Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Article
Engram Cell Excitability S
tate Determines theEfficacy of Memory RetrievalHighlights
d Memory recall induces a transient increase in engram cell
excitability
d The engram excitability increase is mediated by Kir2.1
channel internalization
d The engram high-excitability state is associated with
enhanced context recognition
d Engram-specific exogenous expression of Kir2.1 channels
impairs context recognition
Pignatelli et al., 2019, Neuron 101, 274–284January 16, 2019 ª 2018 Elsevier Inc.https://doi.org/10.1016/j.neuron.2018.11.029
Authors
Michele Pignatelli, Tomas J. Ryan,
Dheeraj S. Roy, Chanel Lovett,
Lillian M. Smith, Shruti Muralidhar,
Susumu Tonegawa
[email protected] (M.P.),[email protected] (S.T.)
In Brief
Memory exists across different time
scales. Long-term memories can last a
lifetime, while working memory persists
for mere seconds. In between, memories
that are newly formed—or recently
recalled—are more vivid than latent ones.
How does this happen? The act of
memory recall increases the excitability
of whole engram cells for about one hour.
This short-term increase of engram cell
excitability facilitates synaptic access to
memory content in response to relevant
environmental cues.
Neuron
Article
Engram Cell Excitability StateDetermines the Efficacy of Memory RetrievalMichele Pignatelli,1,2,5,* Tomas J. Ryan,1,2,3,5 Dheeraj S. Roy,1,4 Chanel Lovett,1,2 Lillian M. Smith,1 Shruti Muralidhar,1
and Susumu Tonegawa1,2,6,*1RIKEN-MIT Center for Neural Circuit Genetics at the Picower Institute for Learning and Memory, Department of Biology and Department of
Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA2Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, MA 02139, USA3Current address: School of Biochemistry and Immunology and Trinity College Institute of Neuroscience, Trinity College Dublin, Dublin D02
PN40, Ireland4Current address: Broad Institute of MIT and Harvard, Massachusetts Institute of Technology, Cambridge, MA 02139, USA5These authors contributed equally6Lead Contact
*Correspondence: [email protected] (M.P.), [email protected] (S.T.)
https://doi.org/10.1016/j.neuron.2018.11.029
SUMMARY
Animals need to optimize the efficacy of memoryretrieval to adapt to environmental circumstancesfor survival. The recent development of memoryengram labeling technology allows a precise investi-gation of the processes associated with the recall ofa specific memory. Here, we show that engram cellexcitability is transiently increased followingmemoryreactivation. This short-term increase of engramexcitability enhances the subsequent retrieval ofspecific memory content in response to cues and ismanifest in the animal’s ability to recognize contextsmore precisely and more effectively. These resultsreveal a hitherto unknown transient enhancementof context recognition based on the plasticity ofengram cell excitability. They also suggest that recallof a contextual memory is influenced by previous butrecent activation of the same engram. The state ofexcitability of engram cells mediates differentialbehavioral outcomes upon memory retrieval andmay be crucial for survival by promoting adaptivebehavior.
INTRODUCTION
Engrams are the specific changes in the brain that are formed by
experience (Semon, 1921) and stored in a quiescent state that
becomes functional under conditions that lead to retrieval
(Tulving, 1983). Recent studies have identified distinct popula-
tions of neurons that hold engrams for particular memories, the
engram cells (Han et al., 2009; Josselyn et al., 2015; Liu et al.,
2012; Ramirez et al., 2013; Tonegawa et al., 2015a). These
engram cells are reactivated by specific cues associated with
the training experience (Deng et al., 2013; Denny et al., 2014;
Guzowski et al., 1999; Ramirez et al., 2013; Ryan et al., 2015),
274 Neuron 101, 274–284, January 16, 2019 ª 2018 Elsevier Inc.
and their stimulation can directly elicit memory recall (Liu et al.,
2012; Ryan et al., 2015). An engram is formed and stored as a
specific neuronal connectivity pattern between cells (Choi
et al., 2018; Redondo et al., 2014; Roy et al., 2017; Ryan et al.,
2015; Tonegawa et al., 2015b). Stability of engram cell connec-
tivity contributes to memory storage and is therefore necessary
for the normal process of memory retrieval. But the efficacy of
memory retrieval is strongly influenced by environmental factors
such as cue availability and cue competition (Riccio et al., 1984).
The success of memory retrieval can influence the balance be-
tween avoidance of threatening situations and exploitation of
the environment, both necessary processes for the survival of
an animal (Passano, 1957).
Under natural circumstances, when cue availability is limited,
memory retrieval may take advantage of a mnemonic process
called pattern completion, the ability to retrieve an entirememory
from a partial or degraded sensory cue (Marr, 1971). Further-
more, situations may arise in which multiple conflicting cues
are present, and their discrimination requires an additional pro-
cess called pattern separation, in which the overlap between
representations is minimized (O’Reilly and McClelland, 1994).
While stability of synaptic connectivity between specific engram
cells is a necessary condition for the retrieval of a specific mem-
ory, plasticity of whole-engram-cell excitability could, in princi-
ple, optimize the efficacy of retrieval by maximizing separation
and completion, resulting in an improvement of accuracy and ef-
ficiency of recognition.
Excitability is an intrinsic physiological property of neurons
that sets the threshold for spike generation and controls signal
transmission (Hille, 1992). Plasticity of the degree of excitability
is regulated through changes in membrane conductance that
can be induced by Hebbian forms of synaptic activation (Aizen-
man and Linden, 2000; Brons and Woody, 1980; Debanne and
Poo, 2010; Disterhoft et al., 1986; Fan et al., 2005; Frick et al.,
2004; Ganguly et al., 2000; Li et al., 2004; Zhang and Linden,
2003). Behavioral training is known to induce broad changes in
neuronal excitability in both vertebrates and invertebrates (Brons
and Woody, 1980; Disterhoft et al., 1986; McKay et al., 2009;
Mozzachiodi et al., 2008), and it has been proposed that the
excitability state may contribute to memory formation (Disterhoft
and Oh, 2006; Schrader et al., 2002), consolidation (Zhang and
Linden, 2003), and allocation (Cai et al., 2016; Josselyn and
Frankland, 2018; Lisman et al., 2018; Rashid et al., 2016; Yiu
et al., 2014). Nevertheless, it is unknown whether recall cues
can drive excitability changes in engram cells and how such
changes may influence memory retrieval.
Here, we investigated the dynamics of excitability of dentate
gyrus (DG) engram cells upon their reactivation by recall cues af-
ter contextual fear conditioning.We found that reactivation of DG
engram cells by contextual cues resulted in a transient enhance-
ment of excitability, which was specific to engram cells, immedi-
ately after recall. This rapid but transient change in the state of
the engram cell excitability increased the accuracy and effi-
ciency of context recognition and thereby augmented the effi-
cacy of memory retrieval.
RESULTS
Transient Increase of Engram Cell Excitability State inResponse to Memory RecallWe employed our previously established method for labeling the
DG component of a contextual fear memory engram with ChR2-
EYFP (Liu et al., 2012). An AAV9-TRE-ChR2-EYFP was injected
into the DG of transgenic mice engineered to express tTA under
the control of the c-fos promoter (c-fos-tTA) (Reijmers et al.,
2007). Temporal control over the expression was achieved by
administering a doxycycline diet (DOX, 40 mg/Kg) (Figures 1A
and S1A–S1C). Mice were taken off of DOX 24 hr prior to contex-
tual fear conditioning (CFC) to tag DG engram cells with ChR2-
EYFP. After CFC, mice were returned to the home cage on a
DOX diet. On a second day, mice were re-exposed to the condi-
tioned context to elicit memory recall and then sacrificed to
examine the physiological properties of engram (ChR2-EYFP+)
and non-engram (ChR2-EYFP�) cells by ex vivo patch-clamp re-
cordings. Intrinsic properties were determined in whole-cell
configuration in three experimental groups (Figures 1B and 1C):
a no recall group (NR), a 5-min-post recall group (5m) and a
3-hr-post recall group (3h) (Figures 1D–1F). The resting mem-
brane potential and action potential threshold were unaltered
across engram and non-engram cells of all groups (Figure S2A).
Membrane resistance and rheobase displayed similar values be-
tween engram and non-engram cells for the NR and 3h groups.
However, we detected a significant increase in membrane resis-
tanceanda reduction in the rheobase in engramcells from the5m
group (Figure 1E). These results revealed an enhanced neuronal
excitability due tomembrane changes in response to engramcell
reactivation induced by re-exposure to the conditioned context
(Figures S1D–S1G). Analysis of the current to spike frequency
discharge confirmed the result (Figure S2B). Spike rate accom-
modation (Figure S2B), action potential after hyperpolarization
(AHP) amplitude and duration (Figure S2C), and post-burst AHP
(Figure S2D) were unaltered across engram and non-engram
cells of all groups. Membrane excitability wasn’t correlated to
the intercellular distance between engram and non-engram cells
(Figure S2E). This engram-cell-specific increase in membrane
excitability was still detected at 10 min, 30 min, and 1 hr after
recall and decayed to baseline in 2 hr (Figure 1G). Importantly,
these changes were still triggered by recall of a purely contextual
memory formed in the absence of associative conditioning (Fig-
uresS3A–S3E).Whenanovel context (contextC)wassubstituted
for the conditioned context (context A) at the recall step, mem-
brane properties of engram cells were not altered, demonstrating
the context specificity of the transient increase in engram cell
excitability (Figures S3F and S3G). Direct optogenetic activation
of engram cells in a novel context did not alter engram cell mem-
brane properties as measured 5 min post-stimulation (Figures
S3H and S3I). The recall-induced increase in membrane
excitability of engram cells was abolished by injection of an
NMDA receptor antagonist before recall (Figure 1H), and its
decay was prevented by injection of a protein synthesis inhibitor
immediately after recall (Figure 1I), suggesting the importance of
avoiding a prolonged hyper-excitation of engram cells.
