1Department of Psychiatry, University of Michigan, 1500 E. Medical Ctr Dr UH-9D, Box 0118,
Ann Abor MI 48109-0117, USA 2Department of Psychiatry, University of Michigan, VA Med Ctr, Research (11R) 2215 Fuller Road,
Ann Arbor MI 48105, USA 3Department of Psychiatry and Mental Health Research Institute 205 Zina Pitcher Place, University of
Michigan Ann Arbor MI 48109, USA
Abstract: Posttraumatic stress disorder (PTSD) is a potential consequence of being exposed to or witnessing an event provoking fear, helplessness, or horror. It is characterized by several debilitating symptoms including persistent hyperarousal, unwanted memories and thought intrusions, and hyperavoidance of stimuli or situations associated with the original trauma. The neurobiological mediators of these symptoms, however, still require elucidation, and animal models are particularly well suited for investigation of these mechanisms. Although the behavioral literature contains a large number of models that involve exposing animals to intense stressors, only a few of these are able to reproduce the biological and behavioral features of PTSD characteristic of a pathophysiological stress response. Among these are models involving single episodes of inescapable shock, which produce several bio-behavioral effects characteristic of PTSD, including opioid-mediated analgesia, noradrenergic sensitization as well as fear-conditioning effects. Single prolonged stress, involving the sequential exposure of rats to restraint, forced swim, and ether anesthesia, is able to produce an enhanced negative feedback of the HPA axis, as observed in PTSD patients, as well as a sustained exaggeration of the acoustic startle response. Predator exposure models, which invoke a significant threat of injury or death to particular animals, are effective in producing behavioral changes potentially analogous to hyperarousal symptoms, as well as changes in amygdala sensitization and long-term potentiation. In addition to these PTSD-specific models, several putative models, including single episodes of restraint/immobilization, forced swim, or early-life maternal separation, show promise as potential PTSD models, but require further validation and characterization before they might be considered specific to this disorder. Given the complexity of PTSD, both in terms of causal factors and symptoms, it is becoming apparent that multiple, independent physiological pathways might mediate this disorder. Future research may, therefore, wish to focus on endophenotypic models, which attend to one specific physiological pathway or neurobiological system, rather than attempt to reproduce the broad range of PTSD symptomatology. Combining information from numerous such models may prove a more efficient strategy.
Posttraumatic stress disorder (PTSD) is a psychiatric condition resulting from exposure to a traumatic event. The defining characteristic of a traumatic experience is its capacity to provoke fear, helplessness or horror. Usually this occurs in response to experiencing, confronting or even witnessing the
"Corresponding author. Tel.: + 1734-764-9527; Fax: + 1734936-7868; E-mail: [email protected]
threat of death, serious injury, or the loss of physical integrity. Some of the better known psychophysiological symptoms of PTSD include exaggerated startle, impaired sleep, intrusive memories or flashbacks and the persistent avoidance of situations or stimuli associated with the trauma.
Animal research has played a significant role in the current understanding of biological and bio-behavioral factors involved in PTSD. Although clinical studies have yielded important findings regarding the symptoms of PTSD, research in human patients is limited by ethical and practical concerns. Thus, providing detailed measurement of the bio-behavioral stress response, delineating specific neural mechanisms, characterizing the relevant features of stressor exposure that contribute to symptom development, and assessing potential therapeutic compounds can all be effectively and efficiently accomplished using animal models. Having said this, animal models of PTSD are still faced with a number of difficult challenges. Mainly because PTSD symptoms involve a significant subjective or experiential component, certain aspects of the disorder are difficult to model in animals. Furthermore, PTSD symptoms are often characterized by a highly variable time course and a high degree of comorbidity, effects which are also difficult to reproduce in models. The current chapter provides an overview of the most relevant animal models and their respective contributions to the current understanding of PTSD pathophysiology. It starts with a brief characterization of PTSD, including features that present particular challenges for animal research.
