Stress and emotional memory: a matter of timing

Transcription

1 Review Stress and emotional memory: a matter of timing Marian Joëls 1, Guillen Fernandez 2 and Benno Roozendaal 3 1 Department Neuroscience and Pharmacology, Division of Neuroscience, UMC Utrecht, Rudolf Magnus Institute, Utrecht, The Netherlands 2 Department for Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behavior, Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands 3 Department Neuroscience, Section Anatomy, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands Stressful events activate the amygdala and a network of associated brain regions. Studies in both humans and rodents indicate that noradrenaline has a prominent role in this activation. Noradrenaline induces a hypervigilant state that helps to remember the event. This mnemonic effect is enhanced when the situation is so stressful that substantial amounts of corticosteroids are released and reach the amygdala. The combination of the two hormones leads to optimal strengthening of contacts and thus memory. Yet, rises in corticosteroid levels that are not precisely synchronized with noradrenaline release do not act synergistically but rather prevent or suppress the effect of noradrenaline. This dynamic interaction illustrates the adaptive and potentially protective capacity of corticosteroids regarding traumatic memories. Timing is essential: the model In daily life people go through many situations that potentially impose a physical or psychological threat. The subjective experience of these threats is called stress. For stress to occur, the threatening situations do not have to be factual; very often, anticipation of the event is sufficient to initiate a biological response meant to adapt to the conditions [1,2]. One of the key features of successful adaptation is remembering the important (often emotionally challenging) aspects of the stressful situation for future use. Limbic areas, in particular the amygdala, play an important role in establishing such memories. The stress response is highly conserved among vertebrates (Box 1). As part of the response, neurons including principal cells in the amygdala receive a strong noradrenergic input from the locus coeruleus [3]. Studies in rodents revealed that this wave of noradrenaline (see Glossary) is relatively short lived (lasting less than 3 minutes) [4]. The same neurons also receive high levels of the adrenocortical hormone corticosterone (in most rodent species; cortisol in humans). The kinetic properties of corticosteroid exposure are slower than those of noradrenaline; peak corticosteroid levels in brain are probably not reached earlier than 2 minutes after stress onset, and normalization takes place only after 1 2 hrs [5]. Nevertheless, there seems to be a window in time during which Corresponding author: Joëls, M. (m.joels@umcutrecht.nl). limbic cells are exposed to elevated levels of both noradrenaline and corticosteroid hormones (Figure 1). It should be emphasized that although noradrenaline and corticosteroids are key elements in the stress response, they accomplish their full effect in concert with several other transmitters and hormones (Box 1). The partially overlapping presence in time and space of noradrenaline and corticosteroids sets the stage for interactions, provided that signals transduced by receptors of these ligands act in the same time frame [6]. Noradrenaline signals are mediated via G-protein-coupled membrane receptors that, on activation, change neuronal function within seconds. Activity is terminated when ligand concentrations drop below the levels necessary for substantial binding. Although receptor activation can secondarily give rise to transcriptional changes, this is not the main pathway through which noradrenaline is thought to act. Corticosteroids, by contrast, bind to Glossary Corticosteroid hormones: steroid hormones produced in the cortex of the adrenal glands. The main corticosteroid hormone in humans is cortisol whereas corticosterone prevails in most rodents. Corticosteroids bind to two types of intracellular receptors: the mineralocorticoid and the glucocorticoid receptor. On binding of the hormone, these receptors translocate to the nucleus where they either bind as dimers to specific recognition sites in the DNA (glucocorticoid response elements) or via protein protein interactions influence transcription factors that interact with the DNA. A small fraction of the receptors is thought to move to the plasma membrane rather than the nucleus. Genomic effects: corticosteroid receptors act as transcriptional regulators: they alter the transcription of a set of responsive genes (an estimated 1 2% of all genes). Eventually this leads to changes in the levels of the protein for which the responsive gene encodes. The process of transcription, translation and in some cases post-translational modification is thought to take at least one hour to be accomplished. Nongenomic effects: neurotransmitters (e.g. glutamate or noradrenaline) generally act through G-protein-coupled receptors or ligand-gated ion channels. This is a fast-acting mechanism that does not involve gene transcription: nongenomic. Recently it has become evident that corticosteroid hormones can also act through signaling pathways that bypass gene transcription and thus change neuronal activity over the course of minutes rather than hours. Noradrenaline: catecholamine that is enzymatically derived from dopamine and can be converted into adrenaline. Noradrenaline is produced to a large extent in the locus coeruleus, with additional production sites in the brainstem. Noradrenaline binds to a- and b-adrenoceptors which through G-proteins and second messengers induce changes in membrane potential at a time scale of seconds. Downstream effector molecules can secondarily change gene transcription inducing slow changes in neuronal function /$ see front matter ß 211 Elsevier Ltd. All rights reserved. doi:1.116/j.tics Trends in Cognitive Sciences, June 211, Vol. 15, No. 6

