Content uploaded by Oliver Hardt
Author content
All content in this area was uploaded by Oliver Hardt
Content may be subject to copyright.
63 0 VOLUME 13 | NUMBER 5 | MAY 2010 nature neurOSCIenCe
a r t I C l e S
Persistence of many kinds of long-term memor y and long-term
potentiation (LTP) requires the constitutive activity of the atypi-
cal protein kinase C isoform M ζ (PKMζ)1. It has been shown that
PKMζ is involved in long-term memory maintenance in various
behavioral tasks and brain regions, including allocentric spatial
memory1,2 and object location3 in the hippocampus, conditioned
taste aversion in the insular cortex4 and auditory fear conditioning in
the amygdala2. It is not understood how PKMζ maintains long-term
memory. Recent results studying late LTP in hippocampal slices,
however, indicate that PKMζ might regulate N-ethylmaleimide–
sensitive factor (NSF)/glutamate subunit 2 (GluR2)-dependent
AMPA receptor trafficking, thereby increasing the amount of post-
synaptic AMPA receptors5. As the interaction of NSF with GluR2
seems to stabilize AMPA receptors in the postsynaptic membrane
by preventing their internalization6–8, PKMζ may maintain long-
term memory by persistently inhibiting AMPA receptor removal
from postsynaptic sites.
We decided to directly test the plausibility of this mechanism
by blocking AMPA receptor synaptic removal and simultaneously
inhibiting PKMζ. If PKMζ’s ability to maintain memory is a result
of the downregulation of AMPA receptor removal from postsynaptic
sites, then blocking the removal process should abolish the long-term
memory loss that is usually observed after inhibiting PKMζ. We found
that PKMζ maintains long-term memories by blocking a GluR2-
dependent pathway for removal of postsynaptic AMPA receptors to
persistently promote increased levels of GluR2-containing AMPA
receptors at postsynaptic sites.
RESULTS
PKM inhibition disrupts established auditory fear memory
We first confirmed the ability of the cell-permeable, myristoylated PKMζ
pseudosubstrate inhibitory peptide (ZIP), the most selective PKMζ
blocking compound, to impair the retention of previously established
auditory fear memory2. We bilaterally infused either ZIP (20 nmol) or
the scrambled inactive version of the peptide (Scr-ZIP) into the basal
lateral amygdala (BLA) 1 d after auditory fear conditioning. Memory
retention was tested the next day. As expected, ZIP infusions impaired
conditioned fear memory (ZIP, 6.7 ± 4.4% freezing; Scr-ZIP, 58.2 ± 12.9%
freezing; Mann-Whitney U test, U = 0.00, P = 0.008, n = 5) to the extent
that the behavior of these rats was similar to that of untrained ones.
Administering ZIP to another group of rats 7 d after training had the
same effect on conditioned fear (ZIP, 6.7 ± 4.1% freezing; Scr-ZIP,
48.7 ± 11.0% freezing; Mann-Whitney U test, U = 3.50, P = 0.031, n = 5).
We then retrained the Scr-ZIP and ZIP-infused rats, and both groups
showed similar, normal conditioned-fear responses, indicating that ZIP
infusions, which retrogradely disrupted memory retention, did not com-
promise the subsequent ability of the BLA to anterogradely acquire and
maintain new fear memories (Supplementary Fig. 1).
We next examined whether PKMζ maintains long-term auditory fear
memory by regulating GluR2-dependent AMPA receptor trafficking. We
used a synthetic peptide derived from the GluR2 carboxy tail, GluR23Y,
fused to the cell membrane transduction domain of the HIV 1 Tat pro-
tein for cell permeability9. GluR23Y is the most selective compound
that can block GluR2-dependent endocytosis without decreasing basal
synaptic transmission or LTP in the BLA10. Biochemical analyses of the
1Department of Psychology, McGill University, Montreal, Quebec, Canada. 2Brain Research Center and Department of Medicine, University of British Columbia,
Vancouver, British Columbia, Canada. 3Departments of Physiology and Pharmacology, and of Neurology, The Robert Furchgott Center for Neural and Behavioral
Science, SUNY Downstate Medical Center, Brooklyn, New York, USA. Correspondence should be addressed to P.V.M. (virginia.migues@mcgill.ca).
Received 23 November 2009; accepted 1 March 2010; published online 11 April 2010; doi:10.1038/nn.2531
PKM maintains memories by regulating GluR2-
dependent AMPA receptor trafficking
Paola Virginia Migues1, Oliver Hardt1, Dong Chuan Wu2, Karine Gamache1, Todd Charlton Sacktor3,
Yu Tian Wang2 & Karim Nader1
The maintenance of long-term memory in hippocampus, neocortex and amygdala requires the persistent action of the atypical
protein kinase C isoform, protein kinase M (PKM). We found that inactivating PKM in the amygdala impaired fear memory
in rats and that the extent of the impairment was positively correlated with a decrease in postsynaptic GluR2. Blocking the
GluR2-dependent removal of postsynaptic AMPA receptors abolished the behavioral impairment caused by PKM inhibition
and the associated decrease in postsynaptic GluR2 expression, which correlated with performance. Similarly, blocking this
pathway for removal of GluR2-containing receptors from postsynaptic sites in amygdala slices prevented the reversal of long-
term potentiation caused by inactivating PKM. Similar behavioral results were obtained in the hippocampus for unreinforced
recognition memory of object location. Together, these findings indicate that PKM maintains long-term memory by regulating
the trafficking of GluR2-containing AMPA receptors, the postsynaptic expression of which directly predicts memory retention.
