Download - d-Serine signalling in the brain: friend and foe

Full text provided by www.sciencedirect.com

D-Serine signalling in the brain:friend and foeMagalie Martineau1, Gerard Baux2 and Jean-Pierre Mothet1

1 Laboratoire de Neurobiologie Morphofonctionnelle, INSERM U378, 146 Rue Leo Saignat, 33077 Bordeaux, France2 Institut de Neurobiologie Alfred Fessard, Laboratoire de Neurobiologie Cellulaire et Moleculaire, CNRS UPR 9040,

1 Avenue de la Terrasse, 91198 Gif-sur-Yvette, France

Review TRENDS in Neurosciences Vol.29 No.8

Neurons and glia talk to each other at synapses. Gliasense the level of synaptic activity and consequentlyregulate its efficacy via the release of neuromodulators.One such glia-derived modulator is D-serine, an aminoacid that serves as an endogenous ligand for the strych-nine-insensitive glycine-binding site of NMDA glutamatereceptors. Here, we provide an overview of recent find-ings on the mechanisms of its synthesis, release andclearance at synapses, with an emphasis on the dichot-omy of behaviour of this novel messenger in the brain.The discovery of the good and ugly faces of this glio-transmitter is an important issue of modern neu-roscience that has repercussions for the treatment ofbrain disorders.

IntroductionClassically, chemical synapses are viewed as polarizedelements, and neurotransmitters are seen as neuron-derived substances that are released upon depolarizationof the nerve terminal and bind to specific receptors on thepostsynaptic target cell. However, the CNS is made up ofneurons and glia, with glia being by far the more numer-ous. In an ascending phylogenic scale, the numeric pre-ponderance of glia over neurons is already notable inrodents, greatly increases in primates and reaches its peakin humans with a 4:1 ratio [1]. Glia are well positioned tocommunicate with neurons at synapses, where chemicalcommunication occurs via their fine processes that are inclose proximity to synapses [2]. Over the past decade, it hasbecome evident that this intimate structural relationshipis the locus of bidirectional communication between neu-rons and glia [3,4]. Thus, the emerging concept of thetripartite synapse considers astrocytes as dynamic part-ners of neurons at synapses, controlling synaptogenesis [5]and synaptic transmission [6]. Astrocytes are thought tocontrol these processes by sensing the level of synapticactivity and, in turn, influencing synaptic activity by theregulated release of neuromodulators [3,4]. Although glu-tamate andATP are themost well known ‘gliotransmitters’mediating this astrocyte–neuron crosstalk, it is nowobvious that D-serine, another amino acid, can be addedto the list [7]. The discovery of D-serine in the CNS revo-lutionised our thinking and forced us to reconsider the longcherished dogma that only L-isomers of amino acids occurin mammalian tissues and body fluids. The present review

Corresponding author: Mothet, J-P. ([email protected]).Available online 30 June 2006

www.sciencedirect.com 0166-2236/$ – see front matter � 2006 Elsevier Ltd. All rights reserve

highlights the most recent findings about the molecularmechanisms controlling D-serine availability in the brain,which have led to the discovery that this atypical messen-ger not only has a vital role in promoting neuronal migra-tion and synaptic plasticity but also behaves as a pro-deathsignal during excitotoxic and neuroinflammatory insults.

How does the CNS make and degrade D-serine?Little attention was paid to D-serine function in the CNSuntil the identification of the glial pyridoxal 50-phosphate(PLP)-dependent serine racemase (SR) [8]. This enzymedirectly converts L-serine into D-serine, L-serine being theonly source for endogenous D-serine in the brain. SR alsoconverts D-serine into L-serine, albeit with lower affinity.Different genes for SR have been identified in mice, ratsand humans [9–11] (Figure 1a). The distribution of SR isvery similar to that reported for endogenous D-serine, withthe highest expression in the forebrain [9,12]. In the CNS,SR is expressed mostly by glial fibrillary acidic protein(GFAP)-positive astrocytes [9,12,13] (Figure 1b), withsome expression by quiescent and activated microglia[13,14]; in the peripheral nervous system SR is expressedby Schwann cells [15]. These observations could leave uswith the idea that SR and then D-serine are strict markersof glia and that they never occur in neurons. But this is notthe case. A more liberal model is needed because SR and D-serine have been found in some neurons of the cerebralcortex [13,16] and in hindbrain glutamatergic neurons[13]. Thus, neurons constitute a source for D-serine that,althoughmuch smaller than the glial one, is not negligible.

In addition to PLP, SR activity is regulated in manydifferent ways by different cellular compounds. Mg2+ andATP are the most prevalent physiological cofactors of theenzyme that increase the rate of D-serine synthesis [17–19]. In the presence of Mg2+, ATP half-maximally activatesSR at 10 mM, which is far below the millimolar range ofATP levels found in astrocytes. Ca2+ might also representanother important SR cofactor because it binds to theenzyme, and production of D-serine is positively influencedby increases in intracellular Ca2+ concentrations in astro-cytes [20]. By contrast, glycine and a series of metabolitesrelated to L-aspartic acid (L-aspartic acid, L-asparagine anda,b-threo-3-hydroxyaspartic acid) competitively inhibit theenzyme [21,22] (Figure 1c). Because glycine concentrationsin astrocytes are �3–6 mM, glycine would constitutivelyinhibit SR activity, unless glycine and SR show differentcompartmentalization within the astrocyte cytosol [22].

d. doi:10.1016/j.tins.2006.06.008

mailto:[email protected]

http://dx.doi.org/10.1016/j.tins.2006.06.008

482 Review TRENDS in Neurosciences Vol.29 No.8

Figure 1. D-Serine cycling and energy metabolism in astrocytes. (a) All genes for serine racemase (SR) comprise seven exons (red), the first exon containing the lysine

residue (Lys56) that forms an internal Schiff base and the pyridoxal 50-phosphate (PLP)-binding region [9–11]. Human and mouse SR genes respectively encode 340 and 339

amino acid proteins, whereas the rat SR has a truncated C-terminal sequence of six residues relative to the human SR. The three proteins share 89% identity in their amino

acid sequence and all contain the consensus sequence ELFQKTGSFKIRGA for PLP binding at the N terminus [9–11]. Mutation of Lys56 inside this sequence abolishes

racemization of L-serine into D-serine [9,23]. In agreement with the presence of alternative exons in the SR gene, different transcripts encode SR proteins of different sizes in

the brain and other organs, indicating that different splice forms are dominant in different tissues [12]. Gene structures were retrieved from NCBI using the SNP gene view.

Blue lines and blue boxes represent the 50- and 30-untranslated regions of the SR genes. (b) Double staining of the primate hippocampus for SR (red) and the astrocytic

marker glial fibrillary acidic protein (GFAP, green) shows that the enzyme is expressed only in astrocytes. Regions of colocalization appear yellow in the overlay panel.

Nuclei were visualized using 40,6-diamidino-2-phenylindole (DAPI, blue). Modified, with permission, from Ref. [12]. (c) In astrocytes, SR converts L-serine into D-serine. L-

Serine can originate from the diet, or from glial glycolysis through the glucose-3-phosphoglycerate-3-phosphoserine biosynthesis pathway or from glycine through the

hydroxymethyltransferase pathway. In addition to its racemase activity, SR catalyzes the a,b-elimination of water from both L-serine and D-serine to form pyruvate. Then,

SR-derived pyruvate can be shuttled to the mitochondrial tricarboxylic acid (TCA) cycle, where it is decarboxylated to give intermediates in the production of diverse amino

acids (e.g. GABA, glutamate and glycine) and ATP; ATP in turn stimulates the a,b-elimination and racemization of D-serine. SR-derived pyruvate can also be converted by

lactate dehydrogenase into lactate, thus providing energy for neurons. Furthermore, SR-derived pyruvate can be used by pyruvate carboxylase, a specific glial enzyme, to

form aspartate. Both glycine and aspartate can in turn inhibit SR [21]. D-Serine can also be deaminated by D-amino acid oxidase (DAAO) to produce pyruvate, which can be

used in all the pathways described for SR-derived pyruvate.

