do not facilitate cognitive flexibility

Francesco Errico,

Robert Nisticò,

Giuseppe Palma,

Mauro Federici,

Andrea Affuso,

Elisa Brilli,

Enza Topo,

Diego Centonze,

Giorgio Bernardi,

Yuri Bozzi,

Antimo D'Aniello,

Roberto Di Lauro,

Nicola B. Mercuri,

and Alessandro Usiello

a_{Laboratory of Behavioral Neuroscience, CEINGE Biotecnologie Avanzate, Naples, Italy} b_{Centro Europeo per la Ricerca sul Cervello (CERC)/Fondazione Santa Lucia, Rome, Italy} c_{Department of Pharmacobiology, Università della Calabria, Arcavacata di Rende, Cosenza, Italy} d_{IRGS, Biogem s.c.ar.l., Ariano Irpino, Avellino, Italy}

e

Istituto di Neuroscienze del CNR, Pisa, Italy

f

Department of Neurobiology, Stazione Zoologica“A. Dohrn”, Naples, Italy

g

Clinica Neurologica, Università Tor Vergata, Rome, Italy

h

Department of Health Science, Università del Molise, Campobasso, Italy Received 17 May 2007; revised 27 September 2007; accepted 27 September 2007 Available online 5 October 2007

In the present study, we demonstrate a direct role forD-aspartate in regulating hippocampal synaptic plasticity. These evidences were ob- tained using two different experimental strategies which enabled a non- physiological increase of endogenousD-aspartate levels in the mouse hippocampus: a genetic approach based on the targeted deletion ofD- aspartate oxidase gene and another based on the oral administration of D-aspartate. Overall, our results indicate that increasedD-aspartate content does not affect basal properties of synaptic transmission but enhances long-term potentiation in hippocampal slices from both genetic and pharmacological animal models. Besides electrophysiological data, behavioral analysis suggests that altered levels ofD-aspartate in the hippocampus do not perturb basal spatial learning and memory abilities, but may selectively interfere with the dynamic NMDAR- dependent processes underlying cognitive flexibility.

© 2007 Elsevier Inc. All rights reserved.

Keywords:D-aspartate; Hippocamus; NMDA receptors; Long-term poten- tiation; Morris water maze; Cognitive flexibility

Introduction

Chirality is a basic property of amino acids. Most of them exist in L- and D-forms even though the L-form was considered, until some decades ago, the sole configuration required in vertebrate tissues

(Corrigan, 1969). Only in the last 20 years, the development of sensitive chromatographic techniques revealed the presence of unexpected amounts ofD-aspartate andD-serine in the brain and

many peripheral tissues of mammals (Dunlop et al., 1986;

Hashimoto et al., 1992). Several findings pointed out a clear role for D-serine as an endogenous co-agonist of NMDA receptors (NMDARs) and its involvement in different brain functions (Martineau et al., 2006). On the other hand, the physiological role of the otherD-enantiomer present in mammalian tissues,D-aspartate, is still controversial.

In endocrine glands,D-aspartate levels increase during postnatal and adult life in concomitance with their functional maturation influencing several endocrine responses (D'Aniello, 2007). In contrast to the emerging role forD-aspartate in endocrine physio- logy, the function of thisD-amino acid in the central nervous system (CNS) remains largely unknown. High levels of freeD-aspartate appear to occur throughout the brain during early development and in newborns but rapidly decrease afterwards (Dunlop et al., 1986; Hashimoto et al., 1995; Neidle and Dunlop, 1990; Wolosker et al., 2000). The postnatal reduction ofD-aspartate levels in the CNS has been correlated with the concomitant increase ofD-Aspartate Oxi- dase (DDO) activity (Van Veldhoven et al., 1991), the only known enzyme able to metabolize selectively bicarboxylicD-amino acids (Errico et al., 2006; Hamilton, 1985; Huang et al., 2006). In particular, the adult rat hippocampus shows a very low concentration ofD-aspartate but strongly expresses DDO (Schell et al., 1997; Zaar et al., 2002), suggesting that a strict control of the enzyme over its substrate must occur in this area. Furthermore, biochemical assays demonstrated thatD-aspartate is able to bind and stimulate NMDARs (Fagg and Matus, 1984; Olverman et al., 1988; Watkins and Evans, 1981), expressed throughout this brain region.

www.elsevier.com/locate/ymcne Mol. Cell. Neurosci. 37 (2008) 236–246

⁎ Corresponding author. Alessandro Usiello, CEINGE—Biotecnologie Avanzate, Via Comunale Margherita 482, 80145 Naples, Italy. Fax: +39 08 13 73 78 08.