Posttraumatic stress disorder is currently the 4th most common psychiatric disorder, with 5-6% of men and 10-14% of women in the United States having been diagnosed at one point in their lives (Kessler et al, 1995; Breslau et al, 1998, 1999). Although PTSD is commonly associated with military veterans, the greatest likelihood of PTSD development occurs following rape. Other vulnerable groups include individuals with combat exposure, victims of physical assault, as well as individuals experiencing the sudden death of a loved one. This last group highlights the fact that not only direct exposure to trauma leads to symptoms, but witnessing violent injury or unnatural death can also constitute a traumatogenic experience. The incidence of PTSD suggests that only a minority of individuals exposed to trauma actually develop the disorder. Kessler et al. (1995) reported that 60% of men and 50% of women will experience a traumatic event at one point in their lives. Thus PTSD appears to reflect an abnormal response to stress rather than the normative one, although the greater the severity of the stressor, the greater the probability of developing PTSD. The natural history of PTSD suggests that symptoms of the illness start soon after the traumatic exposure and decrease in severity with time, although the course of symptoms can be quite variable. For instance, it is not unusual for symptoms to emerge months or even years after the initial trauma. Sometimes symptoms that had been dormant for years can suddenly reemerge in response to stressors that may or may not be related to the initial stress. PTSD is also often comorbid with other psychiatric disorders. In the National Comorbidity study, Kessler et al. (1995) reported that approximately 80% of male and 70% of female PTSD patients were diagnosed with at least one other psychiatric condition. Thus a significant percentage of PTSD patients suffer from mood disorders, other anxiety disorders (e.g., General Anxiety Disorder, Panic Disorder) as well as substance abuse and/or dependence.
The behavioral/psychological symptoms of PTSD are generally clustered into three groups. The first involves frequent reexperiencing of the traumatic event by thought intrusions, flashbacks, nightmares, or sensorimotor triggers. These effects often lead to a second set of symptoms that involve persistent avoidance of stimuli associated with the trauma. This avoidance can include simple behavioral withdrawal, but can also manifest in an inability to recall important aspects of the trauma, as well as experiencing feelings of detachment or estrangement from others. Another aspect of avoidance is a restricted range of emotional experience often expressed as an emotional numbing. A final group of symptoms fall under the category of hyperarousal. This includes an exaggerated startle response, hypervigilence, as well as other indirect effects such as increased irritability, insomnia, and a decreased ability to concentrate.
A number of these symptoms clearly involve subjective components that depend largely on self-report. They are therefore difficult to model in animals without considerable behavioral inference, which has led to two implications. Firstly, many animal models have been designed around their ability to reproduce biological symptoms of PTSD rather than psychological ones, since these are more readily observable and less ambiguous to interpret. Secondly, the selection of behavioral endpoints has relied largely on face validity. Thus, increased startle or fear-potentiated startle has been taken to represent symptoms of hyperarousal. Heightened conditioned fear responses (for instance to situational reminders) could represent reexperiencing symptoms or trigger avoidance symptoms. Decreased exploration in an open field and/or avoidance of novel stimuli can be analogous to avoidance symptoms. Decreased responding for rewards, or a differential response to analgesia and anesthesia has been taken as an indication of emotional numbing. Of these, startle and fear-potentiated startle are probably the most direct analogies and thus have been most extensively studied in the context of PTSD models. Up to 90% of PTSD patients complain of exaggerated startle (Shore et al., 1989; Davidson et al, 1991), which is characterized in part by heightened autonomic arousal in response to audiovisual presentations of traumatic scenes (Pitman et al, 1987; Shalev et al, 1992) and an increased cardiac and electrodermal response to acoustic stimuli (Paige et al, 1990; Shalev et al, 1992; Orr et al, 1995). The startle response is likely a survival mechanism of alarm, which rapidly alerts and arouses the organism to sudden stimuli or loud noises. In patients with PTSD, it is possible that this response is heightened or sensitized.
These symptoms are particularly well suited for study using animal models since over 50 years of research has provided extensive characterization of the biological and behavioral mediators of the acoustic startle response in rodents and other species. In addition to studying mechanisms underlying normal startle responses, a well-established phenomenon in the animal literature has involved fear-potentiated startle. This involves presenting the auditory startle stimulus in the presence of a cue (e.g., a light), which had been previously paired with an aversive event (i.e., a shock). Fear-potentiated startle is thus operationally defined as an elevated startle amplitude in the presence of the fear-associated cue.
Fear-potentiated startle has several attractive features, including its simplicity and intuitive face validity with regard to PTSD. However, it shares with all conditioning paradigms the confound of the memory processes involved. Also relevant to the study of PTSD are behavioral paradigms that assess the acoustic startle response to unconditioned stimuli. PTSD patients commonly exhibit exaggerated startle in the absence of fear-specific cues or in situations (such as at night), which represent an unconditioned aversive environment. In this regard, paradigms such as light-potentiated acoustic startle may be relevant since they assess baseline startle responses in dark and light environments. Bright lights are an unconditioned aversive stimuli for rodents and the acoustic startle response is, in fact, heightened in this environment (Walker and Davis, 2002). Furthermore, this light potentiation is inhibited by anxiolytic compounds, such as benzodiazepines and propranalol, at doses comparable to those that attenuate fear-potentiated startle (Walker and Davis, 2002). Recently, Khan and Liberzon (2004) demonstrated that animals undergoing a single traumatic stress (involving extended restraint, forced swim, and ether anesthesia) developed a sustained exaggeration in the acoustic startle response in both light and dark environments. Thus this particular startle paradigm may have considerable relevance to the study of hyperarousal symptoms in PTSD.