SR activity is also controlled by protein–protein inter-actions. Yeast two-hybrid screens have revealed that SRbinds to glutamate-receptor-interacting protein (GRIP), ascaffolding protein for AMPA receptors. GRIP contains sixPDZ domains, a motif associated with protein–proteininteractions. SR binds selectively to PDZ-6 through theextreme part of its C-terminal portion that contains thePDZ-binding consensus sequence VSV.Mutation to glycineof the C-terminal valine residue of SR abolishes interac-tions with GRIP. Viral infection of cells with GRIP elicits afivefold increase in the production of D-serine [23]. Thephysiological relevance of the GRIP–SR interactions isfurther supported by the fact that activation of AMPAreceptors elicits a strong increase in SR activity [23]. Itis proposed that activation of AMPA receptors leads tophosphorylation of the receptor, causing dissociation ofGRIP that finally binds to SR (Figure 2). The relationshipbetween SR and AMPA receptors raises an intriguing

www.sciencedirect.com

issue. AMPA receptor currents are a predominant featureof GFAP-positive astrocytes and oligodendrocytes that dis-play large outward-rectifying currents and no or poorglutamate transport, and they are absent in glia thatexhibit variably-rectifying currents [24]. This differencemight reflect different physiological functions because onlyglia with variably-rectifying currents are capable of fastuptake of extracellular K+ [24]. Furthermore, in the hip-pocampus, astrocytes that have different current profilesare anatomically segregated [24]. Thus, SR expressionmight be restricted to a subset of ‘non-passive’ astrocytesexpressing AMPA receptors. Nevertheless, there is noevidence for such coexpression of SR and AMPA receptorsin situ and this issue requires further investigation.Recently, Fujii and colleagues [25] have identified proteininteracting with C-kinase (PICK1) as another protein thatinteracts with SR. Again, the binding of PICK1 to SRrequires the PDZ domain of PICK1 and the C terminus

Review TRENDS in Neurosciences Vol.29 No.8 483

Figure 2. D-Serine: a gliotransmitter in the core of glutamatergic synapses. Upon depolarization of nerve terminals (1), glutamate (large blue dots) is released into the

synaptic space where it activates non-NMDA receptors (NMDARs) (2) on the membrane of perisynaptic astrocytes. This activation leads to influx of Ca2+ (small red dots)

either through AMPA/kainate receptors or the reverse mode of the Ca2+–Na+ exchanger (not shown), and to release of Ca2+ from the endoplasmic reticulum (ER) in case of

metabotropic glutamate receptors (3). Activation of AMPA receptors triggers their phosphorylation by protein kinase C (PKC) bound to protein interacting with C-kinase

(PICK1), and causes dissociation of the scaffolding glutamate-receptor-interacting protein (GRIP) from the receptor. Subsequently, GRIP associates with and activates serine

racemase (SR) [23]. PICK1 can also bind the enzyme or bring PKC close to SR, leading to phosphorylation of SR [25]. How GRIP and PICK1 interact to regulate SR remains

unclear. Nevertheless, the coexpression of GRIP with SR causes a large increase in synthesis of D-serine (white triangles) [23]. D-Serine is consequently released, either from

a cytosolic pool by a transporter (T) [38] or from a vesicular pool by a Ca2+-dependent and SNARE-dependent mechanism (4) [35]. Once in the synaptic cleft, D-serine, in

concert with glutamate, activates NMDA receptors at the membrane of postsynaptic neuron, leading to the opening of ion channels (5). NMDA receptors bind to the

postsynaptic density protein PSD95, which in turn binds to neuronal nitric oxide synthase (NOS). Ca2+ entry through NMDA receptors activates NOS by a Ca2+/calmodulin-

dependent mechanism. Nitric oxide (NO) produced by NOS can diffuse to neighbouring cells, where it inhibits SR through nitrosylation and activates DAAO [31,32], which

reduces D-serine levels. Clearance of D-serine from the synaptic space is assured by Na+-dependent and Na+-independent transporters (T) on the membrane of astrocytes

and neurons (6). Although glia-derived D-serine predominates, neurons that also express SR release the amino acid upon activation of glutamate receptors, notably NMDA

receptors [16]. D-Serine is released from neurons by a non-vesicular mechanism that involves an unidentified channel or transporter. Dotted lines represent putative

pathways.

of SR. Mutation of a lysil–aspartate dipeptide within thePDZ domain of PICK1 abolishes the binding to SR, andmutation of SR by adding a tyrosine residue to the Cterminus, to obscure the C-terminal valine residue, alsoabolishes the interactions of the two proteins [25]. Incontrast to GRIP, the role of PICK1 in regulating SR isnot established. Fujii and colleagues have proposed thatPICK1 might escort protein kinase C to its target SR,resulting in phosphorylation of the enzyme (Figure 2).

An intriguing feature of SR is that it catalyzes not onlythe production of D-serine but also that of pyruvate andammonia, via the a,b-elimination of water from L-serine[17–19,26]. a,b-Elimination activity towards L-serine ishigher than the racemization activity, resulting in thesynthesis of three molecules of pyruvate per molecule ofD-serine obtained through racemization [19,22,26]. Thea,b-elimination and racemization of L-serine might beregulated differently; for example, ATP complexed toMg2+ preferentially stimulates the a,b-elimination [19].Perhaps SR switches between forming D-serine and form-ing pyruvate depending on the energy status of the cell. Inglia, D-serine production is linked to the energymetabolism


pathway and to the metabolism of other amino acids(Figure 1c). Finally, SR-derived pyruvate is a potentialsource of lactate for neurons, providing energy duringperiods of enhanced synaptic activity or neuroprotectionagainst oxidative damage and Zn2+ neurotoxicity [19].

Brain D-serine exhibits a half-life of �16 h [19] but themetabolic pathway responsible for its degradation remainselusive. Mammalian D-serine can be metabolized by theperoxisomal flavoprotein D-amino acid oxidase (DAAO), anenzyme present in astrocytes (Figure 1c) of the hindbrainand cerebellum [27–29]. Adult DAAO-deficient mice dis-play increased D-serine levels, especially in areas where itslevels are normally low [30]. Furthermore, D-serine levelsare inversely related to the regional expression of DAAOduring development [27–30]. However, DAAO levels arealmost undetectable in D-serine-rich forebrain, and inDAAO-deficient mice D-serine levels appear relativelyunchanged in this region [30]. Thus, other mechanismsprobably regulate D-serine concentrations in this brainarea. Indeed, SR also catalyzes a,b-elimination of waterfrom D-serine [19,22]. Although a,b-elimination fromD-serine is less effective than that from L-serine, astrocytes


might physiologically regulate their content of D-serine inthis way [19]. This novel function of SR represents analternative mechanism for enzymatic removal of D-serinein regions of the brain where DAAO is absent.

Notably, two recent studies provide some evidence thatDAAO and SR, when present in the same cells or at least atthe same synapse, do not work in isolation: their activitiesare regulated in opposite ways by the gaseous transmitternitric oxide (NO) [31,32]. NO inhibits SR [31] but enhancesDAAO activities [32], thus tightly controlling the levels ofD-serine in glia. In turn, D-serine might inhibit NOsynthase in glia [31]. Owing the importance of NO atglutamatergic synapses [33], these findings, although pre-liminary, might have important functional consequences:they imply that NO is a potential switch-off signal thatorchestrates the termination of D-serine signalling(Figure 2).

