Unlike hippocampal cell loss, dendritic retraction of hippocampal neurons in response to chronic stress is expressed in several species and can be reversed. Chronic stress produces dendritic retraction within the hippocampus in rats (Watanabe, Gould, Cameron, Daniels, & McEwen, 1992; Watanabe, Gould, Daniels, Cameron, & McEwen, 1992; Watanabe, Gould, & McEwen, 1992) and tree shrews (Magariños, McEwen, Flügge, & Fuchs, 1996) and returns to control levels within 10 days after the termination of chronic stress (Conrad, Magariños, LeDoux, & McEwen, 1999). For this reason, many have referred to dendritic retraction as “remodeling.” Dendritic retraction may be a compensatory response to chronic stress that goes unnoticed during uneventful periods because it recovers shortly after stress ceases. However, hippocampal neurons expressing dendritic retraction may be vulnerable to other life events that occur at the same time. For example, rats that were chronically stressed by restraint to produce dendritic retraction within the hippocampal CA3 region show exacerbated hippocampal cell death following a neurotoxin challenge compared to non-stressed controls (Conrad, Jackson, & Wise, 2004). This may explain why the hippocampus is vulnerable to damage when chronic stress precedes or coincides with other conditions, such as AIDS (Oberfield et al., 1994), obesity (Raber, 1998), depression (Sheline, Wang, Gado, Csernansky, & Vannier, 1996), and Alzheimer’s disease (de Leon et al., 1993). These studies suggest that chronic stress and GCs contribute to but are not the sole determinant for hippocampal cell death in rodents and humans (see Sapolsky, 2000b). In human conditions that hypersecrete GCs, such as depression (Sheline et al., 1996), Cushing’s disease (Starkman, Gebarski, Berent, & Schteingart, 1992), and in a subset of aged individuals (Lupien et al., 1998), hippocampal volumes are reduced (Sheline et al., 1996; Starkman et al., 1992). Treatment with antidepressants can protect against hippocampal volume loss in depression (Sheline, Gado, & Kraemer, 2003), and attenuating GC elevations can help increase hippocampal volumes in Cushing’s disease (Starkman et al., 1999). Thus, chronic stress-induced hippocampal dendritic retraction in rodents is a useful model for understanding the dynamics between chronic stress/GC exposure and the hippocampus in humans.

The purpose of this review is to discuss the relevance of studying structural changes to the dendritic arbors of neurons, delineate the functional significance of chronic stress-induced changes in hippocampal neurons, and propose a mechanism by which chronic stress influences hippocampal function.

Chronically activating the stress response can produce maladaptive changes, which have been postulated to contribute to disease (for review, see de Kloet, Joels, & Holsboer, 2005; McEwen & Wingfield, 2003; Smith, 1996). A transition into maladaptive changes includes dendritic remodeling, which can be produced under a variety of chronic stress conditions: 6 hours of daily restraint in wire mesh for 21 days (Watanabe, Gould, Cameron, et al., 1992; Watanabe, Gould, Daniels, et al., 1992; Watanabe, Gould, & McEwen, 1992), daily predator stress combined with a high-fat diet for 21 days (Baran et al., 2005), social defeat stress every other day for 21 days (Kole, Costoli, Koolhaas, & Fuchs, 2004), continual social stress and competition for resources over 14 days (McKittrick et al., 2000), 2 hours in bag restraints daily for 10 days (Vyas, Mitra, Rao, & Chattarji, 2002), and daily activity stress combined with food restriction over 6 days (Lambert et al., 1998). Although chronic stress predominately reduces dendritic arbors in CA3 neurons, dendritic retraction has been observed in other brain regions. When chronic unpredictable stress continues for 4 weeks, CA1 and dentate gyrus neurons express dendritic retraction (Sousa, Lukoyanov, Madeira, Almeida, & Paula-Barbosa, 2000). In addition, CA3 and CA1 dendritic retraction occurs in as little as 6 days following activity stress combined with food restriction (Lambert et al., 1998). Prefrontal cortical neurons also express dendritic retraction following 3 hours of restraint for 1 to 3 weeks (S. M. Brown, Henning, & Wellman, 2005; Radley et al., 2004). Chronic stress-induced CA3 dendritic remodeling has been proposed to be a maladaptive response because it is associated with susceptibility to damage and cognitive dysfunction (discussed later). However, another interpretation is that CA3 dendritic remodeling may be a compensatory response to protect against extended excitatory amino acid stimulation, which can compromise and kill neurons.