Inward Rectifier Potassium Channels Control theEngram Cell Excitability StateMembrane excitability can be regulated by changes in intrinsic
conductances, and most commonly, inward-rectifier potassium
channels (Kir2 family) can influence the membrane resistance of
DG granule cells (Young et al., 2009). DG granule cells displayed
a strong inward rectification (Figures 2A and 2B).We performed a
voltage-clamp recording of DG granule cells, holding the mem-
brane potential at �70 mV and recording the current evoked by
a negative step voltage (�30 mV). Bath application of a Kir2.1blocker (ML133, 50 mM) suppressed 68%of the total current (Fig-
ure 2C). Pharmacological profiling that targeted the contribution
of each leaky potassium conductance confirmed the specificity
of the result (Figure S4A). The Kir2.1 current was significantly
correlated with the membrane resistance, suggesting a direct
control of this specific ion channel over themembrane excitability
(Figure 2D). Importantly, 5min after recall we found a reduction of
the engram cell leaky current relative to that of non-engram cells.
This reduction was dependent on the decrease of Kir2.1 current
and not on a change of residual non-Kir2.1 current (Figures 2E–
2H). The dominant contribution of Kir2.1 channels in regulating
DG cell excitability and the effect of protein-synthesis inhibitor
in abolishing the return to baseline excitability level suggested a
mechanism for an engram cell high-excitability state based on
a downregulation of Kir2.1 channels. To test this hypothesis, we
injected an AAV9-TRE-ChR2-EYFP into the DG of c-fos-tTA
mice, subjected them to CFC while off DOX, and re-exposed
them to the conditioned context to elicit memory recall. Mice
were grouped according to previous criteria (NR, 5m, 3h) with
the addition of an anisomycin group (ANI 150 mg/Kg). We per-
formed a Kir2.1 immunostaining and compared Kir2.1 expression
in pairs of adjacent engramandnon-engramcells. Confocal anal-
ysis revealed a downregulation of this channel exclusively in
engram cells of the 5m and ANI groups, but not of the NR or 3h
groups relative to non-engram cells (Figures 2I–2L, Figures S4B
and S4C), providing a molecular mechanism for the recall-
induced transient increase of engram cell excitability.
Behavioral Correlates of the Engram Cell High-Excitability StateWe then investigated a potential role of the recall-induced
transient increase of engram cell excitability in recall-associated
Neuron 101, 274–284, January 16, 2019 275
Figure 1. Engram Cell-Specific High-Excitability State in Response to Memory Recall
(A) Schematic of the experimental strategy (see text and STAR Methods).
(B) Behavioral protocol: on day 0, mice are taken off doxycycline diet (Dox Off). On day 1, mice are subjected to contextual fear conditioning (CFC) and the
expression window is closed (Dox On). On day 2, mice are re-exposed to the conditioned context A and sacrificed 5 min (5m) or 3 hr (3h) after recall to perform
ex vivo patch-clamp recording of dentate granule (DG) cells. An additional control, a no recall group (NR), was not subjected to context re-exposure.
(C) Confocal image of two biocytin-filled DG cells, ChR2-EYFP+ and ChR2-EYFP�.(D–F) Intrinsic electrophysiological properties:membrane resistance (Rm)vs rheobaseofDGgranule cells belonging toNR (D), 5m (E), and3h (F) groups (top, example
traces).ChR2-EYFP+engramcells are ingreen; nonengramcells are inblack.Redbox: increasedRmand rheobase in theChR2-EYFP+engramcellsof the 5mgroup.
(G) Temporal dynamic of the excitability state. Top: behavioral protocol; bottom: membrane resistance of ChR2-EYFP+ (green) and ChR2-EYFP� (black) cells.
(H and I) The high-excitability state is NMDA dependent (H) and its decay to baseline is protein synthesis dependent (I). Top: behavioral protocol. Mice were
injected with CPP (5 mg/Kg i.p.) before recall in context A (H) or with Anisomycin (150 mg/Kg i.p.) after recall in context A (I). Mice were sacrificed for patch-clamp
recordings 5 min (H) or 24 hr (I) after recall in context A.
Data are represented asmean ±SEM, and statistical significance is assessed by two-tailed unpaired t test for D–F, H, and I (t test *p < 0.05, **p < 0.01, ***p < 0.001)
and two-way ANOVA with Sidak correction for multiple comparison for Figure 1G: df = 283, Frow(6, 171) = 9.000, Ftime(1, 114) = 21.81 (*p < 0.05, **p < 0.01).
Number of mice used: NR: 6, 5m: 10, 10m: 5, 30m: 3, 1h: 8, 2h:3, 3h: 7, 24h: 6, CPP: 3, and ANI: 4.
276 Neuron 101, 274–284, January 16, 2019
Figure 2. Kir2.1 Regulates the State of The DG Granule Cell Membrane Excitability
(A) DG granule cell (inset) response to current steps.
(B) Somatic voltage clamp analysis of current-voltage (I-V) relationship revealed an inwardly rectifying current.
(C) Leaky current in response to a negative voltage step is sensitive to ML133 (50 mM), a Kir2.1 blocker. Right: paired analysis, control in red and ML133 in black
(n = 15, paired t test ***p < 0.001) and percentage of leaky current sensitive to ML133 and residual one.
(D) Correlation between membrane resistance (Rm) and the pharmacologically isolated Kir2.1 current.
(E) Top: schematic of the experimental strategy (see text and STAR Methods). Bottom: behavioral protocol (same as Figure 1B). DG cells ChR2-EYFP+ and
ChR2-EYFP� were recorded 5 minutes after recall.
(F) Confocal image of two biocytin-filled DG cells, ChR2-EYFP+ and ChR2-EYFP�.(G) Inward current in response to a negative voltage step before and after ML133 application in ChR2-EYFP� and ChR2-EYFP+ cells.
(H) Quantification of total leaky current, Kir2.1 current and residual (non Kir2.1) leaky current in ChR2-EYFP� and ChR2-EYFP+ cells.
(I) Schematic of the experimental strategy (same as Figure 1B). Anisomycin (ANI) was injected immedialtely after training and mice were perfused 24 hr later.
(J) DG granule cell ChR2-EYFP+ (green) stained with a Kir2.1 antibody (red).
(K) Confocal images of DG granule cells ChR2-EYFP� (double-arrowhead) and ChR2-EYFP+ (single-arrowhead) stained for Kir2.1 (red).
(L) Paired analysis of Kir2.1 expression, ChR2-EYFP� vs ChR2-EYFP+ (mean in red, two-tailed paired t test, **p < 0.01).
behaviors. We first examined context discrimination by training
mice in a CFC paradigm in context A. We assessed freezing in
a similar context, AB, in which the prominent visual cues of
context A were retained but other cues including odor, lighting,
floor material, cage shape, and experimental room were derived
from a distinct context B (Figure 3A). On day 2, mice were either
directly exposed to context AB (no recall, NR), or were exposed
to context AB 5 min (5m) or 3 hr (3h) after the exposure to
context A. In context A, the 5m and 3h groups froze at similar
levels (Figure 3B). In context AB, the NR and the 3h group
showed robust and similar levels of freezing; however, the 5m
group displayed negligible freezing (Figure 3B), a result also
confirmed in a 1 hr group (Figures S5A and S5B). On day 3, all
three groups showed robust and similar levels of freezing to
context AB. On day 4, all groups showed similar freezing levels
in context A, and the slightly lower levels of freezing displayed
by the 5m and 3h groups can be attributed to fear extinction
(Quirk and Mueller, 2008) due to repeated exposures to
Neuron 101, 274–284, January 16, 2019 277
Figure 3. Behavioral Correlates of High-
Excitability State of Engram Cells
(A) Schematic: protocols to test pattern separation
with a pre-exposure to the conditioned context. On
day 1, mice were fear conditioned in context A
(red). On day 2, a NR group (n = 8) was exposed to
context AB (white), and a 5m group (n = 8) and a 3h
group (n = 8) were exposed to context A, removed,
and exposed to context AB after 5 min or 3 hr,
respectively. On day 3, mice were exposed to
context AB to test discrimination. On day 4, mice
were exposed to context A to test memory recall.
Contexts A and AB were characterized by the
same visual cues (see STAR Methods), different
odor cues, different floors, and different experi-
mental rooms.
(B) Freezing score in response to exposure to
contexts A (red) and AB (white) on days 2, 3, and 4.
Significance assessed by one-way ANOVA with
Tukey’s test for multiple comparisons: *p < 0.05,
***p < 0.001).
(C) Schematic: protocols for pattern completion.
On day 1, mice were exposed to context A for
10 min to form a contextual memory. On day 2, a
NR group was subjected to an immediate shock
test in context A while a 5m and a 3h group were
subjected to a 3-min recall session in context A
5 min or 3 hr before the immediate shock experi-
ment, respectively. Each of the NR, 5m, and 3h
groups was further divided into two subgroups
(n = 10) of mice and a 2 s shock of 1.5 mA was
administered with different placement-to-shock-intervals (PSI: 5 s or 10 s). On day 3, mice were re-exposed to context A to test memory recall.
(D) Freezing score in response to re-exposure to context A on day 3 (significance assessed by one-way ANOVA with Tukey’s test for multiple comparisons:
*p<0.05, **p<0.01).
(E) Freezing score as a function of the PSI for the NR (black), 5m (red), and 3h (gray) groups. Significance assessed by two-way ANOVA with Sidak correction for
multiple comparison: df = 56, Frow(1, 22) = 20.42, Ftime(2, 36) = 14.28 (*p < 0.05).