E-mail address:usiello@ceinge.unina.it(A. Usiello).

1_{F. E. and R. N. contributed equally to this work.}

Available online on ScienceDirect (www.sciencedirect.com). 1044-7431/$ - see front matter © 2007 Elsevier Inc. All rights reserved.

Fig. 1. Facilitatory role of increasedD-aspartate on NMDAR-dependent LTP in Ddo−/−mice. (A) Ddo−/−mice had normal basal synaptic transmission. The input–output curves were obtained from 40 data points for the two experimental groups. Although Ddo+//+and Ddo−/−mice showed different trendlines (y = 2.30x, R2= 0.95 for Ddo+/+; y = 2.58x, R2= 0.82 for Ddo−/−), the values for individual data point were not significantly different (pN0.05). Each data point is an average of two recordings. (B) Pooled data showing the paired-pulse ratio ± S.E.M., against the paired-pulse interval in Ddo+//+(6 slices, 6 mice) and Ddo−/− mice (6 slices, 6 mice). No significant differences in paired-pulse facilitation were observed between the genotypes, at every tested interval (pN0.1 for all comparisons). (C) Superimposed pooled data showing the normalized changes in field potential slope (± S.E.M.) in Ddo+/+vs Ddo−/−mice induced by TBS protocol. The degree of potentiation measured 60 min after TBS was significantly higher in Ddo−/−(n = 6) compared to Ddo+/+_{(n = 6) mice. The fEPSP traces}

were taken before and 60 min after the conditioning train. Calibration bars: 0.5 mV, 10 ms. (D) Superimposed pooled data showing the normalized changes in field potential slope (± S.E.M.) in Ddo+/+_{vs Ddo}−/−_{mice induced by depotentiation protocol vs baseline. The fEPSPs from Ddo}+/+_{were greatly depotentiated}

15–20 min following LFS, whereas mutant slices were not persistently depotentiated. (E) Inward currents (left panel) recorded from CA1 pyramidal neurons of Ddo+/+_{mice following local pressure application of}

D-aspartate (arrows), are reduced byD-AP5 in a concentration-dependent and reversible manner, and persistently diminished by MK801 (10μM). Each bar chart (right panel) represents the normalized mean±S.E.M. ofD-aspartate induced currents taken from 6 to 237 F. Errico et al. / Mol. Cell. Neurosci. 37 (2008) 236–246

Taken together, these evidences suggest that the hippocampus may represent an elective brain region in which alterations ofD- aspartate levels might unmask a potential role of such a molecule in the physiology of the mammalian CNS. To this purpose, in the present study two alternative strategies were used to increase D- aspartate levels in the mouse hippocampus: a genetic approach based on the targeted deletion of Ddo gene (Errico et al., 2006) and another based on the exogenous administration ofD-aspartate. In these two experimental animal models, we examined hippocampal-dependent synaptic plasticity, learning and memory. Our results indicated that D-aspartate enhances NMDAR-dependent long-term potentiation (LTP) without facilitating spatial cognitive flexibility.

Results

Ddo genetic inactivation induces a strong increase ofD-aspartate levels in the hippocampus without affecting its morphology

