Cortico-amygdala circuits: Role in the conditioned stress response

Abstract

The amygdala plays a crucial role in the orchestration and modulation of the organism response to aversive, stressful events. This response could be conceived as the result of two interdependent components. The first is represented by sets of visceral and motor responses aimed at helping the organism to cope with the present event. The second is the acquisition and modulation of memories relative to the stressful stimulus and its context. This latter component contributes to the instatement of conditioned stress responses that are essential to the capability of the organism to predict future exposures to similar stimuli in order to avoid them or counteract them effectively. In the amygdala, these two components become fully integrated. Massive networks link the amygdala to the hypothalamus, midbrain and brainstem. These networks convey visceral, humoral and nociceptive information to the amygdala and mediate its effects on the hypothalamic-pituitary-adrenal axis as well on autonomic and motor centers. On the other hand, interactions between the amygdala and interconnected cortical networks play a crucial role in acquisition, consolidation and extinction of learning relative to the stressful stimulus. Within the scope of this review, current evidence relative to the interaction between the amygdala and cortical networks will be considered in relationship to the integration of the conditioned response to stress.

• Amygdala
• basolateral complex
• associative learning
• central nucleus
• fear conditioning
• memory
• posttraumatic stress disorder
• prefrontal cortex

Introduction

The amygdala, a set of nuclei located within the temporal lobe, plays an important role in the regulation of adaptive behavior. This regulation includes several interrelated domains such as the regulation of homeostasis, feeding, sexual and parenting behaviors. The amygdala accomplishes this regulation through networks of neural connections that include visceral centers as well as complex cortical circuits (Charney and Deutch 1996; McDonald 1998; Pitkanen 2000; Price 1999). By bringing these neural networks together, the amygdala mediates the integration of simple, phylogenetically old behaviors with more sophisticated ones, including the interpretation of complex information on the basis of learning relative to previous exposure to similar events. In this manuscript, the functional role of cortico-amygdalo-cortical circuits is considered in the context of the response to stress, understood as direct or perceived severe challenges to the organism's homeostasis.

Exposure to stressful stimuli elicits a complex set of physiological responses that is highly regulated at multiple levels of the central nervous system. The amygdala is at the core of this regulation (Charney 2003; Herman et al. 2003; McEwen 2003; Sapolsky 2003; Shekhar et al. 2003). Its role reflects the integration of temporally and hierarchically organized functions (Herman et al. 2003). The coordination of behavioral, neuroendocrine and neurochemical responses to a stressful stimulus is accompanied by adaptive mechanisms that include associative memory, memory consolidation and regulation of attention. The role of the amygdala in Pavlovian fear conditioning, a relatively simple instance of such integration, has been particularly well investigated (Davis and Whalen 2001; LeDoux 2000; LeDoux et al. 1990; Maren and Quirk 2004; Pape and Stork 2003; Pare et al. 2004). The most commonly used experimental design involves the use of a tone or light as an emotionally neutral conditioned stimulus (CS) and a footshock as an aversive unconditioned stimulus (US). When the CS is paired with US, the CS eventually becomes sufficient to elicit behavioral and visceral responses to danger similar to those triggered by the US. Growing evidence indicates that the amygdala receives nociceptive stimuli (US) as well as auditory stimuli from the thalamus and auditory cortex (CS) (LeDoux et al. 1990; Mascagni et al. 1993; McDonald 1998; Romanski and LeDoux 1993). Plastic events within the amygdala are thought to be necessary to enable the CS to acquire emotional valence and to trigger visceral and behavioral stress responses (for review see LeDoux 2003). Pavlovian fear conditioning could be thought of as an example of the co-existence and integration within the amygdala of the initial response, acquisition and modulation stages of the conditioned stress response. As such, it will be used in this review to discuss the multifaceted roles that amygdalar subregions and their cortical connections play in this response.