2 [()TD$FIG] Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 Stress hrs Hormone levels in the BLA Noradrenaline Corticosteroids Rapid and delayed noradrenaline effects Rapid and delayed corticosteroid effects Potential windows for functional interactions CORT: Suppressive CORT: Synergistic Encoding up CORT: Normalization Cognitive control up (delayed) Quick reset (rapid) TRENDS in Cognitive Sciences Figure 1. Shortly after stress, noradrenaline levels (yellow) in the BLA are transiently elevated. Corticosteroids (blue) reach the same area somewhat later and remain elevated for approximately 1 2 hrs. For a restricted period of time BLA neurons are exposed to high levels of both hormones (upper panel). Noradrenaline primarily works through a rapid G-protein coupled pathway (pale yellow) but secondary genomic effects requiring gene transcription might develop (bright yellow). By contrast, effects of corticosteroid hormones are mostly accomplished via nuclear receptors that mediate slow and persistent actions (bright blue), although rapid nongenomic actions have also been described in the BLA (pale blue). The lower panel reflects the windows in time during which the two hormones might functionally interact (green). Shortly after stress, corticosteroids are thought to promote effects of noradrenaline (upward arrow) enabling encoding of information. Later on, corticosteroid hormones normalize BLA activity (delayed effect via GR), a phase in which higher cognitive controls seem to be restored. When corticosteroids are given out of sync with noradrenaline they generally work suppressive. This is reflected by the quick reset of BLA activity (nongenomic GR effects) by corticosterone in organisms with a recent history of stress exposure. It is also evident from experiments in which corticosteroids were given several hours in advance of stress exposure or noradrenaline administration and exerted suppressive effects. The latter is a pharmacological situation but could potentially serve therapeutic goals. discretely localized nuclear receptors that act as transcriptional regulators (Box 1). Until recently it was thought that corticosteroids generally change brain function through these nuclear receptors, via a process that is very slow inonset yet longlasting [2]. If so, corticosteroids would act at a time when noradrenergic effects have already subsided. However, over the past decade it has become evident that corticosteroid hormones can also quickly change neuronal cell function in a nongenomic manner [6,7]. This results in a picture in which shortly after stress noradrenaline reaches limbic cells and exerts its rapid actions that after some minutes are potentially modulated by corticosteroid hormones (Figure 1). Simultaneously, a gene-mediated cascade is started primarily by corticosteroids but possibly also by monoamines or neuropeptides released after stress that several hours later normalizes the activity of these neurons. This portrays the neuronal stress response as a fine-tuned hormonal interplay in which time is of the essence. Below we will summarize the current evidence from neurophysiological, animal behavior and human neuroimaging studies that this fine-tuned scheme of hormone exposure indeed results in optimal functioning of the amygdala in response to stress, ensuring that stress-related information is retained very well, particularly those aspects with a strong emotional load. In this model, all hormone exposure that occurs out of sync is expected to attenuate the functional outcome. AMPAR signaling: a common target for noradrenaline and corticosteroids Seminal work by LeDoux and coworkers has established that the basolateral nucleus of the amygdala (BLA) is an essential hub in the formation of emotional memory [8], particularly memories related to fearful or emotionally arousing situations that are associated with the release of noradrenaline and (if sufficiently arousing) corticosteroid hormones [9]. The amygdala is core to a larger network of brain regions that on the one hand are involved in better consolidation of information in efferent brain regions, including the hippocampus and neocortical areas [1], and on the other hand in cognitive control, for example the medial prefrontal cortex (mpfc) [8,11]. Noradrenaline and corticosteroid hormones might interact at several (sub)cellular endpoints in this system, including calcium-dependent potassium channels [2,12], 281

3 Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 Box 1. Stress-induced release of noradrenaline and corticosteroids The various aspects of a stressful event e.g. visual, auditory and nociceptive stimuli are processed in the corresponding areas in the brain and eventually this information is projected to the paraventricular nucleus of the hypothalamus (for reviews see [1,2]). From there, neurons in the brainstem are activated that quickly give rise to activation of the sympathetic nervous system. The most prominent consequence is a quick release of adrenaline from the adrenal medulla that allows the organism to liberate energy sources for immediate action. Noradrenaline is released in the brain via the vagal nerve and stimulation of the nucleus tractus solitarius [34]. There also seems to be a direct stress-induced shift in the firing pattern of locus coeruleus noradrenergic neurons that involves corticotrophin releasing hormone (CRH) and opioids [69]. As a consequence of these pathways, amygdala neurons are exposed to high levels of noradrenaline shortly after stress [4]. Activation of a different set of cells in the paraventricular nucleus causes release of CRH that via portal vessels leads to secretion of adrenocorticotropin hormone (ACTH) in the circulation. In the adrenal cortex, exposure to ACTH increases synthesis and release of corticosterone (in most rodents) or cortisol (in humans). In the periphery, corticosteroids help the organism to restore from stress exposure, for example by replenishing energy resources. Corticosteroids also feed back on the pituitary gland and paraventricular nucleus, thereby suppressing the hypothalamo pituitary adrenal axis. The combination of slow release of corticosteroids from the adrenal glands with their negative feedback gives rise to an inherent oscillating pattern of corticosteroid release, with ultradian pulses approximately once per hour, on top of which surges of corticosteroid hormones are