Randall R Sakai and Kellie LK Tamashiro

Department of Psychiatry, University of Cincinnati Medical Center, 2170 E. Ga/braith Road, Bldg 43/UC-E, Cincinnati, OH 45237, USA


The field of stress research has generated growing interest in response to the escalating number of psychopathologies associated with chronic stress in humans. A better understanding of the mechanisms through which chronic stress leads to these pathologies will enhance the development, effectiveness and efficiency of rational clinical therapies. As such, a variety of animal models have been developed to study the consequences of acute and chronic stress as they may relate to human conditions of stress. These models include electric foot shock, cold, forced swim, restraint, and forced running, each of which produces many of the same neuroendocrine changes seen in humans. Animal models of stress have facilitated the understanding of the behavioral and physiological mechanisms that are common and shared by various stressors. Although these models are useful and commonly used as laboratory stressors, many are physical in nature and represent an immediate threat to an animal's physiological homeostasis. Thus, these "physical" stressors are extremely artificial to the test animal and are sometimes used without consideration of the animal's normal ethology and often do not resemble the challenges that these animals normally face within their natural environment. This can be contrasted with "social" or "psychological" stress models, such as social defeat or social hierarchy formation, that have been found to elicit different behavioral and physiological responses (Herman and Cullinan 1997; Martinez et al, 1998; Sawchenko et al, 1996).

Corresponding author: Phone: + 1513-558-6589; Fax: +1513558-5783; E-mail: randall.sakai (รค ue.edu

The most prevalent type of stress encountered by humans occurs through social interactions, and a number of detrimental health conditions have been associated with chronic social stress (Brown, 1989). The relatively nonphysical nature of most social stress animal models renders them useful in examining stress-related pathology in humans since few people in contemporary society will experience severe physical stressors in their lifetimes while psychological stressors are encountered routinely. These observations highlight the importance of carefully evaluating animal models for studies of stress-related conditions in humans to assess their usefulness in scientific research. The use of an animal model that closely represents the behavioral and physiological effects of stress that are relevant to the animal can be a valuable tool in both basic and clinical studies. It provides a vehicle to investigate the etiology of social stress-induced pathologies and to develop and evaluate the effectiveness of possible treatment for such disorders.

In general, animal models of human disorders can be judged by three criteria (Willner, 1991): The first is face validity, referring to how well the model reproduces symptoms of the related human condition. The second is predictive validity, assessing how well the animals in the model respond favorably to the same drugs that humans do under analogous conditions. The final criterion, construct validity, addressing the question of how consistent the model is with theoretical rationale. That is, are the human and animal responses similar (e.g., do they share the same pathways or substrates), such that the animal model may be used to extend findings to human conditions. In evaluating animal models according to this set of criteria, while no model is expected to be perfect, it is evident that some may be better suited for studies of social stress than others.

There are several models of social stress that are commonly used and some of the most popular include the resident-intruder model of social defeat followed by the colony model of subordination stress. Both paradigms produce characteristic behavioral, physiological and neurochemical changes in defeated or subordinate animals. In this chapter we provide an overview of animal models that have been developed to study the effects of social stress in a laboratory setting. In particular, we will focus on an animal model of social stress that results from dominance hierarchy formation in rats housed under ethologi-cally pertinent conditions to this species. We will include a description and discussion of the acute behavioral, physiological and biochemical consequences of social subordination as well as some of the long-term consequences after recovering from chronic stress and its health implications.

