Animal Models of Anxiety and Anxiolytic Drug Action
Animal models of anxiety attempt to represent some aspect of the etiology, symptomatology, or treatment of human anxiety disorders, in order to facilitate their scientific study. Within this context, animal models of anxiolytic drug action can be viewed as treatment models relevant to the pharmacological control of human anxiety. A major purpose of these models is to identify novel anxiolytic compounds and to study the mechanisms whereby these compounds produce their anxiolytic effects. After a critical analysis of “face,” “construct,” and “predictive” validity, the biological context in which animal models of anxiety are to be evaluated is specified. We then review the models in terms of their general pharmacological profiles, with particular attention to their sensitivity to 5-HT1A agonists and antidepressant compounds. Although there are important exceptions, most of these models are sensitive to one or perhaps two classes of anxiolytic compounds, limiting their pharmacological generality somewhat, but allowing in depth analysis of individual mechanisms of anxiolytic drug action (e.g., GABAA agonism). We end with a discussion of possible sources of variability between models in response to 5-HT1A agonists and antidepressant drugs.
KeywordsAnxiety Animal models Face validity Construct validity Biological validity Anxiolytic Antidepressant Drug receptors
1 Animal Models of Anxiety and Anxiolytic Drug Action: Introduction
The validation and development of animal models of anxiety (e.g., the elevated plus-maze) has relied mainly on the putative models’ sensitivity to well-established anxiolytic drugs, such as the benzodiazepines (e.g., diazepam). It is important to emphasize, however, that the purpose of this work was to establish the predictive validity of the animal model with respect to known anxiolytic compounds, and not to establish the clinical utility of indirect GABAA receptor agonists such as diazepam. A pharmacologically valid model also had to discriminate between “anxiolytic” compounds (e.g., diazepam) and “nonanxiolytic” compounds, such as imipramine, an antidepressant drug. For example, from the late 1960s to the late 1980s, if an antidepressant drug registered the same as diazepam in an animal model of anxiety, it was designated a “false positive.” The more “false positives” a putative model accumulated, the less “valid” it became in terms of predictive validity, and the less useful as a potential “screening test” for finding novel anxiolytic compounds (Treit 1985a, b).
Unfortunately, for animal models of anxiety the pharmacological treatment of human anxiety disorders began to change, almost imperceptibly, after an early report of the successful treatment of panic disorder with imipramine (Klein 1964). The treatment of anxiety disorders such as panic with antidepressant drugs steadily increased through the 1990s (e.g., Tyrer and Tyrer 1994), and by the new millennium they had emerged as the frontline pharmacological treatment for all human anxiety disorders (e.g., Hoffman and Mathew 2008; Borsini et al. 2002). Animal models of anxiety had been validated mainly on their differential sensitivity to benzodiazepine anxiolytics, and were, not surprisingly, insensitive to the anxiolytic effects of antidepressant drugs. More often than not, these animal models completely failed to detect antidepressant drugs, or registered them as “anxiogenic” (Treit et al. 2003; Borsini et al. 2002).
This deficiency, and the unpredictable effects in these models of 5-HT1A anxiolytics such as buspirone before it, has raised serious questions about the “validity” of so-called animal models of anxiety (e.g., Treit et al. 2003). Apart from the general failure of researchers to study the chronic effects of these compounds, which are clinically more relevant than their acute effects, we do not believe that the problem is specific to antidepressant drugs, for the following reason. Imagine for a moment that the history of therapeutic drug discovery had been reversed, because the focus of clinical interest at the time had differed slightly. We know, for example, that clinicians in the early 1950s had noticed that the anti-tuberculosis drug iproniazid had a peculiar side-effect in patients: it elevated the patient’s mood, sometimes to the point of euphoria (Pletscher 1991). Imagine now that what impressed these clinicians even more was another “side-effect” that gradually emerged after chronic treatment with iproniazid: patients became progressively less anxious about their condition, and more tractable than patients treated with other anti-tuberculosis drugs. If the second (perhaps fictitious) observation had attracted more clinical attention than “mood elevation,” iproniazid and its congeners might have been studied in psychiatric patients suffering from anxiety, and eventually have become the first, clinically effective “anxiolytic” drugs. Other drugs would have been assessed for their “anxiolytic” potential, and some, such as imipramine, would emerge serendipitously. In any case, a clinical precedent would have been set. In addition, the careful study of the effects of these drugs on brain function would lead to the following generalization: compounds that possessed “anxiolytic” effects in patients invariably elevated brain monoamines. After repeated demonstrations of this common effect, the “monoamine hypothesis of anxiety” would emerge and become dominant. As this “monoamine hypothesis” was tested further, at both behavioral and neural levels, the importance of reliable “animal models of anxiety” would be reinforced. Their “predictive validity” would depend on their relative sensitivity to the effects of chronic iproniazid and other “standard” anxiolytic drugs (e.g., imipramine), and their relative insensitivity to “nonanxiolytic” drugs. At some point along this not entirely fanciful path to drug discovery, the predictive validity of these early animal models of anxiety would be challenged by a little known, indirect GABAA receptor agonist called diazepam, which registered as a “false positive” in the models, but was eventually shown to be anxiolytic in clinical populations.
The point of this modest revision of history, of course, is not to disparage our inability to see into the future, but to illustrate the fundamental weakness of the pharmacological “validation” of animal models of anxiety: it is always constrained by historical accident, and is completely post hoc. If and only if the test drug has been validated clinically can it be used to evaluate the “predictive” validity of a particular animal model of anxiety. A novel compound that has anxiolytic-like effects in the model (e.g., diazepam in the above scenario) would initially be viewed as a “false positive” and would detract from the predictive validity of the model – at least, until overwhelming clinical evidence suggested otherwise.
In summary, what counts as a “false positive” for an animal model of anxiety is time-sensitive and subject to clinical revision. “False negatives” are not subject to these historical vagaries, but in their case, attempts must be made to show that the model can, with appropriate tweaking, be made sensitive to the previously unrecognized classes of anxiolytic drug (e.g., antidepressants). In an absolute sense, then, no animal model of “anxiety” can be summarily invalidated simply on the basis of a “false positive” or a “false negative.”
As the notion of pharmacological or “predictive” validity becomes more ambiguous, other criteria such as “face” validity and “construct” validity become more attractive. At the very least, these aspects of validity may offer some distraction from the uncomfortable circularity of predictive validity (for original guidelines, definitions, and applications of “face” and “construct” validity in the social sciences, see Cronbach and Meehl 1955; for recent quantitative approaches see Westen and Rosenthal 2003). “Face” validity, however, means only that the test or model appears to measure anxiety, not that it actually measures anxiety. Face validity, in this light, is far from an unambiguous substitute for predictive validity. Predictive validity is at least based on the quantifiable behavior of experimental animals under the influence of a chemically defined set of molecules, not just the opinion of a set of observers. It seems curious, then, that “face validity,” as used in the experimental literature, often implies something more concrete than a common “understanding” among observers, expert, or otherwise (e.g., Bourin et al. 2007).
If face validity is a weak criterion, “construct validity” is a strong criterion (Cronbach and Meehl 1955), mainly because the latter is central to all other “validity” issues (e.g., Westen and Rosenthal 2003). Construct validity specifically addresses the question: “does the test or model actually measure or represent what it was intended to measure or represent?” This question rarely can be answered by experimental data alone, and must be evaluated in the context of a scientific theory of the construct in question. For animal models of anxiety, however, this amounts to an almost insurmountable task for three reasons. First, there is no widely accepted or established scientific theory of “anxiety” in humans. Second, there is no unambiguous translation of such a theory – if it existed – to animal models of anxiety.1 Third, even if a tentative translation of this hypothetical theory existed, our empirical understanding of anxiety does not yet have the precision needed for a scientific theory, at least in comparison with other areas of science in which the term “theory” is used (e.g., cell theory).
