Stress and glucocorticoid receptor-dependent mechanisms in long-term memory: from adaptive responses to psychopathologies

Abstract

A proper response against stressors is critical for survival. In mammals, the stress response is primarily mediated by secretion of glucocorticoids via the hypothalamic-pituitaryadrenocortical (HPA) axis and release of catecholamines through adrenergic neurotransmission. Activation of these pathways results in a quick physical response to the stress and, in adaptive conditions, mediates long-term changes in the brain that lead to the formation of long-term memories of the experience. These long-term memories are an essential adaptive mechanism that allows an animal to effectively face similar demands again. Indeed, a moderate stress level has a strong positive effect on memory and cognition, as a single arousing or moderately stressful event can be remembered for up to a lifetime. Conversely, exposure to extreme, traumatic, or chronic stress can have the opposite effect and cause memory loss, cognitive impairments, and stress-related psychopathologies such as anxiety disorders, depression and post-traumatic stress disorder (PTSD). While more effort has been devoted to the understanding of the effects of the negative effects of chronic stress, much less has been done thus far on the identification of the mechanisms engaged in the brain when stress promotes long-term memory formation. Understanding these mechanisms will provide critical information for use in ameliorating memory processes in both normal and pathological conditions. Here, we will review the role of glucocorticoids and glucocorticoid receptors (GRs) in memory formation and modulation. Furthermore, we will discuss recent findings on the molecular cascade of events underlying the effect of GR activation in adaptive levels of stress that leads to strong, long-lasting memories. Our recent data indicate that the positive effects of GR activation on memory consolidation critically engage the brain-derived neurotrophic factor (BDNF) pathway. We propose and will discuss the hypothesis that stress promotes the formation of strong long-term memories because the activation of hippocampal GRs after learning is coupled to the recruitment of the growth and pro-survival BDNF/cAMP response element-binding protein (CREB) pathway, which is well-know to be a general mechanism required for long-term memory formation. We will then speculate about how these results may explain the negative effects of traumatic or chronic stress on memory and cognitive functions.

1. Introduction

Stress triggers physiological responses that are necessary for organisms to adapt to a changing environment and respond to immediate perturbation, threat, or danger. Animals’ survival not only depends on their immediate response to a stressor, but also relies on their ability to memorize and integrate the information learned about the stressor in order to effectively respond to similar demands in the future.

In addition to the rapid physiological responses that include increases in blood pressure, heart rate, and pulmonary ventilation and a hypervigilance state, stress produces long-lasting changes in the central nervous system (CNS) that are responsible for memorization of the event. These reactions are governed by acute adrenergic neurotransmission in the sympathetic nervous system and, following activation of the hypothalamic-pituitary-adrenal (HPA) axis, the release of glucocorticoids from the adrenal glands. As a result, adrenergic neurotransmission and glucocorticoid secretion activate specific brain regions that include the hippocampus, amygdala, and prefrontal cortex. These regions are enriched in adrenergic and glucocorticoid receptors (GRs), which, in rodents as in humans are known to play critical roles in encoding, processing, and retaining the information of emotional events (de Kloet, Joels & Holsboer, 2005; Lupien, Maheu, Tu, Fiocco & Schramek, 2007; McIntyre, McGaugh & Williams, 2012; Roozendaal, Okuda, de Quervain & McGaugh, 2006).

The various parameters that characterize emotional experiences, such as arousal or stress intensity, duration, chronicity, predictability, and controllability are known to critically affect memory and cognition (Lupien, Maheu, Tu, Fiocco & Schramek, 2007). Whereas optimal levels of stress or arousal stimulate cognitive performance and the formation of a strong long-term memory by mediating and modulating consolidation, the process whereby an experience becomes a strong and long-lasting memory, exposure to severe or chronic stress can lead to cognitive impairments and the development of psychopathologies such as anxiety disorders, depression, and post-traumatic stress disorders (PTSD) (de Kloet, Joels & Holsboer, 2005; McEwen, 2000b). In this review, we will discuss the mechanisms by which one of the major pathways activated by stress, the release of glucocorticoids and activation of GRs, promotes long-term memory formation when stress and/or arousal are adaptive. Based on our recent data, we will propose the hypothesis that the activation of GRs promotes memory consolidation and strengthening because it critically and directly engages the activation of the brain-derived neurotrophic factor (BDNF)/cAMP response element-binding protein (CREB)-dependent pro-survival/growth pathway as a response to stress. We propose that this survival response to stress has been selected by evolution in brain cells as a fundamental mechanism that mediates memory storage. We will then summarize the knowledge of how high levels of stress and/or glucocorticoids may lead to memory impairments. We will conclude with speculations about how the knowledge of the molecular mechanisms activated by GRs in adaptive conditions may also explain the negative effects of stress on cognition, which lead to cognitive impairments and psychopathologies.

2. Stress, glucocorticoids and activation of GRs

2.1 Glucocorticoids and their receptors

Glucocorticoids are steroid hormones synthesized from cholesterol in the adrenal cortex. The predominant glucocorticoid in humans is cortisol and, in rodents, corticosterone. Their release fluctuates with circadian and ultradian rhythms made up of pulses at approximately hourly intervals. Cortisol and corticosterone are also the primary hormones responsible for the stress response because their release is regulated by the HPA axis activated in response to stress. In particular, glucocorticoids play a key role in restoring homeostasis following exposure to stress, and they also modulate important physiological responses such as ion transport, glycogenolysis, immune response, and memory.

Because of their lipophilic properties, glucocorticoids can cross plasma membranes and activate two different intracellular receptors, mineralocorticoid receptors, (MRs) and GRs, also known, respectively, as Type I and Type II receptors. MRs and GRs are homologous in their structural domains, which consist of the N-terminal transactivation domain (TAD), the DNA binding domain (DBD), and the C-terminal ligand binding domain (LBD) (Lu, Wardell, Burnstein, Defranco, Fuller et al., 2006). In the absence of ligand, cytoplasmic MRs and GRs are bound to chaperone protein complexes, including heat shock protein 70 (hsp70), heat shock protein 90 (hsp90), and FK506 binding protein 5 (FKBP5) (Grad & Picard, 2007). On ligand binding, the receptors undergo conformational changes that lead to their dissociation from the chaperone complexes, their homodimerization and nuclear translocation, or their binding to other cytoplasmic proteins. In the nucleus, both MRs and GRs can bind to specific sequences of 15 nucleotides in the promoter of target genes, known as the glucocorticoid response elements (GREs) and directly activate transcription of target genes (Zalachoras, Houtman & Meijer, 2013). Nuclear MRs and GRs can also interact with other transcription factors to control gene expression (Sandi, 2004). In addition, MRs and GRs can control rapid cellular responses by mechanisms that are independent of nuclear translocation and gene expression regulation, but instead occur through genomic-independent actions (Groeneweg, Karst, de Kloet & Joels, 2011; Prager & Johnson, 2009). We will further discuss the genomic and non-genomic mechanisms of GR below.

