Animals, anxiety, and anxiety disorders: How to measure anxiety in rodents and why

Behavioural Brain Research xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr

Review

Animals, anxiety, and anxiety disorders: How to measure anxiety in rodents and why☆ Jaanus Harro Division of Neuropsychopharmacology, Department of Psychology, Estonian Centre of Behavioural and Health Sciences, University of Tartu, Ravila 14A Chemicum, 50411, Tartu, Estonia

A R T I C L E I N F O

A B S T R A C T

Keywords: Anxiety Measurement Animal models Anxiolytics Screening Anxiogenic drugs

Measurement of anxiety is desirable for the benefit of drug development and understanding the brain function and mental well-being. Animal models offer the advantages of detailed neurobiological analysis, experimental manipulation of specific components in the brain circuits that underlie psychopathology, and the possibility of screening novel drugs with clinical potential. A large variety of animal models of anxiety and screening tests of anxiolytics is currently in use. While their value in advancing the knowledge and predicting therapeutic success of drugs is unquestionable, the expectations have grown much higher, and the frustration over absence of novel successful drug concepts is rising. It is argued that the multitude of factors that can interfere with animal behaviour in anxiety tests, and the complexity of neurobiology of the various anxiety disorders, present high demands on validation of each anxiety test within each specific laboratory condition. Anxiety models should be explicitly related to a theoretical paradigm on underlying neurobiology, because there is a diversity in concepts, and validation of the model and the selection of behavioural readouts is critically dependent on the neurobiological model. Environmental conditions during the model production and anxiety testing need more attention, including the less considered factors such as ultrasounds. More attention is required to the differences in anxiety neurobiology between males and females, and inter-individual differences in coping strategies.

1. Introduction Only little effort is needed to find an explanation why anxiety measurement in animal models is of paramount importance: Anxiety disorders are highly prevalent throughout adulthood [1], have an early onset [2], affect a significant proportion of young adults [1], remain an enormous burden on health care resources [3], and are co-morbid with a variety of medical conditions including neurological and cardiovascular illnesses, with resultant major reduction in quality of life [4]. Furthermore, anxiety is a component in a variety of diseases [5], appears to be associated with accelerated aging [6], current treatment options are perceived as inadequate [7], and our understanding of the pathogenesis is thought of as far from sufficient [8,9]. Owing to the high costs of clinical trials, and especially so with regard to the CNS therapeutics [10], preclinical models of the disorder and drug screening tests that would assure of the relevance of the selected molecular target, or predict clinical efficacy of the substance in development, are of obvious significance. Consistently with the importance of the theme, many overviews of anxiety tests and anxiety modelling have been published. The reader is

suggested to consult with Cryan and Sweeney [9] for a recent comprehensive overview of available tests, as well as other important recent reviews, each with somewhat different focus [11–14]. These treatises present the state-of-the-art while having somewhat different emphasis, and discuss the pertinent issue of predictive, face, convergent, etiological, construct and population validity, eventually concluding that animal models are indispensable in psychiatric research, that much progress has been made, but that much more has been desired. Furthermore, the issue of increasingly uncritical use of the animal tests in attempt to “translate” complex neurobiology and its consequences between species has been clearly raised [15]. This uncritical use may however have real-life incentives, as Georg Wilhelm Friedrich Hegel would have reminded us with his words: “What is reasonable is real; that which is real is reasonable.” [16]. Indeed, a recognition is due that while in the ideal world we pursue pertinent research questions with persistence guided by the best of all accumulated knowledge, in reality modern research is highly fragmented by the temporary nature of grants, student projects, post-doc jobs, increasing administrative burden on senior investigators, the lure of emergent technologies, changes in priorities of the institutions, and else. The resultant shift

☆

Dedicated to the memory of Jaak Panksepp (1943–2017). E-mail address: [email protected].

http://dx.doi.org/10.1016/j.bbr.2017.10.016 Received 14 July 2017; Received in revised form 12 October 2017; Accepted 14 October 2017 0166-4328/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Harro, J., Behavioural Brain Research (2017), http://dx.doi.org/10.1016/j.bbr.2017.10.016

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

reflected in a single dimension, or because of the realism of the anxiety researchers who have observed that studies on depression, more often than not, stubbornly refuse to accept the criticism on over-interpretation of the findings in screening tests? What may be commonly assumed but rarely made explicit is the logic that the difference between tests and models lies in the way how anxiety is brought about: In anxiety tests, the assumption is relatively similar anxiety level in all animals before the experiment; in anxiety models, some organisms are known – or thought of – in advance as having persistently higher anxiety owing to genetic background, developmental factors, or adverse environmental events. Obviously, anxiety tests must be used to make the assessment of anxiety in anxiety models, and if testing itself would produce lasting changes in the brain that will lead to persistent expression of anxiety upon repeated measurement, we have arrived at transforming a test into a model of anxiety.

Fig. 1. Annual output of anxiety research in animals: The number of items in PubMed database from the first item in 1962–2015, by using search term “anxiety animal model”.

2.1.1. Tests for the measurement of anxiety 2.1.1.1. Geller-Seifter test. Geller-Seifter test is the prototypical “conflict” test that provides an animal the option to obtain food when hungry by pressing a lever that can also elicit electric footshock [19]. Hence, the response rate to food is inhibited by response-contingent punishment. This test was found to predict clinical efficacy of anxiolytic drugs and further refined by introducing incremental shock levels [20]. The Geller-Seifter test excellently predicts not only clinical efficacy but also the clinically effective dose among benzodiazepines and barbiturates. Under standard conditions it nevertheless has low sensitivity for other anxiolytic treatments. The Geller-Seifter test requires training of the animals and can be affected by other effects of the drug on motivations, such as an analgesic or orexigenic action. These are accounted for by measuring drug action during time slots for unpunished responding, but it should be noted that responding under conflict situation is usually proportionally very much lower, and hence the impact of strong motivational effects (e.g., orexigenic) may not be entirely under control.

towards short-term goals and the overwhelming increase in the amount of scientific information, including that on anxiety tests and models in animals (Fig. 1) are together implicitly guiding the efforts of research towards what seem to be the low-hanging fruit. Alas, some of the fruits can be part of a large-scale self-created mirage, as evidenced by the fact that anxiolytic drugs with any novel mechanism of action have been slow to emerge, while this is not owing to shortage of good intention. As it has been succinctly put in a characterization of the historical development in the field: “It is somewhat ironic that as the tests employed became more sophisticated, the development of anxiolytic drugs has not greatly increased” [9]. Given the sheer volume of publications on anxiety testing and all the complexity of findings in the available literature, it is likely that a newcomer to the field increasingly feels pressed toward the realization that there is too much of previous thoughts and practice to consider, and the best is to just start experimenting right away. With this recognition in mind, this article will first 1) present a condensed overview of the often used animal tests of anxiety, providing historical and current key references and including a few field notes that may be relevant to further discussion and for testing refinement; and only then 2) discuss conceptual issues that arise in development of tests of anxiety and models of human anxiety disorders; 3) focus on potential sources of failure and success in using animal models; and 4) consider some recent issues in model development such as sex, genetic background, interindividual differences, environmental factors etc that have the potential of either confound or, if properly addressed, enrich further studies on anxiety. While recognizing that research on anxiety, screening for anxiolytics and indeed attempts to understand the whole spectrum of psychiatry is being conducted on many different species, and increasingly on “simpler” organisms such zebrafish [17,18] this paper will heavily rely on rodents not only owing to the fact that most of the literature published to date has been dealing with rats, mice and other members of this order (that comprises about 40% of all mammal species) but also because with inclusion of the potential of the whole animal kingdom, the conceptual issues as discussed below would further broaden to unnecessary extent.

2.1.1.2. Vogel test. Another conflict test developed for anxiolytic screening, the Vogel water-lick suppression test, uses thirst motivation instead of hunger [21]. Here, less training is needed as compared to the Geller-Seifter test, but sensitivity to pain and the potential analgesic effect of drugs remain to be controlled for. Numerous procedural variations exist of the Vogel test and have been discussed in detail [22]. The Geller-Seifter, Vogel, and other classic tests involving training of and learning by the animals are time-consuming and have appeared to be less sensitive to systemic administration of drugs other than those acting directly on molecular targets in the GABA-ergic system. For these reasons it is only natural that many attempts have been made to devise principally different and apparently more simple methods. Spontaneous behaviour based tests have been spearheaded by the elevated plus-maze test and the redefinition of the open field paradigm. 2.1.1.3. Open field test. In this context, open field is meant to be a circular or rectangular well-lit arena that is several times larger than the home-cage and is supposed to be novel, strange and mildly aversive [23]. What is measured is the locomotion in terms of the number of squares crossed and rearings, and several refinements made to the open field test pay particular attention to more central areas of the arena and quantitate freezing. The open field test principle has been used to assess anxiety in an amazing variety of living creatures such as rats, mice, gerbils, hamsters, ferrets, foxes, dogs, cattle, sheep, pigs, chicken, quail, and several species of fish [24]. This has however not always been meant to study anxiety in terms translatable to contemporary psychiatry. Indeed, the PubMed database reveals very few references to explicitly anxiety-related open field research until the 90ies (Fig. 2; see 3.2.1). Of course the open field test had been around before that,

2. Major animal tests and models of anxiety 2.1. Tests and models In the closely related field of depression models much discussion has been directed at the need to distinguish models of depression and antidepressant screening tests. Curiously, while reading the anxiety literature it appears that such a discussion on distinction between drug screens and anxiety models is largely absent. Could it be owing to the perception that anxiety is a way simpler phenomenon and better 2

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

manner [34]. While the holeboard test may rather be a suitable tool to include into a battery of tests that aims at separating anxiety, exploratory, activity and learning responses [35], it is often used as an anxiety test [12]. 2.1.1.6. Elevated plus-maze and other elevated mazes. The highly successful elevated plus-maze test of anxiety was developed from the Y-maze test of Montgomery [36] and named alternatively either elevated X-maze [37] or elevated plus-maze [38]. Originally used for rats, the mouse version [39] was created soon after the very beginning of success of the elevated plus-maze test. The elevated plus-maze consists of two open and two enclosed arms connected by a central square, and entries into and time spent in open arms are indicative of non-anxious behaviour. More rigorously, it is the proportion of open arm entries and proportion of time in open arms that indicates anxiety, and the total number of arm entries is often taken as a measure of general locomotor activity. It should nevertheless be noted that the anxiety measures in plus-maze correlate (negatively) with the total number of entries [40], and if the number of open arm entries is low, the statistical analysis of the data is not straightforward [27]. Modifications of the measures recorded as well as the plus-maze itself and measurement procedure are manifold, including different materials for building of the maze and introduction of side lips to open arms, consideration of activity at the central platform [41,42], and sophisticated ethological analysis of all of the behaviours displayed [43]. One more commonly accepted modification is the evaluation of risk assessment in plus-maze [44], and the graded plus-maze test [45] aims at bolstering the extreme responses from individual animals to permit more normal distribution of anxiety measures. The elevated platform principle has been further used in form of the elevated T-maze test [46], the canopy test [47] and the elevated zeromaze test [48,49]; while not becoming as popular as their predecessor test the rat elevated zero-maze seems to be increasingly used in most recent years. Here the area to explore is circular and consists of four segments of equal length, two exposed and two with walls.

Fig. 2. Papers in Medline with search term “open field anxiety”. Note that while the open field test was well characterized before the 2nd World War, it became an “anxiety test” after several more recent inventions but gained popularity very fast.

but the psychological constructs under investigation were bearing labels like timidity, fearfulness and emotionality. Whatever the precise neurobiology behind these constructs that had been explored with the open field test, they were found to be unreliably measured and several reviews published at the end of the past millennium had concluded that any generic, context-independent use of the open field test can only be discouraged [25–27]. Nonetheless, the apparent simplicity of the open field test brought about its rebirth with many modifications aiming at increasing the validity of the anxiety component measures in the open field behaviour and these have led to more optimistic evaluations of the technique [28]. What is observed in the open field test is the so-called forced exploration, meaning that the animal has no escape from the test area. Enforcement facilitates production of anxiety but measurement of locomotion in the simplistic environment confuses the measurement of the two principal motivational components of exploration, the fear of novel or open spaces and curiosity or the drive to know the environment by learning about it. Critical aspects of open field test include the measurement time, size of the arena, and light levels. Open field test is often refined to include the separate measure of centre activity but it should not be taken as granted that thigmotaxis or preference of the closeness to walls reflects anything related to pathological anxiety since this is the natural behaviour of rodents. (Parenthetically, one can often observe thigmotaxis in human behaviour, e.g., while selecting a table in a café.) Freezing, defecation and urination, the measures of the emotionality research tradition, may indeed carry valuable information.

