MINIREVIEW
D-Amino acids in the brain: the biochemistry of brain
serine racemase
Florian Baumgart and Ignacio Rodrı
´
guez-Crespo
Departamento de Bioquı
´
mica y Biologı
´
a Molecular, Facultad de Ciencias Quı
´
micas, Universidad Complutense de Madrid, Spain
The initial purification of brain mammalian serine
racemase was performed by Wolosker, Snyder and
coworkers using 60 brains obtained from rats [1]. This
seminal work permitted the isolation of a homo-
geneous protein preparation that displayed the ability
to isomerize l-serine into its enantiomeric d-serine
counterpart. In addition, the authors established the
molecular mass of the enzyme, its pH and temperature
dependence, the presence of bound pyridoxal-5¢ phos-
phate (PLP) and the exquisite activity regulation
exerted by reagents that react with free SH groups,
such as oxidized glutathione. The subsequent mole-
cular cloning of mouse brain serine racemase, as well
as the comparison with PLP-containing racemases
from other organisms, led to the identification of K56
as the lysine residue that formed the Schiff base with
the PLP moiety [2]. The first recombinant expression
and purification experiment was performed by Wolos-

N-methyl d-aspartate receptors. Mammalian brain serine racemase is a
pyridoxal-5¢ phosphate-containing enzyme that catalyzes the racemization
of l-serine to d-serine. It has also been shown to catalyze the a,b-elimina-
tion of water from l-serine or d-serine to form pyruvate and ammonia.
Serine racemase is included within the group of type II-fold pyridoxal-5¢
phosphate enzymes, together with many other racemases and dehydratases.
Serine racemase was first purified from rat brain homogenates and later
recombinantly expressed in mammalian and insect cells as well as in
Escherichia coli. It has been shown that serine racemase is activated by
divalent cations like calcium, magnesium and manganese, as well as by
nucleotides like ATP, ADP or GTP. In turn, serine racemase is also
strongly inhibited by reagents that react with free sulfhydryl groups such
as glutathione. Several yeast two-hybrid screens for interaction partners
identified the proteins glutamate receptor interacting protein, protein inter-
acting with C kinase 1 and Golga3 to bind to serine racemase, having
different effects on its catalytic activity or stability. In addition, it has also
been proposed that serine racemase is regulated by phosphorylation. Thus,
d-serine production in the brain is tightly regulated by various factors
pointing at its physiologic importance. In this minireview, we will focus on
the regulation of brain serine racemase and d-serine synthesis by the
factors mentioned above.
Abbreviations
[Ca
2+
]
cyt
, cytosolic calcium concentration; AMPA, a-amino-3-hydroxy-5-methylisoxazole-4-propionic acid; AMPAR, AMPA receptor; GluR2,
glutamate receptor subunit 2; Golga3, Golgin subfamily A member 3; GRIP, glutamate receptor interacting protein; GSNO, S-nitroso-
glutathione; NO, nitric oxide; PDZ, PSD95 ⁄ disc large ⁄ ZO-1; PICK1, protein interacting with C kinase 1; PKC, protein kinase C;
PLP, pyridoxal-5¢ phosphate.

calcium activated the racemase activity of the enzyme
at a half-maximal concentration (EC
50
) of about
26 lm, although using changes in tryptophan fluores-
cence a binding constant for calcium to serine race-
mase was narrowed down to about 6 lm [5]. The
physiological activation of brain serine racemase by
divalent cations is described in detail below.
Very recently, no fewer than six PLP-containing
enzymes having broad sequence homology with human
brain serine racemase have been cloned and recombi-
nantly expressed. Three recombinant plant serine race-
mases have recently been characterized: those of
Arabidopsis thaliana, Hordeum vulgare (barley) and
Oryza sativa (rice) [7,8]. An aspartate racemase that has
a very high homology with brain serine racemase has
recently been cloned and characterized from a bivalve
mollusk [9]. The so-called serine racemase from Saccha-
romyces cerevisiae has been recombinantly expressed in
E. coli and characterized, and its properties seem to
indicate that it is a paralog rather than an ortholog of
mammalian serine racemases [10]. Likewise, the serine
racemase from the hyperthermophylum Pyrobacu-
lum islandicum was both purified and recombinantly
expressed, and the isolated enzymes were characterized
[11]. Finally, the coordinates of the 3D structure of ser-
ine racemase from Schizosaccharomyces pombe, another
enzyme that displays high homology with mammalian
serine racemase, have been recently deposited (Protein

