Role of stress, corticotrophin releasing factor (CRF) and amygdala plasticity in chronic anxiety

Abstract

Stress initiates a series of neuronal responses that prepare an organism to adapt to new environmental challenges. However, chronic stress may lead to maladaptive responses that can result in psychiatric syndromes such as anxiety and depressive disorders. Corticotropin-releasing factor (CRF) has been identified as a key neuropeptide responsible for initiating many of the endocrine, autonomic and behavioral responses to stress. The amygdala expresses high concentrations of CRF receptors and is itself a major extrahypothalamic source of CRF containing neurons. Within the amygdala, the basolateral nucleus (BLA) has an important role in regulating anxiety and affective responses. During periods of stress, CRF is released into the amygdala and local CRF receptor activation has been postulated as a substrate for stress-induced alterations in affective behavior. Previous studies have suggested that synaptic plasticity in the BLA contributes to mechanisms underlying long-term changes in the regulation of affective behaviors. Several studies have shown that acute glutamate receptor-mediated activation, by either GABA-mediated disinhibition or CRF-mediated excitation, induces long-term synaptic plasticity and increases the excitability of BLA neurons. This review summarizes some of the data supporting the hypotheses that stress induced plasticity within the amygdala may be a critical step in the pathophysiology of the development of chronic anxiety states. It is further proposed that such a change in the limbic neural circuitry is involved in the transition from normal vigilance responses to pathological anxiety, leading to syndromes such as panic and post-traumatic stress disorders.

• Stress disorders
• basolateral nucleus
• corticotropin-releasing factor
• psychiatric syndromes

Stress initiates a cascade of hormonal and behavioral responses that enable an organism to adapt to new environmental pressures. However, chronic stress may lead to pathological states, which can result in several psychiatric syndromes such as anxiety and depressive disorders. Corticotropin-releasing factor (CRF), a 41 amino acid neuropeptide, has been identified as a key neuropeptide responsible for initiating many of the endocrine, autonomic and behavioral responses to a variety of stressors. In addition to its effects on regulating adrenocorticotropin (ACTH) release (Vale et al. 1981; Rivier and Plotsky 1986), CRF is a putative neurotransmitter that is implicated in the stress/anxiety responses [see (Dunn and Berridge 1990)]. CRF injections into the cerebral ventricles induce anxiety-like responses in several animal tests of anxiety, such as the conflict test (Britton et al. 1986), social interaction test (Dunn and File 1987), acoustic startle (Swerdlow et al. 1986) and elevated plus maze (File et al. 1988). Transgenic mice that overproduce CRF shows increased anxiety-like behaviors that are reversed by central administration of CRF antagonists (Stenzel-Poore et al. 1994). Currently, there are at least two different CRF receptors identified (the CRF₁ and the CRF ₂ receptors). The anxiogenic behaviors produced by CRF in rats may be due to stimulation of the CRF₁ receptor. Recent evidence suggests that dysfunction of systems regulating CRF function in the CNS contributes to the etiology of several psychiatric disorders such as anxiety, panic disorder, posttraumatic stress disorder (PTSD) and depression (Arborelius et al. 1999; Boyer 2000).

Neural substrates of anxiety

The neural pathways involved in generating and regulating anxiety-like responses are still poorly understood. A number of CNS structures have been implicated. A large number of studies suggest a primary role for the amygdala in anxiety responses (Hilton 1963; Kaada 1972; Davis 1992; Kapp et al. 1992; LeDoux 1992; Sanders and Shekhar 1995a,b; Adolphs et al. 2002; Davis and Myers 2002). In addition, the septo-hippocampal system (Gray 1996), and the brain stem centers of the global monoamine projections such as the locus coeruleus (LC) and the raphé nuclei appear to have important regulatory roles in these responses (Price and Amaral 1987; Charney et al. 1992; Price et al. 1995; Aston-Jones et al. 1996; Valentino et al. 2001; Baxter and Murray 2002; Linthorst et al. 2002; Lowry 2002; Keck et al. 2005). Thus, a network of CNS sites extending from the neocortex (particularly frontal and temporal regions) down to the brainstem monoamine pathways appear to form an intricately connected neural axis that may regulate anxiety and panic responses (Shekhar et al. 2002). A regulatory dysfunction at one or more of the key sites in this network could lead to an abnormal pattern of affective responses and the development of recurrent emotional disorders.

