Psychobiological Mechanisms of Resilience and Vulnerability: Implications for Successful Adaptation to Extreme Stress

Testosterone has been among the most studied of all hormones in terms of its relationships to specific behaviors. Aggression is the aspect of human behavior most often linked to testosterone concentrations (105). Preclinical studies consistently show that low levels of testosterone are associated with submissive behavior. In mandrils and squirrel monkeys, social rank correlates with testosterone levels (105, 106). In human subjects, the personal experience of success, as well as the feeling of dominance in a competitive situation, is associated with higher testosterone levels (107, 108). Increased levels of testosterone have been found in male prison inmates with frequent episodes of violent behavior (109–111). Psychological stress is associated with decreases in testosterone levels. For example, elite soldiers participating in a physically and psychologically stressful training exercise show a lowering of their testosterone levels (63).

The mechanism by which testosterone is reduced by physical and psychological stress remains to be elucidated. It is unclear whether the decrease in testosterone from exposure to mental stress is caused by decreased leuteinizing hormone-releasing hormone (LH-RH) synthesis at the hypothalamus or leuteinizing hormone (LH) secretion in the pituitary (105). Perhaps a more likely mechanism involves a recently identified hypothalamic-testicular pathway that is independent of the pituitary but travels through the spinal cord. This pathway appears to mediate the effect of CRH to decrease testosterone levels. Thus, hypothalamic increases in CRH produced by psychological stress may be associated with decreased testosterone by stimulating the neural pathway that interferes with Leydig cell function independently of the pituitary. It is important to establish the relative role of the LH-RH/LH axis and the hypothalamic testicular axis in modulating the influence of specific stressors on testosterone release (112).

There is a recent report of reduced CSF testosterone levels in PTSD patients that was negatively correlated with CSF CRH concentrations (113). There was no correlation between plasma and CSF testosterone levels (113). The data from studies measuring plasma testosterone levels in PTSD patients are mixed (114).

Depressed men have been found to have decreased serum or plasma testosterone in some studies (115), but not all, because of confounding factors. Hypogonadal men often experience depressive symptoms, which are improved by testosterone-replacement therapy (116). Clinical trials of depressed men with decreased testosterone have produced contradictory results. However, a recent placebo-controlled study (115) found testosterone gel to be effective for men with treatment-resistant depression and low testosterone levels when added to an existing antidepressant regimen. Testosterone administration may be helpful for patients with low testosterone secondary to chronic severe psychological stress.

There is abundant preclinical and clinical literature demonstrating consistent gender differences in stress responsiveness (117). Most of the work focused on HPA responses to stressors. Female rats consistently show greater increases in corticosterone and ACTH in response to acute and chronic stressors. These differences have generally been attributed to the activational effects of gonadal steroids on elements of the HPA axis in females (118). Several studies suggest that estradiol plays a role in enhanced stress responses in female rats, based upon increased HPA axis responses to stress when ovariectomized rats are treated with estradiol (119). A possible mechanism for these findings is that estrogen (as well as progesterone) produces a relative resistance to glucocorticoid feedback (120).

However, a recent investigation by Young and colleagues, studying the effects of estrogen antagonists and physiological doses of estradiol, found that estradiol reduced the ACTH response to restraint stress in female rats (118). The estrogen antagonists had the opposite effect. These data suggest that physiological doses of estradiol are inhibitory to stress responsiveness and that blocking estradiol on gonadally intact, normally cycling female rats leads to exaggerated stress responsiveness. The contrast with prior studies seems to relate to the dosage of estradiol and the duration of administration. Considered together, the studies indicate that short-term exposure to low doses of estrogen can suppress HPA axis responses to stress but higher doses and more prolonged treatment enhances HPA axis responses (117, 118). The mechanism underlying these effects could be due to enhanced negative feedback or decreases in the stimulatory aspects of the system, related to either CRH or ACTH. This remains to be elucidated, since studies examining the effects of estradiol on mineral corticoid receptor and glucocorticoid receptor binding and mRNA expression and on CRH have not yielded consistent results, perhaps due to variability in doses and duration of treatment regimens.