2.1.1.7. Social interaction test. File and Hyde proposed placing two unfamiliar rats in a test box and recording time spent in social activity as an animal model of anxiety [50]. Social interaction test is sensitive to light levels and the familiarity of the chamber. This is of course true for all other tests of anxiety based on unlearned behaviour, but for the social interaction test comparison of effects of drugs in familiar/ unfamiliar and high/low light levels has been explicitly used to validate the drug effect [51]. It has been emphasized that behaviour of one rat influences that of the other and hence using separate scores for each member of the pair distorts the conditions on which the usual statistical treatment of the data is based, so the pair has to receive similar treatment and be considered as a unit [52]. It may be worth taking into account that individual rats have largely variable but characteristic average levels of expression of social behaviour if tested with multiple partners, so the social interaction score of a pair is in moderate positive correlation with this inherent sociability of animals [53].

2.1.1.4. Light/dark compartment test. The light/dark compartment test provided a clear conflict situation to exploration-based testing. Here the rodent can avoid a brightly illuminated area but venture out of the dark compartment as the curiosity drives and anxiety permits [29]. Various subsequent modifications of the test have been linked to inconsistencies in findings, however [30]. Not only the testing conditions but also the readouts taken (e.g., the original measure of transitions between light and dark vs. the more common time in the light compartment) have been varied between laboratories. Rats have a strong preference for the dark compartment so the test is mostly used with mice.

2.1.1.8. Distress vocalizations. Rodents produce distress vocalizations in the ultrasonic range, and there is a large variety in approaches to their measurement. Separation distress in young animals is reflected in characteristic separation calls and may be particularly useful in studying developmental aspects of anxiety [54]. Measurement of separation calls has been reported as a particularly powerful technique in chicks and guinea-pigs, and used to characterize the role of glutamate- and CRF-ergic pathways or assessment of anxiolytic drug effect [55,56]. In adults the 22-kHz ultrasonic vocalization is easily produced by stressful stimuli and can be pharmacologically reduced [57–59].

2.1.1.5. Holeboard test. The holeboard test was introduced as an exploration test by Boissier and Simon [31] in 1962. Thereafter it appeared in psychopharmacology for studies on the role of muscarinic acetylcholine receptors in basic forms of learning (habituation) [32]. Subsequently this test was used for the demonstration of the role of 5HT-ergic neurons in response to aversive environments [33] and substantially modified to reflect exploration in a more specific 3

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

overeating and increased sexual activity. Hence this paradigm has a potential for use in anxiety research with focus on inter-individual differences (see 3.3.2).

2.1.1.9. Hyponeophagia. Anxiolytics enhancing GABA-ergic neurotransm ission were found to increase food intake of fasted rats in aversive surroundings similarly to handling or habituation to the testing environment [60]. Also the water motivation has been used in this agoraphobia-suppressed behavioural paradigm in the form of the open field drink test [61], where time spent drinking by a thirsty rat was found as the most sensitive measure of effects of both diazepam and FG7142, an anxiogenic ß-carboline. The principle has been used in many modifications that manipulate with the relative familiarity of the testing environment and the food [62–64], including the seed-finding test for hamsters [65].

2.1.1.14. Four-plate test. A classic making use of the strong exploration/activity drive in mice, this test applies the conflict situation brought about by delivery of electric footshocks to locomotion on an open field like arena that has a floor made out of four metal plates [82]. Crossing from one plate to another is punished with electric shock. Originally the test was validated for use on GABAergic drugs but it has enjoyed recent surge in popularity, including studies on the 5-HT2A receptor function in anxiety [83].

2.1.1.10. Conditioned emotional response/conditioned fear/conditioned response suppression. This basic learning paradigm with many names comprises presentation of footshocks in an alien environment that is first allowed to be explored briefly, and recording freezing and/or locomotion in this testing arena the following day [66,67]. Fear conditioning can also be reliably elicited with administration of anxiogenic drugs. It can be a helpful method to help distinguish anxiogenic action from sedative effect [68]. For example, a sedative dose of diazepam would reduce activity in a novel compartment but there will be no carry-over of this effect when subsequently tested drugfree, instead higher activity could be seen; the opposite can be observed with anxiogenic drugs.

2.1.1.15. Staircase test. A variant of exploration-based tests, this method places animals to an enclosed staircase and the number of steps climbed and the number of rearings are counted [84,85]. Low rearing is supposed to reflect anxiety in contrast to the more locomotion-indicating measure of step climbing. 2.1.1.16. Mirrored chamber. An atypical exploration-based anxiety test in mice comes in the form of a chamber covered with mirrors [86] that has been further modified in order to increase exploration of the mirrored area [87]. One within-laboratory comparison of the mirrored chamber and elevated plus-maze tests reported a higher predictive validity for the latter [88] but the authors used the original, not the modified design.

2.1.1.11. Fear-potentiated startle. An auditory stimulus produces larger startle reflex if previously associated with aversive stimuli [69]. Chi [70] reported in 1965 that sodium amobarbital reduced conditioned fear in the form of the potentiated startle reflex. Davis subsequently demonstrated the efficacy of morphine and benzodiazepines against the potentiated startle [71,72], showing the low doses of diazepam were sufficient to reduce the expression of the potentiation. Different neurobiologies are involved in fear that is specific to a cue and that is raised by fearful general context that need not be learned [73]. 2.1.1.12. Conditioned and unconditioned defensive burying. With the aim to exploit mimicking of learning for survival in the wild, conditioned defensive burying of sources of adverse stimulation was introduced for measurement of anxiety [74,75]. Covering of electrified shock-prod with bedding material has been shown to predict anxiety-related effects of a variety of drugs well, and presented as a suitable model to unravel neurobiology of active and passive coping responses that relate to anxiety [76]. Defensive burying occurs also in response to innocuous objects. The marble burying test was proposed as a convenient method to separate action of major vs. minor tranquillizers [77] (referring to antipsychotic and anxiolytic actions) by comparison of the drug effect on burying marbles and on swim-induced grooming. Whether marble burying represents general anxiety response is questionable as the marbles are not avoided [78] and the test has been suggested to correspond better to requirements for an animal model of obsessivecompulsive disorder than to a drug screening test [79].

2.1.1.17. Antipredator/defense tests. Recording of behavioural responses reflecting fear or anxiety has a long tradition in primate research [89], and often the threats in experiments were of social nature. More recently a variety of tests has been developed for use in mice and rats. Rodents have been exposed to predator (cats, foxes, etc) odours that they avoid, and certain other odours produce both unconditioned and conditioned avoidance, whereas it may be significant what is the natural source of the odour [90]. In a different approach, a dead or heavily anaesthetized rat is presented and the defense behaviour (risk assessment, flight response, defensive attack) of a mouse is observed [91]. Such an exposure to the predator also facilitates behavioural conditioning. Other predators have been made use of, including e.g., applying snakes that consume fellow rodents, a method that has found strong objection on ethical grounds. A remarkable yet underused achievement is the Visible Burrow System [92]. The VBS developed by Robert Blanchard provides a colony of rats with a relatively complex environment with open space and dark burrows and tunnels that permits to assess the strategies of dominant and subordinate animals [93]. At another stage a predator animal is placed into the open area, and over a prolonged period a number of behaviours are recorded. Some of these provide information on anxiety state of the rats, including locomotion-related measures as well as 22kHz USV-s. Development of the mouse defense test battery [94] followed the rat models.

2.1.1.13. Frustrative nonreward. One interesting approach that is falling out of the list of “classic” anxiety tests as its popularity plummets is the frustrative nonreward test. Its apparent advantage lies in the fact that what is measured is behavioural activation, not inhibition that can be hard to distinguish from sedation or loss of interest. It has however been suggested that the frustrative nonreward test rather represents activity in the neural circuits of rage and aggressiveness, not anxiety, as more medial aspects of corpus amygdala are involved [80]. Indeed, benzodiazepines that work well in this paradigm have not only anxiolytic but also marked anti-aggressive properties. Cloninger [81] has proposed that in conditions of frustrative nonreward those individuals who have high reward dependence are susceptible to compensatory noradrenergic hyperactivity and states of agitated dysphoria associated with reward-seeking behaviors such as

2.1.1.18. Stress-induced hyperthermia. While the measures of anxiety are usually behavioural readouts, anxiety in humans is not quantified in this way. Obviously paper and pencil self-reports do not apply to animal tests and models, but research on human emotion also exploits physiological measurements. Of these, research in animals has most often used measurement of stress-induced hyperthermia. Anxiolytics such as benzodiazepines and 5-HT1A receptor agonists reduce hyperthermia [57]. Production of stress-induced hyperthermia can make use of the stress by measurement itself, so that the first measure serves as the baseline and the second provides information on the increase. The first measurement can be immediately followed by the administration of the drug under investigation. Alternatively, also mice removed from the cage as the last develop hyperthermia that is attenuated by anxiolytic drugs [95]. A distinct development is the 4

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

stress [130,131]. Electric foot-shock has been advocated as a stressor with the advantage of not being habituated to, and easy to control regarding intensity and duration [132] that may be helpful in accommodating the technique into the laboratory. Post-traumatic stress disorder has also been imitated by exposure of animals to predator-based psychosocial stress [133]. Chronic psychosocial stressful events as risk factors for anxiety disorders are increasingly receiving much interest in animal modeling [134,135] as studies on the underlying neurobiology suggest that single and repeated, intermittent and continuous defeat experiments can reveal the dynamic adaptive response in the brain [136].

possibility to use of telemetry for time-sensitive monitoring. Of course, drugs have the potential to directly interfere with regulation of body temperature and that need be separately addressed [96]. 2.1.2. Animal models of anxiety 2.1.2.1. Genetic models. A large number of studies on anxiety have been carried out in animals with different genetic background and differences in their behaviour in some of the tests used for the measurement of anxiety. Good description of selected models can be found elsewhere [9,97]. Some models are based on outbred strains like the Wistar-Kyoto (WKY) rat [98] that is often compared with the less vigilant Sprague-Dawley strain [99]. Several rat models emerged as more anxious strains were systematically bred, such as the Roman High and Low Avoidance rats [100–102], Maudsley Reactive and Nonreactive rats [103,104], Fawn Hooded rat [105], Floripa High and Low rats [106], Tsukuba High and Low Emotional rats [107], HAB/LAB rats [108], High and Low 8-OH-DPAT responding rats [109,110], High and Low ultrasonic vocalization rats [111], Syracuse High and Low Avoidance rats [112], and Carioca High and Low Conditioned Freezing rats [113]. Findings with these lines suggest some consistency across anxiety tests but there is considerable heterogeneity. Eight of these ten genetic lines based rat models of anxiety were recently reviewed [114] with regard to the performance in eleven anxiety tests. In the positive vein, many consistencies emerged, but the several results that were at variance to what was expected may eventually be even more helpful in understanding the neurobiology of different aspects of anxiety (see 3.2). A few mouse lines have also been described as more anxious, such as the inbred BALB/c mouse [115,116], BTBR T + tf/J mouse [117] and 129 mouse [118]. In an analogy to the plus-maze selection based HAB/ LAB rat lines [119], the mouse HAB/LAB was bred from CD-1 mice [120]. In addition, a large variety of single gene modification animal models have been proposed and the technologies for their production are becoming more sophisticated [9,97]. While several lines have been brought about by spontaneous mutations, many have targeted candidate genes, following the example of one of the first, the CRF overexpressing mouse that of course aimed at enhancing the activity of the HPA axis [121]. As a matter of fact, far too many genetically engineered mice appear as anxious in some of the anxiety tests. In some cases this may be related to their background strain [122,123].

3. Animals and anxiety Obviously, the tests described in the previous section have all been made good use of, as they reappear time and time again in new peerreviewed publications, each presenting findings that seem to make perfect sense. The issue is that all the evidence put together has not convinced much of the research community in that the methods and models enable us to make the expected progress in understanding human pathological anxiety, and in providing us with tools for preventing it or to mitigate it. Hence this other section of the paper that differs from the previous to a significant degree as it is devoted to controversies. 3.1. “Know thyself”: how do we handle human anxiety makes a difference Anxiety is not easy to describe and to connect with observable human behaviour: The many changes in treatment of anxiety in the psychiatric classifications throughout their history, including the most recent switch from DSM-IV to DSM-5 in 2013 that replaced some of the “former” anxiety disorders, make excellent witnesses. It is challenging for the animal experimenter to “tie the complex and poorly understood symptomatology in humans to observable behaviors in animals” [92]. Anxiety is hard to read out from human behaviour, but the behavioural repertoire and facial expressions of rodents are even more limited, hence the use of behavioural tests. Any behavioural readout resulting from a task challenging for the animal is however likely to involve simultaneously a constellation of different drives and tap a variety of brain processes (Fig. 3). Therefore conceptual clarity in what exactly is supposed to be observed, and how this relates to other relevant constructs, becomes important. While a model of the disorder as a nosological entity is highly desirable, this may not meet all expectations, and some of the reasons for failure are basic.