In addition to the behavioral symptoms listed above, there are a growing number of established biological signs associated with PTSD. It should be noted, however, that the biological study of PTSD is still in its relative infancy. Thus, many of the reported changes in response to traumatic stress have not yet met with consistent replication. As well, not all changes in response to stress are necessarily linked to the pathophysiology of PTSD, as many could be linked to other conditions or simply represent adaptive responses. The current section focuses on those changes that have been most reliably supported in conjunction with PTSD and where a strong theoretical link exists with particular symptoms.
Posttraumatic stress disorder patients show an elevated catecholamine response to trauma-related stimuli as indicated by changes in noradrenaline (Blanchard et al., 1991; Liberzon et al., 1999a) and epinephrine (McFall et al., 1990; Murgburg et al., 1994). PTSD patients also demonstrate increased 24-h circulating levels of noradrenaline and epinephrine in urine samples (De Bellis et al., 1999a). In concert, studies have shown low platelet a2 receptor concentrations in patients with PTSD (Perry et al., 1990, 1994) although this has not met with uniform replication (Gurguis et al., 1999). More convincing data about the role of al receptors, however, comes from studies showing potent anxiogenic effects for the a2 antagonist yohimbine. Yohimbine increases the startle response of PTSD patients to acoustic stimuli (Morgan et al., 1995) and, when administered IV, increases anxiety, panic attacks, and flashbacks (Southwick et al., 1993). Orally administered yohimbine in a natural setting also appears to exacerbate PTSD symptoms (Southwick et al., 1999a). PTSD patients appear particularly vulnerable to these effects as the effect on panic attacks and flashbacks could not be produced in normal controls. Given that a principle pharmacological effect of yohimbine would be to increase synaptic noradrenaline levels through inhibition of pre-synaptic autoreceptors, these findings further implicate the role of noradrenaline and oc2 receptors in PTSD. Indeed, the a2 agonist clonidine has shown clinical effectiveness in treating a number of PTSD symptoms (Kolb et al., 1984; Kinzie and Leung, 1989; Perry et al., 1990).
Noradrenaline cell bodies are densely localized in the locus coeruleus (LC) region of the brain stem and play a well-established role in mediating arousal. Studies have indeed shown that a single traumatic stress can lower the threshold of activation of LC-noradrenaline neurons (Curtis et al., 1999). Thus it is conceivable that a hypersensitive noradrenaline system could be linked to hyperarousal symptoms. Furthermore, LC-noradrenaline neurons also project to the amygdala and hippocampus, raising the possibility that these neurons also influence fear and memory responses to trauma. It has been suggested that retrieval of unpleasant memories in PTSD patients may be facilitated by elevated noradrenaline levels at the amygdala and hippocampus (Southwick et al., 1999b).
The most extensively characterized neuroendocrine change associated with PTSD involves abnormalities in the hypothalamic-pituitary-adrenal (HPA) axis. A wide variety of psychological and physiological stressors are known to produce acute activation of this axis (Herman and Cullinan, 1997) and termination of HPA activation is accomplished through a negative feedback system involving stimulation of glucocortiocoid receptors by Cortisol at the level of the hippocampus, hypothalamus, and/or pituitary. PTSD patients are shown to have lower 24-h circulating levels of Cortisol in some studies (Mason et al., 1986; Yehuda et al., 1995b) although others have found no sustained baseline differences (Mason et al., 2002). Their HPA axis is also characterized by an enhanced negative feedback system. A number of studies have demonstrated increased suppression of plasma Cortisol in PTSD patients following administration of low doses of the glucocorticoid agonist dexamethasone (Yehuda et al., 1993; Goenjian et al., 1996). In addition to these effects, PTSD patients also show increased concentrations of glucocorticoid receptors in plasma lymphocytes, suggesting systemic alterations in receptor regulation (Yehuda et al., 1995a).