Returning to the issue of “correlational” validity and the sensitivity of animal models of anxiety to one particular subclass of anxiolytic compounds (e.g., benzodiazepines)—pharmacological validity can be bolstered by documenting behavioral correlates such as increased vocalization or thigmotaxis (e.g., Treit and Fundytus 1988), and physiological correlates such as increased serum corticosteroids or brain norepinephrine release (e.g., Pellow et al. 1985; Bondi et al. 2007; De Boer et al. 1990). Furthermore, selective sensitivity to only one class of anxiolytic compounds (e.g., benzodiazepines) may in fact be advantageous for exploring the molecular mechanisms that underlie the anxiolytic effects of a particular drug class. If there are significant variations in drug response within these “one drug-class” animal models, they may reflect the degree to which the model engages particular receptors underlying the drugs’ effects. For example, there may be differences in drug response that reflect differences in the expression of particular receptor subunits (α β γ δ), or differences in the specific subunit combinations of a given receptor (e.g., Kaufmann et al. 2003). Conversely, receptors and their subunits can be directly manipulated using genetic (e.g., “knock-out” or “knock-in”) techniques, and the behavioral effects measured in animal models of anxiety (e.g., Finn et al. 2003; Low et al. 2000; McKernan et al. 2000; Marowsky et al. 2004; Kaufmann et al. 2003; Cryan and Holmes 2005). Working in parallel, biochemists can rapidly synthesize hundreds of structural analogues of known anxiolytic compounds2, based on their relative affinity for particular receptor subtypes or receptor subunits (e.g., Caliendo et al. 2005). With these points in mind, our general approach here is to provisionally accept the utility of pharmacologically valid animal “screening” tests of anxiolytic drugs (i.e., purely correlational models), whether or not they are sensitive to antidepressant drugs. However, the overall relevance of pharmacologically valid models to “anxiety” must be assessed in the light of three biological imperatives, which together form a heuristic framework for evaluating their specific “validity” as animal models of anxiety.
First, there should be some formal correspondence between the behavioral expressions of fear in the animal model (e.g., avoidance), the physiological correlates of these behaviors (e.g., increased corticosterone), and their expressions in humans (i.e., isomorphism). Second, in the absence of clear isomorphism, there should be some continuity of biological function between behavioral responses in the animal model and fear responses in humans (e.g., defense). Third, and specifically at the neural level, there should be considerable conservation of the underlying brain mechanisms of fear and anxiety across mammalian species (either by homology or analogy), and these should be engaged during specific fear reactions. Together, these biological constraints are dictated by the fundamental role that fears or anxieties normally play in the survival of organisms in their natural habitats. There is no reason to believe that a model that does not conform to any of these biological imperatives has any relationship to anxiety. For example, PTZ-induced convulsions in mice are exquisitely sensitive to suppression by benzodiazepine anxiolytics, but the convulsive response itself has little or no isomorphism with fear or anxiety, is functionally unrelated to defense, and is unlikely to reflect the same brain mechanisms as those evolutionarily conserved for survival. As such, the PTZ test is not a plausible “animal model of anxiety,” despite its pharmacological validity as a “test” or predictor of anxiolytic drug action (Treit 1985a, b). At the same time, there is no reason why the PTZ “test” could not serve as a valuable “model” of epileptic convulsions, their neurochemical mechanisms and pharmacological treatments (e.g., Luszczki 2008). In some sense, then, “tests” and “models” can only be distinguished by their use in specific experimental contexts (Kalueff and Murphy 2007).
In summary, it is from this overarching biological perspective that we now review the validity of current animal models of anxiety and anxiolytic drug action.
1.1 Conflict Tests
In the “Geller” and the “Vogel” conflict tests, a food or water deprived rat is punished with electric shock if it makes a response (e.g., bar-pressing, licking) that other wise results in reward (e.g., food, water). Punishment inhibits rewarded responding, and standard anxiolytic drugs such as diazepam selectively disinhibit punished responding. In so far as conflict between opposing drives may be indicated in human anxiety, “conflict” models may also have some degree of isomorphism and homology (Treit 1985a, b).
There are a number of procedural variations that add to the complexity of conflict tests, and thus affect the ease with which anxiolytic compounds can be assessed (Treit 1985a, b). For example, most versions of the standard Geller conflict test take weeks of training before drug testing can begin, whereas the Vogel conflict procedures take little or no pretraining. The advantage of the more complex procedures, however, is that control measures for nonspecific drug effects are embedded into the test (e.g., periods of nonpunished responding). Use of less complex conflict procedures, however, has gradually become more prevalent over the past 20 years. For example, our ISI Web of Science search of research papers published between 1988 and 2008 showed that the Vogel test is cited nearly five times more than the Geller test. This trend has prevailed mainly because the Vogel test (1) is procedurally more practical, (2) produces results that differ little from the Geller test, and (3) allows immediate and rapid testing of target compounds. Since there is little to distinguish the two tests other than experimental expedience, the results from each are combined below.
Summary of findings with Vogel-type and Geller-type conflict tests following peripheral administration of different classes of anxiolytic drugs
Ellis et al. (1990), Giusti et al. (1991), Hascoet and Bourin (1997), Kennett et al. (1998), Rudzik et al. (1973), Soderpalm et al. (1989), Kapus et al. (2008), Mathiasen et al. (2008), Gleason and Witkin (2007), Wesolowska and Nikiforuk (2007), Popik et al. (2006), Moreira et al. (2006), Tatarczynska et al. (2004), Ballard et al. (2005), Busse et al. (2004)
Dekeyne et al. (2000), Deren-Wesolek et al. (1998), McCloskey et al. (1987), Meneses and Hong (1993), Pich and Samanin (1986), Sanger (1992), Schefke et al. (1989), Weissman et al. (1984), Yamashita et al. (1995), Young et al. (1987), Stefanski et al. (1992), Bojarski et al. (2006), Vaidya et al. (2005), Jurczyki et al. (2004), Liao et al. (2003)
Chaki et al. (2005)
Commissaris and Hill (1995), Fontana and Commissaris (1988), Fontana et al. (1989), Vigliecca et al. (2007)
Fontana et al. (1989), Commissaris et al. (1995)
Costello et al. (1991)
Costello et al. (1991), Gardner (1986), Kennett et al. (1998), Sanger (1990), Witkin and Perez (1990)
In contrast to their reliable sensitivity to the anxiolytic effects of benzodiazepines and antidepressant drugs, the sensitivity of the conflict tests to the anxiolytic effects of 5-HT1A drugs such as buspirone has been erratic (Howard and Pollard 1990). Furthermore, when anxiolytic effects of 5-HT1A drugs are detected, they are often small, and occur over a much narrower dose range (e.g., Dekeyne et al. 2000; Deren-Wesolek et al. 1998; Sanger 1992; Schefke et al. 1989). Howard and Pollard (1990) studied the effects of buspirone in the conflict test under a wide variety of experimental conditions and failed to find a robust anxiolytic effect under any condition. Other researchers have reported no significant effects of buspirone in the conflict tests, or even dose-related decreases in punished responses (e.g., Brocco et al. 1990; Costello et al. 1991; Gardner 1986; Vaidya et al. 2005; Witkin and Perez 1990; but see Bojarski et al. 2006).
Despite their questionable sensitivity to 5-HT1A compounds, conflict tests have broad utility for detecting the anxiolytic effects of benzodiazepine and antidepressant drugs. The exact conditions under which reliable anxiolytic effects of 5-HT1A compounds can be detected in these tests remain to be determined.
