www.sciencemag.org/cgi/content/full/331/6019/924/DC1

Supporting Online Material for

Early Tagging of Cortical Networks is Required for the Formation of Enduring Associative Memory

Edith Lesburguères, Oliviero L. Gobbo, Stéphanie Alaux-Cantin, Anne Hambucken, Pierre Trifilieff, Bruno Bontempi*

*To whom correspondence should be addressed. E-mail: b.bontempi@cnic.u-bordeaux1.fr

Published 18 February 2011, Science 331, 924 (2010)

DOI: 10.1126/science.1196164

This PDF file includes:

Materials and Methods

SOM Text

Figs. S1 to S13

References

1

Supporting Online Material for

Early tagging of cortical networks is required for the formation of enduring associative memory

Edith Lesburguères, Oliviero L. Gobbo, Stéphanie Alaux-Cantin,

Anne Hambucken, Pierre Trifilieff, Bruno Bontempi*

Institut des Maladies Neurodégénératives, CNRS UMR 5293, Universités de Bordeaux 1 et 2, Avenue des Facultés, 33405 Talence, France.

*To whom correspondence should be addressed. E-mail: bruno.bontempi@u-bordeaux2.fr

Materials and Methods Animals

Male Sprague Dawley rats (Janvier breeding center, Le Genest Saint Isle, France)

weighing 225-250 g at the beginning of experiments were used throughout. All rats

were housed individually in Plexiglas cages and maintained on a 12:12 h light-dark

cycle. Food and water were freely available except during behavioral training where

rats were food-deprived to 85% of their free-feeding body weight. All behavioral and

surgery experiments were conducted during the light phase of the cycle and were in

accordance with official European Guidelines for the care and use of laboratory

animals (86/609/EEC).

Stereotaxic surgery

Animals were implanted bilaterally under deep general anesthesia (mixture of

ketamine 100 mg/kg, and xylazine 12 mg/kg, injected i.p.) with stainless steel guide

cannulae using the following stereotaxic coordinates (S1): (i) Hippocampus (HPC):

anteroposterior (AP) relative to bregma, –3.8 mm; lateral (L) to midline, ±2 mm;

ventral (V) from the skull surface, –2 mm. (ii) Orbitofrontal cortex (OFC): AP, +4.2

mm; L, ±2 mm; V, -2.7 mm. Rats were allowed a minimum of two weeks to recover

before being submitted to memory testing.

2

Cumin

Demonstrator

Observer

Cumin

Exposure

Social interaction

Test

Delay

Thyme

The social transmission of food preference (SFTP) task

The STFP task used to assess associative olfactory memory was performed as

described by Clark et al. (S2) with slight modifications. Rats underwent the classical

three-step procedure.

The STFP paradigm with the cumin/thyme flavored pair used to assess associative olfactory memory

Briefly, demonstrator rats were food-deprived and then habituated to eating

plain or cumin powdered chow for three days (30 min session). Observer rats were

also shaped for three days to consume plain powdered chow from two cups placed in

their home cage. For the interaction session, the demonstrator rat was allowed 30

min access to one cup filled with plain, cumin (0.5 %, i.e. 0.5 g of cumin mixed in

99.5 g of plain powdered chow) or cocoa (2%) powdered chow (water was removed

from the cage) and was then moved to the observer’s cage fitted with a stainless

steel wire mesh divider. Food-deprived observer rats were kept in the opposite side

of their cage for a 15 min interaction period at the end of which the divider was

removed, allowing the two rats to interact freely for another 15 min. At the end of this

30 min interaction period, the demonstrator rat was removed from the cage.

Observer and demonstrator rats were always unfamiliar with each other. At test, after

a selected retention interval (1, 7, 15 or 30 days depending on the experiment, see

below), food-deprived observer rats were presented in home cage with a choice of

two cups containing a novel food (0.75% thyme or 1% cinnamon powdered chow)

and the familiar food that the demonstrator rat had consumed before interacting with

the observer rat (0.5% cumin or 2% cocoa powdered chow). The position of the

familiar food (left or right) in the cage was counterbalanced across the different

3

groups. After 20 min, cups were removed, weighed and olfactory associative memory

performance was expressed as percentage of familiar food eaten (% cumin or %

cocoa) using the following formula: (amount of familiar food eaten / amount of total

food) x 100.

Flavor concentrations within each food pair (cumin/thyme and

cocoa/cinnamon) were chosen in pilot experiments to induce an innate preference for

one given flavor (i.e. thyme and cinnamon, respectively). Use of these two biased

flavored pairs enabled to decrease the chance level at test and thus to optimize the

possibility of detecting changes in memory performance across our various

treatments. At the concentrations used for the cumin/thyme flavor pair for example,

we found that rats naturally prefer thyme over cumin. However, interaction with a

demonstrator that has eaten cumin powdered chow could reverse this innate

preference so that observers chose cumin over thyme (up to 80% of the total food

eaten, chance level of ~20%).

By interacting with a demonstrator rat that has recently eaten a novel flavored

food (e.g. cumin), note that the observer rat forms an association between this food

odor and some constituents of the demonstrator’s breath. Subsequently, when

submitted with a choice between cumin and a new flavored food, the observer rat

expresses a memory for this association by preferentially choosing the same food

odor that was present in the demonstrator’s breath. Interestingly, this paradigm does

exhibit some of the key features of declarative memory that is information about

potential food sources can be encoded rapidly and expressed flexibly in a test

situation different from the circumstances encountered during initial learning (S3).

The STFP task has been shown to be dependent on hippocampal function (S2, S4)

and is particularly well-suited for the investigation of remote memory formation

because a single training session produces long-lasting memories resistant to

forgetting (fig. S2).

The fact that encoding of associative olfactory memory occurs within only one

brief training session provided rigorous control over the time-course of hippocampal-

cortical interactions underlying systems-level memory consolidation and avoided

repeated (over days) initial training sessions as is often the case in complex spatial

tasks. In addition, we were careful in testing the observer rat in its home cage kept in

the animal facility, thereby reinforcing the nonspatial component of the STFP task.

4

vCA3

vCA1

vDG

dCA1

dDG dCA3

PAR

MO

VO

LO

DLO

+4.20mm -6.04mm-3.80mm

vCA3

vCA1

vDG

vCA3

vCA1

vDG

dCA1

dDG dCA3

PARdCA1

dDG dCA3

PARdCA1

dDG dCA3

PAR

MO

VO

LO

DLO

MO

VO

LO

DLO

+4.20mm -6.04mm-3.80mm

This enabled us to better isolate the functional implication of the hippocampus in

systems-level consolidation by minimizing hippocampal-dependent processing of

spatial information.

Immunocytochemistry and brain imaging

Observer rats were terminally anesthetized with pentobarbital (300 mg/kg i.p.) and

perfused transcardially with 0.9% saline and 4% paraformaldehyde 90 min after

completion of retention testing. Brains were removed and prepared for

immunocytochemistry on free-floating sections as previously described (S5) using

anti-Fos (1:5000) and anti-acetyl-histone H3 (Lys9, 1:2000) primary rabbit polyclonal

antibodies. A biotinylated goat anti-rabbit antibody (1:2000) was used as secondary.

Staining was revealed using the avidin-biotin peroxydase method (ABC kit) coupled

to diaminobenzidine as chromogen. Synaptophysin immunofluorescence was

achieved by using a primary mouse monoclonal antibody (1:500) and an Alexa Fluor

488-conjugated goat anti-mouse secondary antibody (1:500). For co-visualization of

nuclei, 1.5 μg/ml of 4,6-diamidino-2-phenylindole (DAPI) was included in the

mounting medium. Quantitative analyzes of positively labeled nuclei were performed

using a computerized image-processing system coupled to a microscope. Structures

were anatomically defined according to the Paxinos and Watson atlas (S1).