How do astrocytes regulate synaptic D-serine?As yet, there is no consensus about how astrocytes regulateD-serine levels at synapses. Pioneer experiments con-ducted on astrocytes in culture revealed that activationof non-NMDA receptors, notably AMPA/kainate subtypeglutamate receptors, is the main stimulus triggering theefflux of D-serine from these cells [27] (Figure 2). However,we still do not know whether this occurs in vivo. In the ratstriatum for example, no changes in D-serine extracellularconcentrations were noted in response to application ofdifferent agonists and antagonists of glutamate receptors,particularly those of AMPA/kainate receptors, which wereeffective at modulating ambient levels of glutamate [34].

Still, in vitro activation of AMPA/kainate and metabo-tropic receptors can trigger release of D-serine fromastrocytes that depends on Ca2+ and the soluble N-ethyl-maleimide-sensitive factor attachment protein receptor(SNARE) [35]. Inhibition of the vacuolar proton ATPasereduced D-serine release, probably by disrupting the uptakeof the amino acid into vesicles. These results are consistentwith vesicular storage and release of D-serine and supportthe existence of specific storing organelles and a vesiculartransportmechanismthathas yet to be identified (Figure2).Recent observations have already shown that astrocytesrelease glutamate and ATP through Ca2+-dependent exo-cytosis [4,36,37], as would be expected to occur for D-serine.Glial D-serine and glutamate releases are affected to thesame extent by tetanus neurotoxin, Ca2+ removal and inhi-bition of vesicular uptake, supporting the idea that therelease of these two gliotransmitters shares common fea-tures. Are glutamate and D-serine co-stored and/or co-released? D-Serine immunoreactivity has indeed been foundin vesicles bearing the vesicular transporter for glutamate[35]. Furthermore, activation of glutamate receptors causesthe release not only of glutamate but also of D-serine [35,37].It is not yet known whether there is a strict interplaybetween D-serine and glutamate but storage of these twogliotransmitters in the same vesicles would represent theperfect cocktail to activate NMDA receptors.

Nevertheless, Ca2+-dependent vesicular release ofD-serine from glia does not exclude the possibility thatthe amino acid is released via other mechanism, especiallyconsidering that the majority of D-serine is free in the


cytoplasm [23,35]. This would fit with the report that trans-fection of GRIP into astrocytes dramatically increases thebasal and the AMPA-evoked efflux of D-serine with noapparent storage [23]. There is mounting evidence thatastrocytes can use a chemical gradient to drive the releaseof cytosolic ATP and glutamate through activated P2X7

receptors, hemichannels or anion channels [4]. One canimagine that a similar phenomenonmight account for effluxof cytosolic D-serine. In addition, release of cytosolic D-serinecan operate through neutral amino acid transporters pre-sent on the membrane of astrocytes [38]. In primary cul-tures, D-serinefluxes are coupled to counter-movements of L-serine and, to a lesser extent, of other small neutral aminoacids, suggesting an antiporter mechanism for D-serinetransport (Figure 2). In this context, efflux of preloaded D-serine is induced by physiological concentrations of L-serinemore efficiently than by kainate, with features of the Na+-dependent B-type alanine–serine–cysteine transporter(ASCT) system [38]. Whether the vesicular and/or non-vesicular pathways are involved in the release of glial D-serine, and which of these occurs in vivo, awaits investiga-tion. However, glia do not constitute a unique releasablesource of D-serine. Indeed, Ciriacks and Bowser [34] havenoted that in vivo activation of NMDA receptors inducesrelease of D-serine in the rat striatum even though ratastrocytes do not express these receptors [39]. The second-ary source of D-serine might be neurons, at least in someregions of the brain [13,16]. Indeed, a recent study hasshown that neurons of the cerebral cortex can release D-serine upon activation of glutamate receptors, notablyNMDA receptors [16]. Nevertheless, it seems that theneuronal release of D-serine is of non-vesicular origin, incontrast to the glial release [16] (Figure 2).

Similar to the action of others neurotransmitters, that ofD-serine normally should be terminated by its clearancefrom the synaptic cleft by transporter proteins expressedby neurons and/or glia. Several candidate transporters forD-serine have been identified on the membranes of glia andneurons [40–44]. Glia express a Na+-dependent transpor-ter that has low affinity for D-serine and L-serine [40], thecharacteristics of which resemble those of the ASCT sys-tem, which carries D-serine in cultured astrocytes and inisolated retina [38,43]. Another neutral amino acid trans-porter, which is Na+-independent, has also been identified.The alanine–serine–cysteine transporter 1 (Asc-1) has ahigh affinity for D-serine and is confined to the presynapticterminals, dendrites and somata of neurons. The cellularlocalization of Asc-1 suggests that this transporter contri-butes to the synaptic clearance of D-serine by neurons[45,46]. Finally, a novel Na+–Cl�-sensitive transporterhas been described in rat brain synaptosomes [41,42]. Incontrast to the ASCT system, which has broad substrateselectivity, this serine transporter has limited affinityfor other neutral amino acids including cysteine and ala-nine. It is conceivable, therefore, that multiple transportsystems contribute simultaneously to the regulation ofD-serine concentrations at the synapse (Figure 2).

D-Serine, an active modulator of synaptic transmissionA key advance in our appreciation of the role of D-serine inthe CNS derives from observations that D-serine is found in


astrocytes that ensheathe NMDA-receptor-bearing neu-rons and that levels of D-serine parallel the ontogeny ofthese receptors [33,47]. In vitro studies teach us that D-serine is released from astrocytes upon activation of theirglutamatergic receptors [27]. All these observationsstrongly suggest that, in some regions of the brain, gluta-mate released from the nerve terminal triggers glial D-serine efflux, which in turnmodulates theNMDA receptorsat postsynaptic sites (Figure 2). The hippocampus providesan exquisite model for studying the function of D-serinebecause high densities of D-serine and NMDA receptorsoccur in the subiculum and CA1 and CA3 regions [47]. Inculture preparations of hippocampal neurons, specificenzymatic degradation of released D-serine by exogenouslyapplied DAAO considerably reduces agonist-evoked andspontaneous NMDA-receptor-driven currents (Box 1). Thehippocampus is one site where long-term potentiation(LTP) relies on NMDA receptor activation [48]. Therefore,because D-serine is an endogenous ligand for NMDA recep-tors, it was not surprising to discover that D-serine releasefrom astrocytes is involved in the induction of LTP in CA1pyramidal cell synapses [49]. Pre-treatment using exogen-ous DAAO inhibited this LTP, further supporting the idea

Box 1. The NMDA receptor with its modulatory sites – the role o

NMDA-sensitive ionotropic glutamate receptors are tetrameric com-

plexes formed by the assembly of two subunits [67,88]. Three subunit

families, designated NR1, NR2 and NR3, have been cloned [89]. NR1

occurs as eight distinct isoforms owing to three independent sites for

alternative splicing. NR2 and NR3 families consist of four and two

subunits respectively that have several splice variants. Each subunit

exhibits four functional domains (Figure Ia). The C-terminal domain

(CTD) is the locus of numerous protein–protein interactions determin-

ing trafficking and the synaptic organization of the receptor [88]. The

channel pore is formed by three transmembrane domains (TMD) and

a hairpin bend within the membrane. The N terminus and the loop

between the two transmembrane domains nearest to it contain the

S1–S2 region, which forms the binding pocket for agonists [67,88–92].