1.2 Fear Potentiated Startle
The magnitude of rats’ naturally-occurring startle reflex to an acoustic stimulus is “potentiated” when the acoustic stimulus is presented together with a cue (e.g., light) that has previously been paired with shock (Brown et al. 1951; Chi 1965; Davis 1986a, b; Richardson et al. 1999). The fact that the startle response has a nonzero baseline makes it possible to distinguish the effects of a drug on base-line startle (acoustic stimulus alone) from its effects on “potentiated startle” (acoustic stimulus plus conditioned fear cue). Insofar as human fears and phobias can be acquired through experiences analogous to “fear conditioning” in rats, and insofar as conditioned fear stimuli (e.g., light) can “potentiate” unconditioned responses such as startle, the model appears to have some degree of isomorphism and homology with human anxiety. It should be noted, however, that Seligman (1970) and others have made persuasive arguments that conditioned fears to artificial stimuli such as lights and tones may have little relevance to the etiology of human fears and phobias.
Summary of findings with the fear-potentiated startle test following peripheral administration of different classes of anxiolytic drugs
Berg and Davis (1984), Brodkin et al. (2002), Davis (1979), Hijzen and Slangen (1989), Joordens et al. (1996), Martin et al. (2002), Pietraszek et al. (2005), Risbrough et al. (2003), Santos et al. (2005), Schulz et al. (2001), Serradeil-Le Gal et al. (2005), Tizzano et al. (2002), Winslow et al. (2007)
Davis et al. (1985), Joordens et al. (1998), Mansbach and Geyer (1988), Nevins and Anthony (1994), Paschall and Davis (2002), Risbrough et al. (2003)
Brodkin et al. (2002), Winslow et al. (2007)
Joordens et al. (1996)
Casella and Davis (1985), Hijzen et al. (1995)
Antidepressant compounds such as TCAs and SSRIs, however, have been ineffective in the fear-potentiated startle test (Casella and Davis 1985; Hijzen et al. 1995; Joordens et al. 1996). Of even more concern is the finding that yohimbine and FG-7142, drugs that have been shown to be anxiogenic in humans, also suppress “fear” potentiated startle (Risbrough and Geyer 2005). Until this paradoxical result can be resolved, the pharmacological validity of this model is questionable. Nevertheless, fear-potentiated startle, and fear conditioning paradigms in general have been extremely valuable behavioral models for dissecting the role of amygdala in fear and anxiety (e.g., Lang et al. 2000; LeDoux 2000).
On the other hand, the fear-conditioning models have difficulty differentiating drug or lesion effects on learning and memory from specific effects on anxiety responses (see Engin and Treit 2008 for a discussion). A variant of the fear-potentiated startle model called the light-enhanced startle has been developed that appears to overcome this deficiency (Walker and Davis 1997). In this version of the model, rats show a potentiated startle response in a brightly illuminated environment, compared to a dark environment, and this “unconditioned” potentiation is used as a measure of fear. Light-enhanced startle was found to be sensitive to the effects of benzodiazepines and 5-HT1A agonists (de Jongh et al. 2002; Walker and Davis 2002), but SSRIs were ineffective, as in the standard fear-potentiated startle model (de Jongh et al. 2002).
Commissaris et al. (2004) have argued that another problem with the original fear-potentiated startle model is that the testing occurs in the extinction phase (i.e., the light is no longer paired with the shock). It follows that the model is not suitable for repeated testing of a drug, as the fear response itself is diminishing as a function of extinction test trials. Consequently, they have proposed a “startle-potentiated” startle paradigm, where the shock is removed from the classical model and noise-only versus light+noise groups are subjected to the test repeatedly. Although the two groups did not differ in startle amplitude initially, repeated trials led to an increase in the startle amplitude of the light+noise group, supposedly through the anticipation of the noise stimulus. The startle amplitude for the light+noise group in subsequent testing was reduced by buspirone or mixed anxiolytic-antidepressant alprazolam. However, it should be noted that these drugs seemed to influence the baseline (i.e., noise only) startle in this version of the test. Thus, further pharmacological characterization of the “startle-potentiated” startle test is required before it can be employed as a model of anxiolytic drug action.
Overall, the startle models are broadly sensitive to benzodiazepine and 5-HT1A-type anxiolytics, and have the advantage of providing a baseline acoustic reflex to which the fear-potentiated reflex can be compared. This simplifies the problem of distinguishing drug effects on anxiety from nonspecific drug effects on the baseline startle reflex itself. However, antidepressants, which have clinically-proven anti-anxiety effects, fail to register in any version of the model, thus restricting its generality as a test of anxiolytic compounds.
1.3 Defensive Burying
The defensive burying model is based on a species-typical response of rodents confronted with a nociceptive or predator-related stimulus (for reviews see Treit 1985a, b; De Boer and Koolhaas 2003; Treit and Pinel 2005; Lapiz-Bluhm et al. 2008). Rodents spray substrate materials (e.g., sand) toward the threat stimulus (e.g., snake) with rapid, alternating movements of the forepaws, while avoiding direct contact (i.e., “burying behavior”; Pinel and Treit 1978). Burying can be induced in the laboratory with an electrified shock-probe, which protrudes through one of the walls of a Plexiglas chamber. Rats shocked once from the stationary probe stereotypically spray bedding material on the floor of the chamber toward and over the probe, while avoiding further contact with the probe (e.g., Pinel and Treit 1978; Treit 1985a, b; Treit et al. 1994; Echevarria et al. 2005; Bondi et al. 2007; Engin and Treit 2008). The amount of time spent burying the shock probe is taken as the primary index of anxiety, and the number of contacts with the probe is a secondary index (Treit 1990; Treit et al. 1981). A genuine “antianxiety” effect in this model is indicated by decreased burying behavior and/or increased probe contacts, independent of changes in pain sensitivity or general activity levels (Treit et al. 1990).
Plasma levels of corticosteroid and catecholamines are increased during the burying test, and these increases can be suppressed by standard anxiolytic drugs (de Boer et al. 1990). Concomitant changes in these “stress” hormones in response to shock and anxiolytic drugs add to the correlational validity of the model. More recent studies have corroborated these findings, documenting increases in heart rate, blood pressure, catecholamines, prolactin, ACTH, and corticosteroids in response to the shock-probe (Lapiz-Bluhm et al. 2008). Interestingly, burying behavior is also sensitive to the delayed anxiogenic effects of repeated, unpredictable stress, similar to the delayed effects of stress in human anxiety disorders (e.g., post-traumatic stress disorder; Matuszewich et al. 2007). In addition to these stress correlates, burying behavior also seems to have biological validity insofar as it represents an evolved adaptation of rodents to natural threat stimuli (homology). Although it has limited isomorphism with human anxiety responses – or to avoidance responses in general – the burying response may have functional relevance insofar as it can block or neutralize dangerous or threatening stimuli. Finally, many of the defining features of anxiety in humans (sweating, dizziness, tachycardia, trembling) may only emerge full-blown when approach is required to a threatening stimulus or situation (McNaughton and Corr 2004).
Summary of findings with the defensive burying test following peripheral administration of different classes of anxiolytic drugs
Beardslee et al. (1990), Blampied and Kirk (1983), Degroot and Nomikos (2004), Fernandez-Guasti and Martinez-Mota (2003), Fernandez-Guasti et al. (2001), Gomez et al. (2002), Perrine et al. (2006), Picazo et al. (2006), Rohmer et al. (1990), Sikiric et al. (2001), Treit (1985, 1987, 1990), Treit and Fundytus (1988), Treit et al. (1981), Tsuda et al. (1988), Violle et al. (2006), Wilson et al. (2004)
Fernandez-Guasti and Lopez-Rubalcava (1998), Fernandez-Guasti and Picazo (1990, 1997), Fernandez-Guasti et al. (1992), Groenink et al. (1995, 1997), Korte and Bohus (1990), Lopez-Rubalcava (1996), Lopez-Rubalcava et al. (1996, 1999), Picazo et al. (2006), Treit et al. (2001), Treit and Fundytus (1988)
Degroot and Nomikos (2004), Treit et al. (2001)
Bondi et al. (2007), Fernandez-Guasti et al. (1999), Martinez-Mota et al. (2000)
Craft et al. (1988)
Beardslee et al. (1990)
A variant of the defensive burying model – “marble burying” in mice – was developed by Broekkamp et al. (1986) and Njung’e and Handley (1991a, b). In this test, 20 glass marbles are distributed evenly on the bedding material that covers the floor of a plexiglass chamber. The mice are placed individually in the chamber and are taken out 30 min later. The number of marbles “buried” is used as an index of anxiety, although the actual “burying behavior” of mice is not measured in this version of the test. A large number of classical anxiolytics, such as diazepam, chlordiazepoxide, alprazolam, clonazepam, and flunitrazepam, have been shown to reduce marble burying in mice (Broekkamp et al. 1986; Li et al. 2006; Young et al. 2006; Nicolas et al. 2006). Although 5-HT1A agonists also reduce marble burying (Young et al. 2006; Nicolas et al. 2006; Njung’e and Handley 1991a, b), the effect was observed only at high doses that tend to reduce activity levels in general (Li et al. 2006; Nicolas et al. 2006).