2.1.2.2. Developmental and environmental models. Developmental and environmental models can not principally differ as the developmental models have at some period included an environmental intervention, and any environmental intervention brings about pathogenetic changes; what may be used as a distinction is the relative age of the animals if the alterations after reaching adulthood are not considered developmental. In anxiety research by far the most broadly accepted technique has been maternal separation. As described in 2.1.1.8, separation calls indicate anxiety in young animals and can be used as an anxiety test, including the studies on drugs. However, if mouse or rat pups are repeatedly separated from the dam, they will display anxietylike behaviours as adults [124]. Separation anxiety paradigm has a conceptually different variant with early weaning that itself comes in a variety of modifications, with similar variation in behavioural outcomes [125,126]. While maternal separation could be considered an intervention establishing general aspects of anxiety, models of specific disorders are also in use. Post-traumatic stress disorder, whatever its current status in diagnostic classifications, is a clinical condition that involves aspects of anxiety and that is produced by stimuli that are in use of anxiety tests. Animal models of PTSD apply stressors to produce a state relevant to the psychological trauma in humans, and guidelines for model development have been compiled [127]. Physical stressors include restraint, underwater holding and electric shock [128]. Restraint or immobilization stress procedures [129] can lead to behaviour reflecting anxiety in several conventional tests. A variant is brief underwater

Factors moderaƟng neurobiology underlying A readout

Test/Model for A

for B

Fig. 3. Anxiety tests and models in relation to other domains of CNS activity and behaviour. For simplicity, the relationship of only two constructs, A (anxiety) and B (say, depression) is depicted. The constructs more often than not have a common ground. A test or model is a well-described approximation of aspects of a construct that aims at being as specific as possible to the construct in question. Intrinsic and extrinsic factors that impact on the neural circuits relevant to behaviour in test for A can (rarely) make the test even more pertinent to the construct than original, or rather carry it away from its target, or shift the method toward detection of the essence of related construct (here, B).

5

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

community appears to believe that animals are anxious in the anxiety tests so that such a sharp separation between measurements in pathogenetic models and in drug screening tests is not a necessity. But even if this assumption is correct, and can be explicitly shown in validation experiments, anxiety tests and even the anxiety models should not be easily taken as models of anxiety disorders. If the human brain comprises unique aspects, so must do its disorders. Elaborating the proposal of Mathysse [145] that animal models in psychiatric research should apply psychological processes in a cross-species manner and not to attempt full psychiatric disorder modeled in non-human species, measures of specific psychophysiological processes have been suggested as the readouts in translation. As such processes, startle sensitivity, CO2 sensitivity and fear generalization have been proposed [14]. This essentially endophenotype-based approach can be complemented with other processes and measures covering e.g., impaired extinction of aversive memories, sleep disturbances, autonomic hyperarousal and avoidance of potentially risky territories [8]. The practical importance of bearing in mind the difference between anxiety and anxiety disorders is highlighted by the controversy of anxiolytics vs. anti-anxiety drugs. The term “anxiolytic” was a development following the “minor tranquillizers” (later simply “tranquillizers”), as drugs such as buspirone came to use that did not tranquil like barbiturates and benzodiazepines. Subsequently it however became evident that drugs classified as antidepressants (as they were developed primarily against affective disorders) are clinically superior in anxiety disorders, and these became the first-line treatment. This was not at all predicted by animal experiments; indeed, acute treatment with many antidepressants produces anxiogenic-like effects in a number of anxiolytic screening tests [146]. But indeed antidepressants do not have the immediate tranquillizing effect that characterizes the anxiolytic drugs, proving a clear separation for anxiety tests from anxiety disorder models. Parenthetically, several investigators have acutely applied low doses of antidepressants in order to validate their novel anxiety tests, despite of the absence of evidence that such a treatment is anxiolytic in humans. If a patient receives a psychiatric diagnosis, it is often not the only one: Current diagnostic systems promote co-morbidity. Co-morbidity probably is a real phenomenon that results from the multi-level alterations in the CNS starting from genetic vulnerability and non-specific and specific effects of lifetime environmental impacts. It has however become an argument in animal modeling, as co-morbidity in psychiatry has been exploited to promote off-label use of models, e.g., advocating the spontaneously hypertensive rat alternatively as a model of panic disorder or ADHD or depression. In a rigorous approach each animal model should be developed for one disorder. Yet, co-morbidity of anxiety disorders and affective disorders is so exceedingly high [147] that one may question the separation line between corresponding animal models altogether (Fig. 4). A more pressing issue is making distinction between normal, adaptive vs. pathological anxiety. The models we commonly use appear to reflect anxiety within the range that is entirely normal, not

One form of the discontent with the status of anxiety tests and models has been clearly expressed and explained by Rodgers [137] who referred to labeling of many procedures as anxiety models with the following criticism: ”… This is an unfortunate misnomer, not only because of the apparent inability of many tests to detect novel anxiolytics consistently, but also because the term implies that anxiety is a unitary emotion." Indeed, animal models have been derived from theoretical considerations that are not entirely mutually exclusive but nevertheless differ to an extent that they lead to very distinct experimental approaches. For example, the difference between Gray’s behavioural inhibition and activation systems based approach [138], the workings of Davis [73] and LeDoux [139] on fear conditioning, and the affective neuroscience concept of Panksepp [80] are easily distinguishable and rather contrasting not only in their explanation how behaviours emerge but also in the outlined neuroanatomy for fear and anxiety. Indeed, important differences must exist in neurobiology behind behaviour in principally different tests. For example, the septal-hypothalamic pathways appeared as critically involved in defensive burying of the shockprobe but not in plus-maze behaviour, suggesting that defensive responses toward discrete aversive stimuli and toward potential threats can be distinguished with these tests [140]. What these and other conceptual approaches would agree with, however, is the principle that the anxiety response in the CNS is an important mechanism by which we adapt and respond to real dangers [9]. But this mechanism becomes activated also in a variety of conditions where real dangers are not present and hence the heterogeneity in anxiety-like behaviour and underlying brain activities. Much effort has been made to distinguish anxiety and fear, or generalized context-related anxiety from fear derived from a distinct threat. A recent review concluded, however, that the majority of studies have nevertheless failed to differentiate fear and anxiety in terms of behaviour [141]. It is probably rather in terms of neurophysiological measures that will eventually permit distinction, especially as both fear and anxiety are dynamic, and neither can probably exclude the presence of the other. Even less has been elaborated on potential distinction of anxiety and panic fear that would be translatable to human panic disorder in its unclear origin and often predominantly vegetative symptomatology. Here Panksepp [80] has drawn a few clear lines and aligned the neurobiological panic system expressed in separation calls with the depression neuropathology. Parenthetically, in a large population representative human sample we could, by using principal component analysis, find the common fears grouping into two clusters that highly resembled what would have been predicted by the FEAR and PANIC systems [142]. In preclinical research on depression, a clear distinction is made between models of depression and antidepressant-screening tests. (Having said this, such a statement of course ignores the important issue of modeling vulnerability to depression [143,144], and the concern over the very fact that many depression researchers keep labeling such drug screening tests as tail suspension and the Porsolt's test as models of depression [15]). With anxiety it is different: The research

Fig. 4. Messy tests and models that can confuse or inform. A. When relationship of two (or more) conditions, e.g., anxiety and affective disorders, involves overlapping aspects, animal models may include aspects of each and give rise to eventually misleading predictions. B. Animal models can purposefully target the common aspects of two (or more) conditions. In case of anxiety, the potential anti-anxiety drugs should be effective in tests that reflect the common aspects of anxiety and affective disorders.

6

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

of anxiety tests because representatives of separate classes most probably measure different neural functions. Factor analysis of such principally different anxiety tests as the elevated plus-maze, Vogel test and social interaction test did not suggest a universal anxiety factor [154]. Behaviour in the social interaction test does not correlate well strainwise with performance in other tests of anxiety [155]. Inconsistencies abound however even while comparing tests based on a single behavioural domain such as exploration: For example, rearing activity of rats in an open field test does not correlate with their performance in the elevated plus-maze test [156]. It may be worthwhile to look into these correlations using methods that permit non-linear relationships to be discovered [157].

pathological [148]. If anxiety models are mostly taxing the coping resources [149], a critically important task is to reveal how true pathological anxiety overruns these resources. Hence, what can help in understanding how anxiety works is the identification of neural responses to threats that need to be coped with, and the range of responses that reflect successful coping. Fanselow has outlined distinct neurobiological systems that respond to nociceptive stimuli, unconditioned stimuli, conditioned stimuli, and species-specific danger signals [150]. This and potentially other classification systems can be helpful in systematic research that could eventually describe what is common in these different stimuli all leading to what we recognize as pathological anxiety. To this end we need to rely on well-validated anxiety tests, however. Of course, this way of reasoning is not excluding the possibility of drug screening tests that would be specific for any given anxiety disorder. Nevertheless, in order to show such a test to possess predictive validity we would need to compare ranges of drugs with efficacy different across the disorders, and this is yet to be done. Meanwhile, the neuroscience community is exploiting a variety of tests that are based on concepts with limited common ground besides trivial. These tests, in turn, are used in many modifications, and little attention is paid in publications as to why a particular test was selected for a specific experiment. A few years ago Haller and Aliczki carried out a most needed systematic review on studies using anxiolytic drugs [151]. They defined the “classical era” of testing that included elevated plus-maze and earlier developed methods and counted fourteen such tests (open field, social interaction, light-dark box, holeboard, elevatedplus-maze, Geller-Seifter test, Vogel test, four plate, defensive burying, hyponeophagia, distress vocalization, active/passive avoidance, fearpotentiated startle, and conditioned fear). Subsequent years had brought many further tests but in the analysis period of 2010–2011, the “classical” tests were used in more than 80% of studies; nevertheless, in many cases under unconventional conditions and with amendments that were mostly specific to the laboratory. Because modifications bring about major changes in anxiety behaviour [152], the heterogeneity in findings is approaching unmanageability. The large variety of available tests could also be seen as a resource, until a more general consensus is reached how to conduct the measurement of anxiety. Use of anxiety test batteries instead of single tests has been advocated for [153]. Indeed, given that predictions from tests differ and individual performances co-vary poorly across tests, this appears reasonable for the time being, because the probability of false negative findings would otherwise be high. However, we ought to ponder how many of false positive findings we can afford without losing the credibility, and how will the findings from batteries of tests, that are bound to bring in contradictory results, be interpreted so that mere cherry-picking were prevented. As the researchers have become ever more creative in use of the variety of known anxiety tests and in further modifying these, the consensus on interpretation of the findings obtained with the test batteries will become a target on its own. How would one decide on the value of a potential anti-anxiety drug if it appears as anxiolytic-like in one or two out of say, eight anxiety tests? The preference for Type 1 instead of Type 2 error is not always reasonable, but in such an instance the researcher may wish to study a broader dose range or modify experimental design for those tests where the drug has initially been ineffective. It is also important to look at the effect size of the positive finding, and, of course, to replicate.

