Keywords

1 Introduction

Animal models are an important topic of preclinical research on neurobiology of psychiatric disorders and help in screening putative drugs for treating the disorder and permit a better comprehension of mechanisms implicated. The choice of the most appropriate animal model according to the condition to be studied is delicate and fundamental in order to be able to validate the extrapolation that will be made to the man afterwards. The ideal animal model would not only replicate the essential features of depression but also reliably predict antidepressant activity in a novel compound. However, an animal model in psychopharmacology must be able to predict the validity of the action of the substance studied in the pathology considered. Moreover, what is considered by the face validity must show similarities or other behaviours between the animal model and the disease that it wants to represent. Finally, construct validity must demonstrate the theoretical rationality of the model (Willner 1984). Animal models cannot take into account all aspects of a mental illness, depression any more than another one. Animal models of depression can take into account one or more symptoms of the disease, such as anhedonia, the retardation or certain biochemical deficiencies caused by the illness.

Observation of symptoms presented in patients presenting with depression has led to various hypotheses concerning the aetiology of depression. Among these hypotheses researchers have built animal models given corresponding more or less to the core symptoms of depression. The existing antidepressants, mainly the tricyclic drugs at that time, were used to define the models. Thus, some models of depression have been constructed according to the mechanism of action of tricyclic antidepressants which are supposed to increase noradrenaline (NA), serotonin (5-HT) or dopamine (DA) in the synaptic cleft. This theory has been modified to take into account the adaptation of receptors which appear to correlate with the onset of the antidepressant response and the action of antidepressants in blocking the inhibitory feedback, their action at the second messenger level as well as a postsynaptic action (Bourin and Baker 1996).

In the absence of spontaneously depressed animals, psychopharmacologists have used pharmacological interactions, stress models or brain damage models to predict such an activity. Historically, pharmacological interactions were used to predict antidepressant activity (Bourin 1990). The antagonism of the effects of reserpine was one of the first depression models used (Costa et al. 1960). Reserpine depletes central stores of 5-HT, NA and DA via blockade of uptake into vesicles, and drugs are assumed to reverse reserpine-induced symptoms by elevating the intersynaptic concentrations of these amines. Different parameters were measured (Bourin et al. 1983), e.g. antagonism of hypothermia, ptosis and antagonism of reduced motor activity. This pharmacological interaction model was abandoned due to its lack of specificity for an antidepressant effect and its sensitivity to non-antidepressant drugs (Willner 1984). Other pharmacological models were used: the antagonism of oxotremorine-induced effects, the antagonism of high-dose apomorphine (Puech et al. 1981) and the yohimbine test (Quinton 1963). But in the mid-1970s, the behavioural animal models of depression replaced the pharmacological interaction methods. Among the animal models used in screening antidepressant activity, we have chosen to discuss the following rodent behavioural models, the forced swimming test (FST) (Porsolt et al. 1977), the tail suspension test (TST) (Steru et al. 1985), the chronic mild stress model (CMS) (Willner et al. 1992), the learned helplessness (LH) model (Seligman et al. 1975) and one paradigm based on neuronal deficits, the olfactory bulbectomy (OB) (Cairncross et al. 1977), as they are the most commonly utilised models.

2 Description of the More Frequently Used Models in Depression

2.1 Stress Models

The hypothesis that chronic stress can induce depression is questionable. It seems that vulnerability to depression in humans could be compared to behavioural conditions induced by stress in animals. It has been shown that chro nic stress causes an increase in stress hormone cortisol and that in parallel it decreases 5-HT and DA and probably other important neurotransmitters in the depressive symptom. In fact clinical symptoms of depression in human are more complex than those induced by stress which is more in the field of anxiety. The rodent models of anxiety are more built on fear. Animal models based on the hypothesis that depression is induced by stress include the mouse/rat FST , the T ST, the CMS and the LH . In these models, animals are exposed to uncontrollable stress resulting in maladaptive behaviours. The shocks necessary to induce a depressive symptomatology like in animals must be uncontrollable by the latter and especially unpredictable.