© 2010 Nature America, Inc. All rights reserved.
nature neurOSCIenCe VOLUME 13 | NUMBER 5 | MAY 2010 631
a r t I C l e S
surface expression of AMPA receptor subunits have shown that this
effect of GluR23Y is a result of abolishing the reduction of AMPA recep-
tor levels without affecting their basal surface expression10. We therefore
tested whether blocking GluR2-dependent internalization by infusing
GluR23Y (15 pmol) or its scrambled inactive version (Scr-GluR23Y)
into the BLA 1 h before ZIP infusion prevents the loss of auditory fear
memory that we observed. We tested memory retention at both 1 and 10 d
after the peptide infusions, which were administered 1 d after training
(Fig. 1a,b). The Kruskal-Wallis analysis by ranks indicated a signifi-
cant difference among the groups at both 1 d (H(3, N = 45) = 15.246,
P = 0.002) and 10 d (H(3, N = 43) = 13.380, P = 0.004). Post hoc compari-
sons revealed that the fear responses of rats infused with Scr-GluR23Y and
ZIP (n = 12) were significantly lower than the responses of the control
group infused with both scrambled peptides (n = 14, 1 d, P = 0.003; 10 d,
P = 0.013) and the group infused with both GluR23Y and ZIP (n = 11, 1 d,
P = 0.017; 10 d, P = 0.022). The GluR23Y before ZIP group and the two
scrambled peptide group were not different from each other (1 and 10 d,
P = 1.000). GluR23Y alone had no effect on memory retention, as there
was no difference between the rats that received GluR23Y and Scr-ZIP
(n = 8) and the ones that received two scrambled peptides (1 and 10 d,
P = 1.000). The fear responses observed after 1 d were not different from
those observed after 10 d (Wilcoxon matched paired test, T = 282.000,
P = 0.423). Fear memories, which may last the lifetime of an animal, can
persist without any decrease in the fear response11. Taken together, these
results indicate that blocking GluR2-dependent AMPA receptor synap-
tic removal with GluR23Y prevents the memory impairment induced by
ZIP. Moreover, the effect was not transient, but persisted for at least 10 d.
The data suggest that the loss of memory caused by PKMζ inactivation
may be a result of the loss of postsynaptic AMPA receptors through a
GluR2-dependent pathway that is persistently inhibited by the activity
of PKMζ after training.
Although rather unlikely, one alternative interpretation of these data
is that a drug-drug interaction between GluR23Y and ZIP decreased
the efficacy of ZIP at inhibiting PKMζ. Ruling out this alternative, we
found that the half-maximal inhibitory concentration of ZIP on PKMζ
activity was not changed by GluR23Y (Supplementary Fig. 2).
PKM inactivation decreases synaptic GluR2 in trained rats
To confirm that the effect of PKMζ inactivation by ZIP is indeed
mediated by the removal of GluR2-dependent AMPA receptors
from postsynaptic sites after training and that GluR23Y prevents this
removal, we replicated the experiment and then isolated subcellular
fractions from the BLA 1 d after the memory test. We examined
the levels of GluR2 (Fig. 2a) and GluR1 (Fig. 2b) in postsynaptic
densities and in fractions containing extrasynaptic membranes by
western blot after training and after the behavioral effect of training
was disrupted by ZIP. Discriminating between surface and intra-
cellular proteins by biotin-labeling and neutravidin pull-down assay
revealed that the postsynaptic density fractions contained only surface
AMPA receptors, whereas the extrasynaptic fractions contained both
0
10
20
30
40
50
60
0
10
20
30
40
50
60
Freezing response (%)
a b
Freezing response (%)
Scr-GluR2
3Y
+ Scr-ZIP Scr-GluR2
3Y
+ ZIP
GluR2
3Y
+ ZIPGluR2
3Y
+ Scr-ZIP
*
*
Figure 1 Blocking GluR2-dependent AMPA receptor synaptic
removal prevents memory impairment induced by PKMζ inactivation.
(a,b) GluR23Y or Scr-GluR23Y was infused into the amygdala 1 d after
the training session. ZIP or Scr-ZIP was infused 1 h later. Memory was
tested 1 d (a) or 10 d (b) after the infusions. Data represent the mean
percentage of the freezing time during the tone. Error bars represent
s.e.m. ZIP infusion abolished the freezing response (*P < 0.001), but
infusions of both GluR23Y and ZIP led to a performance that was similar
to that exhibited by the inactive, scrambled peptide–infused controls, as
determined by the Kruskal-Wallis analysis of ranks test.
0
20
40
60
80
100
0
20
40
60
80
100
20 40 60 80 100 120 140 160 20 40 60 80 100 120 140 160
20 40 60 80 100 120 140 160 20 40 60 80 100 120 140 160
0
20
40
60
80
100
120
140
a
GluR2
GluR1
Postsynaptic
Relative percentage
Extrasynaptic
b
0
20
40
60
80
100
120
140
Postsynaptic
Relative percentage
Extrasynaptic
*
1 2 3 4 5 1 2 3 4
5
1 2 3 4 5 1 2 3 4
5
Postsynaptic 115
kDa
12345
12345
Extrasynaptic
115
kDa
Postsynaptic
Extrasynaptic
2: Scr-GluR2
3Y
+ Scr-ZIP
3: Scr-GluR2
3Y
+ ZIP
4: GluR2
3Y
+ Scr-ZIP
1: Untrained
5: GluR2
3Y
+ ZIP
GluR1 (relative percentage)
Freezing (%)
d
Postsynaptic
Extrasynaptic
c
Extrasynaptic
GluR2 (relative percentage)
Freezing (%)
Postsynaptic
Figure 2 Postsynaptic GluR2 levels are decreased in fear-conditioned rats after PKMζ inactivation and the levels of postsynaptic GluR2 correlate with
the magnitude of freezing during memory retention. (a,b) GluR23Y or Scr-GluR23Y was infused into the amygdala 1 d after the training session. ZIP
or Scr-ZIP was infused 1 h later. Memory was tested 1 d after the infusions and rats were killed 1 d later for BLA extraction. Representative western
blots (left) and quantification of GluR2 (a) and GluR1 (b) protein levels (right) in the indicated subcellular fractions from BLA are shown. The full-
length western blots are shown in Supplementary Figure 6. Data were normalized to the Scr-GluR23Y and Scr-ZIP group mean. Bars represent the
group means, and error bars represent s.e.m. The levels of postsynaptic GluR2 in the group that received Scr-GluR23Y and ZIP were significantly
lower (*P < 0.05) than in the group infused with both scrambled control peptides and in the group that received both GluR23Y and ZIP as determined
by the post hoc Tukey’s HSD test after significant one-way ANOVA. (c,d) Relationship between GluR2 (c) and GluR1 (d) levels in the postsynaptic or
extrasynaptic membrane fractions and freezing levels during the test of trained animals. A significant correlation was observed only for the postsynaptic
GluR2 fraction (R = 0.65, P < 0.001) by ANOVA overall goodness of fit.