Data are displayed as mean ± SEM.
context A. Thus, we have shown that a transient increase of the
engram cell excitability caused by a re-exposure to the condi-
tioned context correlates with an increase in the animal’s ability
to discriminate between the conditioned context and its
degraded variant (pattern separation). The recall-induced
decrease in freezing in the 5m group only occurred in degraded
context AB, not in the intact context A (Figures S6A and S6B).
Exposure to a novel context C did not attenuate freezing to
context AB 5 min later (Figures S6C and S6D), demonstrating
the context specificity of this effect and ruling out any non-mne-
monic explanation such as arousal.
We then investigated the effect of the prior exposure to the
conditioned context on the strength of subsequent pattern-
completion-based recall. For this purpose, we modified the
pre-exposure-dependent CFC paradigm (Fanselow, 1986,
1990). On Day 1, we exposed all mice to context A for 10 min
without shock to allow the formation of a saturated level of
contextual memory (Figure 3C). On Day 2, mice were subjected
to either no recall (NR group) or recall (5m and 3h groups) of
context A memory with a 3-min-long re-exposure to context A.
Then, half of each group was subjected to a 5 s and the other
half to a 10 s ‘‘immediate shock’’ procedure (Fanselow, 1990).
This step occurred either 5min (5m group) or 3 hr (3h group) after
memory recall in context A (Figure 3C). On day 3, all groups of
mice subjected to 10 s PSI immediate shock conditioning dis-
278 Neuron 101, 274–284, January 16, 2019
played a general increase in freezing compared to the groups
subjected to 5 s PSI conditioning (Figure 3D), revealing an
enhanced pattern completion. However, the 5m group displayed
significantly more freezing than NR and 3h groups, which dis-
played similar levels of freezing (Figures 3D and 3E). This result
was also confirmed in a 1 hr group (Figures S5C and S5D). These
data suggest that the transient increase of excitability of context
A memory engram cells correlates with a greater pattern
completion capability of conditioned stimuli.
Causal Link between Engram Cell High-ExcitabilityState and Enhanced RecognitionTo test whether the high-excitability state of engram cells is
causally linked with the enhanced context recognition capabil-
ities, we expressed exogenous Kir2.1 ion channels in DG engram
cells formed during CFC in context A using an AAV9-TRE-Kir2.1-
RFP virus (Figure 4A). When tested 5min after recall in context A,
Kir2.1-RFP+ engram cells showed reduced input resistance and
increased rheobase relative to Kir2.1-RFP� non-engram cells.
This demonstrated the effect of the exogenous Kir2.1 channel
expression in preventing the excitability increase of DG engram
cells by recall (Figures 4B and 4C). The expression of Kir2.1 in DG
engram cells during CFC had no effect on encoding, memory
consolidation, reconsolidation, maintenance, or further learning
(Figures S7A–S7C). However, Kir2.1 expression in DG engram
Figure 4. Reversion of Behavior by Exogenous Expression of Kir2.1 in Engram Cells
(A) Experimental procedure: AAV9-TRE-Kir2.1-RFP injected into the DG of c-Fos tTA mice to express Kir2.1-RFP in engram cells. Bottom: confocal images of the
hippocampus of a mouse on a DOX diet (right) and off DOX (left). Right: quantification of Kir2.1-RFP+ cell number (two-tailed unpaired test, **p < 0.01).
(B) Confocal images of biocytin-filled DG granule cells. Insets: top, Kir2.1-RFP� cells; bottom, Kir2.1-RFP
+ cells.
(C) Intrinsic electrophysiological properties of Kir2.1-RFP� cells (black, n = 17 cells) and Kir2.1-RFP
+ cells (red, n = 13 cells) in response to current injection (gray)
frommice sacrificed 5 min after re-exposure to context A. Bottom: membrane resistance (left) and rheobase (right) from Kir2.1-RFP� non-engram and Kir2.1-RFP
+
engram cells (two-tailed unpaired t test, **p < 0.01).
(D) Top: protocol used to test the effect of the exogenous Kir2.1 expression on the discrimination of recall cues. Day 0: virus-injected c-fos-tTAmice were taken off
DOX. Day 1: RFP and Kir2.1 groups of mice were subjected to CFC in context A (red) to label DG engram cells with RFP (n = 16 mice) and Kir2.1-RFP (n = 22 mice),
respectively, and then placed on DOX. Day 2: mice were exposed to context A for a 3-min recall session. 5 min post-recall in context A, mice were exposed to
context AB for 3 min and freezing score was assessed. Day 3: mice were exposed to context AB for 3 min. Day 4: mice were exposed to context A for 3 min.
Bottom: comparison between Kir2.1-RFP and RFP groups.
(E) Top: protocol used to test the effect of the exogenous Kir2.1 expression on pattern-completion-mediated recall. Day 0: virus-injected c-fos-tTA mice were
taken off DOX. Day 1: RFP and Kir2.1 groups were exposed to context A for 10 min (no shock) to label DG engram cells (RFP, n = 12 mice and Kir2.1, n = 13 mice),
then placed on DOX. Day 2: mice were exposed to context A for a 3-min recall. 5 min post-recall mice were subjected to a 10-s-long PSI conditioning in context A.
Day 3: mice were given a 5-min recall test in context A. Bottom: comparison between Kir2.1-RFP and RFP groups.
Freezing scores are displayed as mean ± SEM and the statistical significance is assessed by two-tailed unpaired t test (**p < 0.01, ***p < 0.001).
Neuron 101, 274–284, January 16, 2019 279
Figure 5. Engram Cell States of Excitability and Their Role in Memory Retrieval
(A) Summary schematic showing the plasticity of engram cell excitability in response to memory recall (timeline on the left). Engram cells are labeled by exposure
to context A (black cells). When a mouse is kept in the home cage (gray box), engram cells display a low-excitability state comparable to non-engram cells (gray
cells). A brief re-exposure to context A (red box) evokes the activation of the perforant path (red fibers) (Kitamura et al., 2015; Tamamaki and Nojyo, 1993), which in
turn activates engram cells (note the gradient scale bar from light red to bright red, representing the intensity of the perforant path input). The synaptic activation of
engramcells triggers a state of high excitability (green cells: note the gradient scale bar from green to black, representing the excitability state). When themouse is
placed back in the home cage (gray box), synaptic inputs cease while excitability persists for an hour.
(B) Schematic showing the differential effect of the engram cell excitability states onmemory retrieval. In a low-excitability state (black), retrieval is highly effective
only in presence of full cues. Exposure to partial cues results in diminished completion due to partial retrieval. Exposure to conflicting cues results in diminished
discrimination due to a partial retrieval. In a high-excitability state (green), retrieval is highly effective not only in presence of full cues but also with partial cues and
conflicting cues, resulting in enhanced completion and separation, respectively.
cells specifically abolished the recall-induced enhancement of
pattern separation as well as the augmented pattern completion
(Figures 4D and 4E). This effect was context-specific, since
exogenous expression of Kir2.1 channels in engram cells specific
to a different context C did not abolish the recall-induced
enhancement of pattern separation as well as the augmented
pattern completion related to context A (Figures S7D and S7E).
DISCUSSION
In this study, we investigated the memory retrieval process using
contextual fear conditioning as amodel of episodic-like memory.
We identified a previously unknown engram cell state defined by
heightened intrinsic excitability and induced rapidly and specif-
ically in engram cells by re-exposure to conditioned stimuli.
This rapid increase in DG engram cell excitability is mediated
by an internalization of Kir2.1 inward-rectifier potassium chan-
nels. The high-excitability state is transient and reverts to the
baseline state within 2 hr through a process that requires protein
synthesis. This process ensures the reset of engram cell excit-
ability to a steady state. We found that this transient window of
high engram cell excitability was causally linked with a short-
term increase of the animal’s recognition accuracy for relevant
contextual configurations, discriminating them from similar but
irrelevant ones. Furthermore the same transient high-excitability
280 Neuron 101, 274–284, January 16, 2019
state causally enhanced the strength of cue recognition, which
manifested as a greater capacity for pattern-completion-medi-
ated recall.
Memory Retrieval Is Optimized by a Change in EngramCell Excitability StateOur study revealed that recall of an episodic memory consists of
two consecutive functional states of engram cells. In the first
state, which corresponds to the quiescent period, the excitability
of the engram cells is at a baseline level, and exposure to condi-
tioned stimuli activates them, causing neural transmission and a
consequent behavioral reaction. The reactivation of the engram
cells triggers a second state by increasing their cellular excit-
ability for 1 hr (Figure 5A). In the first engram cell state, the base-
line excitability level does not allow fully accurate and efficient
recognition of the context in the presence of conflicting cues or
when cues are partially available (Figure 5B). In contrast, the sec-
ond state enhances the engram cell response to partial cues,
augmenting the capability of context recognition (Figure 5B).
Mechanism Controlling the Engram CellExcitability StateRecall cues selectively reactivate engram cells supported by the
neuronal connectivity pattern established during learning (Poo
et al., 2016; Ryan et al., 2015). Our results suggest that the
synaptic activation of NMDA receptors triggers a cascade of
events causing the internalization of Kir2.1 channels and resulting
in an increase of engram cell excitability. Several leaky conduc-
tances can contribute to the regulation of the excitability state,
and DG granule cells are influenced by many of these conduc-
tances (Aller and Wisden, 2008; Karschin et al., 1996; Karschin
et al., 1997; Young et al., 2009). The dominant role of Kir2.1 ion
channels, the correlation of their expression with DG granule
cell excitability, and their association with the engram cell excit-
ability state suggests a deterministic role of these channels in
regulating the excitability of engram cells. Kir2.1-2.3 subunits
share several trafficking domains involved in ER-export
(Ma et al., 2001), Golgi-export (Stockklausner and Klocker,
2003) and endocytosis (Tong et al., 2001), which control surface
expression. Kir2.1 ion channels can be internalized through a
Rac1-mediated signaling pathway and inhibition of the endoge-
nous GTPase Rac1 can reduce surface expression of Kir2.1(Boyer et al., 2009). Consistent with an internalization-based
mechanism, our confocal analysis confirmed a reduced surface
expression of Kir2.1 ion channels in highly excitable engram cells.