Both tricyclics and SSRIs have been shown to reduce marble burying (Broekkamp et al. 1986; Li et al. 2006; Njung’e and Handley 1991a, b; Ichimaru et al. 1995; Nicolas et al. 2006; Harasawa et al. 2006). However, the suppressive effect of tricyclic antidepressants on marble burying is not always robust (e.g., Ichimaru et al. 1995; Nicolas et al. 2006), and in studies where the effect is clear, the reduction in burying is produced at doses that also suppress general locomotor activity (e.g., Broekkamp et al. 1986). Thus, while the SSRIs have a relatively selective inhibitory effect on marble burying, the effect of tricyclics is less clear. To our knowledge, only one study (Nicolas et al. 2006) reported the effects of MAOIs, and in this case phenelzine reduced marble burying.
Overall, marble burying in mice has been fairly well validated as a correlational model of anxiolytic drug action. However, there are several nonpharmacological issues that detract from its use as an animal model of anxiety. The main problem is that the marbles used in the test may not in fact be anxiogenic. The number of marbles buried is strongly correlated with digging and burrowing behavior, which mice display in the absence of marbles, or any other anxiogenic stimulus; this behavior is nondirected but can coincidentally cover glass marbles. On the basis of these careful observations, Gyertyan (1995) concluded that mice are not “burying” an aversive or threatening stimulus, merely engaging in a species-typical response elicited by a movable substrate. In line with this, Costa et al. (2006) have reported that mice classified into two groups according to their performance in the marble-burying test (i.e., low-burying, high-burying) did not differ on any measure of anxiety in three other models (i.e., elevated plus-maze, light–dark, hole board).
Njung’e and Handley (1991a, b) suggested that marble burying is a useful screening test for the detection of anxiolytic drugs, even if it did not represent an isomorphic fear response. As a screening test, however, marble burying is less than ideal, since nonanxiolytic and anxiogenic compounds alike can produce “false positives” (Nicolas et al. 2006; Broekkamp et al. 1986). The suggestion that marble burying in mice is more appropriately viewed as a specific model of obsessive-compulsive disorder (Gyertyan 1995; Millan et al. 2002) is interesting, but seems to push the problem of an undefined anxiogenic stimulus beyond the limits of observation, and into the “obsessive thoughts” that presumably drive the mice’s “compulsive” digging! This notion is not only fanciful; it cannot be falsified and therefore is not subject to scientific test. Furthermore, more recent research has shown that clomipramine, an “antiobsessional” compound, reduces marble-burying behavior only at doses that also reduce general activity (Nicolas et al. 2006; Li et al. 2006). Thus, the pharmacological evidence for marble burying as a model of obsessive-compulsive disorder is less than compelling.
1.4 Light–Dark Exploration
In the light–dark exploration test, rodents avoid the brightly lit side of a two-compartment chamber, spending most of their time exploring the dimly lit side. Anxiety reduction in this test is indicated by increased transitions between the two compartments and/or increased exploration (i.e., time spent and number of line crossings) in the bright compartment, whereas nonspecific effects are indicated by changes in general locomotor activity (Blumstein and Crawley 1983; for a recent review, see Bourin and Hascoet 2003). The test takes advantage of rodents’ natural aversion to brightly illuminated spaces, which itself may represent an adaptive defensive response against daytime predation. Nevertheless, photophobia is not given as an example of a “specific phobia” in the DSM-IV-revised edition, and in humans it is normally related to other medical conditions. As a defensive against predation, however, it may have some homology with other human fears and phobias (e.g., spiders, snakes).
Summary of findings with light–dark exploration test following peripheral administration of different classes of anxiolytic drugs
Both et al. (2005), Costall et al. (1987), Costall et al. (1988a, b, 1989), Costanzo et al. (2002), Crawley (1981), Crawley and Goodwin (1980), de Angelis (1992), Kilfoil et al. (1989), Mi et al. (2005), Shimada et al. (1995), Uriguen et al. (2004), Zanoli et al. (2002)
Bill et al. (1989), Costall et al. (1989), Crawley (1981), Hascoet and Bourin (1988), Imaizumi et al. (1994a, b), Lopez-Rubalcava et al. (1992), Onaivi and Martin (1989), Uriguen et al. (2004), Young and Johnson (1991a, b)
Hascoet et al. (2000b)
Bourin et al. (1996), Shimada et al. (1995), Young and Johnson (1991a, b)
Bourin et al. (1996), De Angelis and Furlan (2000)
Griebel et al. (1994)
Costall et al. (1989)
Crawley (1981), De Angelis and Furlan (2000)
Bill et al. (1989)
Brourin et al. (1996)
Onaivi and Martin (1989)
1.5 Social Interaction
In the social interaction test, naïve rats are placed in pairs in an open arena, and the time they spend in active social interaction (e.g., sniffing, grooming) is measured. Social interaction is suppressed when animals are tested under bright lights or in an unfamiliar test environment, relative to low light/familiar conditions. This suppression is the index of anxiety (File and Hyde 1978). Line crossings are counted as a measure of nonspecific changes in locomotor activity. Disinhibition of social interaction is the measure of anxiety-reduction in this test (for a review see File 2003). In so far as anxious humans display “social phobia,” the model may have some degree of isomorphism. Avoidance of social conflict is in some way related, and may be under positive selection pressure in many social species, from rats to humans.
Table 4 shows that benzodiazepine and 5-HT1A anxiolytics were generally effective in increasing social interaction, whereas antidepressants had varying effects. The only exception to this general pattern of drug effects is a study by Rex et al. (2004), where diazepam was anxiolytic in Wistar but not Sprague Dawley rats. Since many other studies have administered benzodiazepines to Sprague Dawley rats and produced anxiolytic effects in the social interaction test (e.g., Kantor et al. 2005), Rex et al.’s failure to do so is suspect. SSRIs have been uniformly ineffective or anxiogenic following acute administration in the social interaction test, but anxiolytic when administered chronically for 21 days before the test. Tricyclic antidepressants did not produce anxiolytic effects in this test, even following chronic administration (e.g., Popik and Vetulani 1993). Nonanxiolytic drugs such as amphetamine, caffeine, yohimbine, or naxolone did not produce anxiolytic-like effects in the social interaction test (File and Hyde 1979; Pellow et al. 1985; File 1980).
In addition to lighting levels and familiarity of the environment, the novelty of the testing partner can act as an anxiety-inducing parameter (Gardner and Guy 1984; Guy and Gardner 1985). Gardner and Guy (1984) reported that the suppression of social interaction when faced with a novel rather than a familiar partner was reversed by benzodiazepines and the mixed anxiolytic–antidepressant drug alprazolam, whereas nonanxiolytic agents did not affect this measure. Another more recent variant, based on the general idea that anxiogenic stimuli reduce social interaction, is the “stress-induced social avoidance” test (Haller et al. 2002; Haller et al. 2003). In this test, rats subjected to electric shocks or conspecific aggression avoid social contact for up to 10 days. Initial validation studies indicate that the test may be sensitive to the effects of benzodiazepines, 5-HT1A agonists, and SSRIs (Leveleki et al. 2006). These are promising developments and suggest that some variant of the social interaction test may yet be found sensitive to all major classes of anxiolytic drug.