Role of the amygdala in anxiety states

The amygdala is particularly thought to be critical for providing affective salience to sensory information (Adolphs 1999). Recent fMRI studies report that increased amygdala activation is associated with several anxiety and mood states (Adolphs et al. 2001; Baxter and Murray 2002; Anand and Shekhar 2003). The activity of the amygdala is regulated by a balance between glutamate induced excitation and GABA-mediated inhibition (Rainnie et al. 1991a,b), and this is particularly important for the regulation of anxiety responses (Sajdyk and Shekhar 1997a,b). Experimental evidence suggests that the amygdala may be a critical site for the anti-stress effects of the benzodiazepine (BZs) group of drugs (Petersen et al. 1985; Sanders and Shekhar 1995a,b). The role of the specific nuclei of the amygdala in generating negative affective responses has been further characterized by studying the development of conditioned fear responses (Davis 1992; LeDoux 1992). Within the amygdala, the basolateral (BLA) and the central (CE) nuclei have long been known to regulate affective responses (Hilton 1963; Kaada 1972; Davis 1992; Kapp et al. 1992; LeDoux 1992; Sanders and Shekhar 1995a,b; Adolphs et al. 2002; Davis and Myers 2002). The CE appears to be the primary efferent site of the amygdala in eliciting conditioned fear responses (LeDoux et al. 1988; Kapp et al. 1992). Disrupting the function of the BLA inhibits both the acquisition of conditioned fear/negative affective responses as well as retrieval of information necessary for the expression of emotion (Campeau and Davis 1995). Hence, it has been proposed that the BLA serves as a major integrator and relay center for the sensory and memory information necessary for anxiety responses (Campeau and Davis 1995). While there are many excellent reviews of the function of the different regions of the amygdala in conditioned fear (Davis et al. 1997a,b; LeDoux 2003), reward salience (Holland and Gallagher 2004), drug abuse (Koob 1999; See et al. 2003) and other behavioral disturbances (Stevens 1999; Post 2002), the present review will specifically focus on the BLA and its role in anxiety-like behaviors.

While the role of the amygdala in anxiety-like behaviors is generally supported, there are significant aversive behaviors that could be independent of the amygdala. For example, adaptation of hypothalamic–pituitary–adrenal responses to chronic restraint stress appears to be independent of the amygdala (Carter et al. 2004). Lesions of the amygdala are also not uniformly effective in disrupting all types of anxiety or emotional behaviors (Treit et al. 1998; Amaral 2002; Anderson and Phelps 2002; Blundell et al. 2003). Thus, this review will primarily focus on the specific anxiety-like behaviors such as non-cued, exploratory and conditioned anxiety models that have a strong implication of amygdala involvement.

CRF function in the amygdala and its role in emotional salience to stress cues

The amygdala is also a major extra-hypothalamic source of CRF-containing neurons and has high expression levels for the two cognate CRF receptors (Reul and Holsboer 2002). During periods of stress, CRF is released into the amygdala (Koob and Heinrichs 1999) and local CRF receptor activation has been postulated as a substrate for stress-induced alterations in affective behavior (Dunn and Berridge 1990; Lee et al. 1994, 1996; Lee and Davis 1997; Sajdyk et al. 1999; Sajdyk and Gehlert 2000). Our previous studies have suggested that the BLA is a critical site of synaptic plasticity that contributes to changes in affective behavior, and we have shown that acute glutamate receptor mediated activation, by either GABA disinhibition or CRF excitation, induces long-term synaptic plasticity and increases the excitability of BLA neurons (Sanders et al. 1995; Sajdyk et al. 1999; Sajdyk and Gehlert 2000; Shekhar et al. 2003).

Not all the studies implicating CRF in the amygdala in stress and anxiety are pharmacological or genetic manipulation studies. The increases in anxiety-like responses following restraint stress appear to be mediated by CRF receptor activation in the amygdala (Gehlert et al. 2005). Restraint or drug withdrawal stress releases CRF in the amygdala (Merlo Pich et al. 1995), as does neonatal stress (Cratty et al. 1995) and CRF mRNA is increased in the amygdala following stress (Herringa et al. 2004). Chronic, repeated stress results in specific synaptic plasticity within the amygdala (Vyas et al. 2002). These data suggest that a stress-induced increase of CRF release in the amygdala may contribute to altered emotional states often associated with chronic stress.

