP and Memory

same fixative for 6 h at 4°C. After washout of the fixative, the tissue blocks were cryoprotected with 15% sucrose in PBS with 0.1% sodium azide for several days at 4°C. Serial 15-␮m-thick sections through the rostrocaudal extent of the hippocampus were cut on cryostat microtome. After blocking with 5% normal goat serum, the sections were permeabilized with 0.3% Triton X and incubated with monoclonal antibodies against synaptophysin (1 ␮g/ml; Oncogene Research Products, Cambridge, MA) overnight at 4°C. The bound antibodies were visualized with FITC-conjugated secondary antibodies (4 ␮g/ml; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). Images from the stratum radiatum of the hippocampal CA1 region of blindcoded immunolabeled sections were digitized using a confocal microscope (LSM Laser Scan; Zeiss, Germany). Settings for gain, aperture, contrast, and brightness were optimized initially and kept constant throughout the study to maintain the same illumination conditions for all sections. A systematic random sampling approach was used to select sites for analysis (area of each digitized image was 3600 ␮m 2). Data analysis was performed with the use of Scion Image software. Hippocampal Neuron Culturing and Immunostaining Hippocampi were dissected from newborn mouse pups and treated with 0.25% trypsin in Hanks’ balanced salt solution (HBSS; Life Technologies, Gaithersburg, MD) for 15 min. Neurons were then triturated, washed in HBSS, and suspended in Neurobasal medium (Life Technologies) supplemented with 2% B27 (Life Technologies), 25 ␮M l-glutamic acid (Sigma–Aldrich), and 0.5 mM l-glutamine (Life Technologies). The neurons were seeded on poly-l-lysine (Sigma–Aldrich)-coated (50 ␮g/ml) coverslips (35,000 cells/cm 2) and grown for 3 weeks. The neurons were fixed with 2% paraformaldehyde in PBS for 30 min and stained with anti-syndecan-3 antibodies (Kinnunen et al., 1998) or with 10 ␮M DiI (Molecular Probes, Eugene, OR). For detection of AMPA- and NMDA-type glutamate receptors, the neurons were first permeabilized with 0.1% Triton X-100 in PBS for 10 min and then stained with polyclonal rabbit anti-GluR1 antibodies (Pharmingen, San Diego, CA) and monoclonal mouse anti-NMDAR1 antibodies (Pharmingen). In Vitro Electrophysiology Transverse hippocampal slices (400 ␮m) from 1.5- to 2-month-old mice were cut on a Vibratome. Slices were allowed to recover at room temperature at least for 1 h

before being transferred into an interface-type recording chamber (volume 1 ml, temperature 32°C). Perfusion solution containing 124 mM NaCl, 3 mM KCl, 2 mM CaCl 2, 25 mM NaHCO 3, 1.1 mM NaH 2PO 4, 2 mM Mg SO 4, and 10 mM glucose was bubbled with the mixture of 5% CO 2/95% O 2 and applied with the constant rate of 1 ml/min. A bipolar stimulating electrode was used to stimulate Schaffer collateral afferents and elicit fEPSPs recorded from the CA1 region of hippocampus. Recordings were made using glass microelectrodes (2–10 M⍀) filled with 150 mM NaCl and placed in the stratum radiatum. Baseline stimulation frequency was 0.05 Hz, pulse duration 0.1 ms. After input– output data was collected the stimulus intensity was adjusted to evoke half-maximal fEPSP amplitude. LTP was induced by 100-Hz tetanic stimulation for 1 s, during which the pulse length was doubled. The slope of fEPSP was used as an indicator of synaptic efficacy and was calculated between 20 and 80% of the maximal amplitude. The level of LTP was measured as a percentage increase of the fEPSP slope, averaged at a 5-min interval 60 min after the tetanus, and compared to the averaged baseline fEPSP slope. For the LTD studies, 1 Hz/900 pulses stimulation was used. For paired-pulse stimulation, interpulse intervals from 20 to 200 ms were tested. The LTP program (www.ltp-program.com) (Anderson and Collingridge, 2001) was used for data acquisition and analysis. Student’s t test and ANOVA were used for statistical analysis of the data. Repeatedmeasures ANOVA was applied to 5-min blocks of data recorded 1 h after HFS in order to determine differences between genotypes in LTP experiments. Differences were considered significant when P ⬍ 0.05. HB-GAM was pressure injected (⬃0.2 ␮l) into the dendritic area of hippocampal CA1 region close to the recording electrodes 10 min before train stimulation (as described in Lauri et al., 1998). Control experiments with PBS injections were carried out to ensure that the procedure did not interfere with the baseline synaptic response or induction of LTP. Recombinant HB-GAM was produced with a baculovirus expression system and purified as described previously (Raulo et al., 1992). HB-GAM was dialyzed against PBS and used at the concentration of 100 ␮g/ml. Behavioral Analysis Behavioral characterization of syndecan-3 knockout mice was carried out as described in Vo˜ ikar et al. (2001) with some modifications. We performed the tests assessing general neurological functions, nociception (hot plate), coordination (rotarod), locomotor activity (open