Most functional NMDA receptors in the mammalian CNS are formed

by the combination of NR1 and NR2 subunits, containing the co-

agonist and glutamate recognition sites, respectively (Figure Ia). NR3

subunits can assemble with NR1–NR2 complexes to depress NMDA

receptor responses [93]. At resting membrane potentials, the NMDA

Figure I. Allosteric modulation of NMDA receptors by D-serine. (a) Model of the NR1 an

(b) Amino acids involved in binding of D-serine (DS) to NR1. (c) Application of DAAO d

domain. Panels (a), (b) and (c) are modified, with permission, from Refs [67], [91] an


that D-serine, rather than glycine, is the endogenous ligandof NMDA receptors in this area of the brain. It is commonlybelieved that senescence is associated with impairedNMDA-receptor-dependent synaptic plasticity and notablyLTP [50]. In linewith the crucial role of D-serine in synapticplasticity, defective LTP recorded in a senescence-acceler-ated mouse strain was rescued to control levels when D-serine was given as a supplement [51]. However, this studydid not address the molecular and cellular mechanismsunderlying these synaptic plasticity deficiencies or the linkwith metabolism of D-serine. This has been addressedrecently [52] with the finding that deficient LTP in senes-cent rats is due primarily to a significant reduction in theproduction of D-serine rather than to a diminution indensity of D-serine-binding sites or affinity of D-serine forNMDA receptors. Furthermore, in agreement with theemerging role of D-serine as the major ligand for themodulatory glycine-binding site of NMDA receptors, sucha deficit in LTP is not associated with reduced levels ofglycine [52]. The ability of D-serine to control NMDA-receptor-dependent neurotransmission has been con-firmed using DAAO-deficient mice. The highest increasesin D-serine levels displayed by these mice are in the

f D-serine

receptor channel is blocked by Mg2+; depolarization removes this

block and enables ion flux when the channel binds glutamate and the

co-agonist. NMDA receptors are permeant to Ca2+, Na+ and K+.

Protons and polyamines such as spermine and spermidine bind to the

N-terminal domain (NTD) to increase permeability of the receptor

complex to cations. In addition, most NMDA receptors are inhibited

by Zn2+ in a voltage-dependent manner and are influenced by

oxidation–reduction state of the NR1 subunit.

Although glycine has generally been assumed to be the co-agonist

for glutamate at NMDA receptors [94,95], it is now obvious that D-serine

can substitute for glycine to activate these receptors. The crystal

structure of the NR1 S1–S2 ligand-binding core reveals that binding of

D-serine is similar to that of glycine and involves the same electrostatic

interactions with side chains of Arg523 and Asp732 in the loop [91]

(Figure Ib). Direct demonstration that D-serine produced in the brain can

activate NMDA receptors is derived from the observations that

application of purified D-amino acid oxidase (DAAO), the D-serine

degrading enzyme, decreased NMDA-induced currents [54] (Figure Ic).

d NR2 NMDA receptor subunits, depicting functional domains and binding sites.

ecreases NMDA-induced currents. Additional abbreviation: ABD, agonist-binding

d [54], respectively.


brainstem and spinal cord [53]. As expected, NMDA-recep-tor-mediated excitatory postsynaptic currents recordedfrom spinal cord dorsal horn neurons were significantlypotentiated inmutantmice [53]. Finally, knockoutmice forthe neuronal transporter Asc-1 display NMDA-receptor-dependent hyperexcitability, presumably resulting fromelevated extracellular D-serine levels [44]. The fact thatD-serine is present in a subset of neurons [13,16] would fitperfectly with the presence of Asc-1 at the surface ofneurons.

Electrophysiological data obtained in the hippocampus[54] have shown that the level of occupancy of the NMDAreceptor glycine-binding site is higher at synaptic than atextrasynaptic receptors. Whether this is related to theexistence of microdomains for gliotransmitter release, tothe proximity of synaptically released glutamate trigger-ing D-serine release, or both, remains to be determined.Glia have functional compartments or microdomainswhere localized high increases in Ca2+ concentration occur;these tightly enwrap synapses and contain the vesicles forglutamate [4,37]. By analogy to glutamate, it is tempting tospeculate that D-serine is released from fine glial processesthat contact synapses bearing NMDA receptors.

From the preceding discussion, it is obvious that astro-cytic D-serine modulates NMDA-receptor-dependentneurotransmission and synaptic plasticity in the CNS.D-Serine might also be involved in the coding and proces-sing of sensory information. For example, in the retina, SRlocalized in radial Muller glia controls NMDA-receptor-mediated responses resulting from application of NMDA orlight stimulation, through the production of D-serine [55].Another potential model is the peripheral vestibular sys-tem, where there are high levels of SR, DAAO and D-serinein the sensory epithelia [56]. Finally, and not at leastimportantly, astrocytes are not the only glia that synthe-size and release D-serine: microglia do too [14]. The func-tion of D-serine secreted by quiescent microglia is not yetknown. But there is growing evidence that neurons in non-pathological states are frequently contacted by satellitemicroglia, which might promote synaptogenesis andsynaptic plasticity using a large repertoire of secretedfactors, such as brain-derived neurotrophic factor (BDNF)or neurotrophin 3 (NT-3) [57].

D-Serine, a motility-promoting signal duringdevelopmentRadial migration of immature granule cells in the devel-oping cerebellum is one of the best-characterized instancesof the participation of NMDA receptors in neuronal migra-tion [58]. As they migrate through the molecular layer,immature neurons are guided by Bergmann glia (Figure 3).Real-time observation of cell migration in acute cerebellarslices revealed that glutamate, acting on NMDA receptors,has a crucial, modulatory effect on promoting the motilityof granule cells through the molecular layer [59]. Evidenceincludes that granule cells start their radial migrationafter the expression of the NMDA receptors on the plas-malemmal surface, and that antagonists of these receptorssignificantly decrease the rate of glia-guided radial neuro-nal migration. By contrast, the rate of granule cell move-ment is increased by removal of Mg2+ or by application of


NMDA or the co-agonist glycine [59]. How NMDA recep-tors of migrating immature neurons are activated remainscontroversial because migrating neurons do not formsynapses before complete translocation to the internalgranule layer. An attractive hypothesis is that glutamatereleased by Bergmann glia activates immature NMDAreceptors in a non-synaptic, paracrine mode [59]. MightD-serine have a role in this?

In fact, SR is present in the Bergmann glia of thedeveloping cerebellum (Figure 3) and D-serine levels peakat postnatal day (P)14, the time of intense granule cellmigration, and subsequently diminish [33]. D-Serinereleased by Bergmann glia seems to be essential in pro-moting the migration of granule cells through activation ofNMDA receptors [23]. Inhibition of SR, using a new seriesof specific inhibitors or by applying DAAO on cerebellarslices to degrade extracellular released D-serine, blockedthe migration of granule cells via inhibition of NMDA-receptor-mediated Ca2+ influx [23]. In addition, GRIP ade-noviral infection of the developing cerebellum increases D-serine levels through activation of SR (Figure 2) and con-comitantly increases the rate at which granule cellsmigrate to their final location. D-Serine might also parti-cipate in thematuration (i.e. synaptogenesis) of developingneural network because its ontogeny in Bergmann gliaparallels expression of the NR2A and NR2B NMDA recep-tor subunits in Purkinje cells [47].

The motility-promoting role of D-serine is probably notrestricted to the postnatal development of the cerebellum.Indeed, SR is present in the perireticular nucleus, a tran-sient area of the human foetal brain that is thought to beinvolved in the guidance of corticofugal and thalamocor-tical fibres [60]. Additionally, D-serine synthesized in theplacenta is exported into the foetal circulation through theamino acid transporter ATB [61]. NMDA receptors arepresent early during gestation [62], so D-serine is wellpositioned both spatially and temporally to controlNMDA-receptor-mediated neuronal migration and synap-togenesis. Because blocking NMDA receptors during neo-corticogenesis [63] or genetically induced alterations inthese receptors [64] results in severe abnormal corticaldevelopment, disrupting D-serine metabolism duringembryonic and early postnatal life might lead to the samedevelopmental defects. Notably, impairment in cerebellardevelopment and maturation fits with a specific shut-downof DAAO gene expression [65], supporting the hypothesisthat alteration in the signalling of D-serine and/or NMDAreceptors promotes neuronal degeneration and inhibitionof synaptogenesis.