Based on the evidences that DDO plays a crucial role in the metabolism of D-aspartate (Errico et al., 2006; Hamilton, 1985; Huang et al., 2006), we first analysed the neurochemical conse- quences of Ddo gene ablation by measuringD-aspartate levels in the hippocampus of 10–12 weeks old mice. HPLC analysis indicated a strong increase of hippocampalD-aspartate content (pb0.0001) in knockout (Ddo−/−) mice (384.0 ± 25.2 nmol/g tissue) compared to their wild type (Ddo+/+) littermates (29.2 ± 2.9 nmol/g tissue). Then, in order to assess putative histological changes in the hippocampus of mutants, we performed Nissl staining and histochemistry for the pan-neuronal marker NeuN on brain sections of 10–12 weeks old

animals. No gross abnormalities were detected in hippocampal morphology of Ddo−/−mice as compared to controls. Indeed, the thickness, the size and the apparent packing density of the neurons were identical in both genotypes. Moreover, in Ddo−/−hippocampi we did not find any signs of increased cell death, neither by detecting pyknotic nuclei in Nissl or NeuN-stained sections nor by TUNEL staining (data not shown).

Hippocampal NMDAR-dependent LTP increases in Ddo−/−mice Previous studies have shown thatD-aspartate can be taken up into cells through high affinityL-glutamate andL-aspartate transporters (Davies and Johnston, 1976), stored in secretory granules and synaptically released in a calcium-dependent manner (Savage et al., 2001; Waagepetersen et al., 2001). Once released, D-aspartate regulates NMDAR-mediated transmission through the binding at the glutamate site of the NMDAR (Fagg and Matus, 1984; Olverman et al., 1988). Moreover, it has been recently shown that, in acutely isolated rat hippocampal neurons,D-aspartate also inhibits AMPA receptors (Gong et al., 2005), suggesting that thisD-amino acid may bind multiple receptor sites in addition to NMDARs.

Here, we first evaluated in vitro the basal synaptic transmission at Schaffer collateral-CA1 synapses of hippocampal slices derived from Ddo+/+_{and Ddo}−/−_{mice. In both experimental groups, the field}

excitatory postsynaptic potential (fEPSP) was unaffected by the

NMDA glutamate receptor antagonist D-AP5 (50 μM) but

completely abolished by the AMPA/kainate glutamate receptor antagonist CNQX (10μM) (data not shown). Moreover the input– output relationship, namely fEPSP slopes versus their corresponding

Fig. 2. Expression of NMDAR subunits is not altered in the hippocampus of Ddo−/−mice. (A) Representative ISHs on coronal brain sections showing NR1A, NR2A and NR2B mRNA expression in CA1 and CA3 subfields from Ddo+/+_{and Ddo}−/−_{mice. Hippocampal areas, genotypes and probes are as indicated. Scale}

bar = 250μm (NR1A, NR2A) and 200 μm (NR2B). (B) Representative immunoblots showing NR1A, NR2A and NR2B protein content in the whole hippocampus from Ddo+/+_{and Ddo}−/−_{mice. GAPDH was used as an internal standard for protein quantification. (C) Quantification of NR1A, NR2A and NR2B}

protein content in the whole hippocampus from Ddo+/+_{and Ddo}−/−_{mice. For each NMDAR subunit, O.D. levels are normalized to GAPDH O.D. values. The}

presynaptic fiber volley amplitudes, elicited at different stimulus strengths, was not significantly different (pN0.1). This data indicates that increased endogenous levels of D-aspartate do not perturb basic synaptic response properties (Fig. 1A). We next examined the integrity of the presynaptic machinery by analyzing a form of short-term synaptic plasticity using a paired-pulse stimula- tion protocol. At all interpulse intervals (20–500 ms), no significant differences in the paired-pulse ratio were observed between the two genotypes (pN0.1 for all comparisons) (Fig. 1B).