Schematic drawings of rat coronal sections (adapted from S1) showing the regions of interest

(filled areas) selected for measurement of Fos-positive nuclei.

Numbers indicate the distance (in millimeters) of the section from bregma. dCA1: CA1 field of dorsal hippocampus; vCA3: CA3 field of dorsal hippocampus; dDG: dorsal part of dentate gyrus; vCA1: CA1 field of ventral hippocampus; vCA3: CA3 field of ventral hippocampus; vDG: ventral part of dentate gyrus; MO: medial orbital cortex; VO: ventral orbital cortex; LO: lateral orbital cortex; DLO: dorsolateral orbital cortex; PAR: parietal cortex. Fos counts for the following regions were expressed as the pooled means of the listed subregions. Dorsal hippocampus (dorsal HPC): dCA1, dCA3, dDG; Ventral hippocampus (ventral HPC): vCA1, vCA3, vDG; Orbitofrontal cortex (OFC): MO, VO, LO, DLO.

5

Immunoreactive neurons were counted bilaterally using a minimum of three

sections by an experimenter blind to the treatment condition. For each rat, the mean

count in a given structure was divided by the mean count in that region for the

respective food preference control group. The normalized scores were then

expressed as a percentage and averaged across rats to give the final means of each

group.

Western blotting

Rats were decapitated 90 min after recent (Day 1) or remote (Day 30) memory

testing. Brains were rapidly removed and frozen in isopentane, and 150-250 μm-thick

sections were cut with a cryostat. Regions of interest were then punched and kept

frozen until protein extraction. Proteins were extracted by application of boiled 1%

SDS in TBS buffer (Tris 0.01 M, NaCl 0.1 M, pH 7.4) followed by brief sonication and

denaturation 5 minutes at 95°C. Lysates (20 μg per lane) were separated by 10%

SDS-PAGE before electrophoretic transfer onto polyvinylidene difluoride membranes

(0.45 μm Hybond P). Blots were treated as previously described (S6). Briefly,

membranes were saturated for 1 h at room temperature with 5% BSA (Fraction V)

and incubated overnight at 4°C with anti-synaptophysin antibody (1:1000). On the

second day, blots were incubated for 2 h at room temperature with goat anti-rabbit

horseradish peroxidase-conjugated secondary antibody before exposure to the ECL

substrate and photographic processing. Detection of ERK2s protein with (1:5000)

was used as a loading control. Western blotting quantification was conducted by

measuring optical density directly on photographic films.

Intracerebral infusion procedure

Various drugs (see detailed experiments below) were infused using an injection

cannula projecting 1.5 mm beyond the tip of the guide cannula. For hippocampus, 1

µl was injected at a rate of 0.8 µl/min; for OFC, 0.8 µl was injected at a rate of 0.6

µl/min. Anatomical specificity is a critical issue when using pharmacological

techniques to inactivate brain regions. Therefore, only animals with cannula tips

correctly located within targeted structures, and whose extent of neuronal inactivation

was verified to be circumscribed to the region of interest via control of the expression

of the activity-dependent gene c-fos expression, were included in the study (S5).

6

Golgi-Cox impregnation procedure

Golgi-Cox staining was used to assess memory-induced morphological modifications.

Rats were terminally anesthetized with pentobarbital (300 mg/kg i.p.) and

transcardially perfused with a solution of 0.9% saline. Brains were dissected and

impregnated using a Golgi-Cox solution according to the method described by Glaser

& Van der Loos (S7). Briefly, they were first immersed in the Golgi-Cox solution at

room temperature for 14 days and then transferred to a 30% sucrose solution for 3

days before sectioning using a vibratome. Coronal sections (100 μm thick) were

mounted on gelatinized slides, stained according to the Gibb & Kolb method (S8),

and coverslipped with Permount. Spine density was analyzed on pyramidal neurons

located in the OFC defined according to the Paxinos and Watson atlas (S1). Fully impregnated neurons were reconstructed using a computer-assisted

morphometry system consisting of a microscope equipped with a motorized stage

and a video camera interfaced with a computer loaded with the NeuroLucida

software. This system allowed for accurate mapping, tracing and reconstruction of

the neurons and their dendrites in three-dimension. Neurons were first located within

the OFC using a 20X objective. Only neurons which satisfied the following criteria

were chosen for analysis in each of the experimental groups: (1) presence of

untruncated dendrites, (2) consistent and dark impregnation along the entire extent of

all of the dendrites and (3) relative isolation from neighboring impregnated neurons to

avoid interference and ensure accuracy of dendritic spine counting.

Because memory processes trigger specific intracellular signaling cascades

which support neuronal plasticity phenomena, memory-induced changes in spine

density are likely to be restricted to discrete subsets of neurons. Therefore, these

changes stand a high probability of being ‘diluted’ if neuronal samples are randomly

obtained across the general neuronal populations in the area of interest, in our case,

the OFC. To better control for this confounding factor, we took advantage of our brain

imaging approach using the activity-dependent gene c-fos which identified precisely

the subregions of the OFC exhibiting increased Fos protein expression at the time of

remote memory retrieval. Our pilot studies revealed that this increased Fos

expression correlated with the expression of the synapse density marker

synaptophysin, suggesting that it could be attributed, at least in part, to

7

synaptogenesis, which increases the complexity and extent of cortical networks

involved in memory storage. We therefore concentrated our search for impregnated

neurons in the OFC subregions exhibiting the highest Fos expression, namely the

VO and DLO areas.

The selected impregnated neurons were then drawn using a 100X objective

with a numerical aperture of 1.4. A live image was captured by the Neurolucida

software and shown on the computer screen. Mapping was then performed by

moving the microscope stage in 1-μm steps through the z axis along the length of

each dendrite using a joystick. Spines were plotted at the same time. Thus, the x, y, z

coordinates of each dendrite and spine were recorded to enable subsequent three-

dimensional representation and rotation of each reconstructed neuron. A total of six

neurons were fully reconstructed from each brain (three per hemisphere), thereby

minimizing biased sampling associated with random selection of only partial

segments of each dendrite. For each neuron, apical and basal dendrite

reconstructions were performed separately and started at the second branching node

in order to exclude spine depleted zones which arise from the cell body. Only

protuberances with a clear connection of the head of the spine to the shaft of the

dendrite were counted as spines.

Representative example of a three-dimensional computer-assisted reconstruction of one basal dendrite tree of a neuron taken in the OFC. Scale bar: 50 μm.

Because this method has proven to hold reliable results (S9), no attempt was

made to introduce a correction factor for hidden spines. Density measures were

pooled for each dendrite category (basal and apical) to generate the final spine

density results expressed as number of spine per μm for each neuron. To examine

8

how our various treatment-induced changes in spine density affected the entire

population of reconstructed neurons, we also divided our total population of neurons

into 3 subpopulations exhibiting low (<1 spine/µm), intermediate (between 1 and 1.1

spines/µm) or high (>1.1 spines/µm) density of dendritic spines and plotted the

frequency distribution (expressed in %) of each of these categories. All

measurements were performed by an experimenter blind to the experimental

conditions.

Experiments

Experiment 1. HPC is required for acquisition of the STFP task (see fig. S1)

Observer rats implanted with guide cannulae aimed at the dorsal HPC were

inactivated bilaterally with the AMPA receptor antagonist 6-cyano-7-nitroquinoxaline-

2,3-dione (CNQX, 3 mM, in artificial cerebrospinal fluid (aCSF)) one hour prior to

being allowed to interact with demonstrators exposed to cumin. Rats injected with

aCSF served as controls. Independent groups of rats were tested for retrieval at Day

7.