170 field), emotionality (elevated plus maze, light– dark exploration), and memory (spatial navigation in water maze, contextual and cued fear conditioning, conditioned taste aversion). The experimenter was unaware of the mouse genotypes. Water maze. The system consisted of a black circular water tank and computer-interfaced camera tracking system (Columbus Instruments, OH) as described previously (Vo˜ ikar et al., 2001). The training was carried out in daily blocks of trials (four per block, intertrial interval 60 s) with a hidden escape platform in a constant location. The starting positions were varied randomly across the trials. The escape latency (climbing onto the platform) was measured by stop-watch in each trial and averaged for the given training block. A mouse was allowed to stay on the platform for 15 s. If the platform was not found within 60 s the mouse was guided to the platform by the experimenter and kept there for 15 s. Thereafter the animal was taken to the transfer cage until the next trial was started. Twentyfour hours after the fourth and the seventh training blocks transfer tests (60 s) were performed with platform removed from the pool. The degree of spatial memory was estimated by the swimming time in different quadrants and the number of annulus crossings at the platform location compared to the respective annuli in the other quadrants. The data of intermediate quadrants (left and right of the trained one) were averaged. Thigmotaxis was defined as swimming in the outermost annular ring 10 cm from the wall. Cued learning was tested after completing the spatial version of the water maze. The platform was made visible with a yellow flag, and two blocks of four trials were performed. Platform location was changed for every trial and escape latencies were measured. Fear conditioning. The fear-conditioning experiments were carried out employing a computer-controlled fear-conditioning system (TSE, Bad Homburg, Germany). Training was performed in a clear acrylic cage (35 ⫻ 20 ⫻ 20 cm) within a constantly illuminated (⬃550 lx) fear-conditioning box. A loudspeaker provided a constant background noise (70 dB) for 120 s followed by a 10-kHz tone (CS, 75 dB, pulsed 5 Hz) for 30 s. The tone was terminated by a footshock (US, 0.8 mA, 2 s, constant current) delivered through a stainless steel floor grid (␾ 4 mm, distance 9 mm). Two CS-US pairings were separated by a 30-s pause. Contextual memory was tested 24 and 72 h after training in the fear-conditioning box for 180 s without tone stimulation. Freezing was scored as a behavioral parameter. Total time of freezing (absence of any move-

Kaksonen et al.