The basolateral complex, comprising the BLA and the lateral nucleus, receives highly processed multimodal information from several cortical regions including the visual, auditory, somatosensory and olfactory cortex. The basolateral complex is believed to process these inputs and allocate emotional salience to the sensory inputs (LeDoux; McDonald 1998; Pitkanen et al. 2000). The process of allocating emotional salience to sensory stimuli within the BLA is thought to play an integral role in emotional learning, and consequently any disruption of this nucleus disrupts the memory enhancing effects of emotionally salient cues (Ferry et al. 1999; Roozendaal et al. 2001; McGaugh and Roozendaal 2002; Goosens and Maren 2003; LaLumiere et al. 2003; Roesler et al. 2003). Furthermore, accumulating evidence suggests that CRF release within the amygdala and subsequent activation of its cognate receptors (especially in the BLA) may be critical for stress or negative cue-induced behavioral responses (Heilig et al. 1994; Lee et al. 1994; Gray and Bingaman 1996; Sajdyk et al. 1999; Roozendaal et al. 2002; Shekhar et al. 2003). Therefore, activation of CRF receptors in the BLA, of which the CRF₁ receptor seems to be the primary subtype (Rainnie et al. 1992; Sanchez et al. 1999; Van Pett et al. 2000), appears critical for attributing emotional salience to a variety of cues eliciting “anxiety” or negative emotions (Sajdyk et al. 1999; Roozendaal et al. 2002; Rainnie et al. 2004).

The exact source of CRF input to the BLA is still somewhat unclear. There are CRF positive cell bodies in the Ce and the bed nucleus of the stria terminalis (BNST), both of which project to the BLA and other autonomic regulatory areas (See Figure 1; (Swanson et al. 1983; Van Bockstaele et al. 1999; Van Pett et al. 2000). Thus, it is likely that the bulk of the CRF input to the BLA comes from within the extended amygdala. There are sparse CRF cells in the cortex and dentate regions (Palkovits et al. 1983), which could also contribute CRF input to the BLA.

Figure 1 A schematic representation of the sites of action of the CRF system within the amygdala and within an extended “emotional” network that may be involved in normal stress responses and may be critical in the development of chronic stress induced anxiety states. Many other regions modulated by the basolateral amygdala such as the nucleus accumbens and orbitofrontal cortex are not shown for the sake of simplicity and the limited focus on regions critical for anxiety-like behaviors. The central nucleus and the bed nucleus of the stria terminalis are presented as a single group since they have similar afferents from the BLA and efferent projections out of the extended amygdala. Projection neurons are presented as circles within the different brain regions. Abbreviations: BLA, basolateral amygdala; BNST, bed nucleus of the stria terminalis; CE, central nucleus of the amygdala; CRF, corticotrophin releasing factor; CRF1 and 2, CRF receptor types1 and 2; GABA, gamma aminobutyric acid; Glu, glutamate; 5-HT, serotonin; mPFC, medial prefrontal cortex; NE, norepinephrine.

Once sensory stimuli have acquired emotional salience, via neural mechanisms involving the BLA, this information is then relayed to efferent target sites within the amygdala (see Figure 1), specifically the Ce and BNST (McDonald 1998; Pitkanen et al. 2000) for some types of anxiety behaviors, and elsewhere for other functions, such as nucleus accumbens for positive reward cues (Louilot and Besson 2000; Floresco et al. 2001), hippocampus for memory consolidation (Akirav and Richter-Levin 2002; McGaugh et al. 2002; Roozendaal et al. 2002, 2003) and frontal cortex for stimulus coding (Schoenbaum et al. 1998; Schoenbaum 2003). These other functions of the BLA are outside the focus of the current review and information can be obtained from several excellent reviews (Baxter and Murray 2002; McGaugh 2002; Holland and Gallagher 2004).