To ensure that CNQX did not change the innate preference of the animals for

thyme or cumin (or cocoa and cinnamon) and to establish chance levels

experimentally for the flavored pair, additional food preference control groups infused

with aCSF or CNQX were generated. These important control groups were added to

all experiments which required intracerebral infusions of drugs. Throughout all our

pharmacological experiments, note that we did not observe any significant difference

in the percentage of familiar food eaten between food preference groups receiving

aCSF or a given drug. Therefore, their performance was pooled to generate the

experimental chance level represented on graphs by a dotted line and its associated

standard error mean.

Assessing memory performance in the social transmission of food preference

task relies exclusively on the amount of food eaten by the animals. To rule out the

possibility that targeted pharmacological inactivation interfered with motivational

processes, our experiments were designed to control as much as possible for this

potential confounding factor by adding the relevant control groups for each targeted

brain region (i.e. vehicle-injected groups, use of two or more retention delays for a

given targeted brain region enabling to show a delay-dependent effect of inactivation

9

on memory performance in the absence of any treatment effect on total food

consumption).

Experiment 2. Time-dependent involvement of HPC and OFC in recent and remote

memory retrieval of associative olfactory memory (see fig. S2)

To identify brains regions involved in processing recent and remote

associative olfactory information, we tracked the expression of the transcription factor

c-fos classically used as an indirect correlate of neuronal activity. Observers rats

were allowed to interact with demonstrator rats exposed to cumin and were then

tested for food preference either 1 day (recent memory group) or 30 days (remote

memory group) later. Innate food preference was assessed using observer rats that

interacted with a demonstrator rat exposed to plain powdered chow and

subsequently submitted, recently or remotely, to a choice between cumin and thyme

powdered chow. These food preference control groups served to establish the

chance level associated with the cumin/thyme flavored pair at the selected delays

and to isolate changes in Fos protein expression associated with nonspecific aspects

of the testing procedure (locomotor activity, perception of the novel odor food, i.e.

thyme). Observer rats from these food preference groups were treated exactly as

experimental animals. Brains were collected 90 min after testing and processed for

Fos immunocytochemistry as described above. Fos expression induced by the social

interaction was examined in observer rats submitted to an interaction with a

demonstrator fed with cumin (experimental group) or plain food (food preference

group). Following the 30 min interaction session, rats were returned to their home

cage. They were not given a choice between cumin and thyme powdered chow and

their brains were collected 90 min later.

Experiment 3. Effects of region-specific pharmacological inactivation of HPC and

OFC on recent and remote memory retrieval of associative olfactory memory (see

figs S2 and S3)

Because of the correlative nature of the imaging approach used in Experiment

1, we performed region-specific pharmacological inactivation of HPC and OFC using

the selective sodium channel blocker tetrodotoxin (TTX) or the AMPA receptor

antagonist CNQX. We favored this pharmacological approach to minimize the

occurrence of compensatory phenomena within memory systems classically

10

associated with irreversible lesion techniques. Observer rats implanted with guide

cannulae aimed at the dorsal HPC or OFC were allowed to interact with

demonstrators exposed to cumin before being tested for food preference either 1 or

30 days later. One hour prior to retention testing, animals in each group were infused

bilaterally either with TTX (10 ng per site diluted in aCSF), or aCSF used as vehicle.

As a sodium channel blocker, TTX suppresses neuronal activity of both excitatory

and inhibitory neurons in the targeted region, but also affects fibbers of passage. To

circumvent this latter effect, additional animals were also injected with the AMPA

receptor antagonist CNQX which interferes with the excitatory glutamatergic

transmission in a reversible and time-restricted manner (CNQX, 3 mM)(see fig. S3).

Experiment 4. Effects of systems-level memory consolidation of associative olfactory

memory on structural plasticity in the OFC (see fig. S4)

In addition to inducing changes in the efficacy of synaptic transmission

between existing synapses (‘weight’ plasticity), embedding of memories into cortical

networks are also expected to induce rewiring of the connectivity between neurons

(‘wiring’ plasticity)(S10). To examine this latter form of cortical plasticity, we

measured the level of expression of the presynaptically localized protein

synaptophysin revealed by fluorescent immunocytochemistry. Observer rats were

allowed to interact with demonstrators exposed to cumin before being tested for food

preference either 1 or 30 days later. Brains were collected 90 min after testing and

processed for fluorescent immunocytochemistry as described above. Morphological

changes at post-synaptic sites were also evaluated using the Golgi-Cox method. To

rule out the possibility that retrieval processes, in addition to memory consolidation

per se, may have contributed to the observed structural changes within cortical

networks, experimental and food preference underwent the STFP procedure without

being subsequently tested for recent or remote memory retrieval at Day 1 or Day 30.

Brains were collected at these two delays and directly processed for Golgi-Cox

staining as described above.

Experiment 5. Effects of chronic inactivation of HPC or OFC during specific post-

interaction periods on remote memory retrieval (see Fig.1 and figs S5-S7)

Observer rats implanted with guide cannulae aimed at the dorsal HPC or OFC

were allowed to interact with demonstrator rats exposed to cumin before being

11

chronically inactivated with CNQX (3 mM) during an early (from immediately after

social interaction to Day 12 or from Day 1 to Day 12) or a late (from Day 15 to Day

27) post-acquisition period. Remote memory testing occurred at Day 30. Infusions of

aCSF or CNQX were carried out every other day during the selected post-interaction

period to avoid excessive drug dosage. Length of these two post-acquisition periods

was chosen based on pilot Fos imaging experiments that have examined the time-

course of hippocampal disengagement during the 30-day total period. We chose

CNQX as a blocker of neuronal activation due to its time-restricted effects (fig. S3).

This enabled to target the consolidation period without affecting memory retrieval

processes. A subset of brains from each group was processed for Golgi-Cox staining

as described above.

To examine whether chronically inactivated HPC and OFC animals could

relearn and consolidate, we submitted them 7 days after respective testing to a

second social interaction using a cocoa/cinnamon flavor pair. Rats inactivated

chronically in the HPC were tested for retrieval 1 day following the second interaction

with a demonstrator rat fed with cocoa (choice between cups filled with cocoa or

cinnamon). Rats inactivated chronically in the OFC were tested for retrieval 30 days

following the second interaction, a delay which matches the recruitment of OFC

function in remote memory retrieval.

Experiment 6. Effects of OFC inactivation with CNQX upon encoding on remote

memory retrieval (see Fig. 2 and fig. S8)

Observer rats implanted with guide cannulae aimed at the OFC were

inactivated bilaterally with CNQX one hour prior to being allowed to interact with

demonstrators exposed to cumin. Independent groups of rats were tested for retrieval

at Day 7 or Day 30. A subset of brains from this latter group was processed for Golgi-

Cox staining as described above.

To examine whether animals injected with CNQX upon interaction and tested

at Day 7 could rely only on olfactory information for successful performance, we

generated one additional group of rats (OLF group) that could smell the cumin flavour

without access to the powdered chow. Instead of interacting with a demonstrator, a

small glass jar containing cumin (6 g of flavored powdered chow corresponding to the

average amount of food eaten by demonstrator rats) and with holes drilled in the

metal lid was introduced in their cage. Rats were tested 7 days later.

12

Experiment 7. Effects of OFC inactivation with AP-5 upon encoding on remote

memory retrieval (see Fig. 2 and fig. S11)

The experimental design of this experiment was similar to that of experiment 6

except that the selective NMDA receptor antagonist DL-2-Amino-5-phosphonovaleric

acid (AP-5, 0.8 µl of a 5 µg/µl solution) was infused into the OFC 1 hour before social

interaction. All rats were tested for remote memory retrieval at Day 30. Their brains

were collected after completion of testing and processed for synaptophysin

fluorescence immunocytochemistry as described above.