Furthermore, our study demonstrated a homeostatic role of pro-
tein synthesis in returning the membrane excitability back to a
baseline level, suggesting a mechanism based on the de novo
synthesis of new Kir2.1 ion channels. Phosphorylation is also
known to modulate Kir2.1 conductances (Wischmeyer et al.,
1998), a mechanism which could synergistically contribute to
alter the state of excitability of engram cells.
Mechanisms Underlying Recall-Induced PatternSeparation and Pattern CompletionPattern separation is crucial for the formation of specific mem-
ories. The DG has long been attributed to this process on ac-
count of its anatomical structure. The strong signals provided
by DG granule cells during encoding could enhance the distinc-
tion betweenmemories (O’Reilly andMcClelland, 1994). The role
of DG granule cells in pattern separation has been proposed to
be based on their relatively high cell number, their lack of recur-
rent collaterals, and their sparse connectivity with the down-
stream CA3 pyramidal cells (Bernier et al., 2017; Danielson
et al., 2016; Kheirbek et al., 2013; Leutgeb et al., 2007; McHugh
et al., 2007; McNaughton and Morris, 1987; O’Reilly and
McClelland, 1994). In this study, we have shown that an animal’s
ability to discriminate environmental cues can be qualitatively
enhanced following retrieval by raising the excitability of adult
DG engram cells. This result reveals a new mnemonic phenom-
enon and atttributed mechanisms of engram-specific pattern
separation that occur following memory recall and enhance
discrimination. Post-natally generated young granule cells
have also been shown to play a crucial role in pattern separation
(Clelland et al., 2009; Nakashiba et al., 2012), and it has been hy-
pothesized that their hyper-excitability may contribute to
enhance the signal-to-noise ratio by recruiting interneurons (Ai-
mone et al., 2011; Poo et al., 2016). Similar circuit mechanisms
may play a role in the recall-induced pattern separation medi-
ating discrimination of recall cues.
Pattern completion allows recall with a limited set of relevant
cues and it has been attributed to highly plastic recurrent net-
works such as the one present in the hippocampal CA3 field
(Marr, 1971; Nakazawa et al., 2002). The present study reveals
a newmechanism that operates at an individual cell level, namely
a heightened excitability of DG engram cells. The excitability
increase in an engram cell is likely to compensate for the reduc-
tion of synaptic inputs due to limited cue availability. These
network and individual cell-based mechanisms could operate
synergistically to cope with situations that demand fast and effi-
cient memory retrieval.
The enhancement of pattern separation and pattern comple-
tion by the shared mechanism of increased engram cell excit-
ability is consistent with, and extends, current unifying models
of these dual computational processes (Hainmueller and Bartos,
2018; Neunuebel and Knierim, 2014; Tanaka et al., 2018; Yassa
and Stark, 2011). When DG engram cells are highly excitable,
they are likely be more able to separate signals arriving from
the entorhinal cortex through the perforant path, as they will be
more responsive than neighboring non-engram cells to glutama-
tergic transmission. At the same time, highly excitable DG
engramcells will promote completion by enhancing transmission
to CA3 engram cells through the mossy fiber pathway. In this
manner, the processes of pattern separation and completion
are enabled by hippocampal anatomy and determined by the
nature of the perceptual input, but both are significantly influ-
enced in quality and magnitude by the plasticity of engram cell
excitability.
Engram Cell Excitability State in Relation toReconsolidation and ExtinctionWhile recall-induced enhancement of engram excitability im-
proves memory access in the short term, it does not perma-
nently alter the strength of memory recall to either full or partial
cues. Other studies have shown that brief recall trials can
permanently strengthen the conditioned fear response to full
contextual cues (Inda et al., 2011; Khalaf et al., 2018; Kim
et al., 2014). We did not observe any long-term strengthening
in context response due to recall under our experimental con-
ditions, but our results suggest that these reported strength-
ening processes are mediated by mechanisms other than
changes in intrinsic excitability. Nevertheless, because our
pattern completion experiments were based on an immediate
shock updating paradigm (Fanselow, 1986, 1990), it is clear
that heightened engram excitability provides the opportunity
for enhanced learning of new information to be enduringly
associated with a pre-existing engram. We speculate that this
enhanced updating may be mediated by memory reconsolida-
tion (Nader et al., 2000), but future experimental studies will be
required to determine whether heightened engram excitability
is a mechanistic prerequisite for memory reconsolidation. It is
noteworthy that heightened engram excitability does not lead
to any apparent enhancement of memory extinction, since
both the 5 min and 3 hr groups show a similar reduction in
freezing following exposure to the conditioned stimulus in the
absence of the unconditioned stimulus.
Survival ValueThe transient optimization of recognition capabilities in
response to recall cues is reminiscent of the process of facilita-
tion taking place in the motor system (Del Castillo and Katz,
Neuron 101, 274–284, January 16, 2019 281
1953) or sensitization for the sensory system (Castellucci and
Kandel, 1976). Indeed, the short-term enhancement of engram
cell excitability due to engram cell reactivation increases the
future recognition capability of specific cues. The result is a
new form of short-term memory based on the same long-
term engram. Thus, as in the cases of facilitation and sensitiza-
tion, the engram cell state of high excitability will have a strong
survival value in cases where the releasing stimulus can be ex-
pected to recur (Lorenz, 1973). By reversibly altering the effi-
cacy of memory retrieval according to the recent history of
the animal’s experience, the two engram cell states could
determine behavioral outcomes that are appropriate for imme-
diate environmental circumstances. Thus, plasticity of engram
cell excitability may be crucial for survival by facilitating rapid
adaptive behavior without permanently altering the funda-
mental nature of the long-term engram.
STAR+METHODS
Detailed methods are provided in the online version of this paper
and include the following:
d KEY RESOURCES TABLE
d CONTACT FOR REAGENT AND RESOURCE SHARING
d EXPERIMENTAL MODEL AND SUBJECT DETAILS
d METHOD DETAILS
B Engram labeling strategy
B Virus-mediated gene expression
B Stereotactic injection
B Behavior
B Histological verification
B Ex vivo patch clamp recordings
B Ex vivo electrophysiology
B Ion channels immunostaining
d QUANTIFICATION AND STATISTICAL ANALYSIS
d DATA AND SOFTWARE AVAILABILITY
SUPPLEMENTAL INFORMATION
Supplemental Information includes seven figures and can be found with this
article online at https://doi.org/10.1016/j.neuron.2018.11.029.
ACKNOWLEDGEMENTS
We thank A. Arons, F. Bushard, A. Hamalian, S.Y. Huang, D. King, and C. Twiss
for help with experiments; A. Beyeler, A. Bari, J. Kim, T. Kitamura, C. MacDon-
ald, and M. Morrisey for comments on the manuscript; L. Brenner for proof-
reading; and all members of the Tonegawa laboratory for their support. This
work was supported by the RIKEN Brain Science Institute, the Howard Hughes
Medical Institute, and the JPB Foundation (S.T.).
AUTHOR CONTRIBUTIONS
Conceptualization:M.P., T.J.R., and S.T.; Investigation: M.P. and T.J.R.; Phys-
iology: M.P., S.M., D.J.R., and T.J.R.; Behavior: T.J.R. and M.P.; Vector
design: T.J.R., Stereotactic surgery: L.M.S.; Histology: M.P., T.J.R., and
C.L.; Confocal analysis: M.P.; Data analysis: M.P. and T.R.; Data interpretation
and manuscript writing: M.P., T.J.R., and S.T.
DECLARATION OF INTERESTS
The authors declare no competing interests.
282 Neuron 101, 274–284, January 16, 2019
Received: March 12, 2018
Revised: October 12, 2018
Accepted: November 15, 2018
Published: December 11, 2018
REFERENCES
Aimone, J.B., Deng, W., and Gage, F.H. (2011). Resolving new memories: a
critical look at the dentate gyrus, adult neurogenesis, and pattern separation.
Neuron 70, 589–596.
Aizenman, C.D., and Linden, D.J. (2000). Rapid, synaptically driven increases
in the intrinsic excitability of cerebellar deep nuclear neurons. Nat. Neurosci. 3,
109–111.
Aller, M.I., and Wisden, W. (2008). Changes in expression of some two-pore
domain potassium channel genes (KCNK) in selected brain regions of devel-
oping mice. Neuroscience 151, 1154–1172.
Bernier, B.E., Lacagnina, A.F., Ayoub, A., Shue, F., Zemelman, B.V., Krasne,
F.B., and Drew, M.R. (2017). Dentate Gyrus Contributes to Retrieval as well
as Encoding: Evidence from Context Fear Conditioning, Recall, and
Extinction. J. Neurosci. 37, 6359–6371.
Boyer, S.B., Slesinger, P.A., and Jones, S.V. (2009). Regulation of Kir2.1 chan-
nels by the Rho-GTPase, Rac1. J. Cell. Physiol. 218, 385–393.
Brons, J.F., and Woody, C.D. (1980). Long-term changes in excitability of
cortical neurons after Pavlovian conditioning and extinction. J. Neurophysiol.
44, 605–615.
Cai, D.J., Aharoni, D., Shuman, T., Shobe, J., Biane, J., Song, W., Wei, B.,
Veshkini, M., La-Vu, M., Lou, J., et al. (2016). A shared neural ensemble links
distinct contextual memories encoded close in time. Nature 534, 115–118.
Castellucci, V., and Kandel, E.R. (1976). Presynaptic facilitation as a mecha-
nism for behavioral sensitization in Aplysia. Science 194, 1176–1178.