The mechanism underlying stress-induced CA3 dendritic retraction is illustrated in Figure 1 and is discussed in other reviews (McEwen, Magarinos, & Reagan, 2002; Smith, 1996). GCs acting through the GR mediate stress-induced CA3 dendritic retraction. Chronic administration of stress levels of GCs produces CA3 dendritic retraction (Magariños, Orchinik, & McEwen, 1998; Woolley, Gould, & McEwen, 1990), and GC attenuation with cyanoketone blocks CA3 dendritic retraction (Magariños & McEwen, 1995b). These studies indicate that corticosterone, rather than other hormones of the HPA axis or enhanced activity of the sympathetic nervous system, mediates CA3 dendritic retraction.

Excitatory amino acid release influences stress-induced CA3 dendritic retraction. Phenytoin (Dilantin), an antiepileptic that interferes with excitatory amino acid release and action (Crowder & Bradford, 1987; Griffith & Taylor, 1988; Skerritt & Johnston, 1983), prevents stress-induced CA3 dendritic retraction (Magariños et al., 1996; Watanabe, Gould, Cameron, et al., 1992). These data are consistent with stress-enhancing glutamate release (Gilad, Gilad, Wyatt, & Tizabi, 1990; Lowy, Gault, & Yamamoto, 1993) and up-regulating glutamate transporter expression within glia of the CA3 region (Reagan et al., 2004).

Selective changes in regions showing dendritic remodeling may help identify glutaminergic afferents contributing to CA3 dendritic plasticity following chronic stress because glutaminergic afferents are topographically organized. The mossy fibers from the dentate gyrus granule cells form synapses on excrescences, postsynaptic structures located proximal to the CA3 soma (T. H. Brown & Zador, 1990; Martinez & Barea-Rodriguez, 1997). Afferents located distally from the CA3 soma form synapses on “spines” and include the commissural fibers from the contralateral hippocampus, recurrent axon collaterals from CA3 neurons, and the perforant path from the entorhinal cortex (T. H. Brown & Zador, 1990; Martinez & Barea-Rodriguez, 1997; Witter & Amaral, 2004). Chronic stress produces drastic remodeling within the middle part of the apical CA3 dendritic tree (Kole et al., 2004), which corresponds to the region expressing chronic stress-induced changes in the N-methyl-D-aspartate (NMDA) glutaminergic receptor sensitivity (Kole, Swan, & Fuchs, 2002). Consequently, these two studies implicate the commissural-associational collaterals as contributing to CA3 dendritic retraction. Another study showed that entorhinal cortical lesions, which project to the distal portions of the apical CA3 region, protect against CA3 dendritic retraction following chronic stress (Sunanda, Meti, & Raju, 1997). Moreover, dendritic retraction typically occurs on the CA3 apical region, not the CA3 basal tree (Conrad et al., 1999; Magariños et al., 1996; Magariños & McEwen, 1995a; McKittrick et al., 2000), and the former region receives the majority of mossy fiber inputs from the dentate gyrus, indicating that selective damage to the dentate may also prevent stress-induced CA3 dendritic remodeling. Dendritic retraction has been observed in both the apical and basal regions, but this robust dendritic retraction typically occurs under extreme conditions that combine chronic restraint with food restriction (Kleen, Sitomer, Killeen, & Conrad, in press) or ovariectomy in females (McLaughlin, Baran, Wright, & Conrad, 2005). Finally, chronic stress-induced decreases in CA3 dendritic complexity occur in conjunction with increased density of both excrescences and spines (Sunanda, Rao, & Raju, 1995), implicating compensatory mechanisms of both mossy fiber and nonmossy fiber afferents. Altogether, these studies indicate that chronic stress interacts synergistically with a variety of glutaminergic afferents to remodel CA3 dendritic morphology.