2.2 Stress-mediated secretion of glucocorticoids and activation of GRs in the brain

A stressful experience triggers the acute release of catecholamines (adrenaline and noradrenaline) from the sympathetic nervous system, as well as activation of the HPA axis by first engaging secretion of corticotropin-releasing hormone (CRH) from the paraventricular nucleus of the hypothalamus, then adrenocorticotropic hormone (ACTH) from the anterior lobe of the pituitary gland, and finally glucocorticoids from the adrenal cortex into the blood circulation. Once the stress ends, hormonal levels return to homeostasis by the negative feedback action of glucocorticoids on the HPA axis. In addition to the immediate reaction to the stressor by catecholamines and glucocorticoids, which evoke rapid physical responses (e.g., “fight or flight” response in the case of a threat), the release of glucocorticoids activate MRs and GRs in the brain. MRs and GRs are ubiquitously expressed throughout the brain in both glial cells and neurons, with highest levels in the hippocampus and amygdala, two areas that play critical roles in memories of fear, threat, and stressful experiences (de Kloet, Joels & Holsboer, 2005; Lupien, Maheu, Tu, Fiocco & Schramek, 2007). Importantly, MRs have a tenfold higher affinity for glucocorticoids than GRs and are largely occupied by the ligand in basal conditions, whereas GRs occupation highly depends on increases in glucocorticoid levels following stress response (de Kloet, Joels & Holsboer, 2005; Lupien, Maheu, Tu, Fiocco & Schramek, 2007). In this review, we will mainly discuss the mechanisms mediated by GRs.

For many years, the effects of glucocorticoids on synaptic plasticity and memory were believed to result exclusively from the classical genomic-dependent pathway of GR activation. However, it has recently been shown that many effects of GRs are also mediated by rapid, genomic-independent mechanisms (Groeneweg, Karst, de Kloet & Joels, 2011; Prager & Johnson, 2009).

As mentioned earlier, the genomic action of GRs occurs in the nucleus, where these receptors can directly activate transcription by binding to the GREs in the promoter of target genes (Karst, Karten, Reichardt, de Kloet, Schutz et al., 2000; Prager & Johnson, 2009). However, GRs can control gene expression also by interacting with other transcription factors, including activator protein 1 (AP1), nuclear factor κB (NF-κB), transcription factor IID (TFIID), signal transducer and activator of transcription 5 (STAT5), and CREB (Sandi, 2004). Studies using DNA microarray or serial analysis of gene expression (SAGE) in cultured hippocampal neurons or the hippocampus in vivo demonstrated that activation of GRs leads to the transcription of various genes, including calcium binding proteins, synaptosomal-associated proteins (SNAPs), neuronal cell-adhesion molecules (NCAMs), dynein, neurofilaments, β-actin, LIM domain kinase 1 (LIMK1) and profilin. These genes have key functions in intracellular signal transduction, metabolism, neuronal structure, synaptic plasticity, and memory, suggesting that, indeed, they may be target genes regulated by GR in long-term memory formation (Datson, Morsink, Meijer & de Kloet, 2008; Datson, van der Perk, de Kloet & Vreugdenhil, 2001; Morsink, Steenbergen, Vos, Karst, Joels et al., 2006; Sandi, 2004).

Although GR-mediated transcriptional activation is necessary for long-term synaptic changes in the hippocampus, studies have shown that genomic-independent actions of GRs rapidly control glutamate release and modulate synaptic transmission and plasticity (Groeneweg, Karst, de Kloet & Joels, 2011; Haller, Mikics & Makara, 2008; Prager & Johnson, 2009; Tasker, Di & Malcher-Lopes, 2006). In addition, several investigations provided evidence of genomic-independent action of GRs in modulation of the endocannabinoid system (Atsak, Roozendaal & Campolongo, 2012). While glucocorticoid-mediated release of endocannabinoids in the hypothalamus regulates activation and termination of the HPA axis (Di, Malcher-Lopes, Halmos & Tasker, 2003), endocannabinoid signaling in both the basolateral amygdala (BLA) and hippocampus appear to control cognitive processes such as emotional memory encoding (Atsak, Roozendaal & Campolongo, 2012; Hill, Patel, Campolongo, Tasker, Wotjak et al., 2010). In particular, it has been shown that genomic-independent mechanisms of GRs lead to activation of the endocannabinoid system in the BLA and hippocampus, which, in turn, enhances the consolidation of emotional memories (Bucherelli, Baldi, Mariottini, Passani & Blandina, 2006; Campolongo, Roozendaal, Trezza, Hauer, Schelling et al., 2009; de Oliveira Alvares, de Oliveira, Camboim, Diehl, Genro et al., 2005).

2.3 Non-genomic and genomic effects of GRs on glutamate transmission

Glucocorticoids are critical in modulating glutamatergic neurotransmission in several brain regions, including the hippocampus, amygdala, and medial prefrontal cortex (mPFC). Glucocorticoid-mediated regulation of the glutamatergic system engages rapid non-genomic action, as well as long-lasting genomic mechanisms controlled by GRs and directly affects synaptic transmission, plasticity, learning, and memory (Popoli, Yan, McEwen & Sanacora, 2012; Sandi, 2011).

First, glucocorticoids regulate glutamate transmission by non-genomic actions. Specifically, glucocorticoids rapidly enhance presynaptic glutamate release in the hippocampus, amygdala, and mPFC (Lowy, Gault & Yamamoto, 1993; Moghaddam, 1993; Venero & Borrell, 1999) via rapid non-genomic action of GRs (Musazzi, Milanese, Farisello, Zappettini, Tardito et al., 2010) as well as MRs (Karst, Berger, Turiault, Tronche, Schutz et al., 2005; Olijslagers, de Kloet, Elgersma, van Woerden, Joels et al., 2008). Glucocorticoids also rapidly modulate the trafficking of postsynaptic AMPA receptor subunits via genomic-independent mechanisms. Further, activation of MRs leads to lateral diffusion of GLUA1 and GLUA2 subunits to postsynaptic sites, thus increasing the frequency of hippocampal AMPA receptor-mediated current in CA1 neurons (Groc, Choquet & Chaouloff, 2008; Krugers, Hoogenraad & Groc, 2010). As a consequence, the rapid non-genomic effects of glucocorticoids on glutamate neurotransmission increase the frequency of miniature excitatory postsynaptic currents (mEPSCs) in hippocampal and amygdala neurons (Karst, Berger, Erdmann, Schutz & Joels, 2010; Olijslagers, de Kloet, Elgersma, van Woerden, Joels et al., 2008), thereby promoting long-term memory formation (Yuen, Liu, Karatsoreos, Feng, McEwen et al., 2009; Yuen, Liu, Karatsoreos, Ren, Feng et al., 2011).