Several lines of converging evidence support the role of the extended amygdala target areas in the expression of some forms of aversive emotional responses, particularly anxiety-like behaviors (LeDoux et al. 1988; LeDoux 1992; Campeau and Davis 1995; Davis 1998; Hall et al. 2001; Royer and Pare 2002; Kalin et al. 2004). Significantly, recent evidence suggests that CRF receptor activation within the BNST and Ce also plays a role in modulating negative affect (Davis et al. 1997; Merali et al. 1998; Richter-Levin and Akirav 2000; Yilmazer-Hanke et al. 2002; Bale and Vale 2004; Campbell et al. 2004).

Internal circuitry of BLA

Based on a number of anatomical similarities, the BLA has been proposed to closely resemble the adjacent neocortex (McDonald and Pearson 1989; McDonald 1992). Hence, the BLA consists of two major classes of neurons, the large pyramidal (projection) cells and the smaller non-pyramidal cells (McDonald 1982, 1992a,b). The non-pyramidal cells of the BLA are similar to their cortical counterparts and are immuno-positive for GABA, acetylcholine and a variety of peptides such as somatostatin (SOM), neuropeptide-Y (NPY), vasoactive intestinal peptide (VIP) and cholecystokinin (CCK) (McDonald and Pearson 1989; Roberts 1992). It appears that in the BLA, similar to the cortex, the non-pyramidal cells impinge on the pyramidal cells and modulate the activity of these projection neurons (McDonald and Mascagni 2001a,b,c). In contrast, projection neurons of the BLA are excitatory in nature and utilize the excitatory amino acid (EAA) glutamate as their primary neurotransmitter (McDonald 1996).

The BLA receives substantial extrinsic EAA inputs from cortical and subcortical structures, as well as intrinsic EAA inputs from the lateral nucleus of the amygdala (LA) (Price and Amaral 1987; McDonald 1996). Not surprisingly, therefore, both pyramidal neurons and non-pyramidal neurons of the BLA express the N-methyl-d-aspartate (NMDA) and the non-NMDA subtypes of EAA receptors (McDonald 1994). Moreover, stimulation of either extrinsic or intrinsic afferent pathways elicits a fast-excitatory postsynaptic potential (f-EPSP) i.e. mediated by non-NMDA receptor activation and a slow-EPSP that is mediated by NMDA receptor activation in pyramidal neurons of the BLA (Rainnie et al. 1991b). The BLA is also rich in GABA_A and GABA_B receptors. Similar to the glutamate mediated EPSPs, stimulation of the afferent pathways to the BLA also elicit GABA_A receptor-mediated fast inhibitory postsynaptic potentials (f-IPSPs) and GABA_B receptor-mediated slow inhibitory postsynaptic potentials (s-IPSPs) in the BLA projection neurons (Rainnie et al. 1991a). Thus, a coordinated balance of glutamate-mediated excitation and GABA-mediated inhibition of the BLA projection neurons appear to be critical for regulation of behavioral responses. Disruption of this balance appears to result.

Significantly, electrophysiological studies have demonstrated that the firing activity of the pyramidal neurons is tonically regulated by inhibitory GABAergic interneurons of the BLA (Gean and Chang 1991; Rainnie et al. 1991a,b; Smith and Dudek 1996; Mahanty and Sah 1998; Rainnie 1999). These interneurons constitute only 15% of the total neuronal population in the BLA (McDonald 1982, 1992a,b) and yet a network of at least four distinct subclasses of interneurons exist, based on their expression of calcium binding protein and neuropeptide content (McDonald and Betette 2001; McDonald 2001a,b,c, 2002; Mascagni and McDonald 2003; Muller et al. 2003). In addition to their unique combination of protein/neuropeptide content, these interneurons also appear to be functionally distinct. Hence, each class of interneuron can be further differentiated by its efferent connectivity (McDonald and Betette 2001; McDonald and Mascagni 2002; McDonald et al. 2002; Muller et al. 2003a,b), and by its neurotransmitter receptor profile (Morales and Bloom 1997; McDonald and Mascagni 2001a,b,c; Blurton-Jones and Tuszynski 2002).