Animal models of social stress Social defeat (resident-intruder)

The resident-intruder paradigm of social defeat is a popular and widely used animal model of social stress that is applicable to a wide range of animal species. This model is based on the establishment of a territory by a resident male, such that it subsequently defends the territory against unfamiliar male intruders. Although there are a number of variations of this model, the basic principle remains the same. A male animal ("intruder") is introduced into the cage of another male of the same species ("resident") and the two are allowed to interact for a short time (minutes). The initial exposure involves physical attack and usually ends with the intruder being defeated by the resident. Thus, the intruder is sometimes referred to as the "defeated animal" or the "subordinate" and the resident the "victor" or the "dominant". Acute defeat experiments end with a single defeat. In chronic defeat paradigms, the subordinate is usually removed from direct physical contact with the dominant and is separated and thus protected by a wire cage or partition. This procedure minimizes further wounding from physical attack by the resident; however, since the subordinate retains visual, olfactory, and auditory contact with the dominant, social stress during this phase is derived from the threat of attack. The time period of subsequent exposures can vary from minutes to weeks and may be intermittent or continuous.

There are numerous variations of this paradigm. As described above, most social defeat models involve two phases: physical attack and threat of attack. Some include both physical and threat of attack (Korte et al, 1990; Tornatzky and Miczek, 1994; Martinez et al, 1998) while others only rely on threat of attack (Miczek and Mutschler, 1996; Sgoifo et al, 1998; Fuchs and Flugge, 2002) in producing psychosocial stress. In the mouse sensory contact model (Kudryavtseva, 1991, 2000), an intruder male is continuously housed in the home cage of the resident separated only by a clear partition. The partition is removed for a few minutes each day during which the resident is allowed to attack the intruder.

Juvenile rainbow trout are extremely territorial animals and when grouped in an arena will engage in agonistic activity that produces a dominant-subordinate relationship (Jonsson et al, 1998; Winberg and Lepage, 1998; Overli et al, 1999). In a typical study, fishes are pair-housed resulting in one dominant and one subordinate. Furthermore, due to the nature of their interactions, there is no wounding that results in trout interactions, unlike other models of group-housed animals that develop a dominance hierarchy.

It is important to note that depending on the social defeat paradigm, the experimenter can play a significant role in manipulating the conditions of the test to generate the desired outcome in resident-intruder models. First, the resident is usually selected with higher body weight and aggression to give him a greater advantage in fighting. In many cases the resident male is also housed with a female such that he is more willing and eager to defend his territory (Flannelly and Lore, 1977). Second, residents are used repeatedly, thereby giving them experience of victory and further increasing their aggressiveness. However, if naive animals are used, it is also important to consider whether there are preexisting conditions that predispose an animal to becoming dominant or subordinate in the social situation.

Social hierarchy in group-housed animals

Social stress is common in many animal species and typically results from disputes over resources such as space, access to a reproductive partner, food or water. A number of models take advantage of the natural tendency of different species to form social hierarchies when housed in groups. These species include sugar gliders (Mallick et al., 1994; Jones et al., 1995), rats (Barnett, 1958; Barnett et al., 1960; Taylor et al., 1987; Blanchard and Blanchard, 1989; de Goeij et al., 1992; Dijkstra et al., 1992; Fokkema et al., 1995; Blanchard et al., 1995; Stefanski et al., 2001), mice (Ely and Henry, 1978), Cynomolgous macaques (Fontenot et al., 1995; Shively et al., 1997a, 1997b; Shively, 1998; Kaplan et al., 2002), and baboons (Sapolsky, 1990; Virgin and Sapolsky, 1997). Establishing dominance in a group setting is psychologically and physically stressful for both the dominant and subordinate animals. Furthermore, since animals are group housed for extended periods of time, the members of the group are continuously exposed to social stress as opposed to the intermittent exposures often used in social defeat paradigms.

Housing conditions, group gender composition and number often vary among colony models. Some models use large open areas (Mallick et al., 1994) while others use burrow and tunnel systems that model the animals' natural living environment (Ely and Henry, 1978; Blanchard and Blanchard, 1989; Dijkstra et al., 1992; Blanchard et al., 1995; Tamashiro et al., 2004). Some researchers, such as Sapolsky and colleagues, studied wild baboons in their natural habitat in Africa (Sapolsky, 1990). Different types of housing conditions are used in the various models and it appears that with a larger and more natural habitat such as that used in colony models, animals will fight more and display more distinct characteristics of dominance and subordination (Blanchard and Blanchard, 1990). In addition, one should also consider that different housing conditions might produce gender-dependent differences in stress. For example, male rats have higher corticosterone levels when housed under crowded conditions while females have higher levels when housed individually (Brown and Grunberg, 1995).