Second, glucocorticoids affect glutamate neurotransmission via genomic-dependent mechanisms (Yuen, Liu, Karatsoreos, Feng, McEwen et al., 2009; Yuen, Liu, Karatsoreos, Ren, Feng et al., 2011). For example, in cultured hippocampal neurons, glucocorticoids regulate surface expression of GLUA2 subunits by a genomic-mediated process that results in increased GLUA2-containing AMPA receptors at the synapse (Martin, Henley, Holman, Zhou, Wiegert et al., 2009). Although there is no evidence that this effect is due to direct genomic regulation of GRs on AMPA receptor subunits (Martin, Henley, Holman, Zhou, Wiegert et al., 2009), indirect mechanisms have been suggested. Specifically, GRs enhance transcription of the immediate early gene serum-glucocorticoid-inducible kinase 1 (SGK1), which in turn leads to activation of the Rab4-GDI complex that facilitates AMPA receptors recycling at the synapse (Liu, Yuen & Yan, 2010; Yuen, Liu, Karatsoreos, Ren, Feng et al., 2011). Increased synaptic expression of GLUA2-containing AMPA receptors then leads to lasting enhancement of hippocampal synaptic transmission, spine formation, and long-term memory (Conboy & Sandi, 2010; Passafaro, Nakagawa, Sala & Sheng, 2003; Saglietti, Dequidt, Kamieniarz, Rousset, Valnegri et al., 2007). In line with these findings, genomic-mediated effects of GRs also regulate calcium signaling in hippocampal neurons, thus contributing to stronger firing accommodation and high frequency long-term potentiation (LTP) (Joels & Karst, 2012). GR-mediated increases in intracellular calcium levels were found to be dependent on the activation of NMDA receptors (Takahashi, Kimoto, Tanabe, Hattori, Yasumatsu et al., 2002) and L-type voltage-gated calcium channels (VGCCs) (Chameau, Qin, Spijker, Smit & Joels, 2007).

Conversely, in conditions of high or prolonged stress, activation of GRs can have a negative effect on glutamatergic transmission through the activation of NMDA receptors (Coussens, Kerr & Abraham, 1997; Shors & Servatius, 1995). In particular, high glucocorticoid concentrations in the hippocampus lead to GR-mediated activation of extrasynaptic NR2B-containing NMDA receptors (Yang, Huang & Hsu, 2005). This mechanism, which also results in endocytosis of GLUA2-containing AMPA receptors, increases hippocampal long-term depression (LTD) and impairs spatial memory retrieval (Howland & Cazakoff, 2010).

3. Glucocorticoids and memory consolidation

As mentioned earlier, a long-lasting memory is formed through a process known as consolidation, which, over time, converts a new labile memory trace into a stronger one that is resilient to disruption (Dudai, 2012; McGaugh, 2000; Squire, Stark & Clark, 2004). Memory consolidation requires an initial phase of de novo gene expression and protein synthesis, which leads to long-term synaptic plasticity and morphological changes (Alberini, 2008; 2009; Kandel, 2001; Lamprecht, Farb, Rodrigues & LeDoux, 2006). Arousal and moderate levels of stress facilitate memory consolidation and, consistent with this, emotionally arousing experiences are better remembered than neutral ones (Phelps, 2006; Roozendaal & McGaugh, 2011). Extensive evidence indicates that the release of glucocorticoids induced by arousal or stress and the consequent activation of GRs in specific brain regions critically mediates memory consolidation and modulates memory retention (McGaugh & Roozendaal, 2002).

3.1 Glucocorticoids, GRs, and their effect on memory

The contribution of glucocorticoids to the regulation of memory was first found in adrenalectomized rats, which showed impaired corticosterone expression, as well as spatial and contextual fear memory deficits (Pugh, Tremblay, Fleshner & Rudy, 1997; Roozendaal, Portillo- Marquez & McGaugh, 1996). In line with these results, systemic inhibition of glucocorticoid synthesis by administration of metyrapone impairs contextual fear conditioning, as well as spatial and inhibitory avoidance (IA) memories (Cordero, Kruyt, Merino & Sandi, 2002; Roozendaal, Bohus & McGaugh, 1996). Moreover, contextual and auditory fear conditioning, spatial and novel-object recognition, and IA memories are all enhanced by posttraining administration of corticosterone or the synthetic glucocorticoid dexamethasone (Akirav, Kozenicky, Tal, Sandi, Venero et al., 2004; Pugh, Tremblay, Fleshner & Rudy, 1997; Roozendaal & McGaugh, 1996; Roozendaal, Okuda, Van der Zee & McGaugh, 2006). In humans, oral administration of cortisol during learning or within one hour of stimulus presentation strengthens declarative memory for neutral and emotionally arousing information (Abercrombie, Kalin, Thurow, Rosenkranz & Davidson, 2003; Buchanan & Lovallo, 2001), whereas inhibition of glucocorticoid synthesis reduces long-term declarative memory (Maheu, Joober, Beaulieu & Lupien, 2004).

As described earlier, glucocorticoid-mediated effects engage both MRs and GRs, which, however, seem to have different roles in the acquisition, storage, consolidation, and retrieval of arousing information. Activation of MRs regulates the initial phase of memory encoding, including the response to novelty, whereas GRs are important in memory consolidation (de Kloet, Oitzl & Joels, 1999; ter Horst, van der Mark, Arp, Berger, de Kloet et al., 2012). Indeed, systemic inhibition of GRs with antagonists, much like targeted disruption of GR in transgenic mice, suppresses long-term spatial and contextual fear memories (Conrad, Lupien & McEwen, 1999; Cordero & Sandi, 1998; Oitzl & de Kloet, 1992; Oitzl, Reichardt, Joels & de Kloet, 2001; Pugh, Fleshner & Rudy, 1997; Revest, Di Blasi, Kitchener, Rouge-Pont, Desmedt et al., 2005). The degree of GRs activation in addition to that of MRs is critical for the effect of stress on cognitive performance, with memory facilitation occurring when high affinity MRs are fully occupied and low affinity GRs only partially activated. Whereas intermediate activation of GRs is necessary for memory consolidation, saturation of GRs has been shown to lead to memory impairments (de Kloet, Oitzl & Joels, 1999; Lupien, Maheu, Tu, Fiocco & Schramek, 2007). These studies indicate that while a relatively limited amount of glucocorticoids leads to positive effects on cognition, high concentration and/or prolonged exposure to glucocorticoids produce impairing effects.

In line with these observations, several studies have investigated the role of both MRs and GRs in glucocorticoid-mediated synaptic transmission and hippocampal LTP (Diamond, Bennett, Fleshner & Rose, 1992; Joels & Krugers, 2007). In particular, activation of MRs was found to enhance synaptic potentiation and hippocampal LTP, whereas saturation of GRs after treatment with high doses of glucocorticoids attenuated LTP and enhanced LTD (McEwen & Sapolsky, 1995; Pavlides, Ogawa, Kimura & McEwen, 1996). Other evidence indicates that enhanced activation of GRs dampens the ability of hippocampal neurons to induce LTP and elevates the threshold for synaptic strengthening, suggesting that activation of GRs may play a role in reducing the accessibility of novel information to the same neural network (Diamond, Park & Woodson, 2004; Joels, Pu, Wiegert, Oitzl & Krugers, 2006; Wiegert, Pu, Shor, Joels & Krugers, 2005).