1.6 Elevated Plus-Maze
In the elevated plus-maze, rodents normally avoid the two open arms of the maze, and restrict most of their activity to the two closed arms. Open-arm avoidance appears to be driven by an aversion to open spaces, leading to thigmotaxic behavior (Treit et al. 1993). An antianxiety effect is indicated by an increase in the proportion of activity in the open arms of the maze (i.e., an increase in the percentage of time spent in the open arms and in the percentage of entries into the open arms). Changes in total entries and/or changes in the number of closed arm entries indicate nonspecific drug effects on locomotor activity. (For reviews of procedures and methods see Pellow 1986; Hogg 1996; Treit et al. 2003; Carobrez and Bertoglio 2005.) Other ethologically driven behaviors such as “risk assessment” (i.e., the “stretched attend” posture) have also been measured in the elevated plus-maze to complement the original spatial measures of anxiety (e.g., Rodgers and Dalvi 1997), although their current use is not widespread. Anxious humans can also display fear of heights and open spaces, and may even display thigmotaxis and risk assessment under these conditions, suggesting some degree of isomorphism or homology. In smaller animals, avoidance of open spaces may have evolved as a defense against larger mammals and/or avian predators (Treit and Fundytus 1988).
Benzodiazepine anxiolytics increase the proportion of activity in the open arms, whereas nonanxiolytic agents (e.g., amphetamine, caffeine) generally do not (Baldwin et al. 1989; Handley and Mithani 1984; Johnston and File 1989; Pellow et al. 1985). Mixed anxiolytic-antidepressant compounds such as alprazolam also have reliable anxiolytic effects in the elevated plus-maze (Griebel et al. 1996; Johnston and File 1988; Jones et al. 1994; Pellow and File 1985). However, the effects of standard antidepressant drugs in the plus-maze have been inconsistent. On the one hand, the tetracyclic antidepressant mianserin produced significant anxiolytic effects after chronic administration (Rocha et al. 1994), and MAOIs such as phenelzine and befloxatone produced anxiolytic effects whether given acutely (Caille et al. 1996; Paslawski et al. 1996) or chronically (Johnston and File 1988). On the other hand, both acute and chronic administration of TCAs (imipramine, amitriptyline) failed to produce anxiolytic profiles in the plus-maze (e.g., Cole and Rodgers 1995; Lister 1987; Luscombe et al. 1990), and SSRIs such as fluoxetine have been reported to be anxiogenic (e.g., Handley and McBlane 1992; Silva and Brandao 2000; Silva et al. 1999), anxiolytic (e.g., Cadogan et al. 1992; Kurt et al. 2000), or ineffective (e.g., Linnoila et al. 1987). The inconsistent effects of SSRIs have been found after both acute (e.g., Griebel et al. 1999; Silva and Brandao 2000) and chronic (Kurt et al. 2000; Silva et al. 1999) drug administrations.
The effects 5-HT1A-type compounds in the elevated plus-maze are also mixed. There are reports of clear anxiolytic effects (e.g., Dunn et al. 1989; Griebel et al. 1997; Hallar et al. 2000), or no anxiolytic effects (e.g., Pellow and File 1986; Pellow et al. 1987; Silva and Brandao 2000), even after chronic drug administration (e.g., Moser 1989; Moser et al. 1990). Although chronic regimens with buspirone or ipsapirone did not produce anxiolytic effects in the plus-maze (e.g., Moser 1989), there is some evidence that these negative findings may have been related to dose. Soderpalm et al. (1993) found that 5 weeks of buspirone (10 mg kg−1 b.i.d.) significantly increased open-arm activity whereas the same regimen at lower doses (2.5 or 5.0 mg kg−1) was without effect. A number of other studies support the hypothesis that high doses of 5-HT1A compounds may be necessary for their anxiolytic effects to emerge after chronic treatment in the elevated plus-maze (Cole and Rodgers 1994; Maisonnette et al. 1993; Motta et al. 1991; Silva and Brandao 2000).
In summary, the elevated plus-maze is clearly sensitive to benzodiazepine-type anxiolytics. However, the effects of antidepressant drugs (both chronic and acute) are mixed, as are the effects of 5-HT1A compounds. There is some evidence that high doses of chronically administered 5-HT1A compounds may be necessary to detect their anxiolytic effects in the elevated plus-maze.
A recent modification of the original-plus maze test is the “stress-potentiated plus-maze” test, in which subjects are prestressed before exposure to the standard elevated plus-maze test (Korte and DeBoer 2003). The addition of prestress (e.g., restraint, social defeat, electric shock) seems reminiscent of the fear-potentiated startle paradigm, although its motivation seems to have been to increase the sensitivity of the test to the anxiolytic effects of experimental compounds. For example, demonstrating the putative anxiolytic effects of CRF receptor antagonists in models such as the standard plus-maze has proven difficult (Martins et al. 2000). By adding prestress, it was possible to shown that intra-PAG CRF receptor antagonists can block fear behavior in the elevated plus-maze, whereas this was not possible in the standard test (Martins et al. 2000). Nevertheless, the precise role of “pretest stress” in the anxiolytic effects of CRF antagonists remains to be determined. This is especially true given the fact that the sensitivity (or lack thereof) of the “fear-potentiated” plus-maze test to the anxiolytic effects of standard anxiolytic compounds such as diazepam has not yet been demonstrated. Furthermore, “prestress” effects seem to occur only when baseline levels of open-arm activity (the index of fear) are particularly high, indicating low anxiety (for examples, see Korte and DeBoer 2003). If these problems can be rectified or rationalized, however, the stress-potentiated plus-maze paradigm may see special use in unveiling the anxiolytic effects of neuropeptide antagonists and other experimental compounds.
1.7 Separation-induced Ultrasonic “Distress” Vocalization
Rat pups emit high frequency (30–50 kHz) “distress calls” when separated from mother and littermates, which elicits retrieval behavior from the mother (Noirot 1972). In this model, a reduction in the high-frequency calls in the absence of behavioral sedation is taken as the index of anxiety reduction (Insel et al. 1986; Gardner 1985). The eliciting stimulus (separation), under certain conditions (e.g., dependency) seems capable of producing fear or anxiety in humans, and in this sense the model may be analogous to separation anxiety (for a review see Igor et al. 2001).
Summary of findings with the social interaction test following peripheral administration of different classes of anxiolytic drugs
Bhattacharya et al. (2000), Cheeta et al. (2001), Eguchi et al. (2001), File and Hyde (1978, 1979), File et al. (2001), Kantor et al. (2005), Kita and Furukawa (2008), Louis et al. (2008), Millan et al. (2001), Mizowaki et al. (2001), Rex et al. (2004), Salome et al. (2006), Si et al. (2005), Wood et al. (2001)
Cheeta et al. (2001), Costall et al. (1992), Cutler (1991), Dunn et al. (1989), Haller et al. (2000, 2001), Louis et al. (2008), Millan et al. (2001), Picazo et al. (1995), Salome et al. (2006)
Bristow et al. (2000), Dekeyne et al. (2000), Duxon et al. (2000) (chronic), File et al. (1999), Lightowler et al. (1994) (chronic), Starr et al. (2007) (chronic), To and Bagdy (1999), To et al. (1999)
Pellow and File (1987)
Cheeta et al. (2001), Bagdy et al. (2001)
Johnston and File (1988)
Rex et al. (2004)
Lightowler et al. (1994), Duxon et al. (2000), Salome et al. (2006), Louis et al. (2008)
Eguchi et al. (2001), File (1985), Popik and Vetulani (1993)(chronic)
Blumberg et al. (2000) have proposed that ultrasonic vocalizations in rat pups are not anxiety responses per se, but by-products of the “abdominal compression reaction,” which increases venous return to the heart when its return is compromised. Blumberg et al. demonstrate that cold temperatures can elicit ultrasonic vocalizations through an increase in venous pressure, as can clonidine, an α2 adrenoceptor agonist. Whether or not this cardiovascular mechanism can explain the suppression of ultrasonic vocalizations by anxiolytic drugs is questionable. It is just as likely that these cardiovascular perturbations themselves are aversive or anxiogenic, or a by-product of anxiety, in which case these data only reinforce the model’s general utility.