Choi, J.H., Sim, S.E., Kim, J.I., Choi, D.I., Oh, J., Ye, S., Lee, J., Kim, T., Ko,
H.G., Lim, C.S., and Kaang, B.K. (2018). Interregional synaptic maps among
engram cells underlie memory formation. Science 360, 430–435.
Clelland, C.D., Choi, M., Romberg, C., Clemenson, G.D., Jr., Fragniere, A.,
Tyers, P., Jessberger, S., Saksida, L.M., Barker, R.A., Gage, F.H., and
Bussey, T.J. (2009). A functional role for adult hippocampal neurogenesis in
spatial pattern separation. Science 325, 210–213.
Danielson, N.B., Kaifosh, P., Zaremba, J.D., Lovett-Barron,M., Tsai, J., Denny,
C.A., Balough, E.M., Goldberg, A.R., Drew, L.J., Hen, R., et al. (2016). Distinct
Contribution of Adult-Born Hippocampal Granule Cells to Context Encoding.
Neuron 90, 101–112.
Debanne, D., and Poo, M.M. (2010). Spike-timing dependent plasticity beyond
synapse - pre- and post-synaptic plasticity of intrinsic neuronal excitability.
Front. Synaptic Neurosci. 2, 21.
Del Castillo, J., and Katz, B. (1953). Statistical nature of facilitation at a single
nerve-muscle junction. Nature 171, 1016–1017.
Deng, W., Mayford, M., and Gage, F.H. (2013). Selection of distinct popula-
tions of dentate granule cells in response to inputs as a mechanism for pattern
separation in mice. eLife 2, e00312.
Denny, C.A., Kheirbek, M.A., Alba, E.L., Tanaka, K.F., Brachman, R.A.,
Laughman, K.B., Tomm, N.K., Turi, G.F., Losonczy, A., and Hen, R. (2014).
Hippocampal memory traces are differentially modulated by experience,
time, and adult neurogenesis. Neuron 83, 189–201.
Disterhoft, J.F., and Oh, M.M. (2006). Learning, aging and intrinsic neuronal
plasticity. Trends Neurosci. 29, 587–599.
Disterhoft, J.F., Coulter, D.A., and Alkon, D.L. (1986). Conditioning-specific
membrane changes of rabbit hippocampal neurons measured in vitro. Proc.
Natl. Acad. Sci. USA 83, 2733–2737.
Fan, Y., Fricker, D., Brager, D.H., Chen, X., Lu, H.-C., Chitwood, R.A., and
Johnston, D. (2005). Activity-dependent decrease of excitability in rat hippo-
campal neurons through increases in I(h). Nat. Neurosci. 8, 1542–1551.
Fanselow, M.S. (1986). Associative vs topographical accounts of the immedi-
ate shock-freezing deficit in rats: Implications for the response selection rules
governing species-specific defensive reactions. Learn. Motiv. 17, 16–39.
Fanselow, M.S. (1990). Factors governing one-trial contextual conditioning.
Animal Learning &. Behaviour 18, 264–270.
Frick, A., Magee, J., and Johnston, D. (2004). LTP is accompanied by an
enhanced local excitability of pyramidal neuron dendrites. Nat. Neurosci. 7,
126–135.
Ganguly, K., Kiss, L., and Poo, M. (2000). Enhancement of presynaptic
neuronal excitability by correlated presynaptic and postsynaptic spiking.
Nat. Neurosci. 3, 1018–1026.
Guzowski, J.F., McNaughton, B.L., Barnes, C.A., and Worley, P.F. (1999).
Environment-specific expression of the immediate-early gene Arc in hippo-
campal neuronal ensembles. Nat. Neurosci. 2, 1120–1124.
Hainmueller, T., and Bartos, M. (2018). Parallel emergence of stable and dy-
namic memory engrams in the hippocampus. Nature 558, 292–296.
Han, J.-H., Kushner, S.A., Yiu, A.P., Hsiang, H.-L.L., Buch, T., Waisman, A.,
Bontempi, B., Neve, R.L., Frankland, P.W., and Josselyn, S.A. (2009).
Selective erasure of a fear memory. Science 323, 1492–1496.
Hille, B. (1992). Ionic Channels of Excitable Membranes, Second Edition
(Sunderland, MA: Sinauer Associates Inc).
Inda, M.C., Muravieva, E.V., and Alberini, C.M. (2011). Memory retrieval and
the passage of time: from reconsolidation and strengthening to extinction.
J. Neurosci. 31, 1635–1643.
Josselyn, S.A., and Frankland, P.W. (2018). Memory Allocation: Mechanisms
and Function. Annu. Rev. Neurosci. 41, 389–413.
Josselyn, S.A., Kohler, S., and Frankland, P.W. (2015). Finding the engram.
Nat. Rev. Neurosci. 16, 521–534.
Karschin, C., Dissmann, E., St€uhmer, W., and Karschin, A. (1996). IRK(1-3) and
GIRK(1-4) inwardly rectifying K+ channelmRNAs are differentially expressed in
the adult rat brain. J. Neurosci. 16, 3559–3570.
Karschin, C., Ecke, C., Ashcroft, F.M., and Karschin, A. (1997). Overlapping
distribution of K(ATP) channel-forming Kir6.2 subunit and the sulfonylurea re-
ceptor SUR1 in rodent brain. FEBS Lett. 401, 59–64.
Khalaf, O., Resch, S., Dixsaut, L., Gorden, V., Glauser, L., and Gr€aff, J. (2018).
Reactivation of recall-induced neurons contributes to remote fear memory
attenuation. Science 360, 1239–1242.
Kheirbek, M.A., Drew, L.J., Burghardt, N.S., Costantini, D.O., Tannenholz, L.,
Ahmari, S.E., Zeng, H., Fenton, A.A., and Hen, R. (2013). Differential control of
learning and anxiety along the dorsoventral axis of the dentate gyrus. Neuron
77, 955–968.
Kim, J., Kwon, J.T., Kim, H.S., Josselyn, S.A., and Han, J.H. (2014). Memory
recall and modifications by activating neurons with elevated CREB. Nat.
Neurosci. 17, 65–72.
Kitamura, T., Sun, C., Martin, J., Kitch, L.J., Schnitzer, M.J., and Tonegawa, S.
(2015). Entorhinal Cortical Ocean Cells Encode Specific Contexts and Drive
Context-Specific Fear Memory. Neuron 87, 1317–1331.
Leutgeb, J.K., Leutgeb, S., Moser, M.-B., and Moser, E.I. (2007). Pattern sep-
aration in the dentate gyrus and CA3 of the hippocampus. Science 315,
961–966.
Li, C.Y., Lu, J.T., Wu, C.P., Duan, S.M., and Poo, M.M. (2004). Bidirectional
modification of presynaptic neuronal excitability accompanying spike timing-
dependent synaptic plasticity. Neuron 41, 257–268.
Lisman, J., Cooper, K., Sehgal, M., and Silva, A.J. (2018). Memory formation
depends on both synapse-specific modifications of synaptic strength and
cell-specific increases in excitability. Nat. Neurosci. 21, 309–314.
Liu, X., Ramirez, S., Pang, P.T., Puryear, C.B., Govindarajan, A., Deisseroth,
K., and Tonegawa, S. (2012). Optogenetic stimulation of a hippocampal
engram activates fear memory recall. Nature 484, 381–385.
Lorenz, K. (1973). Behind the mirror (Harvest/HBJ Book).
Ma, D., Zerangue, N., Lin, Y.F., Collins, A., Yu, M., Jan, Y.N., and Jan, L.Y.
(2001). Role of ER export signals in controlling surface potassium channel
numbers. Science 291, 316–319.
Marr, D. (1971). Simplememory: a theory for archicortex. Philos. Trans. R. Soc.
Lond. B Biol. Sci. 262, 23–81.
McHugh, T.J., Jones, M.W., Quinn, J.J., Balthasar, N., Coppari, R., Elmquist,
J.K., Lowell, B.B., Fanselow, M.S., Wilson, M.A., and Tonegawa, S. (2007).
Dentate gyrus NMDA receptors mediate rapid pattern separation in the hippo-
campal network. Science 317, 94–99.
McKay, B.M., Matthews, E.A., Oliveira, F.A., and Disterhoft, J.F. (2009).
Intrinsic neuronal excitability is reversibly altered by a single experience in
fear conditioning. J. Neurophysiol. 102, 2763–2770.
McNaughton, B.L., and Morris, R.G.M. (1987). Hippocampal synaptic
enhancement and information storage within a distributed memory system.
Trends Neurosci. 10, 408–415.
Mozzachiodi, R., Lorenzetti, F.D., Baxter, D.A., and Byrne, J.H. (2008).
Changes in neuronal excitability serve as a mechanism of long-term memory
for operant conditioning. Nat. Neurosci. 11, 1146–1148.
Nader, K., Schafe, G.E., and Le Doux, J.E. (2000). Fear memories require pro-
tein synthesis in the amygdala for reconsolidation after retrieval. Nature 406,
722–726.
Nakashiba, T., Cushman, J.D., Pelkey, K.A., Renaudineau, S., Buhl, D.L.,
McHugh, T.J., Rodriguez Barrera, V., Chittajallu, R., Iwamoto, K.S., McBain,
C.J., et al. (2012). Young dentate granule cells mediate pattern separation,
whereas old granule cells facilitate pattern completion. Cell 149, 188–201.
Nakazawa, K., Quirk, M.C., Chitwood, R.A., Watanabe, M., Yeckel, M.F., Sun,
L.D., Kato, A., Carr, C.A., Johnston, D., Wilson, M.A., and Tonegawa, S. (2002).
Requirement for hippocampal CA3 NMDA receptors in associative memory
recall. Science 297, 211–218.