© 2010 Nature America, Inc. All rights reserved.
63 2 VOLUME 13 | NUMBER 5 | MAY 2010 nature neurOSCIenCe
a r t I C l e S
surface and intracellular receptors (Supplementary Fig. 3), indicat-
ing that changes in receptor content in postsynaptic fractions repre-
sent changes in expression of surface receptors. A one-way ANOVA
revealed a significant difference in the levels of postsynaptic GluR2
(F3,24 = 7.522, P = 0.001) and post hoc comparisons (Tukey’s hon-
estly significant differences (HSD) test) confirmed that GluR2 levels
decreased in the postsynaptic membranes of fear-conditioned rats
infused with Scr-GluR23Y and ZIP (n = 6) compared with rats infused
with Scr-GluR23Y and Scr-ZIP (n = 8, P = 0.002). Consistent with
our behavioral results, infusions of GluR23Y prevented the decrease
of postsynaptic GluR2, as the GluR2 levels of rats that received both
GluR23Y and ZIP (n = 9) were similar to the levels of rats that received
both scrambled peptides (P = 0.986), but were different from the lev-
els of rats infused with Scr-GluR23Y and ZIP (P = 0.004). These results
indicate that ZIP causes a decrease in postsynaptic levels of GluR2
in trained rats and blocking AMPA receptor synaptic removal pre-
vents this decrease. GluR23Y alone had no effect on the postsynaptic
content of GluR2 (GluR23Y + Scr-ZIP versus Scr-GluR23Y + Scr-ZIP,
P = 0.995, n = 5), consistent with the finding that the GluR23Y peptide
does not affect AMPA receptor basal surface expression10,12. There was
no difference in extrasynaptic GluR2 levels between the treatment
groups (one-way ANOVA, F4,29 = 0.367, P = 0.829; Fig. 2a). Finally,
GluR1 levels did not differ among the groups at either postsynap-
tic (F4,17 = 0.038, P = 0.996) or extrasynaptic sites (F4,19 = 0.103,
P = 0.979) (Fig. 2b).
Further examination of the relationship between behavior after train-
ing and postsynaptic GluR2 expression revealed a positive correlation
(R = 0.658, F = 19.905, P < 0.001): stronger fear responses indicated
stronger postsynaptic GluR2 expression (Fig. 2c). A correlation was
not found between the fear response and the expression of GluR2 in
the extrasynaptic membranes (R = 0.031, F = 0.025, P = 0.875) or of
GluR1 (Fig. 2d) in either postsynaptic (R = 0.166, F = 0.428, P = 0.522)
or extrasynaptic fractions (R = 0.155, F = 0.393, P = 0.539).
Notably, the total levels of postsynaptic GluR2 in the BLA between
untrained and trained rats were indistinguishable (Scr-GluR23Y–Scr-ZIP
untrained versus Scr-GluR23Y–Scr-ZIP trained, Mann-Whitney
U test, U = 35, P = 0.656). One possible explanation for this find-
ing is that decreases in GluR2 in some synapses in the BLA might
have occurred following training, compensating for PKMζ-mediated
increases in other synapses. Such counterbalanced changes between
the mechanisms for increasing and decreasing synaptic strength have
been observed after training in the hippocampus13,14. However, as
ZIP injections into the BLA would reverse the mechanisms that lead
to increases in postsynaptic GluR2, but not
those mechanisms that lead to the compen-
satory decrease of postsynaptic GluR2 in
other synapses, a detectable difference can
be observed between trained animals infused
with ZIP and those infused with the control
peptide. Thus, injecting ZIP may reveal
AMPAR trafficking in trained animals that is
normally masked by compensatory responses.
Consistent with this hypothesis, ZIP had a
downward trend, but no significant effect on
the level of postsynaptic GluR2 in the BLA
of untrained rats (P = 0.308; Supplementary
Fig. 4). Thus, ZIP has a selective effect on post-
synaptic AMPARs of trained rats, suggesting
that the number of synapses potentiated by
PKMζ from prior experience in naive rats is
relatively small and therefore below the level
of detection. This is consistent with previous observations that ZIP
has no substantial effect on baseline synaptic transmission1,15.
A loss of AMPA receptors from postsynaptic sites also occurs in
NMDA receptor–dependent long-term depression (LTD)10,12. Thus,
although ZIP disrupts memory storage without overt behavioral cues,
it is possible that an unintentional signal (or ZIP itself) induces LTD
and may contribute to the memory loss. However, we found that
blocking NMDA receptors with d(−)-2-amino-5-phosphonovaleric
acid (AP5) before ZIP infusion did not diminish the disruption of
memory by ZIP (Supplementary Fig. 5), suggesting that the removal
of synaptic AMPA receptors by PKMζ inhibition is not dependent on
NMDA receptor activation, as in LTD.
Loss of postsynaptic GluR2 mediates LTP disruption by ZIP
We asked whether ZIP could impair LTP in the BLA as it does in the hippo-
campus and, if so, whether GluR23Y could prevent this impairment
as it prevented ZIP-mediated amnesia in our behavioral experiments.
Excitatory postsynaptic currents (EPSCs) were evoked by stimulating the
auditory thalamic synaptic inputs in BLA slices. To ensure the GluR23Y
effect is the result of blocking postsynaptic endocytosis, we perfused
GluR23Y postsynaptically into amygdala neurons through a whole-cell
recording patch pipette. Once stable EPSCs were obtained, ZIP or scram-
bled ZIP was applied to the bath and LTP was induced.