It is important to emphasize that cortico-amygdala networks are preferentially involved in the response to specific types of stressors, such as those that do not directly compromise homeostasis but induce stress on the basis of previous experiences (Figueiredo et al. 2003). It is also important to note that the characteristics of the response to stress change dynamically in relationship with the duration of exposure to the stressor. The effects of chronic stress are qualitatively different from those induced by an acute stressor. In the present review, we chose to focus predominantly on the responses to acute stressors, in order to highlight the role played by dynamic interactions among cortico-limbic regions in the conditioned response to stress. It should not be forgotten, however, that chronic stress induces profound morphological and neurochemical changes in brain regions such as the medial prefrontal cortex (mPFC), amygdala and hippocampus (Cook and Wellman 2004; Czeh et al. 2005; Joels et al. 2004; McEwen 2005; McEwen et al. 2002; Radley et al. 2004; Salm et al. 2004; Stewart et al. 2005; Trentani et al. 2002; Vyas et al. 2003, 2004; Wommack and Delville 2002). In turn, chronic stress-induced changes are likely to alter acute cortico-amygdalo-cortical stress response mechanisms (Correll et al. 2005).

Amygdalar networks and their role in the response to stress

The amygdala consists of a functionally integrated complex of cellular groups. Its thirteen nuclei and cortical areas evolved during different phylogenetical stages and present distinct cytoarchitectonic and neurochemical features as well as specific patterns of connections. Within the scope of this review, we will focus primarily on the central nucleus (CeA) and the basolateral complex (BLC), whose roles in conditioned stress, fear and anxiety have been investigated extensively (Bhatnagar et al. 2004; Cardinal et al. 2002; Charney 2003; Davis 1992a; Goldstein et al. 1996; Herman et al. 2003; LeDoux 2003; Shekhar et al. 2003; Shors and Mathew 1998). Importantly, other components of cortico-amygdalo-cortical networks, such as the medial nucleus of the amygdala, are also known to participate in responses to stress (Crane et al. 2005; Flugge et al. 1992; Ma and Morilak 2005; McDougall et al. 2004; Walker et al. 2005).

The CeA has cytoarchitectonic features that are typical of the striatum, with GABAergic medium spiny projection neurons as its prevalent cell type (McDonald 1982). The BLC includes the lateral (Ln), basal (Bn) and accessory basal (Abn) nuclei. The cytoarchitectonic profile of these cell groups strongly resembles that of the cortex, with a prevalence of glutamatergic pyramidal projection neurons, and subgroups of smaller GABAergic interneurons (McDonald 1985; Millhouse and DeOlmos 1983; Washburn and Moises 1992). Despite their differences, the BLC and CeA function as an integrated unit in processing and responses to emotionally relevant stimuli.

The central nucleus and its autonomic network

The CeA is embedded in visceral circuits (reviewed in Pitkanen 2000) which enable it to play a pivotal role in the coordination of the autonomic, hormonal, neurochemical and motor aspects of the stress response. From this point of view, it has rightfully been described as the autonomic center of the amygdala (Swanson and Petrovich 1998). The CeA receives visceral, humoral and nociceptive information from a broad range of inputs, arising from the hypothalamus, brainstem, midline thalamus, and spinal cord (Cliffer et al. 1991; Ge et al. 2001; Menetrey and De Pommery 1991; Ottersen 1980, 1981; Ottersen and Ben-Ari 1979; Rinaman and Schwartz 2004; Saper and Loewy 1980; Veening 1978b). In turn, it provides substantial inputs to the lateral and medial hypothalamus, which mediate sympathetic and hypothalamo-pituitary-adrenocortical (HPA) axis activation, respectively (Gray et al. 1989; Prewitt and Herman 1998; Price and Amaral 1981; Rosen et al. 1991). The CeA also innervates several midbrain, pons and medulla nuclei, including the peduncolopontine tegmental and parabrachial nuclei, nucleus of the trigeminal nerve, nucleus reticularis pontis caudalis and the dorsal nucleus of the vagus (Fort et al. 1994; Krettek and Price 1978; Petrovich and Swanson 1997; Rosen et al. 1991; Semba and Fibiger 1992; Zhang et al. 2003). These projections mediate the CeA effects on several autonomic functions as well as reflex motor responses such as startle and freezing. The CeA is reciprocally connected to the bed nucleus of the stria terminalis (BNST) (Dong et al. 2001; Krettek and Price 1978; Veinante and Freund-Mercier 1998) and to neurochemically distinct regions such as the ventral tegmental area and substantia nigra pars compacta (dopamine), the raphe system (serotonin) and the locus coeruleus (norepinephrine) (Asan 1998; Commons et al. 2003; Li et al. 1990; Ottersen 1981; Petrov et al. 1994; Peyron et al. 1998; Wallace et al. 1992).

