Content uploaded by Steve Standley
Author content
All content in this area was uploaded by Steve Standley on Mar 28, 2014
Content may be subject to copyright.
794 nature neuroscience • volume 4 no 8 • august 2001
Glutamate receptors mediate most rapid excitatory neurotrans-
mission in the CNS and are important in neuronal development,
synaptic plasticity, and learning and memory. Glutamate receptors
are highly enriched at postsynaptic sites, and it is thought that the
appropriate clustering of receptors at the postsynaptic membrane is
critical for efficient synaptic transmission. Whereas the specific type
and number of glutamate receptors at glutamatergic synapses must
be precisely regulated, the mechanisms underlying glutamate recep-
tor targeting and stabilization are not well understood. The
ionotropic glutamate receptors are multimeric complexes made up
of homologous subunits, and are subdivided into NMDA recep-
tors (NR1, NR2A–D), AMPA receptors (GluR1–4) and kainate
receptors (GluR5–7, KA1 and 2)
1,2
. Most glutamatergic synapses
express several subtypes of glutamate receptors, with the exact com-
bination of subunits and subtypes expressed defining the physio-
logical response to glutamate released from the presynaptic cell.
The expression of NMDA receptors precedes that of AMPA
receptors at most excitatory glutamatergic synapses
3
. The cur-
rently accepted model for the maturation of central glutamater-
gic synapses postulates that NMDA receptors are initially
expressed in the absence of AMPA receptors, forming ‘silent
synapses’ that are not activated unless the cell is depolarized, and
that synaptic activity triggers AMPA receptor insertion into the
postsynaptic membrane, resulting in mature excitatory synapses
4
.
Even at mature glutamatergic synapses, it is thought that synap-
tic AMPA receptor expression is dynamic, with receptors rapid-
ly cycling in and out of the postsynaptic membrane
5
. The
membrane fusion protein NSF (N-ethyl-maleimide sensitive fac-
tor) directly interacts with the AMPA receptor subunit GluR2
and stabilizes synaptic currents, consistent with the regulation
of insertion of AMPA receptors into synaptic sites
6–8
. In addi-
tion, disruption of the NSF-AMPA receptor interaction has been
implicated in long-term depression
5,9
.
Endocytosis also regulates AMPA receptor density at postsy-
naptic sites; disruption of endocytosis modulates synaptic plas-
Molecular determinants of NMDA
receptor internalization
Katherine W. Roche
1
, Steve Standley
1
, Jennifer McCallum
1
, C. Dune Ly
1
, Michael D. Ehlers
2
and
Robert J. Wenthold
1
1
Laboratory of Neurochemistry, National Institute on Deafness and Other Communication Disorders, National Institutes of Health,
Building 36, Room 5B25, Bethesda, Maryland 20892, USA
2
Department of Neurobiology, Duke University Medical Center, Box 3209, Durham, North Carolina 27710, USA
Correspondence should be addressed to K.W.R. (rochek@nidcd.nih.gov)
Although synaptic AMPA receptors have been shown to rapidly internalize, synaptic NMDA receptors
are reported to be static. It is not certain whether NMDA receptor stability at synaptic sites is an inher-
ent property of the receptor, or is due to stabilization by scaffolding proteins. In this study, we demon-
strate that NMDA receptors are internalized in both heterologous cells and neurons, and we define an
internalization motif, YEKL, on the distal C-terminus of NR2B. In addition, we show that the synaptic
protein PSD-95 inhibits NR2B-mediated internalization, and that deletion of the PDZ-binding domain
of NR2B increases internalization in neurons. This suggests an involvement for PSD-95 in NMDA recep-
tor regulation and an explanation for NMDA receptor stability at synaptic sites.
ticity in a variety of protocols
5,10,11
. Clathrin-dependent endo-
cytosis remains the most prevalent and best understood type of
regulated endocytosis
12
. Synaptic vesicle recycling accounts for
the extraordinary number of clathrin-coated vesicles (CCVs) in
brain, and it is becoming apparent that clathrin-mediated endo-
cytosis is also important at postsynaptic sites. Both
clathrin-coated pits and non-clathrin coated pits are frequently
seen in dendritic spines
13
. In addition, there is strong evidence
that G-protein-coupled receptors undergo endocytosis by both
clathrin-dependent and clathrin-independent mechanisms
14
.
Studies indicate that postsynaptic AMPA receptors internalize
using a clathrin-dependent mechanism
5,10,15
.
In contrast to AMPA receptors, NMDA receptors have been
reported to be stably expressed in the postsynaptic mem-
brane
5,16,17
. Although experimental evidence supports the notion
that NMDA receptors do not rapidly cycle in and out of the post-
synaptic membrane under normal conditions, it is likely that the
surface expression of NMDA receptors is also tightly regulated.
For example, NMDA receptors cluster at synaptic sites, but also
cluster at nonsynaptic sites on the soma and proximal dendrites
of cultured neurons
18,19
and form functional extrasynaptic recep-
tors
20,21
. This distribution can be modulated by synaptic activi-
ty, as chronic treatment with NMDA receptor antagonists leads to
an increase in surface clusters of NMDA receptors and a shift to
a more synaptic localization
18
.
The interest in defining mechanisms that regulate synaptic
expression of receptors has resulted in the identification of a
rapidly expanding number of glutamate receptor binding pro-
teins thought be critical in receptor localization and signaling.
Recent proteomic analysis has estimated the NMDA receptor
complex to include as many as 77 different proteins
22
. The NR1
subunit directly interacts with calmodulin, α-actinin, yotiao,
intermediate filaments and protein kinases
23
. The NR2 subunits
interact directly with PSD-95 (ref. 24), a modular protein
highly enriched in the postsynaptic density (PSD)
25,26
. PSD-95
articles
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
articles
nature neuroscience • volume 4 no 8 • august 2001 795
construct containing a FLAG epitope within the extracellular
N-terminus (NR2B
FLAG
32
), allowing specific labeling of surface-
expressed receptor. Therefore, we expressed NR2B
FLAG
/NR1-1b
receptor complexes in HeLa cells, labeled surface-expressed recep-
tors with FLAG antibodies on ice and either fixed the cells imme-
diately (Fig. 1, left) or after an incubation in conditioned media
for 15 minutes at 37°C (Fig. 1, right). We observed the internal-
ization of NR2B
FLAG
/NR1-1b complexes after incubation at 37°C
for 15 min (Fig. 1, right), 30 min or 60 min (data not shown).
These data demonstrated that NMDA receptors are efficiently
internalized in heterologous cells.
Motifs necessary for NMDA receptor internalization
Of all of the NR1 and NR2 cytosolic domains, NR2B contains
the strongest consensus internalization motifs
33
. Both tyrosine
(YXX∅) and dileucine (LL) motifs signal clathrin-mediated
endocytosis in a variety of proteins. The NR2B receptor subunit
contains both a tyrosine-based (YEKL) and dileucine (LL) con-
sensus sequence within its distal C-terminus (Fig. 2), making it an
excellent candidate subunit for regulating NMDA receptor inter-
nalization. The tyrosine-based motif is located only several amino
acids upstream from the PDZ-binding motif located at the NR2B
extreme C-terminus.