One-way ANOVA revealed a significant group effect (F3,21 = 5.138,
P = 0.007; Fig. 3). This group effect was the result of application
of Scr-GluR23Y and ZIP (n = 6) impairing LTP (n = 8), as deter-
mined by post hoc comparisons (Tukey’s HSD, versus Scr-GluR23Y +
Scr-ZIP, n = 6, P = 0.019; Fig. 3a,c), whereas application of GluR23Y
(n = 6) prevented the effect of ZIP (P = 0.032 versus Scr-GluR23Y +
ZIP) and led to a similar effect as applying both scrambled control
peptides (P = 0.993) (Fig. 3b,c). EPSCs in the presence of GluR23Y
alone (GluR23Y + Scr-ZIP, n = 5) were not significantly different from
the responses recorded after application of both scrambled peptides
(P = 0.999; Fig. 3b,c). The time course of the PKMζ-dependent
phase that we observed was consistent with the rapid onset of the
protein synthesis–dependent phase previously described for LTP in the
amygdala16. Thus, GluR2-dependent postsynaptic removal of AMPA
receptors mediates the effect of ZIP on LTP in amygdala slices.
PKM-mediated GluR2 trafficking sustains location memory
We then asked whether the mechanism by which PKMζ maintains
long-term memory may represent a general principle that can be
found in other brain regions and tasks. To maximize the difference
a
Normalized EPSC amplitude
0.5
1.0
1.5
2.0
2.5
3.0
5 10 15 20 25 30 35 40 45
Scr-GluR23Y + ZIP
Scr-GluR23Y + Scr-ZIP
Normalized EPSC amplitude
0.5
1.0
1.5
2.0
2.5
3.0
5 10 15 20 25 30 35 40 45
GluR23Y + ZIP
GluR23Y + Scr-ZIP
Normalized EPSC amplitude
0
0.5
1.0
1.5
2.0
Scr-GluR23Y
+ Scr-ZIP
Scr-GluR23Y
+ ZIP
GluR23Y
+ Scr-ZIP
GluR23Y
+ ZIP
b c
*
Figure 3 Blocking GluR2-dependent AMPA receptor synaptic removal prevents the LTP impairment
induced by PKMζ inactivation in BLA slices. (a) Bath application of ZIP, but not Scr-ZIP, reversed
LTP induced by 200 pulses at 2 Hz paired with postsynaptic depolarization to −5 mV. (b) The
reversal of LTP caused by ZIP was prevented by intracellular perfusion of GluR23Y. (c) Averaged
EPSC amplitude obtained at 40 min after LTP induction. The samples that received both Scr-
GluR23Y and ZIP were significantly different from all the other groups (*P < 0.05, post hoc Tukey’s
HSD test after significant one-way ANOVA).
© 2010 Nature America, Inc. All rights reserved.
nature neurOSCIenCe VOLUME 13 | NUMBER 5 | MAY 2010 633
a r t I C l e S
between tasks and memory systems, we chose object location memory,
which rats acquire in an emotionally neutral task by unrestricted and
unguided exploration in the absence of any external reinforcement
or punishment. Long-term memory for object location in this task
engages brain areas that are different from those recruited in auditory
fear conditioning17 and depends on persistent PKMζ activity in the
dorsal hippocampus 1 and 7 d after acquisition3.
We bilaterally infused GluR23Y (15 pmol) or its scrambled con-
trol peptide into dorsal hippocampus of rats 1 d after training and
then infused ZIP (10 nmol) or its scrambled control peptide 1 h
later. Memory for object location was assessed the next day. One-
way ANOVA detected a significant difference among the groups
in novel object-location exploration (F3,22 = 9.484, P < 0.001;
Fig. 4a), whereas the overall time spent exploring was not
significantly different among the groups (F3,22 = 0.671, P = 0.579;
Fig. 4b), indicating that differences in novel object-location explora-
tion were not the results of basic differences in exploratory activity.
Post hoc comparisons (Tukey’s HSD) determined that the group
infused with Scr-GluR23Y and ZIP (n = 7) explored the object at
the novel location less than the group infused with both scrambled
peptides (P = 0.001, n = 7) and the group infused with GluR23Y and
ZIP (P = 0.001, n = 7), which were not different from each other
(P = 0.952). The group infused with GluR23Y and Scr-ZIP (n = 5) was
also not different from the control group infused with both scram-
bled peptides (P = 0.797), indicating that GluR23Y alone had no
effect on the normal performance. Comparing exploratory activity
against what would have been expected by chance alone revealed
that only the rats infused with Scr-GluR23Y and ZIP explored both
objects equally (P = 0.123, t test), whereas all other groups preferred
the object at the new location to the object at the old location
(P < 0.001). These data indicate that GluR23Y prevents the impair-
ment of object location memor y induced by ZIP and suggest that
the mechanism by which PKMζ maintains long-term memory is
task and memory system independent.
DISCUSSION
Our results, across brain areas (amygdala and hippocampus), tasks
(auditory fear memory and object location memory) and both in
amygdala slices and in vivo, indicate that PKMζ maintains long-term
memories by blocking a GluR2-dependent pathway that removes post-
synaptic AMPA receptors to persistently promote increased levels of
GluR2-containing AMPA receptors at postsynaptic sites. It is always
possible, however, that other mechanisms may operate at later time
periods after memory acquisition, especially in the hippocampus,
where memory has been suggested to undergo a systems consolida-
tion process that can last for several months after learning18,19.
A previous study found that PKMζ causes synaptic potentiation
in LTP by upregulating the NSF/GluR2-dependent trafficking path-
way and focused on the initial trafficking of the receptor from extra-
synaptic to synaptic sites5; however, the role of reducing postsynaptic
GluR2/AMPA receptor clearance as the mechanism of maintaining
these changes has not been examined. This latter role is consistent
with observations that the interaction of NSF with GluR2 probably
stabilizes surface AMPA receptors in the postsynaptic membrane
by preventing the internalization of AMPA receptors (rather than
membrane insertion)6–8,20. In addition, NSF seems to be involved in
restricting GluR2-containing AMPARs from lysosomal degradation
and maintaining them in a recycling pathway21. Our results are there-
fore compatible with the idea that PKMζ acts persistently through
NSF/GluR2 interaction to prevent the removal of AMPA receptors
from postsynaptic sites and to maintain the constitutive recycling
of receptors. By blocking these mechanisms, ZIP could induce the
removal of the receptors from postsynaptic sites and/or facilitate
their degradation. Our biochemical results may therefore be con-
sistent with the possibility that training induces counterbalanced
PKMζ-maintained increases of GluR2-containing AMPA receptors
in some synapses and non–PKMζ mediated decreases in GluR2 in
others and/or a model in which training shifts the regulatory mecha-
nisms of overall or subpopulations of postsynaptic GluR2 trafficking
from a PKMζ-independent to a PKMζ-maintained mechanism. Such
a shift of regulatory trafficking mechanisms would provide a way to
maintain memories without affecting the receptor trafficking that
maintains baseline synaptic transmission.