Group composition is also an important factor in the formation of a dominance hierarchy. The presence of females results in a higher aggression level among males than when females are not present (Flannelly et al., 1982; Taylor et al., 1987). If females are not included in the rat colonies, there is no evidence of a clear hierarchy in males (Barnett, 1963; Tamashiro, 2004).

Other models of social stress: variations on a theme

There are other paradigms of social stress that do not fit perfectly into the social defeat or colony hierarchy models. Some of these variations include housing two males with one female, where over time the formation of a dominant and subordinate male is established (Sachser and Lick, 1989, 1991; Sachser et al., 1994). Other models involve introducing a previously dominant male animal into an established colony where the new male now loses his former dominant status in a similar fashion to the resident intruder paradigm (Willner et al., 1995). Finally, there are other variations of social defeat that consist of placing an intruder male in an established mixed-gender colony where the intruder is attacked by the dominant member of the colony (Williams and Lierle, 1988).

Female models of social stress

It is well recognized that the prevalence, etiology and response to treatment of psychiatric disorders is gender dependent (Earls, 1987). Although studies involving females are highly desirable given the preponderance of anxiety and depressive disorders in women, there is a serious lack of animal models to address this issue. Social defeat and subordination are effective social stressors for males; however very few of these studies are done using females. It is not for the simple reason that females do not respond to stress; in fact, females show more pronounced stress-induced changes in social instability and social disruption models than in dominant-subordinate models (Haller et al., 1998). Additionally, other investigators report higher anxiety-like behavior and higher corticosterone levels in individually housed females and group-housed males suggesting that males and females perceive housing conditions differently (Brown and Grunberg, 1995; Palanza et al., 2001). The influence of social stress on behavioral alterations in females typically occurs only in a maternal care or in the context of protection of young (von Saal et al., 1995). It is apparent from these observations that females respond to different stressors and in a different manner compared to males. Therefore, unique models must be developed to address these gender-specific issues.

A few studies have incorporated original features of female stress responses in their animal models including nonpregnant females that are attacked by another female opponent (Scholtens et al., 1990) or a lactating female (Haney and Miczek, 1993). Other studies have shown that the social stress of housing two pregnant female rodents can influence intrauterine mortality suggesting that an alpha-female may exist in this unique experimental setting (Wise et al., 1985). Although resident-intruder social stress models in females are few in number due to the relatively low defensive behavior in females, some investigators have taken advantage of the situations in which females do display higher aggressive behavior. Lactating females will actively defend their nests against intruders (Sgoifo et al., 1995) and ovariectomized females also tend to be more aggressive and have been used as residents (Huhman et al., 1990). Dominance hierarchies in most animal species are more distinct in males than they are in females; however, Shively et al. have successfully developed a model of social hierarchy in the Cynomolgus macaque (Shively et al., 1997a, 1997b; Shively, 1998). It is possible that female hierarchies are more pronounced or evident in higher species than in rodents. Obviously, more research using females in these various stress models should be encouraged to determine whether the same physiological and neurochemical changes due to stress seen in males are also evident in females.

Considerations in choosing an animal model of social stress

The use of animal models in the laboratory has greatly facilitated the progress of basic research and is fundamental to gaining a better understanding of biological processes and the mechanisms by which these processes may be disrupted to produce pathological states. The following is a summary of some of the important experimental conditions that we believe should be considered when selecting and using an animal model of social stress in a laboratory setting.