2.1.1 The Mouse Forced Swimming Test (FST)

The mouse F ST is a model that was built to predict the antidepressive action of new molecules now for years, using tricyclic antidepressants as reference (Porsolt et al. 1977). Mice are individually placed into glass cylinders (height, 25 cm; diameter, 10 cm) containing 10 cm of water maintained at 23–25 °C and leaving them there for 6 min. After vigorous activity, swimming attempts cease and the animal adopts a characteristic immobile posture. The animal is judged to be immobile when it floats in an upright position and makes only minimal movements to keep its head above water. This state of immobility has been named “behavioural despair”, because probably of an anthropomorphic idea based on the assumption that the animals have the feeling they cannot escape from the cylinder (Petit-Demouliere et al. 2005). The decrease of the duration of immobility, which is recorded during the last 4 min of the 6-min test period, leads to think that the drug assessed is potentially an antidepressant. FST is very useful to study neurobiological mechanisms to better understand through the drug responses what is depression in humans (Porsolt 2000; Lucki et al. 2001). This behavioural test, far of the reality of depression clinical features, is a good translational approach (Bourin 2010).

The researches in discovering new drugs for treating depression are performed with FST as a core behavioural model. As we know neurotrophic factor could be a potential antidepressant agent. So an infusion of brain-derived neurotrophic factor (BDNF) was injected in the ventral tegmental area in mice. As a result, it induced a shorter latency to immobility relative to control animals, in the FST in rats (Eisch et al. 2003). Other researchers pointed out when using FST, a significant decrease in the immobility time compared to vehicle-infused controls after BDNF infusion (Siuciak et al. 1997). On the other hand, other potential mechanisms of antidepressants using FST K+ as channel openers and K+ channel blockers were studied (Guo et al. 1995, 1996; Redrobe et al. 1996; Slattery et al. 2004). The FST is not only for screening antidepressant-like effects, but as well to better understand neurobiology of depression, mainly to study the role of monoamines. Nevertheless, this model of depression is not only linked to monoamines. A classic treatment of depression, the electroconvulsive seizures, was performed on animals in FST (Nestler et al. 2002) and was effective in increasing the swimming time.

The FST showed this ability to use genetically modified animals, which are useful to understand the mechanisms of action of antidepressants (Gardier et al. 2001; Holmes et al. 2002; Cryan et al. 2001). These studies are the following of those using specific ligands (Redrobe and Bourin 1998a)

2.1.2 The Rat Forced Swimming Test

The rat FST was performed before the one in mouse. The main problem with the rat is the fact that the animal dives to the bottom of the tank. The typical procedure involves placing a rat in a cylinder for 15 min on day 1 of the test. As a result of this exposure, escape attempts eventually cease and the animal is immobile by the end of the session. The following day, the animal is returned to the cylinder for a period of 5 min, at which point immobility time is recorded. Drug treatment (single or repeated) is performed between the first and the second exposure to the cylinders. Effective antidepressant treatments reduce immobility time in the second exposure. More recently, an improvement to the rat FST was proposed: with swimming time, two other parameters are scored, immobility and climbing behaviour. Lucki (1997) suggested that climbing behaviour is involved in initial response to the test situation, whereas swimming may be secondary exploratory behaviour associated with escape. SSR Is reduce immobility and increase swimming, whereas the selective noradrenaline reuptake inhibitors reduce immobility and decrease climbing without affecting swimming.

2.1.3 The Tail Suspension Test (TST)

The T ST is based on the observation that a mouse suspended by the tail shows alternate periods of agitation and immobility, similar but not identical to that observed in the mouse FST (Steru et al. 1985). So it is almost the same paradigm to the FST . In the TST immobility is induced in mice simply by suspending them, using adhesive Scotch tape, to a hook connected to a strain gauge that picks up all movements of the mouse and transmits them to a central unit, which calculates the total duration of immobility during a 6-min test. This test has been automated (ITEMATIC-TST) and measures duration of immobility and the energy expended by each animal, the power of the movements (Steru et al. 1987) which can distinguish different classes of psychotropic activity. The TST procedure bypasses several problems of the swimming model: the immobility is objectively measured and no hypothermia is induced by immersion in cold water. The mouse TST can predict antidepressant activity of numerous components. It will be shown later on in this chapter that the combination of both tests (TST and FST ) can help in discrimination of mechanisms of action of antidepressants when used in a purpose of screening.