This complex arrangement of inhibitory neurons may be key to understanding the function of the BLA in emotional regulation. Deficits in this inhibitory modulation may be linked to the etiology and maintenance of several pathophysiological conditions (Rainnie et al. 1991a,b; Sanders and Shekhar 1991; Rainnie et al. 1992a,b; Davis et al. 1994; Pitkanen et al. 1997; Benes and Berretta 2001). In particular, blocking the GABAergic inhibition in the BLA with bicuculline methiodide (BMI) results in anxiety-like responses (Sanders and Shekhar 1991; Sajdyk and Shekhar 1997a,b). Moreover, repeated disruption of the inhibition of the BLA leads to the development of long-term synaptic facilitation and a behavioral state of chronic anxiety and susceptibility to panic-like reactions (Sanders et al. 1995; Sajdyk and Shekhar 2000). Furthermore, increased stimulation of the BLA by repeatedly injecting the CRF receptor agonist urocortin I (UCN I) leads to a persistent state of anxiety- or panic-like symptoms in the rat (Sajdyk et al. 1999; Shekhar et al. 2003a,b).

Although, the work cited above demonstrates that an obvious link exists between the steady state activity of the BLA circuitry and expression of anxiety, little is known of the mechanism by which the inhibitory tone is altered in the BLA during pathological anxiety. For example, we recently demonstrated that UCN I-induced stimulation of the BLA results in persistent expression of anxiety-like behaviors that is associated with a reduction in the steady state inhibition of BLA pyramidal cells by GABAergic interneurons, and that this effect was associated with a reduction in GABA_A receptor-mediated IPSPs in projection neurons (Rainnie et al. 2004). However, the mechanism responsible for this reduction in inhibition has not yet been determined. Further characterizing the pattern of activation of specific neuronal populations within the amygdala following treatments that lead to anxiety-like and depression-like behaviors in the rat would be the next critical step in elucidating the role of the amygdala in these behaviors.

Long-term synaptic plasticity in the BLA

The basal activity of the projection neurons in the BLA is maintained by a balance between the excitatory (both non-NMDA and NMDA-mediated) input, and local inhibitory input from GABAergic non-pyramidal cells (GABA_A and GABA_B receptor-mediated inhibition). Upon stimulation of the afferent inputs to the BLA, NMDA receptor-mediated excitation of BLA projection neurons temporally overlaps with the GABA_A receptor-mediated inhibition and, as a consequence, is generally unable to drive the cell towards NMDA receptor-mediated stimulation (Rainnie et al. 1991a,b). This GABA_A receptor-mediated inhibition of NMDA receptor activation is a key step in limiting the NMDA receptor-mediated calcium flux into the amygdala projection neurons and thus preventing the initiation of plasticity within these cells. However, If for any reason this inhibitory GABAergic tone onto BLA projection neurons is reduced, this could unmask the NMDA receptor-mediated excitation, leading not only to greater activation of the cell but also to long-term changes in synaptic plasticity that are dependent on NMDA receptor activation (Rainnie et al. 1991b).

The BLA exhibits a high degree of synaptic plasticity such as long-term potentiation (LTP), which is believed to be a cellular substrate for synaptic changes that may occur during learning, and which results from a cascade of events that are initially triggered by EAA receptor activation (Collingridge et al. 1983; Brown et al. 1988; Collingridge and Bliss 1995). Although, most studies of LTP have been conducted in the hippocampus (Brown et al. 1988), the amygdala, and in particular the BLA and LA, (Maren 1999) are also known to exhibit LTP, both in vitro (Chapman et al. 1990; Chapman and Bellavance 1992), and in vivo (Clugnet and LeDoux 1990; McKernan and Shinnick-Gallagher 1997). These forms of long-term synaptic changes are thought to be critical in learning and memory and are primarily brought about by an EAA receptor-mediated cascade of events (Collingridge et al. 1983; Brown et al. 1988; Collingridge and Bliss 1995).