But, D-serine, a pro-death signalShould we conclude that D-serine is a good Samaritan thatsubtly regulates NMDA receptor activity? It is well knownthat NMDA receptors can cause cell death in many neu-ropathological conditions when they are intensely orchronically activated [66–68]. Increased extracellularlevels of glutamate, resulting from downregulation of itsuptake system [69] or from active release [70], are theprimary cause of neuronal death following excessiveNMDA receptor activation [68]. Because D-serine regulatesNMDA receptor activity, and glia are suspected to support


Figure 3. Granule cell migration during postnatal development of the cerebellum, and its regulation by D-serine. (a) In the early postnatal cerebellum, granule cells migrate

from the germinal zone, represented by the external granule cell layer (EGL), to the inner granule cell layer (IGL) and Purkinje cell layers. The migrating cells adopt two types

of trajectory. In the EGL (stages 1–5), immature granule cells migrate tangentially to the bottom of the layer. During their tangential migration, they extend two horizontal

processes near the top of the EGL and develop a vertical process when approaching the border between the EGL and the molecular layer (ML). In the ML (stages 6–8),

granule cells undergo radial migration along the processes of Bergmann glia (Bg). In stages 9–11, granule cells migrate radially in the IGL, independently of glia, completing

their migration in the middle or the bottom of the layer. Additional abbreviations: G, Golgi cell; g, postmigratory granule cell; P, Purkinje cell. (b) Relationship between the

activity of granule cell NMDA receptors and the position of the granule cells along the migratory pathway. The histogram represents the mean frequency (�SEM) of

spontaneous NMDA receptor channel activity in the EGL, ML + PCL and IGL of developing cerebellar slices. (c–e) D-Serine released by Bergmann glia triggers the radial

migration of granule cells in cerebellar explants. DAAO and SR inhibitors decrease granule cell migration. (c) Pretreatment of cerebellar slices with DAAO before the

migration assay (white bars) or continuous treatment with DAAO (grey bars) considerably affects granule cell migration. D-Serine and sodium benzoate (NaBENZ), an

inhibitor of DAAO, rescue granule cell migration from the inhibitory effect of DAAO. (d) Phenazine methosulfate (Met-Phen) and phenazine ethosulfate (Et-Phen), two

inhibitors of SR, block granule cell migration whereas phenazine exerts modest inhibitory effects. The inhibitory effects of SR inhibitors are reversed by adding D-serine. (e)

Inhibiting SR blocks granule cell Ca2+ transients in three different cells (1,2 and 3), most likely through reduction of NMDA receptor activity [59]. Panels (a), (b) and (c–e) were

modified with permission from Refs [59], [96] and [23], respectively.

the development of neurotoxicity [71,72], D-serine mightcompromise neuronal survival and function by exacerbat-ing the effect of glutamate when its extracellular levels arealtered. Research over the past two years has offered muchcredibility to this notion, particularly in relation to twopathological conditions.

Various lines of evidence support a central role of amy-loid b-peptide (Ab), the major component of neuritic pla-ques, in the pathogenesis of Alzheimer disease (AD) [73].Ab causes an inflammatory phenotype in microglia, whichin turn triggers neuronal death by excitotoxicity [15,73,74].Furthermore, the activity of NMDA receptors is increasedin the AD brain and memantine, an antagonist with mod-erate affinity for these receptors, is neuroprotective [68].Recent observations also support the idea that D-serinemight participate in the pathogenic signatures of AD,excitotoxicity and neuroinflammation. Indeed, it has beendiscovered that the hippocampus of AD patients displayshigher SR activity and that Ab can stimulate, in vitro, thesynthesis and the release of neurotoxic levels of D-serine


from microglia [14] (Figure 4). Conversely, DAAO canprotect neurons against Ab-induced Ca2+ overload andneurotoxicity, thus providing evidence that D-serine is adeath signal induced byAb [14]. How does Ab stimulate SRactivity? Analysis of the first intron of the SR gene revealsthe presence of several activator protein-1 (AP-1)-bindingsites for transcriptional regulation [15]. Deletion and site-specific mutagenesis show that two AP-1-binding sites are,in fact, responsible for responsiveness to Ab and to lipo-polysaccharide (LPS), another relevant proinflammatorystimulus [14,15] (Figure 4). These proinflammatory stimuliinduce transcription of SR through activation of the Jun N-terminal kinase (JNK) pathway, implicating c-Fos andJunB as the interacting proteins at the AP-1 sites [15].Ab provokes Ca2+ transients in microglia [75] and mighttherefore also regulate the activity of SR at a post-transla-tion level, because the enzyme activity is influenced byCa2+ [20]. All these in vitro observations fit with thedramatic increase in D-serine levels measured in the cere-brospinal fluid of AD patients [76]. Nevertheless, no


Figure 4. Role of glial-derived D-serine in the neurotoxicity associated with neuroinflammation. (a) Schematic model. Normally, glia (microglia and astrocytes) are

quiescent. Chronic CNS inflammation in Alzheimer’s disease, caused by the amyloid b-peptide (Ab) or acute inflammation by the endotoxin lipopolysaccharide (LPS),

activates glia (1), leading to release of cell-death mediators such as glutamate, cytokines and/or reactive oxygen species (ROS) [71,74] (2). In addition, inflammatory stimuli

such as Ab and LPS stimulate serine racemase (SR) transcription through two AP-1 elements present in the upstream region of the SR promoter. Increased SR transcription

leads to generation and efflux of deleterious levels of D-serine [14,15] (2). Subsequent to its release, D-serine exacerbates the effects of glutamate at synapses, leading to

neuronal damage through over-activation of NMDA receptors (3), causing Ca2+ overload [14,15] and strong mitochondrial ROS generation. Oxygen free radicals contribute

to membrane lipid peroxidation, damage to DNA and proteins, and production of inflammatory mediators (4) that can in turn reinforce the inflammation reciprocally (5).

This model of disrupted glia-to-neuron D-serine signalling might be associated with inflammation resulting from ischemia [83,84] or from opioid (morphine) administration

[81]. Blue text represents the prime pathological stimuli, and red text the downstream cascades leading to neuron death. (b) Results supporting the model in (a). (i)

Enhanced activity of SR regulatory elements (AP-1) in response to LPS. (ii) Enhanced D-serine levels in microglial culture medium in response to Ab or LPS. (iii)

Chromatograms obtained from control microglia and Ab-treated microglia media illustrate accumulation of D-serine. (iv) Ab(1–42)-activated microglia induce neurotoxicity

and this deleterious effect on hippocampal neurons is prevented by treatment of the cultures with DAAO. Photomicrographs document neuronal integrity over 24 h. Results

in (b,i,ii) and chromatograms in (b,iii) are modified, with permission, from Refs [14,15]; photographs in (b,iv) are reproduced courtesy of Steve Barger [14].

change has been noted in D-serine levels in the brainparenchyma or serum of such patients [77,78]. The appar-ent discrepancy between the different studies might bemulti-factorial. Notably, levels of D-serine itself might beexpected to increase early in AD progression; thus, anyelevation might be difficult to detect after the disease hasprogressed to its final stages.

The demonstration that Ab activates the JNK-mediatedmodulation of AP-1-binding activity might have more gen-eral repercussions for our understanding of other neurode-generative and inflammatory diseases. Considerable datahave shown that JNK is a crossroads for cellular stress andapoptosis-inducing stimuli [79], and agents targeting thispathway are neuroprotective [80]. In this context, the recentreport that acute morphine administration in rats [81]increases expression of SR transcripts illustrates thisnotion, as does the finding that the AP-1–JNK signallingpathway is activated in analgesia [82].