Experiment 8. Is the synaptic tagging process in the OFC information-specific? (see

Fig. 2)

Observer rats implanted with guide cannulae into the OFC were submitted to

two consecutive social interactions with two different flavored pairs, cocoa/cinnamon

and cumin/thyme. The OFC was inactivated bilaterally with CNQX one hour prior to

the second interaction which occurred 7 days after the first interaction. Remote

memory for each pair was assessed 30 days after each interaction.

Experiment 9. Effects of partial versus extensive inactivation of the OFC upon

encoding on remote memory retrieval (see Fig. 2)

Observer rats implanted with guide cannulae aimed at the OFC were

inactivated with CNQX one hour prior to being allowed to interact with demonstrators

exposed to cumin. Animals of the extensive inactivation group received 0.8 µl of

CNQX (3 mM solution) into the OFC while animals from the partial inactivation group

received only 0.3 µl. Control rats received aCSF as vehicle. Choice of these two

volumes was based on pilot experiments that have examined the extent of neuronal

inactivation induced by the infusions via control of Fos expression blockade.

Independent groups of rats were tested for retrieval at Day 7, 15, 30. Brains were

collected 90 min after testing and processed for Fos immunocytochemistry as

described above to examine the effects of OFC blockade on the temporal dynamics

of hippocampal-cortical interactions.

13

Experiment 10. Effects of encoding of associative olfactory memory on levels of

histone H3 acetylation in the HPC and OFC (see Fig.3 and fig. S9)

Observer rats were allowed to interact with demonstrators exposed to cumin

(experimental group) or plain powdered chow (food preference group). Their brains

were collected 1 hour later and processed for histone H3 acetylation

immunocytochemistry as described above using an antibody directed against

acetylated Lysine 9.

Experiment 11. Effects of blockade of the MAPK/ERK signaling cascade in the OFC

on remote memory retrieval (see Fig. 3 and fig. S11)

Observer rats implanted with guide cannulae aimed at the OFC were

inactivated bilaterally 1 h prior to social interaction with 1,4-diamino-2,3-dicyano-1,4-

bis(2-aminophenylthio)-butadiene (U0126 dissolved in aCSF containing 30% DMSO,

1.6 µg in 0.8 µl). This dose of U0126 was chosen based on previous findings

showing that it selectively and reliably impairs the activation of the MAPK/ERK

pathway by inhibiting the mitogen-activated and extracellular signal-regulated kinase

(MEK) (S11). Controls animals were injected with the same vehicle. All rats were

tested for remote memory retrieval at Day 30.

We used the same experimental design to block the mitogen- and stress-

activated protein kinase 1 (MSK1) in the OFC. In the absence of a specific inhibitor of

MSK1 activity, we used the non-selective protein kinase inhibitor (N-[2–p-

bromocinnamylamino-ethyl]-5-isoquinolinesulfonamide H89 (S12). Rats were infused

bilaterally with H89 dissolved in aCSF (2 µg in 0.8 µl). Controls animals were injected

with the same vehicle. Since the two experiments were conducted conjointly and

each vehicle group exhibited similar performance in the STFP task (F < 1), these two

groups were pooled to generate the final graph and statistics reported in main Fig.

3B.

Brains in these two experiments were collected upon completion of remote

memory testing and processed for synaptophysin fluorescence immunocytochemistry

as described above.

14

Experiment 12. Effects of blockade of MSK1 on the level of histone H3 acetylation

induced by social interaction (see fig. S10)

Observer rats implanted with guide cannulae aimed at the OFC were

inactivated bilaterally 1 h prior to social interaction with H89. Controls animals were

injected with aCSF. Their brains were collected 1 hour after completion of the social

interaction and processed for histone H3 acetylation immunocytochemistry as

described above using an antibody directed against acetylated Lysine 9.

Experiment 13. Effects of intra-OFC infusion of the histone deacetylase inhibitors

sodium butyrate (NaB) and Trichostatin A (TSA) on the level of histone H3

acetylation induced by social interaction (see Fig. 3 and fig. S12)

Observer rats implanted with guide cannulae aimed at the OFC were infused

with NaB or TSA immediately upon completion of social interaction (NaB diluted in

aCSF, 1 µg in 0.8 µl; TSA diluted in a mixture of aCSF and 30% DMSO, 3.5 µg in 0.8

µl). Controls animals were injected with the corresponding vehicle. Their brains were

collected 4 hours later and processed for histone H3 acetylation

immunocytochemistry as described above using an antibody directed against

acetylated Lysine 9.

Experiment 14. Effects of maintaining elevated levels of histone H3 acetylation in the

OFC during specific post-interaction periods on remote memory retrieval (see Fig. 3

and figs S12 and S13)

Observer rats implanted with guide cannulae aimed at the OFC were allowed

to interact with a demonstrator exposed to cumin before being chronically infused

with NaB (1 µg in 0.8 µl of aCSF) during an early (from immediately after social

interaction to Day 6) or a late (from Day 15 to Day 21) post-acquisition period.

Remote memory testing occurred at Day 30. To maximize the possibility of detecting

a NaB-induced memory improvement at this delay (i.e. to prevent a ‘ceiling ‘effect),

the stainless steel wire mesh divider was maintained during the entire 30 min

interaction period. While still inducing a robust associative olfactory memory

significantly above chance level, (early period: F1,15 = 6.90; p < 0.02; late period: F1,12

= 14.62; p < 0.01 versus FP controls) retrieval performance was slightly reduced

compared to the typical performance of rats given free access to the demonstrator

during the last 15 min of the interaction period (~45% versus ~70%). This difference

15

in remote memory performance induced by the two testing procedures was clearly

time-dependent as pilot experiments indicated that performance assessed 24 hours

after social interaction was similar across the two groups.

Infusions of aCSF or NaB were carried out every other day during the selected

post-interaction period to avoid excessive drug dosage. Brains were collected upon

completion of remote memory testing and processed for synaptophysin fluorescence

immunocytochemistry as described above.

We also tested a second histone deacetylase inhibitor, Trichostatin A, to

confirm the observed facilitatory effect of sodium butyrate on remote memory

retrieval following “early” drug infusion. Observer rats implanted with guide cannulae

aimed at the OFC were allowed to interact with a demonstrator exposed to cumin

before being chronically infused with TSA (3.5 µg in 0.8 µl of a mixture of aCSF and

25% DMSO) during the early (from immediately after social interaction to Day 6)

post-acquisition period. Remote memory testing occurred at Day 30.

Statistical analyses

Results were expressed as means ± SEM. Data analyses were performed using

analyses of variance (ANOVAs) followed by post-hoc paired comparisons using

Neuman-Keuls F-tests or Student’s t-tests where appropriate. Values of p < 0.05

were considered as significant.

16

Supporting text

The hippocampus as a consolidation organizing device

Studies of memory in humans and animal models, coupled to various

correlative and invasive techniques including neuroimaging, brain lesions,

electrophysiological recordings and mapping of gene expression, have provided

complementary evidence that the medial temporal lobe, which includes the

hippocampus, operates with neocortex to establish and maintain remote memories.

Further supporting the concept of the hippocampus as a crucial consolidation

organizing device, findings from our inactivation experiments provide converging

evidence that the hippocampus constitutes at least one indispensable node within the

medial temporal lobe network of interconnected brain regions in charge of

coordinating the embedding of remote associative olfactory memories within

networks of the OFC. However, the possibility remains that a broader subcortical

network extending beyond the boundaries of the medial temporal lobe system (e.g.

the thalamic system), might guide or exert a modulatory role on the plasticity-related

events we observed in the OFC during the course of systems-level memory

consolidation (S13).