We then investigated hippocampal synaptic plasticity by studying LTP in the CA1 area. LTP was studied by conditioning the Schaffer commissural pathway with a theta burst stimulation (TBS) protocol (Arai et al., 2004; Morgan and Teyler, 2001). The post-tetanic potentiation (PTP), measured as the peak response elicited by TBS, was similar between genotypes, whereas the resulting LTP differed significantly. In fact, Ddo+/+slices showed a continually decaying potentiation that recovered almost to baseline, whereas the same stimulation protocol produced a more stable LTP in slices obtained from Ddo−/−mice (fEPSP slope % of baseline measured 1 h after TBS, Ddo+/+vs Ddo−/−: 109.2 ± 8% vs 147.3 ± 14%, pb0.001, n=6,Fig. 1C). It is well established that, after the induction of LTP by high frequency stimulation (HFS), a low frequency stimulation (LFS) can reverse a stable LTP to pre-LTP levels (Kemp and Bashir, 2001). Therefore, we analysed synaptic depotentiation following the induction of LTP by HFS at CA1 synapses in both Ddo+/+ and Ddo−/− mice. Surprisingly, whereas LTP was readily induced by tetanic stimulation in both groups, LFS only reversed LTP in Ddo+/+compared to Ddo−/−mice (fEPSP slope % of baseline measured 1 h after LFS, Ddo+/+vs Ddo−/: 113.2 ± 7% vs 138.3 ± 3%, pb0.001, n=4,Fig. 1D). These results suggest thatD- aspartate plays a role in the maintenance of hippocampal LTP by preventing time-dependent depotentiation.

Moreover, we carried out further experiments using single- electrode voltage-clamp recordings to confirm a direct involvement ofD-aspartate on NMDAR responses in hippocampal slices obtained from Ddo+/+animals. As shown inFig. 1E, puff applications ofD- aspartate caused an inward current that was antagonized in a concentration-dependent and reversible manner byD-AP5 (D-AP5

50, 150, 450μM vs washout: 57.5±5%, 37.8±5%, 28.3±5% vs

83.4 ± 6%) or by MK801 (19.4 ± 3%), competitive and non- competitive blockers of NMDARs, respectively. However, a resi- dual current still persisted even in the presence of high concentra- tions of these antagonists, suggesting that another target, yet unidentified, could be involved in the actions ofD-aspartate. Expression of NMDAR subunits is not altered in Ddo mutant hippocampus

A critical role for the induction of long-term synaptic plasticity, including LTP, is played by NMDARs through their activation and the subsequent influx of calcium into the postsynaptic neurons (Zukin and Bennett, 1995). NMDARs act as multimeric complexes formed from a common NR1 subunit and one of four NR2 subunits (NR2A-2D), which potentiate and differentiate the function of NMDARs (Dingledine et al., 1999). While NR1 subunit is ubiquitously ex- pressed, mRNAs for different NMDAR2 subunits show overlapping, but differential expression patterns in the brain even though NR2A and 2B subunits are the most prominently expressed in the hippo- campus (Ishii et al., 1993). To evaluate whether increasedD-aspartate levels may alter the expression of the NMDAR and, consequently,

−/−

mice, we performed in situ hybridization (ISH) and immunoblot experiments in the hippocampus of wild type and mutant animals. ISH studies showed comparable levels of NR1A, NR2A and NR2B mRNA expression in CA1, CA3 (Fig. 2A) and dentate gyrus (not

Fig. 3. Spatial learning and memory of Ddo−/−are not altered in the Morris water maze task. (A) Ddo+/+_{(n = 10) and Ddo}−/−_{(n = 10) mice were trained for}

5 consecutive days (acquisition phase, days 1–5) with the submerged platform located in the north-west quadrant, as indicated at day 1. Escape time, ex- pressed in seconds, was used as dependent variable. Two-way ANOVA, with escape latency as repeated measure, revealed a significant days effect [F(4, 72)=

29.396, pb0.0001], a non-significant genotype effect [F(1, 72)= 0.416,

p = 0.5271] and a non-significant genotype × days interaction [F(4, 72)= 0.075,

p = 0.9895]. Acquisition phase was followed by 3 days of reversal phase (days 6–8) in which the submerged platform was moved to the opposite position of the pool (south-east), as indicated at day 6. Two-way ANOVA, with repeated measures, displayed a significant days effect [F(2, 36)= 16.073, pb0.0001], no

genotype effect [F(1, 36)= 0.040, p = 0.8445] and no interaction between

genotype and days [F(2, 36)= 0.115, p = 0.8914]. A 60 s transfer test was

performed after acquisition, at day 6, and showed similar memory aptitudes between Ddo+/+and Ddo−/−mice, as revealed by (B) the percent of time spent in quadrants [one-way ANOVA: Ddo+/+, F(3, 27)= 54.820, pb0.0001; Ddo−/−,