ments for more than 5 s) was measured by infrared light barriers scanned continuously with a frequency of 10 Hz. Tone (CS)-dependent memory was tested 2 h after the first contextual memory test in a novel context. New context was a similarly sized acrylic box. The light intensity was reduced to 100 lx, the floor was plain (without shock grid), and the background color was black (as opposed to white color in training context). After 120 s of free exploration in novel context the CS was applied for 120 s and freezing was measured again. Conditioned taste aversion. The mice were singly housed, water deprived, and adapted to a specific drinking schedule (two drinking sessions/day for 4 days). During each session (lasting 20 min) two bottles filled with tap water were attached to the cage. The amount of liquid consumed was determined by weighing the bottles before and after each drinking session. On day 5, one bottle with sucrose solution (15%) was presented. One hour after drinking the sucrose solution the mice were injected intraperitoneally with lithium chloride (LiCl, 0.14 M) at a dose of 2% of body weight. The choice tests were performed 24 and 72 h after conditioning. The mice could choose between drinking sucrose solution or tap water. The percentage of sucrose solution consumed as a proportion of total fluid intake (sucrose ⫹ water) was taken as an aversion index. Later the same animals were subjected to the choice between water and quinine solution (0.1%) in order to further test the gustatory function and the same formula was applied for calculating the quinine aversion. Statistical analysis. The behavioral data were analyzed by means of one-way or repeated-measures ANOVA with genotype as an independent variable, and post hoc comparisons were performed by Newman–Keuls test where appropriate. The data in figures and table are expressed as group means ⫾ standard error of mean.

ACKNOWLEDGMENTS This study has been supported by the Academy of Finland and the Technical Research Centre of Finland (Programme of Molecular Neurobiology) and the Sigrid Juse´ lius Foundation. We thank Eeva-Liisa Saarikalle for her excellent technical assistance and Susanna So¨ derlund for assistance in performing the behavioral experiments.

REFERENCES Abel, T., Martin, K. C., Bartsch, D., and Kandel, E. R. (1998). Memory suppressor genes: Inhibitory constraints on the storage of long-term memory. Science 279: 338 –341.

Syndecan-3 in Hippocampal LTP and Memory

Alexander, C. M., Reichsman, F., Hinkes, M. T., Lincecum, J., Becker, K. A., Cumberledge, S., and Bernfield, M. (2000). Syndecan-1 is required for Wnt-1-induced mammary tumorigenesis in mice. Nat. Genet. 25: 329 –332. Anderson, W. W., and Collingridge, G. L. (2001). The LTP Program: A data acquisition program for on-line analysis of long-term potentiation and other synaptic events. J. Neurosci. Methods 108: 71– 83. Amet, L. E., Lauri, S. E., Hienola, A., Croll, S. D., Lu, Y., Levorse, J. M., Prabhakaran, B., Taira, T., Rauvala H., and Vogt, T. F. (2001). Enhanced long-term potentiation in mice lacking heparin-binding growth-associated molecule. Mol. Cell. Neurosci. 17: 1014 –1024. Asundi, V. K., and Carey, D. J. (1995). Self-association of N-syndecan (syndecan-3) core protein is mediated by a novel structural motif in the transmembrane domain and ectodomain flanking region. J. Biol. Chem. 270: 26404 –26410. Bernfield, M., Go¨ tte, M., Woo Park, P., Reizes, O., Fitzgerald, M. L., Lincecum, J., and Zako, M. (1999). Functions of cell-surface heparan sulfate proteoglycans. Annu. Rev. Biochem. 68: 729 –777. Bliss, T. V. P., and Collingridge, G. L. (1993). A synaptic model of memory—Long-term potentiation in the hippocampus. Nature 361: 31–39. Brun, V. H., Ytterbo, K., Morris, R. G. M., Moser, M.-B., and Moser, E. I. (2001). Retrograde amnesia for spatial memory induced by NMDA receptor-mediated long-term potentiation. J. Neurosci. 21: 356 –362. Carey, D. J., Conner, K., Asundi, V. K., O’Mahony, D. J., Stahl, R. C., Showalter, L., Cizmeci-Smith, G., Hartman, J., and Rothblum, L. I. (1997). cDNA cloning, genomic organization, and in vivo expression of rat N-syndecan. J. Biol. Chem. 272: 2873–2879. Chernousov, M. A., and Carey, D. J. (1993). N-syndecan (syndecan 3) from neonatal rat brain binds basic fibroblast growth factor. J. Biol. Chem. 268: 16810 –16814. Cohen, A. R., Woods, D. F., Marfatia, S. M., Walther, Z., Chishti, A. H., and Anderson, J. M. (1998). Human CASK/LIN-2 binds syndecan-2 and protein 4.1 and localizes to the basolateral membrane of epithelial cells. J. Cell. Biol. 142: 129 –138. Dobrunz, L. E., and Stevens, C. F. (1997). Heterogenity of release probability, facilitation, and depletion at central synapses. Neuron 18: 995–1008. Echtermeyer, F., Streit, M., Wilcox-Adelman, S., Saoncella, S., Denhez, F., Detmar, M., and Goetinck, P. (2001). Delayed wound repair and impaired angiogenesis in mice lacking syndecan-4. J. Clin. Invest. 107: R9 –R14. Ethell, I. M., and Yamaguchi, Y. (1999). Cell surface heparan sulfate proteoglycan syndecan-2 induces the maturation of dendritic spines in rat hippocampal neurons. J. Cell Biol. 144: 575–586. Everall, I. P., DeTeresa, R., Terry, R., and Masliah, E. (1997). Comparison of two quantitative methods for the evaluation of neuronal number in the frontal cortex in Alzheimer disease. J. Neuropathol. Exp. Neurol. 56: 1202–1206. Gerlai, R., Henderson, J. T., Roder, J. C., and Jia, Z. (1998). Multiple behavioral anomalies in GluR2 mutant mice exhibiting enhanced LTP. Behav. Brain. Res. 95: 37– 45. Grootjans, J. J., Zimmermann, P., Reekmans, G., Smets, A., Degeest, G., Durr, J., and David, G. (1997). Syntenin, a PDZ protein that binds syndecan cytoplasmic domains. Proc. Natl. Acad. Sci. USA 94: 13683–13688. Houpt, T. A. (2000). Molecular neurobiology of ingestive behavior. Nutrition 16: 827– 836. Hsueh, Y. P., and Sheng, M. (1999). Regulated expression and subcellular localization of syndecan heparan sulfate proteoglycans and