Ischemia provides another pathological situation forstudying the role of D-serine in neuroinflammation andexcitotoxicity. Exposing neurons to oxygen-free and glu-cose-free conditions (simulated ischemia) causes NMDA-receptor-dependent cell death that is prevented by NMDAreceptor antagonists [66,68]. Katsuki et al. have providedevidence that endogenous D-serine has a significant role inneuronal damage resulting from simulated ischemia [83]:application of DAAO to ischemic brain slices protectedneurons from death, probably by reducing extracellular


D-serine levels, although D-serine levels were not moni-tored in this study. Katsuki et al. have also demonstrated aprotective effect of exogenous DAAO against NMDA-induced excitotoxicity. This is consistent with the recentreport by Wolosker and colleagues [84] that D-serine dea-minase offers protection against NMDA-induced neuronalinsults by shutting-down the activity of NMDA receptors tophysiological levels. Nevertheless, these studies did notaddress the source (astrocytes, microglia and/or neurons)of D-serine and the mechanisms leading to deleteriousextracellular levels of the amino acid. It is plausible thatin these cell types, ischemia and NMDA might upregulateSR expression, and induce release of excessive amounts ofD-serine.

It has been shown that extrasynaptic NMDA receptorstrigger neuronal death by shutting-off the pro-survivalactivity of cAMP-response-element-binding protein(CREB), whereas synaptic NMDA receptor activation pro-motes the survival of neurons by opposing a positive actionon CREB activity [66]. Amajor focus for future work will beto define whether this is the molecular basis for the exci-totoxic influence of D-serine when present in excess.

Concluding remarksRecent literature has unveiled multiple roles for D-serine inthe brain. This atypical amino acid can serve as a gliotrans-mitter that modulates neurotransmission at glutamatergicsynapses, and is a motility-promoting signal important for


development and maturation of the CNS. However, it canalso cause cell deathwhen in excess, through overactivationof NMDA receptors in neuropathological conditions. Thefeatures of D-serine activity thus parallel those of NMDAreceptors.Of course,manyquestions remainunsolved. Is toolittle D-serine also deleterious for neurons? This is likely,because hypofunction of NMDA receptors render neuronsmore vulnerable to trauma and causes apoptosis duringdevelopment [66], and D-serine metabolism deficiency islinked to the etiology of schizophrenia [25,85,86].

Although much evidence strongly supports the notionthat D-serine is a major endogenous ligand for the glycine-binding site of NMDA receptors, this does not mean thatglycine is not also a ligand. Clearly, ambient glycine levelswould be sufficient in most brain areas to account forNMDA receptor activation, and genetically targeted inac-tivation of glycine transporters leads to major changes inglutamate-mediated synaptic transmission [87]. Thereremains much to be done to delineate the respective con-tributions of D-serine and glycine at glutamatergicsynapses in physiological and pathological conditions. Itis a fascinating but complicated task, because there aremanyNMDA receptor subtypes and these all have intrinsicproperties that control their trafficking, their pharmaco-logical features and their expression during development.In addition, NMDA receptors are central to many physio-logical and pathological signalling events. Use of geneticanimal models to disrupt D-serine metabolism and thedevelopment of new tools to visualize D-serine and glycinein vivo should aid the translation of our cell biologicalknowledge into a more physiological context, and help todefine the role of each agonist in regulating NMDA-recep-tor-dependent cell death, cell survival or physiologicalpathways.

AcknowledgementsWe thank Drs Dionysia Theodosis and Elisabeth Traiffort for graciouslyproviding critical readings of the manuscript. We also acknowledge LydieCollet and Marielle Rimard for their technical assistance in preparing thefigures. We apologize to those whose work we were unable to cite owing tospace limitations. J.P.M. is supported by grants from the CNRS andServier laboratories. M.M. is a recipient of a PhD fellowship from the‘Ministere de l’Enseignement, de la Recherche et de la Technologie’.

References1 Nedergaard, M. et al. (2003) New roles for astrocytes: redefining the

functional architecture of the brain. Trends Neurosci. 26, 523–5302 Ventura, R. and Harris, K.M. (1999) Three-dimensional relationships

between hippocampal synapses and astrocytes. J. Neurosci. 19, 6897–6906

3 Haydon, P.G. (2001) Glia: listening and talking to the synapse. Nat.Rev. Neurosci. 2, 185–193

4 Volterra, A. and Meldolesi, J. (2005) Astrocytes, from brain glue tocommunication elements: the revolution continues.Nat. Rev. Neurosci.6, 626–640

5 Pfrieger, F.W. (2002) Role of glia in synapse development. Curr. Opin.Neurobiol. 12, 486–490

6 Oliet, S.H. et al. (2004) Glial modulation of synaptic transmission:Insights from the supraoptic nucleus of the hypothalamus. Glia 47,258–267

7 Wolosker, H. et al. (2002) Neurobiology through the looking-glass: D-serine as a new glial-derived transmitter.Neurochem. Int. 41, 327–332

8 Schell, M.J. (2004) The N-methyl D-aspartate receptor glycine site andD-serine metabolism: an evolutionary perspective. Philos. Trans. R.Soc. Lond. B Biol. Sci. 359, 943–964


9 Wolosker,H. et al. (1999)Serineracemase: a glial enzymesynthesizingD-serine to regulate glutamate-N-methyl-D-aspartate neurotransmission.Proc. Natl. Acad. Sci. U. S. A. 96, 13409–13414

10 DeMiranda, J. et al. (2000) Human serine racemase:molecular cloning,genomic organization and functional analysis. Gene 256, 183–188

11 Konno, R. (2003) Rat cerebral serine racemase: amino acid deletion andtruncation at carboxy terminus. Neurosci. Lett. 349, 111–114

12 Xia, M. et al. (2004) Characterization and localization of a humanserine racemase. Mol. Brain Res. 125, 96–104

13 Williams, S.M. et al. (2006) Immunocytochemical analysis of D-serinedistribution in the mammalian brain reveals novel anatomicalcompartmentalizations in glia and neurons. Glia 53, 401–411

14 Wu, S.Z. et al. (2004) Induction of serine racemase expression and D-serine release from microglia by amyloid b-peptide. J.Neuroinflammation 1, 2

15 Wu, S. and Barger, S.W. (2004) Induction of serine racemase byinflammatory stimuli is dependent on AP-1. Ann. N. Y. Acad. Sci.1035, 133–146

16 Kartvelishvily, E. et al. (2006) Neuron-derived D-serine: novel means toactivate N-methyl-D-aspartate receptors. J. Biol. Chem. 281, 14151–14162

17 De Miranda, J. et al. (2002) Cofactors of serine racemase thatphysiologically stimulate the synthesis of the N-methyl-D-aspartate(NMDA) receptor coagonist D-serine. Proc. Natl. Acad. Sci. U. S. A. 99,14542–14547

18 Neidle, A. and Dunlop, D.S. (2002) Allosteric regulation of mouse brainserine racemase. Neurochem. Res. 27, 1719–1724

19 Foltyn, V.N. et al. (2005) Serine racemase modulates intracellular D-serine levels through an a,b-elimination activity. J. Biol. Chem. 280,1754–1763

20 Cook, S.P. et al. (2002) Direct calcium binding results in activation ofbrain serine racemase. J. Biol. Chem. 277, 27782–27792

21 Dunlop, D.S. and Neidle, A. (2005) Regulation of serine racemaseactivity by amino acids. Mol. Brain Res. 133, 208–214

22 Strisovsky, K. et al. (2005) Dual substrate and reaction specificity inmouse serine racemase: identification of high-affinity dicarboxylatesubstrate and inhibitors and analysis of the b-eliminase activity.Biochemistry 44, 13091–13100