To characterize NR2B internalization further and to define the
sequence determinants necessary for internalization, we made a
chimera of the distal C-terminus of NR2B (amino acids
1315–1482) and the surface integral membrane protein Tac
(TacNR2B; Fig. 2). We could directly compare targeting infor-
mation encoded by different cytosolic domains by appending their
sequences to Tac, a protein that is normally stable on the plasma
membrane. The use of a high-affinity monoclonal antibody to
the Tac extracellular domain eliminated variability between
chimeras. In addition, the use of the Tac chimeras allowed us to
circumvent the problem of cytotoxicity encountered when express-
ing full-length NMDA receptor complexes, making it possible to
use more robust assays of internalization. We compared the inter-
nalization of Tac and TacNR2B. These constructs were
expressed in HeLa cells and labeled with Tac antibodies
on ice. The cells were returned to conditioned media at
37°C to allow internalization. Whereas wild-type Tac was
expressed exclusively on the plasma membrane,
Fig. 2. The distal C-terminus of NR2B (1315–1482). The con-
sensus endocytic motifs are indicated in bold type (LL, YEKL),
and the PDZ-binding motif is indicated in bold italics (ESDV).
The sites of the TacNR2B(7) and TacNR2B(11) truncations
are indicated with arrows. Note the close proximity of the
YEKL endocytic motif to the PDZ-binding domain on the
extreme C-terminus.
is one member of a family of related proteins that also includes
PSD-93/chapsyn-110, SAP97/hdlg and SAP102. The family mem-
bers all possess three PDZ-domains, which are highly conserved
domains consisting of 80–100 amino acids that mediate pro-
tein–protein interactions, particularly between proteins localized
to areas of cell–cell contact
23,27,28
. Several ion channels, including
K
+
channels and NMDA receptors, interact with the PDZ
domains of PSD-95 family proteins through their C-termini. In
Drosophila, the PSD-95 homolog, discs-large, is critical for synap-
tic localization of K
+
channels at the neuromuscular junction, as
loss-of-function mutations disrupt the normal channel cluster-
ing
29
. Although the mammalian homologs of discs-large can
induce both K
+
channel and NMDA receptor clustering in het-
erologous cells, mice lacking PSD-95 (ref. 30) display normal
NMDA receptor distributions, and it is still uncertain what func-
tional role PSD-95 subserves.
Here we investigated the surface expression and internalization
of NMDA receptors. We found that neuronal NMDA receptors
are robustly internalized in primary neuronal cultures early in
development, but a decrease occurs as the cultures mature. We
also found that recombinant NR1/NR2B receptor complexes are
rapidly internalized in heterologous cells. Using a chimera of the
stable surface protein Tac and the distal NR2B tail, we demon-
strated that the NR2B C-terminal sequence is sufficient to induce
internalization. We defined an internalization signal, YEKL, near
the extreme C-terminus of NR2B, and showed that NR2B-medi-
ated internalization is blocked by coexpression with the synap-
tic protein PSD-95. In addition, deletion of the PDZ-binding
domain of NR2B increased NR2B-mediated internalization.
These findings reveal an involvement for PSD-95 in NMDA
receptor regulation and provide an explanation for the stability of
NMDA receptors at synaptic sites.
R
ESULTS
NMDA receptor internalization in heterologous cells
To investigate NMDA receptor internalization, we expressed
recombinant receptors in heterologous cells. Endogenous NMDA
receptors are composed of an NR1 subunit and one or more NR2
subunits, and both NR1 and NR2B subunits must be coexpressed
for these subunits to reach the cell surface
31
. We decided to char-
acterize the turnover of surface NMDA receptors using an NR2B
Fig. 1. Full-length NR2B/NR1 complexes are internalized in HeLa
cells. HeLa cells were transiently cotransfected with NR2B
FLAG
and
NR1-1b. Cells were incubated with FLAG monoclonal antibodies on
ice for one hour and either immediately fixed (left, NR2B surface) or
washed and returned to conditioned media for 15 minutes at 37°C
before fixation (right, NR2B internalized). All cells were permeabilized
and immunoreactivity was visualized using Cy3 conjugated anti-mouse
secondary antibodies.
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
TacNR2B was internalized to punctate vesicular structures (Fig.
3a and b). These data demonstrate that the C-terminus of NR2B
was sufficient to induce the internalization of the stable surface
protein Tac. This effect was specific for the NR2B C-terminus, as
a construct containing the AMPA receptor GluR1 C-terminus
appended to Tac was not internalized (data not shown).
By comparing secondary antibodies applied before perme-
abilization of the cells (Fig. 3c) or after permeabilization of the
cells (Fig. 3d), we demonstrated that the vesicular structures were
indeed intracellular. To define the intracellular compartment
containing internalized TacNR2B, we loaded the cells with rho-
damine-conjugated transferrin during the 15-minute 37°C incu-
bation, allowing us to visualize transferrin-positive endosomes.
We found considerable colocalization of the internalized trans-
ferrin with TacNR2B (Fig. 3e and f; arrows indicate two exam-
ples of colocalization), illustrating that internalized TacNR2B
moves through this common pathway.
To define a motif or domain that regulates the NR2B-induced
internalization, we next used a quantitative assay of internaliza-
Fig. 3. A tyrosine motif within the
NR2B C-terminus regulates inter-
nalization of the TacNR2B chimera
to endosomes. (a–f) The NR2B
C-terminal tail is sufficient to
induce internalization of the sur-
face protein Tac into transferrin-
positive endosomes. HeLa cells
were transiently transfected with
Tac or TacNR2B. Cells were incu-
bated with Tac monoclonal anti-
bodies on ice for one hour,
washed, and returned to condi-
tioned media for one hour (a, b) or
15 min (c–f) at 37°C. The cells
were fixed and permeabilized, and
the immunoreactivity was visual-
ized using fluorescent-conjugated
anti-mouse secondary antibodies
(a, b) to assess the distribution of
surface-labeled protein following
the 37°C incubation. A second
protocol was used to distinguish
surface-expressed protein from
intracellular pools (c, d). A FITC-
conjugated anti-mouse secondary
antibody was applied before per-
meabilization (c) followed by a
rhodamine-conjugated anti-mouse
secondary antibody applied after permeabilization (d) confirming the
intracellular location of the endosomes. In a third protocol, early endo-
somes were loaded with rhodamine-conjugated transferrin (f) during
the 37°C incubation (described in Methods), and considerable colocal-
ization with internalized TacNR2B was observed (e, arrowheads).
(g) Identification of a tyrosine motif, YEKL, that regulates the internal-
ization of TacNR2B. Cells were incubated with Tac monoclonal antibod-
ies on ice for one hour, washed, and returned to conditioned media for
15 min at 37°C. Internalized cells were stripped of surface-bound anti-
bodies by low pit treatment, while the ‘total’ group was not. Both ‘inter-
nalized’ and ‘total’ cells were fixed, personalized and incubated with
HRP-conjugated anti-mouse secondary antibodies. HRP activity was
measured using a calorimetric assay. Experiments were performed in
quadruplicate. Values are expressed as a ratio of internalized/total
(mean ± s.e.m., n = 3 experiments). Truncation of the last 11 amino
acids significantly inhibited internalization for TacNR2B(11) and Tac
NR2B(11)L1395A (p < 0.05; Student’s unpaired t-test).