HPA axis abnormalities may also be linked to PTSD symptoms. For example, the HPA response to stress is an important component in returning organisms to physiological homeostasis following stressor exposure and represents part of the organism's adaptive response. Hypervigilance symptoms or symptoms of irritability and quick anger may reflect an inability in PTSD patients to adapt or adjust properly to external stressors or changes in homeostasis. Furthermore, adrenalectomized rats have been shown to demonstrate increased spontaneous firing rate at the LC, and thus low circulating glucocorticoid levels may also be linked to hyperarousal symptoms (Pavcovich and Valentino, 1997). It is critical to note at this point, in the context of developing valid animal models of PTSD, that the HPA dysfunction observed in PTSD patients is distinct from that observed in unipolar depression. Depression is in fact associated with high Cortisol levels and blunted dexamethasone suppression. This distinction provides a highly useful tool for differentiating those stress models that produce symptoms more relevant to depression than to PTSD.
Posttraumatic stress disorder patients demonstrate lower pain thresholds and increased opioid-mediated analgesia (Van der Kolk et al., 1989; Pitman et al., 1990). These observations have led to the hypothesis that increased central nervous system (CNS) opioid activity exists in patients with PTSD. Combat veterans with PTSD show a decreased sensitivity to pain when being exposed to traumatic reminders (Van der Kolk et al., 1989) and this effect is reversible with the opioid antagonist naloxone (Pitman et al., 1990), suggesting a heightened analgesic effect in this group. Although PTSD patients may have lower (Hoffman et al., 1989) or normal (Hamner and Hitri, 1992) resting levels of plasma beta-endorphins, they show higher levels in cerebral spinal fluid (CSF) as compared to controls (Baker et al., 1997). There is also a negative correlation between beta-endorphin levels in CSF and PTSD intrusive and avoidant symptoms (Baker et al., 1997). Similarly, our group (Phan et al., 2002) found higher amygdala (i-opioid binding in PTSD patients, as compared to combat controls, and a strong negative correlation was observed between [i-opioid receptor binding at the amygdala and the severity of PTSD symptoms and PTSD-related anxiety.
Neuroanatomical changes associated with traumatic stress exposure have been demonstrated in limbic and paralimbic regions, such as the amygdala, hippocampus, anterior cingulate gyrus, and insula (Rauch et al., 1996; Shin et al., 1999). The amygdala, a region well-established in mediating the emotional response to fear, appears more readily activated in PTSD patients exposed to traumatic stimuli (Liberzon et al., 1999c; Rauch et al., 2000). Decreased hippocampal volume and/or cell density in PTSD patients perhaps contributing to memory deficits were reported in some studies (Bremner et al., 1995), but not others (De Bellis et al., 1999b; Bonne et al., 2001). In animals, physiological inhibition of long-term potentiation (LTP) at the hippocampus has also been observed following a variety of behavioral stressors including restraint and inescapable tailshock (Shors et al., 1989; Shors and Dryver, 1994; Kim et al., 1996) and predator exposure (Mesches et al., 1999). The presence of hippocampal abnormalities in PTSD does remain controversial, since methodological differences in various studies have yielded inconsistent reports. Changes in glucocorticoid receptor densities have also been observed at hippocampal CA1 neurons in response to a single traumatic stress (Liberzon et al., 1999b), which may mediate the increased sensitivity of the glucocorticoid negative feedback system.
In the behavioral literature there are currently a large number of models that involve exposing animal subjects to intense stressors. Such models provide an opportunity for investigating the normal behavioral and physiological responses to fear and anxiety. However, it is important to note here that PTSD represents the abnormal response to stress rather than the normal response, as only a fraction of individuals exposed to trauma actually develop the disorder. Thus exposing animals to an intense or abnormal stressor is not, in itself, sufficient for consideration as a valid PTSD model. Rather it is also necessary to reliably reproduce the behavioral symptoms and/or biological signs distinctive to PTSD, which are indicative of a pathophysiological anxiety response.
At this point, a number of models have shown an initial promise as they appear to produce symptoms and signs characteristic of PTSD. In addition, there are several putative models, which possess a degree of face validity and produce some symptoms potentially relevant to PTSD. The following paragraphs will describe both putative and PTSD-specific animal models. Putative models, which involve single stress episodes such as restraint and forced swim, appear to incorporate some of the defining characteristics of a traumatic stress, including the ability to provoke fear and helplessness, as well as posing a significant threat of injury or death. They also produce a number of specific changes in stress-response systems (see below) that are highly relevant to both symptom and neurobiological alterations seen in PTSD. Similarly, models involving early-life environmental stress appear analogous to early-life traumas, such as abuse or neglect, which have also been linked to development of PTSD. Despite their similarities, however, these models remain putative as they have yet to consistently reproduce symptoms distinctive to PTSD. In contrast, PTSD-specific models, which will be described afterwards, also involve exposure to single stress episodes but have been more successful in producing symptoms characteristic of a pathophysiological stress response.