Studies in human populations suggest that female subjects respond with greater HPA activation to stressors involving interpersonal concerns (social rejection) and male subjects to achievement-oriented stressors (117). The role of estrogen in these differential responses remains to be studied. Estrogen has been shown to blunt HPA axis responses to psychological stress in postmenopausal women (121, 122) and to blunt the ACTH response to CRH in postmenopausal women with high levels of body fat. In addition, 8 weeks of estrogen supplementation to perimenopausal women blunted systolic and diastolic blood pressure, cortisol, ACTH, plasma epinephrine and norepinephrine, and norepinephrine responses of the entire body to stress (120).

Although the mechanisms responsible for the effect of estrogen on glucocorticoid levels are not fully defined, it appears that it acts by means of ACTH and thus the pituitary or hypothalamus rather than directly on the adrenal gland. This is consistent with evidence obtained from women with hypothalamic amenorrhea, in whom a blunted response to CRH administration and increased cortisol levels were observed (123). These effects could be explained by a direct action of estrogen on CRH gene expression or glucocorticoid receptor numbers or function (124).

The mechanisms by which estrogens affect catecholamine levels are also uncertain. The effects of estrogen may be due to actions on the adrenal gland or central or peripheral neuronal pathways. Neuronal pathways seem more likely (125), although several different mechanisms may be involved, including effects on α₁-noradrenergic (126) and β-noradrenergic (127) receptors and modulation of norepinephrine release. Estrogen has also been shown to up-regulate the GABA_A benzodiazepine receptor (128).

The effects of estrogen on mood and anxiety may be mediated in part by the serotonin system (129). Estrogen has complex effects on functioning of the serotonin system, including increased tryptophan hydroxylase gene and protein expression (130), decreased expression of the serotonin transporter (131), and increased 5-HT_2A binding (132). Perhaps most important are studies relevant to the 5-HT_1A receptor. Estrogen in both rats and monkeys decreases 5-HT_1A in RNA and 5-HT_1A binding in both presynaptic (dorsal raphe) and postsynaptic sites (133). Estrogen also decreases the inhibitory G proteins involved in intracellular signal transduction mediated by the 5-HT_1A receptor (134, 135).

Women appear to be more sensitive to the effects of traumatic stress. One survey found that 31% of women and 19% of men develop PTSD when exposed to major trauma (136). However, the role of estrogen in the development of PTSD has not been investigated. Based upon these data, short-term increases in estrogen after exposure to stress might be beneficial because of its ability to blunt the HPA axis and noradrenergic response to stress. However long-term stress-related elevation in estrogen might be detrimental because of estrogen-induced decreases in 5-HT_1A receptor numbers and function.

The last section identified 11 possible mediators of the psychobiological response to extreme stress and how each may contribute, alone or through functional interactions, to resilience or vulnerability (Table 1 and Figure 1). In the beginning of this article, the concept of allostatic load was introduced as a measure of the cumulative physiological burden borne by the body from attempts to adapt to stressors and strains of life’s demands (137). McEwen and Stellar (3) hypothesized that the cumulative impact on health risk from modest dysregulations in multiple systems can be substantial, even if they individually have minimal and insignificant health effects. Thus, they defined allostatic load as a cumulative measure of physiological dysregulation over multiple systems (3).