796 nature neuroscience • volume 4 no 8 • august 2001
articles
tion to analyze the effect different trun-
cations of TacNR2B had on endocyto-
sis. The TacNR2B chimeras were
expressed in HeLa cells, labeled with
Tac antibodies on ice, and the cells were
returned to conditioned media for 15
minutes at 37°C to allow internaliza-
tion (Fig. 3g). We found that an
11-amino acid C-terminal truncation
of TacNR2B (TacNR2B(11)), which removes the YEKL motif,
dramatically inhibited internalization, whereas truncation of the
last 7 amino acids (TacNR2B(7)), which removed the PDZ-bind-
ing domain but left the tyrosine motif intact, had no effect
(Fig. 3g). Furthermore, we mutated the dileucine motif (L1395A)
contained within the TacNR2B(11) C-terminus and found no
additional inhibition of internalization. Comparable results were
observed after a 1-hour incubation at 37°C (data not shown).
Truncation of the tyrosine motif results in 50% inhibition of
internalization, and the residual internalization suggests the pres-
ence of additional internalization signals within the NR2B
C-terminus. These data demonstrate that the YEKL motif regu-
lates internalization of TacNR2B in heterologous cells.
NMDA receptor internalization is clathrin-dependent
The tyrosine-based internalization motif we identified at the
C-terminus of NR2B, YEKL, is a strong consensus sequence for
binding to AP-2 adaptors that link internalized cargo to clathrin
33
.
In addition, the colocalization of internalized TacNR2B with
transferrin-positive endosomes is consistent with trafficking
through a clathrin-dependent pathway, suggesting that NMDA
receptor internalization is also clathrin-mediated. Internaliza-
tion of TacNR2B chimera was blocked by coexpression with a
dominant-negative form of dynamin (K44A), a protein neces-
sary for clathrin-mediated endocytosis (Fig. 4); 96% of cells
expressing this form of dynamin exhibited no internalization
(n = 50). In contrast, internalization of TacNR2B was not inhib-
ited by coexpression with wild-type dynamin (Fig. 4). These data
suggest that NMDA receptor internalization, like AMPA recep-
tor internalization
10,15
, is mediated by clathrin-coated pits.
Glutamate receptor internalization in neurons
Having characterized the internalization of NMDA receptors in
heterologous cells, we next wanted to investigate NMDA recep-
tor internalization in neurons. We first took a biochemical
approach, using the reversible biotinylation of surface proteins in
cultured cortical neurons. At early stages of development (7 days
a
b
c
d
e
f
g
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
articles
nature neuroscience • volume 4 no 8 • august 2001 797
in vitro; d.i.v.), NMDA receptors undergo robust endocytosis;
21.6 ± 1.7% of surface NMDA receptors were endocytosed after
30 minutes (Fig. 5a). NMDA receptor endocytosis progressively
decreased in more mature cultures; 10.3 ±1.5% of surface NMDA
receptors were internalized after 30 min in 12 d.i.v. neurons and
5.0 ±1.2% were internalized at 18 d.i.v. (Fig. 5b). These findings
extend previous studies, which showed only limited endocytosis
(∼5% after 30 min) of NMDA receptors in mature cortical neu-
rons
16
, by demonstrating a developmental decline in NMDAR
endocytosis as neurons mature.
PSD-95 blocks NR2B-mediated internalization
Previous studies reported that NMDA receptors are stably
expressed at synaptic sites. Our biotinylation experiments did
not differentiate between synaptic and extrasynaptic receptors,
and we wanted to investigate the role that synaptic proteins such
as PSD-95 might play in NMDA receptor surface expression. We
had demonstrated that TacNR2B was internalized in HeLa cells
and that internalization was regulated by a tyrosine-based motif
located on the NR2B C-terminus (Fig. 3). PSD-95 binds to a
series of amino acids (ESDV) on the extreme C-terminus of
NR2B just a few amino acids downstream of the critical tyrosine
motif. The close proximity of these consensus sequences raised
the possibility of a physiological interaction between scaffolding
proteins such as PSD-95 and internalization of NMDA receptors.
Therefore, we next investigated whether coexpression with
PSD-95 had an effect on NR2B-mediated internalization. We
coexpressed PSD-95 with TacNR2B in HeLa cells, labeled surface
TacNR2B with Tac antibodies on ice, and returned the cells to
conditioned media for 1 hour at 37°C to allow internalization.
We found that coexpression of PSD-95 with TacNR2B blocked
TacNR2B internalization into endosomes as detected by immuno-
fluorescence (Fig. 6a–h). Instead, PSD-95 expression resulted in
the clustering of TacNR2B into patches on the cell surface (Fig. 6a
and b). In contrast, coexpression of PSD-95 with TacNR2B(7),
which does not contain the PSD-95 binding domain, did not
affect TacNR2B(7) internalization into endosomes (Fig. 6c
and d). As in Fig. 3, we analyzed the distribution of TacNR2B
following internalization by applying secondary antibodies both
before (Fig. 6e) and after (Fig. 6f) permeabilization. When
TacNR2B was coexpressed with PSD-95, it clustered on the sur-
face of the cell and could easily be visualized without permeabi-
lization (Fig. 6e), unlike the intracellular endosomes (Fig. 3c
and d). In addition, we labeled surface TacNR2B with Tac anti-
bodies on ice and immediately fixed the cells without the 37°C
incubation to allow internalization. Clusters were clearly visible
on the cell surface (Fig. 6g and h). Along with the microscopic
examination of endocytosis, we used a quantitative colorimetric
assay to measure the effect of PSD-95 on TacNR2B internaliza-
tion. We found that PSD-95 significantly inhibited internaliza-
tion and stabilized TacNR2B on the cell surface (Fig. 6i). These
results were striking, especially considering that PSD-95 is often
expressed only in a subset of cells expressing TacNR2B as detect-
ed by immunofluorescence (data not shown).
NR2B-mediated internalization in neurons
Finally, we wanted to characterize NR2B-mediated internalization
in neurons using the TacNR2B chimeras. We had constructed
chimeras lacking the extracellular domain of the NR2B subunit to
selectively assay the targeting/internalization information includ-
ed on the NR2B C-terminal domain and also to avoid assembly
with native receptors. Therefore, we transiently transfected pri-
mary hippocampal cultures with TacNR2B, TacNR2B(7) or
TacNR2B(11) and monitored internalization using a surface-label-
ing/stripping protocol (Fig. 7; see Methods). TacNR2B internal-
ization was typically observed in very few transfected neurons,
whereas TacNR2B(7) internalization was seen in most transfect-
ed neurons. In addition, TacNR2B(7) consistently displayed robust
internalization, both in the soma and extending into distal process-
es (Fig. 7e–h). This is consistent with the PDZ-binding domain
regulating TacNR2B internalization. TacNR2B, like TacNR2B(7),
Fig. 5. Endogenous NMDA receptors undergo endocyto-
sis, which decreases during development. (a) Biotinylation
assays of NMDAR endocytosis were done on cultured
cortical neurons at 7, 12 and 18 d.i.v. Internalized (biotiny-
lated) NMDARs were isolated by neutravidin precipitation
and detected by anti-NR1 immunoblot. Internalized recep-
tors (middle) were compared to a constant plate percent-
age of total surface receptor (right) to quantify percent
internalization. Molecular mass markers are in kilodaltons.