While our findings point to the existence of early tagging of cortical networks

as a prerequisite for the formation of enduring memories through bidirectional

hippocampal-cortical interactions, it is important to note that not all types of memory

require the hippocampus or the medial temporal lobe for their consolidation and

future studies will need to determine whether an early neuronal tagging process is

also required as a gateway for the formation of remote memories within other

memory systems involved in memory consolidation.

The cortical tagging process

The nature of our different complementary experiments does not enable to

pinpoint precisely the mechanisms involved in the ‘tagging’ of cortical neurons at the

synaptic level. Our experiments were not designed to favor particularly the synaptic

tagging and capture (STC) hypothesis first introduced by Frey and Morris in 1998

(tagged synapses can capture plasticity-related proteins that stabilize synaptic

17

modifications (S14)), but only to use it as a conceptual framework in the context of

systems-level memory consolidation.

Over the last decade, accumulated evidence has considerably widened the

theoretical framework initially set by the STC hypothesis and unraveled a series of

additional but non mutually exclusive cellular and molecular mechanisms that most

likely act in concert to support associativity in synapse-specific plasticity processes

(for an expanded view of the synaptic tagging and capture model, see S15). Local

protein synthesis and degradation at activated synapses (S16), synapse sensitization

enabling distribution of specific plasticity-related proteins across all synapses of a

neuron (S15, S17), clustered plasticity among synapses within a dendritic branch

(S18), intracellular protein trafficking (S19), structural changes such as formation of

new active presynaptic terminals or widening of synaptic spines (S20) are all relevant

mechanisms which may underlie the early cortical tagging process we have

indentified in the present study.

Role of the OFC in recent memory

We found that transitory silencing of neuronal activity in the OFC either prior to social

interaction or prior to recent memory retrieval (Day 1) did not impair recent memory

retrieval, indicating that the OFC is neither required for acquisition of the social

transmission of food preference (STFP) task nor for retrieval of recent memory. While

confirming previous findings which have used irreversible lesions of the OFC (S21),

this result contrasts with the study by Ross et al. (S22) showing that acetylcholine in

the OFC is necessary for the acquisition of the STFP task.

The difference in the experimental methodology used to impair OFC function

in the studies mentioned above (e.g. transitory neuronal inactivation versus

irreversible acetylcholine depletion), and especially anatomical specificity, may

account for the reported contrasting findings. Indeed, the observation that (192)IgG-

saporin used in the Ross study to destroy cholinergic neurons is transported

retrogradely (S23) makes it possible that the intra-OFC injection of (192)IgG-saporin

led to dysfunction of additional brain regions compared to our targeted

pharmacological inactivation of the OFC using CNQX or to the irreversible OFC

excitotoxic lesion approach used by Smith et al. (S21). While Ross and colleagues

have examined the status of cholinergic neurons in a few brain regions in close

vicinity to the OFC (i.e. anterior cingulate cortex, insular cortex) and reported a lack

18

of effect of the toxin, it cannot formally be excluded that cholinergic neurons in a

broader network of cortical and subcortical regions were affected due to the fact that

cholinergic neurons in the cortex exhibit complex axonal branching (S24). Thus,

additional cholinergic depletions may have led to memory deficits in the STFP task

unrelated to OFC dysfunction (but see S21 for additional discussion).

Does the cortex exert a top-down inhibitory control over hippocampal function upon retrieval of remote associative memory?

Partial versus extensive inactivation of the OFC at the time of encoding

differentially modulated the kinetic of hippocampal-cortical activation during the 30

days post-learning period (Experiment 9, Fig. 2G, H). In control rats injected with

aCSF, systems-level consolidation of associative olfactory memory involved the

gradual disengagement of the hippocampus associated with a concomitant

recruitment of OFC. Concurring with the hippocampal pattern of activation seen in

our imaging approach (fig. S2), hippocampal Fos expression was reduced below

control level after the remote memory test, suggesting that hippocampal activity may

be inhibited during remote memory retrieval.

Previous findings have led us to suggest the possibility that upon remote

memory retrieval, the cortex may exert a top-down inhibitory feedback over

hippocampal function to prevent encoding of redundant information already

consolidated in the cortex (S4). In line with this possibility, it is interesting to note that

in rats with a previously inactivated OFC upon encoding and which did not express

remote memory at Day 30, hippocampal activity remained significantly above control

levels at the 30-day remote time point. This may reveal an absence of cortical

inhibitory feedback due to a degraded or impaired remote memory embedded within

cortical networks consecutive to an altered post-learning hippocampal-cortical

dialogue.

A link between epigenetic transcriptional changes and structural synapse-specific mechanisms

Our pharmacological approach using histone deacetylase (HDAC) inhibitors

infused into the OFC upon encoding translated into improved remote memory and

enhanced structural plasticity within this cortical region. This suggests that cell-wide

epigenetic transcriptional changes are capable of impacting on synapse-specific

19

mechanisms, possibly by modulating the expression and life span of experience-

relevant cortical tags which will in turn support the progressive changes in the

connectivity of specific cortical networks (via ‘weight’ and ‘wiring’ plasticity) during the

course of systems-level memory consolidation. Alternatively, potentiating histone

acetylation might enhance transcription (quantitatively and/or in terms of duration)

and result in an increase of the pool of available RNAs coding for plasticity-related

proteins to be captured by activated (“tagged”) synapses. These proteins could in

turn facilitate the strengthening, or weakening, of synapses in a site-specific manner.

Note that these findings do not mean that epigenetic marks in the cortex

underlie the hippocampal-cortical dialogue. They only represent a regulatory

mechanism involved in the initial setting of the appropriate cortical tags and it is the

status of these tags which may in turn modulate hippocampal inputs to the cortex

(the so-called hippocampal-cortical dialogue), and in fine ensure proper stabilization

of synaptic weights within cortical networks. Thus, epigenetic mechanisms could act

as a permissive mechanism for hippocampal influence onto cortical networks but

would not be modulated by hippocampal activity per se.

We can only speculate that the expression of an array of genes coding for

specific tags might be altered by intra-OFC injections of the HDAC inhibitors sodium

butyrate or Trichostatin A. Interestingly, and contradictory to the assumption that a

large number of memory genes might be altered non specifically by these

compounds, a study by Vecsey and collaborators (S25) suggests that the effect of

HDAC inhibition is mediated in a gene-specific manner. However, it is still unclear

how one specific gene region can be targeted for acetylation or any other type of

epigenetic marks and translates into a change in memory function (but see S26 for

putative actions of cell-wide epigenetic marks on neuronal function, e.g. these marks

can affect the responsivity of neurons making them more or less sensitive to existing

inputs and capable of stabilizing synaptic weights).

Future experiments using genome-wide analyses of gene transcription

coupled to various classes of HDAC inhibitors will be required at various time points

following training to identify, in consolidation-relevant brain regions (e.g.

hippocampus and cortex), which effector genes and signalling pathways represent

key targets involved in the cortical tagging process and the associated changes in

structural plasticity.

20

Is early cortical tagging a process employed by all types of associative memories?

While our findings point to the existence of early tagging of cortical network as

a gateway for the formation of enduring memories, future studies will need to explore

whether this neurobiological process that can drive the formation of all types of long-

lasting associative memories (presumably in specific cortical regions), be they

appetitive or aversive in nature. For instance, just as for the appetitive social

transmission of food preference task in our study, Fos expression was also found to

be increased in the anterior cingulate cortex early (90 minutes) after acquisition of

fear conditioning training (S27), raising the possibility that early modifications of

cortical synaptic activity, and thereby cortical tagging, are susceptible to occur upon

acquisition of the fear conditioning procedure to subsequently drive structural

plasticity and the formation of enduring aversive remote memories in the anterior

cingulate cortex (S28).