3.2.1. Always more than anxiety Each anxiety test taps from a number of CNS domains. Fortunately it has become increasingly common to examine the learning abilities and perception (e.g., vision, hearing, and olfaction) in animals tested for anxiety as these cognitive abilities may interfere with anxiety readouts. Effects on memory and learning abilities interfere with anxiety-related behavior in multiple ways [158,159]. This is particularly important with genetically modified animals that are being made available in large variety. Owing to the pleiotropic effects of genes the probability of emergence of unexpected confounding factors is significant. What may need more consideration is that the neurobiology of fear is inevitably intertwined with neurobiologies of other emotive systems. For example, the conflict-involving conditions may elicit frustration and hence evoke not only anxiety but, even more prominently, anger, and thus recruit a distinct neural circuit. Perhaps a most telling case of emotive conflict is in the pattern emerging from tests that measure exploratory behaviour. Indeed these comprise the majority of animal anxiety studies, yet it has been left imprecisely defined exactly when a change in exploratory behaviour is reflecting anxiety. Exploratory behaviour appears as a reflection of conflict situation between neophobia (fear, anxiety, negative emotionality) and exploratory drive (seeking, reward sensitivity, positive emotionality). The balance between neural activities leading to expression of these fundamental traits is shifted by a variety of drugs in a predictable manner but not independent of other internal and external variables that are less well controllable. It may come as a surprise to many how recent is the use of the open field test for the measurement of “anxiety” and for anxiolytic drug screening. In earlier years, the open field test was used to measure the construct of “emotionality” that was its original destination [160]. The rationale of the open field test was that many mammals freeze when exposed to strange stimuli, and they may defecate [23]. But the advent of open field test as an anxiety measure has rather been based on measurement of units of locomotion that had previously been convincingly shown to be a very unreliable correlate of “emotionality” [25,26]. The open field test had been criticized as a forced exploration test, one that offers no behavioural choice measure, and measures the whole adaptive spectrum through essentially one channel, hence ignoring the existence of alternative approaches to novelty. Therefore, tests that emphasize free choice to enter observation area such as the Hughes' box [161] and the free-exploration paradigm by Griebel and colleagues [162] have been put to use. While these tests have a better rationale from the ethological perspective, they have remained underused in anxiety research, possibly because their applicability in drug screening has remained limited. The free choice tests appear to draw heavily from trait anxiety [163] but this has recently received too little attention. The potential shift in balance between concurring drivers of behaviour makes proper pharmacological validation of any test conditions absolutely necessary. How drug effects on exploration drive confound anxiety measurements can be exemplified in the case story of studies on neuropeptides in anxiety. The highly promising initial findings suggestive that cholecystokinin (CCK) receptor antagonists could present a principally novel type of anxiolytics [164] led many pharmaceutical

3.2. Toward universally accepted (batteries of) anxiety tests There is more than one way to build a framework to the available anxiety tests and models. Panksepp [80] classified the tests by two factors, use of punishment and use of learning in the test. Cryan and Sweeney [9] divided anxiety tests into ethological tests, conflict tests, cognition-based tests, physiological tests, vocalization-based tests and hyponeophagia. Such classifications will be useful for devising batteries 7

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

anxiolytic properties, and this paradigm also predicted that CRF1 receptor antagonists can rather affect general adaptive strategies than to possess a true anxiolytic effect [182,183].

companies to test novel non-peptide CCK antagonists, eventually with no gain. This contrasted heavily with the many independent reports on the remarkable anxiolytic potential of both CCKA and, more frequently, CCKB receptor antagonists in animal models. Nevertheless, the drugs had been effective mostly in exploration-based tests only. Furthermore, receptor subtype selective CCK receptor antagonists were entirely ineffective against the anxiogenic action of low dose of picrotoxin [165], no convincing evidence was provided that benzodiazepine anxiolytics would modify the state produced by CCK agonists, and the studies that could find anxiolytic affect of CCK receptor antagonists on their own had used animals with low baseline anxiety which is contra-intuitive for an anxiolytic drug [166]. Truly anxiogenic effects such as produced by reduction of NPY Y1 receptor-mediated signaling are readily antagonized by benzodiazepines at low doses [167]. In hindsight, the known potential of interference of the potent regulatory effect of CCK on dopaminergic neurotransmission [168] and consequent modification of locomotor activity [169] was overlooked, and false conclusions drawn. The relationship of baseline activity levels with the validity for anxiety measurement is probably underestimated. Non-peptide CRF1 receptor antagonists was another case of large commercial investment but in animal experiments their efficacy was largely limited to conditions of excessive CRF activity [170] and the clinical efficacy was eventually not found [171]. Another case when this issue has surfaced was the difficulty with interpretation of plus-maze behaviour after repeated administration of MDMA: Wistar rats with low anxiety levels responded to 5-HT-ergic lesion with an apparent anxiety response, while the Dark Agouti rats, with high innate anxiety, displayed an apparent anxiolytic effect [172]. It has recently been shown that the effect of MDMA on plus-maze behaviour can also vary dramatically from anxiogenic-like to highly anxiolytic-like dependent on changes in feeding schedule [173]. Indeed, in a variety of other conditions that by construct are unlikely to represent reduced anxiety, e.g., sensitization to nicotine [174], the traditional readouts of plus-maze behaviour would permit such a conclusion. In juvenile stroke-prone spontaneously hypertensive rats not only were ambulation and rearing in an open field higher than in the comparison strain, but also entries into open arms of the plus-maze, making the authors to suggest this as a measure of impulsive behaviour [175]. Impulsive responding in a novel environment that provokes anxiety can be read out as anxiolytic effect [176]. Indeed, many peculiar findings in behavioural tests are all compatible with enhanced impulsivity that is strongly associated with reduced serotonergic function. This could explain another controversy: While sleep deprivation produces anxiety, but also impulsivity, in humans, it has a consistent anxiolytic-like profile in animal experiments, the majority of these conducted using the plus-maze test and indeed almost all of the rest exploring other tests based on exploratory behaviour [177]. In turn, drugs that act on serotonergic targets have shown highly divergent profiles if examined in a variety of principally different tests of anxiolytic activity, and inter-individual differences matter. For example, eltoprazine, a drug acting at several subtypes of the 5-HT1 and 5-HT2 receptors as either full or partial agonist can produce a full range of effects in anxiety tests from anxiolytic to anxiogenic [178], and ketanserin, a 5-HT2A antagonist, produced opposite effects in Carioca High- vs. Low-Conditioned rats [179]. One robust example of exploration choice tests providing a clearly different prediction as compared to elevated plus-maze and other forced exploration tests is the exploration box test [68,180] that has been developed to take into account distinct aspects of measurement of exploratory behaviour in order to more fully characterize acute and repeated psychotropic drug action. Effect of noradrenergic denervation had proved to be hard to detect consistently in tests of anxiolytic/anxiogenic activity [181], but in the exploration box the rats with selective locus coeruleus denervation by DSP-4 treatment were uniformly non-exploratory and this effect was completely eliminated by administration of a low dose of diazepam [180]. Furthermore, effect profile of a CCKB receptor antagonist was very clearly differentiated from the

3.2.2. Anxiety, anxiety disorder, and vulnerability to anxiety All anxiety disorders involve anticipation of future threat, they are highly comorbid [184], and include dimensional aspects [185]. At the same time it is well acknowledged that anticipation of future threats has adaptive value, hence it is of concern only if appearing in somehow inappropriate ways. In acute experiments we can not know whether a rat in plus-maze is experiencing adaptive or maladaptive anxiety, and hence the anxiolytic drugs predicted with anxiety tests are likely to inhibit both. How to make a distinction in animal tests is another challenge for the future. The brain mechanisms of excessive acute anxiety and persistently excessive anxiety termed as anxiety disorder are likely different, and these differences can be captured by neurobiological studies addressing anxiety tests and anxiety models in comparison. Nevertheless, something else appears in animal experiments that is representing neither non-adaptively deep acute condition of anxiety nor aspects of anxiety disorder. This aspect becomes visible in higher expression of anxiety in test situations and can be categorized in some instances as trait anxiety and in others as vulnerability to anxiety. In human studies, trait anxiety predicts higher probability of anxiety (and affective) disorders [186] and has been proposed as an integrative component in conceptual model building for prediction of anxiety disorders [187]. Trait anxiety as proneness to experience anxiety, and vulnerability to anxiety disorders that may be an overlapping concept but does not necessarily include persistent expression of anxiety can serve as a resource for anxiety modeling, as proposed for diathesis-stress studies in depression [143,144]. On the contrary, trait anxiety can appear in anxiety tests as state anxiety, with unexpected consequences such as relatively low sensitivity to anxiolytics. 3.2.3. Validation stepwise Each test is expected to possess construct, face, and predictive validity. Beyond these common validity concerns, issues of ecological validity, response variability, robustness and specificity require attention. When reading the reported findings we judge over their value on the basis of whether the method appears as valid or not. This is a categorical decision but admittedly the level of validation comes not only from touching on all or most of the traditional aspects of validity but also the strength and extent of measures taken to validate. This varies enormously for novel procedures and is often very weak for their modifications. The scope of this review does not permit any detailed discussion of all aspects of validation, but it may be worth of taking into consideration what is the object of validation. That is, tests and models for the different purposes may have different validation focus. This is obvious if we think of the symptoms of, say, schizophrenia and anxiety disorders, because for the former the core symptoms are not accessible by observation of spontaneously emerging behaviour. Not surprisingly, a psychopharmacologist would propose that the most critical validation experiments come from the pharmacological approach. Drugs that have relatively specific effects on anxiety, at least in a well-selected dose range, are many and can be easily compared with each other and with other psychoactive compounds. Most importantly, available drugs include anxiolytics as well as anxiogenic compounds, making it feasible to elicit effects in either direction and hence reducing the possibility of unspecific effects considered as anxiety-related. The pharmacological validation has its own, oft-ignored pitfalls. For anxiety tests, only drugs that have acute tranquillizing effect can be used, and those that possess anti-anxiety effect in clinical studies after weeks of administration should be administered appropriately, pharmacokinetics considered, and are applicable rather for anxiety models than tests. This means that antidepressants, despite of their first-line 8

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

pre-existing factors also potentially interact with the treatment applied, so there is always minimally a two-way interaction taking place. As Denenberg has put it, “As long as I work in this context, I feel secure in making these particular interpretations. However, this does not mean that these same conceptual meanings may be attributed to open-field performance in other research contexts” [23]. This probably applies to other methods to measure anxiety as well. Thus, any method validated in well-defined conditions is possibly good, and any often-used method taken over in a hasty manner is likely to turn out invalid. Our feelings are not particularly stable, and the time-course of anxiety-relevant CNS states in animals is likely not to be either. This necessitates careful consideration of the time window for anxiety measurement. In pharmacological studies, an additional factor to consider is the dependence of the outcome on dose. Comparison of the effect of many drugs in a variety of anxiolytic drug screening tests has revealed that hormetic-like biphasic dose responses are common, with the magnitude width of the response similar for each drug across tests and the testing conditions under consideration [195]. Treatments may have floor effects and ceiling effects, so the range of behavioural readout that can be expected should be known. The broader the range, the better, but within ethologically reasonable constraints: For example, it makes no sense to prefer a potentially dangerous area, so the animal that excessively ventures to open arms of an elevated plus-maze may not be very tranquil but rather simply out of its mind. All these precautions lead to the need of optimization of the implemented anxiety test. Throughout the optimization process that begins with learning about the factors that can not be modified in the laboratory and with selecting the strain, one has to bear in mind the risk of over-optimization, changing the conditions to an extent that the behavioural readout would not any longer reflect the originally targeted construct. Practical considerations require the possibility of reusing animals in the anxiety tests. While methods exist to reduce the potential confound from retesting, some studies have provided evidence that second exposure to the same test can be qualitatively different [13], or the size of treatment effects may change systematically [56,196]. As the experiments of Denenberg [23] demonstrated, repeated testing in open field conditions reveals that behaviour during the first exposure to an environment of uniquely different from the subsequent performances, reflecting something called emotional reactivity and strongly associated with defecation measures. Of importance here is that activity in an entirely novel environment appears as a mixture of flight and freezing responses, so that the more anxious non-handled animals can present high activity in response to stress that is reflected in the rise of corticosterone levels but measures of locomotion are not reliable. In contrast, the social interaction test possibly can be repeated in the same animals, probably because sociability is a persistent trait but its expression is modified by the partner animal [53], so there is always a novelty factor involved.

role in anxiety management, should not be used as acute treatment in animal experiments. It is advisable to induce state of anxiety with a standard procedure before trying the effect of an anxiolytic, because otherwise the anxiety measure may tap rather the trait anxiety less malleable by acute drug treatment. Drugs have therapeutic window so if first used in a given set and setting a range of doses needs to be used. Inclusion of negative control drugs is important while bearing in mind that a drug not considered anxiety treatment owing to their unbearable side effects (e.g., opiates and ethyl alcohol) may still have acute anxiolytic effect! At low doses, the anxiogenic drugs, by their activating properties, may increase activity even in the central quadrates of an open field [188]. 3.3. The challenge of measurement 3.3.1. Recording and confounders When we have arrived at actual measurement of possibly anxious behaviour, the training of experimenters is very important, as well a healthy routine in the experimental setting that prevents fluctuations in the environment in the broadest sense. Taking the measures involves subjective factors. These are more obvious in case of direct observation and measurement of inter-rater reliability would be a good practice. What is less obvious is that interrater reliability is not so much about the measure but the measurer [189]. Most novices can reliably record highly complex behaviours such as those displayed in the social interaction, but some observers have a biased focus even after years of experience in studies involving laboratory animals. An experimenter is also limited in how many distinct behaviours can be simultaneously observed. It is hard to deny that more information offers more possibilities, and with automated recording and computerized data analysis the apparent treasure trove of measures is opened, and the subjective aspects apparently mitigated. While this has undeniable advantages [152,190,191], there are also a few issues. Automatization of recording is cost-effective but may provide a somewhat different readout as compared to the originally intended. The bias may be minor in baseline conditions but grow to make a difference after intervention. The benefit of ability to record enormous number of variables can be offset by limited insight of the experimenter of what really happens [192]: behaviours compete with each other, they vary with regard of dynamics, and much of this may be at random and not helpful. Trivial as it sounds but the animal can show only a limited set of behaviours at a time, and it may not be so important, which. A number of factors can influence response to the test [189]: internal such as genetic background, sex, health condition, age, weight, body composition, nutrition, gut microflora, and external such as season, time of the day, diet, sensory stimuli, caging and density in cage and resultant activity levels, temperature, humidity, handling and other potential effects of the experimenter. Animal houses have tremendously changed in recent years as laboratory animal science has made maintenance a research field of its own and additional regulations are being applied. This is likely to have an impact that may change the comparability with some results obtained in the past. Caging systems, including the ventilation of the cages, has found to affect basal levels of behaviour in anxiety tests [193]. Whether testing has proceeded during the dark or the light phase can make an impact [194], but this may further depend on what have been the light conditions in living quarters of the animal house. Thus, we are aware of a large variety of environmental factors that have the potential to interfere with expression of anxiety-like states, and these may produce even more confounds by interacting with each other, and with the treatment factor. Besides the known confounders we are only beginning to consider the “unknown unknowns”, and in common practice certainly do not as yet pay sufficient attention to several important factors such as smell, and even less on ultrasound that has found increasing use as a measure of CNS conditions in rodents. The