(b) Quantification of NMDAR internalization. Data repre-
sent means ± s.e.m. (n = 6; pairwise p-values, 7 versus
12 d.i.v. and 7 versus 18 d.i.v., p < 0.001; 12 versus 18 d.i.v.,
p < 0.05, t-test).
Fig. 4. TacNR2B internalization is dynamin-dependent. HeLa cells were
transiently transfected with TacNR2B and either wild-type dynamin
(DynWT) or a dynamin dominant-negative mutant (DynDN). Both
dynamin constructs were HA-tagged. Cells were incubated with Tac
monoclonal antibodies on ice for one hour, washed and returned to
conditioned media for one hour at 37°C. The cells were fixed, perme-
abilized, and incubated with HA rabbit polyclonal antibodies.
Immunoreactivity was visualized using a combination of rhodamine con-
jugated anti-mouse secondary antibodies and FITC conjugated anti-rab-
bit secondary antibodies.
a
b
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
798 nature neuroscience • volume 4 no 8 • august 2001
articles
Fig. 6. PSD-95 inhibits internalization of TacNR2B and clusters it on the
cell surface. (a–h) HeLa cells were transiently transfected with PSD-95
and TacNR2B or TacNR2B(7). Cells were incubated with Tac monoclonal
antibodies on ice for one hour, washed and returned to conditioned
media for one hour (a–d) or 15 min (e–h) at 37°C. The cells were fixed,
permeabilized and incubated with PSD-95 rabbit polyclonal antibodies.
Immunoreactivity was visualized using a combination of rhodamine-conju-
gated anti-mouse secondary antibodies and FITC-conjugated anti-rabbit
secondary antibodies. A second protocol was used to distinguish surface
expressed TacNR2B from intracellular pools (e, f). A FITC-conjugated
anti-mouse secondary antibody was applied before permeabilization
(e) followed by a rhodamine-conjugated anti-mouse secondary antibody
applied after permeabilization (f) confirming the cell surface location of
the TacNR2B clusters. In a third protocol, cells were fixed immediately
after the one-hour incubation with Tac antibody on ice. A FITC-conju-
gated anti-mouse secondary antibody was applied before permeabiliza-
tion to reveal surface-localized TacNR2B (g). The cells were incubated
with PSD-95 antibodies following permeabilization and PSD-95 was visu-
alized with a rhodamine-conjugated anti-rabbit secondary antibody (h).
(i) HeLa cells were transiently transfected with TacNR2B or TacNR2B(7)
and either empty vector or PSD-95. Cells were incubated with Tac mon-
oclonal antibodies on ice for one hour, washed, and returned to condi-
tioned media for 15 min at 37°C. The amount of internalized protein was
quantitated as described in Methods. Experiments were performed in
quadruplicate. Values are expressed as a ratio of internalized/total (mean
± s.e.m., n = 3 experiments). Co-expression with PSD-95 significantly
inhibited the internalization of TacNR2B (p < 0.05; Student’s paired t-test).
always internalized robustly in the transfected glia (Fig. 7c). This is
again consistent with specific inhibition of NR2B-mediated inter-
nalization by PSD-95 or some other neuron-specific PDZ-domain-
binding protein. We therefore used the robust internalization of
TacNR2B in glia as an internal control to confirm efficient surface-
labeling of the TacNR2B construct in each experiment. In addi-
tion, unstripped controls were used in which we replaced the low
pH washes with PBS to confirm that TacNR2B was efficiently sur-
face-labeled in neurons (data not shown).
To quantitate the differences in internalization that we
observed in transfected neurons, we analyzed the captured images
of TacNR2B, TacNR2B(7) and TacNR2B(11). We compared the
ratio of internalized signal to background signal for the captured
images (Fig. 7, see Methods) and found that TacNR2B internal-
ization was close to background (0.9/1), whereas TacNR2B(7)
internalization was more intense (5.8/1) and TacNR2B(11) was
intermediate (3.9/1). In addition, we calculated the percentage
of transfected cells that displayed any internalization above
threshold (Fig. 7i). Whereas 67% of the TacNR2B(7)-expressing
neurons displayed internalization, this was true for only 15% of
the TacNR2B-expressing neurons. TacNR2B(11)-expressing neu-
rons displayed an intermediate phenotype, with internalization in
about half of the cells (55%). The intermediate phenotype that
we routinely observe for TacNR2B(11) internalization is due to
the incomplete inhibition of internalization following deletion
of the tyrosine motif. We found that truncation of the tyrosine
motif results in a 50% inhibition of TacNR2B internalization in
HeLa cells (Fig. 3g), indicating that there is at least one additional
internalization motif encoded in the NR2B C-terminus. Finally,
a separate evaluation of an independent experiment using an
observer rating system confirmed the difference in the amount
of internalization in cells expressing TacNR2B(7) (78%, n = 36)
compared to TacNR2B (43%, n = 44). These data confirmed our
qualitative observations presented in Fig. 7a–h illustrating the
importance of the NR2B PDZ-binding domain for inhibiting
TacNR2B internalization in neurons.
D
ISCUSSION
The number and position of glutamate receptors on the postsy-
naptic membrane dictates the responsiveness of a postsynaptic
cell to glutamate released from a presynaptic cell; thus, the regu-
lation of receptor surface expression is particularly important.
Although AMPA receptors have been demonstrated to constitu-
tively cycle in and out of the postsynaptic membrane
10,11,16,34
,
electrophysiological analyses of synaptic responses suggest that
NMDA receptors do not
5
. Even so, the clustering of NMDA recep-
tors at synaptic sites can be altered in an activity-dependent man-
ner
18
, demonstrating that the expression of surface NMDA
a
b
c
d
e
f
g
h
i
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
articles
nature neuroscience • volume 4 no 8 • august 2001 799
receptors is also tightly regulated. We observed and characterized
NMDA receptor internalization in both heterologous cells and in
neurons using a combination of molecular and biochemical tech-
niques. Our data support a model in which NMDA receptors are
capable of constitutive internalization from the cell surface, but
upon binding to PSD-95 are anchored at synaptic sites.
Tyrosine-based internalization motifs bind to AP-2 adaptor
complexes, thereby linking cargo to clathrin and regulating their
inclusion to clathrin coated pits and ultimately to clathrin-coated
vesicles
35
. Tyrosine-based internalization signals have been char-
acterized in a wide variety of integral membrane proteins and are
typically located very near the extreme cytosolic terminal domain
of transmembrane proteins
35
. The structural determinants dic-
tating the binding preferences for adaptor medium chains (µ1
and µ2) have been characterized using a combinatorial peptide
library, and data from these experiments revealed that certain
residues are preferred for binding and internalization
33
. NR2B
contains a strong consensus sequence, YEKL, for binding to µ2,
a subunit of the AP-2 adaptor complex that regulates clathrin-
dependent protein internalization from the plasma membrane
35
,
making it a good candidate for regulating the internalization of
NMDA receptor complexes. In addition to the tyrosine-based
motif, NR2B also contains a dileucine motif on its distal C-ter-
minus. Dileucine motifs also mediate internalization of plasma
membrane proteins via clathrin-coated pits. We found that dele-
tion of the YEKL motif significantly inhibited the NR2B-regulat-
ed internalization; however, mutation of the dileucine motif did
not result in any further decrease in the level of internalization.