21

Fig. S1. Inactivation of the dorsal HPC prior to social interaction induces anterograde amnesia. (A) Rats infused with CNQX into the HPC were impaired when tested 7 days later compared to rats injected with artificial cerebrospinal fluid (aCSF) used as vehicle (F1,17 = 7.84; *p < 0.05 versus aCSF). Their performance was no longer different from that of pooled food preference control rats (dotted line) injected with either aCSF or CNQX (p = 0.14, NS). (B) Total food eaten during test at Day 7 was similar across all groups (F < 1). n = 4-6 rats per group.

020406080

100

Day 7

% c

umin

eat

en aCSF

CNQX*

A

02468

10

Day 7

Tota

l foo

d ea

ten

BHPC HPC

22

600700800900

100011001200

Fos

coun

ts

020406080

100

Day 1 Day 30

% c

umin

eat

enFP

EXP

020406080

100

Day 1 Day 30

% c

umin

eat

enFP

EXP

020406080

100

Day 1 Day 30

% c

umin

eat

en

aCSFTTX

02468

10

Day 1 Day 30

Tota

l foo

d ea

ten

02468

10

Day 1 Day 30

Tota

l foo

d ea

ten

C

* *

02468

10

Day 1 Day 30

Tota

l foo

d ea

tenFP

EXP

50

100

150

200

Day 1 Day 3050

100

150

200

Day 1 Day 30Fos

coun

ts (%

cont

rol) OFC PAR

50

100

150

200

Day 1 Day 3050

100

150

200

Day 1 Day 30Fos

coun

ts (%

cont

rol) dorsal HPC ventral HPC

B

*†

*

*†

* *

*

*†

Day 1

Day 30

HPC OFC

HPC

*

aCSFTTXaCSFTTX

020406080

100

Day 1 Day 30

% c

umin

eat

en

aCSF

TTXOFC

*aCSFTTXaCSFTTX

D

OFC

E OFC

EXPFPEXPFP*

A

Fig. S2. Time-dependent involvement of HPC and OFC in remote memory storage and retrieval of associative olfactory memory. (A) Recent (Day 1) and remote (Day 30) memory performance of experimental (EXP) and food preference (FP) controls tested in the STFP paradigm (left panel). After training, preference for the familiar food increased to ~80% and was significantly above chance (F1,42 = 341.83; p <

23

0.0001 versus FP rats). This acquired preference for cumin did not differ when tested either 1 or 30 days following training and was not due to a difference in appetitive motivation between groups as the amount of total food eaten by EXP and FP rats was similar (right panel, F < 1); *p < 0.01 versus FP. (B) Corresponding temporal patterns of Fos counts relative to controls in HPC (dorsal and ventral parts), OFC and parietal cortex (PAR) (left panel). ANOVA revealed region-specific changes within hippocampal-cortical networks as a function of the age of the associative olfactory memory (brain region x time x group interaction: F3,176 = 53.66, p < 0.0001). Increasing the retention interval from 1 day to 30 days resulted in a significant decrease in Fos protein expression in the HPC associated with an increased Fos expression in the OFC. It should be noted that Fos expression in the HPC at Day 30 was significantly lower than that of paired control subjects, which raises the possibility that inhibitory influences may ultimately control the level of engagement of the hippocampal formation in memory consolidation (S5). Not all cortical regions are expected to be disproportionally involved in processing associative olfactory information acquired either recently or remotely. One such region is the parietal cortex which has been shown to be predominantly involved in processing spatial information (S29). Accordingly, no significant time-dependent change in neuronal activity was observed in this region, contrasting with those observed in the HPC and OFC. Photomicrographs showing increased Fos-positive nuclei in OFC on day 30 as compared with day 1 (right panel). *p < 0.01 versus FP (100% dotted line); †p < 0.01 versus Day 1. Scale bar, 100 μm. (C) Differential effects of targeted pharmacological inactivation of HPC and OFC by tetrodotoxin (TTX) on recent and remote memory retrieval. Blocking the propagation of action potentials in the HPC impaired recently acquired memory while sparing remote memory (interaction treatment x delay: F1,35 = 15.33, p < 0.001). Conversely, inactivating the OFC selectively blocked retrieval of remote memory (interaction treatment x delay: F1,43 = 5.0, p < 0.05). Impaired rats performed similarly as respective FP controls (dotted line, innate preference, (p > 0.30 for all comparisons, NS). *p < 0.01 versus aCSF. (D) For a given region (HPC or OFC), the total food eaten by rats injected with aCSF or TTX during recent (Day 1) and remote (Day 30) memory tests was similar across groups (a delay-dependent effect on memory performance was observed in the absence of any treatment effect, p > 0.10 for all comparisons, NS), indicating that a change in appetitive motivation was not a confounding factor in these pharmacological experiments. (E) Experimental (EXP) rats showed increased Fos expression in the OFC compared to food preference (FP) controls 90 minutes after social interaction (F1,9 = 6.58; *p < 0.05). This suggests that the OFC is activated upon encoding but does not participate in recent memory retrieval (Day 1, panels B, C), confirming previous findings (S30, S31). The OFC however exhibits a growing importance over time and becomes involved in processing remote associative olfactory memory (Day 30, panels B, C). n = 4-13 rats per group.

24

020406080

100

Day 1

% c

umin

eat

en

aCSF

CNQX

020406080

100

1 h 24 h 72 h

% c

umin

eat

en

aCSF

CNQX

HPC

OFC

02468

10

1 h 24 h 72 h

Tota

l foo

d ea

ten

CNQX or aCSF

Day 30TESTat 72h

TESTat 24h

TESTat 1h

or or

OFC

A B

02468

10

Day 1

Tota

l foo

d ea

ten

aCSFCNQXaCSFCNQX

*

*

Fig. S3. Effects of hippocampal and cortical infusions of CNQX on retrieval of recent and remote olfactory associative memory. (A) Just like for TTX (fig. S2, C-D), rats infused into the HPC with CNQX prior to recent memory retrieval at Day 1 were impaired compared to rats injected with aCSF (F1,12 = 7.56, p < 0.05). Total food eaten was similar across groups (F < 1). (B) Experimental design used to assess the pharmacological reversibility of CNQX injected into OFC prior to remote memory retrieval at Day 30 (top). Concurring with the effects of TTX on remote memory (Fig. S2, C-D), animals were severely impaired when tested 1 hour after CNQX infusion compared to aCSF rats (F1,12 = 28.22, p < 0.001). This deleterious effect was time-limited as rats tested at 24h or 72 h post-infusion were no longer impaired (F2,27 = 2.56; p > 0.09), confirming the previously reported time-limited suppression of neuronal activity induced by CNQX (S32)(middle). Total food eaten was similar across groups (p>0.20 for all comparisons, NS)(bottom). *p < 0.05 versus aCSF, n = 3-11 rats per group.