171 the syndecan-binding protein CASK/LIN-2 during rat brain development. J. Neurosci. 19: 7415–7425. Hsueh, Y. P., Yang, F. C., Kharazia, V., Naisbitt, S., Cohen, A. R., Weinberg, R. J., and Sheng, M. (1998). Direct interaction of CASK/ LIN-2 and syndecan heparan sulfate proteoglycan and their overlapping distribution in neuronal synapses. J. Cell Biol. 142: 139 –151. Ishiguro, K., Kadomatsu, K., Kojima, T., Muramatsu, H., Tsuzuki, S., Nakamura, E., Kusugami, K., Saito, H., and Muramatsu T. (2000). Syndecan-4 deficiency impairs focal adhesion formation only under restricted conditions. J. Biol. Chem. 275: 5249 –5252. Kim, C. W., Goldberger, O. A., Gallo, R. L., and Bernfield, M. (1994). Members of the syndecan family of heparan sulfate proteoglycans are expressed in distinct cell-, tissue-, and development-specific patterns. Mol. Biol. Cell 5: 797– 805. Kinnunen, T., Kaksonen, M., Saarinen, J., Kalkkinen, N., Peng, H. B., and Rauvala, H. (1998). Cortactin–Src kinase signaling pathway is involved in N-syndecan-dependent neurite outgrowth. J. Biol. Chem. 273: 10702–10708. Lauri, S. E., Kaukinen, S., Kinnunen, T., Ylinen, A., Imai, S., Kaila, K., Taira, T., and Rauvala, H. (1999). Regulatory role and molecular interactions of a cell-surface heparan sulfate proteoglycan (N-syndecan) in hippocampal long-term potentiation. J. Neurosci. 19: 1226 –1235. Lauri, S. E., Rauvala, H., Kaila, K., and Taira, T. (1998). Effect of heparin-binding growth-associated molecule (HB-GAM) on synaptic transmission and early LTP in rat hippocampal slices. Eur. J. Neurosci. 10: 188 –194. Lauri, S. E., Taira, T., Kaila, K., and Rauvala, H. (1996). Activityinduced enhancement of HB-GAM expression in rat hippocampal slices. NeuroReport 7: 1670 –1674. Liao, D., Zhang, X., O’Brien, R., Ehlers, M. D., and Huganir, R. L. (1999). Regulation of morphological postsynaptic silent synapses in developing hippocampal neurons. Nat. Neurosci. 2: 37– 43. Maeda, N., and Noda, M. (1998). Involvement of receptor-like protein tyrosine phosphatase zeta/RPTPbeta and its ligand pleiotrophin/ heparin-binding growth-associated molecule (HB-GAM) in neuronal migration. J. Cell Biol. 142: 203–216. Malenka, R. C., and Nicoll, R. A. (1999). Long-term potentiation: A decade of progress? Science 285: 1870 –1874. Malleret, G., Haditsch, U., Genoux, D., Jones, M. W., Bliss, T. V., Vanhoose, A. M., Weitlauf, C., Kandel, E. R., Winder, D. G., and Mansuy, I. M. (2001). Inducible and reversible enhancement of learning, memory, and long-term potentiation by genetic inhibition of calcineurin. Cell 104: 675– 686. Manabe, T., Wyllie, D. J., Perkel, D. J., and Nicoll, R. A. (1993). Modulation of synaptic transmission and long-term potentiation: Effects of paired pulse facilitation and EPSC variance in the CA1 region of the hippocampus. J. Neurophysiol. 70: 1451–1459. Mansour, S. L., Thomas, K. R., and Capecchi, M. R. (1988). Disruption of the proto-oncogene int-2 in mouse embryo-derived stem cells: A general strategy for targeting mutations to non-selectable genes. Nature 336: 348 –352. Mazurkiewicz, J. E., and Nakane, P. K. (1972). Light and electron microscopic localization of antigens in tissues embedded in polyethylene glycol with a peroxidase-labeled antibody method. J. Histochem. Cytochem. 20: 969 –974. McMillan, A. M., and Sorensen, F. B., (1992). The efficient and unbiased estimation of nuclear size variability using the ‘selector’. J. Microsc. 165: 433– 437. Minichiello, L., Korte, M., Wolfer, D., Kuhn, R., Unsicker, K., Cestari, V., Rossi-Arnaud, C., Lipp, H. P., Bonhoeffer, T., and Klein, R.

