Experimental evidence of the cellular engram theory

In the last century fear conditioning has been one of the most used models to study memory in animals. It has also been used as experimental framework to define and test the principles of engrams. In contextual fear conditioning mice learn to associate a context (novel environment, the CS) to an aversive stimulus (foot shocks, the US). Mice immediately respond to the shock with an innate behavior called freezing (CR). After the association is learned, exposure to the conditioned context alone (without US presentation), induces the conditioned response (Figure 13). This associational learning is measurable, context dependent and requires exploration of the context.

Figure 13. Fear conditioning paradigm in rodents, adapted from Maren et al. 2013.

A. Rodent exposed to a novel context will encode the context by exploring it and will not show freezing responses when they are represented with the same context. B. In contextual fear conditioning, the context (conditioned stimulus, CS) is associated with the presentation of foot-shocks (unconditioned stimulus, US), the rodent reacts to the shocks. When the rodent is presented to the CS, it shows freezing responses (conditioned response, CR), but exposure to a different context will not induce the CR. C. Legend explaining mice behaviors.

Recent studies in the field have used different strategies for the investigation of the cellular engram: the identification of neurons belonging to the engram (A) and the tagging and consequent manipulation of these cells (B) (Figure 14).

Figure 14. Different strategies for the study of memory engrams.

Left, identification: the mouse’s experience is translated into an increase in neuronal activity of a sub-population of cells that can be detected with specific markers such as IEGs. Right, tagging & manipulation:

transgenic mice combining IEGs promoters together with specific effector genes can be used to label and consequently manipulate specific sub-population of neurons activated by the mouse’s experience.

Identification Tagging &

Manipulation Allocation &

Manipulation

Neuronal activity markers (IEGs)

tTA

tTA* TetO ChR2 doxycycline c-Fos

+

neuronal activity

(CREB)

+

Experience Experience Experience

(A) Identification of the cellular engram:

Engram neurons are defined as being active during the encoding of a particular experience. As we have seen in section 4.2, the neuronal activity markers c-fos and Arc/arg3.1 can be used for the identification of recently activated neurons following specific learning and memory tests (Guzowski et al., 2005). Indeed, c-fos and Arc/Arg3.1 expression is very low in neurons of resting animals but is rapidly and transiently upregulated by synaptic activity in a fraction of neurons immediately after behavior.

Exploration of a novel environment induces c-fos and Arc/arg3.1 expression in the hippocampus in a region specific manner, reflecting different activity patterns in each region (Jenkins et al., 2003; Vazdarjanova et al., 2006; Per Andersen, 2007). Spatial exploration drives c-fos expression in only a small fraction of cells in the DG thus allowing a very clear identification of active neurons. In the CA3 and CA1 areas, the same behavior trigger c-fos expression in a higher number of cells (Stefanelli, 2014).

IEGs allow to test the first criterion for a cell population to be an engram: the learning-induced activity trigger molecular and functional cellular changes. For example, Nonaka et al. revealed that only a subset of amygdala neurons, identified by means of increased Arc/Arg3.1 expression, were recruited during contextual fear learning and showed increased synaptic plasticity (Nonaka et al., 2014). They found a pre-synaptic potentiation that was confirmed by an increase in mEPSC frequency and a reduction in the paired-pulse ratio. These learning-induced synaptic changes occurred only in amygdala fear-activated neurons. Another study performed by Ryan et al., confirmed using c-fos as activity marker, that fear labelled DG engram cells showed greater synaptic strength than non-engram cells (Ryan et al., 2015). They revealed that synaptic plasticity in c-fos+

engram cells was due to an increase in EPSC as well as by an increased AMPA/NMDA ratio, confirming that these synapses were potentiated.

The properties of IEGs have prompted researchers to developed transgenic mouse lines with IEGs promoters to drive the expression of other genes (i.e. LacZ, GFP,…) and allow in vivo permanent labelling of cells activated in a particular time window (Smeyne et al., 1992; Barth et al., 2004; Wang et al., 2006; Reijmers et al., 2007).