⁄ Mg
2+
racemase
activation observed for these enzymes [7,8,10]. Con-
versely, the absence of these residues in the bivalve
and Pyrobaculum serine racemases is in agreement with
the absence of increased racemization of these enzymes
induced by Ca
2+
or Mg
2+
[9,11].
The enzyme activation by nucleotides is somehow
more puzzling. For instance, the activation of the
homologous protein E. coli Thr dehydratase by AMP
was observed as early as 1949 [13]. However, in this
bacterial enzyme the nucleotide exerted an allosteric
role, promoting protein oligomerization and activating
the enzyme [14]. In the absence of AMP, the K
m
of
E. coli Thr deaminase for Thr was 70 mm and it
decreased to 5 mm in the presence of the nucleotide
[15]. The residues participating in the binding of nucle-
otides (shown in red in Fig. 1) could also be predicted
because the crystal structure of the S. pombe serine
racemase has the nucleotide AMPpcp bound [12].
Interestingly, most of the nucleotide-binding sites are
conserved in all the enzymes. However, although mam-
malian serine racemase is strongly activated by nucleo-

the case of E. coli Thr dehydratase mentioned above,
we were unable to observe changes in the oligomeriza-
tion state of recombinant mouse serine racemase in the
presence and absence of added ATP [12]. In fact, all
the recombinant versions of serine racemase mentioned
above are either homodimers [7–9] or homotrimers [11]
in the absence of the nucleotide. Consequently, it is
unlikely that ATP might be regulating the quaternary
structure of serine racemase. In this regard, as noted by
Wolosker and coworkers [4] ATP is not hydrolyzed
during catalysis, because both ADP and a nonhydro-
lyzable analog of ATP are able to activate the enzyme
to a similar extent. Furthermore, in the cytosol the
ATP concentrations are in the 3–6 mm range, an obser-
vation suggesting that serine racemase is always satu-
rated with enough nucleotide to exert its racemase
activity because 100 lm ATP is more than enough to
result in full activity [4,6].
Perhaps it is even more important to know if serine
racemase is activated by Ca
2+
or Mg
2+
in vivo.In
principle, if the Mg
2+
concentration in the cellular
cytosol is indeed 600 lm [4,16], the brain serine race-
mase would always be ‘on’. However, when type II
astrocytes were loaded with radioactive d-serine, its

with different calcium concentrations, hence regulating
its enzymatic activity. For instance, direct coupling of
serine racemase to the AMPA receptor (AMPAR) via
glutamate receptor interacting protein (GRIP) binding
might be one way to regulate its d-serine synthesizing
activity (see below).
Nitrosylation of serineine racemase
Only scant data are available on possible post-transla-
tional modifications of serine racemase in vivo. The
observation that both oxidized glutathione [1,5] and cys-
tamine [5] could inhibit serine racemase provided some
evidence that reactive cysteine residues should be pres-
ent that are essential for serine racemase function. When
we tested if the nitric oxide (
•
NO) donor DETA
NONOate [(z)-1-[2-(2-aminoethyl)-N-(2-ammonioethyl)
amino]diazen-1-ium,2-diolate] could alter serine race-
mase activity, we obtained a negative result [5]. Quite
recently, reports of
•
NO as an inhibitor of serine race-
mase in a glioblastoma cell line added a new aspect to
d-serine-dependent modulation of the glutamatergic
synapse. The authors propose that NMDAR-mediated
calcium entry into postsynaptic neurons entails cal-
cium ⁄ calmodulin-dependent activation of neuronal
nitric oxide synthase and the consequent liberation of
•
NO. Serine racemase is subsequently nitrosylated and

in the mouse and human serine racemase sequence) can
become nitrosylated, both in the recombinant enzyme
and in transfected cells, using the
•
NO donor, S-nitroso-
glutathione (GSNO). A molecular model of mouse ser-
ine racemase, based on a yeast homolog, reveals that
residue 113 lies in proximity to the putative ATP-bind-
ing region of the enzyme. Nitrosylation would therefore
lead to impaired nucleotide binding and inactivation of
the enzyme. It is noteworthy that GSNO is known to
modify cysteines not only with
•
NO but also with gluta-
thione, leading to protein glutathionylation, another
post-translational modification occurring under condi-
tions of oxidative ⁄ nitrosative stress. In fact, GSNO is
very frequently used in glutathionylating studies [23]. It
is thus conceivable that purified serine racemase
becomes modified by glutathione together with
•
NO.
Experiments with milder nitrosylating reagents that lack
a glutathione moiety would unambiguously demon-
strate if serine racemase is, in fact, modified by
•
NO.
Serine racemase-interacting proteins:
GRIP, PICK1 and Golga3
The carboxy-terminal end of both mouse and human