3.2 Spatial and temporal activation of GRs in memory formation and retrieval

Memory is encoded by the concerted interplay of several brain areas and networks that interact for proper memory acquisition, consolidation, and expression (McIntyre, McGaugh & Williams, 2012; Schwabe & Wolf, 2013). Glucocorticoids are critical regulators in the activation and cooperation of these areas (Roozendaal, 2003). For example, stress-mediated secretion of glucocorticoids and/or activation of GRs directly affects hippocampal functions, thus modulating the consolidation of several types of hippocampal-dependent memories, including spatial and contextual memories in rodents and declarative memory in humans (Donley, Schulkin & Rosen, 2005; Eichenbaum, 2000; Gabrieli, 1998; Kim & Diamond, 2002; Lupien & Lepage, 2001; Roozendaal & McGaugh, 1997a; Roozendaal, Nguyen, Power & McGaugh, 1999; Squire, 2004).

Stress-induced secretion of glucocorticoids also targets the amygdala (Kim, Lee, Han & Packard, 2001; Roozendaal & McGaugh, 1997a; Roozendaal, Nguyen, Power & McGaugh, 1999). In particular, activation of GRs in the amygdala is important for fear memory encoding and hippocampal modulation. Inhibition of GRs in the BLA impairs long-term spatial and contextual fear memories, whereas infusion of GR agonist into the BLA enhances IA-mediated memory (Donley, Schulkin & Rosen, 2005; Roozendaal & McGaugh, 1997b; Roozendaal, Quirarte & McGaugh, 2002). Here we should remind that in addition to glucocorticoids, adrenergic neurotransmission in the amygdala is key element in modulating fear memories, as shown by contextual and spatial memory enhancements after amygdala infusion of norepinephrine or β-adrenergic receptors agonists (Ferry, Roozendaal & McGaugh, 1999; Hatfield & McGaugh, 1999). In contrast, inhibition of adrenergic receptors in the amygdala blocks memory enhancement induced by systemic or intrahippocampal corticosterone injections (Quirarte, Roozendaal & McGaugh, 1997; Roozendaal, Okuda, Van der Zee & McGaugh, 2006). Hence, a concerted action of adrenalin/noradrenalin and glucocorticoid is critical for memory formation and modulation. Cortical areas such as the entorhinal, parietal, perirhinal, insular, and prefrontal cortices are modulated by BLA activity, and all contribute to fear memory consolidation (McGaugh, 2002). Recent evidence indicates that glucocorticoids have a direct effect in these regions, as activation of GRs in the insular cortex and mPFC enhances long-term memory consolidation (Barsegyan, Mackenzie, Kurose, McGaugh & Roozendaal, 2010; Fornari, Wichmann, Atucha, Desprez, Eggens-Meijer et al., 2012; Roozendaal, McReynolds, Van der Zee, Lee, McGaugh et al., 2009).

Notably, the influence of stress on memory retention largely depends on the phase of memory processing the arousing or stressful event is presented with. Thus, exposure to mild stress during or immediately after learning has a positive effect on the consolidation of long-term memory, typically when the stressor is part of the learning event (Roozendaal, 2000). However, in both humans and rodents, if stress is presented shortly before retrieval, impairments in memory retention can occur (Cahill, Gorski & Le, 2003; McGaugh & Roozendaal, 2002; Roozendaal, 2002). In particular, stress is known to have an inhibitory effect on memory retrieval, as shown, for example, in rodents that exhibit retrograde amnesia when exposed to a stressor before a spatial memory retention test (de Quervain, Roozendaal & McGaugh, 1998; Diamond, Park, Heman & Rose, 1999). Activation of glucocorticoid and noradrenergic pathways in the hippocampus and BLA are important in the impairing effects of stress on memory retrieval (de Quervain, Roozendaal & McGaugh, 1998; Roozendaal, Hahn, Nathan, de Quervain & McGaugh, 2004). Also, the severity of memory impairments has been correlated with the concentration of circulating plasma corticosterone at the time of testing (de Quervain, Roozendaal & McGaugh, 1998; Diamond, Park, Heman & Rose, 1999).

In humans, the results of many studies support the importance of stressor’s timing with regard to memory retention. Increases in cortisol levels during acquisition positively correlate with the strength of memory consolidation (Abercrombie, Kalin, Thurow, Rosenkranz & Davidson, 2003; Abercrombie, Speck & Monticelli, 2006; Smeets, Sijstermans, Gijsen, Peters, Jelicic et al., 2008; Zorawski, Blanding, Kuhn & LaBar, 2006), whereas stress exposure or administration of synthetic glucocorticoids before memory testing significantly impairs retrieval of declarative memories, particularly when they are associated with emotionally arousing information (de Quervain, Roozendaal, Nitsch, McGaugh & Hock, 2000; Kuhlmann, Piel & Wolf, 2005). Hence, the effect of stress on memory performance largely depends on the timing, context, and convergence of stress hormones in the brain (Joels, Fernandez & Roozendaal, 2011; Roozendaal, 2002).

4. GR-dependent molecular mechanisms critical for long-term memory consolidation

As we discussed earlier, in adaptive conditions, the release of glucocorticoids and activation of GRs mediate and modulate memory consolidation and the storage of strong and long-lasting memories of salient events. Which are the molecular mechanisms underlying the effect of glucocorticoids and GR activation on memory consolidation? Do GRs interact with the identified transcriptional, translation and post-translational mechanisms required for memory consolidation? The understanding of the molecular mechanisms mediated by GRs in promoting memory consolidation is still partial, possibly because of the complexity of GR-mediated responses and the multiple cell types and brain regions targeted. The identification of these mechanisms will provide important information for developing new strategies for memory strengthening or weakening, hence treating memory and cognitive disorders.

Mechanisms found to be activated by GR stimulation in cell culture or by pharmacological treatments do not necessarily inform about the identity and regulation of the mechanisms occurring in vivo after an adaptive stressful/arousing experience that will be consolidated into a long-term memory. It is in fact well known that GR activation is involved in many different stress conditions and is a highly regulated mechanism that can lead to many different types of responses depending on the stress levels, duration and type. Hence, it is important to identify which mechanisms occur physiologically in vivo after learning as a consequence of GR activation, and also characterize their temporal and spatial progression as well as regulation.

A few studies, including a recent one from our laboratory (Chen, Bambah-Mukku, Pollonini & Alberini, 2012) have demonstrated that, after learning, GRs regulate several intracellular signaling pathways known to be required for memory consolidation. These include the pathways activated by CREB, mitogen-activated protein kinase (MAPK), calcium/calmodulin-dependent protein kinase II (CamK II), and BDNF. In addition, GRs control epigenetic modifications that influence long-term memory. Specifically, activation of GRs by different types of psychological stress such as forced swimming or predator exposure increase the phosphorylation and acetylation of histone H3 in dentate gyrus granule neurons (Bilang- Bleuel, Ulbricht, Chandramohan, De Carli, Droste et al., 2005; Chandramohan, Droste, Arthur & Reul, 2008). The concomitant activation of GRs, NMDA receptors, and MAPK signaling pathway is required for phospho-acetylation of histone H3 in the dentate gyrus (Bilang-Bleuel, Ulbricht, Chandramohan, De Carli, Droste et al., 2005; Chandramohan, Droste, Arthur & Reul, 2008). Phospho-acetylation of histone H3 in turn modulates stress-induced behavioral responses such as immobility after forced swimming (Bilang-Bleuel, Ulbricht, Chandramohan, De Carli, Droste et al., 2005) and object recognition memory (Stefanko, Barrett, Ly, Reolon & Wood, 2009). In the insular cortex, activation of GRs enhances the interaction between phospho-CREB and CREB-binding protein (CBP), leading to the histone deacetylase (HDAC)-mediated chromatin modification necessary for memory consolidation for novel object recognition and object location (Roozendaal, Hernandez, Cabrera, Hagewoud, Malvaez et al., 2010).