As shown in Figure 1, the CeA receives projections from several amygdalar nuclei (Pare' et al. 1995; Petrovich et al. 1996; Pitkanen and Amaral 1998; Pitkanen et al. 1997). Contrary to the common pattern of intra-amygdalar connections, these projections are not reciprocal. This pattern strongly suggests that the CeA may integrate information from intra- as well as extra- amygdalar sources and indeed represent the gateway between the BLC and the visceral centers. Growing evidence, however, indicates that the CeA may in itself represent a crucial site of plasticity. Various forms of conditioned learning, including the acquisition of contextual conditioning, conditioned suppression and conditioned taste aversion have been shown to require the integrity of the CeA (Bahar et al. 2003; Goosens and Maren 2001; Killcross et al. 1997; Selden et al. 1991). Within this context, it is important to note that moderate projections from sensory-related cortical areas, entorhinal cortex and prefrontal cortex impinge directly on the CeA (Mascagni et al. 1993; McDonald 1998; McDonald and Mascagni 1997; McDonald et al. 1996; Veening 1978a). These direct inputs may provide processed sensory information necessary for specific forms of plasticity in this nucleus. Furthermore, the CeA also modulates the attentional aspects of stimulus processing (Holland and Gallagher 1993, 1999Holland and Gallagher 1993, 1999). This function may largely depend on the CeA projections to cholinergic neurons in the nucleus basalis magnocellularis, which in turn innervates several cortical areas including the posterior parietal cortex (Bucci et al. 1998; Chiba et al. 1995; Han et al. 1999; Holland 1997).

Figure 1 Simplified schematic diagram showing intrinsic connections within the BLC, cortical nucleus (CO), CeA and MeA. Note the intricate pattern of reciprocal connections among Ln, Bn and Abn and between these nuclei and CO and MeA (lighter gray). In contrast, connections of these nuclei with the CeA are generally not reciprocal (darker gray). Other amygdalar nuclei, including the ITC cell masses, were not included for the sake of simplicity. References are in the text.

The basolateral complex and its cortical network

The BLC receives strong projections from a broad array of sensory-related associative cortical areas, including the gustatory dysgranular insular cortex, visceral dysgranular insular cortex, temporal auditory and visual cortical regions, parietal ventral and parietal rhinal cortices (somatosensory) (McDonald 1998) (Figure 2). Massive reciprocal connections link the BLC to the mPFC (infralimbic and prelimbic), perirhinal cortex, hippocampus/entorhinal system and orbitofrontal cortex (Bacon et al. 1996; Barbas and De Olmos 1990; Berretta et al. 2001; Cassell and Wright 1986; Cunningham et al. 2002; Ghashghaei and Barbas 2002; Krettek and Price 1977; McDonald 1998; McDonald and Mascagni 1997; McDonald et al. 1996; Pitkanen et al. 2000; Porrino et al. 1981; Sesack et al. 1989; Shi and Cassell 1999; Vertes 2004; Verwer et al. 1996). As shown in Figure 1, intricate connections link nuclei within the BLC to each other and to other amygdala cells groups (Pitkanen 2000; Pitkanen et al. 1997). Additional BLC inputs include projections originating from the midline and posterior thalamic nuclei, ventral globus pallidus, and from some hypothalamic and pontine nuclei (Haber et al. 1985; Kelley et al. 1982; Moga et al. 1995; Ottersen1980; Ottersen and Ben-Ari 1979; Turner and Herkenham 1991; Woolf and Butcher 1982). Neurochemically distinct ventral tegmental area, raphe system and locus coeruleus also project to the BLC (Asan 1998; Nitecka et al. 1980; Vertes 1991). The BLC sends substantial projections to the nucleus accumbens, caudoputamen, BNST, substantia innominata, and mediodorsal nucleus of the thalamus (Cornwall and Phillipson 1988; Groenewegen 1988; Grove 1988; Kelley et al. 1982; Kita and Kitai 1990; Krettek and Price 1978; McDonald 1987, 1991; Ray and Price 1992; Russchen and Price 1984; Weller and Smith 1982).