released after stress [7]. Corticosteroid hormones easily enter the brain and bind to highaffinity MRs and somewhat lower-affinity GRs (for reviews see [1,2[). MRs are enriched in all principal cells of the hippocampus, the lateral septum and to a lesser extent other limbic regions, for example the central amygdala and anterior cingulate cortex. Low concentrations of corticosteroids, corresponding to levels circulating during the nadir of the pulses, suffice to translocate these receptors to the nucleus where they act as transcriptional regulators. GRs are much more widespread in brain both in neurons and glial cells. These receptors require stress levels of corticosteroids to become fully occupied and translocate to the nucleus where they transiently bind to response elements in the DNA. Over the past years it has become evident that both MR and GR can also reside in or close to the plasma membrane where they mediate rapid nongenomic actions [6,7]. These rapid effects (also via MRs) seem to be evoked with corticosteroid concentrations that are only achieved during stress conditions. In addition to the two compounds discussed in this viewpoint, that is noradrenaline and corticosteroid hormones, many more transmitters and hormones play a role in the central response to stress, for example CRH, acetylcholine, GABA and endogenous opioids [6]. However, for these compounds aspects of timing have hardly been investigated so they will not be discussed here. remodeling of nuclear chromatin [13] and the function of a- amino-3-hydroxy-5-methyl-4-isoxazole propionate receptors (AMPARs). Here we elaborate on the latter. Glutamate transmission in the BLA and other limbic regions, specifically via AMPARs, is crucial for establishing emotional memories [14]. AMPARs are composed of various combinations of subunits, of which the GluA1 and GluA2 subunits are tightly linked to memory formation [15]. It is thought that on exposure to emotionally arousing situations, locally released neurotransmitters cause phosphorylation of several serine residues of the GluA1 subunit by protein kinase A (PKA) or CamKII/protein kinase C [16]. Consequently, GluA1 is transported to the postsynaptic density, a process that is necessary and sufficient to induce long-term potentiation (LTP) of the synapses [12,16]. Possibly (as seen in the hippocampus), GluA1 subunits are internalized at a later stage and replaced by GluA2 subunits [15], thus establishing stable contacts, enabling the downstream formation of memories. Noradrenaline is one of the transmitters that trigger this cascade [16]. The exact mechanism still needs to be resolved in the amygdala but it might resemble that in the mpfc where b 2 -adrenoceptors form a complex with GluA1 subunits, PKA and other proteins [17]. Activation of b 2 - adrenoceptors leads to phosphorylation of GluA1 subunits and increases the amplitude of AMPAR-mediated miniature postsynaptic currents (mepscs), i.e. currents caused by spontaneous release of one glutamate-containing synaptic vesicle [17]. Corticosterone also targets the AMPA signaling pathway [18,19]. In its slow, genomic mode it enhances surface expression of GluA2 and, with a higher concentration, GluA1 subunits in the hippocampus, via the lower-affinity glucocorticoid receptor (GR). Moreover, corticosterone slowly increases the amplitude of hippocampal mepscs [19,2], similar to the rapid effects by noradrenaline. Enhanced AMPAR surface expression and mepsc amplitude by corticosterone and stress were also observed in the mpfc [21]. This could point to a generalized slow enhancement in mepsc amplitude by corticosterone, including in the amygdala (Box 2). Yet, corticosterone can also rapidly nongenomically change AMPAR signaling via the mineralocorticoid receptor (MR). The hormone quickly promotes lateral diffusion Box 2. Questions for future research To what extent do noradrenaline and corticosterone interact on AMPAR signaling when tested in the same BLA cell (or the same cell in another brain region)? How do various concentrations of corticosterone given approximately 2 minutes after a pulse of noradrenaline (i.e. the naturally occurring time frame of both compounds after stress) affect signal transduction in the BLA? What is the effect of timed corticosterone and noradrenaline administration into the BLA on the various phases of memory? Does metaplasticity of BLA function (i.e. a switch in response after a second stressor) also happen at the network and behavioral level? Exactly how do endocannabinoids affect BLA function after stress? Are the effects of corticosteroid administration several hours before noradrenaline on animal behavior comparable to those indicated by the electrophysiological studies and, if so, are they valence-dependent as in human subjects? Are some of the discrepant findings explained by the fact that moderate concentrations of the two hormones interact differently from high doses? What are the timed effects of ACTH and CRH on amygdala tissue in individuals that are stressed or receive corticosteroids systemically? Are individual differences in processes controlling amygdala responsiveness crucial for differences in stress sensitivity and resilience? By which mechanisms do adverse environmental factors change human amygdala sensitivity? 282