The permanent labeling of cells opened a door to test another aspect of engram definition: the specificity for a particular experience, by precisely determining the overlap between the neuronal ensembles activated by the exploration of a novel and a previously visited environment. Reijmers et al. showed that amygdala neuronal ensembles activated during the formation of a contextual fear memory were reactivated during the re-exposure of the conditioning context that trigger memory retrieval (Reijmers et al., 2007). Xie et al.

identified with a EGR1-EGFP transgenic mouse, that cortical layer II can form sparse and task-specific cell populations over a long period of time, confirming the role of the cortex for long-term memory storage (Xie et al., 2014).

Different studies have shown that recall induce reactivation of most of the cells of the engram in the DG and CA1 (Denny et al., 2014; Tanaka et al., 2014; Ramirez et al., 2013).

However, some recent works have revealed that time is a relevant criterion to consider when it comes to engram formation (Cai et al., 2016; Rashid et al., 2016). In these studies, it has been shown that memories for events occurring close in time will recruit similar cellular engrams because of the strong competition between them, while distinct cellular engrams will be selected for events occurring apart in time.

Altogether these studies have assessed two criteria defining the engrams: the cellular changes induced by specific behaviors by using IEGs as activity markers and their specificity to code different experiences across time. However, the use of IEGs based technologies to study the cellular engram raises three different concerns: 1. What is the mnemonic role of cells that do not express c-fos (i.e. INs or glia)? 2. What type of neuronal activity drives c-fos expression? 3. What are the limits of IEG’s temporal resolution? It is likely that some neurons are very active during phases of the process of memorization but do not express c-fos or detectable levels of c-fos protein (Estabrooke et al., 2001); as an example, there is evidence showing that during sleep, when memory consolidation takes place, c-fos expression is very low throughout the brain (with the exception of the ventrolateral preoptic area, (Cirelli and Tononi, 2000)), although we know that some sleep phases, like the rapid eye movement (REM) sleep, induce high neuronal activity (Estabrooke et al., 2001). If this is true, this methodology may neglect a great population of cells that participate in memory processes. In addition, IEGs have very rapid kinetics (between 15 minutes to hours for c-fos proteins expression), but still do not fully reflect the immediate reality of cell activity (these topics will be further developed in the discussion).

Altogether, although we still do not fully understand the relationship between brain activity and IEGs expression, IEGs have been extremely helpful to identify and study cellular changes induced by specific behaviors and their role in coding different experiences across time.

(B) Tagging and subsequent manipulations of the cellular engram:

To define the remaining criteria of engrams: sufficiency and necessity, different approaches are needed. Transgenic mice have also been used to manipulate active neuronal populations by coupling IEGs as promoters with functional effector molecules

(rhodospins like channelrhodopsins and halorhodopsin; ChR2, NpHR) to alter neuronal’

function. For example, Denny et al. have developed a transgenic mouse in which the expression of a tamoxifen-inducible Cre recombinase (CreERT2) is driven by the IEG Arc/Arg3.1 promoter (Denny et al., 2014). These techniques have been used to determine whether behaviorally activated ensembles of neurons are necessary and/or sufficient to recall the associated memory.