Figure 2 Simplified diagram depicting BLC-cortical interactions within the context of acquisition, modulation and consolidation of stress-related memories. Inputs from a broad range of associative sensory cortical areas, representing all sensory modalities, converge in the BLC. Multifaceted information relative to specific events can then be associated with its emotional valence. The BLC mediates long-term consolidation of stress-related memories through its projections to several cortical (and subcortical, not shown) regions. Finally, projections from prefrontal cortex to the amygdala are thought to modulate the expression of learned responses on the basis of the current valence of specific stimuli. For the sake of simplicity, the list of the cortical areas included is not intended to be comprehensive and the complexity of their connections with the amygdala is not fully represented. References are in the text.

Thus the BLC is embedded within a vast network of cortical regions and is strongly connected with the basal ganglia and limbic thalamus. These circuits are crucial to the regulation of the stress response. Growing evidence indicates that the main contributions of the BLC to this response are the formation and retrieval of associative memories and the modulation of memory consolidation relative to stressful events.

Projections from sensory association cortical areas to the BLC—associative memory

Associative learning could be defined as neural plastic events resulting from the convergence of sensory stimuli and biologically relevant information (Fanselow and Poulos 2005). The idea that the amygdala plays a crucial role in associative learning is a relatively old one (Jones and Mishkin 1972). The involvement of this nuclear complex in multiple forms of associative learning, including cue-, place- and object- reward associations, conditioned taste aversion, appetitive conditioning and conditioned fear and anxiety, is now well recognized (for reviews see Baxter and Murray 2000; Davis 2000; Easton and Gaffan 2000; Gallagher 2000; Holland et al. 2002; Holland and Gallagher 1999; for reviews see McGaugh 2002; White and McDonald 2002). Pavlovian fear conditioning has been used extensively to investigate the role of the BLC/CeA in associative learning. In the simple form of this paradigm illustrated above, thalamic inputs relative to CS and US converge in the Ln (LeDoux 2003). While information originating from the thalamus is likely to support conditioning of simple sensory stimuli, sensory association and mesiotemporal cortical areas provide the BLC with a remarkable array of highly processed information related to complex sensory stimuli (Figure 2). The BLC is thus in an ideal position to integrate sensory information both within and across modalities and to use it to attach emotional value to newly acquired experiences in order to guide adaptive behavior. During a stressful event, the convergence in the BLC of complex information relative to explicit sensory stimuli and environmental context (conditioned) and visceral (unconditioned) information relayed by hypothalamic and midline thalamic nuclei, may generate multiple orders of conditioned learning. This newly learned information will be used in the future to predict and react appropriately to future presentation of similar events.