There is much work being done to understand the intra-cellular mechanisms involved in the development of synaptic plasticity in the amygdala. Most studies support the role of the NMDA receptors in the initiation of the synaptic plasticity (Miserendino et al. 1990; Gewirtz and Davis 1997; Rogan et al. 1997; Adamec 1998; Maren 1999; Rainnie et al. 2004) although exceptions have been noted (Chapman and Bellavance 1992). There have been many intracellular second messenger cascades implicated further in the development of plasticity and LTP within the amygdala including activation of several types of kinases, including mitogen-activated protein (MAP) kinase (Huang et al. 2000; Schafe et al. 2000), calcium calmodulin kinases (Wei et al. 2002; Rainnie et al. 2004), protein kinase A (Huang and Kandel 1998; Huang et al. 2000), and extracellular signal-regulated kinase (Schafe et al. 2000). These kinases affect other signaling molecules such as cAMP response element-binding protein (CREB) and growth factors (Lamprecht et al. 1997; Hall et al. 2001; Josselyn et al. 2001; Rattiner et al. 2005), which eventually induce specific genomic changes (Lin et al. 2003a,b). Such molecular mechanisms of amygdala synaptic plasticity are reviewed elegantly elsewhere (Blitzer et al. 2005; Levenson and Sweatt 2005; Xia and Storm 2005).

Proposed intrinsic BLA circuit regulating anxiety responses

Based on the neuroanatomical and functional organization of the BLA outlined above, a simple BLA circuit can be proposed that is capable of integrating incoming sensory information and assigning the appropriate salience and vigilance to that sensory information, as shown in Figure 2.

Figure 2 A schematic drawing showing a highly simplified circuit within the BLA regulating emotional responses. The BLA is represented as a series of inhibitory cells ( − ) arranged around the projection neurons to integrate two sets of inputs: a stream of processed and direct sensory information (+; directly excitatory on the projection neurons) and the cortical input about predicted ability to cope with the given sensory inputs (inhibitory to the projection neurons; probably via the local inhibitory inputs). During chronic stress, in vulnerable individuals, the balance of these two inputs onto the projection neuron is disrupted, resulting in pathological anxiety and mood responses. Ascending serotonin and norepinephrine pathways provide modulatory tone that regulates this balance of stimulus and coping inputs, thus capable of up- or down-regulating the individuals ability to adapt to stress. The disruption of these modulatory mechanisms in extreme stress or in vulnerable populations, e.g. such as those with genetic polymorphisms in the monoamine modulatory systems, may also result in pathological stress reactivity. While this is clearly an oversimplification, similar principles can be applied to model the BLA function with multiple feedback loops containing many neurotransmitters and neuropeptides. Abbreviations: 5-HT, serotonin; BLA, basolateral amygdala; BNST, bed nucleus of the stria terminalis; CB, calbindin; CCK, cholecystokinin; CE, central nucleus of the amygdala; Cg, cingulate cortex; CR, calretinin; DRN, dorsal raphe nucleus; GABA, gamma-aminobutyric acid; HC, hippocampus; LA, lateral amygdala; LC, locus coeruleus; NE, norepinephrine; PFC, prefrontalcortex; PRh, perirhinal cortex; PV, parvalbumin; SOM, somatostatin; VIP, vasoactive intestinal peptide.

Under basal conditions, glutamatergic projection neurons of the BLA are tonically inhibited by local GABAergic interneurons and in a majority of BLA projection neurons baseline recordings exhibit rhythmic spontaneous inhibitory post-synaptic potentials (IPSPs) (Rainnie et al. 1991a). The occurrence of rhythmic spontaneous IPSPs in BLA projection neurons is thought to facilitate the synchronized output of ensembles of these neurons in response to sensory input (Rainnie 1999). Hence, highly processed sensory information is relayed to BLA projection neurons via excitatory afferent pathways from the lateral amygdala (LA), perirhinal cortex (PRh) and the hippocampal formation (Price and Amaral 1987; Amaral et al. 1992; McDonald 1998). These excitatory inputs activate both the glutamatergic projection neurons and also the local GABAergic interneurons, most likely via a collateral excitatory pathway (see Figure 2). Thus, stimulation of primary sensory afferents to the BLA results in a direct glutamate-mediated excitation and a secondary indirect GABA-mediated inhibition of projection neurons, such that there is a rapid activation followed by inhibition of the projection neurons (Rainnie et al. 1991a,b). At the same time, the BLA receives concurrent information about the organism's ability to “cope” with the sensory stimuli from the executive centers of the prefrontal circuit. The input from the prefrontal cortex is thought to gate information flow within the BLA in such a way as to facilitate attentional bias to salient sensory input (Rosenkranz and Grace 2001, 2002) and initiate the activation of downstream structures involved in maintaining vigilance or “anxiety” states, or conversely inhibit this process when the input is not salient. At this stage, stress-induced activation of CRF in the amygdala could shift the network towards greater excitability and make it more resistant to prefrontal inhibition, thus reducing the ability to cope with the stimulus (Rogan et al. 1997; Maroun and Richter-Levin 2003). Such a mechanism is supported by the fact that studies involving stimulating the mPFC and recording from either interneurons or projection neurons of the amygdala found increased EPSPs in the interneurons and increased IPSPs in the projection neurons following mPFC stimulation. Furthermore, the IPSPs in the projection neurons were blocked by pretreatment with a GABA_A receptor antagonist while the temporal relation of the stimulation to the increase in the occurrence of IPSPs was consistent with activation of an interneuron (Rosenkranz and Grace 2002). Some of the downstream structures that could be activated by salient sensory stimuli include the BNST for arousal/anxiety, the CE for fear and autonomic responses, the hippocampus for “contextual” memory, the medial nucleus of the amygdala (ME) for reproductive and endocrine responses, and nucleus accumbens (NA) for reward. Thus, the differential effects of the BLA on these different output pathways are likely to influence the physiological and behavioral responses to specific salient sensory stimuli, and modulate the animal's behavior.