Serotonin may be implicated in CA3 dendritic retraction. Chronic stress increases brain serotonergic levels (Chaouloff, 1993) and down-regulates the transporters that remove serotonin from the synapse, a process that can enhance extracellular serotonin (McKittrick et al., 2000). Reducing extracellular serotonin by enhancing its uptake with tianeptine prevents stress-induced CA3 dendritic retraction (see Figure 2; Conrad et al., 1999; Magariños, Deslandes, & McEwen, 1999; Watanabe, Gould, Daniels, et al., 1992). Antidepressants that enhance serotonin, such as fluoxetine, fail to alter stress-induced CA3 dendritic retraction (Magariños et al., 1999). However, recent studies suggest that tianeptine may prevent CA3 dendritic retraction through actions on the glutaminergic system (Fuchs, Czeh, Kole, Michaelis, & Lucassen, 2004; McEwen & Chattarji, 2004), which includes altering NMDA receptor sensitivity (Kole et al., 2002) and changed expression of glutamate transporter within glial cells (Reagan et al., 2004). These studies show that enhanced serotonergic tone, or perhaps actions on the glutaminergic system, contributes to chronic stress-induced CA3 dendritic retraction.

Enhancing the inhibitory tone may also attenuate stress-induced CA3 dendritic retraction. The majority of interneurons within the hippocampus are inhibitory (Feldblum, Erlander, & Tobin, 1993; Woodson, Nitecka, & Ben-Ari, 1989), and activating gamma-aminobutyric acid (GABA)_A receptors with adinazolam, a benzodiazepine agonist, block CA3 dendritic retraction caused by chronic stress (Magariños et al., 1999). Paradoxically, chronic stress enhances GABAergic systems (Bowers, Cullinan, & Herman, 1998; Qin, Karst, & Joels, 2004), but these effects may be region specific. For example, GABA can be altered within the interneurons of the dentate gyrus without changing GABA in other hippocampal regions (Herman, Renda, & Bodie, 2003). This latter finding supports the interpretation that altered inhibitory tone is not uniform and that region-specific effects may disrupt overall inhibitory tone.

Neurotrophic factors, such as brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), neurotrophin (NT)–3, and NT-4, can also contribute to hippocampal dendritic plasticity. Postsynaptic neurons release neurotrophic factors, which are retrogradely transported to the nucleus of the presynaptic neuron to influence gene expression (Smith, 1996). The hippocampus expresses the highest levels of BDNF mRNA in the brain (Hofer, Pagliusi, Hohn, Leibrock, & Barde, 1990), and BDNF promotes neuronal connectivity (Klintsova & Greenough, 1999; Lewin & Barde, 1996), especially dendritic differentiation and complexity in nonpyramidal interneurons of the hippocampus (Marty et al., 1996; Vicario-Abejon, Collin, McKay, & Segal, 1998), neocortex (Jin, Hu, Mathers, & Agmon, 2003), striatum (Ventimiglia, Mather, Jones, & Lindsay, 1995), and cerebellum (Mertz, Koscheck, & Schilling, 2000). Indeed, GABAergic influence is facilitated by BDNF and prevented by BDNF inhibitors (Jin et al., 2003). Although chronic stress also enhances GABAergic systems (Bowers et al., 1998), stress or isolation down-regulates BDNF (Scaccianoce et al., 2006) and up-regulates NGF and NT-3 (Smith, Makino, Kvetnansky, & Post, 1995). The effects of stress on NGF and NT-3 are proposed to be compensatory mechanisms (Smith, 1996). Overall, many of the changes produced by BDNF are opposite those produced by glutamate and chronic stress (Smith, 1996).

To assess the functional consequences of chronic stress-induced CA3 dendritic retraction, we selected a Y-maze task as described by Dellu, Mayo, Cherkaoui, Moal, and Simon (1992). The Y-maze consists of three equilaterally intersecting Plexiglas arms with spatial cues positioned outside the maze. Rats are placed in the Y-maze with one of the arms blocked to prevent access, allowing the rats to explore the remaining two arms. After the rats have explored the two arms for 15 minutes, they are removed from the maze and returned to their home cages. During this intertrial interval, the maze is rotated and the bedding on the floor of the maze is mixed to reduce the use of olfactory cues within the maze, which can facilitate navigation without using the hippocampus. After a delay of 4 hours, rats are reintroduced to the spatial location where they were placed in the first trial and allowed to explore all three arms for 5 minutes. Rats with functional spatial memory will enter the arm located in the unexplored spatial location more, relative to the arms in locations that they had already visited in the initial trial, whereas rats with impaired spatial memory will enter the previously explored and unexplored arms similarly (Conrad, Galea, Kuroda, & McEwen, 1996; see Figure 6).