BLC projections to the cortex—modulation of memory consolidation

Newly acquired short-term memories can be transformed, over time, into more stable long-term memories (Davis and Squire 1984; McGaugh 2000). The capability of turning the experience of aversive events into long lasting memories is of obvious importance for the survival of the organism. It is thus not surprising that stress hormones and neuromodulators, namely glucocorticoids and norepinephrine, have powerful effects on memory consolidation (Gold and Van Buskirk 1975; Hui et al. 2004; Izquierdo and Dias 1983; Lupien and McEwen 1997; McEwen and Sapolsky 1995; McGaugh 2004; Okuda et al. 2004; Roozendaal 2002; Roozendaal et al. 2004). Overwhelming evidence indicates that the BLC plays a crucial role in mediating the effects of stress hormones on memory consolidation (Barros et al. 2002; Bianchin 1999; Lalumiere et al. 2004; McGaugh 2002; McGaugh 2004; Schafe and LeDoux 2000). Exposure to stress is associated with release of norepinephrine in the BLC, mediated by the nucleus of the solitary tract directly or via its projections to the locus coeruleus, which in turn innervates the BLC (Herman et al. 2003; Myers and Rinaman 2002; Williams et al. 1998). Stress also results in increased release of glucocorticoids from the adrenal cortex. Although centrally-released norepinephrine and glucocorticoids reach, and profoundly affect, several brain regions, it is the concurrent activation and interaction of beta-adrenergic and glucocorticoid receptors within the BLC that mediates their effects on memory consolidation (for review see McGaugh 2004). Within BLC intrinsic circuits, this interaction is also affected by other neurotransmitters such as acetylcholine, GABA and glutamate (McGaugh 2004; Pare 2003; Power et al. 2000, 2003).

A growing body of evidence supports the idea that these neuromodulatory interactions affect the output of the BLC to several of its target regions and that it is in these regions that memory consolidation ultimately occurs (for review see McGaugh 2004; Pare 2003) (Figure 2). BLC effects on memory consolidation in the nucleus accumbens, hippocampus, entorhinal and insular cortices have been described (Liang and McGaugh 1983; Miranda and McGaugh 2004; Roesler et al. 2002; Roozendaal and McGaugh 1997; Roozendaal et al. 1999; Setlow et al. 2000). Emerging from these studies is the notion that the BLC modulates long-term memory consolidation relative to different forms of learning occurring in distinct brain regions. Interestingly, the BLC has also been found to mediate other forms of stress hormone-dependent effects on memory, such as impairment of memory retrieval and working memory (Nathan et al. 2004), functions typically attributed to the hippocampus and prefrontal cortex, respectively. Such highly diversified effects suggest that the BLC may play a far-reaching role in the modulation of stress-related learning in several cortical regions.

The mechanisms underlying the modulation of stress-related memories are not well understood at present. In some cortical regions, such as the hippocampus, effects of the amygdala on specific plastic events have been demonstrated (Akirav and Richter-Levin 1999; Frey et al. 2001; Ikegaya et al. 1995; Pelletier and Pare 2004). The hippocampus receives both direct and indirect inputs from the BLC (Berretta et al. 2001; Pitkanen et al. 2000). Activation of β-adrenoceptors in the BLC affects memory consolidation in the hippocampus (Roozendaal et al. 1999). For example, such activation has been recently shown to be associated with the induction of the immediate early gene Arc in the hippocampus as well as with changes in performance of an inhibitory avoidance task (McIntyre et al. 2005). Electrophysiological evidence indicates that stimulation of the BLC facilitates long-term potentiation in the dentate gyrus, while intra-amygdala infusion of β-adrenergic or muscarinic receptor antagonists has the opposite effect (Akirav and Richter-Levin 1999; Frey et al. 2001; Ikegaya et al. 1997, 1995, 1996). Modulation of synaptic plasticity may in part account for BLC effects on memory consolidation in the hippocampus. In addition, synchronized oscillatory activity in the BLC and hippocampus has also been proposed as a possible mechanism for enhancing memory consolidation (Pape and Stork 2003; Pelletier and Pare 2004). Morphological and neurochemical changes may also accompany such effects. Infusion of a GABA_A receptor antagonist in the BLC profoundly alters hippocampal inhibitory intrinsic circuits (Berretta et al. 2004, 2001). These effects are remarkably long lasting as substantial changes affecting specific hippocampal interneuronal subpopulations are detectable 96 h following an acute infusion (Berretta et al. 2004). Interestingly, these intrinsic circuits are known to control oscillatory patterns in the hippocampus (for review see Freund and Buzsaki 1996). It is also important to note that GABA has been shown to inhibit norepinephrine release within the BLC (Hatfield et al. 1999), suggesting that blockade of GABA_A receptors may instead increase norepinephrine release and thus mimic a state of emotional arousal. This possibility is supported by evidence showing that infusion of GABA receptor antagonists in the amygdala also induce physiologic and behavioral changes associated with a defense reaction (Sajdyk and Shekhar 1997; Sanders et al. 1995; Sanders and Shekhar 1995; Thielen and Shekhar 2002). It is possible that long-lasting changes relative to specific hippocampal intrinsic GABAergic networks may represent a component of the stress-induced enhancement of memory consolidation mediated by the amygdala.