investigated. Therefore, other proteins that become
associated with GRIP, using some of the other six
PDZ domains, might modulate the activity of serine
racemase. Conversely, d-serine might also change the
activity of some GRIP-associated proteins. When ser-
ine racemase ⁄ GRIP interactions were first studied, it
was proposed that GRIP was released from AMPARs
when they became stimulated and phosphorylated [18],
which would lead to GRIP interacting with serine
racemase in the cytosol where it would bind to and
activate serine racemase. With our own results we were
able to confirm the interaction of GRIP with serine
racemase via PDZ6 [12]. However, we observed that
binding to PDZ6 alone was not sufficient for activa-
tion. Rather, the presence of the rest of the C-terminal
region of GRIP, that is the PDZ7 module and a link-
ing segment between PDZ6 and PDZ7, was required
for full activation of serine racemase, both in vitro and
in vivo. Although these results do not necessarily pre-
clude a translocation process to AMPARs mediated by
GRIP, they do show the direct activation of serine
racemase by GRIP as a result of the concerted inter-
action of several PDZ modules, independent of the
subcellular localization. This PDZ crosstalk, where an
isolated PDZ domain is insufficient to carry out a
specific function, has also been observed in other
examples, for instance in the requirement of both
PDZ4 and PDZ5 for GRIP binding to GluR2 [27,28].
Interestingly, the activating effect of GRIP on serine
racemase results mainly in a change in V

GRIP-interacting protein that also binds to PDZ6
might disrupt the serine racemase–GRIP interaction,
hence diminishing the activity of the former.
In a similar yeast two-hybrid screen using a human
hippocampal cDNA library, a different PDZ domain-
containing protein was found to interact with serine
racemase, also requiring the C-terminal binding motif
[30]. Protein interacting with C kinase 1 (PICK1) con-
tains one PDZ domain that is required for interacting
with protein kinase C (PKC) [30,31] or serine racemase.
It also contains a Bin ⁄ amphiphysin ⁄ Rys domain,
important for the interaction with lipids, and a coiled-
coil domain. Furthermore, it has been shown recently
that the PDZ domain of PICK1 is also capable of inter-
acting with lipid membranes, a property crucial for the
clustering of AMPAR and synaptic plasticity [32].
There are no data available regarding the effect of the
binding of PICK1 on serine racemase activity. There-
fore, biochemical characterization of the role of the
interaction of serine racemase and PICK1 is needed to
judge the importance of these observations. Surely the
interaction of PICK1 with PKC leads to the temptation
to speculate on a possible phosphorylation of serine
racemase by PKC [33]. As yet, however, there are no
data available, either on the details of the interaction of
PICK1 with serine racemase, or on the phosphorylation
of serine racemase.
Considering that the phosphorylation of Ser880 of
the GluR2 subunit of the AMPA receptor, positioned
at the carboxy-terminal end of the polypeptide chain,

C
C
C
C
C
L-Ser
D-Ser
C
C
P
C
PP
P
AB
C
D
L-Ser
D-Ser
L-Ser
D-Ser
Fig. 3. Proposed modes of interaction
among serine racemase, GRIP and the
AMPAR. (A) A trimeric complex is assumed.
(B) Phosphorylation of the GluR2 subunit of
the AMPAR at Ser880 dissociates GRIP
binding, which remains bound to serine
racemase. (C) Serine racemase is active in
the cytoplasm and does not interact with
GRIP, whereas the latter associates with
the GluR2 subunit. (D) The simultaneous

tion of serine racemase from GRIP. Nevertheless, in
the absence of the association with the AMPAR,
recombinant purified GRIP alone is able to increase
the activity of recombinant purified serine racemase
[12]. We have shown that certain GRIP amino acids,
present further down in the sequence than PDZ6, are
responsible for the majority of the observed activation
of serine racemase by GRIP [12].
In another study to identify binding partners of ser-
ine racemase using the yeast two-hybrid technology,
the Golgi-localized protein, Golgin subfamily A mem-
ber 3 (Golga3), was found to interact with serine race-
mase [34]. In this case, however, no PDZ interactions
with the C-terminal amino acid triplet of serine were
crucial for binding, but instead, the interaction was
established with its N-terminal 66 residues. Binding of
Golga3 increases d-serine synthesis. Intriguingly, this is
achieved through a decrease in ubiquitin ⁄ proteasomal
degradation of serine racemase, rather than by modula-
tion of the catalytic rate. Serine racemase was shown to
have an average half-life of about 4.5 h. When Golga3
and serine racemase were cotransfected, both serine
racemase stability and d-serine synthesis increased con-
siderably. Thus, it is important to note that in addition
to the modulators mentioned beforehand which directly
influence the catalysis of serine racemase, indirect
effects such as protein stability or subcellular localiza-
tion should be taken into account when investigating
the precise regulation of serine racemase-dependent
d-serine levels at glutamatergic synapses.