23 Kim, P.M. et al. (2005) Serine racemase: activation by glutamateneurotransmission via glutamate receptor interacting protein andmediation of neuronal migration. Proc. Natl. Acad. Sci. U. S. A. 102,2105–2110

24 Zhou, M. and Kimelberg, H.K. (2001) Freshly isolated hippocampalCA1 astrocytes comprise two populations differing in glutamatetransporter andAMPA receptor expression. J. Neurosci. 21, 7901–7908

25 Fujii, K. et al. (2006) Serine racemase binds to PICK1: potentialrelevance to schizophrenia. Mol. Psychiatry 11, 150–157

26 Strisovsky, K. et al. (2003) Mouse brain serine racemase catalyzesspecific elimination of L-serine to pyruvate. FEBS Lett. 535, 44–48

27 Schell, M.J. et al. (1995) D-serine, an endogenous synaptic modulator:localization to astrocytes and glutamate-stimulated release. Proc.Natl. Acad. Sci. U. S. A. 92, 3948–3952

28 Moreno, S. et al. (1999) Immunocytochemical localization of D-aminoacid oxidase in rat brain. J. Neurocytol. 28, 169–185

29 Urai, Y. et al. (2002) Gene expression of D-amino acid oxidase incultured rat astrocytes: regional and cell type specific expression.Neurosci. Lett. 324, 101–104

30 Hamase, K. et al. (2005) Sensitive determination of D-amino acids inmammals and the effect of D-amino-acid oxidase activity on theiramounts. Biol. Pharm. Bull. 28, 1578–1584

31 Shoji, K. et al. (2006) Regulation of serine racemase activity by D-serineand nitric oxide in human glioblastoma cells.Neurosci. Lett. 392, 75–78

32 Shoji, K. et al. (2006)Mutual regulation between serine and nitric oxidemetabolism in human glioblastoma cells. Neurosci. Lett. 394, 163–167

33 Boehning, D. and Snyder, S.H. (2003) Novel neural modulators. Annu.Rev. Neurosci. 26, 105–131

34 Ciriacks, C.M. and Bowser, M.T. (2006) Measuring the effect ofglutamate receptor agonists on extracellular D-serine concentrations inthe rat striatum using online microdialysis-capillary electrophoresis.Neurosci. Lett. 393, 200–205

35 Mothet, J.P. et al. (2005) Glutamate receptor activation triggers acalcium-dependent and SNARE protein-dependent release of thegliotransmitter D-serine. Proc. Natl. Acad. Sci. U. S. A. 102, 5606–5611


36 Coco, S. et al. (2003) Storage and release of ATP from astrocytes inculture. J. Biol. Chem. 278, 1354–1362

37 Bezzi, P. et al. (2004) Astrocytes contain a vesicular compartment thatis competent for regulated exocytosis of glutamate. Nat. Neurosci. 7,613–620

38 Ribeiro, C.S. et al. (2002) Glial transport of the neuromodulator D-serine. Brain Res. 929, 202–209

39 Verkhratsky, A. et al. (1998) Glial calcium: homeostasis and signalingfunction. Physiol. Rev. 78, 99–141

40 Hayashi, F. et al. (1997) Uptake of D- and L-serine in C6 glioma cells.Neurosci. Lett. 239, 85–88

41 Yamamoto, N. et al. (2001) Uptake of D-serine by synaptosomal P2fraction isolated from rat brain. Synapse 42, 84–86

42 Javitt, D.C. et al. (2002) A novel alanine-insensitive D-serinetransporter in rat brain synaptosomal. Brain Res. 941, 146–149

43 O’Brien, K.B. et al. (2005) D-Serine uptake by isolated retinas isconsistent with ASCT-mediated transport. Neurosci. Lett. 385, 58–63

44 Xie, X. et al. (2005) Lack of the alanine-serine-cysteine transporter 1causes tremors, seizures, and early postnatal death in mice. Brain Res.1052, 212–221

45 Helboe, L. et al. (2003) Distribution and pharmacology of alanine–serine–cysteine transporter 1 (ASC-1) in rodent brain.Eur. J. Neurosci.18, 2227–2238

46 Matsuo, H. et al. (2004) High affinity D- and L-serine transporter ASC-1:cloning and dendritic localization in the rat cerebral and cerebellarcortices. Neurosci. Lett. 358, 123–126

47 Schell, M.J. et al. (1997) D-Serine as a neuromodulator: regional anddevelopmental localizations in rat brain glia resemble NMDAreceptors. J. Neurosci. 17, 1604–1615

48 Nicoll, R.A. (2003) Expression mechanisms underlying long-termpotentiation: a postsynaptic view. Philos. Trans. R. Soc. Lond. BBiol. Sci. 358, 721–726

49 Yang, Y. et al. (2003) Contribution of astrocytes to hippocampal long-term potentiation through release of D-serine. Proc. Natl. Acad. Sci. U.S. A. 100, 15194–15199

50 Barnes, C.A. (2003) Long-term potentiation and the ageing brain.Philos. Trans. R. Soc. Lond. B Biol. Sci. 358, 765–772

51 Yang, S. et al. (2005) D-serine enhances impaired long-termpotentiation in CA1 subfield of hippocampal slices from agedsenescence-accelerated mouse prone/8. Neurosci. Lett. 379, 7–12

52 Mothet, J.P. et al. (2006) A critical role for the glial-derivedneuromodulator D-serine in the age-related deficits of cellularmechanisms of learning and memory. Aging Cell 5, 267–274

53 Wake, K. et al. (2001) Exaggerated responses to chronic nociceptivestimuli and enhancement of N-methyl-D-aspartate receptor-mediatedsynaptic transmission in mutant mice lacking D-amino-acid oxidase.Neurosci. Lett. 297, 25–28

54 Mothet, J.P. et al. (2000) D-serine is an endogenous ligand for theglycine site of the N-methyl- D -aspartate receptor. Proc. Natl. Acad.Sci. U. S. A. 97, 4926–4931

55 Stevens, E.R. et al. (2003) D-serine and serine racemase are present inthe vertebrate retina and contribute to the physiological activation ofNMDA receptors. Proc. Natl. Acad. Sci. U. S. A. 100, 6789–6794

56 Dememes, D. et al. (2006) Cellular distribution of D-serine, serineracemase and D-amino acid oxidase in the rat vestibular sensoryepithelia. Neuroscience 137, 991–997

57 Nakajima, K. et al. (2001) Neurotrophin secretion from culturedmicroglia. J. Neurosci. Res. 65, 322–331

58 Hatten, M.E. (1999) Central nervous system neuronal migration.Annu. Rev. Neurosci. 22, 511–539

59 Yacubova, E. and Komuro, H. (2003) Cellular and molecularmechanisms of cerebellar granule cell migration. Cell Biochem.Biophys. 37, 213–234

60 Hepner, F. et al. (2005) Mass spectrometrical analysis of human serineracemase in foetal brain. J. Neural Transm. 112, 805–811

61 Chen, Z. et al. (2004) Serine racemase and D-serine transport in humanplacenta and evidence for a transplacental gradient for D-serine inhumans. J. Soc. Gynecol. Investig. 11, 294–303

62 Ritter, L.M. et al. (2001) Ontogeny of ionotropic glutamate receptorexpression in human fetal brain. Dev. Brain Res. 127, 123–133

63 Reiprich, P. et al. (2005) Neonatal NMDA receptor blockade disturbsneuronal migration in rat somatosensory cortex in vivo. Cereb. Cortex15, 349–358


64 Gressens, P. (2000) Mechanisms and disturbances of neuronalmigration. Pediatr. Res. 48, 725–730

65 Taharaguchi, S. et al. (2003) Impaired development of the cerebellumin transgenic mice expressing the immediate-early protein IE180 ofpseudorabies virus. Virology 307, 243–254