The concept of allostatic load has proven to be useful as a predictor of functional decline in elderly men and women. Seeman and colleagues (138) developed a measure of allostatic load based on 10 markers reflecting levels of physiological activity across a range of important regulatory systems, which individually have been linked to disease based upon data from a longitudinal community-based study of successful aging (138). The markers were the following:

For each of the 10 markers, the subjects were classified into quartiles based upon the distribution of scores in the baseline cohort. Allostatic load was measured by summing the number of parameters for which the subject fell into the highest-risk quartile (top quartile for all markers except HDL cholesterol and DHEA-S, for which the lowest quartile corresponds to the highest risk). In two follow-up studies encompassing 2.5 and 7 years, none of the 10 markers of allostatic load exhibited significant associations on their own with health outcomes. However, the summaried measure of allostatic load was found to be significantly associated with four major health outcomes: 1) new cardiovascular events, 2) a decline in cognitive functioning, 3) a decline in physical functioning, and 4) mortality. Thus, these data are consistent with the hypothesis that although modest abnormalities in a single physiological system may not be predictive of poor health outcome, the cumulative effect of multiple abnormalities in the physiological system is prognostic of poor physical health (11, 138).

The allostatic load concept has not been used to investigate neurobiological risk factors related to psychopathology. Perhaps an analogous approach that involves the identification of a group of biological markers that will relate to psychobiological allostasis and psychobiological allostatic load and, consequently, to resilience and vulnerability to the effects of extreme psychobiological stress will be fruitful. It is in this context that this review of the neurochemical response patterns to stress can provide a framework for developing a measure for psychobiological allostatic load. The finding that many of these measures have important functional interactions is supportive of the concept of developing a more integrative measure. One prediction is that individuals in the highest quartile for measures of HPA axis, CRH, locus coeruleus-norepinephrine, dopamine, and estrogen activity and the lowest quartile for DHEA, neuropeptide Y, galanin, testosterone, and 5-HT_1A receptor and benzodiazepine receptor function will have the highest index for psychobiological allostatic load and an increased risk for psychopathology after exposure to stress. It is possible that psychobiological allostatic load will relate to vulnerability to the effects of chronic, mild, intermittent stressors as well as extreme psychological trauma. In contrast, a resilient profile will be characterized by individuals in the highest quartile for measures of DHEA, neuropeptide Y, galanin, testosterone, and 5-HT_1A receptor and benzodiazepine receptor function and the lowest quartile for HPA axis, CRH, and locus coeruleus-norepinephrine activity (Table 1). The mediators of the stress response identified in this review are not meant to be an exhaustive or definitive list. For example, glutamate and neurotrophic factors, such as brain-derived neurotrophic factor, and neuropeptides, such as substance P and cholecystokinin, could have been included. Longitudinal community-based surveys of successful adaptation to extreme stress should be considered to determine if markers such as these or others can be used to develop a measure of psychobiological allostatic load that will be of predictive value.

Fear conditioning in many patients with PTSD and major depression causes vivid recall of memories of traumatic events, autonomic hyperarousal, and even flashbacks elicited by sensory and cognitive stimuli associated with prior traumas. Consequently, patients may begin to avoid these stimuli in their everyday lives, or a numbing of general emotional responsiveness may ensue. Resilience to the effects of severe stress may be characterized by the capacity to avoid overgeneralizing specific conditioned stimuli to a larger context, reversible storage of emotional memories, and facilitated extinction.

Classical fear conditioning is a form of associative learning in which subjects come to express fear responses to a neutral conditioned stimulus that is paired with an aversive unconditioned stimulus. The conditioned stimulus, as a consequence of this pairing, acquires the ability to elicit a spectrum of behavioral, autonomic, and endocrine responses that normally would only occur in the context of danger (156). Fear conditioning can be adaptive and enable efficient behavior in dangerous situations. The individual who can accurately predict threat can engage in appropriate behaviors in the face of danger. In the clinical situation, specific environmental features (conditioned stimuli) may be linked to the traumatic event (unconditioned stimuli), such that reexposure to a similar environment produces a recurrence of the symptoms of anxiety and fear. Patients often generalize these cues and experience a continuous perception of threat to the point that they become conditioned to context.