3.3.2. The individual behind the measurement Anxiety studies have highlighted the importance of several stratifying factors. In many experiments, strain (and often substrain) has had a significant effect, including e.g., on the expression of startle response and in exploration-based tests such as holeboard, light/dark compartment, and elevated plus-maze [194,196,197]. While the strain has become increasingly better controlled, including as a background confound in genetic studies, there is an emerging theme in neuroscience with regard to profound differences in male and female brain. Male and female rodents robustly differ in a variety of behavioural tests, and this has impact on responses to drugs [198]. For example, male and female rats have important differences in exploratory behaviour while the expression of these differences depend on external factors such as the length of exposure [199]. In general, females have been found more explorative [180,200,201]. Similar findings have been obtained with hamsters and it has been proposed that females habituate to novelty 9

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

applying a modification of a test, conducted in insufficiently characterized unique conditions, and earning the access to publicity mainly by presenting group level differences that meet the conventional level of statistical significance. In recent years, much has changed with regard to scrutiny with which the use of statistics is assessed [217]. While vitally important, nominal statistical significance can not serve the role reserved by laws of nature for replication experiments.

more rapidly [202]. This sex difference also holds for common exploration-based anxiety tests such as elevated plus-maze while females appeared as more anxious in the social interaction test and in a Vogel test [203]. These differences may be of great interest for the study of underlying neurobiology but certainly of the validity of the test. In the same experimental setting, inescapable shocks affected behaviour of male and female rats in holeboard and elevated plus-maze differently [204], as does chronic social instability [205] or maternal separation [206]. The latter can also have profoundly different effect on male and female offspring if tested as adults [207]. Other stratifications exist that have remained in the realm of ‘unknown unknowns’, and these express themselves as unexplained persistent quantitative or qualitative inter-individual differences in responding. Suggestions to take such inter-individual differences into account in neuropsychopharmacology [41,208] had however remained rather non-influential until the differences in the response social defeat [209] became well recognized. The inter-individual variability in anxiety tests is systematic [210] and can strongly interfere with the effect of a variety of psychoactive drugs [211]. These differences, owing to their neurobiological substrates [148,212], can contribute to better understanding of the varieties of anxiety. Inter-individual differences are of interest to the extent their emergence can be explained and predictive value for future behaviour can be harvested. It is in this realm where the answer to the divergent stress responses should probably be found. The contribution that adversities during childhood make to the later incidence of anxiety disorders [213] has led to approaches aiming at producing anxiety models by applying stress during post-weaning and pre-pubertal stages. It has however been found that some aspects of juvenile stress rather facilitate the development of resilience [214]. Behavioral profiling after underwater-trauma has been reported to yield distinct phenotypes of stress response [130] and such profiling attempts appear as a highly promising strategy.

Acknowledgements Relevant research in the author’s laboratory has been supported by the Hope for Depression Research Foundation and the Institute for the Study of Affective Neuroscience, the EC 6FP Integrated Project NEWMOOD, and the Estonian Ministry of Education and Science project IUT20-40. References [1] O. Remes, C. Brayne, R. van der Linde, L. Lafortune, A systematic review of reviews on the prevalences of anxiety disorders in adult populations, Brain Behav. 6 (2016) e00497, http://dx.doi.org/10.1002/brb3.497. [2] R.C. Kessler, P. Berglund, O. Demler, R. Jin, K.R. Merikangas, E.E. Walters, Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication, Arch. Gen. Psychiatry 62 (2005) 593–602. [3] A. Gustavsson, M. Svensson, F. Jacobi, C. Allgulander, J. Alonso, E. Beghi, et al., Cost of disordes of the brain in Europe 2010, Eur. Neuropsychopharmacol. 21 (2011) 718–779. [4] A.A. Shah, J.Y. Han, Anxiety, Continuum 21 (2015) 772–782. [5] L. Culpepper, Generalized anxiety disorder and medical illness, J. Clin. Psychiatry 70 (S2) (2009) 20–24. [6] G. Perna, G. Iannone, A. Alciati, D. Caldirola, Are anxiety disorders associated with accelerated aging? A focus on neuroprogression, Neural Plast. 206 (2016) 8457612, http://dx.doi.org/10.1155/2016/8457612. [7] M.J. Millan, G.M. Goodwin, A. Meyer-Lindenberg, S.O. Ögren, Learning from the past and looking to the future: emerging perspectives for improving the treatment of psychiatric disorders, Eur. Neuropsychopharmacol. 25 (2015) 599–656. [8] J.F. Cryan, A. Holmes, The ascent of mouse: advances in modelling human depression and anxiety, Nat. Rev. Drug Discov. 4 (2005) 775–790. [9] J.F. Cryan, F.F. Sweeney, The age of anxiety: role of animal models of anxiolytic action in drug discovery, Br. J. Pharmacol. 164 (2011) 1129–1161. [10] S. Frantz, Therapeutic area influences drug development costs, Nat. Rev. Drug Discov. 3 (2004) 466–467. [11] A. Ramos, Animal models of anxiety: do I need multiple tests, Trends Pharmacol. Sci. 29 (2008) 493–498. [12] J. Haller, M. Aliczki, K.G. Pelczer, Classical and novel approaches to the preclinical testing of anxiolytics: a critical evaluation, Neurosci. Biobehav. Rev. 37 (2013) 2318–2330. [13] M. Bourin, Animal models for screening anxiolytic-like drugs: a perspective, Dialogues Clin. Neurosci. 17 (2015) 295–303. [14] K.L. Hoffman, New dimensions in the use of rodent behavioral tests for novel drug discovery and development, Expert Drug Opin. Discov. 11 (2016) 343–353. [15] S.C. Stanford, Confusing preclinical (predictive) drug screens with animal models of psychiatric disorders, or disorder-like behaviour, is undermining confidence in behavioural neuroscience, J. Psychopharmacol. 31 (2017) 641–643. [16] G.W.F. Hegel, Elements of the Philosophy of Right, Cambridge University Press, 2017. [17] A. Stewart, S. Gaikwad, E. Kyzar, J. Green, A. Roth, A.V. Kalueff, Modeling anxiety using adult zebrafish: a conceptual review, Neuropharmacology 62 (2012) 135–143. [18] K.M. Khan, A.D. Collier, D.A. Meshalkina, E.V. Kysil, S.L. Khatsko, T. Kolesnikova, et al., Zebrafish models in neuropsychopharmacology and CNS drug delivery, Br. J. Pharmacol. 174 (2017) 1925–1944. [19] I. Geller, J.T. Kulak Jr., J. Seifter, The effects of chlordiazepoxide and chlorpromazine on a punishment discrimination, Psychopharmacologia 31 (1962) 374–385. [20] G.T. Pollard, J.L. Howard, The Geller-Seifter conflict paradigm with incremental shock, Psychopharmacology (Berl) 62 (1979) 117–121. [21] J.R. Vogel, B. Beer, D.E. Clody, A simple and reliable conflict procedure for testing anti-anxiety agents, Psychopharmacologia 21 (1971) 1–7. [22] M.J. Millan, M. Brocco, The Vogel conflict test: procedural aspects, gamma-aminobutyric acid, glutamate and monoamines, Eur. J. Pharmacol. 463 (2003) 67–96. [23] V.H. Denenberg, Open-field behavior in the rat: what does it mean, Ann. N. Y. Acad. Sci. 159 (1968) 852–859. [24] S.D. Gosling, From mice to men: what can we learn about personality from animal research, Psychol. Bull. 127 (2001) 45–86. [25] J. Archer, Tests for emotionality in rats and mice: a review, Anim. Behav. 21 (1973) 205–235. [26] R.N. Walsh, R.A. Cummins, The open-field test: a critical review, Psychol. Bull. 83 (1976) 482–504.

4. What should we do today? Many of the guidelines above require much work and the fruits will be available some time in future. There is no way we could postpone the immediate research questions that can be answered only by using anxiety tests and models until all the critical voices have become silent. Trivial as it may sound, in the first place it is important to be explicit in the goal why measure anxiety in animals: Is it for drug development, elucidating neurobiology of anxiety, increase of awareness of confounding variables, or overall behavioural profiling of a new genetic variant. If the intention is to eventually treat anxiety disorder or anxiety symptoms then it would be wise to use a panel of anti-anxiety drugs for full pharmacological profiling. If understanding of anxiety is the aim, then one has to compare performance over a spectrum of motivational conflict types and levels. In case there is a wish to develop an entirely new model of anxiety, first specify its primary intended use: for drug screening or study of pathophysiology, state or trait, fear or anxiety or panic, and then examine relationship with other emotive systems such as anger and curiosity. Phenotyping a new genetically modified mouse would initially benefit of most often used tests to facilitate comparison. The specific conditions for performing the test should be brought to awareness and kept constant. If the latter ceases to be feasible, like while moving to a new animal quarters or inevitable switching to another strain, one has to define and perform critical control experiments. Technological developments now allow to increase the usefulness of physiological measures in anxiety studies [215,216], and telemetric observation of physiological signs of anxiety, at least during the validation phase, should grow in importance. While the physiological measures of anxiety are not entirely reliable, they are not worse than behaviour and provide an additional dimension. It should also be acknowledged that too much of information on anxiety is hanging on a very thin evidence: findings from an experiment 10