Although the Y-maze has its advantages and disadvantages, we believe that the advantages outweigh the disadvantages when studying the effects of chronic stress on spatial memory. First, spatial memory must be determined relatively quickly because CA3 dendritic arborization is dynamic, and the dendritic arbors can return to their prestress condition within 10 days after the termination of the chronic stressor (Conrad et al., 1999). In this regard, the Y-maze task can be used to train the rat in one trial, and spatial recognition memory can be determined one trial later, usually within several hours. Second, motivating rats to perform on the Y-maze does not require food restriction, which can be an added confound when investigating the effects of chronic stress on performance. Chronic stress decreases sucrose preference (Rygula et al., 2005) and bar-pressing rate for food at higher fixed ratio schedules (Kleen et al., in press). Moreover, food deprivation can increase circulating corticosterone levels (Broocks, Schweiger, & Pirke, 1990; Stewart, Meaney, Aitken, Jensen, & Kalanr, 1988). Third, the Y-maze is as not overtly stressful as other testing paradigms that involve foot shock (i.e., fear conditioning) or immersion in water (Morris water maze).

Potential disadvantages of the Y-maze are that chronic stress decreases motivation, which may influence the stressed rats’ motivation to explore. However, we have found that chronic stress decreases food seeking, whereas Y-maze exploration is unaffected (Kleen et al., in press). Thus, motivation to seek food and sucrose may involve separate neurobiological constructs than those guiding Y-maze exploration. Nonetheless, care is taken to validate that motivation is similar among treatment groups. In the Y-maze, motivation to explore can be determined by comparing the total number of entries into all arms across groups. For example, nonstressed controls may make nine entries overall but show a preference for the novel arm by entering the novel arm six times and the remaining arms three times. In contrast, chronically stressed rats may make nine entries overall but fail to show an arm preference by entering each arm three times. In this example, motivation is suggested to be similar because rats made a similar number of entries even though arm preference differed. Related to this behavior, chronic stress enhances anxiety and produces neophobia, which may interfere with the goal of seeking out the novel arm in the Y-maze. However, rats must recognize the novel arm to avoid it. Nonetheless, some studies have observed that an animal exhibiting fear responses may show an increased alternation rate because it tries to avoid the arm from which it was previously picked up (Gerlai, 2001). Under these conditions, chronically stressed rats may show performance at chance levels to avoid the experimenter. This concern is addressed in several ways. First, animals are handled daily in novel and familiar locations to expose them to novelty. Second, an investigator handling the rats is different from the investigator restraining the rats to reduce potential negative associations between the investigator and restraint process. Third, we have recently demonstrated that chronically stressed rats will perform well in the Y-maze by entering the novel arm more than the previously explored arm when the task demands are made easier by reducing the intertrial interval from 4 hours to 1 minute (Bellani, Luecken, & Conrad, 2006; Kleen et al., in press) and placing salient cues within the maze (Wright & Conrad, 2005), which is similar in principle to other paradigms that include intramaze cues (Ramos, 2000). Rats that attempt to avoid the novel arm or the investigator should exhibit poor performance regardless of the demands of the Y-maze. However, our findings demonstrate that chronically stressed rats will seek out the novel arm when the demands of the task reduce the reliance on the hippocampus. The Y-maze is probably successful for determining spatial memory in rats because the relatively narrow walls of the Y-maze (19 cm wide) enable rats to exhibit thigmotaxis, the preference to remain near walls and sides of a room. Thus, rats can explore and seek out novelty without interference from competing demands, such as avoiding open spaces that would trigger anxiogenic responses.