It is often proposed that the PSD-95 family of proteins is respon-
sible for clustering NMDA receptors at synaptic sites. This is based
mainly on data showing that discs-large regulates the synaptic clus-
tering of K
+
channels at the neuromuscular junction in Drosophila
29
.
However, the evidence that PSD-95 mediates clustering in mam-
malian cells is less convincing. We now provide evidence that PSD-
95 regulates the internalization of NMDA receptors, suggesting a
different role for this family of proteins. In support of our findings,
a previous study reports that PSD-95 blocks the internalization of
K
+
channels in heterologous cells
36
. This suggests a common func-
tional role for PSD-95 in both NMDA receptor and K
+
channel reg-
ulation. However, unlike NR2B, Kv1.4 does not have a consensus
internalization motif in close proximity to the PDZ-binding
domain; therefore, it is likely that the mechanism of PSD-95 inhi-
bition of internalization of K
+
channels is distinct from the one we
have proposed for NMDA receptors. There are recent findings that
PSD-95 itself contains targeting information
37
. PSD-95 contains a
tyrosine-based motif, YHKV, near its C-terminus that is important
for synaptic localization of PSD-95. Furthermore, the C-terminus
of PSD-95 including this tyrosine-based motif was sufficient to
induce internalization of a Tac-PSD-95 chimera, suggesting that
PSD-95 is capable of interacting with clathrin adaptor proteins,
thus affecting its localization. These findings lead to two different
scenarios of PSD-95 regulation of channel–receptor surface expres-
sion. Under certain conditions, the NMDA receptor–PSD-95 inter-
action may be disrupted, leading to NR2B-regulated internalization
via the YEKL motif on the NR2B C-terminus. Under different cir-
cumstances, the NMDA receptor-PSD-95 or K
+
channel–PSD-95
interaction might remain intact while PSD-95 binding to the
cytoskeleton is disrupted, leading to PSD-95-regulated internaliza-
tion. This second scenario would be particularly relevant to PSD-
95’s binding to and regulation of channels and receptors that do
not contain cis-acting internalization motifs, such as K
+
channels.
Based on our results, one might imagine that NMDA recep-
tors would be free to internalize in a variety of situations. This
would certainly be true at extrasynaptic sites at which PSD-95 is
Fig. 7. TacNR2B internalization in neurons is regulated by the PDZ-
binding domain. Truncation of the PDZ-binding domain results in robust
internalization of TacNR2B(7) in neurons. Hippocampal neurons, 4 d.i.v.,
were transfected with TacNR2B or TacNR2B(7) using the calcium phos-
phate coprecipitation method. At 10 d.i.v., the cells were incubated with
Tac antibodies for 30 min at room temperature, antibody was removed,
and cells were returned to conditioned media for 15 min at 37°C to
allow internalization. Neurons were stripped of surface-bound antibod-
ies by low pH treatment, fixed, permeabilized, and incubated with FITC-
conjugated secondary antibodies to visualize internalized protein (left).
The cells were washed, incubated again with Tac primary antibodies, and
incubated with rhodamine-conjugated secondary antibodies to label
total transfected protein (right). Neurons (a, b, e, f) and glia (c, d, g, h)
expressing each construct are shown. (i) Truncation of the PDZ-binding
domain leads to an increase in TacNR2B internalization. Cells were
transfected with TacNR2B, TacNR2B(7) or TacNR2B(11) and stained as
described above.
a
b
c
d
e
f
g
h
i
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
800 nature neuroscience • volume 4 no 8 • august 2001
articles
not expressed. NMDA receptors cluster at both synaptic and
extrasynaptic sites in hippocampal cultures, with colocalization
with PSD-95 predominantly occurring at synaptic sites
18,19
. Thus,
surface expression of NMDA receptors at regions of the plasma
membrane lacking PSD-95 might be followed by rapid receptor
internalization, providing less stable expression of these receptors
at functionally irrelevant sites. The expression of NR2B decreases
as PSD-95 expression increases during development in vivo
38
,
consistent with the developmental downregulation of endocyto-
sis we observe. This supports a role for NMDA receptor internal-
ization during synapse formation throughout development.
It is also possible that NMDA receptor association with
PSD-95 could be disrupted at synaptic sites in response to
dynamic events at the PSD. One potential mechanism for this
would be the phosphorylation of the receptor, a modification that
could inhibit its interaction with PSD-95. For example, it has been
shown that the phosphorylation of inwardly rectifying K
+
chan-
nels inhibits the binding to PSD-95 (ref. 39), and more recently,
that phosphorylation of NR2B also inhibits binding to PSD-95
(Richard Huganir, personal communication). In addition, the
NMDA receptor-PSD-95 complex is disrupted following transient
cerebral ischemia in rat
40
, supporting a dynamic model of the PSD.
Finally, disruption of the cytoskeleton during spine rearrangement
could also destabilize NMDA receptors at synaptic sites and per-
mit their internalization. Although our studies do not address the
ultimate fate of internalized NMDA receptors, future studies inves-
tigating the endocytic pathways involved will be important, as this
is clearly regulated for AMPA receptors
16,17,41
.
Although our data have provided evidence that NR2B is piv-
otal in regulating the surface expression of native NMDA recep-
tors, it also is likely that other NMDA receptor subunits
participate in this process. The NR1 subunit interacts with many
cytosolic proteins that are important for functional regulation,
such as calmodulin, protein kinases and proteins important for
linkage to the PSD and cytoskeleton such as α-actinin and inter-
mediate filaments
23
. Therefore, the activity of the receptors and
their stability within the PSD is likely to be affected by NR1 inter-
acting proteins as well.
In addition to binding to NMDA receptors, PSD-95 and its
family members bind to a variety of different receptors and chan-
nels, including kainate receptors
42
, the GluR1 AMPA receptor
subunit
43
, delta glutamate receptors
44
and both voltage-gated
45
and inwardly rectifying K
+
channels
39
. Therefore, it is likely that
PSD-95 family members regulate protein localization in a variety
of systems. The tyrosine-based internalization motif is
located just a few amino acids away from the PDZ-binding
domain of NR2B, and is not conserved in kainate receptors, GluR1
AMPA receptors, delta 2 receptors or K
+
channels, indicating a
unique functional interaction of PSD-95 with NMDA receptors.