6 Neidle A & Dunlop DS (2002) Allosteric regulation of
mouse brain serine racemase. Neurochem Res 27, 1719–
1724.
7 Fujitani Y, Nakajima N, Ishihara K, Oikawa T, Ito K
& Sugimoto M (2006) Molecular and biochemical char-
acterization of a serine racemase from Arabidopsis thali-
ana. Phytochemistry 67, 668–674.
8 Fujitani Y, Horiuchi T, Ito K & Sugimoto M (2007)
Serine racemases from barley, Hordeum vulgare L., and
other plant species represent a distinct eukaryotic
group: gene cloning and recombinant protein character-
ization. Phytochemistry 68, 1530–1536.
9 Abe K, Takahashi S, Muroki Y, Kera Y & Yamada
RH (2006) Cloning and expression of the pyridoxal
5¢-phosphate-dependent aspartate racemase gene from
the bivalve mollusk Scapharca broughtonii and charac-
terization of the recombinant enzyme. J Biochem
(Tokyo) 139, 235–244.
10 Wada M, Nakamori S & Takagi H (2003) Serine race-
mase homologue of Saccharomyces cerevisiae has
l-threo-3-hydroxyaspartate dehydratase activity. FEMS
Microbiol Lett 225, 189–193.
11 Ohnishi M, Saito M, Wakabayashi S, Ishizuka M,
Nishimura K, Nagata Y & Kasai S (2008) Purification
Biochemistry of brain serine racemase F. Baumgart and I. Rodrı
´
guez-Crespo
3544 FEBS Journal 275 (2008) 3538–3545 ª 2008 The Authors Journal compilation ª 2008 FEBS
and characterization of serine racemase from a hyper-
thermophylum Pyrobaculum islandicum. J Bacteriol 190,

Ghosh A & Snyder SH (2005) Serine racemase: activa-
tion by glutamate neurotransmission via glutamate
receptor interacting protein and mediation of neuronal
migration. Proc Natl Acad Sci U S A 102, 2105–2110.
19 Mothet JP, Pollegioni L, Ouanounou G, Martineau M,
Fossier P & Baux G (2005) Glutamate receptor activa-
tion triggers a calcium-dependent and SNARE protein-
dependent release of the gliotransmitter d-serine. Proc
Natl Acad Sci U S A 102, 5606–5611.
20 Shoji K, Mariotto S, Ciampa AR & Suzuki H (2006)
Regulation of serine racemase activity by d-serine and
nitric oxide in human glioblastoma cells. Neurosci Lett
392, 75–78.
21 Shoji K, Mariotto S, Ciampa AR & Suzuki H (2006)
Mutual regulation between serine and nitric oxide
metabolism in human glioblastoma cells. Neurosci Lett
394, 163–167.
22 Mustafa AK, Kumar M, Selvakumar B, Ho GP, Ehm-
sen JT, Barrow RK, Amzel LM & Snyder SH (2007)
Nitric oxide S-nitrosylates serine racemase, mediating
feedback inhibition of d-serine formation. Proc Natl
Acad Sci U S A 104, 2950–2955.
23 Klatt P & Lamas S (2000) Regulation of protein func-
tion by S-glutathiolation in response to oxidative and
nitrosative stress. Eur J Biochem 267, 4928–4944.
24 Hung AY & Sheng M (2002) PDZ domains: structural
modules for protein complex assembly. J Biol Chem
277, 5699–5702.
25 Sakarya O, Armstrong KA, Adamska M, Adamski M,
Wang IF, Tidor B, Degnan BM, Oakley TH & Kosik

33 Mustafa AK, Kim PM & Snyder SH (2004) D-Serine as
a putative glial neurotransmitter. Neuron Glia Biol 1,
275–281.
34 Dumin E, Bendikov I, Foltyn VN, Misumi Y, Ikehara
Y, Kartvelishvily E & Wolosker H (2006) Modulation
of d-serine levels via ubiquitin-dependent proteasomal
degradation of serine racemase. J Biol Chem 281,
20291–20302.
F. Baumgart and I. Rodrı
´
guez-Crespo Biochemistry of brain serine racemase
FEBS Journal 275 (2008) 3538–3545 ª 2008 The Authors Journal compilation ª 2008 FEBS 3545