4 [()TD$FIG] Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 (a) Frequency (Hz) Frequency (Hz) st 1st Hippocampus CA1 2nd Amygdala BLA 2nd (b) Frequency (Hz) After restraint stress nm CORT Time (min) (c) Response (% of baseline) ISO CORT preincubated CORT preincubated before ISO Time (in min) TRENDS in Cognitive Sciences Figure 2. Corticosterone administration several hours after or before stress/noradrenaline treatment attenuates the functional outcome. In the presence of corticosterone (gray/ green bars in (a) mepsc frequency in the hippocampus is increased. A similar response is seen after a second pulse of corticosterone. In the BLA, however, the first pulse lastingly increases mepsc frequency that is quickly reset by a second pulse. Also after restraint stress (b), corticosterone suppresses mepsc frequency. Based on [23]. (c) Isoproterenol administration (horizontal bar) causes LTP in response to weak tetanic stimulation (arrow) that by itself does not induce LTP (not shown). Pretreatment with corticosterone 1 4 hours before isoproterenol application, ineffective by itself, attenuates the potential of the b-agonist to induce LTP. Adapted, with permission, from [25]. of GluA2 subunits in cultured hippocampal cells [18] and increases mepsc frequency rather than amplitude [22]. Interestingly, in the BLA rapid MR-dependent increases in mepsc frequency develop into long-lasting changes through a slow process requiring protein synthesis [23]. This alters the state of BLA neurons such that after one surge of corticosterone, for example after experiencing stress, they respond to renewed exposure of corticosterone with a quick decrease in mepsc frequency, resetting the BLA cell activity (Figure 2); this involves GRs and endocannabinoid signaling. What does this mean for the combined effects of noradrenaline and corticosterone on limbic cell function (Figure 1)? In its rapid mode, corticosterone seems to target GluA2 subunits whereas noradrenaline seems more effective on GluA1 subunits, although studies directly addressing this issue are lacking so far (Box 2). If so, effects of the two hormones might be additive rather than synergistic. Electrophysiological data support this to some extent. In the dentate gyrus corticosterone accelerated the effect of co-administered isoproterenol (a b- adrenoceptor agonist) on LTP [24].Yet,corticosteronedid not rapidly facilitate isoproterenol effects in the BLA, neither with respect to LTP nor AMPAR signaling [25,26]. Nevertheless, given the quick but particularly long-lasting increase in amygdala mepsc frequency by corticosterone [23], this hormone could indirectly enable effects of noradrenaline and prolong the window during which stress-related memory is encoded. The enabling effects of corticosterone in the short term contrast with its slow genomic effects. Corticosterone was shown to gradually normalize the response of BLA neurons earlier raised by noradrenaline [25], potentially allowing a more prominent higher cognitive control in the aftermath of stress. Simultaneously, corticosterone slowly changes the state of BLA neurons such that their response to a new stressor is different, a phenomenon reminiscent of cellular metaplasticity [23]. This might curtail the activity of BLA neurons on reactivation of the circuit. When corticosterone is given several hours in advance of isoproterenol a condition that will not take place physiologically but might be exploited therapeutically it reduces the efficacy of the b-adrenoceptor agonist to facilitate LTP and evoked responses via AMPARs, both in the dentate gyrus and BLA [24 26] (Figure 2), possibly due to convergence of the two hormones on the same signaling pathways. In this mode corticosterone could prevent the formation of emotional memories. 283

5 [()TD$FIG] Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 Training trial Retention trial A1 A2 A3 B Interval? amygdala-dependent behavior (e.g. 28,29). For instance, post-training corticosterone administration enhances spatial and aversive memory formation, an effect that is blocked by concurrent intra-bla infusions of a b-adrenoceptor antagonist [29 32]. Corticosterone also enhanced memory of object recognition training when administered to naïve rats, which again was blocked by the b-adrenoceptor antagonist propranolol [32]. Importantly, corticosterone was ineffective in rats with reduced trainingassociated emotional arousal due to prior habituation to the experimental context [33] (Figure 3). The effect of emotional arousal was mimicked in well-habituated rats by administration of the a 2 -adrenoceptor antagonist yohimbine immediately after object recognition training, presumably increasing noradrenaline levels in the brain [32]. Collectively, these studies indicate that noradrena- Withhabituation Withouthabituation Corticosterone (sc) Day Habituation Training Retention Discrimination index (%) Without-habituation ** ** Key: Vehicle Cort.3 mg/kg Cort 1. mg/kg Cort 3. mg/kg Discrimination index (%) With-habituation ** -1 Propranolol (3. mg/kg) -1 Yohimbine (.3 mg/kg) TRENDS in Cognitive Sciences Figure 3. Glucocorticoid effects on memory consolidation for object recognition training require arousal-induced noradrenergic activation. Rats were either habituated to the training context or not for seven days. On day 8, they were given a 3-minute training trial during which they could freely explore 2 identical objects (top) followed by systemic drug administration. Retention was tested 24 hours later by placing the rats back into the apparatus for 3 minutes in which one object was similar to that explored during training whereas the other object was novel. Corticosterone injection selectively enhanced memory consolidation in context-naïve rats that showed a greater emotional arousal response during training (bottom left). Systemic post-training administration of the b-adrenoceptor antagonist propranolol blocked this corticosteroneinduced enhancement of object recognition memory in naïve (emotionally aroused) rats. Coadministration of the a 2 -adrenoceptor antagonist yohimbine with corticosterone enhanced object recognition memory in habituated (emotionally nonaroused) rats (bottom right). Data represent discrimination index (%) on the 24-hour retention trial, expressed as mean SEM. The discrimination index was calculated as the difference in time exploring the novel and familiar object expressed as the ratio of the total time spent exploring both objects., P <.1 vs. vehicle. Adapted, with permission, from [1]. Noradrenaline and corticosterone in emotional memory formation These inferences about hormonal influences on emotional memory formation based on observations at the cellular level are to a considerable degree supported by studies at the behavioral level (but see Box 2). The presence of noradrenaline is important for the facilitation of emotional memory in rodents. For instance, noradrenaline levels in the amygdala following aversively motivated inhibitory avoidance training correlate nicely with retention latencies tested 24 hours later [4]. Moreover, post-training infusions of noradrenaline or b-adrenoceptor agonists into the BLA produce a dose-dependent enhancement of memory consolidation [1,27]. For decades, however, it has been known that corticosterone interacts with noradrenaline in the modulation of 284