To summarize, our results suggest a model of NMDA receptor
surface expression regulation in which NMDA receptors are teth-
ered to the cytoskeleton through an interaction with PSD-95 at
the PSD. The disruption of the NMDA–PSD-95 complex desta-
bilizes the NMDA receptor, thereby allowing rapid internaliza-
tion. The tyrosine motif located on the NR2B C-terminus
regulates the rapid internalization of surface NMDA receptors
by binding to AP-2 adaptor complexes, and internalization pro-
ceeds in a clathrin- and dynamin-dependent fashion. This model
accounts for the robust internalization of NMDA receptors
observed in neuronal cultures when monitoring the total pool of
surface expressed NMDA receptors, including both synaptic and
extrasynaptic receptors. Our model also provides an explanation
as to why the TacNR2B chimera is internalized efficiently in glia,
but not in neurons that express endogenous PSD-95. In contrast,
TacNR2B(7), which lacks the PDZ-binding domain, undergoes
pronounced internalization in both neurons and glia.
M
ETHODS
Antibodies and DNA constructs. The primary antibodies FLAG M2
(Sigma, St. Louis, Missouri), Tac 7G7 (ATCC, Manassas, Virginia),
GluR2/3 (ref. 46), PSD-95 T60 antisera
38
and NR1 54.1 (N. Brose, Max-
Planck Institute, Gottingen, Germany) were used for immunofluores-
cence. All secondary antibodies were obtained from Jackson Laboratories
(West Grove, Pennsylvania), with the exception of Alexa 488 and Alexa
568 conjugated anti-mouse antibodies, which were obtained from Mol-
ecular Probes (Eugene, Oregon).
The following full-length cDNA constructs in mammalian expression
vectors were all donated: NR1-1b (M. Hollmann, Max-Planck Institute,
Gottingen, Germany), NR2B
FLAG
(F.A. Stephenson, School of Pharmacy,
London, UK), PSD-95 (D. Bredt, UCSF, San Francisco, California), Tac
(J. Bonifacino, NIH, Bethesda, Maryland) and HA-tagged wild-type and
dominant-negative (K44A) mutant forms of dynamin (S. Schmid, The
Scripps Research Institute, La Jolla, California). The GluR1 C-terminus
(amino acids 827–907), NR2B distal C-terminus (amino acids 1315–1482),
and NR2B(11) (amino acids 1315–1471) were amplified using PCR and
subcloned into the cytosolic portion of Tac using XbaI and EcoRV restriction
sites as previously described
47
. The L1395A mutation in the TacNR2B(11)
chimera and the stop codon resulting in the NR2B(7) construct (amino
acids 1315–1475) were introduced using the QuickChange site-directed
mutagenesis kit (Stratagene, La Jolla, California). The sequences of all PCR
products were confirmed by automated sequence analysis.
Immunofluorescent internalization assays. HeLa cells (HeLa 229 cells;
ATCC), grown on glass coverslips, were transfected with the cDNAs indi-
cated (4 µg DNA per well, 6-well dish) using the calcium phosphate copre-
cipitation method. Transfected cells were analyzed 24 h after transfection.
HeLa cells expressing full-length NMDA receptor subunits were maintained
in 100 µM AP5 to reduce cell death. However, even under these conditions,
the NR2B
FLAG
/NR1-1b-transfected cells appeared less healthy as compared
to untransfected HeLa cells or HeLa cells expressing AMPA receptors.
Transfected cells were washed in PBS, incubated with anti-Tac anti-
bodies for 1 h on ice, washed with PBS and returned to conditioned
media at 37°C for 15 min, 30 min or 1 h, as indicated. The cells were
washed with PBS, fixed in 4% paraformaldehyde in PBS for 15 min,
washed in PBS and permeabilized in 0.25% Triton-X-100 in PBS for
5 min. The coverslips were washed in PBS, incubated with
Cy3-conjugated anti-mouse secondary antibodies for 30 min at room
temperature, and washed and mounted with Vectashield mounting media
(Vector Laboratories, Burlingame, California). In addition, to differentiate
surface versus internalized pools of proteins, secondary antibodies were
applied both before (surface, FITC-conjugated anti-mouse) and after
(total, rhodamine conjugated anti-mouse) permeabilization. When dou-
ble-labeling for PSD-95, the cells were incubated with anti-PSD-95 pri-
mary antibodies for one hour at room temperature following
permeabilization. The cells were washed in PBS, incubated with sec-
ondary antibodies (1:500) for 30 min at room temperature, washed in
PBS, and mounted with Vectashield mounting media (Burlingame, Cal-
ifornia). When comparing surface to total immunoreactivity of TacNR2B
coexpressed with PSD-95, the expression of PSD-95 was confirmed using
a third channel (AMCA-conjugated anti-rabbit antibodies).
To label early endosomes, we first incubated the cells expressing
TacNR2B with Tac antibody on ice and washed the cells twice with
serum-free media supplemented with 10 mg/ml BSA. The cells were then
incubated in serum-free media with BSA containing rhodamine-conju-
gated transferrin for 15 min at 37°C to allow internalization of both the
surface-labeled TacNR2B and the fluorescent transferrin. The cells were
then fixed, permeabilized and visualized as described above.
Quantitative assay of internalization. We used a modification of a col-
orimetric assay to quantitate internalized receptors
10,47
. Briefly, HeLa
cells were transiently transfected with TacNR2B, TacNR2B(7),
TacNR2B(11) or TacNR2B(11)L1395A. The cells were then incubated
with Tac monoclonal antibodies on ice for one hour, washed and returned
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
to complete media for 15 min at 37°C. Cells in the ‘internalized’ group
were stripped of surface-bound antibodies by low pH treatment
15
, and
cells in the ‘total’ group were not. Both ‘internalized’ and ‘total’ cells were
fixed, permeabilized and incubated with HRP-conjugated anti-mouse
secondary antibodies. HRP activity was measured using a colorimetric
assay
47
. Experiments were performed in quadruplicate. Values are
expressed as a ratio of internalized/total.
Biotinylation assay of receptor endocytosis with cleavable biotin. High-
density cortical neuron cultures were incubated with 100 µg/ml of the lyso-
somal protease inhibitor leupeptin for 30 min before biotinylation with
1.5 mg/ml sulfo-NHS-SS-biotin (Pierce, Rockford, Illinois) for 10 min at
4°C
16
. Neurons were then incubated at either 4°C to block membrane traf-
ficking or 37°C for 30 min to allow endocytosis to occur. The remaining
surface biotin was then cleaved by reducing its disulfide linkage with glu-
tathione cleavage buffer (50 mM glutathione in 75 mM NaCl and 10 mM
EDTA containing 1% BSA and 0.075 N NaOH) twice for 15 min at 4°C.
Cell membranes were prepared, and biotinylated proteins were pre-
cipitated with Ultralink-neutravidin (Pierce) essentially as described
48
.
Biotinylated receptors were detected by immunoblot (ECL Plus, Amer-
sham, Piscataway, New Jersey) and quantified on a phosphorimager
(Storm 860, Molecular Dynamics, Sunnyvale, California) using Image-
Quant 5.0 software (Sunnyvale, California). The percent of receptors
internalized was determined by measuring the band intensity after 37°C
incubation, subtracting the nonspecific band intensity obtained after 4°C
incubation (always less than 5%), and comparing to a constant percent-
age of total surface receptor obtained from sister cultures.