F(3, 27)= 40.644, pb0.0001] and by (C) the number of annulus crossings [one-

way ANOVA: Ddo+/+, F(3, 27)= 31.603, pb0.0001; Ddo−/−, F(3, 27)= 14.965,

pb0.0001]. Another 60 s probe test was performed after the reversal phase, at day 9, and did not evidence any difference in spatial memory between genotypes. (D) Percent of time spent in quadrants [one-way ANOVA: Ddo+/+_,

F(3, 27)= 16.782, pb0.0001; Ddo−/−, F(3, 27)= 28.345, pb0.0001] and (E)

number of annulus crossings [one-way ANOVA: Ddo+/+_{, F}

(3, 27)= 24.549,

pb0.0001; Ddo−/−, F(3, 27)= 29.845, pb0.0001] were measured. The dashed

lines in panels B and D indicate the chance level of quadrant search. All values 239 F. Errico et al. / Mol. Cell. Neurosci. 37 (2008) 236–246

shown) from Ddo+/+ and Ddo−/− mice. In accordance with these findings, immunoblotting experiments revealed comparable levels of NR1A, NR2A and NR2B proteins in the whole hippocampus of wild type and mutant animals (Figs. 2B and C).

Abnormal levels ofD-aspartate in Ddo−/−mice do not alter hippocampus-dependent learning and memory

The synaptic changes that underlie hippocampal LTP are thought to represent the cellular substrate of some forms of learning and memory (Bliss and Collingridge, 1993). Therefore, we addressed whether enhanced synaptic plasticity in Ddo−/−mice altered spatial learning and memory. To clarify this, we used a reference memory version of the Morris water maze test (Morris et al., 1982). The results obtained during the five days acquisition phase (Fig. 3A, days 1–5) indicated that increased levels ofD-aspartate in Ddo−/−mice did not affect the learning of a hidden platform location, as revealed by similar escape latencies between genotypes (analysis of variance (ANOVA): genotype effect and genotype × days interaction, pN0.1). Then, in order to evaluate spatial memory in mutants after the acquisition phase, 24 h later the last trial, mice performed a 60 s retention test. The percentage of time spent in the correct quadrant and the number of the target annulus crossings were used as indexes of memory (Tsien et al., 1996). Statistical analysis indicated a significant difference in both genotypes between the percentage of time spent in the correct quadrant and the percentage of time spent in the other three quadrants (ANOVA: quadrants effect, pb0.0001 for each genotype) (Fig. 3B). Similar results were found for the number of crossings in the correct annulus (ANOVA: annuli effect, pb0.0001 for each genotype) (Fig. 3C). Thus, during the transfer test, both Ddo+/+_{and Ddo}−/− _{mice evidenced preferential spatial}

search in the goal quadrant and on the goal annulus compared to the others, but no differences between genotypes were seen (Fisher's post-hoc comparisons: goal quadrant, pN0.1; goal annulus, pN0.1). We next performed a reversal version of the Morris water maze in which the platform was moved to the opposite position (Fig. 3A,

days 6–8). This task was performed in order to evaluate the ability of mice to suppress the old spatial information and elaborate a new spatial navigation strategy (Brambilla et al., 1997). The results obtained in the reversal-learning task, displayed a similar decrease in the escape latency between genotypes throughout the 3 days training period (ANOVA: genotype effect and genotype × days interaction, pN0.1). Finally, also the spatial memory assessed in the reversal probe test, performed at day 9, showed a comparable bias spatial search between Ddo+/+ and Ddo−/− animals (Fisher's post-hoc comparisons: goal quadrant, pN0.1; goal annulus, pN0.1) in target quadrant and annulus (ANOVA: quadrants effect, pb0.0001 for each genotype; annuli effect, pb0.0001 for each genotype) (Figs. 3D and E).