66 Hardingham, G.E. and Bading, H. (2003) The Yin and Yang of NMDAreceptor signalling. Trends Neurosci. 26, 81–89

67 Kemp, J.A. and McKernan, R.M. (2002) NMDA receptor pathways asdrug targets. Nat. Neurosci. 5, 1039–1042

68 Lipton, S.A. (2004) Failures and successes of NMDA receptorantagonists: molecular basis for the use of open-channel blockerslike memantine in the treatment of acute and chronic neurologicinsults. NeuroRx 1, 101–110

69 Rao, S.D. et al. (2003) Disruption of glial glutamate transport byreactive oxygen species produced in motor neurons. J. Neurosci. 23,2627–2633

70 Takano, T. et al. (2001) Glutamate release promotes growth ofmalignant gliomas. Nat. Med. 7, 1010–1015

71 Aschner, M. et al. (1999) Glial cells in neurotoxicity development.Annu. Rev. Pharmacol. Toxicol. 39, 151–173

72 Swanson, R.A. et al. (2004) Astrocyte influences on ischemic neuronaldeath. Curr. Mol. Med. 4, 193–205

73 Butterfield, D.A. and Boyd-Kimball, D. (2004) Amyloid b-peptide(1-42)contributes to the oxidative stress and neurodegeneration found inAlzheimer disease brain. Brain Pathol. 14, 426–432

74 Barger, S.W. and Basile, A.S. (2001) Activation of microglia by secretedamyloid precursor protein evokes release of glutamate by cystineexchange and attenuates synaptic function. J. Neurochem. 76, 846–854

75 Silei, V. et al. (1999) Activation of microglial cells by PrP and b-amyloidfragments raises intracellular calcium through L-type voltagesensitive calcium channels. Brain Res. 818, 168–170

76 Fisher, G. et al. (1998) Free D- and L-amino acids in ventricularcerebrospinal fluid from Alzheimer and normal subjects. AminoAcids 15, 263–269

77 Nagata, Y. et al. (1995) Free D-serine concentration in normal andAlzheimer human brain. Brain Res. Bull. 38, 181–183

78 Hashimoto, K. et al. (2004) Possible role of D-serine in thepathophysiology of Alzheimer’s disease. Prog. Neuropsychopharmacol.Biol. Psychiatry 28, 385–388

79 Bonny, C. et al. (2005) Targeting the JNK pathway as a therapeuticprotective strategy fornervous systemdiseases.Rev.Neurosci. 16, 57–67

80 Gao, Y. et al. (2005) Neuroprotection against focal ischemic brain injuryby inhibition of c-Jun N-terminal kinase and attenuation of themitochondrial apoptosis-signaling pathway. J. Cereb. Blood FlowMetab. 25, 694–712

81 Yoshikawa, M. et al. (2005) Acute treatment with morphine augmentsthe expression of serine racemase and d-amino acid oxidase mRNAs inrat brain. Eur. J. Pharmacol. 525, 94–97

82 Doya, H. et al. (2005) c-Jun N-terminal kinase activation in dorsal rootganglion contributes to pain hypersensitivity. Biochem. Biophys. Res.Commun. 335, 132–138

83 Katsuki, H. et al. (2004) Endogenous D-serine is involved in induction ofneuronal death byN-methyl-D-aspartate and simulated ischemia in ratcerebrocortical slices. J. Pharmacol. Exp. Ther. 311, 836–844

84 Shleper, M. et al. (2005) D-serine is the dominant endogenous coagonistfor NMDA receptor neurotoxicity in organotypic hippocampal slices. J.Neurosci. 25, 9413–9417

85 Chumakov, I. et al. (2002) Genetic and physiological data implicatingthe new human gene G72 and the gene for D-amino acid oxidase inschizophrenia. Proc. Natl. Acad. Sci. U. S. A. 99, 13675–13680

86 Owen, M.J. et al. (2005) Schizophrenia: genes at last? Trends Genet. 21,518–525

87 Martina, M. et al. (2005) Reduced glycine transporter type 1 expressionleads to major changes in glutamatergic neurotransmission of CA1hippocampal neurones in mice. J. Physiol. 563, 777–793

88 Prybylowski, K. and Wenthold, R.J. (2004) N-Methyl-D-aspartatereceptors: subunit assembly and trafficking to the synapse. J. Biol.Chem. 279, 9673–9676

89 Cull-Candy, S. et al. (2001) NMDA receptor subunits: diversity,development and disease. Curr. Opin. Neurobiol. 11, 327–335

90 Foucaud, B. et al. (2003) Structural model of theN-methyl-D-aspartatereceptor glycine site probed by site-directed chemical coupling. J. Biol.Chem. 278, 24011–24017


91 Furukawa, H. and Gouaux, E. (2003) Mechanisms of activation,inhibition and specificity: crystal structures of the NMDA receptorNR1 ligand-binding core. EMBO J. 22, 2873–2885

92 Inanobe, A. et al. (2005)Mechanism of partial agonist action at theNR1subunit of NMDA receptors. Neuron 47, 71–84

93 Das, S. et al. (1998) Increased NMDA current and spine density inmice lacking the NMDA receptor subunit NR3A. Nature 393, 377–381

Articles of interest in o

Neurotransmitter transporters: moleculaUlrik Gether, Peter H. Andersen, Orla

Trends in Pharmacological Science

Theory-based Bayesian models of iJoshua B. Tenenbaum, Thomas L

Trends in Cognitive Sciences D

Bayesian theories of conditioAaron C. Courville, Nathaniel D.

Trends in Cognitive Sciences DO

Probabilistic inference in huMark Steyvers, Thomas L. Gr


Vision as Bayesian inferenceAlan Yuille and D


Probabilistic models of languageNick Chater and Christ


Breakthroughs in the search foLauren M. McGrath, Shelley D. Sm

Trends in Molecular Medicine DOI

The neurobiology of hypocretins (orexins), narcJamie M. Zeitzer, Seiji Nishin


Targeting information-processing deficit in schizopdrug discMihaly H


The nicotinic acid receptor GPR109A (HM74AStefan Offe


AntidepressanJuan A. Mico, Denis Ardid, Esther



94 Johnson, J.W. and Ascher, P. (1987) Glycine potentiates the NMDAresponse in cultured mouse brain neurons. Nature 325, 529–531

95 Kleckner, N.W. and Dingledine, R. (1988) Requirement for glycine inactivation of NMDA-receptors expressed in Xenopus oocytes. Science241, 835–837

96 Rossi, D.J. and Slater, N.T. (1993) The developmental onset ofNMDA receptor-channel activity during neuronal migration.Neuropharmacology 32, 1239–1248

ther Trends journals

r function of important drug targetsM. Larsson and Arne Schousboe

s DOI: 10.1016/j.tips.2006.05.003

nductive learning and reasoning. Griffiths and Charles Kemp

OI:10.1016/j.tics.2006.05.009

ning in a changing worldDaw and David S. Touretzky

I: 10.1016/j.tics.2006.05.004

man semantic memoryiffiths and Simon Dennis

I: 10.1016/j.tics.2006.05.005

: analysis by synthesis?aniel Kersten

I: 10.1016/j.tics.2006.05.002

processing and acquisitionopher D. Manning

I: 10.1016/j.tics.2006.05.006

r dyslexia candidate genesith and Bruce F. Pennington

: 10.1016/j.molmed.2006.05.007

olepsy and related therapeutic interventionso and Emmanuel Mignot

s DOI:10.1016/j.tips.2006.05.006

hrenia: a novel approach to psychotherapeuticoveryajos

s DOI: 10.1016/j.tips.2006.05.005

or PUMA-G) as a new therapeutic targetrmanns

s DOI: 10.1016/j.tips.2006.05.008

ts and painBerrocoso and Alain Eschalier

s DOI: 10.1016/j.tips.2006.05.004