Our results further indicate that both memory maintenance and
erasure are active molecular mechanisms, as both ZIP and GluR23Y
interfere with enzymatically driven processes. NSF/GluR2 inter-
action prevents the removal of postsynaptic AMPA receptors by dis-
rupting the interaction of the carboxyl tail of GluR2 with proteins
that are critical for receptor internalization such as PICK1 and AP2
(refs. 22,23). The binding motif comprised by GluR23Y, however, does
not include the AP2, NSF or the PDZ/PICK1-binding motifs12, indicat-
ing that the GluR23Y peptide does not interfere with the interaction of
GluR2 with these molecules. Instead, GluR23Y comprises a cluster of
tyrosine residues whose phosphorylation is a critical step in the post-
synaptic removal of GluR2-containing receptors through intracellular
internalization12,24 and is required for insulin-stimulated AMPA recep-
tor internalization and some forms of LTD, but not for baseline synaptic
transmission12. Although the exact mechanism by which this binding
motif regulates postsynaptic receptor removal is still poorly understood,
our results suggest that, were it not for the persistent action of PKMζ, this
mechanism would actively eliminate the postsynaptic pool of GluR2-
containing receptors by which LTP maintenance and memory storage
are expressed. Although enzymatically driven, the postsynaptic AMPA
receptor removal mediating the memory erasure by ZIP does not appear
to require the activation of NMDA receptors, as in LTD, because the
amnestic effect of ZIP is not prevented by NMDA receptor blockade.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Scr-GluR23Y + Scr-ZIP
0
5
10
15
20
25
30
35
*
ab
Exploration ratio
Total exploration time (s)
GluR2
3Y
+ ZIP
GluR23Y + Scr-ZIP
Scr-GluR23Y + ZIP
Chance
Figure 4 Blocking GluR2-dependent, postsynaptic AMPA receptor removal
prevents the impairment of object location memory induced by PKMζ
inactivation in the dorsal hippocampus. GluR23Y or Scr-GluR23Y was
infused into the dorsal hippocampus 1 d after the training session. ZIP or
Scr-ZIP was infused 1 h later. Memory was tested 1 d after the infusions.
(a) Values represent the mean ratio of the time that the rats spent
exploring a familiar object in a novel location over the time they spent
exploring a familiar object in the same location as in the training
session. ZIP infusions led to a total loss of memory for object location
(*P < 0.001), as determined by post hoc Tukey’s HSD test after
significant one-way ANOVA, such that the rats explored both objects at
the level of chance. GluR23Y infusions before ZIP infusions prevented
the memory loss. (b) Values represent the total exploration time. No
significant differences were observed among the treatment groups.
All error bars represent s.e.m.
© 2010 Nature America, Inc. All rights reserved.
63 4 VOLUME 13 | NUMBER 5 | MAY 2010 nature neurOSCIenCe
a r t I C l e S
We also found that GluR2-containing AMPA receptors are crucial
for the expression and maintenance of long-term memories. Several
studies have shown that synaptic insertion of GluR1-containing
receptors, believed to be GluR1/2 heteromers, is necessary for memory
formation and LTP induction in the amygdala10,25 , hippocampus13
and barrel cortex26,27. In the barrel cortex, however, GluR1-containing
receptors are replaced by GluR2-containing receptors that lack
GluR1, which are thought to be GluR2/3 heteromers26,27, within
24 h of training. Our findings that disrupting memory by PKMζ
inactivation decreased the levels of GluR2, but not GluR1, and that
GluR2 levels predicted memory performance are therefore consist-
ent with these results26–28. It is therefore possible that early phases
of memory formation are characterized by the initial postsynaptic
insertion of GluR1/2-containing receptors and that persistent post-
synaptic increases of GluR2/3-containing receptors allow memory to
be maintained in the long term28. A recent study found that GluR2
not only regulates synaptic function, but also promotes spine growth
and maintenance in neuronal cultures through its interaction with
N-cadherins29. Whether these morphological changes are sustained
by the same persistent action of PKMζ that maintains postsynaptic
GluR2 expression and the persistence of behavioral memory is an
important question for future study.
METHODS
Methods and any associated references are available in the online version
of the paper at http://www.nature.com/natureneuroscience/.
Note: Supplementary information is available on the Nature Neuroscience website.
ACKNOWLEDGMENTS
This work was supported by the Canadian Institutes of Health Research and the
Natural Sciences and Engineering Research Council of Canada (K.N. and Y.T.W.),
the E.W.R. Steacie Foundation (K.N.) and by US National Institutes of Health
grants R01 MH53576 and MH57068 (T.C.S.). O.H. was supported by the German
Research Society (Deutsche Forschungsgemeinschaft).
AUTHOR CONTRIBUTIONS
All of the authors contributed to the design of experiments, interpretation of
results and editing of the manuscript. P.V.M. and O.H. conducted the behavioral
studies. P.V.M. carried out the biochemical studies. D.C.W. conducted the
electrophysiological studies. K.G. and O.H. performed the stereotaxic surgeries.
P.V.M., O.H., T.C.S. and K.N. wrote the manuscript.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Published online at http://www.nature.com/natureneuroscience/.
Reprints and permissions information is available online at http://www.nature.com/
reprintsandpermissions/.
1. Pastalkova, E. et al. Storage of spatial information by the maintenance mechanism
of LTP. Science 313, 1141–1144 (2006).
2. Serrano, P. et al. PKMzeta maintains spatial, instrumental, and classically
conditioned long-term memories. PLoS Biol. 6, 2698–2706 (2008).
3. Hardt, O., Migues, P., Hastings, M., Wong, J. & Nader, K. PKMzeta maintains one
day and six-day old long-term object location, but not object identity memory in
dorsal hippocampus. Hippocampus published online, doi:10.1002/hipo.20708
(5 October 2009).