Putative stress models
Single episode of restraint/immobilization
In restraint models, the movement of animals is severely restricted for a prolonged period of time through containment in a small, whole-body chamber (for instance a Plexiglas restraining tube or a small plastic bag). In some situations, total immobilization can be achieved by tying each individual limb to a solid surface such as a wooden board. A well-replicated consequence of prolonged restraint is a reduction in locomotor activity in novel environments. A number of studies have shown long-lasting decreases in locomotion following a single episode, lasting in some cases for several weeks (e.g., Carli et al, 1989). Although most of these studies used restraint durations in excess of 30min, Shinba et al. (2001) found that a relatively brief 8-min restraint episode produced an 80% decrease in exploratory behavior in a novel open field, lasting up to 14 days. Interestingly, this effect was reversed by pretreatment with the a2 agonist clonidine. Since a2 receptors are primarily autoreceptors, clonidine has the effect of reducing noradrenaline activation, and in fact has demonstrated clinical efficacy in treating PTSD symptoms.
Marti et al. (2001) found that a single exposure to restraint or immobilization also modified the HPA response to the same stressor several days later.
Specifically, when compared to control animals, a faster recovery of circulating ACTH and corticoster-one was observed following reexposure to the same stressor. This was taken to indicate an enhanced termination of the HPA response. This is similar to PTSD, which is also characterized by enhanced termination. However, these authors also found that this effect was only observed if the identical stressor was repeated. This might, therefore, suggest a stressor-specific adaptive response, rather than a generalized change in HPA responsiveness. Further investigation involving glucocorticoid administration and/or quantification of glucocorticoid receptors in brain or plasma would help to clarify the nature and extent of the HPA-related changes.
In forced swim paradigms, animals are placed in cold or room temperature water and required to swim continuously in order to keep their head above the water surface. Such tests are presumably quite stressful since continuous swim is necessary to avoid drowning. Curtis et al. (1999) demonstrated that a 20-min forced cold water swim produced significant immobility as compared to control rats. Interestingly, it also had the effect of lowering the threshold of excitation of LC neurons to CRF stimulation. Stress-induced sensitization of noradrenaline neurons at the LC had previously been observed in response to chronic stress. For instance rodents in chronically cold environments (5°C) demonstrate sensitized LC-noradrenaline activation to acute footshock (Mana and Grace, 1997) and central CRF administration (Jedema et al, 2001). Cold stress also leads to a heightened noradrenaline release at the hippocampus, which is a major projection site for LC-noradrenaline neurons (Nisenbaum et al, 1991). The findings by Curtis et al. (1999) suggest that a single intense stress exposure can also produce physiological supersensivity of the LC-noradrenaline neurons, and could possibly be linked to subsequent hyperarousal symptoms. Interestingly, Gesing et al. (2001) observed that 24 h following a 15-min forced swim (in 25°C water), rats also showed an upregulation in mineralocorticoid receptors (MRs) at the hippocampus, an effect inhibited by preswim administration of CRF antagonists. Rats exposed to forced swim also showed enhanced MR-mediated inhibition of HPA activity in response to a challenge test with an MR antagonist. Although these neuroendocrine changes are similar to that observed in PTSD, the upregulation in glucocorticoid receptors was not sustained, and returned to prestress levels within 48 h. Although these results show promise, further study is required to determine if forced swim produces behavioral symptoms of hyperarousal (such as enhanced acoustic startle), and whether sensitization of LC- noradrenaline neurons also leads to systemic changes in catecholamine levels, as is found in PTSD patients.
Childhood abuse and/or neglect can be a significant early-life source of trauma for many individuals, and can often lead to PTSD symptoms in adulthood. A number of developmental animal models have been developed to attempt to assess the effects of neonatal and early-life experience on subsequent behavioral and biological stress responses. However, although these models have considerable face validity, their ability to reproduce a PTSD-relevant picture is yet to be demonstrated.