Cue-specific conditioned stimuli are transmitted to the thalamus by external and visceral pathways. Afferents then reach the lateral amygdala by means of two parallel circuits: a rapid subcortical path directly from the dorsal (sensory) thalamus and a slower regulatory cortical pathway encompassing the primary somatosensory cortices, the insula, and the anterior cingulate/prefrontal cortex. Contextual conditioned stimuli are projected to the lateral amygdala from the hippocampus and perhaps the bed nucleus of the stria terminalis. The long loop pathway indicates that sensory information relayed to the amygdala undergoes substantial higher-level processing, thereby enabling assignment of significance based on prior experience to complex stimuli. Cortical involvement in fear conditioning is clinically relevant because it provides a mechanism by which cognitive factors will influence whether symptoms are experienced or not following exposure to stress (157).

During the expression of fear-related behaviors, the lateral amygdala engages the central nucleus of the amygdala, which, as the principal output nucleus, projects to areas of the hypothalamus and brainstem that mediate the autonomic, endocrine, and behavioral responses associated with fear (158). The molecular and cellular mechanisms that underlie synaptic plasticity in amygdala-dependent learned fear are an area of active investigation (159). Long-term potentiation in the lateral amygdala appears to be a critical mechanism for storing memories of the association between conditioned stimuli and unconditioned stimuli (156). A variety of behavioral and electrophysiological data have led LeDoux and colleagues (157, 158) to propose a model to explain how neural responses to the conditioned stimuli and unconditioned stimuli in the lateral amygdala could influence long-term potentiation-like changes that store memories during fear conditioning. This model proposes that calcium entry through NMDA receptors and voltage-gated calcium channels initiates the molecular processes to consolidate synaptic changes into long-term memory (156). Short-term memory requires calcium entry only through NMDA receptors and not voltage-gated calcium channels.

This hypothesis leads to several predictions that may have relevance to psychological responses to stress. It suggests that blocking NMDA receptors in the amygdala during learning should impair memory of short- and long-term fear. This has been demonstrated in rodents (160, 161). Valid human models of fear conditioning and the availability of the NMDA receptor antagonist memantine should permit this hypothesis to be tested clinically (162). If memantine impairs the acquisition of fear in humans, it may have use in the prevention and treatment of stress-induced disorders such as PTSD. Blockade of voltage-gated calcium channels appears to block long-term but not short-term memory (163). Therefore, clinically available calcium channel blockers such as verapamil and nimodipine may be helpful in diminishing the intensity and impact of recently acquired fear memory and perhaps in preventing PTSD as well.

This discussion has focused primarily upon the neural mechanisms related to the coincident learning of the unconditioned stimuli-conditioned stimuli association (i.e., Pavlovian fear conditioning) in the lateral amygdala. However, there is significant evidence that a broader neural circuitry underlies the memory of fear that is modulated by amygdala activity. The inhibitory-avoidance paradigm is used to examine memory consolidation for aversively motivated tasks and involves intentional instrumental choice behavior. Studies using inhibitory avoidance learning procedures have been used to support the view that the amygdala is not the sole site for fear learning but, in addition, can modulate the strength of memory storage in other brain structures (164).

There is evidence that Pavlovian fear conditioning and inhibitory avoidance involve fundamentally different neural mechanisms. Pavlovian fear conditioning and inhibitory avoidance are differentially affected by posttraining pharmacological manipulations. The two types of learning involve different experimental procedures. In Pavlovian fear conditioning, the presentation of the conditioned stimuli and unconditioned stimuli occurs independent of behavior, whereas with inhibitory avoidance shock, delivery is contingent upon an animal’s behavioral response. Inhibitory avoidance may involve a more complex neural network because an animal’s response is contingent upon a number of contextual cues, in contrast to the more specific conditioned stimuli and unconditioned stimuli. The basal lateral amygdala is the primary amygdala nucleus responsible for voluntary emotional behavior based upon aversive emotional events, whereas the central nucleus of the amygdala is more involved in Pavlovian responses to fear-inducing stimuli (165). The relevance of the inhibitory-avoidance paradigm to human fear and anxiety rests on its assessment of a behavioral response to a fear-inducing context (166).