6 Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 line is in the driver seat but that simultaneously released corticosteroids potently boost these actions [34]. To some extent corticosteroids might act by increasing the availability of noradrenaline in the BLA, an effect that cannot be taken into account with in vitro electrophysiological investigations. For instance, corticosterone administered systemically after inhibitory avoidance training rapidly elevated noradrenaline levels within the BLA [35]. Moreover, post-training infusions of a GR agonist into the nucleus of the solitary tract (Box 1) dose-dependently enhanced memory consolidation of inhibitory avoidance training and this memory enhancement was blocked by intra-bla infusions of a b-adrenoceptor antagonist [36]. However, corticosterone certainly does not only affect behavior by increasing noradrenaline levels. Firstly, posttraining intra-bla infusions of the b-adrenoceptor agonist clenbuterol or the camp analog 8-bromo-cAMP enhanced memory consolidation in a dose-dependent fashion [27]. After infusion of a GR antagonist into the BLA shortly before training, a much higher dose of clenbuterol was required to induce comparable memory enhancement [29]. The GR antagonist did not modify the effects of 8-bromocAMP, indicating that camp acts downstream from the locus of interaction of corticosteroids with the b-adrenoceptor camp/pka pathway. Secondly, intra-bla administration of a cannabinoid (CB) receptor-1 antagonist blocked the ability of systemically administered corticosterone to facilitate aversive memory consolidation [37]. Recently, Hill et al. [38] showed that acute administration of corticosterone rapidly increases endocannabinoid levels in the amygdala. This finding indicates that corticosteroids might bind to membrane-located receptors in the BLA that activate a G-protein-coupled signaling cascade inducing endocannabinoid synthesis, as was previously also proposed for hypothalamic neurons [7]. Endocannabinoid ligands then could diffuse and bind to presynaptic CB1 receptors. What the endocannabinoids will do next is unclear at this time (Box 2). In vitro electrophysiological studies have shown that glutamatergic transmission is reduced via the CB1 receptor but only when the organism has recently experienced a period of stress or arousal [23]. Possibly, endocannabinoids target local GABAergic terminals in nonaroused organisms, to inhibit GABA release onto noradrenergic terminals, thus increasing the local release of noradrenaline [37,39]. In an intact system, corticosteroid and noradrenergic actions in the BLA cannot be regarded independent of what happens in associated areas, such as the hippocampus and mpfc, given the connections between these brain regions [11,4]. For instance, interactions between noradrenaline and corticosterone in the BLA time-dependently influence the function of the dentate gyrus [41]. Moreover, corticosteroids administered into the prefrontal or insular cortices enhance memory consolidation [42,43]. Highly comparable to the BLA, infusions of a b-adrenoceptor antagonist or PKA inhibitor into these brain regions prevented the memory enhancement by concurrently administered corticosteroids [44,45]. Corticosterone administered systemically immediately after inhibitory avoidance training increased PKA activity in the mpfc within 3 minutes, supporting the view that facilitatory corticosteroid effects on noradrenergic signaling in the mpfc are rapid and nongenomic. Interactions between noradrenaline and corticosteroids on memory consolidation have almost exclusively been studied when both stress systems were modulated around the training experience. Interestingly, one study shows that pretreatment with corticosteroids dramaticallyreducestheefficacyofadrenalinetoaffectamygdaladependent behavior [28]. This is entirely in line with the slow genomic GR effects observed at the cellular level. Clearly, this observation requires more detailed investigation (Box 2). Noradrenaline and corticosteroid actions in the human brain In humans, functional neuroimaging allows noninvasive monitoring of regional brain activity with good temporal and spatial resolution. Such studies have shown that successful memory formation of emotionally arousing stimuli is correlated with an amygdala response that is not apparent when neutral nonarousing memories are formed [46,47]. This neural correlate of emotional memory formation is causally associated with the emotional enhancement, as elegantly shown by Richardson et al. [48]. Interventional neuroimaging studies, blocking b-adrenoceptors or enhancing noradrenaline activity pharmacologically, as well as genetic neuroimaging studies probing a gene coding for the presynaptic a 2 -adrenoceptor have collectively confirmed that this emotional memory effect observed in the amygdala is controlled by noradrenergic activity [49 52]. These pharmacological and genetic findings in brain activity are closely in line with those of behavioral studies that show that the emotional enhancement effect depends on noradrenaline availability [53,54]. Going beyond single stimuli that are arousing, one can induce more prolonged states of psychological stress or threat, which also go along with heightened amygdala activity [55,56]. This state-dependent effect seems to be controlled by noradrenaline, as indicated by a genetic study in which a putatively higher availability of noradrenaline was found to be associated with greater amygdala responses when individuals are in a stressful but not in a neutral control state [57]. In sum, neuroimaging studies have confirmed the analogous role of the rodent amygdala and noradrenaline in humans: mediating the emotional enhancement effect during memory formation. Neuroimaging is also well suited to assess functional connectivity in macroscopic brain networks, including the amygdala, hippocampus and mpfc. The connectivity between the amygdala and the mpfc is especially increased directly after being confronted with emotional stimuli [58,59] and given the negative control by the mpfc over amygdala activity thus seems instrumental in limiting the amygdala response to stress. Arousing stimuli also increase the connectivity between the amygdala and specific brainstem regions that are the main source of noradrenaline [59]. Thus, analysis of connectivity has delineated a network that is responsible for regulating the amygdala. Cognitive control signals sent by the mpfc and noradrenaline-dependent bottom-up signals sent by brainstem nuclei, such as the locus coeruleus, seem to 285