Using fear conditioning in mice, two studies demonstrated the critical role of GRs in the induction of MAPK phosphorylation, expression of early growth response protein 1 (Egr-1), and phosphorylation of Synapsin-Ia/Ib during fear memory consolidation (Revest, Di Blasi, Kitchener, Rouge-Pont, Desmedt et al., 2005; Revest, Kaouane, Mondin, Le Roux, Rouge-Pont et al., 2010). More specifically, activation of hippocampal GRs recruits the MAPK signaling pathway, which in turn leads to induction of the downstream immediate early gene Egr-1 (also known as Zif268), which is a key transcription factor in memory consolidation (Jones, Errington, French, Fine, Bliss et al., 2001; Kelleher, Govindarajan, Jung, Kang & Tonegawa, 2004). In agreement, exogenous administration of glucocorticoids or constitutive activation of GR in transgenic mice results in activation of MAPK and induction of Egr-1 expression in the hippocampus. In line with these findings, inhibition of the MAPK pathway in the hippocampus abolishes the increase in contextual fear conditioning induced by glucocorticoids (Revest, Di Blasi, Kitchener, Rouge-Pont, Desmedt et al., 2005). GR-mediated activation of MAPK and Egr-1 then leads to the increased expression and phosphorylation of Synapsin-Ia/Ib necessary for contextual memory consolidation (Revest, Kaouane, Mondin, Le Roux, Rouge-Pont et al., 2010). Phosphorylation of Synapsin Ia/Ib has been shown to facilitate activity-dependent release of glutamate from presynaptic vesicles of pyramidal neurons (Chi, Greengard & Ryan, 2003; Jovanovic, Czernik, Fienberg, Greengard & Sihra, 2000). Collectively, these studies have identified the MAPK-activated pathway, with subsequent regulation of Egr-1 expression and phosphorylation of Synapsin-Ia/Ib, as a sequence of events targeted by activation of GRs that leads to fear memory consolidation.

Our laboratory has recently identified several critical molecular mechanisms underlying glucocorticoid-mediated long-term memory consolidation in the rat hippocampus using an IA learning paradigm. We found that recruitment of hippocampal GRs after training controls the activation of several intracellular pathways that are critical for memory consolidation (Chen, Bambah-Mukku, Pollonini & Alberini, 2012). Specifically, GRs control the rapid learning-dependent hippocampal increase of CREB phosphorylation and expression of the immediate early gene-activity-regulated cytoskeleton-associated protein (Arc), as well as the increase in synaptic phospho-CAMKIIα, phospho-Synapsin-1, and GluA1 expression. All of these rapid changes, except Arc induction, result from non-genomic actions of GRs (Chen, Bambah-Mukku, Pollonini & Alberini, 2012). These results extend previous findings that expression of Arc increases in hippocampal synapses following memory-enhancing administration of corticosterone (McReynolds, Donowho, Abdi, McGaugh, Roozendaal et al., 2010); that stress-dependent Arc expression is impaired in the hippocampus of GR-deficient (GR+/−) mice (Molteni, Calabrese, Chourbaji, Brandwein, Racagni et al., 2010); and that corticosterone increases AMPA receptor trafficking in pyramidal neuronal cultures from prefrontal cortex (Liu, Yuen & Yan, 2010). We also found that inhibition of GRs in rat hippocampus significantly reduces phosphorylation of the tropomysosin receptor kinase B (TrkB), extracellular-signal regulated kinase 1/2 (ERK1/2), Akt, and phospholipase C γ (PLCγ) (Chen, Bambah-Mukku, Pollonini & Alberini, 2012). Because these pathways are all canonical activation pathways downstream of BDNF, these findings suggest that the BDNF-dependent pathway is a key downstream effector of GR activation during memory consolidation.

In addition, we established that the long-lasting molecular modifications induced by training that are required for memory consolidation, including the persistent phosphorylation of CREB, CamKIIα, and Synapsyn1a, are all dependent on GR activation, most likely because they require the early and rapid molecular activation described just above. In fact, memory consolidation appears to be mediated by a BDNF-dependent autoregulatory loop that activates the CREB/CCAAT enhancer-binding protein (C/EBP)-dependent gene cascade, which in turn regulates BDNF expression. This BDNF-dependent autoregulatory loop is required to complete the consolidation necessary for long-term memory persistence (unpublished observations). Notably, we then found that intrahippocampal injections of BDNF, but not of other neurotrophins, such as nerve growth factor (NGF) and neurotrophin 3 (NT3), rescue both the molecular impairments and the amnesia caused by GR inhibition (Chen, Bambah-Mukku, Pollonini & Alberini, 2012). The effect is selective for GRs because BDNF does not rescue behavioral deficits caused by the inhibition of β-adrenergic receptors (Chen, Bambah-Mukku, Pollonini & Alberini, 2012), underscoring the critical and specific role of the BDNF-mediated signaling pathway in the GR-dependent molecular activations required for memory consolidation. Our findings extended to memory formation previous observations byJeanneteau et al. (2008) who reported that administration of dexamethasone in the hippocampus of rats or in hippocampal or cortical cell cultures induces TrkB phosphorylation (Jeanneteau, Garabedian & Chao, 2008). However, in these studies, TrkB transactivation was found to require a genomic effect mediated by GR, contrasting with the non-genomic effect of GRs on TrkB phosphorylation observed in our study (Chen, Bambah-Mukku, Pollonini & Alberini, 2012. See also discussion of this manuscript).

If hippocampal GR activation upon learning takes place upstream of the BDNF-dependent activation required for long-term memory, how does that occur? Although a direct interaction between GRs and TrkB has been reported in cortical neurons in vitro (Numakawa, Kumamaru, Adachi, Yagasaki, Izumi et al., 2009), our attempts do not yet indicate any direct interaction between GRs and TrkB in rat dorsal hippocampi in vivo. Hence, we speculate that the non-genomic effect of GRs on TrkB phosphorylation may target the regulation of BDNF release and/or TrkB membrane trafficking. GRs may also regulate BDNF-mediated signaling pathways by mechanisms that depend on cAMP-mediated trafficking and activation of TrkB (Ji, Pang, Feng & Lu, 2005). Moreover, it appears that the persistence of the long lasting molecular changes necessary for memory consolidation recruit additional genomic-dependent mechanisms such as BDNF transcription. Although there is no classical GRE sequence in the BDNF promoter, in situ hybridization demonstrated that BDNF mRNA expression is enhanced after intrahippocampal injection of corticosterone, presumably by the indirect transcriptional effects of GRs (Chao & McEwen, 1994; Hansson, Sommer, Rimondini, Andbjer, Stromberg et al., 2003). In contrast, expression of BDNF is altered in the hippocampus of GR-deficient mice (Alboni, Tascedda, Corsini, Benatti, Caggia et al., 2011; Ridder, Chourbaji, Hellweg, Urani, Zacher et al., 2005).