In 2012, Garner et al. introduced the hM3Dq DREADD receptor (designer receptor exclusively activated by designer drug) into neurons activated by sensory experience (Garner et al., 2012). This technique allowed to increase neuronal activity in hM3Dq expressing neurons in presence of the exogenous ligand clozapine-N-oxide (CNO). When an ensemble of neurons coding for a specific and neutral context was artificially activated during conditioning in a distinct context, mice formed a hybrid (false) memory representation of the second context. Memory recall of the conditioned context could occur only when hybrid memory ensemble was artificially reactivated (with CNO). Liu et al. in 2012 reported very similar results (Liu et al., 2012). They showed that optogenetic reactivation of DG neurons activated during contextual fear conditioning was sufficient to induce memory recall (freezing repose) in a different neutral context (Figure 15 E). They used a c-fos-tTA transgenic mouse to drive the expression of ChR2 in fear conditioned activated GCs (Figure 15 A-D). Light reactivation of these labelled GCs was sufficient to induce fear memory recall. A confirmation of their specific manipulation come with the fact that optogenetic re-activation of GC cells labelled in a neutral context did not induce fear memory recall (Figure 15 F-G). These findings suggest that activation of a specific sparse population of hippocampal GCs, engram neurons, is sufficient to induce the recall of the associated memory. However, the time window for the labelling of behaviorally activated cells in these studies is quite wide; from hours to days in some cases. Thus, the labelled population is very likely to be a mixed population of different active cells: those active before, during and after the behavior.

Figure 15. Optogenetic stimulation of a hippocampal engram activates fear memory recall, adapted from Liu et al. 2012.

a-b. c-fos-tTA mouse in combination with AAV9-TRE-ChR2-EYFP virus implanted with optical fibers in the DG. Removal of doxycycline form mice diet, will allow fear conditioning training to induce tTA expression which will bind to TRE and drive the expression of ChR2-EYFP in DG activated cells (yellow). c.

Experimental paradigm: mice under doxycycline diet are habituated for 5 days in context A with light stimulation. Mice were removed form doxycycline diet for 2 days and subsequently fear-conditioned in context B. Mice were immediately put back on doxycycline diet and tested for 5 days in context A with light stimulation. d. Fear-conditioned activated DG cells expressing ChR2-EYFP. e. Mice trained as shown in paradigm panel c, showed higher levels of freezing responses during bilateral 3-min light-on epochs during 5-day average memory recall (n 5 6, F1,10 5 85.14, ***P, 0.001). f. Experimental paradigm: mice were removed form doxycycline diet and exposed to an open field (OF) in context C. Mice were subsequently fear-conditioned in context B while back on doxycycline (OF-CF). g. Optogenetic stimulation of OF-labeled cells with ChR2-EYFP during memory test did not showed freezing responses (n 5 5; N.S., not significant).

In 2013, a follow-up study by Ramirez et al. showed that it is possible to create a false memory in mice (Ramirez et al., 2013). They labeled with ChR2 either DG or CA1 cells active in a particular context (context A, Figure 16 A-B). Mice were subsequently contextual fear conditioned (context B) while context A cells were optogenetically activated. (Figure 16 C; context A ChR2+ ensembles and non-labeled ensembles of context B). Exposure of mice to context A (A’) without optogenetic reactivation of DG cells resulted in increased fear responses compared to control mice, while exposure to a novel context (context C) did not induce any freezing response in conditioned mice (Figure 16 D). The hybrid (false) and genuine memories both recruited the amygdala (Figure 16 F-G).

This demonstrates that artificial activation of a neuronal ensemble can create a false memory and suggests that c-fos+ ensembles are sufficient to induce memory recall.

Habituation Exposure FC Testing

Figure 16. Creating a false memory in the hippocampus, adapted from Ramirez et al. (2013).