Summary of findings with the elevated plus maze test following peripheral administration of different classes of anxiolytic drugs
Kapus et al. (2008), Lister (1987), Pellow et al. (1985), Naderi et al. (2008), Ognibene et al. (2008), Rocha et al. (2007), Seo et al. (2007), Yoon et al. (2007), Albrechet-Souza et al. (2007, 2005), Stemmelin et al. (2008), Felipe et al. (2007), Wesolowska and Nikiforuk (2007), Wei et al. (2007), Bradley et al. (2007), Lolli et al. (2007), Grundmann et al. (2007), Drapier et al. (2007), Chen et al. (2006), Violle et al. (2006), Byrnes and Bridges (2006), Cui et al. (2007), Vignes et al. (2006), Vargas et al. (2006), Wesolowska et al (2006), Gonzalez-Trujano et al. (2006), Gonzalez-Pardo et al. (2006), Kumar and Sharma (2005), Xu et al. (2006), Perrine et al. (2006), Atack et al. (2006), Mora et al. (2006), Carr et al. (2006), Hagenbuch et al. (2006), Papp et al. (2006), Mora et al. (2005), Tokumo et al. (2006), Cha et al. (2005), Chen et al. (2004, 2005), Yasui et al. (2005), Park et al. (2005), Rabbani et al. (2005), Mi et al. (2005), Fernandez-Guasti et al. (2005), Clenet et al. (2004), Savic et al. (2004), Soman et al. (2004), Peng et al. (2004), Rabbani et al. (2004), Wilson et al. (2004), Molina-Hernandez et al. (2004), Cryan et al. (2004), Klodzinska et al. (2004), Kurt et al. (2003), Huen et al. (2003), Rabbani et al. (2003), Cechin et al. (2003), Dal-Co et al. (2003), Silva and Brandao (2000)
Cao and Rodgers (1997), Critchley and Handley (1987a, b), Dunn et al. (1989), Graeff et al. (1990), Griebel et al. (1997), Hallar et al. (2000), Mendoza et al. (1999), Pokk and Zharkovsky (1998), Soderpalm et al. (1989), Grundmann et al. (2007), Jung et al. (2006), Vaidya et al. (2005), Kim et al. (2004), Peng et al. (2004), Majercsik et al. (2003), Escarabajal et al. (2003), Lin et al. (2003)
Cadogan et al. (1992), Griebel et al. (1994), Griebel et al. (1999), Kurt et al. (2000), Kuan et al. (2008), Matsuzawa-Yanagida et al. (2007), Chaki et al. (2005)
Matsuzawa-Yanagida et al. (2007), Aricioglu and Altunbas (2003)
Caille et al. (1996), Paslawski et al. (1996), Johnston and File (1988)
Handley and McBlane (1992), Koks et al. (2001), Pollier et al. (2000), Silva and Brandao (2000), Silva et al. (1999), Drapier et al. (2007)
Collinson and Dawson (1997), Critchley and Handley (1987a, b), Pellow and File (1986), Pellow et al. (1987), Moser (1989), Moser et al. (1990), Rodgers et al. (1997a)
Handley and McBlane (1992), Linnoila et al. (1987), Rodgers et al. (1997b), Adamec et al. (2004), Holmes and Rodgers (2003)
Cole and Rodgers (1995), File and Johnston (1987), Lister (1987), Luscombe et al. (1990), Pellow et al. (1985), Drapier et al. (2007)
Holmes and Rodgers (2003)
Summary of findings with the ultrasonic vocalizations test following peripheral administration of different classes of anxiolytic drugs
Engel et al. (1987), Fish et al. (2000), Gardner (1985a, b), Hodgson et al. (2008), Iijima and Chaki (2005), Kehne et al. (2000), Miczek et al. (1995), Millan et al. (2001), Olivier et al. (1998), Podhorna and Brown (2000), Rowlett et al. (2001)
Benton and Nastiti (1988), Fish et al. (2000), Hodgson et al. (2008), Iijima and Chaki (2005), Kehne et al. (1991, 2000), Millan et al. (2001), Nastiti et al. (1991), Olivier et al. (1998), Siemiatkowski et al. (2001)
Hodgson et al. (2008), Iijima and Chaki (2005), Kehne et al. (2000), Olivier et al. (1998)
Kehne et al. (2000), Podharma and Brown (2000)
Winslow and Insel (1990)
Iijima and Chaki (2005)
USVs induced by stressors other than foot shock, such as air puffs, conspecific intruders, acoustic startle stimuli, or drug withdrawal are also reduced by benzodiazepine anxiolytics and 5-HT1A compounds, although there is considerable variability (Naito et al. 2003; Vivian and Miczek 1993; Kaltwasser 1990, 1991; Knapp et al. 1993, 1998; but see Becker et al. 2001). Cold-induced USVs can be reduced by several SSRIs (Fish et al. 2004). However, it is hard to imagine that stimuli as diverse as cold temperatures, attack by an intruder, air puffs to the head, and withdrawal from cocaine would produce similar distress, either quantitatively or qualitatively. Thus, variability in drug effects may be associated with the specific stimulus used to induce the USVs.
2 Summary and Conclusions
In summary, the models reviewed in this chapter show at least some sensitivity to a variety of agents known to produce anxiolysis in humans (i.e., the benzodiazepines, antidepressants, and 5-HT1A compounds). All of the models show good sensitivity to benzodiazepine anxiolytics. Light/dark exploration, social interaction, elevated plus-maze, shock-probe/marble burying, and the conflict tests have shown some sensitivity to antidepressants and 5-HT1A compounds, but to varying degrees. The conflict and ultrasonic vocalization tests seem to be broadly sensitive to the anxiolytic effects of antidepressants. Ultrasonic vocalization appears to be sensitive to all classes of therapeutically effective anxiolytic compounds. Fear-potentiated startle, although sensitive to both benzodiazepine and 5-HT1A anxiolytics, has thus far failed to detect the anxiolytic effects of traditional antidepressants.
While the majority of these models showed at least some sensitivity to antidepressant and 5-HT1A compounds, the anxiolytic effects of these drugs were often more variable than the effects of benzodiazepine anxiolytics. In addition, there were a number of instances in which antidepressant and 5-HT1A agents produced effects opposite to those of standard anxiolytics, suggestive of an “anxiogenic” action. There are several possible explanations for these inconsistencies, which have more general implications for animal models of anxiety and anxiolytic drug action.
A drug may have very reliable effects in an animal model of anxiety, but unless that drug also has reliable antianxiety effects in humans, it cannot be used to validate the animal model. Conversely, a drug that has inconsistent or unreliable anxiolytic effects in humans cannot be used to invalidate an animal model of anxiety. In this regard, there is little clinical evidence that 5-HT1A agents, other than buspirone, produce reliable antianxiety effects in humans and even the effects of buspirone appear to be more variable than the effects of benzodiazepine anxiolytics (e.g., Pecknold et al. 1985; Sheehan et al. 1988, 1990; Wheatley 1982). A number of clinical trials (e.g., Olajide and Lader 1984; Sheehan et al. 1990) suggest that the efficacy of buspirone across different human anxiety disorders (see DSM-IV) is not as robust as the benzodiazepines (Hoffman and Mathew 2008). For a summary of comparative clinical findings see; Argyropoulos et al. 2000, Table 1. These clinical data are certainly not definitive but, if anxious humans respond more variably to buspirone than to benzodiazepine anxiolytics, one might expect the effects of 5-HT1A agents in animal models of anxiolytic drug action to be more variable than the effects of benzodiazepines.