172 (1999). Essential role for TrkB receptors in hippocampus-mediated learning. Neuron 24: 401– 414. Migaud, M., Charlesworth, P., Dempster, M., Webster, L. C., Watabe, A. M., Makhinson, M., He, Y., Ramsay, M. F., Morris, R. G., Morrison J. H., O’Dell T. J., and Grant S. G. (1998). Enhanced long-term potentiation and impaired learning in mice with mutant postsynaptic density-95 protein. Nature 396: 433– 439. Moser, E. I., Krobert, K. A., Moser, M. B., and Morris, R. G. (1998). Impaired spatial learning after saturation of long-term potentiation. Science 281: 2038 –2042. Naisbitt, S., Kim, E., Tu, J. C., Xiao, B., Sala, C., Valtschanoff, J., Weinberg, R. J., Worley, P. F., and Sheng, M. (1999). Shank, a novel family of postsynaptic density proteins that binds to the NMDA receptor/PSD95/GKAP complex and cortactin. Neuron 23: 569 –582. Nolo, R., Kaksonen, M., Raulo, E., and Rauvala, H., (1995). Co-expression of heparin-binding growth-associated molecule (HB-GAM) and N-syndecan (syndecan-3) in developing rat brain. Neurosci. Lett. 191: 39 – 42. Pavlov, I., Vo˜ ikar, V., Kaksonen, M., Lauri, S. E., Hienola, M., Taira, T., and Rauvala, H. (2002). Role of heparin-binding growth-associated molecule (HB-GAM) in hippocampal LTP and spatial learning revealed by studies on overexpressing and knockout mice. Mol. Cell Neurosci. 20: 330 –342. Raulo, E., Chernousov, M. A., Carey, D. J., Nolo, R., and Rauvala, H. (1994). Isolation of a neuronal cell surface receptor of heparin binding growth-associated molecule (HB-GAM). Identification as N-syndecan (syndecan-3). J. Biol. Chem. 269: 12999 –13004. Raulo, E., Julkunen, I., Merenmies, J., Pihlaskari, R., and Rauvala, H. (1992). Secretion and biological activities of heparin-binding growth-associated molecule: Neurite outgrowth-promoting and mitogenic actions of the recombinant and tissue-derived protein. J. Biol. Chem. 267: 11408 –11416. Reizes, O., Lincecum, J., Wang, Z., Goldberger, O., Huang, L., Kaksonen, M., Ahima, R., Hinkes, M. T., Barsh G. S., Rauvala, H., and Bernfield, M. (2001). Transgenic expression of syndecan-1 uncovers a physiological control of feeding behavior by syndecan-3. Cell 106: 105–116. Sanderson, R. D. (2001). Heparan sulfate proteoglycans in invasion and metastasis. Semin. Cell Dev. Biol. 12: 89 –98.