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

[61] J.C. Stout, J.M. Weiss, An animal model for measuring behavioral responses to anxiogenic and anxiolytic manipulations, Pharmacol. Biochem. Behav. 47 (1994) 459–465. [62] S.R. Bodnoff, B. Suranyi-Cadotte, R. Quirion, M.J. Meaney, A comparison of the effects of diazepam versus several typical and atypical anti-depressant drugs in an animal model of anxiety, Psychopharmacology (Berl) 97 (1989) 277–279. [63] Z. Merali, C. Levac, H. Anisman, Validation of a simple, ethologically relevant paradigm for assessing anxiety in mice, Biol. Psychiatry 54 (2003) 552–565. [64] S.C. Dulawa, R. Hen, Recent advances in animal models of chronic antidepressant effects: the novelty-induced hypophagia test, Neurosci. Biobehav. Rev. 29 (2005) 771–783. [65] J.A. King, T. Messenger, C.F. Ferris, Seed finding in golden hamsters: a potential animal model for screening anxiolytic drugs, Neuropsychobiology 45 (2002) 150–155. [66] D.J. Sanger, Animal models of anxiety and the screening and development of novel anxiolytic drugs, in: A. Boulton, G. Baker, M. Martin-Iverson (Eds.), Neuromethods: Animal Models in Psychiatry II, Vol. 19 Humana Press, 1991. [67] T.D. Goode, S. Maren, Animal models of fear relapse, ILAR J. 55 (2014) 246–258. [68] M.-H. Otter, V. Matto, R. Sõukand, T. Skrebuhhova, L. Allikmets, J. Harro, Characterization of rat exploratory behavior using the exploration box test, Method Find. Exp. Clin. Pharmacol. 19 (1997) 683–691. [69] J.S. Brwon, H.I. Kalish, I.E. Farber, Conditioned fear as revealed by magnitude of startle response to an auditory stimulus, J. Exp. Psychol. 41 (1951) 317–328. [70] C.C. Chi, The effect of amobarbital sodium on conditioned fear as measured by the potentiated startle response in rats, Psychopharmacologia 7 (1965) 115–122. [71] M. Davis, Morphine and naloxone: effects on conditioned fear as measured with the potentiated startle paradigm, Eur. J. Pharmacol. 54 (1979) 341–347. [72] M. Davis, Diazepam and flurazepam: effects on conditioned fear as measured with the potentiated startle paradigm, Psychopharmacologia 62 (1979) 1–7. [73] M. Davis, D.L. Walker, L. Miles, C. Grillon, Phasic vs sustained fear in rats and humans: role of the extended amygdala in fear vs anxiety, Neuropsychopharmacology 35 (2010) 105–135. [74] J.P.J. Pinel, D. Treit, Burying as a defensive response in rats, J. Comp. Physiol. Psychol. 92 (1978) 708–712. [75] D. Treit, J.P. Pinel, H.C. Fibiger, Conditioned defensive burying: a new paradigm for the study of anxiolytic agents, Pharmacol. Biochem. Behav. 15 (1981) 619–626. [76] S.F. De Boer, J.M. Koolhaas, Defensive burying in rodents: ethology, neurobiology and psychopharmacology, Eur. J. Pharmacol. 463 (2003) 145–161. [77] C.L. Broekkamp, H.W. Rijk, D. Joly-Gelouin, K.L. Lloyd, Major tranquillizers can be distinguished from minor tranquillizers on the basis of effects on marble burying and swim-induced grooming in mice, Eur. J. Pharmacol. 126 (1986) 223–229. [78] K.U. Njung’e, S.L. Handley, Evaluation of marble-burying behavior as a model of anxiety, Pharmacol. Biochem. Behav. 38 (1991) 63–67. [79] N. Abelda, D. Joel, Animal models of obsessive-compulsive disorder: exploring pharmacology and neural substartes, Neurosci. Biobehav. Rev. 36 (2012) 47–63. [80] J. Panksepp, Affective Neuroscience The Foundations of Human and Animal Emotions, Oxford University Press, 1998. [81] C.R. Cloninger, A unified biosocial theory of personality and its role in the development of anxiety states, Psychiatr. Dev. 4 (1986) 167–226. [82] J.R. Boissier, P. Simon, C. Aron, A new method for rapid screening of minor tranquillizers in mice, Eur. J. Pharmacol. 4 (1968) 145–151. [83] M. Bourin, M. Hascoët, Implication of the 5-HT2A receptor subtypes in the mechanism of action of the GABAergic compound etifoxine in the four-plate test in Swiss mice, Behav. Brain Res. 208 (2010) 352–358. [84] L. Molinergo, S. Ricci-Gamalero, The staircase maze and the simple staircase in the analysis of the psychopharmacological action of CNS depressants, Pharmacology 4 (1970) 169–178. [85] J. Simiand, P.E. Keane, M. Morre, The staircase test in mice: a simple and efficient procedure for primary screening of anxiolytic agents, Psychopharmacology (Berl) 84 (1984) 48–53. [86] P.L. Toubas, K.A. Abla, W. Cao, L.G. Logan, T.W. Seale, Latency to enter a mirrored chamber: a novel behavioral assay for anxiolytic agents, Pharmacol. Biochem. Behav. 35 (1990) 121–126. [87] C.L. Kliethermes, D.A. Finn, J.C. Crabbe, Validation of a modified mirror chamber sensitive to anxiolytics and anxiogenics in mice, Psychopharmacology (Berl) 169 (2003) 190–197. [88] N.E. Paterson, M. Iwunze, S.F. Davis, S.A. Malekiani, T. Hanania, Comparison of the predictive validity of the mirror chamber and elevated plus maze tests in mice, J. Neurosci. Methods 188 (2010) 62–70. [89] M. Barros, R.S. Maior, J.P. Huston, C. Tomaz, Predatory stress as an experimental strategy to measure fear and anxiety-related behaviors in non-human primates, Rev. Neurosci. 19 (2008) 157–169. [90] L.G. Staples, Predator odor avoidance as a rodent model of anxiety: learningmediated consequences beyond the initial exposure, Neurobiol. Learn. Mem. 94 (2010) 435–445. [91] G. Griebel, D.C. Blanchard, A. Jung, J.C. Lee, C.K. Masuda, R.J. Blanchard, Further evidence that the mouse defense test battery is useful for screening anxiolytic and panicolytic drugs: effects of acute and chronic treatment with alprazolam, Neuropharmacology 34 (1995) 1625–1633. [92] J.M. Robertson, M.A. Prince, J.K. Achua, R.E. Carpeneter, D.H. Arendt, J.P. Smith, T.L. Summers, T.R. Summers, C.H. Summers, Nuance and behavioral cogency: how the visible burrow system inspired the stress-alternatives model and conceptualization of the continuum of anxiety, Physiol. Behav. 146 (2015) 86–97. [93] R.J. Blanchard, D.C. Blanchard, Antipredator defensive behaviors in a visible

[27] J. Harro, Measurement of exploratory behavior in rodents, in: P.M. Conn (Ed.), Methods in Neurosciences, Vol. 14 Academic Press, San Diego, 1993, pp. 359–377. [28] L. Prut, C. Belzung, The open field test as a paradigm to measure the effects of drugs on anxiety-like behaviors: a review, Eur. J. Pharmacol. 463 (2003) 3–33. [29] J.N. Crawley, F.K. Goodwin, Preliminary report of a simple animal behaviour for the anxiolytic effects of benzodiazepines, Pharmacol. Biochem. Behav. 13 (1980) 167–170. [30] M. Bourin, M. Hascoët, The mouse light/dark box test, Eur. J. Pharmacol. 463 (2003) 55–65. [31] J.R. Boissier, P. Simon, The exploration reaction in the mouse. Preliminary note. (In French.), Therapie 17 (1962) 1225–1232. [32] S.E. File, Are central cholinergic paths involved in habituation of exploration and distraction, Pharmacol. Biochem. Behav. 4 (1976) 695–702. [33] J.F. Deakin, S.E. File, J.R. Hyde, N.K. Macleod, Ascending 5-HT pathways and behavioural habituation, Pharmacol. Biochem. Behav. 10 (1979) 687–694. [34] S.E. File, A.G. Wardill, The reliability of the holeboard apparatus, Psychopharmacologia 44 (1975) 47–51. [35] S.E. File, Factors controlling measures of anxiety and responses to novelty in the mouse, Behav. Brain Res. 125 (2001) 151–157. [36] K.C. Montgomery, The relation between fear induced by novel stimulation and exploratory behavior, J. Comp. Physiol. Psychol. 48 (1955) 254–260. [37] S.L. Handley, S. Mithani, Effects of alpha-adrenoceptor agonists and antagonists in a maze-exploration model of fear-motivated behaviour, Naunyn-Schmiedeberg's Arch. Pharmacol. 327 (1984) 1–5. [38] S. Pellow, P. Chopin, S.E. File, M. Briley, Validation of open:closed arm entries in an elevated plus-maze as a measure of anxiety in the rat, J. Neurosci. Methods 14 (1985) 149–167. [39] R.G. Lister, The use of a plus-maze to measure anxiety in the mouse, Psychopharmacology (Berl) 92 (1987) 180–185. [40] F. Sherif, A. Harro El-Hwuegi, L. Oreland, Anxiolytic-like effect of the GABAtransaminase inhibitor vigabatrin (gamma-vinyl-GABA) on rat exploratory activity, Pharmacol. Biochem. Behav. 49 (1994) 801–805. [41] L. Rägo, R.A. Kiivet, J. Harro, M. Põld, Behavioral differences in an elevated plusmaze: correlation between anxiety and decreased number of GABA and benzodiazepine receptors in mouse cerebral cortex, Naunyn Schmiedebergs Arch. Pharmacol. 337 (1988) 675–678. [42] C. Lee, R.J. Rodgers, Antinociceptive effects of elevated plus-maze exposure: influence of opiate receptor manipulations, Psychopharmacology (Berl) 102 (1990) 507–513. [43] T. Sorregotti, J. Mendes-Gomes, J.L. Rico, R.J. Rodgers, R.L. Nunes-de-Souza, Ethopharmacological analysis of the open elevated plus-maze in mice, Behav. Brain Res. 246 (2013) 76–85. [44] J.C. Cole, R.J. Rodgers, Ethological evaluation of the effects of acute and chronic buspirone treatment in the murine elevated plus-maze test: comparison with haloperidol, Psychopharmacology (Berl) 111 (1994) 288–296. [45] E. Dere, B. Topic, M.A. De Souza Silva, M. Srejic, C. Frisch, T. Buddenberg, J.P. Huston, The graded anxiety test: a novel test of murine unconditioned anxiety based on the principles of the elevated plus-maze and light-dark test, J. Neurosci. Methods 122 (2002) 65–73. [46] M.B. Viana, C. Tomaz, F.G. Graeff, The elevated T-maze: a new animal model of anxiety and memory, Pharmacol. Biochem. Behav. 49 (1994) 549–554. [47] J.K. Shepherd, S.S. Grewal, A. Fletcher, J.D. Bill, C.T. Dourish, Behavioural and pharmacological characterization of the canopy stretched attend posture test as a model of anxiety in mice and rats, Psychopharmacology (Berl) 133 (1997) 29–38. [48] S.S. Grewal, J.K. Shepherd, D.J. Bill, A. Fletcher, C.T. Dourish, Behavioural and pharmacological characterization of the elevated zero-maze as an animal model of anxiety, Psychopharmacology (Berl) 116 (1994) 56–64. [49] V. Matto, J. Harro, L. Allikmets, The effects of cholecystokinin A and B receptor antagonists on exploratory behaviour in the elevated zero-maze in rat, Neuropharmacology 36 (1997) 389–396. [50] S.E. File, J.R. Hyde, Can social interaction be used to measure anxiety, Br. J. Pharmacol. 62 (1978) 19–24. [51] S.E. File, The use of social interaction as a method for detecting anxiolytic activity of chlordiazepoxide-like drugs, J. Neurosci. Methods 2 (1980) 219–238. [52] S.E. File, P. Seth, A review of 25 years of the social interaction test, Eur. J. Pharmacol. 463 (2003) 35–53. [53] M. Tõnissaar, M.A. Philips, M. Eller, J. Harro, Sociability trait and serotonin metabolism in the rat social interaction test, Neurosci. Lett. 367 (2004) 209–312. [54] C.R. Gardner, Distress vocalization in rat pups: a simple screening method for anxiolytic drugs, J. Pharmacol. Methods 14 (1985) 181–187. [55] L. Normansell, J. Panksepp, Glutamatergic modulation of separation distress: profound emotional effects of excitatory amino acids in chicks, Neurosci. Biobehav. Rev. 35 (2011) 1890–1901. [56] L. Groenink, P.M. Verdouw, B. Bakker, K.E. Wever, Pharmacological and methodological aspects of the separation-induced vocalization test in guinea-pig pups; a systematic review and meta-analysis, Eur. J. Pharmacol. 753 (2015) 191–208. [57] B. Olivier, E. Molewijk, R. van Oorschot, G. van der Poel, T. Zethof, J. van der Heyden, J. Mos, New animal models of anxiety, Eur. Neuropsychopharmacol. 4 (1994) 93–102. [58] C. Sanchez, Stress-induced vocalisation in animals: a valid model of anxiety, Eur. J. Pharmacol. 463 (2003) 133–142. [59] H.E. Covington III, K. Miczek, Vocalizations during withdrawal from opiates and cocaine: possible explanations of affective distress, Eur. J. Pharmacol. 467 (2003) 1–13. [60] D.R. Britton, K.T. Britton, A sensitive open field measure of anxiolytic activity, Pharmacol. Biochem. Behav. 15 (1981) 577–582.