Transfected hippocampal cultures. Hippocampal neurons were isolated
as previously described
49
with minor changes. Cells were dissected from
E18 Sprague–Dawley rats (Harlan, Indianapolis, Indiana), dissociated in
0.25% trypsin and incubated with 0.01% DNAse I. Dispersed neurons
were plated onto polyornithine/fibronectin coated glass coverslips
50
at a
density of 200,000 or 350,000 cells per well of a 24-well dish. The neurons
were grown in DMEM/F12 media containing 10% fetal bovine serum and
N2 supplements as previously described
50
, or in serum-free media (Neu-
robasal, Gibco, Gaithersburg, Maryland) with B27 supplement (Gibco).
After 3 days in culture, 3 µM cytosine arabinoside was added to inhibit
proliferation of non-neuronal cells. The cultures were maintained in a 5%
CO
2
incubator until transfected and processed for immunofluorescence.
Hippocampal neurons (3–5 d.i.v.) were transfected with the appro-
priate cDNA using the calcium phosphate coprecipitation method. Three
micrograms of total DNA were used per well of the 24-well dish. Precip-
itate was allowed to form on cells for approximately 45 min and the cells
were washed twice with serum-free media. The cells were returned to
conditioned media and maintained at 37°C in a 5% CO
2
incubator for
an additional 7 days. The use and care of animals used in this study fol-
lowed the guidelines of the NIH Animal Research Advisory Committee.
Internalization and immunocytochemistry of neurons. Transfected neu-
rons were washed in PBS and incubated with Tac antibodies for 30 min
at room temperature to label surface-expressed protein. The antibody was
removed and the coverslips were returned to conditioned media for a 15-
min incubation at 37°C. The cells were washed in PBS and the surface
antibody was stripped off with low pH (ref. 15) using two washes over 8
min. The cells were fixed in 4% paraformaldehyde/4% sucrose in PBS for
15 min. Cells were washed in PBS and permeabilized in 0.25% Triton-X-
100 in PBS for 5 min. Following a 30-min preincubation with 10% normal
goat serum (Vector Laboratories), the cells were incubated with rho-
damine-conjugated anti-mouse secondary antibodies for 30 min at room
temperature to label internalized receptor pools (‘stripped’). The cells
were washed in PBS and again incubated with Tac antibodies (1:5 dilu-
tion of 7G7) for 30 min at room temperature to label total receptor. The
cells were washed three times with PBS and incubated with FITC-conju-
gated anti-mouse secondary antibodies (‘total’). The cells were washed in
PBS and mounted with Prolong mounting media (Molecular Probes) and
visualized on a Zeiss Axiophot microscope (Thornwood, New York).
A quantitative analysis was done to compare the internalization of the
different TacNMDA receptor constructs in neurons. Using standardized
exposure times, we captured images (MetaMorph image analysis pro-
gram, Universal Imaging, Downingtown, Pennsylvania) of internalized
(green) and total (red) immunoreactivity for transfected neurons express-
ing TacNR2B, TacNR2B(7) or TacNR2B(11). We collected at least 15
images for each condition. We also captured at least five images of back-
ground from the same experimental coverslips. (Background images con-
tained cells that were not transfected.) We used the maximum
background signal (green background for ‘internalized’ and red back-
ground for ‘total’) to set the threshold for measuring the signal for both
internalized and total immunoreactivity in transfected cells. We then
determined the ratio of ‘internalization’ signal to ‘internalization’ back-
ground for each of the constructs (Fig. 7i). We confirmed similar expres-
sion levels of the different constructs using the ratio of ‘total’ signal to
‘total’ background between constructs (7.8, 8.7 and 7.8 for TacNR2B,
TacNR2B(7) and TacNR2B(11), respectively). We also calculated the per-
centage of neurons that displayed any detectable internalization above
the threshold (Fig. 7i).
In an independent experiment, we used an observer rating system to
quantitate the percentage of transfected cells with internalization. An
experiment was used that had relatively high transfection efficiencies of
both TacNR2B and TacNR2B(7). At least 30 transfected cells were eval-
uated per construct, and any neuron with detectable internalization was
counted as positive. The results were then expressed as a percentage of
total cells counted.
A
CKNOWLEDGEMENTS
This work was supported by the NIDCD intramural program, and in part by
grants from National Institutes of Health (RO1-NS39402) and the Spinal Cord
Research Foundation (to M.D.E.).
RECEIVED 10 MAY; ACCEPTED 22 JUNE 2001
1. Hollmann, M. & Heinemann, S. Cloned glutamate receptors. Annu. Rev.
Neurosci. 17, 31–108 (1994).
2. Dingledine, R., Borges, K., Bowie, D. & Traynelis, S. F. The glutamate receptor
ion channels. Pharmacol. Rev. 51, 7–61 (1999).
3. Petralia, R. S. et al. Selective acquisition of AMPA receptors over postnatal
development suggests a molecular basis for silent synapses. Nat. Neurosci. 2,
31–36 (1999).
4. Malenka, R. C. & Nicoll, R. A. Silent synapses speak up. Neuron 19, 473–476
(1997).
5. Lüscher, C. et al. Role of AMPA receptor cycling in synaptic transmission and
plasticity. Neuron 24, 649–658 (1999).
6. Nishimune, A. et al. NSF binding to GluR2 regulates synaptic transmission.
Neuron 21, 87–97 (1998).
7. Osten, P. et al. The AMPA receptor GluR2 C terminus can mediate a
reversible, ATP-dependent interaction with NSF and alpha- and beta-SNAPs.
Neuron 21, 99–110 (1998).
8. Song, I. et al. Interaction of the N-ethylmaleimide-sensitive factor with
AMPA receptors. Neuron 21, 393–400 (1998).
9. Lüthi, A. et al. Hippocampal LTD expression involves a pool of AMPARs
regulated by the NSF–GluR2 interaction. Neuron 24, 389–399 (1999).
10. Man, Y. H. et al. Regulation of AMPA receptor-mediated synaptic
transmission by clathrin-dependent receptor internalization. Neuron 25,
649–662 (2000).
11. Wang, Y. T. & Linden, D. J. Expression of cerebellar long-term depression
requires postsynaptic clathrin-mediated endocytosis. Neuron 25, 635–647
(2000).
12. Mellman, I. Endocytosis and molecular sorting. Annu. Rev. Cell Dev. Biol. 12,
575–625 (1996).
13. Spacek, J. & Harris, K. M. Three-dimensional organization of smooth
endoplasmic reticulum in hippocampal CA1 dendrites and dendritic spines
of the immature and mature rat. J. Neurosci. 17, 190–203 (1997).
14. Tsao, P. & Zastrow, M. Downregulation of G protein-coupled receptors. Curr.
Opin. Neurobiol. 10, 365–369 (2000).
15. Carroll, R. C. et al. Dynamin-dependent endocytosis of ionotropic glutamate
receptors. Proc. Natl. Acad. Sci. USA 96, 14112–14117 (1999).
16. Ehlers, M. D. Reinsertion or degradation of AMPA receptors determined by
activity-dependent endocytic sorting. Neuron 28, 511–525 (2000).
17. Lin, J. W. et al. Distinct molecular mechanisms and divergent endocytotic
pathways of AMPA receptor internalization. Nat. Neurosci. 3, 1282–1290
(2000).
18. Rao, A. & Craig, A. M. Activity regulates the synaptic localization of the
NMDA receptor in hippocampal neurons. Neuron 19, 801–812 (1997).