Medial prefrontal cortex modulates the amygdala response to stress

The mPFC has been shown to play an important modulatory role in the response to stress (Amat et al. 2005; Diorio et al. 1993; McDougall et al. 2004; Spencer et al. 2005; Sullivan and Gratton 1999). Growing evidence supports the idea that the amygdala is one of the main targets of this modulation (Correll et al. 2005; Figueiredo et al. 2003). We bring here as an example the role that the mPFC and amygdala play in the phenomenon of extinction of fear conditioning.

As discussed above, associative memory of aversive stimuli is crucial to the organism's capability of predicting danger and responding appropriately. Once instated, conditioned fear associations are especially long lasting, or possibly indelible (LeDoux et al. 1989), even in the absence of further exposure to coincident occurrence of the conditioned and unconditioned stimuli. A new learning event, defined as extinction of fear conditioning, has to take place in order to inhibit the stress response to a CS that no longer predicts danger (Davis et al. 2003; Rescorla 2001). This mechanism is particularly important in light of the fact that responses to aversive stimuli, with their associated behavioral, visceral and hormonal components, imply high energetic costs as well as risks and, thus, have to be weighed against the actual likelihood of the occurrence of a stressful event.

Circuits reciprocally linking the amygdala to the mPFC have been shown to be critically involved in the modulation of fear responses during extinction (for reviews see Maren and Quirk 2004; Pare et al. 2004) (Figure 2). The functional integrity of the amygdala is crucial, as demonstrated by experiments using local pharmacological manipulations (Falls et al. 1992; Lu et al. 2001). Growing experimental evidence shows that the mPFC plays an essential role in long-term extinction memory (Morgan et al. 1993; Quirk et al. 2000; Santini et al. 2004). In rats not exposed to extinction learning, electrical stimulation of the infralimbic cortex during exposure to a CS reduces conditioned responses (Milad and Quirk 2002). These researchers have also demonstrated that electrical stimulation of the infralimbic cortex produces an inhibition of brainstem-projecting CeA neurons (Quirk et al. 2003). Following extinction training, infralimbic neurons respond to the CS with a strong increase of their activity (Milad and Quirk 2002).

Together, these results indicate that the mechanisms mediating extinction of conditioned fear may involve activation of the mPFC and consequent inhibition of the CeA leading to a failure to trigger a stress response through activation of the hypothalamus and brainstem. It has been argued recently that intercalated (ITC) cells of the amygdala are ideal candidates to mediate the infralimbic-driven inhibition of CeA neurons (Pare et al. 2004; Quirk et al. 2003; Royer and Pare 2002). ITC cell masses are clusters of GABAergic neurons that form a net throughout the rostral half of the lateral-basolateral nuclear complex (McDonald and Augustine 1993; Millhouse 1986; Nitecka and Ben-Ari 1987; Pare and Smith 1993a; Pitkanen and Amaral 1994). These neurons are known to receive projections from the BLC and to send inhibitory inputs to the CeA (Delaney and Sah 2001; Millhouse 1986; Pare and Smith 1993a,bPare and Smith 1993a, b; Royer et al. 1999). The infralimbic cortex also sends strong projections to ITC cell masses (McDonald et al. 1996) (Berretta et al. unpublished observation). Experiments from our laboratory, using induction of the immediate early gene Fos as an index of neuronal activation, indicate that infralimbic inputs increase the activity of ITC cells (Berretta et al. 2005). This effect might explain how infralimbic stimulation inhibits CeA neurons (Quirk et al. 2003) and the expression of conditioned fear responses (Milad and Quirk 2002). Thus, infralimbic activation of ITC neurons may play a major role in modulating the response of brainstem projecting CeA neurons to CS-related inputs arising in the BLC.