Prior experience and preexisting conditions

Some colony models of social stress rely on manipulations or disruptions of stable colony behavior. For example, one version of the social disruption model involves introduction of highly aggressive males into stable social groups of animals (Padgett et al., 1998); while in another variation, social groups are formed and allowed to stabilize before group members are randomly mixed. In many cases social disruption models involve prior selection of intruders that are highly aggressive to facilitate colony disruption. These paradigms introduce an additional level of complexity since all of the animals involved have experienced defeat, victory, or both. Prior experiences of winning increase the likelihood of aggressive behavior while experiences of defeat decrease it. Thus, prior victory or defeat experience can influence subsequent behavior and reactions to stress and should therefore be considered when using these types of models.

Preexisting conditions may also influence an animal's subsequent social status and/or responsiveness to stress. As an example, in rainbow trout, fish that had low Cortisol responses (low responsiveness) to stress prior to grouping were more likely to be identified as dominant in a subsequent social interaction than fish with high Cortisol (high responsiveness) responses (Pottinger and Carrick, 2001). Rats that are fed with a high-fat diet prior to social defeat show attenuated behavioral and physiological responses to the stressor (Buwalda et al., 2001). It is clear that preexisting conditions and/or prior experiences can alter subsequent behavioral responses and should be considered in designing social stress experiments and interpreting their results.

Representation of ethological conditions

Resident-intruder studies involve brief exposure of the intruder to the resident and are often not representative of natural conditions or the duration of social interactions. In some variations of this model, the intruder is physically exposed to the resident for a brief time and is then protected from attack by a wire mesh cage in order to minimize physical harm presumably without reducing psychological effects. The frequency and predictability of future re-exposure to the resident may be highly variable between laboratories and/or between experiments. As such, the degree of stress/defeat between studies becomes more difficult to equate. Furthermore, the coping strategies in which the intruder may engage will differ depending on the social paradigm and number of exposures used. It is common in laboratory practice to use an animal model to mimic stress-induced changes in physiology or behavior and extrapolate results to what would occur in nature. However, it is also important to consider whether the model is ethologically relevant in regard to severity, duration, and frequency of the stress and/or interaction with conspecifics. Animals may adopt coping strategies that best fit the situation in which they find themselves or to which they are accustomed based on the context of the situation. Indeed, by changing the testing environment, animals will show different behaviors compared to how they would normally respond. In other words, a restrictive environment may cause the animal to select a different coping style other than its preferred one (Koolhaas et al, 1999).

Appropriate controls and comparisons

What are the appropriate controls for social stress studies? The models summarized above utilize a variety of control groups depending on the model employed. Many of the current models compare subordinates with nonstressed controls (Fuchs and Flugge, 2002). Experiments employing the resident-intruder model often use the resident (dominant) animal repeatedly to generate defeated (subordinate) animals and therefore do not include the dominant resident in comparisons to subordinates or non-stressed controls. Several reports indicate that maintaining dominant status is stressful in and of itself, and in some models the dominant animals also exhibit stress-induced alterations in behavior, physiology, brain neurochemistry and neuronal morphology (Blanchard et al, 1995, McKittrick et al, 2000, D'Amato et al, 2001, Tamashiro et al, 2004). Indeed, in some cases, the differences resulting from social stress are greater between dominants and subordinates than those between nonstressed controls and subordinates (McKittrick et al, 1995). It is therefore worth considering whether the dominant or a nonstressed control is the appropriate group to which to compare subordinates. Most humans experience social stress similar to the dominant, subordinate or a combination of the two, therefore, inclusion of the dominant group is important in order to fully examine the effects of social stress at all levels of the social hierarchy.

The visible burrow system model of social hierarchy

The animal model of chronic social stress routinely used in our laboratory is the visible burrow system (VBS) (Blanchard and Blanchard, 1989; Blanchard et al, 1995). The VBS takes advantage of the dominance hierarchy that naturally develops among male rats housed in mixed-gender colonies. Our laboratory has maintained a long-term collaboration with the behavioral neurobiology laboratory at the University of Hawaii headed by Drs. Bob and Caroline Blanchard who developed the VBS model of animal housing, and over the years we have examined the behavioral, physiological and neuroendocrine correlates of chronic social stress in VBS-housed animals. We recently established the VBS model in our laboratory at the University of Cincinnati and have independently validated many of its most salient aspects (Tamashiro et al, 2003, 2004).