25

0,8

0,9

1,0

1,1

1,2

Day 1 Day 30Cor

tical

spi

nes

/ µm FP

EXP

APICAL BASAL

60% 20% 63% 30%

37% 57% 30% 57%

3% 23% 7% 13%

Day 1 Day 30Day 30 Day 1

LowIntermediate

High

0,8

0,9

1,0

1,1

1,2

Day 1 Day 30Cor

tical

spi

nes

/ µm

****††

C

A B

ERK2

SYN

DAY 1 DAY 30

OFC00,5

1

1,52

EXP0

0,30,60,91,2

EXP

SY

N /

ER

K

OFC

*Day 1Day 30Day 1Day 30

PAR

FPEXPFPEXP

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Fig. S4. (A) Differential levels of expression of the presynaptically localized protein synaptophysin (SYN) in rats tested for retrieval of recent (Day 1) and remote (Day 30) associative olfactory memory. Western blot analysis of total-extracted lysates obtained from OFC showing increased expression of SYN (normalized to ERK2) in experimental rats (EXP) tested at Day 30 compared to Day 1 (t(13) = 3.72, *p < 0.01). Note that the increase in SYN expression was specific to the OFC as it was not observed in the PAR (p > 0.05, NS). (B) Representative Western blots from the same loading showing SYN and ERK2 from OFC extracts of rats sacrificed upon completion of retrieval testing at Day 1 or Day 30 (2 rats per condition are shown). Note that total ERK2 levels were not affected at these delays. (C) Increased dendritic spine density along apical (left panel) and basal (right panel) dendrites of pyramidal neurons of the OFC occurred over time in experimental (EXP) rats submitted to social interaction but not tested for retrieval at Day 1 or Day 30 (*p < 0.01 versus FP; apical dendrites: †p < 0.05 versus Day 1 and basal dendrites: p value close to reaching significance, p = 0.059 versus Day 1). Corresponding distribution of cortical neurons with low, intermediate and high density of dendritic spines is shown below. Note the change in the proportion of these three neuronal populations as memory matured over time. n = 5-9 rats per group.

These findings corroborate the view that structural plasticity, namely cortical rewiring of previously connected or unconnected cell assemblies, is at least in part one generic form of synaptic plasticity responsible for the formation and long-term storage of enduring memories in the cortex (S10). Furthermore, the observation that modifications in structural plasticity in the OFC occurred in both rats tested (synaptophysin experiment) and not tested (Golgi-Cox experiment) for memory retrieval indicates that rewiring of cortical networks can be achieved prior to memory reactivation. In agreement with previous observations (S28), this suggests that offline neural activity triggered by initial learning is sufficient to ensure memory storage during the course of the systems-level consolidation process.

26

02468

10

EARLY LATE

Tota

l foo

d ea

ten aCSF

CNQX

HPC

Early Late02468

10

EARLY LATE

Tota

l foo

d ea

ten OFC

Early Late

Day 15 Day 30

EARLY(Day 1 to Day 13)

CNQX or aCSF TEST

Day 1

A

020406080

100

Day 30

% c

umin

eat

en

HPC OFCB C

*

020406080

100

Day 30

% c

umin

eat

en

*

aCSFCNQX

020406080

100

Day 30

% c

umin

eat

en

HPC OFCB C

*

020406080

100

Day 30

% c

umin

eat

en

*

aCSFCNQXaCSFCNQX

Fig. S5. Total food eaten by rats injected chronically into the HPC or OFC with aCSF or CNQX during the early (Day 0 to Day 12) or the late (Day 15 to Day 27) post-acquisition periods and tested at Day 30 was similar across groups (F < 1). Thus, these chronic treatments did not affect rat motivation during STFP testing. n = 5-12 rats per group.

Fig. S6. Chronic inactivation with CNQX of HPC or OFC during the early post-acquisition period of the STFP task. (A) Experimental design. Note that the first infusion started one day following social interaction. (B) Concurring with data in Main Fig. 1, HPC or OFC chronic inactivation (from Day 1 to Day 13) similarly impaired remote memory retrieval examined at Day 30 compared to rats injected with aCSF (HPC: F1,9 = 17.17; *p < 0.01; OFC: F1,12 = 14.03 *p < 0.01 versus aCSF). n = 4-10 rats per group.

27

Fig. S7. Animals chronically inactivated in the HPC or OFC during the early or late post-acquisition periods of the STFP task can relearn. (A) Experimental design. Animals infused with aCSF or CNQX into the HPC during the early post-acquisition period were submitted to a second interaction one week later (Day 37). Memory for cocoa was assessed 7 days later (Day 44, choice between cocoa and cinnamon). Groups previously injected with aCSF or CNQX exhibited a similar acquired preference for cocoa (F < 1). (B) Experimental design. Animals infused with aCSF or CNQX into the OFC during the early or late post-acquisition periods were submitted to a second interaction one week later (Day 37). Memory for cocoa was assessed this time 30 days later to enable the establishment of remote memory in this region (Day 67, choice between cocoa and cinnamon). Groups previously injected with aCSF or CNQX during the early or the late post-acquisition periods exhibited a similar acquired preference for cocoa (F < 1). Dotted line represents innate preference of rats for cocoa. n = 3-10 rats per group.

020406080

100

EARLY

% c

ocoa

eat

en

PreviousaCSF

PreviousCNQX

020406080

100

EARLY

% c

ocoa

eat

en

PreviousaCSF

PreviousCNQX

Day 30 Day 37 Day 44

Test(cumin/thyme)

Second interaction

(cocoa)

Test(cocoa/cinnamon)

Day 30 Day 37 Day 67

Test(cumin/thyme)

Second interaction

(cocoa)

Test(cocoa/cinnamon)

//

A

B

HPC

PreviouslyCNQX

PreviouslyaCSF

PreviouslyCNQX

PreviouslyaCSF

OFC

020406080

100

EARLY LATE

% c

ocoa

eat

en

PreviousaCSF

PreviousCNQX

PreviouslyaCSF

PreviouslyCNQX

PreviouslyaCSF

PreviouslyCNQX

PreviouslyEarly

PreviouslyEarly

PreviouslyLate

28

020406080

100

Day 7

% c

umin

eat

en

aCSF

CNQX

OLF*

HPC

5001000

15002000

2500

AcH

3 co

unts

FP

EXP

1h

HPC

5001000

15002000

2500

AcH

3 co

unts

FP

EXP

1h500

1000

15002000

2500

AcH

3 co

unts

FP

EXP

1h

PAR

500

10001500

2000

2500

AcH

3 co

unts

FPEXP

1h

PAR

500

10001500

2000

2500

AcH

3 co

unts

FPEXP

1h500

10001500

2000

2500

AcH

3 co

unts

FPEXP

1h

*

Fig. S8. Rats which have smelled cumin only (OLF group) without interacting with a demonstrator did not exhibit an acquired preference for this flavor when tested 7 days later. Performance of the OLF group did not differ from that of food preference rats (dotted line) (p > 0.07, NS). Moreover, its performance was significantly lower than that exhibited by experimental rats injected into the OFC with aCSF or CNQX upon social interaction (F2,24 = 6.21; p < 0.01). *p < 0.02 versus aCSF and CNQX groups. n=7-11 rats per group.

Thus, rats injected with CNQX, whose performance was not different from aCSF controls rats at Day 7 (data presented here and in main Fig. 2A), did express a STFP-induced associative memory and not merely an olfactory memory for cumin.

Fig. S9. STFP-induced levels of histone H3 acetylation in the dorsal HPC and PAR. Experimental rats (EXP) showed increased level of histone H3 acetylation in the HPC one hour following social interaction compared to food preference (FP) rats (F1,10 = 8.76; *p < 0.05) but not in the PAR cortex (F1,10 = 1.45; p > 0.26). n = 6 rats per group.