In view of all these data, we propose a model that explains, at least in part, the nature of the biological mechanisms involved in the formation of long-term memories elicited by stressful or arousing experiences: we suggest that evolution has selected mechanisms of growth and pro-survival response to stress as the fundamental molecular pathways activated by learning and recruited in brain cells to form long-term memories. In other words, as in any other cell, in the brain, adaptive levels of stress (via glucocorticoids) trigger cellular events that are responsible to restore homeostasis, hence adaptation to the new environment. To do so, these cellular events must produce persistent changes in the state of the cell, which lead to a new homoestatic state. Thus, the new homeostatic state, by means of its underlying long-lasting cellular and molecular changes, is the result of long-term changes, in other words result in the formation and persistence of a cellular long-term memory of the salient stimulus. Indeed, glucocorticoids, like hypoxia, metabolic or thermal stresses are cellular stressors and apoptotic promoters. Furthermore, whereas, as described earlier, in animal and humans, adaptive responses to environmental stressors depend on the activation of the HPA axis, in less differentiated organisms and cultured cells it induce the "stress response" (Pagliacci, Migliorati, Smacchia, Grignani, Riccardi et al., 1993). This cellular stress response generally comprises a pro-survival response that promotes growth and survival and protects cells from death. In neurons, the pro-survival response to stress consists of activation of growth pathways, which actually leads, in addition to the pro-survival response, to synaptic growth. Synaptic growth is known to underlie the formation and persistence of memory. Hence, it is plausible that evolution has indeed chosen the cellular response to stress as the fundamental mechanisms that in brain cells promote long-term memory formation.

Based on our data, we suggest that the pro-survival response to stress, which mediates long-term memory formation, occurs in the brain through sequential events: exposure to a salient event leads to the release of glucocorticoids that activate GRs in brain regions such as the hippocampus, which are critical for memory consolidation. GR activation recruits the BDNF/CREB-dependent pathways, which through their downstream as well as additional parallel events lead to cellular growth and promote survival. This activation of survival and growth pathways results in synaptic changes that underlie long-term maintenance of the information (Fig. 1). Our findings experimentally support and are in line with numerous evidence obtained in studies of chronic stress and glucocorticoid treatments indicating that glucocorticoids and BDNF critically influence each other (Bath, Schilit & Lee, 2013; Gray, Milner & McEwen, 2013; Jeanneteau & Chao, 2013; Numakawa, Adachi, Richards, Chiba & Kunugi, 2013; Rothman & Mattson, 2013; Suri & Vaidya, 2013).

Fig. 1. The GR/TrkB model of memory consolidation.

Exposure to a stress triggers release of glucocorticoids through activation of the HPA axis (1). Glucocorticoids released in the circulation cross the blood-brain barrier and activate glucocorticoid receptors (GRs) at the synapse of hippocampal neurons (2). In presynaptic neurons, GRs regulate release of glutamate by genomic-dependent and -independent mechanisms (3). Postsynaptically, GRs stimulate rapid non-genomic increases in synaptic GluA1 expression (4) and phosphorylation of CamKII (5), CREB (6), and TrkB (7), as well as genomic-dependent increases in Arc expression (8). TrkB-mediated signaling pathways activated by BDNF converge on CREB phosphorylation (9). Activation of presynaptic and postsynaptic GRs in hippocampal neurons, together with recruitment of the BDNF-mediated signaling pathways, is necessary for stress-mediated memory consolidation. These data are adapted from Chen, Bambah-Mukku, Pollonini, and Alberini (2012).

Although the GR/BDNF pathway recruitment is critical for proper memory consolidation in conditions at which the stress levels are controllable, evidence indicates that chronic stress and elevated glucocorticoids levels in pathological situations negatively regulate the BDNF pathway (Allen & Dawbarn, 2006). In agreement, chronic exposures to corticosterone or dexamethasone suppress BDNF-mediated release of glutamate in cultured cortical neurons, providing a possible mechanism for the negative impact of chronic stress on cognitive functions (Numakawa, Kumamaru, Adachi, Yagasaki, Izumi et al., 2009). In depression, the negative impact of chronic stress on spine density, synaptic plasticity, and neuronal survival in the hippocampus and prefrontal cortex is mediated, at least in part, by glucocorticoid-dependent downregulation of BDNF expression (Duman, Heninger & Nestler, 1997; Duman & Monteggia, 2006). Our results and working model, together with the literature on chronic stress, suggest that the convergence between GR and BDNF pathways may be an important node of dysfunction in stress-related cognitive impairments and affective disorders.

5. Glucocorticoids, cognitive impairments, and psychopathologies

5.1. Stress and memory: an inverted U-shaped relationship

Chronic stress or even single severely traumatic experiences can have a negative impact on cognitive functions and lead to the development of several psychopathologies (de Kloet, Joels & Holsboer, 2005; de Quervain, Aerni, Schelling & Roozendaal, 2009). The effect of stress on cognitive functions is largely dependent on the characteristics of the stressor. Stress intensity, duration, chronicity, controllability, and predictability are major characteristics that affect cognition and memory (Lupien, Maheu, Tu, Fiocco & Schramek, 2007).

Many studies have established that the intensity of a stressor is a critical factor that modulates cognitive performance. Specifically, in both humans and rodents, stress intensity and memory are known to follow an inverted U-shaped relationship, with maximal memory strength at an intermediate level of stress. In the early 20th century, this relationship was originally described by Yerkes and Dodson in a paradigm that measured the cognitive performance of mice in a discrimination task after exposure to electrical shocks of different intensities (Calabrese, 2008). The so-called Yerkes-Dodson law postulates that an optimal level of stress or arousal leads to maximal performance of a specific cognitive task. Importantly, this law also emphasizes that the inverted U effect shifts to a linear relationship as the task becomes simple. An important component of this theory therefore relies on the complexity of the task and brain structures involved in memory processing (Diamond, Campbell, Park, Halonen & Zoladz, 2007; Sandi & Pinelo-Nava, 2007).