19. Rao, A., Kim, E., Sheng, M. & Craig, A. M. Heterogeneity in the molecular
composition of excitatory postsynaptic sites during development of
hippocampal neurons in culture. J. Neurosci. 18, 1217–1229 (1998).
20. Rumbaugh, G. & Vicini, S. Distinct synaptic and extrasynaptic NMDA
articles
nature neuroscience • volume 4 no 8 • august 2001 801
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
802 nature neuroscience • volume 4 no 8 • august 2001
the Kv1.4 potassium channel is suppressed by clustering interactions with
PSD-95. J. Biol. Chem. 275, 1357–1364 (2000).
37. Craven, S. E. & Bredt, D. S. Synaptic targeting of the postsynaptic density
protein PSD-95 mediated by a tyrosine-based trafficking signal. J. Biol. Chem.
275, 20045–20051 (2000).
38. Sans, N. et al. A developmental change in NMDA receptor-associated
proteins at hippocampal synapses. J. Neurosci. 20, 1260–1271 (2000).
39. Cohen, N. A., Brenman, J. E., Snyder, S. H. & Bredt, D. S. Binding of the
inward rectifier K
+
channel Kir 2.3 to PSD-95 is regulated by protein kinase A
phosphorylation. Neuron 17, 759–767 (1996).
40. Takagi, N., Logan, R., Teves, L., Wallace, M. C. & Gurd, J. W. Altered
interaction between PSD-95 and the NMDA receptor following transient
global ischemia. J. Neurochem. 74, 169–178 (2000).
41. Beattie, E. C. et al. Regulation of AMPA receptor endocytosis by a signaling
mechanism shared with LTD. Nat. Neurosci. 3, 1291–1300 (2000).
42. Garcia, E. P. et al. SAP90 binds and clusters kainate receptors causing
incomplete desensitization. Neuron 21, 727–739 (1998).
43. Leonard, A. S., Davare, M. A., Horne, M. C., Garner, C. C. & Hell, J. W. SAP97
is associated with the alpha-amino-3-hydroxy-5-methylisoxazole-4-
propionic acid receptor GluR1 subunit. J. Biol. Chem. 273, 19518–19524
(1998).
44. Roche, K. W. et al. Postsynaptic density-93 interacts with the delta2 glutamate
receptor subunit at parallel fiber synapses. J. Neurosci. 19, 3926–3934 (1999).
45. Kim, E., Niethammer, M., Rothschild, A., Jan, Y. N. & Sheng, M. Clustering of
Shaker-type K
+
channels by interaction with a family of membrane-
associated guanylate kinases. Nature 378, 85–88 (1995).
46. Wenthold, R. J., Yokotani, N., Doi, K. & Wada, K. Immunochemical
characterization of the non-NMDA glutamate receptor using subunit-
specific antibodies. Evidence for a hetero-oligomeric structure in rat brain.
J. Biol. Chem. 267, 501–507 (1992).
47. Standley, S., Roche, K. W., McCallum, J., Sans, N. & Wenthold, R. J. PDZ-
domain suppression of an ER retention signal in NMDA receptor NR1 splice
variants. Neuron 28, 887–898 (2000).
48. Mammen, A. L., Huganir, R. L. & O’Brien, R. J. Redistribution and
stabilization of cell surface glutamate receptors during synapse formation.
J. Neurosci. 17, 7351–7358 (1997).
49. Roche, K. W. & Huganir, R. L. Synaptic expression of the high-affinity kainate
receptor subunit KA2 in hippocampal cultures. Neuroscience 69, 383–393
(1995).
50. Vicario-Abejón, C. in Current Protocols in Neuroscience (eds. Crawley, J. N.,
Gerfen, C. R., McKay, R., Rogawski, M. A., Sibley, D. R. & Skolnick, P.)
3.2.1–3.2.12 (John Wiley and Sons, New York, 1997).
articles
receptors in developing cerebellar granule neurons. J. Neurosci. 19,
10603–10610 (1999).
21. Tovar, K. R. & Westbrook, G. L. The incorporation of NMDA receptors with a
distinct subunit composition at nascent hippocampal synapses in vitro.
J. Neurosci. 19, 4180–4188 (1999).
22. Husi, H., Ward, M. A., Choudhary, J. S., Blackstock, W. P. & Grant, S. G.
Proteomic analysis of NMDA receptor-adhesion protein signaling complexes.
Nat. Neurosci. 3, 661–669 (2000).
23. Hsueh, Y. P. & Sheng, M. Anchoring of glutamate receptors at the synapse.
Prog. Brain Res. 116, 123–131 (1998).
24. Kornau, H. C., Schenker, L. T., Kennedy, M. B. & Seeburg, P. H. Domain
interaction between NMDA receptor subunits and the postsynaptic density
protein PSD-95. Science 269, 1737–1740 (1995).
25. Cho, K. O., Hunt, C. A. & Kennedy, M. B. The rat brain postsynaptic density
fraction contains a homolog of the Drosophila discs-large tumor suppressor
protein. Neuron 9, 929–942 (1992).
26. Kistner, U. et al. SAP90, a rat presynaptic protein related to the product of the
Drosophila tumor suppressor gene dlg-A. J. Biol. Chem. 268, 4580–4583
(1993).
27. Craven, S. E. & Bredt, D. S. PDZ proteins organize synaptic signaling
pathways. Cell 93, 495–498 (1998).
28. Garner, C. C., Nash, J. & Huganir, R. L. PDZ domains in synapse assembly
and signalling. Trends Cell Biol. 10, 274–280 (2000).
29. Tejedor, F. J. et al. Essential role for dlg in synaptic clustering of Shaker K
+
channels in vivo. J. Neurosci. 17, 152–159 (1997).
30. Migaud, M. et al. Enhanced long-term potentiation and impaired learning in
mice with mutant postsynaptic density-95 protein. Nature 396, 433–439
(1998).
31. McIlhinney, R. A. et al. Assembly intracellular targeting and cell surface
expression of the human N-methyl-D-aspartate receptor subunits NR1a and
NR2A in transfected cells. Neuropharmacology 37, 1355–1367 (1998).
32. Hawkins, L. M., Chazot, P. L. & Stephenson, F. A. Biochemical evidence for
the co-association of three N-methyl-D-aspartate (NMDA) R2 subunits in
recombinant NMDA receptors. J. Biol. Chem. 274, 27211–27218 (1999).
33. Ohno, H. et al. The medium subunits of adaptor complexes recognize distinct
but overlapping sets of tyrosine-based sorting signals. J. Biol. Chem. 273,
25915–25921 (1998).
34. Shi, S. H. et al. Rapid spine delivery and redistribution of AMPA receptors
after synaptic NMDA receptor activation. Science 284, 1811–1816 (1999).
35. Bonifacino, J. S. & Dell’Angelica, E. C. Molecular bases for the recognition of
tyrosine-based sorting signals. J. Cell Biol. 145, 923–926 (1999).
36. Jugloff, D. G., Khanna, R., Schlichter, L. C. & Jones, O. T. Internalization of
© 2001 Nature Publishing Group http://neurosci.nature.com
© 2001 Nature Publishing Group http://neurosci.nature.com