7 [()TD$FIG] Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 (a) Main effect of Drug (b) Drug x Emotion type interaction (c) Placebo Rapid Slow CORT.2 CORT F value F value R R Happy Fear TRENDS in Cognitive Sciences Figure 4. Hydrocortisone was administered orally to male young-adult subjects either 75 minutes (rapid CORT) or 285 minutes (slow CORT) before scanning in a randomized, double-blind, placebo-controlled functional Magnetic Resonance Imaging study. Results revealed that hydrocortisone affects amygdala responsivity in a timeand emotion-specific manner. (a) Main effect of hydrocortisone administration on amygdala activity. Hydrocortisone administration reduced amygdala responsivity in general regardless of the timing of administration. (b) Drug x emotion type interaction in the amygdala. The effects of hydrocortisone administration depended on the emotion type. (c) Extracted parameter estimates from the anatomically defined bilateral amygdala revealed that the drug x emotion type interaction was driven by a larger emotion effect (fearful>happy) in the slow hydrocortisone condition. Error bars represent SEM. Reproduced, with permission, from [62]. Parameter estimates (a.u.) adjust amygdala activity on a moment-by-moment scale of seconds to several minutes to adapt brain responses to rapidly changing environmental demands. This supports the view based on animal behavior of a noradrenalinedriven hypervigilant state associated with high amygdala activity shortly after stress. Cortisol, released simultaneously as part of the stressful conditions, might initially promote such noradrenergic actions on amygdala activity and behavior, as indicated by indirect evidence [55,6]. However, when subjects either received cortisol 3 4 hours before the experiment or were investigated during the diurnal cortisol peak, amygdala activity was reduced compared to a control condition, either during rest or while processing salient stimuli [61 63], indicating normalization of amygdala activity by cortisol via its slow mode of action allowing the expression of higher cognitive control via the mpfc. The two modes by which cortisol might act were recently addressed in detail [62] (Figure 4). In this study, subjects ingested a placebo or a cortisol tablet (1 mg) either 75 or 285 minutes before neuroimaging, preferentially targeting nongenomic and slow genomic effects respectively; it should be noted, however, that a delay of 75 minutes probably does not exclude genomic effects. Irrespective of timing, cortisol reduced amygdala responses to salient stimuli. In the slow mode cortisol increased mpfc amygdala connectivity, indicating augmented cognitive topdown control that is associated with normalized responses to fearful face stimuli but still reduced responses to happy face stimuli. This indicates a valence-specific slow corticosteroid effect that causes diminished amygdala processing for one valence condition only, something that has not been addressed in animal studies so far (Box 2). In these studies the status of the endogenous noradrenergic system was unknown. For some individuals and under some circumstances the experimental procedures might already be stressful [64] whereas other individuals might be in a normal alert state or even in a drowsy state of low motivation. Two studies manipulating both cortisol and noradrenaline levels pharmacologically before neuroimaging indicate that the combination of both drugs causes a strong amygdala-dependent processing of negative information [65,66] but the delay between drug administration and neuroimaging in these studies was ambiguous: it favors development of nongenomic rather than genomic actions. Concluding remarks The overall model presented in the first section of this article largely fits with observations at the cellular level, the animal behavioral level and in human brain studies. Firstly, of the two modulators discussed, primarily noradrenaline promotes the formation of stress-related emotional memories via activation of the (basolateral) amygdala. Secondly, this process is facilitated by corticosteroid hormones that reach the same area shortly after noradrenaline. Thirdly, corticosteroids given hours before presumed release (or administration) of noradrenaline generally suppress amygdala activity and associated behavior. The model is based on very diverse experimental approaches, from single cell to human brain, each with their advantages and limitations. Although human research provides crucial ecological validity because it investigates complex interactions at the brain system level in the species of most interest to clinical neuroscience, it cannot optimally dissect the effects that particular neuromodulators exert on a specific brain region or process because drugs are applied systemically and affect the body and the entire brain. Moreover, restrictions in spatial resolution, for example not allowing distinction between the various amygdala nuclei, are still a rate-limiting factor. Thus, efforts integrating in vitro and in vivo results obtained in animal (tissue) as well as in human research are of utmost importance when striving to understand how stress affects emotional memory formation and consolidation. Using comparable timing paradigms and pharmacological interventions is a necessary step to resolve some of the open questions (Box 2). The additive or synergistic effects between noradrenaline and corticosteroids seem to be an important element in behavioral adaptation. Studies so far indicate that optimal facilitation is achieved with moderate concentrations of the two compounds; at high levels there might be saturation that overall would result in an inverted U-shaped dosedependency [e.g. 32], although this issue requires more indepth investigation, especially in humans. However, what is the role of the antagonistic slow corticosteroid-mediated 286