Chronic corticosterone treatment produces anomalous findings in whether spatial memory is impaired. As described earlier, chronic stress-induced CA3 dendritic retraction is mediated by GCs, such as corticosterone: CA3 dendritic retraction is created by exposure to stress levels of corticosterone for 3 weeks (Magariños & McEwen, 1995b; Watanabe, Gould, Cameron, et al., 1992; Woolley et al., 1990) and prevented by blocking corticosterone secretion with cyanoketone (Magariños & McEwen, 1995b). Thus, if hippocampal dendritic plasticity mediated hippocampal function, then GC-mediated CA3 dendritic retraction should produce spatial learning and memory deficits. However, the literature is mixed regarding chronic GC actions on spatial learning and memory; some studies show deficits (Arbel et al., 1994; Bardgett et al., 1994; Dachir et al., 1993; Endo, Nishimura, & Kimura, 1996; McLay, Freeman, & Zadina, 1998; Ramos-Remus, González-Castañeda, González-Perez, Luquin, & García-Estrada, 2002), others show improvements (Bardgett, Newcomer, & Taylor, 1996; Hebda-Bauer, Morano, & Therrien, 1999), and still others report no significant effects (Bardgett et al., 1996; Bodnoff et al., 1995; Clark et al., 1995; Hebda-Bauer et al., 1999; Krugers et al., 1997; Luine, Spencer, & McEwen, 1993). Some discrepancies may be resolved by the differential treatment protocols. Corticosterone administration beyond a month may compromise neurons and eventually lead to cell death (see Arbel et al., 1994). Other paradigms continue corticosterone treatment during behavioral assessment, which allow GCs to modulate memory through activational effects that are unrelated to dendritic retraction (see Conrad, 2005). Moreover, corticosterone dose and serum levels can greatly vary among studies. Nonetheless, some protocols replicate the conditions in which corticosterone treatment produces CA3 dendritic retraction (Watanabe, Gould, Cameron, et al., 1992; 40 mg/kg daily injections for 3 weeks) without altering spatial memory (Coburn-Litvak, Pothakos, Tata, McCloskey, & Anderson, 2003). Thus, corticosterone-induced CA3 dendritic retraction shows evidence of a disconnection from spatial memory.

In a second example, classical fear conditioning was used to determine whether chronic stress selectively impaired hippocampus-dependent memory compared to hippocampus-independent memory. Lesions to the hippocampus or its afferents attenuate contextual fear conditioning without disrupting cued fear conditioning (Bechara et al., 1995; Kim & Fanselow, 1992; Phillips & LeDoux, 1992, 1995). Thus, freezing to the environment is hippocampus dependent whereas freezing to tone is hippocampus independent. Hence, rats were chronically stressed by restraint (6 hours per day for 21 days) and then submitted to fear conditioning (Conrad et al., 1999). Mild foot shock was paired with a tone, and memory for this event was measured in the environment that was paired with the foot shock (hippocampus dependent) and with the tone (hippocampus independent). We found that chronic stress facilitates fear conditioning to both context and tone (Conrad et al., 1999). In addition, performance appears unrelated to hippocampal CA3 dendritic retraction because tianeptine treatment to prevent CA3 dendritic retraction fails to alter the facilitated fear conditioning of chronically stressed rats (Conrad et al., 1999). Compared to the nonstressed controls, the CA3 dendritic arbors in chronically stressed rats were confirmed to be less complex, an effect that was prevented by tianeptine (Conrad et al., 1999). This last finding was critical because it provided the first demonstration of a disconnection between CA3 dendritic retraction and hippocampal function within the same animals.

The role of the hippocampus in regulating the HPA axis has been demonstrated in a variety of ways. Stimulating the hippocampus generally inhibits the HPA axis (Dunn & Orr, 1984; Mandel & Walter, 1963), and destroying part or all of the hippocampus or its efferents enhances HPA axis activity (Daniels, Jaffer, Engelbrecht, Russell, & Taljaard, 1990; Feldman & Conforti, 1980; Fischette, Komisaruk, Edinger, Feder, & Siegel, 1980; Herman et al., 1989; Sapolsky, Krey, & McEwen, 1984a; Wilson, Greer, & Roberts, 1980). These data indicate that the hippocampus participates in the regulation of the HPA axis through an inhibitory role (for review, see Herman, Figueiredo, et al., 2003).