Following the seminal observation of Yerkes and Dodson, various studies have characterized the nonlinear relationship between stress intensity and cognitive performance in rodent models and humans. For example, variation of the intensity of a stress intrinsic to the learning paradigm, such as water temperature in a radial arm maze, demonstrated the inverted U effect of stress on learning and memory performance in rats (Salehi, Cordero & Sandi, 2010). Similarly, systemic administration of corticosterone shortly after training modulates long-term object-recognition memory with an inverted U effect (Okuda, Roozendaal & McGaugh, 2004), while exposure to electrical footshocks of different intensities accompanied by intrahippocampal administration of corticosterone leads to a similar effect on contextual fear memories (Kaouane, Porte, Vallee, Brayda-Bruno, Mons et al., 2012). Notably, memory-impairing effects of high stress on spatial tasks are largely mediated by the action of glucocorticoids and GRs in the hippocampus and BLA (Roozendaal, Griffith, Buranday, De Quervain & McGaugh, 2003; Roozendaal, Hahn, Nathan, de Quervain & McGaugh, 2004), and become more pronounced as cognitive task gains complexity (Celerier, Pierard, Rachbauer, Sarrieau & Beracochea, 2004; Diamond, Park, Heman & Rose, 1999). In contrast, increasing stressor intensity leads to enhanced fear memory strength for simpler cognitive tasks, as shown in studies using classical Pavlovian contextual and cued fear conditioning (Cordero, Kruyt, Merino & Sandi, 2002; Rau, DeCola & Fanselow, 2005).

In humans, studies also support the conclusion of an inverted U-shaped relationship between stress intensity and performance of complex cognitive tasks (Diamond, Campbell, Park, Halonen & Zoladz, 2007; Lupien, Maheu, Tu, Fiocco & Schramek, 2007). Consolidation of declarative memories follows a bell-shaped curve associated with levels of cortisol secreted after stress exposure, with peak memory performance occurring at an intermediate cortisol level (Andreano & Cahill, 2006). Similarly, injections of increasing doses of cortisol rapidly modulate declarative memory retrieval with a dose-dependent effect that follows an inverted U profile (Abercrombie, Kalin, Thurow, Rosenkranz & Davidson, 2003; Domes, Rothfischer, Reichwald & Hautzinger, 2005; Schilling, Kolsch, Larra, Zech, Blumenthal et al., 2013). In addition, psychological stressors or administration of high doses of cortisol lead to impairment in spatial cognitive tasks and declarative memory, particularly when associated with emotionally laden material (Kirschbaum, Wolf, May, Wippich & Hellhammer, 1996; Kuhlmann, Piel & Wolf, 2005; Newcomer, Selke, Melson, Hershey, Craft et al., 1999). In contrast to the impairing effect of severe stress on complex memories and cognitive functions, it is known that exposure to traumatic events can lead to the development of pathological memories referred as “hypermnesia” or “flashbulb memories” during which subjects experience a particularly strong and vivid autobiographical memory for a specific highly arousing experience (Berntsen & Thomsen, 2005; Tekcan & Peynircioglu, 2002).

Together, studies on animal models and humans have shown that, in line with Yerkes and Dodson’s original observation, memory performance associated with complex cognitive tasks is sensitive to stress in an inverted U fashion, whereas simple forms of fear memory induced by traumatic experiences can be strong and persistent (Fig 2).

Fig. 2. Effect of arousal or stress intensity on cognitive performances and steroid receptor occupancy.

The intensity of a stressor is a critical parameter that modulates cognitive and memory performance. Exposure to an intermediate level of stress that leads to optimal cognitive performance triggers secretion of glucocorticoids in a range that fully activates the high affinity MRs and partially activates the low affinity GRs. An inverted U-shaped relationship between stress level and cognitive performance is observed in rodents and humans for complex tasks (e.g., decision making process, declarative and spatial memories), whereas high stress leads to an asymptotic effect on memory performance for simpler cognitive tasks (e.g., flashbulb memories, fear memories).

5.2. Glucocorticoid-mediated mechanisms of memory impairments

Exposure to an acute strong stress triggers the secretion of high levels of glucocorticoids and leads to memory impairment (de Kloet, Joels & Holsboer, 2005; Kim & Diamond, 2002; Sandi, 2004). For example, impairments of spatial memory have been observed in rats exposed to acute stress, social stress, following administration of corticosterone and in transgenic mice with endogenous elevated corticosterone levels (de Quervain, Roozendaal & McGaugh, 1998; Diamond, Park, Heman & Rose, 1999; Heinrichs, Stenzel-Poore, Gold, Battenberg, Bloom et al., 1996; Luine, Spencer & McEwen, 1993).

Chronic stress also leads to impairment in hippocampal-dependent learning and memory tasks. Chronic immobilization, unpredictable randomized stressors, repetitive social stressors, or chronic injections of corticosterone have all been found to provoke deficits in hippocampal-dependent forms of memory (Liu, Betzenhauser, Reiken, Meli, Xie et al., 2012; Luine, Villegas, Martinez & McEwen, 1994; Luine, Spencer & McEwen, 1993; Yuen, Wei, Liu, Zhong, Li et al., 2012). Interestingly, chronically stressed rats did not exhibit any deficits in certain memory paradigms such as cued and contextual fear conditioning, presumably because of the predominant role of the amygdala in these tasks (Conrad, LeDoux, Magarinos & McEwen, 1999; Sandi, Merino, Cordero, Touyarot & Venero, 2001). Together, these studies indicate that chronic stress mostly impairs hippocampal-dependent mechanisms and functions.

Exposure to severe acute or chronic stress can have dramatic consequences on neuronal morphology in the hippocampus. Chronic stress or corticosterone administration leads to dendritic atrophy in the CA1, CA3, and dentate gyrus (Magarinos & McEwen, 1995; McEwen, 2000a; Vyas, Mitra, Shankaranarayana Rao & Chattarji, 2002), as well as loss of excitatory synapses in the CA3 area (Sousa, Lukoyanov, Madeira, Almeida & Paula-Barbosa, 2000). Acute stress and social defeat stress also lead to neurodegeneration and inhibition of neurogenesis in the adult dentate gyrus (Gould & Tanapat, 1999; Lehmann, Brachman, Martinowich, Schloesser & Herkenham, 2013; Sapolsky, 2000). In particular, binding of glucocorticoids to GRs has been shown to mediate the negative effect of severe or chronic stress on hippocampal morphology and function, therefore likely contributing to the negative effect of stress on hippocampal-dependent memories (Conrad, Lupien & McEwen, 1999; Kirschbaum, Wolf, May, Wippich & Hellhammer, 1996; McEwen, 2000a).

In contrast to what is observed in the hippocampus, severe and chronic stresses enhance neuronal activity, synaptic transmission, spine formation, and dendritic growth in the amygdala (Roozendaal, McEwen & Chattarji, 2009). Stress-mediated neuronal activity and morphological changes in the amygdala in turn lead to increased anxiety-like behavior (Roozendaal, McEwen & Chattarji, 2009). The release of glucocorticoids on exposure to severe stress therefore has contrasting physiological and morphological effects in different brain regions that in turn control specific behavioral responses.