One type of neonatal stress involves exposing young animals to noxious stimuli, such as footshock, temperature extremes, pinprick, or even surgical procedures. During the first two weeks of life, noxious stimuli invoke a subnormal HPA response (Shapiro, 1968; De Kloet et al„ 1988). During this period, stress-induced CRF and ACTH responses are attenuated (Grino et al„ 1989; Walker et al., 1991; Baram et al, 1997) and baseline corticosterone levels are lower than normal (Sapolsky and Meaney, 1986). It is unclear at this point what the significance of these effects might be. They could be related to subsequent vulnerability to stress or stress-related disorders, or they may reflect an adaptive response characteristic of HPA axis development. Further research is required to do determine whether these changes are in fact linked to behavioral symptoms in adulthood.
A commonly used model of early-life trauma involves maternal separation or maternal deprivation. Such models appear analogous to situations involving early childhood neglect or separation. In this paradigm, rat pups are deprived of maternal care by being taken from their mother, prior to weaning, for a set period of time. Although this paradigm induces long-term changes in adult behavior indicative of anxiety, many of the physiological changes produced also closely resemble depressive symptoms. For instance, maternal separation in rodents is characterized by increased circulating glucocorticoids (Kuhn et al, 1990; Pihoker et al, 1993), a decreased sensitivity in the HPA negative feedback system, and lower glucocorticoid receptor densities (Ladd et al, 2000). In addition, maternal separations for 3-6 h per day for two weeks enhanced the ACTH response to relatively mild stressors, such as exposure to novel environments (Plotsky and Meaney, 1993). Twenty-four-hour maternal separation also produces heightened ACTH and corticosterone responses to stressful stimuli (Cirulli et al, 1994; Walker, 1995). In contrast, PTSD is characterized by low-circulating glucocorticoids, enhanced HPA feedback, high GC receptor densities, and a blunted ACTH response. One potential similarity between maternally separated animals and PTSD is high central CRF levels, since maternal separation produces an increase in CRF mRNA at the amygdala (Heim et al, 1997) and hypothalamus (Plotsky and Meaney, 1993) and increased CRF concentrations at the median eminence (Ladd et al, 1996). Similarly, PTSD patients demonstrate high CRF levels in cerebrospinal fluid (Baker et al, 1999). However, high central CRF is not unique to PTSD as increased CRF in the CSF is also found in depression, and centrally administered CRF in animals produces a number of behavioral symptoms similar to that observed in patients with major depression (Nemeroff and Owens, 2002). Thus with respect to HPA axis changes, it appears that maternal separation paradigms might be a better model for depression than PTSD.
Since stress is an etiological factor in the development of both depression and PTSD, one of the challenges in developing PTSD models is discerning whether stress-induced behavioral or biological changes are reflective of one disorder or the other. For instance, immobility produced by single episodes of restraint could be indicative of PTSD-like avoidance behaviors, but they could also reflect a general lethargy or behavioral apathy characteristic of depression. Similarly, the efficient recovery of HPA functioning following repeat restraint stress could reflect normal habituation processes (i.e., a chronic stress response potentially leading to depression) or enhanced negative feedback (PTSD response). As already eluded to, the fact that stress can either decrease (depression) or increase (PTSD) sensitization of inhibitory components of the HPA axis provides a useful tool for dissociating depression models from PTSD models. Thus, decreased baseline circulating corticosterone, rapid HPA inhibition in response to exogenously administered glucocorticoids, and higher glucocorticoid receptor densities are all indicators of increased sensitization and thus are more likely to be reflective of PTSD. On the other hand, effects such as increased concentrations of brain CRF could have dual interpretations and models producing these effects might benefit from further characterization.
PTSD-specific models Inescapable shock (single episode)
It has been hypothesized that an important aspect of traumatic stress exposure that contributes strongly to subsequent symptom development is the low degree of control and predictability over the stressful stimuli (Foa et al, 1992). To incorporate these components, one of the more common PTSD models has involved inescapable shock (IES). Typically in this model, animals are exposed to a single or repeated episode of footshock or tailshock without an opportunity for escape. Often animals are then tested on subsequent days, with shock treatment, to determine if an escape deficit has developed in response to escapable or controllable stressors (this deficit is also called learned helplessness). Learned helplessness paradigms generally require multiple IES episodes, and a repeat shock exposure in the escape tasks. As such, they might be less relevant to PTSD, which in humans often emerges after a single stress episode. Yehuda and Antelman (1993) pointed out that in a valid model of PTSD, even brief stressors should be capable of producing the biological and behavioral sequelae of PTSD. Learned helplessness models, involving multiple shock sessions, tend to produce additional symptoms more analogous to depression than to PTSD. For instance, common effects of chronic IES include decreased movement away from aversive events (Weiss et al., 1981), and reduced body weight and food and water intake (Weiss, 1968). The effect on avoidant behaviors is a particular concern, as PTSD subjects tend to have heightened avoidant response rather than impaired ones. The decrease in food and water intake could represent an emotional numbing or increased anhe-donia, but combined with the decrease in body weight, may also be a symptom of general behavioral apathy more characteristic of depression. Indeed learned helplessness models are among the most common paradigms used for screening putative antidepressant drugs.