4. Shema, R., Sacktor, T.C. & Dudai, Y. Rapid erasure of long-term memory
associations in the cortex by an inhibitor of PKM zeta. Science 317, 951–953
(2007).
5. Yao, Y. et al. PKM zeta maintains late long-term potentiation by N-ethylmaleimide–
sensitive factor/GluR2-dependent trafficking of postsynaptic AMPA receptors.
J. Neurosci. 28, 7820–7827 (2008).
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. Brebner, K. et al. Nucleus accumbens long-term depression and the expression of
behavioral sensitization. Science 310, 1340–1343 (2005).
10. Yu, S.Y., Wu, D.C., Liu, L., Ge, Y. & Wang, Y.T. Role of AMPA receptor trafficking
in NMDA receptor–dependent synaptic plasticity in the rat lateral amygdala.
J. Neurochem. 106, 889–899 (2008).
11. Gale, G.D. et al. Role of the basolateral amygdala in the storage of fear memories
across the adult lifetime of rats. J. Neurosci. 24, 3810–3815 (2004).
12. Ahmadian, G. et al. Tyrosine phosphorylation of GluR2 is required for insulin-
stimulated AMPA receptor endocytosis and LTD. EMBO J. 23, 1040–1050
(2004).
13. Whitlock, J.R., Heynen, A.J., Shuler, M.G. & Bear, M.F. Learning induces long-term
potentiation in the hippocampus. Science 313, 1093–1097 (2006).
14. Xu, L., Anwyl, R. & Rowan, M.J. Spatial exploration induces a persistent reversal
of long-term potentiation in rat hippocampus. Nature 394, 891–894 (1998).
15. Ling, D.S. et al. Protein kinase Mzeta is necessary and sufficient for LTP
maintenance. Nat. Neurosci. 5, 295–296 (2002).
16. Huang, Y.Y., Martin, K.C. & Kandel, E.R. Both protein kinase A and mitogen-
activated protein kinase are required in the amygdala for the macromolecular
synthesis–dependent late phase of long-term potentiation. J. Neurosci. 20,
6317–6325 (2000).
17. Mumby, D.G., Gaskin, S., Glenn, M.J., Schramek, T.E. & Lehmann, H. Hippocampal
damage and exploratory preferences in rats: memory for objects, places and
contexts. Learn. Mem. 9, 49–57 (2002).
18. Kim, J.J. & Fanselow, M.S. Modality-specific retrograde amnesia of fear. Science
256, 675–677 (1992).
19. Winocur, G. Anterograde and retrograde amnesia in rats with dorsal hippocampal
or dorsomedial thalamic lesions. Behav. Brain Res. 38, 145–154 (1990).
20. Noel, J. et al. Surface expression of AMPA receptors in hippocampal neurons is
regulated by an NSF-dependent mechanism. Neuron 23, 365–376 (1999).
21. Lee, S.H., Simonetta, A. & Sheng, M. Subunit rules governing the sorting of
internalized AMPA receptors in hippocampal neurons. Neuron 43, 221–236
(2004).
22. Man, H.Y. et al. Regulation of AMPA receptor–mediated synaptic transmission by
clathrin-dependent receptor internalization. Neuron 25, 649–662 (2000).
23. Hanley, J.G., Khatri, L., Hanson, P.I. & Ziff, E.B. NSF ATPase and alpha-/beta-
SNAPs disassemble the AMPA receptor–PICK1 complex. Neuron 34, 53–67
(2002).
24. Hayashi, T. & Huganir, R.L. Tyrosine phosphorylation and regulation of the
AMPA receptor by SRC family tyrosine kinases. J. Neurosci. 24, 6152–6160
(2004).
25. Rumpel, S., LeDoux, J., Zador, A. & Malinow, R. Postsynaptic receptor trafficking
underlying a form of associative learning. Science 308, 83–88 (2005).
26. Takahashi, T., Svoboda, K. & Malinow, R. Experience strengthening transmission
by driving AMPA receptors into synapses. Science 299, 1585–1588 (2003).
27. McCormack, S.G., Stornetta, R.L. & Zhu, J.J. Synaptic AMPA receptor exchange
maintains bidirectional plasticity. Neuron 50, 75–88 (2006).
28. Kessels, H.W. & Malinow, R. Synaptic AMPA receptor plasticity and behavior. Neuron
61, 340–350 (2009).
29. Saglietti, L. et al. Extracellular interactions between GluR2 and N-cadherin in spine
regulation. Neuron 54, 461–477 (2007).
© 2010 Nature America, Inc. All rights reserved.
nature neurOSCIenCe
doi:10.1038/nn.2531
ONLINE METHODS
Subjects. Male Sprague-Dawley and Long Evans rats (only for the object location
recognition experiments) (275–350 g, Charles River) were housed in pairs in
plastic Nalgene cages with environmental enrichment and maintained on a 12:12-h
light dark cycle. All experiments were carried out during the light phase. Food
and water were provided ad libitum throughout the experiment. All procedures
complied with Canadian Council on Animal Care guidelines and were approved
by McGill University’s Animal Care Committee.
Surgery. Rats were anaesthetized with a mixture of xylazine (3.33 mg ml−1),
ketamine (55.55 mg ml−1) and Domitor (0.27 mg ml−1) by intraperitoneal injec-
tion of a volume of 1 ml per kg of body weight. Using a Kopf Stereotax, we
implanted three jeweler screws into the skull and then placed two steel cannulae
(22 gauge), aiming bilaterally at the BLA (anteroposterior, −3.0 mm; medial-lateral,
±5.1 mm; dorsoventral, −8.0 mm) or dorsal hippocampus (anteroposterior, −3.6 mm;
medial-lateral, ± 3.1 mm; dorsoventral, −2.4 mm, with the cannula aimed
10° away from midline). Dental cement was applied to stabilize the implants.
Obturators inserted in the guides prevented blocking. After surgery, an intra-
muscle injection of analgesic (buprenorphine, 0.324 mg per kg) was given. An
intraperitoneal injection of Antisedan (7.5 mg per kg) suspended anesthesia. The
rats were then allowed 7–10 d to recover from surgery. Rats were handled daily
during the recovery period. At the end of the experiment, cannula placement
was checked by examining 50-µm brain sections stained with formal-thionin
under a light microscope.