D-aspartate-induced NMDAR-independent currents are not involved in spatial learning

Electrophysiological findings indicated that increased levels of D-aspartate in Ddo−/− animals induced an enhancement of hip- pocampal NMDAR-mediated synaptic plasticity. In addition, in CA1 pyramidal neurons we found thatD-aspartate evoked further currents not completely suppressed by administration of both com- petitive and non-competitive NMDAR antagonists (Fig. 1E). In order to unveil whether NMDAR-independent,D-aspartate-specific hippocampal currents may influence the execution of a spatial learning and memory task, we tested mutant animals using the Morris water maze under pharmacological blockade of NMDARs. Thirty minutes before each daily session on training period, MK801 was systemically administered in Ddo+/+and Ddo−/−mice at the doses of 0.05, 0.10 and 0.15 mg/kg (Ahlander et al., 1999). Mice of both genotypes, injected with the doses of 0.10 and 0.15 mg/kg, showed clear motor impairments in water and were, consequently, excluded from the analysis of MK801-induced cognitive deficit. Conversely, the administration of MK801 at the dose of 0.05 mg/kg produced only a learning impairment, regardless of genotypes. In fact, firstly, both Ddo+/+and Ddo−/−mice displayed a reduced

Fig. 4. Comparable effects of MK801 on the Morris water maze task in Ddo+/+and Ddo−/−mice. Animals were intraperitoneally injected with MK801 at the dose of 0.05 mg/kg (n = 9) or vehicle (n = 10) each day, 30 min before the first acquisition trial. (A) Mice were trained for 5 consecutive days (days 1–5) in the hidden platform version of the Morris water maze. Escape time, expressed in seconds, was used as dependent variable. Three-way ANOVA, with escape latency as repeated measure, revealed a significant treatment effect [F(1, 136)= 4.219, p = 0.0477] a non-significant genotype effect [F(1, 136)= 0.540, p = 0.4675], a non-

significant genotype × treatment interaction [F(1, 136)= 0.115, p = 0.7363] and non-significant days × genotype × treatment interaction [F(4, 136)= 0.717,

p = 0.5815]. (B) Swim speed, expressed as centimetres swam per second, was recorded across testing days and used as dependent variable. Three-way ANOVA, with speed as repeated measure, revealed a significant days effect [F(4, 136)= 33.155, pb0.0001] but non-significant treatment effect [F(1, 136)= 0.059,

p = 0.8092], genotype effect [F(1, 136)= 0.564, p = 0.4580], genotype × treatment interaction [F(1, 136)= 0.477, p = 0.4947] and days × genotype × treatment

spatial learning, as shown by increased escape latency in MK801 treated animals compared to their respective saline control groups (ANOVA: treatment effect, pb0.05; treatment×genotype×days interaction, pN0.1) (Fig. 4A). Secondly, the analysis of the swim speed showed no significant differences in both genotypes when compared to their respective saline-treated control groups (ANOVA: treatment effect, pN0.1 for each genotype). Moreover, similarly to the deficit induced on the acquisition of the hidden platform lo- cation, MK801 disrupted spatial memory abilities of Ddo+/+ and Ddo−/−mice, as measured by the transfer test performed 24 h after the last acquisition trial (data not shown).

Oral administration ofD-aspartate increases hippocampal D-aspartate levels and strongly enhances LTP at CA1 synapses

In order to examine the consequences of deregulated levels of D-aspartate on hippocampus-dependent functions, avoiding Ddo gene targeting, we administered for 1 month a solution of 20 mM D-aspartate to 5 weeks old C57BL/6 mice. We used animals from C57BL/6 inbred strain because it represents the genetic background

in which our Ddo mutant mice were backcrossed, in accordance to Bandbury Conference, for 5 generations before being tested (1997). HPLC analysis revealed a substantial increase ofD-aspartate in the hippocampus of C57BL/6 treated animals (102.5 ± 27.5 nmol/g tissue) in comparison to untreated mice (22.4 ± 2.7 nmol/g tissue) (pb0.001) (Fig. 5A).

Next, using the same scheduled oral administration, we ex- plored whether D-aspartate still exerted its facilitatory effect at a synaptic level on hippocampal synaptic plasticity. As previously described, both Ddo+/+and Ddo−/−mice had normal basal synaptic transmission and presynaptic transmission properties. To determine if this was confirmed in D-aspartate treated C57BL/6 mice, we tested the synaptic input–output relationship and the paired pulse facilitation in the CA1 region. As for Ddo−/−animals, also in this