8 Review Trends in Cognitive Sciences June 211, Vol. 15, No. 6 actions in the amygdala? One interpretation is that these actions slowly normalize amygdala activity in the aftermath of stress, comparable to the hippocampus, although at the moment there is limited factual evidence for this interpretation. Tuning down amygdala activity might allow the gradual expression of higher mpfc-driven cognitive control after the initial phase of alert after stress has subsided. Moreover, the slow gene-mediated effects seem to alter the state of BLA cells such that their activity is quickly reset by a subsequent surge of corticosteroids. This indicates that BLA cell activity after stress can be curtailed by renewed elevations of corticosteroid levels, for example during re-experience of the stressful event and could fail to do so with inadequate hormone levels. Although this notion fits with some theories on the vulnerability to posttraumatic stress disorder [67], it strikes one as a rather circumventive way to normalize BLA activity. These corticosteroid actions in the aftermath of stress clearly require more dedicated investigation. Interestingly, all studies (using electrophysiological, animal behavioral or human neuroimaging techniques) agree that corticosteroids given several hours before noradrenergic activation of the amygdala suppress amygdala activity. This is clearly a pharmacological approach, but nevertheless of interest, because this prophylactic suppression of amygdala activity could have applications in the protection of individuals at risk for psychological trauma. Finally, the findings on timed corticosteroid noradrenaline interactions were made in young, healthy laboratory rats or human subjects. Individual variations were neglected but are expected to be of great relevance. Genetic variations but also early life influences, which are known to alter amygdala connections [68], might require adjustments to the here-described model. References 1 McEwen, B.S. and Gianaros, P.J. (21) Stress- and allostasis-induced brain plasticity. Annu. Rev. Med. 62, De Kloet, E.R. et al. (25) Stress and the brain: from adaptation to disease. Nat. Rev. Neurosci. 6, Sara, S.J. (29) The locus coerlueus and noradrenergic modulation of cognition. Nat. Rev. Neurosci. 1, McIntyre, C.K. et al. (22) Amygdala norepinephrine levels after training predict inhibitory avoidance retention performance in rats. Eur. J. Neurosci. 16, Droste, S.K. et al. (28) Corticosterone levels in the brain show a distinct ultradian rhythm but a delayed response to forced swim stress. Endocrinology 149, Joëls, M. and Baram, T.Z. (29) The neuro-symphony of stress. Nat. Rev. Neurosci. 1, Tasker, J. et al. (26) Minireview: rapid glucocorticoid signaling via membrane-associated receptors. Endocrinology 147, LeDoux, J.E. (2) Emotion circuits in the brain. Annu. Rev. Neurosci. 23, Rodrigues, S.M. et al. (29) The influence of stress hormones on fear circuitry. Annu. Rev. Neurosci. 32, Roozendaal, B. et al. (29) Stress, memory and the amygdala. Nat. Rev. Neurosci. 1, Likhtik, E. et al. (25) Prefrontal control of the amygdala. J. Neurosci. 25, Power, J.M. et al. (211) Location and function of the slow afterhyperpolarization channels in the basolateral amygdala. J. Neurosci. 31, John, S. et al. (211) Chromatin accessibility pre-determines glucocorticoid receptor binding patterns. Nat. Genet. 43, Rumpel, S. et al. (25) Postsynaptic receptor trafficking underlying a form of associative learning. Science 38, Isaac, J.T. et al. (27) The role of the GluR2 subunit in AMPA receptor function and synaptic plasticity. Neuron 54, Hu, H. et al. (27) Emotion enhances learning via norepinephrine regulation of AMPA-receptor trafficking. Cell 131, Joiner, M.L. et al. (21) Assembly of a beta2-adrenergic receptor GluR1 signalling complex for localized camp signalling. EMBO J. 29, Groc, L. et al. (28) The stress hormone corticosterone conditions AMPAR surface trafficking and synaptic potentiation. Nat. Neurosci. 11, Martin, S. et al. (29) Corticosterone alters AMPAR mobility and facilitates bidirectional synaptic plasticity. PLoS ONE 4, e Karst, H. and Joëls, M. (25) Corticosterone slowly enhances miniature excitatory postsynaptic current amplitude in mice CA1 hippocampal cells. J. Neurophysiol. 94, Yuen, E.Y. et al. (29) Acute stress enhances glutamatergic transmission in prefrontal cortex and facilitates working memory. Proc. Natl. Acad. Sci. U.S.A. 16, Karst, H. et al. (25) Mineralocorticoid receptors are indispensable for nongenomic modulation of hippocampal glutamate transmission by corticosterone. Proc. Natl. Acad. Sci. U.S.A. 12, Karst, H. et al. (21) Metaplasticity of amygdalar responses to the stress hormone corticosterone. Proc. Natl. Acad. Sci. U.S.A. 17, Pu, Z. et al. (27) Corticosterone time-dependently modulates betaadrenergic effects on long-term potentiation in the hippocampal dentate gyrus. Learn Mem. 14, Pu, Z. et al. (29) Beta-adrenergic facilitation of synaptic plasticity in the rat basolateral amygdala in vitro is gradually reversed by corticosterone. Learn Mem. 16, Liebmann, L. et al. (29) Effects of corticosterone and the beta-agonist isoproterenol on glutamate receptor-mediated synaptic currents in the rat basolateral amygdala. Eur. J. Neurosci. 3, Ferry, B. et al. (1999) Basolateral amygdala noradrenergic influences on memory storage are mediated by an interaction between b- and a 1 - adrenoceptors. J. Neurosci. 19, Borrell, J. et al. (1984) Corticosterone decreases the efficacy of adrenaline to affect passive avoidance retention of adrenalectomized rats. Life Sci. 34, Roozendaal, B. et al. (22) Glucocorticoids interact with the basolateral amygdala b-adrenoceptor-camp/pka system in influencing memory consolidation. Eur. J. Neurosci. 