The clinical efficacy of antidepressant drugs in the treatment of anxiety disorders is far more convincing, but there is still some variation in efficacy (see Tyrer and Tyrer 1994; Hoffman and Mathew 2008; Borsini et al. 2002). There is also some disagreement about whether specific antidepressants are required for particular anxiety disorders (e.g., agoraphobia, panic), or are superior to benzodiazepine anxiolytics for these disorders. Furthermore, anxiety in humans often overlaps with depression, so that interpretation of a therapeutic drug effect as being either anxiolytic or antidepressant can sometimes be difficult. Perhaps the most important clinical finding in this literature, however, is that unlike classical benzodiazepines, the anxiolytic effects of traditional antidepressants in humans are normally delayed (2–4 weeks), and the initial (acute) response may sometimes be an exacerbation of anxiety (Argyropoulos et al. 2000). Thus, acute antidepressant treatment in an animal model of anxiety is of questionable relevance to its pharmacological validation.
Chronicity may be equally relevant to the effects 5-HT1A compounds in these models. Whereas chronic administration of 5-HT1A or antidepressant drugs often resulted in reliable, anxiolytic effects in a variety of animal models (Commissaris et al. 1990; Duxon et al. 2000; Fontana and Commissaris 1988; Griebel et al. 1994; Rocha et al. 1994; Soderpalm et al. 1993; Yamashita et al. 1995), acute administration resulted in less reliable anxiolytic effects, or even anxiogenesis (e.g., Griebel et al. 1994; Handley and McBlane 1992; Moser 1989).
Another possibility is that different animal models represent qualitatively different types of “anxiety” or fear, only some of which are reliably inhibited by 5-HT1A agents or antidepressants. Thus, one could speculate that the social interaction test primarily reflects a type of social phobia, which is reliably suppressed by 5-HT1A agents and certain antidepressants, whereas the elevated plus-maze test reflects a type of acrophobia, which is not as reliably suppressed by 5-HT1A agents or antidepressants. This would imply that animal fears can be pharmacologically dissected, which in turn would support the pharmacological dissection of human anxiety. Although these speculations seem to be consistent with some of the animal data reviewed in this chapter, at this time there is no convincing clinical evidence that specific anxiety disorders are differentially affected by benzodiazepine, 5-HT1A or antidepressant anxiolytics (Hoffman and Mathew 2008; Argyropoulos et al. 2000; Tyrer and Tyrer 1994).
Thus, a number of factors, including clinical effectiveness, chronicity, and model-type, may alter the correspondence between the effects of benzodiazepines, 5-HT1A agents, and antidepressants in animal models of anxiety. On the whole, however, the data summarized above suggest that there is enough correspondence between drug-effects across these tests that future paradigmatic studies may ultimately establish their validity as general models of antianxiety drug action. While all models may not attain this ideal, it should be remembered that “class-specific” models could serve as a valuable tools for studying the mechanisms by which benzodiazepine, 5-HT1A, or antidepressant drugs produce their anxiolytic effects.
From an evolutionary perspective, one could argue that this is putting the cart before the horse: a scientific understanding of anxiety in humans first requires a detailed understanding of its distal and proximal causes in lower animals.
For readers interested in investigational drugs, their anxiolytic properties and mechanisms of action, see Chapters Metabotropic Glutamate Receptors (W. Spooren), Neuropeptides (T. Steckler), and Cannabinoids (C. Wotjak) in the present volume.
- Albrechet-Souza L, Oliveira AR, De Luca MC et al (2005) A comparative study with two types of elevated plus-maze (transparent vs. opaque walls) on the anxiolytic effects of midazolam, one-trial tolerance and fear-induced analgesia. Prog Neuropsychopharmacol Biol Psychiatry 29(4):571–579PubMedCrossRefGoogle Scholar
- Albrechet-Souza L, de Carvalho MC, Franci CR, Brandao ML (2007) Increases in plasma corticosterone and stretched-attend postures in rats naive and previously exposed to the elevated plus-maze are sensitive to the anxiolytic-like effects of midazolam. Horm Behav 52:267–273PubMedCrossRefGoogle Scholar
- American Psychiatric Association (1994) Diagnostic and Statistical Manual of Mental Disorders, (DSM-IVTM), 4th edn. American Psychiatric Association, Washington, DCGoogle Scholar
- Argyropoulos SV, Sandford JJ, Nutt D (2000) The psychobiology of anxiolytic drugs Part 2: pharmacological treatments of anxiety. PharmacolTher 88:213–227Google Scholar
- Aricioglu F, Altunbas H (2003) Agmatine an endogenous anxiolytic. Agmatine and imidazolines: Their novel receptors and enzymes. 1009:136–140Google Scholar
- Bill DJ, Fletcher A, Knight M (1989) Actions of 5HT1A ligands and standard anxiolytics on mouse exploratory behaviour in a two compartment light: dark arena. Br J Pharmacol 98(Suppl):679Google Scholar
- Bojarski AJ, Paluchowska MH, Duszynska B et al (2006) Structure-intrinsic activity relationship studies in the group of 1-imido/amido substituted 4-(4-arylpiperazin-1-yl) cyclohexane derivatives; new, potent 5-HT1A receptor agents with anxiolytic-like activity. Bioorg Med Chem 14:1391–1402PubMedCrossRefGoogle Scholar
- Cadogan AK, Wright IK, Coombs I et al (1992) Repeated paxroxetine administration in the rat produces an anxiolytic profile in the elevated X-maze and decreased [3H]-ketanserin binding. Neurosci Lett 42(Suppl):S8Google Scholar
- Casella JV, Davis M (1985) Fear enhanced acoustic startle is not attenuated by acute or chronic imipramine treatment in rats. Psychopharmacology (Berl) 87:4Google Scholar
- Collinson N, Dawson GR (1997) On the elevated plus-maze the anxiolytic-like effects of the 5-HT1A agonist, 8-OH-DPAT, but not the anxiogenic-like effects of the 5-HT1A partial agonist, buspirone, are blocked by the 5-HT1A antagonist, WAY 100635. Psychopharmacology (Berl) 132:35–43CrossRefGoogle Scholar
- Commissaris RL, Hill RJ (1995) High-dose subchronic imipramine treatment: effects on anxiety-like (conflict) behavior in rats. Anxiety 1:109–113Google Scholar
- Costall B, Hendrie CA, Kelly ME et al (1987) Actions of sulpiride and tiapride in a simple model of anxiety in mice. Neuropharmacology 26:5Google Scholar
- Costall B, Domeney AM, Grrard PA et al (1988a) Zacopride: anxiolytic profile in rodent and primate models of anxiety. J Pharm Pharmacol 40:3Google Scholar
- Costall B, Kelly ME, Naylor RJ et al (1988b) Actions of buspirone in a putative model of anxiety in the mouse. J Pharm Pharmacol 40:6Google Scholar
- Costall B, Domeney AM, Farre AJ et al (1992) Profile of action of a novel 5-hydroxytriptamine1A receptor ligand E-4424 to inhibit aversive behavior in the mouse, rat and marmoset. J Pharmacol Exp Ther 262:8Google Scholar