Corticosteroids have long been known to influence cognition, including spatial memory. The limbic structures, especially the hippocampus, express the highest levels of receptors for GCs in both the rodent brain (McEwen, Weiss, & Schwartz, 1968, 1969) and primate brain when both receptor subtypes are included (Mar Sanchez et al., 2000), causing the hippocampus to be highly sensitive to GCs. Indeed, stress and GCs greatly influence the learning and memory processes that are mediated by the hippocampus (Bremner & Narayan, 1998; Conrad, 2005; Kim & Diamond, 2002; Luine, 1997). GCs influence spatial memory in a biphasic fashion, with low to moderate GC levels positively correlating with spatial memory and moderate to high GC levels negatively correlating with spatial memory (Conrad, 2005; Park, Campbell, Smith, Fleshner, & Diamond, in press; Yau, Olsson, Morris, Meaney, & Seckl, 1995). Moreover, GC actions can differ depending on the stage of memory processing (acquisition, consolidation, retrieval; Conrad, 2005; Conrad, Lupien, Thanasoulis, & McEwen, 1997; Oitzl & de Kloet, 1992) and the receptor subtype activated (Lupien & McEwen, 1997; Oitzl & de Kloet, 1992). Thus, the HPA axis clearly influences spatial memory in a complex, nonlinear fashion (see Figure 7A).

Chronic stress can produce numerous changes throughout the HPA axis, especially within the hippocampus, to disrupt HPA axis activity. The potential for chronic stress and GC hypersecretion to disrupt the HPA axis has been extensively studied and reviewed (McEwen, 2001; Sapolsky, 1992, 1996, 2000a). Briefly, chronic stress and GCs down-regulate mRNA and protein for GC receptors within the hippocampus (Meyer, van Kampen, Isovich, Flügge, & Fuchs, 2001; Nishimura, Makino, Tanaka, Kaneda, & Hashimoto, 2004; Sapolsky, Krey, & McEwen, 1984b) and for corticotropin-releasing factor within the hypothalamus (Albeck, Hastings, & McEwen, 1994; Nishimura et al., 2004; Szot et al., 2004). These imbalances in GR and MR expression are proposed to alter the hippocampus regulation of the HPA axis (Dallman, 1993; de Kloet, 1991). Moreover, chronic stress reduces GABAergic input to hypothalamic neurons that participate in regulating the HPA axis (Verkuyl, Hemby, & Joels, 2004). Indeed, chronically stressed rats show an enhanced HPA axis response to novel stressors (Bhatnagar & Dallman, 1998; Harris et al., 2004) and increased sensitivity to an α₂ adrenergic receptor antagonist (Park et al., 2001), suggesting a sensitization of the sympathetic nervous system. These findings clearly demonstrate that chronic stress alters the HPA axis and the response to novel stressors.

Three of the four criteria have been satisfied by demonstrating that (a) the hippocampus regulates HPA axis activity, (b) altered HPA axis activity influences spatial memory, and (c) chronic stress disrupts HPA axis activity. For the last criterion, we recently reported that manipulating the HPA axis can restore spatial memory in chronically stressed rats (see model, Figure 7B). We followed a paradigm that reliably produces hippocampal CA3 dendritic retraction (Conrad et al., 1999; Magariños & McEwen, 1995a; Watanabe, Gould, & McEwen, 1992), whereby rats were chronically stressed by restraint for 6 hours/day for 21 days and then spatial memory was assessed using the Y-maze 1 day after chronic restraint ended (Wright, Lightner, Bellani, Harman, & Conrad, 2004; Wright et al., 2005; Wright, Lightner, Harman, Meijer, & Conrad, in press). On the day of Y-maze exposure, a subset of rats received a single injection of metyrapone to prevent stress levels of GCs from being released. We found that the highest dose of metyrapone (75 mg/kg) restored spatial memory ability in chronically stressed rats (see Figure 8A). Moreover, chronic stress increased the corticosterone response to the Y-maze (see Figure 8B) and down-regulated GR mRNA within the hippocampus, demonstrating that the HPA axis was dysregulated. A potential concern is that metyrapone could have influenced each stage of memory processing (acquisition, consolidation, retrieval), as well as motivational/motor abilities because metyrapone was given prior to training. However, total entries made during Y-maze exploration were similar among conditions, indicating that motivational differences were unlikely to have caused the improvement in Y-maze performance. Future studies will be performed to determine the stages of memory processing that are influenced by metyrapone treatment. Retrieval is a putative target because injecting metyrapone prior to retrieval restores spatial memory in rats with lesions to the CA3 region (Roozendaal, Phillips, et al., 2001). These studies show that altering the HPA axis can correct memory deficits in individuals with a hippocampus compromised by remodeled dendrites or even neuron loss.