Neunuebel, J.P., and Knierim, J.J. (2014). CA3 retrieves coherent representa-
tions from degraded input: direct evidence for CA3 pattern completion and
dentate gyrus pattern separation. Neuron 81, 416–427.
O’Reilly, R.C., and McClelland, J.L. (1994). Hippocampal conjunctive encod-
ing, storage, and recall: avoiding a trade-off. Hippocampus 4, 661–682.
Passano, M.L. (1957). Prey-predator recognition in the lower vertebrates
(Eugene: University of Oregon Press).
Poo, M.M., Pignatelli, M., Ryan, T.J., Tonegawa, S., Bonhoeffer, T., Martin,
K.C., Rudenko, A., Tsai, L.-H., Tsien, R.W., Fishell, G., et al. (2016). What is
memory? The present state of the engram. BMC Biol. 14, 40.
Quirk, G.J., and Mueller, D. (2008). Neural mechanisms of extinction learning
and retrieval. Neuropsychopharmacology 33, 56–72.
Ramirez, S., Liu, X., Lin, P.-A., Suh, J., Pignatelli, M., Redondo, R.L., Ryan,
T.J., and Tonegawa, S. (2013). Creating a false memory in the hippocampus.
Science 341, 387–391.
Rashid, A.J., Yan, C., Mercaldo, V., Hsiang, H.-L.L., Park, S., Cole, C.J., De
Cristofaro, A., Yu, J., Ramakrishnan, C., Lee, S.Y., et al. (2016). Competition
between engrams influences fear memory formation and recall. Science
353, 383–387.
Redondo, R.L., Kim, J., Arons, A.L., Ramirez, S., Liu, X., and Tonegawa, S.
(2014). Bidirectional switch of the valence associated with a hippocampal
contextual memory engram. Nature 513, 426–430.
Reijmers, L.G., Perkins, B.L., Matsuo, N., and Mayford, M. (2007). Localization
of a stable neural correlate of associative memory. Science 317, 1230–1233.
Riccio, D.C., Richardson, R., and Ebner, D.L. (1984). Memory retrieval deficits
based upon altered contextual cues: a paradox. Psychol. Bull. 96, 152–165.
Roy, D.S., Muralidhar, S., Smith, L.M., and Tonegawa, S. (2017). Silent mem-
ory engrams as the basis for retrograde amnesia. Proc. Natl. Acad. Sci. USA
114, E9972–E9979.
Ryan, T.J., Roy, D.S., Pignatelli, M., Arons, A., and Tonegawa, S. (2015).
Memory. Engram cells retain memory under retrograde amnesia. Science
348, 1007–1013.
Neuron 101, 274–284, January 16, 2019 283
Schrader, L.A., Anderson, A.E., Varga, A.W., Levy, M., and Sweatt, J.D. (2002).
The other half of Hebb: K+ channels and the regulation of neuronal excitability
in the hippocampus. Mol. Neurobiol. 25, 51–66.
Semon, R.W. (1921). The mneme (George Allen).
Stockklausner, C., and Klocker, N. (2003). Surface expression of inward recti-
fier potassium channels is controlled by selective Golgi export. J. Biol. Chem.
278, 17000–17005.
Tamamaki, N., and Nojyo, Y. (1993). Projection of the entorhinal layer II neu-
rons in the rat as revealed by intracellular pressure-injection of neurobiotin.
Hippocampus 3, 471–480.
Tanaka, K.Z., He, H., Tomar, A., Niisato, K., Huang, A.J.Y., and McHugh, T.J.
(2018). The hippocampal engram maps experience but not place. Science
361, 392–397.
Tonegawa, S., Liu, X., Ramirez, S., and Redondo, R. (2015a). Memory Engram
Cells Have Come of Age. Neuron 87, 918–931.
Tonegawa, S., Pignatelli, M., Roy, D.S., and Ryan, T.J. (2015b). Memory
engram storage and retrieval. Curr. Opin. Neurobiol. 35, 101–109.
Tong, Y., Brandt, G.S., Li, M., Shapovalov, G., Slimko, E., Karschin, A.,
Dougherty, D.A., and Lester, H.A. (2001). Tyrosine decaging leads to substan-
284 Neuron 101, 274–284, January 16, 2019
tial membrane trafficking during modulation of an inward rectifier potassium
channel. J. Gen. Physiol. 117, 103–118.
Tulving, E. (1983). Elements of Episodic Memory (Oxford University Press).
Wischmeyer, E., Doring, F., and Karschin, A. (1998). Acute suppression of
inwardly rectifying Kir2.1 channels by direct tyrosine kinase phosphorylation.
J. Biol. Chem. 273, 34063–34068.
Yassa, M.A., and Stark, C.E. (2011). Pattern separation in the hippocampus.
Trends Neurosci. 34, 515–525.
Yiu, A.P., Mercaldo, V., Yan, C., Richards, B., Rashid, A.J., Hsiang, H.-L.L.,
Pressey, J., Mahadevan, V., Tran, M.M., Kushner, S.A., et al. (2014).
Neurons are recruited to amemory trace based on relative neuronal excitability
immediately before training. Neuron 83, 722–735.
Young, C.C., Stegen, M., Bernard, R., M€uller, M., Bischofberger, J., Veh, R.W.,
Haas, C.A., and Wolfart, J. (2009). Upregulation of inward rectifier K+ (Kir2)
channels in dentate gyrus granule cells in temporal lobe epilepsy. J. Physiol.
587, 4213–4233.
Zhang, W., and Linden, D.J. (2003). The other side of the engram: experience-
driven changes in neuronal intrinsic excitability. Nat. Rev. Neurosci. 4,
885–900.
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
Anti-cFos Synaptic System Cat# 226-003
Anti-GFP AbCam Cat# ab13970
Anti-RFP Rockland Cat# 600-901-379
CF555 Streptavidin Biotium Cat# 29038
Anti-GFP Invitrogen Cat# A10262
Anti-Kir2.1 Neuromab Cat# 75-210
Anti-PAN Sigma Cat# S-8809
fAB fragment Jackson Immunoresearch Cat# 115-007-003
Bacterial and Virus Strains
AAV9-TRE-ChR2-EYFP Liu et al. 2012 N/A
AAV9-TRE-RFP Ramirez et al. 2013 N/A
AAV9-TRE-Kir2.1-RFP This article N/A
Biological Samples
Chemicals, Peptides, and Recombinant Proteins
Biocytin Sigma Cat# B4261
CPP Tocris Cat# 0247
Anisomycin Tocris Cat# 1290/10
ML133 Sigma Cat# SML0190
CsCl Sigma Cat# C3011
ZD7288 Tocris Cat# 1000/10
Tertiapin LQ Tocris Cat# 4339/1
Quinine Tocris Cat# 4114/50
Glibenclamide Tocris Cat# 0911/100
Experimental Models: Organisms/Strains
Mouse: cfos-tTA/cfos-shEGFP Jackson https://www.jax.org/strain/018306
Software and Algorithms
ImageJ-Fiji NIH https://fiji.sc/
Excel Microsoft https://products.office.com/en-us/excel
GraphPad Prism https://www.graphpad.com
IgorPro7 Wavemetrics https://www.wavemetrics.com
CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Susumu
Tonegawa ([email protected]).
EXPERIMENTAL MODEL AND SUBJECT DETAILS
All experiments were conducted in accordance with National Institutes of Health (NIH) guidelines, the Massachusetts Institute of
Technology (MIT) Department of Comparative Medicine and MIT Committee of Animal Care. c-fos-tTA transgenic mice were gener-
ated as described in (Liu et al., 2012), by breeding TetTag mice (Reijmers et al., 2007) with C57BL/6J mice and selecting offspring
carrying only the c-fos-tTA transgene. Mice had access to food and water ad libitum and were socially housed in numbers of two
to five littermates until surgery. Following surgery, mice were singly housed. All mice used for the experiments were 4–9 weeks
Neuron 101, 274–284.e1–e5, January 16, 2019 e1
old males at the time of surgery, which had been raised on food containing 40 mg kg-1 doxycycline for at least 20 days before ex-
periments, and remained on doxycycline food for the remainder of the experiments except for the target engram labeling days.
METHOD DETAILS
Engram labeling strategyIn order to label memory engram cells we employed an adeno-associated virus that expressed ChR2 fused to an EYFP fluoro-
phore under the control of a tetracycline-responsive element (TRE)-containing promoter (AAV9-TRE-ChR2-EYFP), which is active
in cells that contain the tetracycline transactivator (tTA). We injected this virus into the hippocampal dentate gyrus (DG) of c-fos-
tTA transgenic mice, which express tTA under the control of a c-fos promoter (Figure S1A) (Reijmers et al., 2007). Because c-fos is
an activity-dependent gene, the c-fos-tTA transgene selectively expresses tTA in active cells. Thus, active DG cells can express
tTA that will then induce the expression of ChR2-EYFP in those cells. In order to restrict activity-dependent labeling of DG cells to
targeted training episodes, mice were fed with doxycycline (DOX, 40 mg/Kg), which sequesters tTA function and prevents ChR2-
EYFP expression. Mice were then taken off DOX one day prior to training in order to permit the labeling of engrams cells with
ChR2-EYFP.
Virus-mediated gene expressionThe recombinant AAV vectors used for viral production were pAAV-TRE-ChR2-EYFP, pAAV-TRE-RFP and pAAV-TRE-Kir2.1-RFP.
Plasmids were serotyped with AAV9 coat proteins and packaged either at University of Massachusetts Medical School Gene
Therapy Center and Vector Core or commercially by Vigene. The recombinant AAV vectors were injectedwith the following viral titers:
1 X 1013 genome copy (GC) ml-1 for AAV9-TRE-ChR2-EYFP, 1.4 X 1013 GC ml-1 for AAV9-TRE-RFP and 2.0 X 1013 GC ml-1 for AAV9-
TRE-Kir2.1-RFP.