In this model, rats are housed in a semi-naturalistic environment resembling an underground burrow system, the natural habitat for rats (Fig. 1). The housing apparatus consists of a large open surface area that is maintained on a 12h:12h light:dark cycle and is connected to a series of tunnels and chambers that is kept in constant darkness and is designed to mimic underground burrow systems. Once a colony is formed, usually by group housing 4 adult males and 2 adult females together in the VBS, a dominance hierarchy quickly develops (usually within a few days) among the males,

Surface area

15-VWitjHbJb 12M2hL:D

1 nthigh bghrt tattle o

AtcuDiet food moflilo'

Fig. 1. Schematic diagram of a visible burrow system (VBS).

producing one dominant and three subordinate male rats. Animals housed in the VBS display behavioral, physiological, and neuroendocrine alterations (Table 1) consistent with severe stress in other models and will be discussed in detail in the remaining sections of this chapter.

As described above, the housing environment in the VBS closely represents the conditions under which rats would live in the wild and serves to facilitate expression of behavioral and physiological responses that would normally occur in that environment. Animals compete for resources such as reproductive partners, rest areas, food and water. Food and water are provided throughout the apparatus in three areas (open surface area and two smaller chambers), but animals must still gain access to the food hoppers and water bottles that may be guarded by one or more of the other 5 animals of the colony. Provisions of food and water in the burrows is a relatively new modification to the VBS protocol and results in decreased wounding in subordinates, but has no effect on other stress-induced changes typically present after VBS housing including body weight loss (Blanchard et al., 2001a). While this may suggest that there are less aggressive encounters between the dominant and subordinates, it also decreases physical stress to the animals that may be associated with higher wounding.

Another feature of the model is that all animals used in VBS studies are naive, and this precludes the necessity to consider the influence of previous experience on behavior and hierarchy formation. It is important to keep in mind that it is possible that preexisting conditions or characteristics could predispose some animals to become dominant or subordinate. However, thus far we have not found any reliable behavioral or endocrine measures that can serve as predictors of social status of male rats prior to colony formation in this model of social stress.

An important consideration that has developed over the years is determining key control groups for appropriate comparisons. The VBS model includes two groups of control males. The first are control males that are single-housed with a female rat and the second are males that are single-housed and food restricted such that their body weights match those of subordinate males in each colony. This second, weight-matched control group is particularly important since it identifies whether any of the physiological and/or neurochemical changes in VBS subordinates may simply be due to differences in body weight and not stress per se. The dominant animals are also used as a comparison group in our studies. Results from the VBS model support the idea that the dominant male also experiences some indices of stress such as elevated basal glucocorticoid secretion, though these stress-induced consequences are independent of body weight and are milder compared to subordinate males.

The VBS is a unique model compared to other stress models since the stress imposed is derived from the natural conspecific interactions between the animals themselves and is devoid of excessive experimental intervention. Certainly this does not imply that the resident-intruder or social defeat models that do involve some degree of intervention by the investigator are not applicable to the study of social stress. Rather, those models present specific situations of defeat at predetermined times arranged by the investigator, and as such the resultant behavioral and physiological responses are specific to social defeat in a contextual and temporal specific manner. In contrast, the stress imposed in the VBS is unpredictable and is based on the behavioral intensity of the social interactions of the males competing for access to the females in the colony.

Table 1. Summary of behavioral, physiological, endocrine and neurochemical consequences of chronic social stress in the VBS.

Measure Effect Reference


Table 1. Summary of behavioral, physiological, endocrine and neurochemical consequences of chronic social stress in the VBS.




Blanchard et al, 1995



Blanchard et al, 1995



Blanchard et al, 1995


Was this article helpful?

+1 0

Post a comment