A. c-fos-tTA mouse in combination with AAV9-TRE-ChR2-EYFP virus implanted with optical fibers in the DG.

B. Removal of doxycycline form mice diet in combination with spatial exploration of a novel context induces the expression of ChR2-mCherry in DG GCs. C. Experimental paradigm: mice were removed form doxycycline diet and allowed to explore context A labeling DG or CA1 cells with ChR2-mcherry. Mice were immediately put back on doxycycline diet and fear-conditioned in context B with simultaneous light stimulation. Mice were subsequently fear-conditioned in context B. Mice were tested for freezing responses in context A (A’) and in a novel context C. The yellow lightning symbol indicate foot shocks while the blue shower symbol and blue light delivery; red circles represent neurons encoding context A labeled with ChR2-mCherry while gray and white circles represent neurons encoding context B and C, respectively; asterisks indicate neurons activated either by exposure to context or light stimulation. D. The graph shows mice freezing responses for the different contexts when DG cells were manipulated. E. Same as in D but CA1 cells were manipulated. F. The graph shows the percentage of c-fos positive cells in the BLA and CeA for three different groups of mice that underwent training as in panel C and were sacrificed after testing either in context B (natural recall), A (false recall) or in C (neutral context). G. Representative image of c-fos labeled cells for the recall of false memory.

Cowansage et al. demonstrated that optogenetic reactivation of a contextual fear memory ensemble in the retrosplenial cortex (RSC) was sufficient to induce context-specific memory recall (Cowansage et al., 2014). Hippocampal neuronal silencing during re-exposure to the conditioned context impaired fear memory recall, while simultaneous optogenetic reactivation of RSC fear-memory ensemble elicited normal contextual fear recall. These results suggest that, although the HPC is needed for the cued-induced contextual memory recall, neuronal ensembles in the cortex are able via other means to retrieve the stored fear memory. Fear-memory ensembles in the RSC are thus sufficient to induce memory recall.

Ryan et al. found that consolidation of learning resulted in increased synaptic plasticity in activated neurons. Interestingly, optogenetic reactivation of these ensembles resulted in memory retrieval, despite the chemical blockade of synaptic plasticity in these cells (Ryan et al., 2015). These findings suggest that memory-ensembles connectivity can overcome the blockade of synaptic plasticity mechanisms normally needed for memory retention and consolidation, again suggesting sufficiency for an engram to recall the associated memory.

All the above-mentioned studies have studied engram properties by activating the

B

C

(EYFP) instead of ChR2-mCherry in the DG that underwent the same behavioral schedule also showed increased freezing in context A (fig. S2A).

New experimental and control groups of mice were taken off Dox in context A in order to label activated cells and then placed in context C on the following day while back on Dox. In this experiment, although conditioning took place after the formation of both context A and context C memories, only those cells encoding context A were reactivated by light during fear condition-ing. Subsequently, all groups of mice displayed background levels of freezing in context C. In contrast, in the context A test the next day, the experimental group showed increased freezing levels as compared with those of the mCherry-only group, confirming that the recall of the false memory is specific to context A (Fig. 2G). This freezing was not observed in another ChR2-mCherry group that underwent the same behav-ioral protocol but without light stimulation during fear conditioning in context B, or in a group in which an immediate shock protocol was admin-istered in context B with light stimulation of con-text A cells (Fig. 2G and fig. S3). In a separate group of animals, we labeled cells active in con-text C rather than concon-text A and repeated similar

experiments as above. These animals showed freezing in context C but not context A (fig. S2B).

The hippocampus processes mnemonic infor-mation by altering the combined activity of sub-sets of cells within defined subregions in response to discrete episodes (11–13). Therefore, we in-vestigated whether applying the same parameters and manipulations to CA1 as we did to the DG could form a false memory. We first confirmed that light could activate cells expressing ChR2-mCherry along the anterior-posterior axis of the CA1 similar to the DG (fig. S1, J to R). Also similar to the DG (Fig. 2, A to E), the overlap of active CA1 cells was significantly lower across contexts (A and C) as compared with that of a reexposure to the same context (A and A). How-ever, the degree of overlap for the two contexts was much greater in CA1 (30%) than in the DG (~1%). When we labeled CA1 cells activated in context A and reactivated these cells with light during fear conditioning in context B, no increase in freezing was observed in the experimental group expressing ChR2-mCherry as compared with the mCherry-only control group in either context A or context C, regardless of whether the animals were exposed to context C or not before fear con-ditioning in context B (Fig. 2, M and N).