On the other hand, single episodes of IES produce a range of behavioral and biological characteristics with similarities to PTSD including exaggerated fear conditioning (Maier, 1990), increased neophobia (Job and Barnes, 1995), decreased social interaction (possibly analogous to avoidant behaviors) (Short and Maier, 1993), and decreased consumption of a palatable food (Griffiths et al., 1992). In addition, IES also produces a pronounced opioid analgesia analogous to that observed in human PTSD patients (Van der Kolk et al, 1985; Maier, 1989).
A noradrenergic hypersensitivity has also been reported following IES with some evidence of time-dependent sensitization. Irwin et al. (1986a) found that a single episode of footshock in mice produced a transient increase in noradrenaline utilization, and subsequent shocks of milder intensity reproduced this. Similarly, Curtis et al. (1995) found that acute (but also) chronic footshock lowered the threshold of activation of LC-noradrenaline neurons to localized CRF administration, also suggesting sensitization effects. Locally administered CRF at the LC is known to excite LC-noradrenaline neurons and is linked to its arousal functions. Servatius et al. (1995) further found an enhanced acoustic startle response in animals receiving 2 h of IES, although this effect was only present on the 7th day poststress.
A highly relevant aspect of IES phenomenon, as pertains to PTSD, is that for particular types of symptoms, brief and intense shocks are capable of producing long-lasting effects. Thus, Van Dijken et al. (1992b) found that one brief IES session, involving ten 50-Hz footshocks (6 s each) over 15min, produced a heightened response to noise stimuli lasting up to 14 days, an enhanced immobility response to a sudden drop in background noise at 21 days (Van Dijken et al, 1992a) and decreased behavioral activity and increased defecation in a novel open field at 28 days (Van Dijken et al, 1992c). Further evidence of the importance of single versus chronic stressors in IES in reproducing PTSD-like biological signs comes from studies showing an increase in GR- and MR-binding capacity at the hippocampus 14 days after a single session of IES (Van Dijken et al, 1993). In contrast, chronic IES appears to produce a downregulation of glucocorticoid receptors, more indicative of depression than PTSD (Sapolsky et al, 1984).
While IES models appear to reproduce a number of symptoms relevant to PTSD, not all effects are consistent with this disorder. For instance, IES models involve a decrease in catecholamines following acute footshock (Van der Kolk et al, 1985; Irwin et al, 1986b), whereas PTSD patients demonstrate a sustained increase in noradrenaline activation. Evidence also shows that a single exposure to IES produces a sensitized ACTH response to subsequent novel stressors (van Dijken et al, 1993). However, human PTSD patients show a blunted ACTH response to CRF administration (Smith et al, 1989) at least in some studies. Another concern of IES models is the transient nature of many of the behavioral changes. For instance, a number of studies have found that for certain anxiety measures, as well as for opioid-mediated analgesia and fear conditioning, IES effects disappear if subjects are tested as little as 1-7 days after stress exposure (Jackson et al, 1979;
Grau et al, 1981; Weiss et al, 1981; Maier, 1990; Short and Maier, 1993). For these models to have a greater relevance to PTSD, the biological or behavioral symptoms produced should be more enduring.
To address this issue, an interesting modification to IES models has involved examining the effect of situational reminders of the initial shock experience. Maier (2001) found that such reminders can greatly increase the duration of IES effects. For instance, symptoms of learned helplessness following a single session of tailshock were extended when subjects were exposed to reminders or cues of the initial shock exposure, such as presentation of the shock chamber. Pynoos et al. (1996) also found that cues associated with the initial brief shock episode produced an increase in aggressive behavior, significantly decreased exploratory activity in an open field as well as producing a progressive increase over time in the magnitude of the startle reflex. This latter effect is particularly interesting since it suggests that instead of habituating to the presentation of conditioned stimuli, animals actually become sensitized.
Thus a considerable potential exists for IES models in assessing specific aspects of PTSD. Further study may need to characterize long-term neuroendocrine changes following single shock episodes to further differentiate PTSD-like effects from depression-like symptoms.