Specific drugs and neurotransmitters infused into the basal lateral amygdala influence consolidation of memory for inhibitory avoidance training. Posttraining peripheral or intra-amygdala infusions of drugs affecting GABA, opioid, glucocorticoid, and muscarinic acetylcholine receptors have dose- and time-dependent effects on memory consolidation (164). Norepinephrine infused directly into the basal lateral amygdala after inhibitory avoidance training enhances memory consolidation, indicating that the degree of activation of the noradrenergic system within the amygdala by an aversive experience may predict the extent of the long-term memory for the experience (167).

Interactions among CRH, cortisol, and norepinephrine receptors have important effects on memory consolidation, which is likely to be relevant to the effects of traumatic stress on memory. Extensive evidence indicates that glucocorticoids influence long-term memory consolidation by means of stimulation of glucocorticoid receptors. The glucocorticoid effects on memory consolidation require activation of the basal lateral amygdala, and lesions of the basal lateral amygdala block retention enhancement of intrahippocampal infusions of a glucocorticoid receptor agonist. Additionally, the basal lateral amygdala is a critical locus of interaction between glucocorticoids and norepinephrine in modulating memory consolidation (168).

There is also extensive evidence consistent with a role for CRH in mediating the effects of stress on memory consolidation. Activation of CRH receptors in the basal lateral amygdala by CRH released from the central nucleus of the amygdala facilitates the effects of stress on memory consolidation. As reviewed, there are important functional interactions between the CRH and norepinephrine systems, including a role in memory consolidation. Memory enhancement produced by CRH infusions in the hippocampus are blocked by propranolol and the noradrenergic toxin DSP-4 (75-R), suggesting that CRH infusions by means of a presynaptic mechanism stimulate norepinephrine release in the hippocampus (169).

Efferent projections from the basal lateral amygdala are also crucial to memory formation. The basal lateral pathway of the amygdala stria terminalis is involved, since lesions of the stria terminalis impair the memory-enhancing effects of intra-amygdala infusions of norepinephrine and systemic dexamethasone, which are presumably acting on the hippocampus. Also, lesions of the nucleus accumbens block the memory-enhancing effects of intra-amygdala infusions of glucocorticoid receptor agonist. Finally, the cortex is also a locus for memory consolidation, since projections from the basal lateral amygdala are essential in the modulation of memory by the entorhinal cortex (165, 170).

These results support the concept that CRH, by means of an interaction with glucocorticoids, interacts with the noradrenergic system to consolidate traumatic memories. Individuals with excessive stress-induced release of CRH, cortisol, and norepinephrine are likely to be prone to the development of indelible traumatic memories and their associated reexperiencing symptoms. Administration of CRH antagonists, glucocorticoid receptor antagonists, and β-adrenergic receptor antagonists may prevent these effects in vulnerable subjects.

Reconsolidation is a process in which old, reactivated memories undergo another round of consolidation (171–173). The process of reconsolidation is extremely relevant to both vulnerability and resiliency to the effects of extreme stress. It is the rule rather than the exception that memories are reactivated by cues associated with the original trauma. Repeated reactivation of these memories may serve to strengthen the memories and facilitate long-term consolidation (174, 175). Each time a traumatic memory is retrieved, it is integrated into an ongoing perceptual and emotional experience and becomes part of a new memory. Moreover, preclinical studies indicate that consolidated memories for auditory fear conditioning, which are stored in the amygdala (176), hippocampal-dependent contextual fear memory (171), and hippocampal-dependent memory associated with inhibitory avoidance (172) are sensitive to disruption upon reactivation by administration of a protein synthesis inhibitor directly into the amygdala and hippocampus, respectively. The reconsolidation process, which has enormous clinical implications, results in reactivated memory trace then returns to a state of lability and must undergo consolidation once more if it is to remain in long-term storage. Some controversies persist regarding the temporal persistence of systems reconsolidation. Debiec and colleagues (171) found that intrahippocampal infusions of anisomycin caused amnesia for a consolidated hippocampal-dependent memory if the memory was reactivated, even up to 45 days after training. Milekic and Alberini (172) however, found that the ability of intrahippocampal infusion of anisomycin to produce amnesia for an inhibitory avoidance task was evident only when the memory was recent (up to 7 days old). Further work is needed to resolve this very important question (173).