2.1.4 Chronic Mild Stress (CMS)

Chronic sequential exposure to a variety of mild stressors (chronic mild stress) has been found to decrease the consumption of and/or preference for a palatable weak sucrose solution in rats or mice (Willner et al. 1992). Animals are exposed to various types of stressors which change over a period of weeks or months, e.g. overnight illumination, cage tilt and change of cage mate, resulting in a decrease in sucrose preference for several weeks, which reflects a generalised decrease in the sensitivity to rewards or anhedonia. Along with a state of anhedonia, various other behavioural changes due to depression are shown, persisting weeks after stimuli cessation (Gorka et al. 1996). The model has predictive validity since the reversal of pathologic behaviour requires 3–4 weeks of treatment, as in human depression. In fact, this model has the ability to demonstrate a potential early onset of action of antidepressant treatment. This test presents with the advantage of the chronicity, it looks more of the treatment of depression which takes several weeks to be active. The problem is that time for rats is difficult to compare with humans.

2.1.5 Learned Helplessness Model

The LH model is the most familiar simulation of depression and also the most controversial. The model mimics some of the main features of depression, particularly of the kind that are precipitated by unfavourable environmental stress. The model, described by Seligman et al. (1975), consists of exposing animals to unavoidable and uncontrollable stressors such as electric foot shock, after which learning deficits on subsequent tests are observed where animals are found to be unable to learn to avoid an aversive stimulus and remain motionless and helpless in such a situation. This state has been named “learned helplessness” and is not found in animals exposed to identical but controllable stress . It has been shown that the persistent immobility of the animal to respond is confined to the learned immobility that has been required during the unavoidable shock situation. Thus the learned helplessness behaviour does not generalise to other types of behaviour that has been learned in the absence of the shock. Seligman and co-authors have suggested that animals learn that responding to uncontrollable shock is futile and that the cognitive and motivational deficits produced in this paradigm are parallel to human clinical depression. The helpless animal enters a learning situation with a generalised associative set that its actions are without consequence. It therefore responds less or not at all. In addition to an acquisition deficit, other features of the helpless animal parallel clinical dimensions of depression, deficits in motivation and emotion .Changes in activity, food intake and weight have also been reported.

2.2 Neuronal Deficit Models

2.2.1 Olfactory Bulbectomy Model (OB)

Apart from models based on stress , there are animal models of depression that are based on the hypothesis that depression is caused by neuronal regulatory deficit. One example in the OB proposes that depression is a biological disorder that develops in individuals who are predisposed due to neural regulatory deficits in the brain (Cairncross et al. 1977). The major brain damage model involves bilateral lesions of the olfactory bulbs, which form part of the limbic system in the rodents. Rats subjected to this operation display a variety of behavioural changes, including irritability, hyperactivity and an elevation of circulating levels of plasma steroids. A disconnection of the olfactory bulbs has shown to produce abnormalities in emotional behaviour, termed “bulbectomy syndrome” due to a disruption in the homeostatic regulation of impulse traffic in the limbic system (Jesberger and Richardson 1985). The hyperactivity exhibited by bulbectomised rats when they are subjected to a stressful novel environment such as the open field, as well as deficits in passive avoidance tasks, is reversed by the chronic administration of antidepressants.

3 The Use of Animal Models to Define Antidepressant Response

Animal models of depression are increasingly being used to better understand the mechanisms of action of antidepressants but as well for screening potential new antidepressants (Darcet et al. 2016). The basic requirement for an animal model of depression is its sensitivity and/or responsiveness to an antidepressant drug and lack of false positives (e.g. neuroleptics, stimulants and/or anxiolytics). All models presented above are called animal models of depression due to their responsiveness to antidepressant drugs. However, it is not possible to choose among these models the one which would be the most specific of the characterisation of one antidepressant rather than another. However, we will see later that the combination of several models associated with different strains of animals can contribute to the discrimination of antidepressants. False positives must not be readily rejected as many non-antidepressant drugs have not been adequately tested for their possible antidepressant activity in controlled clinical trials. Also the co-administration of certain drugs can increase the efficacy of antidepressant drugs and reduce the time of onset of action (e.g. pindolol , buspirone ) (Pérez et al. 1997). It is clear that not all symptoms of human depression can be modelled in animals and no universal model representing all these symptoms as yet exists. The model must however be robust and reproducible.

The OB model can be considered as a good approximation to an aetiological model of depression, but it is not an exact phenomenon of depression since it is a noncognitive explanation of depression. OB has a strong theoretical rationale (as antidepressants are not effective on normal rats), a face and predictive validity in the identification of antidepressants from some chemical classes. However, there is some question about sensitivity and selectivity as certain antidepressants appear to lack activity in this model. Even if the OB model exhibits a few false positive, only a narrow range of non-antidepressants has actually been tested. However, this remains an interesting model as many antidepressants are only active after subchronic or chronic treatment; the duration of treatment is similar to that needed for therapeutic activity to become apparent in depressed patients.