Involvement of cortico-amygdalo-cortical circuits in psychiatric disorders

The cortico-amygdalar networks described above have been found to be involved in a remarkable array of psychiatric diseases, including anxiety disorders, major depression, schizophrenia, and bipolar disorder. Importantly, stress and anxiety appear to be important components of many of these diseases. Recent views on post traumatic stress disorder (PTSD) envision this condition in terms of fear conditioning and postulate that an impairment of extinction of conditioned responses may play an important role in this disease (Gorman et al. 2000; Pitman et al. 2001). Imaging studies in PTSD patients have shown altered activation of the amygdala and of several cortical regions such as the anterior cingulate gyrus and lateral anterior prefrontal cortex as well as occipital and parietal regions that support visual and spatial aspects of stimuli representations (Bremner 2002; Lanius et al. 2002; Lanius et al. 2003; Liberzon and Phan 2003; Pissiota et al. 2002; Rauch et al. 1996; Shin et al. 1997; Shin et al. 2004). Although these results are still controversial, they are generally consistent with current knowledge on conditioned fear brain circuits and support the hypothesis that persistent, intrusive memory of an intensely traumatic event may be mediated by increased amygdalar activity (Grillon et al. 1996; Rauch et al. 2000). An important component of this hypothesis is the failure of cortical inhibitory mechanisms to reduce amygdala activity, and thus mediate extinction learning (Armony and LeDoux 1997; Charney 2003). Recently, increased activation of the visual cortex induced by amygdala hyperactivity has also been suggested to play an important role by mediating emotion-driven retrieval of visual images relative to the traumatic event (Gilboa et al. 2004). In summary, it is reasonable to propose that altered cortico-amygdalo-cortical interactions may represent an important factor in the pathogenesis of PTSD.

Similar mechanisms could be involved in the pathogenesis of panic and other anxiety disorders (Gorman et al. 2000). As an example, imaging investigations suggest the involvement of the amygdala and frontal and temporal cortical regions in panic disorder (Brambilla et al. 2002; Eren et al. 2003; Massana et al. 2002, 2003; Thomas et al. 2001; Wik et al. 1997). The role played by the amygdala in this disorder, and more generally in anxiety, has been investigated extensively using animal models (Davis 1992a,b; Shekhar et al. 2003, 1999). These studies indicate that altered neuroregulatory mechanisms in the amygdala, including disruption of GABAergic inhibition and stimulation of corticotropin releasing factor receptors result in anxiety-like behavior (Sajdyk et al. 1999; Sanders and Shekhar 1995). Interestingly, this latter effect depends on NMDA receptors (Rainnie et al. 2004), suggesting that excitatory inputs to the amygdala may play an important role in these responses.

Conclusions

The amygdala and the massive cortical networks in which it is embedded, regulate multiple stages and aspects of the stress response (Figure 2). Connections with the hypothalamus and brainstem, mediated primarily by the CeA, provide a broad array of visceral, hormonal and humoral information and allow the amygdala to orchestrate comprehensive, multifaceted responses to a stressful event. Within cortico-amygdalo-cortical circuits, such responses are strongly associated with plastic events, from associative learning triggered by initial exposure to an aversive event to long term memory consolidation and, in some cases, to extinction of conditioned responses according to the current predictive value of the CS. Given the importance and complexity of these mechanisms, it is not surprising that cortico-amygdala-cortical circuits have been found to play an important role in a broadening number of psychiatric conditions involving anxiety and vulnerability to stress.