As indicated earlier, abnormalities in glucocorticoid and HPA functioning are among the more robust neurobiological symptoms associated with PTSD. Glucocorticoid dysfunction is not unique to PTSD, as it is also implicated in depression, anxiety, memory, and even cell death. However, neuroendocrine abnormalities in PTSD appear specifically characterized by an enhanced negative feedback of the HPA axis and an upregulation of glucocorticoid receptors. Recently, a specific stress paradigm was developed which successfully reproduced both the enhanced negative feedback and receptor upregulation. In this model, called single prolonged stress (SPS), animals received 2 h of restraint, followed immediately by 20min of forced swim (in 24°C) water, followed by exposure to ether vapors until loss of consciousness. Animals then remained untouched (undisturbed) for 7 days, which proved to be a critical component in subsequent development of PTSD-like characteristics.
Animals were then given a rodent equivalent of the dexamethasone suppression test, involving glucocorticoid injection and subsequent tail-blood collection for measurement of ACTH and corticosterone. SPS-exposed animals demonstrated increased sensitivity of glucocorticoid negative feedback as indicated by a blunted ACTH response to stress (Liberzon et al., 1997). Furthermore, the HPA fast feedback in these subjects was specifically linked to changes in hippocampal glucocortoicoid receptor concentrations (Liberzon et al., 1999a). Rats exposed to SPS thus showed an altered GR/MR receptor ratio characterized by an upregulation in GR and down-regulation of MR receptors at the hippocampus. When SPS animals were compared to those receiving chronic stress, the effects on HPA feedback were not observed. Although SPS produced an upregulation of both GR and MR early on, if the stress was continued chronically glucocorticoid receptor concentrations actually returned to prestress levels. Thus single prolonged, rather than chronic or long-term stress in this model was capable of producing PTSD-relevant neuroendocrine changes.
In addition to neuroendocrine effects, it was recently demonstrated that SPS also produced a long-term exaggeration of the acoustic startle response. In the light-potentiated acoustic startle paradigm, SPS-exposed animals showed a heightened startle response in both light and dark environments 14 days after stressor exposure (Khan and Liberzon, 2004). Kato (2002) also found that SPS increased low-frequency stimulated LTD in hippocampal neurons. The physiological converse to this effect, inhibition of LTP, reliably occurs in hippocampal neurons following a number of stressors, including restraint, inescapable tailshock (Shors et al., 1989; Shors and Dryver, 1994; Kim et al., 1996), and predator exposure (Mesches et al., 1999). Taken together, SPS may provide a useful animal model of PTSD, capable of reproducing behavioral, neuroendocrine, and neurophysiological symptoms (see Fig. 1).
An important component of traumatic stress exposure is the ability to invoke a significant threat of injury or death. Animal models attempting to incorporate these features have often used predator exposure models. Blanchard and Blanchard (1989) were among the first to demonstrate that a brief escapable exposure of rats to a cat produced acute increases in defensive behavior, lasting about 24 h, which included increased withdrawal, immobility, and risk-assessment behavior. Cat odor was also shown to increase anxiety in the elevated plus maze and social interaction tests, although these anxiogenic effects were no longer evident 24 h later (Zangrossi and File, 1992). Adamec and Shallow (1993) found that a single 5-min unprotected exposure of a rat to a cat produced anxious behavior in the elevated plus maze test lasting 3 weeks after the stress. Although the cats were nonaggressive by nature, the rats were unaware of this and could not escape. In addition to effects on the plus maze, this stressor also produced a potentiation of the acoustic startle response 8 days following exposure (Adamec, 1997).
Predator exposure may also sensitize the amygdala to subsequent stressful stimuli, which would be similar to symptoms observed in human PTSD patients (Liberzon et al., 1999c; Rauch et al., 2000). Cook (2002) showed that when sheep were exposed to a predator (in this case a dog) a significant CRF and glucocorticoid activation was observed at the amygdala. Subsequent presentation of a novel stress (forelimb shock) produced an exaggeration in the amygdala CRF response suggesting a sensitization. Interestingly, this sensitization effect was inhibited when animals received a glucocorticoid antagonist prior to the repeated stress. It is thus interesting to speculate that the sensitization effect may be mediated by an upregulation of glucocorticoid receptors, which has been observed in PTSD patients. CRF neurons at the amygdala show a 90% colocalization with glucocorticoid receptors (Honkaniemi et al., 1992), and glucocorticoids injected into the central nucleus of the amygdala produce anxiogenic effects (for instance, in the elevated plus maze) via CRF neurons (Shepard et al., 2000).
Gesing et al. (2001) also found an interaction between CRF and corticosterone in the regulation of
Was this article helpful?