- Costanzo A, Guerrini G, Ciciani G et al (2002) Benzodiazepine receptor ligands. 7. Synthesis and pharmacological evaluation of new 3-esters of the 8-chloropyrazolo[5,1-c][1,2,4]benzotriazine 5-oxide. 3-(2-thienylmethoxycarbonyl) derivative: an anxioselective agent in rodents. J Med Chem 45:5710–5720PubMedCrossRefGoogle Scholar
- Critchley MAE, Handley SL (1987a) 5-HT 1A ligand effects in the X-maze anxiety test. Br J Pharmacol 92:660PGoogle Scholar
- Cryan JF, Kelly PH, Chaperon F et al (2004) Behavioral characterization of the novel GABAB receptor-positive modulator GS39783 (N, N′-dicyclopentyl-2-methylsulfanyl-5-nitro-pyrimidine-4, 6-diamine): anxiolytic-like activity without side effects associated with baclofen or benzodiazepines. J Pharmacol Exp Ther 310(3):952–963PubMedCrossRefGoogle Scholar
- Cutler MG (1991) An ethological study of the effects of buspirone and the 5-HT3 receptor antagonist, BRL 43694 (granisetron) on behavior during social interactions in female and male mice. Neuropharmacology 30:7Google Scholar
- Davis M (1986a) Neural mechanisms of fear conditioning measured with the acoustic startle reflex. Pharmacol Biochem Behav 25:312Google Scholar
- De Boer SF, Slangen JL, van der Gugten J (1990) Plasma catecholamine and corticosterone levels during active and passive shock-prod avoidance behavior in rats: effects of chlordiazepoxide. Pharmacol Biochem Behav 47:1089–1098Google Scholar
- Dekeyne A, Brocco M, Adhumeau A et al (2000) The selective serotonin (5-HT)1A receptor ligand, S15535, displays anxiolytic-like effects in the social interaction and Vogel models and suppresses dialysate levels of 5-HT in the dorsal hippocampus of freely-moving rats. Psychopharmacology (Berl) 152:55–66CrossRefGoogle Scholar
- File SE, Hyde JRG (1979) A test of anxiety that distinguishes between the actions of benzodiazepines and those of other minor tranquilisers and stimulants. Pharmacol Biochem Behav 11:4Google Scholar
- Hallar J, Hallasz J, Makara GB (2000) Housing conditions and the anxiolytic effect of buspirone: the relationship between main and side effects. Behav Pharmacol 11:9Google Scholar
- Handley SL, McBlane JW (1992) Opposite effects of fluoxetine in two animal models of anxiety. Br J Pharmacol 107(Suppl):446PGoogle Scholar
- Imaizumi M, Suzuki T, Machida H et al (1994b) A fully automated apparatus for a light/dark test measuring anxiolytic or anxiogenic effects of drugs in mice. Jpn J Psychopharmacol 14:83–91Google Scholar
- Kaltwasser MT (1990) Acoustic startle induced ultrasonic vocalization in the rat – a novel animal-model of anxiety. Psychopharmacology 101:s28–s28Google Scholar
- Kantor S, Jakus R, Molnar E et al (2005) Despite similar anxiolytic potential, the 5-hydroxytryptamine 2C receptor antagonist SB-242084 [6-chloro-5-methyl-1-[2-(2-methylpyrid-3-yloxy)-pyrid-5-yl carbamoyl] indoline] and chlordiazepoxide produced differential effects on electroencephalogram power spectra. J Pharmacol Exp Ther 315:921–930PubMedCrossRefGoogle Scholar
- Kuan WL, Zhao JW, Barker RA (2008) The role of anxiety in the development of levodopa-induced dyskinesias in an animal model of Parkinson’s disease, and the effect of chronic treatment with the selective serotonin reuptake inhibitor citalopram. Psychopharmacology 197:279–293PubMedCrossRefGoogle Scholar
- Lapiz-Bluhm MD, Bondi CO, Doyen J, Rodriguez GA, Be´dard-Arana T, Morilak DA (2008) Behavioural assays to model cognitive and affective dimensions of depression and anxiety in rats. J Neuroendocrinol 20:1115–1137Google Scholar
- Lister RG (1987) The use of a plus-maze to measure anxiety in the mouse. Psychopharmacology (Berl) 92:180–185Google Scholar
- Luscombe GP, Mazurkiewicz SE, Buckett WR (1990) Evaluation of tricyclic antidepressants in the elevated plus-maze in rats: anxiolytic effects of dothiepin and doxepine. Br J Pharmacol 100(Suppl):356PGoogle Scholar
- Mathiasen LS, Mirza NR, Rodgers RJ (2008) Strain- and model-dependent effects of chlordiazepoxide, L-838,417 and zolpidem on anxiety-like behaviours in laboratory mice. Pharmacol Biochem Behav (in press)Google Scholar
- Nevins ME, Anthony EW (1994) Antagonists at the serotonin-3 receptor can reduce the fear potentiated startle response in the rat: evidence for difference types of anxiolytic activity? J Pharmacol Exp Ther 268:6Google Scholar
- Olajide D, Lader M (1984) A double-blind comparison of buspirone and diazepam in outpatients with chronic anxiety states. Abstract at C.I.N.P. meeting, FlorenceGoogle Scholar
- Pellow S (1986) Anxiolytic and anxiogenic drug effects in a novel test of anxiety: are exploratory models of anxiety in rodents valid? Meth Find Expt Clin Pharmacol 8:557–565Google Scholar
- Popik P, Kostakis E, Krawczyk M et al (2006) The anxioselective agent 7-(2-chloropyridin-4-yl)pyrazolo-[1, 5-a]-pyrimidin-3-yl](pyridin-2-yl)methanone (DOV 51892) is more efficacious than diazepam at enhancing GABA-gated currents at alpha1 subunit-containing GABAA receptors. J Pharmacol Exp Ther 319:1244–1252PubMedCrossRefGoogle Scholar
- Rudzik AD, Hester JB, Tang AH et al (1973) Triazolobenzodiazepines, a new class of central nervous system depressant compounds. In: Mussini E, Randall LO (eds) The Benzodiazepines. Raven Press, New York, pp 285–326Google Scholar
- Shimada T, Matsumoto K, Osanai M et al (1995) The modified light/dark transition test in mice, evaluation of classic and putative anxiolytic and anxiogenic drugs. Gen Pharmacol 26:5Google Scholar
- Stefanski R, Palejko W, Kotowski W (1992) The comparison of benzodiazepine derivatives and serotonergic agonists and antagonists in two animal models of anxiety. Neuropharmacology (Berl) 31:1252–1258Google Scholar
- Treit D, Menard J, Pesold C (1994) The shock probe burying test. Neurosci Protocols Mod 3:9–17Google Scholar
- Treit D, Degroot A, Shah A (2003) Animal models of anxiety and anxiolytic drug action. In: Kasper S, den Boer JA, Ad Sitsen JM (eds) Handbook of depression and anxiety: a biological approach. New York, Marcel Dekker, pp 681–702Google Scholar
- Tsuda A, Ida Y, Tanaka M (1988) The contrasting effects of diazepam and yohimbine on conditioned defensive burying in rats. Psychobiology 16:213–217Google Scholar
- Tyrer P, Tyrer J (1994) Antidepressive drugs for treatment of anxiety disorders – and vice versa. In: den Boer JA, Ad Sitsen JM (eds) Handbook of depression and anxiety: a biological approach. New York, Marcel Dekker, pp 497–514Google Scholar
- Vaidya AH, Rosenthal DI, Lang W et al (2005) Oral buspirone causes a shift in the dose-response curve between the elevated-plus maze and vogel conflict tests in Long-Evans rats: relation of brain levels of buspirone and 1-PP to anxiolytic action. Meth Find Exp Clin Pharmacol 27:245–255CrossRefGoogle Scholar
- Westen D, Rosenthal R (2003) Quantifying construct validity: two simple measures. JPSP 84:608–618Google Scholar
- Wheatley D (1982) Buspirone: multicenter efficacy study. J Clin Psychiat 43:92–94Google Scholar