Many reviews have reported that the amygdala is involved in highly aversive conditions and is considered part of the repertoire of emotional responses (Cahill, 2000; Cahill & McGaugh, 1996; LeDoux, 2000). Amygdalar lesion or inactivation prevents fear conditioning to both cue and context (Conrad, MacMillan, et al., 2004; Phillips & LeDoux, 1992). In contrast, hippocampal lesions impair contextual conditioning, but spare cue conditioning (Phillips & LeDoux, 1992). Thus, the amygdala is necessary for fear conditioning in general, and the hippocampus plays a more limited role in contextual conditioning.

Several studies show that the contribution of the amygdala can have a profound effect on hippocampus-dependent performance. When rats learn a stressful version of a water maze task, the amygdala is activated; rats that fail to learn the stressful version also fail to show activation of the amygdala (Akirav, Sandi, & Richter-Levin, 2001). In another study, amygdalar stimulation facilitates hippocampal long-term potentiation, which is a model of brain plasticity (Akirav & Richter-Levin, 1999). These studies show that hippocampus-dependent learning can be greatly facilitated by the contribution of the amygdala.

The HPA axis is instrumental in the modulation of hippocampus-dependent memory by the amygdala. The amygdala has a rich supply of GC receptors (Reul & de Kloet, 1986; Sarrieau et al., 1985; Warembourg, 1975), indicating that GCs can directly influence the amygdala. Intra-amygdala infusions of GR agonists enhance passive avoidance, a hippocampus-dependent task that uses foot shock (Roozendaal & McGaugh, 1997; Roozendaal, Quirarte, & McGaugh, 2002). Moreover, disruption of the amygdala blocks memory facilitation from GR agonists (Roozendaal, de Quervain, Ferry, Setlow, & McGaugh, 2001; Roozendaal & McGaugh, 1996).

Acute stress is thought to help maintain homeostasis, defined as the stability of physiological systems that sustain life (McEwen & Wingfield, 2003). For example, acute stress helps regulate blood glucose levels and body temperature through tightly controlled mechanisms. Allostasis is defined as achieving stability through change (McEwen & Wingfield, 2003). Thus, dendritic remodeling following chronic stress or GCs may be one mechanism that allows the hippocampus to maintain stability through change. Allostatic load refers to the toll on the body as it tries to reinstate homeostasis, and this toll may make individuals vulnerable to disease. For example, CA3 dendritic retraction may make individuals susceptible to spatial memory deficits and even cell damage following exposure to neurotoxins. These latter events may have life-threatening repercussions if memory is critical for locating food resources (in the former example) or if a metabolic challenge coincides with chronic stress. Other individuals may experience stress-induced CA3 dendritic retraction uneventfully, and they will recover from chronic stress without any significant long-term ramifications of cognitive decline or cell loss.

Genetic predisposition and life events will greatly affect how individuals respond to chronic stress (for review, see de Kloet et al., 2005; see Figure 10). Individuals with genetic anomalies or exposure to unique life events may be more susceptible to the debilitating effects of chronic stress. Using a hypothetical genetic example, the underexpression of BDNF may make one more susceptible to stress-induced CA3 dendritic retraction than if BDNF levels were higher because neurotrophins appear to protect against neuronal damage. In terms of behavioral phenotype, rats that are characterized as high anxiety, based on a commonly used behavioral paradigm that determines emotionality in rodents, show spatial memory deficits following chronic stress compared to low-anxiety rats (Bellani et al., 2006). Regarding life choices, chronic stress combined with a high-fat diet produces robust CA3 dendritic retraction compared to either condition alone (Baran et al., 2005). Hence, highly anxious individuals who prefer high-fat, fast food may be prone to the effects of chronic stress compared to less anxious individuals who eat balanced meals. All of these studies demonstrate that genetics and life events influence susceptibility to chronic stress.