Kaksonen et al.

Shen, L., Liang, F., Walensky, L. D., and Huganir, R. L., (2000). Regulation of AMPA receptor GluR1 subunit surface expression by a 4.1N-linked actin cytoskeletal association. J. Neurosci. 20: 7932– 7940. Swiatek, P. J., and Gridley, T., (1993). Perinatal lethality and defects in hindbrain development in mice homozygous for a targeted mutation of the zinc finger gene Krox20. Genes Dev. 7: 2071–2084. Tang, Y. P., Shimizu, E., Dube, G. R., Rampon, C., Kerchner, G. A., Zhuo, M., Liu, G., and Tsien, J. Z. (1999). Genetic enhancement of learning and memory in mice. Nature 401: 63– 69. Tomita, S., Nicoll, R. A., and Bred D. S. (2001). PDZ protein interactions regulating glutamate receptor function and plasticity. J. Cell Biol. 153: F19 –F24. Tsien, J., Huerta, P., and Tonegawa, S. (1996). The essential role of hippocampal CA1 NMDA receptor-dependent synaptic plasticity in spatial memory. Cell 87: 1327–1338. Uetani, N., Kato, K., Ogura, H., Mizuno, K., Kawano, K., Mikoshiba, K., Yakura, H., Asano, M., and Iwakura, Y. (2000). Impaired learning with enhanced hippocampal long-term potentiation in PTP␦deficient mice. EMBO J. 19: 2775–2785. Uruno, T., Liu, J., Zhang, P., Fan, Y. x., Egile, C., Li, R., Mueller, S. C., and Zhan, X. (2001). Activation of Arp2/3 complex-mediated actin polymerization by cortactin. Nat. Cell Biol. 3: 259 –266. Vo˜ ikar, V., Ko˜ ks, S., Vasar, E., and Rauvala, H. (2001). Strain and gender differences in the behavior of mouse lines commonly used in transgenic studies. Physiol. Behav. 72: 271–281. Walensky, L. D., Shi, Z. T., Blackshaw, S., DeVries, A. C., Demas, G. E., Gascard, P., Nelson, R. J., Conboy, J. G., Rubin, E. M., Snyder, S. H., and Mohandas, N. (1998). Neurobehavioral deficits in mice lacking the erythrocyte membrane cytoskeletal protein 4.1. Curr. Biol. 8: 1269 –1272. Woods, A., and Couchman, J. R. (2000). Integrin modulation by lateral association. J. Biol. Chem. 275: 24233–24236. Yamaguchi, Y. (2001). Heparan sulfate proteoglycans in the nervous system: Their diverse roles in neurogenesis, axon guidance, and synaptogenesis. Semin. Cell Dev. Biol. 12: 99 –106. Zucker, R. S. (1989). Short-term synaptic plasticity. Annu. Rev. Neurosci. 12: 13–31. Received March 13, 2002 Revised May 16, 2002 Accepted May 28, 2002