Drug infusions. ZIP (Myr-SIYRRGARRWRKL-OH, Anaspec), Scr-ZIP (Myr-
RLYRKRIWRSAGR-OH, Anaspec), GluR23Y (TAT(47–57)-869YKEGYNVYG877),
Scr-GluR23Y (TAT(47–57)-AKEGANVAG) and AP5 (Sigma) were dissolved in
100 mM Tris-saline (pH 7.2). They were infused into the BLA (ZIP, 1 µl per
hemisphere; GluR23Y and AP5, 0.5 µl per hemisphere) or dorsal hippocampus
(1 µl per hemisphere) via a microinjector (28 gauge) connected to a Hamilton
syringe with plastic tubing at a rate of 0.25 µl per min. The injector remained
connected for an additional min to allow for drug diffusion away from the tip
of the cannula.
Auditory fear conditioning task. The training chambers (22 × 22 × 22 cm, Med
Associates) were equipped with a stainless steel grid (bar radius = 2.5 mm, spread =
1 cm apart) and the floor was connected to a shocker. The walls were made of
Plexiglass. Three key lights were mounted on the sidewalls. A fan provided a
constant background noise. Behavior was recorded by a digital video camera
mounted in front of the chamber door. To assess only the auditory component
of the fear memory, the testing chambers were different from the training ones
with respect to textual, visual and olfactory cues. The floor of the testing chambers
was made of opaque plastic panels. The front and the back walls were pasted with
wallpapers containing black and white alternating strips. Peppermint scent was
sprayed onto the floor before the rat was placed in the box. A digital camera was
mounted on the ceiling and videotaped the sessions for later analysis.
Rats were habituated to the training and testing chambers by placing them in
each chamber for 20 min with a 5-h interval between them for 2 consecutive days.
The order of chambers was reversed on the second day. Training was performed
the day after habituation. Rats were placed into the box and, after 2 min, a tone
(5 kHz, 75 dB) was presented for 30 s, which co-terminated with a foot shock
(1.5 mA, 1 s). The rats remained in the box for another 30 s and were then returned
to their home cages. Rats were tested for 6 min, during which three 30-s tones
were presented, with an interval of 60 s between each. The first tone was presented
2 min after the rats were placed into the box. Memory was evaluated by averaging
the freezing time during each tone. Because the rats did not freeze before the first
tone, the freezing to the tone accurately represents auditory fear memory.
Object location recognition task. The experiment was carried out in a
40 × 40 × 60 cm open field arena (that is, a box made of white laminated wood).
The walls were high enough to obscure external room cues. Sawdust covered the
floor. Objects were secured in the floor by screws. A video camera above the box
recorded behavior.
Rats were habituated over 3 consecutive days. On day 1, the co-housed pairs
were placed into the open field and were able to explore it for 20 min together.
No objects were present. On day 2, rats were placed individually into the open
field and explored it, again without objects, for 10 min. On day 3, rats were placed
individually into the open field in the absence of objects for 5 min. The sampling
phase started the next day. Two sampling sessions were administered, the first in
the morning and the second in the afternoon. Each session consisted of five 5-min
trials per rat, separated by ~1 h. During each 5-min trial, rats were exposed to two
identical copies of a junk object. The objects were made of different materials and
had different shapes, colors and textures (for example, metal cup, ceramic incense
burner, plastic baby bottle, etc.). The objects were placed in opposing corners
during the ten sampling trials. The objects and their locations remained constant
for each rat, but varied between rats and were counterbalanced. Object location
knowledge was assessed in a 3-min probe trial. Rats were presented with the same
objects that they encountered during sampling; however, one of the objects was
moved to a new location. Object exploration was defined as the time when the
rat’s head was oriented toward the object within 45 degrees and was within 4 cm
of the object. The exploration ratio observed per min was averaged across the
first minute only17. The exploration ratio is defined as the relative time the animal
spends exploring the novel stimulus (that is, the object at the novel location) rela-
tive to the total time the animal engaged in stimulus exploration: tnovel/(tfamiliar +
tnovel)17. A ratio of 0.5 indicates the absence of location memory.
Subcellular fractionation. Rats were anesthetized as for surgeries, decapitated,
and their brains were removed and frozen. The amygdala was dissected from
the frozen brains with a neuro punch (1 mm, Fine Science Tools) and homo-
genized in ice-cold Tris-HCl buffer (30 mM, pH 7.4) containing 4 mM EDTA,
1 mM EGTA and a cocktail of protease inhibitors (Complete, Roche). Subcellular
fractions were prepared as described30. Briefly, the amygdala homogenates were
centrifuged twice at 4 °C at 500g for 5 min to remove nuclei and other debris. The
two supernatants were pooled and centrifuged at 100,000g at 4 °C for 60 min. Pellets
were resuspended in the same buffer containing 0.5% Triton X-100 and incubated
at 4 °C for 20 min, layered over 1 M sucrose and centrifuged at 100,000g for
60 min. Finally, the Triton-soluble fraction that remained above the sucrose layer,
which consists mainly of detergent-soluble membrane components and contains
the extra-synaptic receptors, was collected and the Triton-insoluble material that
sedimented through the sucrose layer, which is highly enriched in postsynaptic
densities, was resuspended in the same buffer and stored at −80 °C. Total protein
concentration was determined by the BCA protein assay kit (Pierce).
Western blotting. Western blotting was performed using 7.5% SDS-polyacrlyl-
amide gel electrophoresis. The proteins were transferred onto nitrocellulose
membranes and incubated with polyclonal antibodies to GluR2 (0.5 µg ml−1,
Millipore), GluR1 (1 µg ml−1, Millipore) or PSD-95 (0.6 µg ml−1, Chemicon), fol-
lowed by incubation with goat antibody to rabbit horseradish peroxidase–linked
IgG. We used the ECL plus immunoblotting system (Amersham) for detection.
Blots were scanned with a Storm Laser Scanner (Molecular Dynamics) and ana-
lyzed with ImageQuant software (ABI).
Slice electrophysiology. Rats were placed under deep anesthesia and decapitated.