Stereotactic injectionAll surgeries were performed using stereotaxic apparatus. Mice were anaesthetized using 500mg kg-1 avertin, or isoflurane. Bilateral
craniotomies were performed using 0.5 mm diameter drill and viruses were injected using a glass micropipette attached to a 10 mL
Hamilton microsyringe (701LT; Hamilton) through a microelectrode holder (MPH6S; WPI) filled with mineral oil. A microsyringe pump
(UMP3; WPI) and its controller (Micro4; WPI) were used to maintain the speed of the injection at 60 nl min-1. The needle was slowly
lowered to the target site and remained for 5 minutes before the beginning of the injection. After the injection the needle stayed for
5minutes before it was slowly withdrawn and the skin was sutured.Mice were given 1.5mg kg-1metacam as analgesic and remained
on a heating pad until fully recovered from anesthesia. Mice were allowed to recover for at least 2 weeks before all subsequent ex-
periments. DG injections were targeted bilaterally to (-2.0mmanteroposterior AP, +/- 1.3mmmediolateral ML, –1.9mmdorsoventral
DV). AAV9-TRE-ChR2-EYFP, AAV9-TRE-RFP and AAV9-TRE- Kir2.1-RFP volumes were 300 nl for DG. AAV9-TRE-ChR2-EYFP was
used undiluted, AAV9-TRE-RFP was used undiluted and AAV9-TRE-Kir2.1-RFP was diluted 1:5 and the percent of engram cells
was confirmed histologically. All injection sites were verified histologically. As criteria we only included mice with ChR2-EYFP,
RFP or Kir2.1-RFP expression limited to the targeted hippocampal region.
For electrophysiology experiments, DG injections of AAV9-TRE-ChR2-EYFP (300 nL, undiluted) or AAV9-TRE-Kir2.1-RFP (300 nL,
diluted 1:5) were targeted unilaterally at coordinates specified above.
BehaviorAll the behavioral experiments were conducted using mice that were 8 to 12 weeks of age. All behavioral experiments designed for
ex vivo physiological analysis were conducted using mice that were 4 to 7 weeks of age. All experiments were conducted during the
light cycle of the day (6.30 am to 6.30 pm).
Apparatus
Context A chambers were 29 X 25 X 22 cm chambers with grid floors, opaque triangular ceilings, bright white lighting, and scented
with 1% acetic acid. All experimental groups were counter-balanced for chamber within contexts. Context B chambers were 30 X
25 X 33 cm chambers with perspex floors, transparent square ceilings, very dim white lighting, and scented with 0.25% benzal-
dehyde. Context A and B chambers were housed in separate rooms in different areas of the facility. For all context discrimination
experiments, semicircular perspex inserts with regular black and white patterns were placed in the chambers of Context A.
Degraded Context AB was created by placing identical perspex inserts in Context B. For context discrimination experiments
all mice were conditioned in Context A, and tested in Contexts A, B, and AB. For pattern completion experiments only Context
A was used. Pilot experiments showed that there was no significant generalization of conditioned response between Context A
and Context B. Four mice were run simultaneously in four identical Context A or Context B chambers, except in the case of pattern
completion experiments and in all experiments intended for ex vivo physiological analysis where one mouse was run at a time.
Floors of chambers were cleaned with Quatricide before and between runs. Mice were transported to and from experimental
room in their home cages using a wheeled cart. The cart and cages remained in an anteroom to the experimental rooms during
all behavioral experiments.
e2 Neuron 101, 274–284.e1–e5, January 16, 2019
Pre-handling
All behavioral subjects were individually habituated to handling by the investigator by handling for one minute on each of three sepa-
rate days. Handling took place in the holding roomwhere the mice lived. Immediately prior to each handling section mice were trans-
ported by wheeled cart to and from the experimental context rooms, to habituate them to the journey.
Training
For context discrimination experiments, mice were trained in Context A using a contextual fear-conditioning paradigm. Training ses-
sionswere 330 s in duration, and three 0.75mA shocks of 2 s duration were delivered at 150 s, 210 s, and 270 s. Immediately after fear
conditioning mice were placed in their home cages, and carted back to the holding room. For engram labeling experiments, mice
were kept on regular food without doxycycline for 24 hours prior to training. When training was complete, mice were switched
back to food containing 40 mg kg-1 doxycycline. For pattern completion experiments mice were exposed to Context A for 10 min
in the absence of shock on day 1. Immediate shock procedures were followed on day 2 in Context A, where a single 1.5 mA shock
of 2 s duration was delivered at either 5 s or 10 s, with all mice removed 30 s after the completion of the shock. For consolidation and
reconsolidation experiments mice were trained according to the same procedure as the context discrimination experiments.
Testing
All testing sessions were 180 s in duration (unless specified differently). Testing conditions were identical to training conditioning,
except that no shocks were presented. In the case of context discrimination experiments, testing in Context AB involved introducing
the visual cues fromContext A into Context B. At the end of all testing session mice were placed in their home cages and carted back
to the holding room.
Quantification of freezing behavior
All behavioral performance was recorded by digital video camera. Data were quantified using FreezeFrame software (ActiMetrics)
with bout size set a 1.25 ms.
Histological verificationMice were dispatched by overdosing with 750–1000 mg kg-1 avertin and perfused transcardially with PBS, followed by 4% parafor-
maldehyde (PFA) in PBS. Brainswere extracted and incubated in 4%PFA at 4�Covernight. Brains were transferred to PBS and 50 mm
coronal slices were taken using a vibratome and collected in PBS. For immunostaining, each slice was placed in PBS + 0.2% Triton
X-100 (PBS-T), with 5% normal goat serum for 1 h and then incubated with primary antibody at 4oC for 24 h. Slices then underwent
three wash steps for 10 min each in PBS-T, followed by 2 h incubation with secondary antibody. Slices then underwent three more
wash steps of 10 min each in PBS, followed bymounting and cover-slipping onmicroscope slides. Antibodies used for staining were
as follows: to stain for ChR2-EYFP, slices were incubated with primary chicken anti-GFP (AbCam, ab13970) (1:1000) and visualized
using anti-chicken Alexa-488 (Invitrogen, A11039) (1:500). For RFP or Kir2.1-RFP, slices were stained using primary chicken anti-RFP
(Rockland, 600-901-379) (1:1000) and secondary anti-chicken Alexa-568 (Invitrogen, A-11041) (1:500). c-Fos was stained with rabbit
anti-c-Fos (1:1000, Synaptic System, 226-003) and goat anti-rabbit Alexa-647 (Invitrogen, A-21245) (1:500) or goat anti-rabbit Alexa-
568 (Invitrogen, A-11011) (1:500).
Cell counting
To quantify the expression pattern of ChR2-EYFP, RFP and Kir2.1-injected c-fos-tTA mice, we counted the number of EYFP/RFP
immunoreactive neurons from 6 coronal slices per mouse (n=6 mice per group). Coronal slices were selected from the dorsal hippo-
campus on coordinates centered on the injection site (–1.82 mm to –2.30 mmAP). Fluorescence images were acquired using a Zeiss
LSM700 confocal microscope (10X magnification). Cell counting analysis was performed using ImageJ-Fiji software. The cell body
layer of DG granule cells upper blade was outlined as a region of interest (ROI) according to the DAPI signal in each slice. The number
of EYFP/RFP-positive cells per section was calculated by thresholding EYFP/RFP immunoreactivity above background levels. All
imaging and analyses were performed blind to the experimental conditions.
Engram cell reactivation
To quantify the percentage of engram cells reactivated duringmemory recall plotted in Figures S1F andS1G, a reactivation indexwas
calculated as (c-Fos+, ChR2-EYFP+ cells / total number of ChR2-EYFP+ cells). Statistical chance was calculated by multiplying the
occurrence of EYFP-positive cells by the occurrence of c-Fos-positive cells.
Ex vivo patch clamp recordingsBehavior
For ex vivo electrophysiology experiments, mice were 28-35 days old at the time of viral injection. Mice weremaintained on food con-
taining 40 mg kg-1 doxycycline and habituated to investigator handling on three consecutive days. One day prior to training, mice
were switched to regular food without doxycycline to enable engram cell labeling. Mice were trained using a contextual fear-condi-
tioning paradigm in Context A (as described above). Two days later, micewere dispatched for patch clamp recording. The 5mand the
3h groups, but not the NR (no recall) group, were subjected to a 3 min recall session in the training chamber, 5 min and 3 hours
respectively, before the ex vivo patch clamp recording.
Animals and slice preparation
Mice (P35-P50) were anesthetized with isoflurane, decapitated and brains were quickly removed. Sagittal slices (300 mm thick)
were prepared in an oxygenated cutting solution at �4�C by using a vibratome (VT1000S, Leica). Slices were then incubated at
Neuron 101, 274–284.e1–e5, January 16, 2019 e3
room temperature (�23�C) in oxygenated ACSF until the recording. The cutting solution contained (in mM): 3 KCl, 0.5 CaCl2,
10 MgCl2, 25 NaHCO3, 1.2 NaH2PO4, 10 D-glucose, 230 sucrose, saturated with 95%O2 - 5%CO2 (pH 7.3, osmolarity 340
mOsm). The ACSF contained (in mM): 124 NaCl, 3 KCl, 2 CaCl2, 1.3 MgSO4, 25 NaHCO3, 1.2 NaH2PO4, 10 D-glucose, saturated
with 95%O2 - 5% CO2 (pH 7.3, osmolarity 300 mOsm). Individual slices were transferred into a submerged experimental chamber
and perfused with oxygenated ACSF warmed at 35�C (± 0.5�C) at a rate of 3 mL min-1 during recordings.