11

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

[123] V. Võikar, S. Kõks, E. Vasar, H. Rauvala, Strain and gender differences in the behavior of mouse lines commonly used in transgenic studies, Physiol. Behav. 72 (2001) 271–281. [124] P.M. Plotsky, M.J. Meaney, Early, postnatal experience alters hypothalamic corticotropin-releasing factor (CRF) mRNA, median eminence CRF content and stressinduced release in adult rats, Mol. Brain. Res. 18 (1993) 195–200. [125] T. Kikusui, Y. Mori, Behavioural and neurochemical consequences of early weaning in rodents, J. Neuroendocrinol. (2009) 427–431. [126] M.C. Fraga, E.G. de Moura, N. da Silva Lima, P.C. Lisboa, F. de Oliveira, J.O. Silva, et al., Anxiety-like, novelty-seeking and memory/learning behavioral traits in male Wistar rats submitted to early weaning, Physiol. Behav. 124 (2014) 100–106. [127] A. Siegmund, C.T. Wotjak, Toward an animal model of posttraumatic stress disorder, N. Y. Acad. Sci. 1071 (2006) 324–334. [128] E.I. Flandreau, M. Toth, Animal models of PTSD: a critical review, Curr. Top. Behav. Neurosci. (2017), http://dx.doi.org/10.1007/7854_2016_65 (in press). [129] T. Buynitsky, D.I. Mostofsky, Restraint stress in biobehavioral research: recent developments, Neurosci. Biobehav. Rev. 33 (2009) 1089–1098. [130] G. Richter-Levin, Acute and long-term behavioral correlates of underwater trauma − potential relevance to stress and post-stress syndromes, Psychiatry Res. 79 (1998) 73–83. [131] G. Ritov, B. Boltyansky, G. Richter-Levin, A novel approach to PTSD modeling in rats reveals alternating patterns of limbic activity in different types of stress reaction, Mol. Psychiatry 21 (2016) 630–641. [132] A. Bali, A.S. Jaggi, Electric foot shock stress: a useful tool in neuropsychiatric studies, Rev. Neurosci. 26 (2015) 655–677. [133] P.R. Zoladz, D.M. Diamond, Predator-based psychosocial stress animal model of PTSD: Preclinical assessment of traumatic stress at cognitive, hormonal, pharmacological and epigenetic levels of analysis, Exp. Neurol. 284 (2016) 211–219. [134] C. Hammels, E. Pishva, J. De Vry, D.L. van den Hove, J. Prickaerts, R. van Winkel, et al., Defeat stress in rodents: from behavior to molecules, Neurosci. Biobehav. Rev. 59 (2015) 111–140. [135] C.R. Pryce, E. Fuchs, Chronic psychosocial stressors in adulthood: studies in mice, rats and tree shrews, Neurobiol. Stress 6 (2017) 94–103. [136] M. Vasconcelos, D.J. Stein, R.M. de Almeida, Social defeat protocol and relevant biomarkers, implications for stress response physiology, drug abuse, mood disorders and individual stress vulnerability: a systematic review of the last decade, Trends Psychiatry Psychother. 37 (2015) 51–66. [137] R.J. Rodgers, Animal models of anxiety: where next, Behav. Pharmacol. 8 (1997) 477–496. [138] J.A. Gray, The Neuropsychology of Anxiety: An Enquiry into the Functions of the Septo-Hippocampal System, Oxford University Press, 1982. [139] J. LeDoux, Fear and the brain: where have we been, and where are we going, Biol. Psychiatry 44 (1998) 1229–1238. [140] S.J. Lamontagne, M.C. Olmstead, J.L. Menard, The lateral septum and anterior hypothalamus act in tandem to regulate burying in the shock-probe test but not open-arm avoidance in the elevated plus-maze, Behav. Brain Res. 314 (2016) 16–20. [141] J.N. Perusini, M.S. Fanselow, Neurobehavioral perspectives on the distinction between fear and anxiety, Learn. Mem. 22 (2015) 417–425. [142] T. Tulviste, E. Kiive, K. Akkermann, J. Harro, Fears in the general population: more frequent in females and associated with the serotonin transporter promoter polymorphism and perceived relationship with mothers, J. Child Neurol. 30 (2015) 1459–1465. [143] P. Willner, P.J. Mitchell, The validity of animal models of predisposition to depression, Behav. Pharmacol. 13 (2002) 169–188. [144] J. Harro, Animal models of depression vulnerability, Curr. Top. Behav. Neurosci. 14 (2013) 29–54. [145] S. Matthysse, Animal models in psychiatric research, Prog. Brain Res. 65 (1986) 259–270. [146] F. Borsini, J. Podhorna, D. Marazziti, Do animal models of anxiety predict anxiolytic-like effects of antidepressants, Psychopharmacology (Berl) 163 (2002) 121–141. [147] R.C. Kessler, A.M. Ruscio, K. Shear, H.U. Wittchen, Epidemiology of anxiety disorders, Curr. Top. Behav. Neurosci. 2 (2010) 21–35. [148] T. Steimer, Animal models of anxiety disorders in rats and mice: some conceptual issues, Dialogues Clin. Neurosci. 13 (2011) 495–506. [149] T. Steimer, The biology of fear- and anxiety-related behaviors, Dialogues Clin. Neurosci. 4 (2002) 231–249. [150] M.S. Fanselow, The midbrain periaqueductal gray as a coordinator of action in response to fear and anxiety, in: A. Depaulis, R. Bandler (Eds.), The Midbrain Periaqueductal Gray Matter, Plenum Press, N.Y, 1991. [151] J. Haller, M. Alicki, Current animal models of anxiety, anxiety disorders and anxiolytic drugs, Curr. Opin. Psychiatry 25 (2012) 59–64. [152] N. Kulesskaya, V. Võikar, Assessment of mouse anxiety-like behavior in the lightdark box and open-field arena: role of equipment and procedure, Physiol. Behav. 133 (2014) 30–38. [153] M.M. Van Gaalen, T. Steckler, Behavioural analysis of four mouse strains in an anxiety test battery, Behav. Brain Res. 115 (2000) 95–106. [154] S.E. File, Behavioral detection of anxiolytic action, in: J.M. Elliot, D.J. Heal, C.A. Marsden (Eds.), Experimental Approaches to Anxiety and Depression, Wiley, Chichester, 1992. [155] A. Ramos, O. Berton, P. Mormede, A. Chaouloff, A multiple-test study of anxietyrelated behaviours in six inbred strains, Behav. Brain Res. 85 (1997) 57–69. [156] C.M. Thiel, C.P. Müller, J.P. Huston, R.K.W. Schwarting, High versus low reactivity to a novel environment: behavioural, pharmacological and neurochemical assessments, Neuroscience 93 (1999) 243–251.

burrow system, J. Comp. Psychol. 103 (1989) 70–82. [94] D.C. Blanchard, G. Giebel, R.J. Blanchard, The Mouse Defense Test Battery: pharmacological and behavioral assays for anxiety and panic, Eur. J. Pharmacol. 463 (2003) 97–116. [95] A. Lecci, F. Borsini, G. Volterra, A. Meli, Pharmacological validation of a novel animal model of anticipatory anxiety in mice, Psychopharmacology (Berl) 101 (1990) 255–261. [96] J.A. Bouwknecht, B. Olivier, R.E. Paylor, The stress-induced hyperthermia paradigm as a physiological animal model for anxiety: a review of pharmacological and genetic studies in the mouse, Neurosci. Biobehav. Rev. 31 (2007) 41–59. [97] D.A. Finn, M.T. Rutledge-Gorman, J.C. Crabbe, Genetic animal models of anxiety, Neurogenetics 4 (2003) 109–135. [98] W.P. Paré, Open field, learned helplessness, conditioned defensive burying and forced-swim tests in WKY rats, Physiol. Behav. 55 (1994) 433–439. [99] J.D. McAuley, A.L. Stewart, E.S. Webber, H.C. Cromwell, R.J. Servatius, K.C.H. Pang, Wistar-Kyoto rats as an animal model of anxiety vulnerability: support for a hypervigilance hypothesis, Behav. Brain Res. 204 (2009) 162–168. [100] G. Bignami, Selection of high rates and low rates of avoidance conditioning in the rat, Anim. Behav. 13 (1965) 221–227. [101] F. Chaouloff, N. Castanon, P. Morméde, Paradoxical differences in animal models of anxiety in the Roman rat lines, Neurosci. Lett. 182 (1994) 217–221. [102] T. Steimer, P. Driscoll, Divergent stress responses and coping styles in psychogenetically selected Roman high-(RHA) and low-(RLA) avoidance rats: behavioural, neuroendocrine and developmental aspects, Stress 6 (2003) 87–100. [103] R.L. Commissaris, G.M. Harrington, A.M. Ortiz, H.J. Altman, Maudsley reactive and non-reactive rat strains: differential performance in a conflict task, Physiol. Behav. 38 (1986) 291–294. [104] A. Paterson, P.J. Whiting, J.A. Gray, J. Flint, G.R. Dawson, Lack of consistent behavioural effects of Maudsley reactive and non-reactive rats in a number of animal tests of anxiety and activity, Psychopharmacology (Berl) 154 (2001) 336–342. [105] M. Altemus, M.A. Smith, V. Diep, C.S. Aulakh, D.L. Murphy, Increased mRNA for corticotrophin releasing hormone in the amygdala of fawn-hooded rats: A potential animal model of anxiety, Anxiety, 1994–1995, 1, 251–257. [106] A. Ramos, E.C. Correia, G.S. Izidio, G.R. Bruske, Genetic selection of two new rat lines displaying different levels of anxiety-related behavior, Behav. Genet. 33 (2003) 657–668. [107] O. Fujita, Behavior-genetic analysis of responses in runway test as measures of emotional reactivity in rats: I. Phenotypic variations and heritability estimates based on offspring-parents regressions, Shrinrigaku Kenkyu 46 (1975) 281–292. [108] G. Liebsch, A. Montkowski, F. Holsboer, R. Landgraf, Behavioural profiles of two Wistar rat lines selectively bred for high or low anxiety-related behaviour, Behav. Brain Res. 94 (1998) 301–310. [109] D.H. Overstreet, A.H. Rezvani, D.J. Knapp, F.T. Crews, D.S. Janowsky, Further selection of rat lines differing in 5-HT1A receptor sensitivity: behavioral and functional correlates, Psychiatry Genet. 6 (1996) 107–117. [110] R.L. Commissaris, P.A. Ardayfio, D.A. McQueen, G.A. Gilchrist, D.H. Overstreet, Conflict behavior and the effects of 8-OHDPAT treatment in rats selectively bred for differential 5-HT1A-induced hypothermia, Pharmacol. Biochem. Behav. 67 (2000) 199–205. [111] S.A. Brunelli, Selective breeding for an infant phenotype: rat pup ultrasonic vocalization (USV), Behav. Genet. 35 (2005) 53–65. [112] F.R. Brush, J.C. Froehlich, P.C. Sakellaris, Genetic selection for avoidance behavior in the rat, Behav. Genet. 9 (1979) 309–316. [113] V.C. Gomes, J. Landeira-Fernandez, Amygdaloid lesions produced similar contextual fear conditioning disruption in the Carioca high- and low-conditioned freezing rats, Brain Res. 1233 (2008) 137–145. [114] V.C. Gomes, W. Hassan, S. Maisonette, L.R. Johnson, A. Ramos, J. LandeiraFernandez, Behavioral evaluation of eight rat lines selected for high and low anxiety-related responses, Behav. Brain Res. 257 (2013) 39–48. [115] H.A. Robertson, Benzodiazepine receptors in emotional and non-emotional mice; comparison of four strains, Eur. J. Pharmacol. 56 (1979) 163–166. [116] C. Belzung, G. Griebel, Measuring normal and pathological anxiety-like behaviour in mice: a review, Behav. Brain Res. 125 (2001) 141–149. [117] R.L.H. Pobbe, E.B. Defensor, B.L. Pearson, V.J. Bolivar, D.C. Blanchard, Blanchard R.J. General, social anxiety in the BTBR T + tf/J mouse strain, Behav. Brain Res. 216 (2010) 446–451. [118] M. Camp, M. Norcross, N. Whittle, M. Feyder, W. D’Hanis, D. Yilmazer-Hanke, et al., Impaired Pavlovian fear extinction is a common phenotype across genetic lineages of the 129 inbred mouse strain, Genes Brain Behav. 8 (2009) 744–752. [119] R. Landgraf, M.S. Kessler, M. Bunck, C. Murgatroyd, D. Spengler, M. Zimbelmann, et al., Candidate genes of anxiety-related behavior in HAB/LAB rats and mice: focus on vasopressin and glyoxalase-I, Neurosci. Biobehav. Rev. 31 (2007) 89–102. [120] S.A. Krömer, M.S. Kessler, D. Milfay, I.N. Birg, M. Bunck, L. Czibere, et al., Identification of glyoxalase-I as a protein marker in a mouse model of extremes in trait anxiety, J. Neurosci. 25 (2005) 4375–4384. [121] S.C. Heinrichs, H. Min, S. Tamraz, M. Carmouché, S.A. Boehme, W.W. Vale, Antisexual and anxiogenic behavioral consequences of corticotropin-releasing factor overexpression are centrally mediated, Psychoneuroendocrinology 22 (1997) 215–224. [122] J.N. Crawley, J.K. Belknap, A. Collins, J.C. Crabbe, W. Frankel, N. Henderson, R.J. Hitzemann, S.C. Maxson, L.L. Miner, A.J. Silva, J.M. Wehner, A. WynshawBoris, R. Paylor, Behavioral phenotypes of inbred mouse strains: implications and recommendations for molecular studies, Psychopharmacology (Berl) 132 (1997) 107–124.