The reconsolidation process involves NMDA receptors and β-adrenergic receptors and requires cAMP response-element binding protein induction. The cAMP response- element binding protein requirement suggests that nuclear protein synthesis is necessary (177). NMDA receptor antagonists and β receptor antagonists impair reconsolidation (174, 178). The effect of the β receptor antagonist propranolol was greater after memory reactivation than when administered immediately after the initial training. These results suggest that reactivation of memory initiates a cascade of intracellular events that involve both NMDA receptor and β receptor activation in a fashion similar to postacquisition consolidation.

This remarkable lability of a memory trace, which permits a reorganization of an existing memory in a retrieval environment, provides a theoretical basis for both psychotherapeutic and pharmacotherapeutic intervention for traumatic stress exposure. Administration of β receptor and NMDA receptor antagonists shortly after the initial trauma exposure as well as after reactivation of memory associated with the event may reduce the strength of the original traumatic memory.

When the conditioned stimuli are presented repeatedly in the absence of the unconditioned stimuli, a reduction in the conditioned fear response occurs. This process is called extinction. It forms the basis for exposure-based psychotherapies for the treatment of a variety of clinical conditions characterized by exaggerated fear responses. Individuals who show an ability to attenuate learned fear quickly through powerful and efficient extinction processes are likely to function more effectively under dangerous conditions. They may also be less susceptible to the effects of intermittent exposure to fear stimuli, which can reinstate fear-conditioned learning. Highly stress-resilient individuals under extreme stress generally experience fear but have the capacity to function well under states of high fear. In addition, individuals in positions that regularly cause them to confront danger need to be able to extinguish learned fears rapidly.

Extinction is characterized by many of the same neural mechanisms as in fear acquisition. Activation of amygdala NMDA receptors by glutamate is essential (179), and L-type voltage-gated calcium channels also contribute to extinction plasticity (180). Long-term extinction memory is altered by a number of different neurotransmitter systems, including GABA, norepinephrine, and dopamine, in a manner similar to fear acquisition (181, 182).

Destruction of the medial prefrontal cortex blocks recall of fear extinction (183, 184), indicating that the medial prefrontal cortex might store long-term extinction memory. Infralimbic neurons, which are part of the medial prefrontal cortex, fire only when rats are recalling extinction—greater firing correlates with reduced fear behaviors (185). It has been suggested that the consolidation of extinction involves potentiation of inputs into the medial prefrontal cortex by means of NMDA-dependent plasticity. The basal lateral amygdala sends direct excitatory inputs to the medial prefrontal cortex, and NMDA antagonists infused into the basal lateral amygdala block extinction. The ability of the medial prefrontal cortex to modulate fear behaviors is probably related to projections from the medial prefrontal cortex by means of GABA interneurons to the basal lateral amygdala (186).

Failure to achieve an adequate level of activation of the medial prefrontal cortex after extinction might lead to persistent fear responses (187). Individuals with the capacity to function well after experiencing states of high fear may have potent medial prefrontal cortex inhibition of amygdala responsiveness. In contrast, patients with PTSD exhibit depressed ventral medial prefrontal cortex activity, which correlates with increased autonomic arousal after exposure to traumatic reminders (unpublished work by Bremner et al.). Consistent with this hypothesis, we (188) showed that PTSD patients had increased left amygdala activation during fear acquisition and decreased activity of the medial prefrontal cortex/anterior cingulate during extinction. It has been proposed that potentiating NMDA receptors using the glycine agonist d-cycloserine may facilitate the extinction process when given in combination with behavioral therapy in patients with anxiety disorders (189).