The brain was rapidly removed and placed in ice-cold slicing solution containing
87 mM NaCl, 2.5 mM KCl, 7 mM MgCl2, 0.5 mM CaCl2, 1.25 mM NaH2PO4,
26 mM NaHCO3, 25 mM glucose and 75 mM sucrose, which was bubbled con-
tinuously with carbogen (95% O2 and 5% CO2) to adjust pH to 7.4. Coronal slices
of 400-µm thickness containing the amygdala were produced using a vibrating
blade microtome and recovered in an incubation chamber with carbogenated
artificial cerebrospinal fluid (ACSF) containing 125 mM NaCl, 2.5 mM KCl,
1.0 mM MgCl2, 2.0 mM CaCl2, 1.25 mM NaH2PO4, 26 mM NaHCO3 and
25 mM glucose for 45 min at 34 °C and were then returned to 20–22 °C for at least
30 min before recording. All experiments were carried out at 20–22 °C.
Recordings were carried out in a chamber perfused continuously by carbogen-
ated ACSF containing bicuculline methiodide (10 µM) to block GABAA receptor–
mediated inhibitory synaptic currents. Whole-cell patch-clamp recordings were
performed using the ‘blind’ method from neurons in the dorsal part of the lateral
amygdala. Recording pipettes were filled with pipette solution containing 122.5 mM
cesium gluconate, 17.5 mM CsCl, 2 mM MgCl2, 10 mM HEPES, 0.5 mM EGTA,
4 mM K-ATP and 5 mM QX-314 (pH 7.2, osmolarity 290–300 mOsm), with or
without 100 µg ml−1 GluR23Y or scrambled GluR23Y. The resistance of electrodes
was typically 4–5 MΩ. After obtaining the whole-cell configuration, clamp
current was used to identify the firing pattern of the cells.
© 2010 Nature America, Inc. All rights reserved.
nature neurOSCIenCe doi:10.1038/nn.2531
After cell characterization, the membrane potential was held at −70 mV. EPSCs
were evoked by stimulating the auditory thalamic synaptic inputs via a con-
stant current pulse (0.05 ms) delivered through a tungsten bipolar electrode
and recorded with a MultiClamp 700B amplifier (Axon Instruments). Synaptic
responses were evoked at 0.05 Hz except during the induction of LTP. Once stable
EPSCs were obtained, ZIP or scrambled ZIP was applied to the bath (5 µM, final
concentration). After obtaining a stable EPSC baseline, LTP was induced by 200
pulses at 2 Hz while depolarizing the cell to −5 mV. The stimulation intensity of
induction was the same as that used during baseline recording. LTP was induced
within 10 min of the establishment of a whole-cell configuration to avoid washout
of intracellular contents.
Slice biotinylation and neutravidin pull-down. Coronal brain slices of 300-µm
thickness containing amygdala and hippocampus were prepared using a
vibratome and maintained in ice-cold carbogenated ACSF containing 124 mM
NaCl, 3 mM KCl, 1.0 mM MgSO4, 2.0 mM CaCl2, 1.25 mM NaH2PO4, 26 mM
NaHCO3 and 10 mM glucose. The biotinylation was performed as described
previously31 with some modifications. Briefly, slices were incubated with NHS-
SS-Biotin (Pierce) for 45 min on ice. Excess biotin was removed by washes with
NH4Cl. Slices were then homogenized at 4 °C in Tris-HCl buffer (30 mM, pH 7.4)
containing 4 mM EDTA, 1 mM EGTA and a cocktail of protease inhibitors
(Complete, Roche), and subcellular fractions were prepared as described above.
The subcellular fractions were solubilized in lysis buffer containing 150 mM
NaCl, 20 mM HEPES (pH 7.4), 2 mM EDTA, 1% Triton X-100 and 0.1% SDS,
sonicated and incubated in a shaker for 2 h at 4 °C. The total protein concentra-
tion was determined using the Pierce BCA kit. The samples were then incubated
with Neutravidin beads for 10 h at 4 °C. Samples were centrifuged and the super-
natant containing the unbound fraction was used to determine the intracellular
content of receptors. The biotinylated proteins were eluted in equal volume as
the unbound fraction with 2× loading buffer at 90 °C for 5 min. The amount of
GluR1, GluR2 and PSD95 was determined by western blot.
Kinase activity. The reaction mixture (50 µl final volume) contained 50 mM Tris-
HCl (pH 7.5), 10 mM MgCl2, 1 mM dithiothreitol, PKMζ (0.015 pmol per min
per assay phosphotransferase activity), 6 µM myelin basic protein as substrate,
and varying concentrations of ZIP and GluR23Y as indicated. The reaction, begun
with the addition of 50 µM ATP (4 µCi [γ-32P] per assay), continued for 10 min at
30 °C, which is in the linear range for time and enzyme concentration (data not
shown). The reaction was stopped by addition of 25 µl of 100 mM ice-cold ATP
and 100 mM EDTA, and 40 µl of the assay was spotted onto phosphocellulose
paper and counted by liquid scintillation. PKMζ activity was measured as the dif-
ference between counts incorporated in the presence and absence of enzyme.
Statistical analysis. All data were tested for normality distribution fitting by the
Shapiro-Wilk W test and for homogeneity of variance by Levene’s test. When
values were normally distributed and groups had identical variance, we used a
one-way ANOVA, followed by Tukey’s HSD post hoc test when appropriate. When
data did not meet the above ANOVA assumptions, we used Kruskal-Wallis analy-
sis of ranks for multiple comparison and Mann Whitney U test for two samples
comparisons. For nonparametric two dependent samples comparisons, we used
the Wilcoxon matched pairs test.
30. Migues, P.V. et al. Phosphorylation of CaMKII at Thr253 occurs in vivo and enhances
binding to isolated postsynaptic densities. J. Neurochem. 98, 289–299 (2006).
31. Feligioni, M., Holman, D., Haglerod, C., Davanger, S. & Henley, J.M. Ultrastructural
localization and differential agonist-induced regulation of AMPA and kainate
receptors present at the presynaptic active zone and postsynaptic density.
J. Neurochem. 99, 549–560 (2006).
© 2010 Nature America, Inc. All rights reserved.