5.3. Glucocorticoids, GRs, and traumatic memories in PTSD: potential clinical applications

Improper regulation of the stress response and subsequent abnormal secretion of cortisol are often associated with stress-related psychopathologies such as anxiety disorders, depression, and PTSD (de Kloet, Joels & Holsboer, 2005; de Quervain, Aerni, Schelling & Roozendaal, 2009). Chronic hypercortisolemia in depression, advanced aging, or Cushing’s disease have been associated with declarative memory impairments (Lupien, de Leon, de Santi, Convit, Tarshish et al., 1998; Parker, Schatzberg & Lyons, 2003; Starkman, Gebarski, Berent & Schteingart, 1992). In PTSD, most studies have found reduced levels of circulating cortisol levels (Anisman, Griffiths, Matheson, Ravindran & Merali, 2001; Delahanty, Raimonde, Spoonster & Cullado, 2003; Neylan, Brunet, Pole, Best, Metzler et al., 2005; Yehuda, 2004; Yehuda, McFarlane & Shalev, 1998), although others did not find such correlation (Lindauer, Olff, van Meijel, Carlier & Gersons, 2006; Meewisse, Reitsma, de Vries, Gersons & Olff, 2007; Muhtz, Wester, Yassouridis, Wiedemann & Kellner, 2008; Pfeffer, Altemus, Heo & Jiang, 2007). In particular, individuals with PTSD exhibit enhanced suppression of cortisol release after administration of low doses of dexamethasone, indicating that chronic hypocortisolemia is caused by excessive negative feedback in the HPA axis (Grossman, Yehuda, New, Schmeidler, Silverman et al., 2003; Newport, Heim, Bonsall, Miller & Nemeroff, 2004; Yehuda, Halligan, Golier, Grossman & Bierer, 2004).

The action of GRs in PTSD has also been shown by human genetic studies (DeRijk & de Kloet, 2005). The BclI polymorphism of the GR gene that leads to glucocorticoid hypersensitivity has notably been associated with the development of PTSD symptoms after cardiac surgery (Hauer, Weis, Papassotiropoulos, Schmoeckel, Beiras-Fernandez et al., 2011; van Rossum, Koper, van den Beld, Uitterlinden, Arp et al., 2003) and at the onset of major depression (van Rossum, Binder, Majer, Koper, Ising et al., 2006). In healthy subjects, the BclI polymorphism is associated with increased memory performance for emotional pictures, suggesting its contribution to interindividual differences in emotional memory formation in nonpathological situations (Ackermann, Heck, Rasch, Papassotiropoulos & de Quervain, 2013). In line with these findings, a polymorphism in the gene encoding the GR chaperone FKBP5 has also been correlated with increased sensitivity of GR to cortisol and risk of PTSD (Mehta, Gonik, Klengel, Rex-Haffner, Menke et al., 2011). Identification of polymorphisms in the GR gene and associations with genetic variability of other regulators of the stress response is particularly important to better characterize the genetic bases underlying trauma-related psychopathologies such as PTSD.

Because glucocorticoids and GRs have key roles in the formation of traumatic memories, potential clinical application of steroid-based therapies for the treatment of stress-related psychopathologies has been investigated. Promising findings have shown that exogenous administration of low doses of cortisol dampens the strength of traumatic memories developed in PTSD, presumably via the inhibitory effect of glucocorticoids on memory retrieval or expression (Aerni, Traber, Hock, Roozendaal, Schelling et al., 2004; Bentz, Michael, de Quervain & Wilhelm, 2010; de Quervain & Margraf, 2008). Other evidence suggests the potential use of cortisol to treat phobic disorders. Cortisol administration in patients with different types of phobias reduces their fear response during the anticipation, exposure, and recovery phases after phobic cue presentation (Soravia, Heinrichs, Aerni, Maroni, Schelling et al., 2006). In addition to inhibiting the fear response, low doses of cortisol may facilitate extinction of a traumatic memory by enhancing the consolidation of a novel corrective experience dissociated from the original memory trace (de Quervain & Margraf, 2008). In mice, administration of glucocorticoids after memory reactivation impairs retrieval of an established fear memory (Cai, Blundell, Han, Greene & Powell, 2006), whereas inhibition of glucocorticoid synthesis during memory reactivation enhances retrieval and inhibits extinction of fear memory in mice (Blundell, Blaiss, Lagace, Eisch & Powell, 2011).

Memory reconsolidation, the process whereby a retrieved memory returns to a fragile state and becomes reconsolidated (Alberini, 2011), can be targeted to weaken traumatic memories. Studies from our laboratory have shown that in rats postretrieval inhibition of amygdalar GRs with the antagonist RU38486 (mifepristone) persistently weakens IA memory (Tronel & Alberini, 2007). Similarly, postretrieval inhibition of GRs by systemic treatment with RU38486 disrupts reconsolidation of an IA traumatic memory in rats, suggesting the importance of GR inhibitors in combination with trauma reactivation as potential novel therapeutic approach (Taubenfeld, Riceberg, New & Alberini, 2009). In agreement with this idea, a recent pilot study reported significant benefit with mifepristone in combat-related PTSD (Golier, Caramanica, Demaria & Yehuda, 2012). Hence, the combination of glucocorticoid-targeting treatments with behavioral or psychological therapy aimed at reactivating the old memory trace and disrupting its reconsolidation or evoking a new memory in a safe context may represent promising leads for novel therapeutic strategies against stress-related psychopathologies. A better understanding of the cascade of molecular events occurring in the brain after activation of GRs in high stress or traumatic conditions will be key to the design of novel selective interventions against these psychopathologies.

6. Conclusion

Stress modulates memory consolidation by orchestrated activation of neuroendocrine pathways in a specific spatial and temporal manner. Consolidating a strong memory after a salient or stressful event is an adaptive response that is necessary for appropriate reactions to similar demands in the future. In part, the molecular changes elicited by glucocorticoids, which are released in response to stress and play a critical role in mediating and modulating long-term memory formation and retention have been elucidated. These changes include rapid nongenomic synaptic modifications as well as long-term genomic changes triggered by GRs. Activation of GRs in adaptive conditions engages multiple intracellular signaling pathways, possibly because GRs target multiple brain areas, cell populations and memory phases. The GR-dependent mechanisms interact with fundamental process of neural transmission and plasticity such as glutamate neurotransmission and neurotrophic factor-mediated long-term responses.

Importantly, in adaptive response to stress, activation of GRs recruits the BDNF/CREB pathways, which in turn mediate and control memory consolidation. We propose that evolution has selected growth and pro-survival mechanisms in response to stress, and particularly those mediated by the BDNF/CREB pathways, as general mechanisms underlying memory consolidation. In contrast, although it is known that glucocorticoids and GRs also play a critical role in memory impairments following chronic stress or traumatic experience, the underlying molecular mechanisms have not yet been identified. Given the negative correlations between BDNF expression and chronic stress or cognitive impairments, our studies agree with the hypothesis proposed by previous authors (Duman, Heninger & Nestler, 1997; Nestler, Barrot, DiLeone, Eisch, Gold et al., 2002) indicating that dysregulation of the GR-mediated pathway may lead to depletion or disruption of BDNF expression and signaling. Such depletion would explain the associated memory and trauma-or stress-induced cognitive impairments. Characterization of the molecular pathways engaged by glucocorticoids in conditions of chronic or maladaptive stress that lead to cognitive impairments will be of particular importance in pursuit of the development for novel specific therapeutic strategies against stress-related psychopathologies.