29

500

1000

1500

2000

2500

1h

AcH

3 co

unts

aCSF

H89

*

A B

500

1000

1500

2000

2500

4h

AcH

3 co

unts

aCSF

NaB

aCSF

NaB*

Fig. S10. Effects of the MSK1 inhibitor H89 and histone deacetylase inhibitor NaB on levels of histone H3 acetylation in the OFC. (A) Rats infused with H89 into the OFC prior to social interaction showed reduced level of histone H3 acetylation compared to rats injected with aCSF at 1 h post-acquisition (F1,8 = 37.41; *p < 0.001 versus aCSF). (B) This graph is reproduced from main Figure 3 and is shown for comparison of the levels of histone H3 acetylation in aCSF rats assessed at 1 h and 4 h post-acquisition. Note that the SFTP-induced increase in acetylation was transitory. n = 5-6 rats per group.

Fig. S11. Blocking the MAPK/ERK signaling pathway in the OFC upon encoding altered the late development of structural plasticity in this region at Day 30. (A) Infusion of the MEK inhibitor U0126 or the MSK1 inhibitor H89 prior to social interaction resulted in reduced expression of synaptophysin in the OFC. A similar effect was observed after OFC infusion of the NMDA receptor antagonist AP-5 upon encoding (F3,25 = 5.85; p < 0.01). Synaptophysin fluorescent labelling was normalized with respect to the number of cells labeled with DAPI (% (SYN / DAPI)). *p < 0.05 versus aCSF.(B) Corresponding photomicropgraphs taken at the level of the OFC. Nuclei were stained with DAPI (blue); Synaptophysin labeling appears in green. Scale bar, 10 μm. n = 5-12 rats per group.

0

10

20

30

Day 30

acsfU0126H89AP-5

*

% (S

YN /

DAP

I )

* *

A B

aCSF U0126

AP-5H89

aCSF U0126

AP-5H89

30

Fig. S12. Effects of the histone deacetylase inhibitor Trichostatin A (TSA) on remote memory retrieval and levels of histone H3 acetylation in the OFC. (A) Experimental design used to investigate the effects of maintaining in the OFC the level of histone H3 acetylation during the early STFP post-acquisition period on remote memory retrieval at Day 30. Just as for sodium butyrate (Fig. 3D), intra-OFC infusion of TSA improved remote memory compared to control rats injected with vehicle (aCSF + DMSO)(F1,16 = 5.36; *p < 0.05). n = 7 rats per group. (B) The SFTP-induced increase in acetylation at 1 h post-interaction was transitory as it decreased over a 4 h post-interaction period. Intra-OFC infusion of TSA immediately upon completion of social interaction prevented such a natural decrease over time, confirming the efficacy of the dose used in the behavioural study shown in panel A (F2,14 = 43.79; p < 0.0001). *p < 0.0001 versus vehicle 4 h. n = 5-6 rats per group.

500

1000

1500

2000

2500

1

AcH

3 co

unts

0

20

40

60

80

100

Day 30

% c

umin

eat

en

*Vehicle 4hTSA 4h

Vehicle 1hVehicle 4hTSA 4h

Vehicle 1hBA

TSA or Vehicle

Day 0 Day 15 Day 30Early

OFC

*

*

Vehicle TSA Vehicle TSA

31

Fig. S13. Effects of chronic infusion of NaB into the OFC during the early or late post-acquisition periods on levels of synaptophysin (SYN) expression in this region. Only infusion of NaB during the early post-acquisition period resulted in an increase of SYN expression in the OFC compared to aCSF rats (F1,8 = 5.09; *p < 0.05). n = 5 rats per group. Note however that the observed rewiring of the connectivity between neurons (i.e. ‘wiring’ plasticity as revealed by the SYN marker) was not correlated with remote associative memory performance at Day 30 (r = 0.30; p > 0.12, NS). This suggests that in addition to affecting wiring cortical plasticity, maintaining elevated levels of histone H3 acetylation-induced by SFTP task during the early hippocampal-cortical dialogue may have also favored changes in the efficacy of synaptic transmission between existing synapses (‘weight’ plasticity), these two forms of plasticity likely acting in concert to support the progressive embedding of remote memories into cortical networks (S10).

References

01020304050

EARLY LATE

aCSFNaB

01020304050

EARLY LATE

aCSFNaB

% (S

YN

/ D

API

)

Early Late

*

32

S1. G. Paxinos, C. Watson, The Rat Brain in Stereotaxic Coordinates (Academic

Press, San Diego, CA, 1998).

S2. R.E. Clark, N.J. Broadbent, S.M. Zola, L.R. Squire, J. Neurosci. 22, 4663 (2002).

S3. P. Alvarez, P.A. Lipton, R. Melrose, H. Eichenbaum, Learn. Mem. 8, 79 (2001).

S4. P.W. Frankland, B. Bontempi, Nat. Rev. Neurosci. 6, 119 (2005).

S5. T. Maviel, T.P. Durkin, F. Menzaghi, B. Bontempi, Science 305, 96 (2004)

S6. P. Vanhoutte et al., Mol. Cell. Biol. 19, 136 (1999).

S7. E.M. Glaser, H. Van der Loos, J. Neurosci. Methods. 4, 117 (1981).

S8. R. Gibb, B. Kolb, J. Neurosci. Methods. 79, 1 (1998).

S9. C.H. Horner, E. Arbuthnott, J. Anat. 177, 179 (1991).

S10. D.B. Chklovskii, B.W. Mel, K. Svoboda, Nature 14, 782 (2004).

S11. A. Fischer et al., J. Neurobiol. Learn. Mem. 87, 149 (2007).

S12. S.P. Davies, H. Reddy, M. Cavano, P. Cohen, Biochem. J. 351, 95 (2000).

S13. J.P. Aggleton, M.W. Brown, Trends Cogn. Sci. 10, 455 (2006).

S14. U. Frey, R.G. Morris, Trends Neurosci. 21, 181 (1998).

S15. A. Barco, M. Lopez de Armentia, J.M. Alarcon, Neurosci. Biobehav. Rev. 32, 831 (2008).

S16. K.C. Martin, K.S. Kosik, Nat. Rev. Neurosci. 3, 813 (2002).

S17. S. Sajikumar, S. Navakkode, T.C. Sacktor, J.U. Frey,

J. Neurosci. 25, 5750 (2005).

S18. A. Govindarajan, R.J. Kelleher, S. Tonegawa, Nat. Rev. Neurosci. 7, 575 (2006).

S19. S.V. Puthanveettil et al., Cell 135, 960 (2008).

S20. M.A. Colicos, B.E. Collins, M.J. Sailor, Y. Goda, Cell 107, 605 (2001).

S21. C.A. Smith, B.S. East, P.J. Colombo, Behav. Brain Res. 208, 243 (2010).

S22. R.S. Ross, J. McGaughy, H. Eichenbaum, Learn. Mem. 12, 302 (2005).

S23. T. Ohtake et al., Neuroscience 78, 123 (1997).

S24. A. Duque, J.M. Tepper, L. Detari, G.A. Ascoli, L. Zaborszky, Brain Struct. Funct.

212, 55 (2007).

S25. C.G. Vecsey et al., J. Neurosci. 27, 6128 (2007).

S26. J.J. Day, J.D. Sweatt, Nature Neurosci. 13, 1319 (2010).

33

S27. P.W. Frankland, B. Bontempi, L.E. Talton, L. Kaczmarek, A.J. Silva, Science

304, 881 (2004).

S28. L. Restivo, G. Vetere, B. Bontempi, M. Ammassari-Teule, J. Neurosci. 29, 8206

(2009).

S29. R.P. Kesner, Neurobiol. Learn. Mem. 91, 197 (2009).

S30. R.S. Ross, H. Eichenbaum, J. Neurosci. 26, 4852 (2006).

S31. C.A. Smith, R.A. Countryman, L.L. Sahuque, P.J. Colombo, Neurobiol. Learn.

Mem. 88, 65 (2007).

S32. M. Day, R. Langston, R.G. Morris, Nature 424, 205 (2003).