The neural network of the BLA is extremely plastic with the potential for LTP following repeated exposure to excitatory stimuli. Thus, under chronic stress, when aversive stimuli repeatedly activate afferent pathways to the BLA, long-term plastic changes could occur within the local BLA circuitry such that the tonic inhibition is reduced. Then, seemingly non-salient stimuli could activate downstream structures and elicit anxiety, fear, or neuroendocrine responses, i.e. result in pathological anxiety responses. Such a mechanism is plausible and has been supported by several of our previous studies (Sanders and Shekhar 1991; Rainnie et al. 1992a, b; Sanders and Shekhar 1995a, b; Sajdyk et al. 1999; Sajdyk and Shekhar 2000) as well as by others (Clugnet and LeDoux 1990; McKernan and Shinnick-Gallagher 1997; Mahanty and Sah 1998; Maren 1999; Huang et al. 2000; Goosens and Maren 2002; Lin et al. 2003a; b). In addition, this balance of excitatory and inhibitory tone in the BLA may be modulated by tonic serotonergic (5-HT) input (Cheng et al. 1998; Rainnie 1999; Chen et al. 2003) from the dorsal raphe nucleus (Abrams et al. 2005), and norepinephrine input (Ellis and Kesner 1983; Koob 1999; McGaugh 2002; Charney 2003; LaLumiere et al. 2003; Morilak et al. 2003) from the locus coeruleus (Fallon et al. 1978).

Plasticity in the BLA and its role in other behaviors

While the above discussion has focused on the role of plasticity within the amygdala, and the resulting enhancement of glutamate-mediated excitation and reduction of GABA-mediated inhibition, on the development of anxiety responses, similar mechanisms have been implicated in other models such as kindling and epilepsy (Post 2002; Wig et al. 2002) or memory consolidation. This is in fact supported by existing evidence, and the amygdala plasticity that results in enhanced anxiety-like behaviors, when pushed to extremes could easily result in seizures and provide models of epilepsy (Sanders et al. 1995; Adamec 2001). However, epileptogenesis generally results in spontaneous firing of the neurons, reduced number of GABAergic neurons (Callahan et al. 1991) and other structural abnormalities (Asprodini et al. 1992) not usually seen with synaptic plasticity that results in anxiety-like behaviors. Thus, while plasticity is a broad neural mechanism that underlies acquisition of a variety of new responses within the CNS, there are emerging differences in the cellular substrates of plasticity leading to the different types of behaviors.

Conclusion

Based on the above data, it is evident that the BLA has an important role in regulating anxiety and affective responses. Existing data summarized in this review support the hypotheses that stress induced plasticity within the amygdala may be a critical step in the pathophysiology of the development of chronic anxiety states, and this change in the limbic neural circuitry is involved in the transition from normal vigilance responses to pathological anxiety, which then often leads to syndromes such as panic and post-traumatic stress disorders.

Acknowledgements

The authors are grateful to the technical assistance of Ms Stephanie Fitz. This work was supported by grants RO1s MH065702 and MH52619.