Ex vivo electrophysiologyWhole cell recordings in current clamp or voltage clampmode were performed by using an IR-DICmicroscope (BX51, Olympus) with
a water immersion 40X objective (N.A. 0.8), equipped with four automatic manipulators (Luigs & Neumann) and a CCD camera (Orca
R2, Hamamatsu Co). For all the recordings borosilicate glass pipettes were fabricated (P97, Sutter Instrument) with resistances of
5MU. For current and voltage clamp recordings, pipettes were filled with the following intracellular solution (inmM): 110 K-gluconate,
10 KCl, 10 HEPES, 4 ATP, 0.3 GTP, 10 phosphocreatine and 0.5% biocytin. The osmolarity of this intracellular solution was 290
mOsm and the pH was 7.25. CsCl (Tocris), Quinine (Tocris), Tertiapin LQ (Tocris), Glibenclamide (Tocris), ZD7288 (Tocris) and
ML133 (Sigma) were bath applied. Recordings were amplified using up to two dual channel amplifiers (Multiclamp 700B, Molecular
Devices), filtered at 2 kHz, digitized (20 kHz), and acquired through an ADC/DAC data acquisition unit (ITC1600, Instrutech) by using
custom made software running on Igor Pro (Wavemetrics). Access resistance (RA) was monitored throughout the duration of the
experiment and data acquisition was suspended whenever the resting membrane potential was depolarized above -50 mV, the
RA was beyond 20 MU, or the spike amplitude didn’t overshoot 0 mV.
Optogenetics
Optogenetic stimulation was achieved with a 460 nm LED light source (XLED1, Lumen Dynamics) driven by TTL input with a delay
onset of 25 ms. Light power on the sample was 33mW/mm2. To test ChR2 expression, slices were stimulated with a single light pulse
of 1 s, repeated 10 times every 5 s. In voltage clampmode cells were held at -70 mV, while in current mode, response to optogenetic
stimulation was measured at resting potential.
Data collection and analysis
DGGranule cells expressing ChR2-EYFP, which in current mode (at resting potential) responded to optogenetic stimulation (1 s light
pulse) with at least one action potential were considered engram cells and were selected for the analysis. The intrinsic electrophys-
iological properties were measured in current clamp, holding the cell at –70 mV. Intrinsic membrane properties were collected
following a precise sequence: 1) resting membrane potential was measured in current mode without current injection. 2) membrane
resistance was measured by injecting a negative 100 pA current step lasting 1 s while holding the membrane potential to -70 mV. 3)
rheobase wasmeasured by injecting a variable positive current step lasting 1 s until the cell discharged a single action potential while
holding the membrane potential to -70 mV. 4) Action potential threshold was tested with a current ramp injection while holding the
membrane potential to -70mV. 5) current vs firing rate (IF curve) wasmeasured by injecting 15 pulses of current lasting 1 s, from 10 to
150 pA, while holding themembrane potential to -70mV.Membrane time constant was estimated through single exponential fit of the
recovery-time from a -10mV voltage deflection of 1 s duration. Afterhyperpolarization (AHP) amplitude wasmeasured starting from a
baseline obtained by averaging 10 ms before the spike to the minimum point occurring after the spike. AHP duration was measured
starting from the crossing point between the falling voltage of the spike and the baseline and ending to the crossing point between the
rising voltage of the AHP and the baseline. For data analysis, the investigator was blind to the group origin.
Statistics
Statistical analysis was performed using Igor (Wavemetrics), Matlab (Mathworks), Excel (Microsoft) and GraphPad (Prism). The dis-
tribution of the data was tested with the Kolmogorov-Smirnov test. A two-sample Kolmogorov-Smirnov test or a two-tailed paired or
unpaired t-test were employed for comparisons according to the need. A two way repeated measures ANOVA with multiple compar-
ison test (Sidak correction), was employed to test the significance of the current vs firing rate relationship across groups or to test the
effect of the after-recall time on membrane resistance.
Post hoc immunocytochemistry after recording
For immunohistochemistry of samples subjected to electrophysiological experiments, brain slices were washed in PBST three times
for 10 min, blocked for 1 h, and then incubated overnight at 4�C with chicken anti-GFP (Invitrogen, A10262) (1:1000) to visualize
ChR2-EYFP fibers and CF555 streptavidin (Biotium, 29038) (1:100) to visualize biocytin-filled neurons. On the next day, slices
were washed in PBST three times for 10 min, and then incubated in chicken anti-GFP (Invitrogen, A10262, 1:1,000) for 2 h at
room temperature. After threemore 10-min PBSTwashes, slides were dried at room temperature for at least 6 h before coverslipping
and mounting using VectaShield mounting solution containing DAPI (Vector Laboratories).
Ion channels immunostainingStaining procedure
Mice were dispatched by overdosing with 750–1000mg kg-1 avertin and perfused transcardially with 0.1M PB, followed by 3%para-
formaldehyde (PFA) in PB for 25 minutes. No postfix was used. Immediately, brains were extracted and 80 mm coronal slices were
taken using a vibratome and collected in 0.1MPB. For immunostaining, each slice was placed in PB + 0.3%Triton X-100 (PBS-T) with
5% normal goat serum for 1 h. Then incubated with fAB fragment goat anti-mouse (Jackson Immunoresearch, 115-007-003) (1:200).
Then slices were washed 3 times for 10 min each in PB and incubated with primary antibody at 4oC for 12 h. Slices then underwent
e4 Neuron 101, 274–284.e1–e5, January 16, 2019
three wash steps for 10 min each in PB-T, followed by 2 h incubation with secondary antibody. Slices then underwent three more
wash steps of 10 min each in PB, followed by mounting and cover-slipping on microscope slides (Vectashield mounting medium).
Antibodies used for staining were as follows: to stain for ChR2-EYFP, slices were incubated with primary chicken anti-GFP (AbCam,
ab13970) (1:1000) and visualized using anti-chicken Alexa-488 (Invitrogen, A11039) (1:500). For Kir2.1 channel, slices were stained
using primary mouse anti-Kir2.1 (Neuromab, 75-210) (1:200) and secondary anti-mouse Alexa-568 (Invitrogen, A-11031) (1:500).
For sodium channel, slices were stained using primary mouse anti-PAN (Sigma, S-8809) (1:500) and secondary anti-mouse
Alexa-568 (Invitrogen, A-11031) (1:500). Kir2.1 antibody was validated by hyperexpressing Kir2.1-mCherry in DG engram cells and
by performing the staining with a secondary antibody tagged by Alexa-488 (Figure S4B). Colocalization between Kir2.1-mCherry
(red) and Kir2.1 antibody signal (green) validated the staining accuracy.
Data acquisition
A total of n=7mouse brains was used for each of the 4 groups (NR, 5m, 3h, ANI) for a total number of 28 brains. Two brains per group
were excluded for technical reasons resulting in 5 brains per group for a total number of 15 brains. From these brains we collected a
total number of 386 images at 40X magnification centered on the DG region of the hippocampus. Settings such as laser intensity at
and gain of the photomultiplier were kept constant for all the samples. Imageswere collected by focusing the confocal microscope on
a ChR2-EYFP positive cell. A snapshot was taken on a plane intersecting the center of the soma. In this way the nuclear region re-
sulted dark and the cell membrane, site of the Kir2.1 channels, was exposed. After the cell detection was accomplished and the scan-
ning plane selected, a second snapshot was taken from three different channels: DAPI, YFP and Alexa-568 signal. The Alexa-568
fluorophore was exposed to the laser exclusively at the moment of the snapshot to avoid photobleaching.
Criteria such as tissue quality, staining quality and success of sodium channel staining (positive control, Figure S4C) were used to
select high quality images.
Data analysis
For image analysis, the investigator was blind to the group origin. Of 386 Chr2-EYFP-positive cells collected, 235 cells were selected
for analysis and the others discarded for technical reasons (see previous criteria). For each sample an engram cell was paired to an
adjacent non engram cell. Analysis consisted in measuring the average membrane fluorescence intensity at 568 nm. The region of
interest was located on the periphery of the DAPI signal for both DG engram and non engram cells and coincided with the YFP-signal
of the cell membrane. Non engram cells were selected randomly based on: close proximity to the engram cell (< 50 mm) and same
focal plane. Background fluorescence was measured from samples without primary Kir2.1-antibody and subtracted from the data.
The resulting value was normalized by the fluorescence intensity of a 50X50 mm area measured on the DG molecular layer. Analysis
was performed using ImageJ-Fiji software. Paired analysis was employed, engram vs non engram cell expression of Kir2.1, and a two-
tailed paired t-test was used to assess statistical significance.
QUANTIFICATION AND STATISTICAL ANALYSIS
Data are displayed as mean ± SEM and the significance threshold was set at a=0.05 (NS, *p<0.05, **p<0.01, ***p<0.001). Statistical
analysis was performed by using Prism (GraphPad), Excel or Igor (Wavemetrics). No statistical method was used to determine sam-
ple size. Statistical comparisons were performed blindly to the experimenter. The normality of data distribution was tested with a
Kolmogorov-Smirnov test. Two-tailed unpaired t-test was used in Figures 1D, 1E, 1F, 1H, 1I, 2H, 4C, 4E, 4D, S1C, S1E, S1G,
S2A, S2B, S2C, S2D, S3E, S3G, S5B, S5D, S7B, S5D, S7C, S7D, S7E. Two-tailed paired t-test was used in Figures 2C and 2L.
One way ANOVA with Tukey’s test for multiple comparisons was used in Figures 3B and 3D. Two way repeated measures
ANOVA with multiple comparison test (Sidak correction) was used in Figure 1G, 3E and S2B.
DATA AND SOFTWARE AVAILABILITY
All data, materials and animals are available from corresponding authors upon request.
Neuron 101, 274–284.e1–e5, January 16, 2019 e5