12

Behavioural Brain Research xxx (xxxx) xxx–xxx

J. Harro

carboline FG 7142 in the open field test, Eur. J. Pharmacol. 200 (1991) 183–185. [189] E.J. Chesler, S.G. Wilson, W.R. Lariviere, S.L. Rodriguez-Zas, J.S. Mogil, Identification and ranking of genetic and laboratory environment factors influencing a behavioral trait, thermal nociception, via computational analysis of a large data archive, Neurosci. Biobehav. Rev. 26 (2002) 907–923. [190] M.M. Sidor, K. Rilett, J.A. Foster, Validation of an automated system for measuring anxiety-related behaviours in the elevated plus maze, J. Neurosci. Methods 188 (2010) 7–13. [191] P.K. Thanos, C. Restif, J.R. O’Rourke, C.Y. Lam, D. Metaxas, Mouse social interaction test (MoST): a quantitative computer automated analysis of behaviour, J. Neural Transm. 124 (2017) 3–11. [192] P.M. Wall, C. Messier, Methodological and conceptual issues in the use of the elevated plus-maze as a psychological measurement instrument of animal anxietylike behavior, Neurosci. Biobehav. Rev. 25 (2001) 275–286. [193] A. Polissidis, S. Zelelak, M. Nikita, P. Alexakos, M. Stasinopoulou, Z.-I. Kakazanis, N. Kostamitsopoulos, Assessing the exploratory and anxiety-related behaviors in mice. Do different caging systems affect the outcome of behavioral tests? Physiol. Behav. 177 (2017) 68–73. [194] M. Labots, X. Zheng, G. Moattari, F. Ohl, H.A. van Lith, Effects of light regime and substrain on behavioral profiles of male C57BL/6 mice in three tests of unconditiones anxiety, J. Neurogenet. 30 (2016) 306–315. [195] E.J. Calabrese, An assessment of anxiolytic screening tests: hormetic dose responses predominate, Crit. Rev. Toxicol. 38 (2008) 489–542. [196] R.J. Rodgers, C. Lee, J.K. Shepherd, Effects of diazepam on behavioural and antinociceptive responses to the elevated plus-maze in male mice depend upon treatment regimen and prior maze experience, Psychopharmacology (Berl) 106 (1992) 102–110. [197] H. Schwegler, P.K. Pilz, M. Koch, M. Fendt, R. Linke, P. Driscoll, The acoustic startle response in inbred Roman high- and low-avoidance rats, Behav. Genet. 27 (1997) 579–582. [198] J. Simpson, J.P. Kelly, An investigation of whether there are sex differences in certain behavioural and neurochemical parameters in the rat, Behav. Brain Res. 229 (2012) 289–300. [199] P.A. Russell, Sex differences in rats' stationary exploration as a function of stimulus and environmental novelty, Anim. Learn. Behav. 5 (1977) 297–302. [200] R.N. Hughes, Behaviour of male and female rats with free choice of two environments differing in novelty, Anim. Behav. 16 (1968) 92–96. [201] D.T. Corey, The determinants of exploration and neophobia, Neurosci. Biobehav. Rev. 2 (1978) 235–253. [202] R.N. Hughes, Sex differences in spontaneous alternation and open-field behavior of hamsters: habituation differences, Curr. Psychol. Res. Rev. 8 (1989) 144–150. [203] A.L. Johnston, S.E. File, Sex differences in animal tests of anxiety, Physiol. Behav. 49 (1991) 245–250. [204] H.L. Steenberger, F. Farabollini, R.P.W. Heinsbroek, N.E. Van de Poll, Sex-dependent effects of aversive stimulation on holeboard and elevated plus-maze behavior, Behav. Brain Res. 43 (1991) 159–165. [205] A.R. Roeckner, A. Bowling, T.R. Butler, Chronic social instability increases anxiety-like behavior and ethanol preference in Long Evans rats, Physiol. Behav. 173 (2017) 179–187. [206] A. Wigger, I.D. Neumann, Periodic maternal deprivation induces gender-dependent alterations in behavioral and neuroendocrine responses to emotional stress in adult rats, Physiol. Behav. 66 (1999) 293–302. [207] J. McIntosh, H. Anisman, Z. Merali, Short- and long-periods of neonatal maternal separation differentially affect anxiety and feeding in adult rats: gender-dependent effects, Dev. Brain Res. 113 (1999) 97–106. [208] P.V. Piazza, J.M. Deminiére, M. Le Moal, S. Simon, Factors that predict individual vulnerability to amphetamine self-administration, Science 245 (1989) 1511–1513. [209] O. Berton, C.A. McClung, R.J. Dileone, V. Krishnan, W. Renthal, S.J. Russo, et al., Essential role of BDNF in the mesolimbic dopamine pathway in social defeat stress, Science 311 (2006) 864–868. [210] C.R. Pawlak, Y.-J. Ho, R.K.W. Schwarting, Animal models of human psychopathology based on individual differences in novelty-seeking and anxiety, Neurosci. Biobehav. Rev. 32 (2008) 1544–1568. [211] L. Valzelli, W. Kozak, O. Giraud, Exploratory behavior and the dual activity of some psychoactive drugs, Part 1 Methods Find. Exp. Clin. Pharmacol. 9 (1987) 437–440. [212] J. Harro, Inter-individual differences in neurobiology as vulnerability factors for affective disorders: implications for psychopharmacology, Pharmacol. Ther. 125 (2010) 402–422. [213] K.A. McLaughlin, K.J. Conron, K.C. Koenen, S.E. Gilman, Childhood adversity, adult stressful life events, and risk of past-year psychiatric disorder: a test of the stress sensitization hypothesis in a population-based sample of adults, Psychol. Med. 40 (2010) 1647–1658. [214] A. Albrecht, I. Müller, Z. Ardi, G. Caliskan, D. Gruber, S. Ivens, et al., Neurobiological consequences of juvenile stress: a GABAergic perspective on risk and resilience, Neurosci. Biobehav. Rev. 74 (2017) 21–43. [215] M.C. Camp, K.P. Macpherson, L. Lederle, C. Graybeal, S. Gaburro, L.M. Debrouse, et al., Genetic strain differences in learned fear inhibition associated with variation in neuroendocrine, autonomic, and amygdala dendritic phenotypes, Neuropsychopharmacology 37 (2012) 1534–1547. [216] R.S. Thompson, J.P. Christianson, T.P. Masianik, S.F. Maier, B.N. Greenwood, M. Fleshner, Effects of stressor controllability on diurnal physiological rhythms, Physiol. Behav. 112–113 (2013) 32–39. [217] J.B. Schulz, M.R. Cookson, L. Hausmann, The impact of fraudulent and irreproducible data to the translational research crisis -solutions and implementation, J. Neurochem. 139 (S2) (2016) 253–270.

[157] M. Kanarik, J. Harro, Sociability trait and regional cerebral oxidative metabolism in rats: predominantly nonlinear relations, Behav. Brain Res. 337 (2018) 186–192. [158] E. Dere, A. Zlomuzica, M.A. De Souza Silva, L.A. Ruocco, A.G. Sadile, J.P. Huston, Neuronal histamine and the interplay of memory, reinforcement and emotions, Behav. Brain Res. 216 (2010) 209–220. [159] J.W.B. Elsey, M. Kindt, Tackling maladaptive memories through reconsolidation: from neural to clinical science, Neurobiol. Learn. Mem. 142 (2017) 108–117. [160] C.S. Hall, Emotional behavior in the rat: I. Defecation and urination as measures of individual differences in emotionality, J. Comp. Psychol. 18 (1934) 385–403. [161] R.N. Hughes, Food deprivation and locomotor exploration in the white rat, Anim. Behav. 13 (1965) 30–32. [162] G. Griebel, C. Belzung, R. Misslin, E. Vogel, The free-exploratory paradigm: an effective method for measuring neophobic behaviour in mice and testing potential neophobia-reducing drugs, Behav. Pharmacol. 4 (1993) 637–644. [163] T. Mällo, A. Alttoa, K. Kõiv, M. Tõnissaar, M. Eller, J. Harro, Rats with persistently low or high exploratory activity: behaviour in tests of anxiety and depression, and extracellular levels of dopamine, Behav. Brain Res. 177 (2007) 269–281. [164] J. Harro, E. Vasar, J. Bradwejn, CCK in animal and human research on anxiety, Trends Pharmacol. Sci. 14 (1993) 244–249. [165] J. Harro, E. Vasar, Cholecystokinin-induced anxiety: how is it reflected in studies on exploratory behaviour, Neurosci. Biobehav. Rev. 15 (1991) 473–477. [166] J. Harro, CCK and NPY as antianxiety treatment targets: promises, pitfalls, and strategies, Amino Acids 31 (2006) 215–230. [167] A. Kask, L. Rägo, J. Harro, Anxiogenic-like effect of the neuropeptide Y Y1 receptor antagonist BIBP3226: antagonism with diazepam, Eur. J. Pharmacol. 317 (1996) R4–R4. [168] J.N. Crawley, Cholecystokinin-dopamine interactions, Trends Pharmacol. Sci. 12 (1991) 232–236. [169] V. Daugé, A. Sebret, F. Beslot, T. Matsui, B.P. Roques, Behavioral profile of CCK2 receptor-deficient mice, Neuropsychopharmacology 25 (2001) 690–698. [170] J.H. Kehne, The CRF1 receptor, a novel target for the treatment of depression, anxiety, and stress-related disorders, CNS Neurol. Disord. Drug Targets 6 (2007) 163–182. [171] S.R. Spierling, E.P. Zorrilla, Don’t stress about CRF: assessing the translational failures of CRF1 antagonists, Psychopharmacology (Berl) 234 (2017) 1467–1481. [172] A.R. Green, I.S. McGregor, On the anxiogenic and anxiolytic nature of long-term cerebral 5-HT depletion following MDMA, Psychopharmacology (Berl) 162 (2002) 448–450. [173] Z. Ebrahimian, Z. Karimi, M.J. Khoshnoud, M.R. Namavar, B. Daraei, Haidari M.R. Behavioral, stereological analysis of the effects of intermittent feeding diet on the orally administered MDMA (ecstasy) in mice, Innov. Clin. Neurosci. 14 (2017) 40–52. [174] P. Olausson, J.A. Engel, B. Söderpalm, Behavioral sensitization to nicotine is associated with behavioral disinhibition; counteraction by citalopram, Psychopharmacology (Berl) 142 (1999) 111–119. [175] K.I. Ueno, H. Togashi, K. Mori, M. Matsumoto, S. Ohashi, A. Hoshino, et al., Behavioural and pharmacological relevance of stroke-prone spontaneously hypertensive rats as an animal model of a developmental disorder, Behav. Pharmacol. 13 (2002) 1–13. [176] J. Harro, Long-term partial 5-HT depletion: interference of anxiety and impulsivity, Psychopharmacology (Berl) 164 (2002) 433–434. [177] G.N. Pires, A.G. Bezerra, S. Tufik, M.L. Andersen, Effects of experimental sleep deprivation on anxiety-like behavior in animal research: systematic review and meta-analysis, Neurosci. Biobehav. Rev. 68 (2016) 575–589. [178] A. Gravius, A. Dekundy, A. Vanaga, L. Franke, W. Danysz, Further pharmacological characterization of eltoprazine: focus on its anxiolytic, anorexic, and adverseeffect potential, Acta Neurobiol. Exp. 77 (2017) 77–85. [179] L.A. León, V. Castro-Gomez, S. Zárate-Guerrero, K. Corredor, A.P. Mello Cruz, M.L. Brandao, et al., Behavioral effects of systemic, infralimbic and prelimbic injections of a serotonin 5-HT2A antagonist in carioca high- and low-conditioned freezing rats, Front. Behav. Neurosci. 11 (2017) 117, http://dx.doi.org/10.2289/ fnbeh.2017.00117. [180] J. Harro, L. Oreland, E. Vasar, J. Bradwejn, Impaired exploratory behaviour after DSP-4 treatment in rats: implications for the increased anxiety after noradrenergic denervation, Eur. Neuropsychopharmacol. 5 (1995) 447–455. [181] M. Davis, D.E. Redmond Jr., J.M. Baraban, Noradrenergic agonists and antagonists: effects on conditioned fear as measured by the potentiated startle paradigm, Psychopharmacology (Berl) 65 (1979) 111–118. [182] J. Harro, M. Tõnissaar, M. Eller, The effects of CRA1000, a non-peptide antagonist of corticotropin-releasing factor type 1, on adaptive behavour in the rat, Neuropeptides 35 (2001) 100–109. [183] T. Mällo, C. Berggård, M. Eller, M. Damberg, L. Oreland, J. Harro, Effect of longterm blockade of CRF1 receptors on exploratory behaviour, monoamines and transcription factor AP-2, Pharmacol. Biochem. Behav. 77 (2004) 855–865. [184] American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 5th edition, American Psychiatric Publishing, Arlington, VA, 2013. [185] T. Insel, B. Cuthbert, M. Garvey, R. Heinssen, D.S. Pine, K. Quinn, et al., Research domain criteria (RDoC): toward a new classification framework for research on mental disorders, Am. J. Psychiatry 167 (2010) 748–751. [186] J.A. Chambers, K.G. Power, R.C. Durham, The relationship between trait vulnerability and anxiety and depressive diagnoses at long-term follow-up of generalized anxiety disorders, J. Anxiety Disord. 18 (2004) 587–607. [187] J.G. Raymond, J.D. Steele, P. Seriés, Modeling trait anxiety: from computational professes to personality, Front. Psychiatry 8 (2017) 1, http://dx.doi.org/10.3389/ fpsyt.2017.00001. [188] J. Bruhwyler, E. Chleide, G. Houbeau, M. Mercier, Stimulant effect of the ß-

13