15, Quirarte, G.L. et al. (1997) Glucocorticoid enhancement of memory storage involves noradrenergic activation in the basolateral amygdala. Proc. Natl. Acad. Sci. U.S.A. 94, Roozendaal, B. et al. (1999) Basolateral amygdala noradrenergic influence enables enhancement of memory consolidation induced by hippocampal glucocorticoid receptor activation. Proc. Natl. Acad. Sci. U.S.A. 96, Roozendaal, B. et al. (26) Glucocorticoid enhancement of memory requires arousal-induced noradrenergic activation in the basolateral amygdala. Proc. Natl. Acad. Sci. U.S.A. 13, Okuda, S. et al. (24) Glucocorticoid effects on object recognition memory require training-associated emotional arousal. Proc. Natl. Acad. Sci. U.S.A. 11, McGaugh, J.L. and Roozendaal, B. (22) Role of adrenal stress hormones in forming lasting memories in the brain. Curr. Opin. Neurobiol. 12, McReynolds, J.R. et al. (21) Memory-enhancing corticosterone treatment increases amygdala norepinephrine and Arc protein expression in hippocampal synaptic fractions. Neurobiol. Learn. Mem. 93, Roozendaal, B. et al. (1999) Glucocorticoid receptor activation in the rat nucleus of the solitary tract facilitates memory consolidation: involvement of the basolateral amygdala. Eur. J. Neurosci. 11, Campolongo, P. et al. (29) Endocannabinoids in the rat basolateral amygdala enhance memory consolidation and enable glucocorticoid modulation of memory. Proc. Natl. Acad. Sci. U.S.A. 16,

9 Review Trends in Cognitive Sciences June 211, Vol. 15, No Hill, M.N. et al. (21) Rapid induction of limbic endocannabinoid signaling by glucocorticoid hormones in vivo. Psychoneuroendocrinol 35, Hill, M.N. and McEwen, B.S. (29) Endocannabinoids: the silent partner of glucocorticoids in the synapse. Proc. Natl. Acad. Sci. U.S.A. 16, Perez-Jaranay, J.M. and Vives, F. (1991) Electrophysiological study of the response of medial prefrontal cortex neurons to stimulation of the basolateral nucleus of the amygdala in the rat. Brain Res. 564, Akirav, I. and Richter-Levin, G. (1999) Biphasic modulation of hippocampal plasticity by behavioral stress and basolateral amygdala stimulation in the rat. J. Neurosci. 19, Miranda, M.I. et al. (28) Glucocorticoids enhance taste aversion memory via actions in the insular cortex and basolateral amygdala. Learn. Mem. 15, Roozendaal, B. et al. (29) Glucocorticoid effects on memory consolidation depend on functional interactions between the medial prefrontal cortex and basolateral amygdala. J. Neurosci. 29, Barsegyan, A. et al. (21) Glucocorticoids in the prefrontal cortex enhance memory consolidation and impair working memory by a common neural mechanism. Proc. Natl. Acad. Sci. U.S.A. 17, Roozendaal, B. et al. (21) Membrane-associated glucocorticoid activity is necessary for modulation of long-term memory via chromatin modification. J. Neurosci. 3, Cahill, L. et al. (1996) Amygdala activity at encoding correlated with long-term, free recall of emotional information. Proc. Natl. Acad. Sci. U.S.A. 93, Canli, T. et al. (2) Event-related activation in the human amygdala associates with later memory for individual emotional experience. J. Neurosci. 2, RC99 48 Richardson, M.P. et al. (24) Encoding of emotional memories depends on amygdala and hippocampus and their interactions. Nat. Neurosci. 7, Strange, B.A. and Dolan, R.J. (24) Beta-adrenergic modulation of emotional memory-evoked human amygdala and hippocampal responses. Proc. Natl. Acad. Sci. U.S.A. 11, Onur, O.A. et al. (29) Noradrenergic enhancement of amygdala responses to fear. Soc. Cogn. Affect. Neurosci. 4, Rasch, B. et al. (29) A genetic variation of the noradrenergic system is related to differential amygdala activation during encoding of emotional memories. Proc. Natl. Acad. Sci. U.S.A. 16, Urner, M. et al. Genetic variation of the a2b-adrenoceptor affects neural correlates of successful emotional memory formation. Hum. Brain. Mapp. doi:1.12/hbm Cahill, L. et al. (1994) Beta-adrenergic activation and memory for emotional events. Nature 371, de Quervain, D.J. et al. (27) A deletion variant of the alpha2badrenoceptor is related to emotional memory in Europeans and Africans. Nat. Neurosci. 1, van Marle, H.J. et al. (29) From specificity to sensitivity: how acute stress affects amygdala processing of biologically salient stimuli. Biol. Psychiatry 66, Mobbs, D. et al. (21) Neural activity associated with monitoring the oscillating threat value of a tarantula. Proc. Natl. Acad. Sci. U.S.A. 17, Cousijn, H. et al. (21) Acute stress modulates genotype effects on amygdala processing in humans. Proc. Natl. Acad. Sci. U.S.A. 17, Klumpers, F. et al. (21) Prefrontal mechanisms of fear reduction after threat offset. Biol. Psychiatry 68, van Marle, H.J. et al. (21) Enhanced resting-state connectivity of amygdala in the immediate aftermath of acute psychological stress. Neuroimage 53, van Stegeren, A.H. et al. (27) Endogenous cortisol level interacts with noradrenergic activation in the human amygdala. Neurobiol. Learn. Mem. 87, Cunningham-Bussel, A.C. et al. (29) Diurnal cortisol amplitude and fronto-limbic activity in response to stressful stimuli. Psychoneuroendocrinology 34, Henckens, M.J. et al. (21) Time-dependent effects of corticosteroids on human amygdala processing. J. Neurosci. 3, Lovallo, W.R. et al. (21) Acute effects of hydrocortisone on the human brain: an fmri study. Psychoneuroendocrinology 35, Tessner, K.D. et al. (26) Cortisol responses of healthy volunteers undergoing magnetic resonance imaging. Hum. Brain Mapp. 27, Kukolja, J. et al. (28) Modeling a negative response bias in the human amygdala by noradrenergic-glucocorticoid interactions. J. Neurosci. 28, van Stegeren, A.H. et al. (21) Interacting noradrenergic and corticosteroid systems shift human brain activation patterns during encoding. Neurobiol. Learn. Mem. 93, Yehuda, R. (29) Status of glucocorticoid alterations in post-traumatic stress disorder. Ann. N.Y. Acad. Sci. 1179, Moriceau, S. et al. (29) Early-life stress disrupts attachment learning: the role of amygdala corticosterone, locus ceruleus corticotropin releasing hormone, and olfactory bulb norepinephrine. J. Neurosci. 29, Valentino, R.J. and Van Bockstaele (28) Convergent regulation of locus coeruleus activity as an adaptive response to stress. Eur. J. Pharmacol. 583, Lightman, S.L. and Conway-Campbell, B.L. (21) The crucial role of pulsatile activity of the HPA axis for continuous dynamic equilibration. Nat. Rev. Neurosci. 11,