The simultaneous availability of two CSs can sometimes result in competitive conditioning; the memory for each individual CS is acquired less strongly as compared with when it is presented alone, and the presentation of two simultaneous CSs to animals trained with a single CS can also lead to decrement in recall (14). In our experi-ments, it is possible that the light-activated DG cells encoding context A interfered with the ac-quisition or expression of the genuine fear mem-ory for context B. Indeed, upon reexposure to context B, the experimental group froze signifi-cantly less than the group that did not receive light during fear conditioning or the group ex-pressing mCherry alone (Fig. 3A and fig. S4).

During light-on epochs in the context B test, freezing increased in the experimental group and decreased in the group that did not receive light during fear conditioning (Fig. 3A and fig. S2C).

We conducted similar experiments with mice in which the manipulation was targeted to the CA1 region and found no differences in the experi-mental or control groups during either light-off or light-on epochs of the context B test (fig. S5A).

Memory recall can be induced for a genuine fear memory by light reactivation of the corre-sponding engram in the DG (8). To investigate

Fig. 2. Creation of a false contextual fear memory. (AtoE) c-fos-tTA mice injected with AAV₉-TRE-ChR2-mCherry in the DG were taken off Dox and exposed to context A in order to label the activated cells with mCherry (red), then put back on Dox and exposed to the same context A [(A) and (C)] or a novel context C [(B) and (D)] 24 hours later so as to let activated cells express c-Fos (green). Images of the DG from these animals are shown in (A) to (D), and the quantifications are shown in (E) (n= 4 subjects each; ***P< 0.001, unpaired Student’sttest). Blue and red dashed lines indicate the chance level of overlap for A-A and A-C groups, respectively. (F) (Top) Training and testing scheme of animals injected with AAV₉-TRE-ChR2-mCherry or AAV₉ -TRE-mCherry. Various symbols are as explained in Fig. 1. (Bottom) Animals’

freezing levels in context A before fear conditioning and in context A and C after fear conditioning [n= 8 subjects for ChR2-mCherry group, andn= 6 subjects for mCherry group; ***P < 0.001, two-way analysis of variance (ANOVA) with repeated measures followed by Bonferroni post-hoc test]. (G) (Top) Training and testing scheme of animals injected with AAV₉

-TRE-ChR2-mCherry or AAV₉-TRE-mCherry. One control group injected with AAV₉ -TRE-ChR2-mCherry did not receive light stimulation during fear conditioning (ChR2-mCherry, no light). (Bottom) Animals’freezing levels in context A and C before and after fear conditioning (n= 11 subjects for ChR2-mCherry group, n= 12 subjects for mCherry, andn= 9 subjects for ChR2-mCherry, no-light groups; ***P< 0.001, two-way ANOVA with repeated measures followed by Bonferroni post-hoc test). (HtoL) Animals underwent the same protocol as in (A) to (E), except the virus injection was targeted to CA1. Representative im-ages of CA1 from these animals are shown in (H) to (K), and the quantifications are shown in (L) (n= 4 subjects each; *P= 0.009, unpaired Student’sttest).

(M) Same as (F), except the viral injection and implants were targeted to CA1 (n= 8 subjects for ChR2-mCherry and mCherry groups; n.s., not significant;

two-way ANOVA with repeated measures followed by Bonferroni post-hoc test). (N) Same as (G), except the viral injection and implants were targeted to CA1 (n= 6 subjects for ChR2-mCherry group andn= 5 subjects for mCherry group). Scale bar in (A) and (H), 250mm.

www.sciencemag.org SCIENCE VOL 341 26 JULY 2013 389

REPORTS

on February 23, 2017http://science.sciencemag.org/Downloaded from

D

E

whether this applies to a false fear memory, we examined fear-memory recall of experimental and control groups of mice in a distinct context

whether this applies to a false fear memory, we examined fear-memory recall of experimental and control groups of mice in a distinct context