The LH is the most criticised paradigm due to its lack of complete specificity and poor reliability across laboratories. Also many of the depression-like phenomena produced are short-lived, most symptoms dissipating in 48–72 h depending on the shock procedure produced (Weiss and Simson 1989). Anxiolytics can reverse the behaviour (GABA injected into the hippocampus ) (Petty and Sherman 1980), and only a few SS RIs (citalopram , fluvoxamine, indalpine and zimelidine) have shown efficacy and then only under particular conditions (Martin et al. 1990). The relevance of this model to depression has been questioned, and Anisman et al. (1980) suggested that the LH model is more a model of stress adaptation than a model of depression.

The CMS model was considered to be the most validated model of depression implicating stress as the aetiological cause of depression (Willner and Papp 1997) but is not selective for antidepressant drugs and it is not used for screening new drugs. The model exhibits poor reliability (D’Aquila et al. 1997) and the behavioural alterations dissipate quickly. The role of chronic stress and anhedonia has been questioned in the aetiology of depression (Breslau and Davis 1986), as an improvement in mood regulation of patients exposed to antidepressants is observed before any improvement in anhedonia. This test presents with the advantage of the chronicity, it looks more of the treatment of depression which takes several weeks to be active. Yet it is difficult to compare time in rat’s life with humans. On the other hand, there are a lot of false positive results because CMS is built mainly on anhedonia which is present as well in schizophrenia; as a result some antipsychotic drugs induce positive results on the test.

The TST is a rapid and convenient test to perform; however, results obtained show variations in the same strain, and for the same treatment (David et al. 2003), it is very important to be careful regarding strain, weight of animals as well as the operating conditions. Hascoët et al. (1991) suggested that the TST is closer to a spontaneous activity model like actimetry than an antidepressant model.

The FST model is extensively utilised to screen drugs to define an antidepressant effect, as it is quick, inexpensive and easy to perform and has proven to be an easy reproducible screening test for pharmacological activity. These qualities allow the possibility of investigating various factors such as age, gender, and strain difference, which bear significant relevance to human depression (David et al. 2001a, b). Modification of the test to look at chronic rather than acute drug treatment and prolongation of the stress has been reported to improve the specificity. In a study Detke et al. (1997) showed that antidepressants chronically administered at lower doses produced a significant decrease of immobility duration in the mouse FST . The value of this test is quite high because it is able to predict very often the dose-response effect in humans.

It is sensitive to all of the major classes of antidepressant drugs presenting with different mechanisms of action. It is necessary to use actimetry to measure the locomotor activity preventing the false positives which are mainly amphetamine drugs. In addition, the mouse FST permits exploration of the possible mechanisms of action of different classes of antidepressants through the use of specific ligands (Redrobe and Bourin 1997; Redrobe et al. 1996, 1998a, b).

For 30 years in our research laboratory, we have invested a lot of efforts in daily use of TST and FST in mice, not only to detect possible antidepressants but also to better understand their mechanisms of action. The use of different ligands, whether they are agonists or antagonists specific for serotonin receptors , allowed us to show the involvement of 5-HT1A and 5-HT1B receptors in the mechanism of S SRIs. Two research papers clearly show the impact of various strains of mice on the response of antidepressants on these two models, these considerations can lead to a pharmacogenomic impact on the efficacy of antidepressants (David et al. 2003; Ripoll et al. 2003). The best effect/size obtained on all the models envisaged is that of the Swiss strain mice on the T ST. As the C57BL/6 and the DBA/2 mice attempted to redress their position (i.e. climbing up their tails previously reported by Mayorga and Lucki 2001 and Ripoll et al. 2003), it was difficult to conclude on their activity in the TST. It is necessary to use different strains of mice to demonstrate the antidepressants acting on noradrenaline or serotonin or dopamine although it is believed that the latter is the common final route of all antidepressants. Swiss mice are the most sensitive strain to detect serotonin and/or noradrenalin antidepressants, whereas C57BL/6 Rj was the only strain sensitive to bupropion (dopaminergic agent) using the FST. In the TS T, all antidepressants studied decreased the immobility time in Swiss and C57BL/6 Rj strains. In order to evaluate the mechanism of action of a substance that may be clinically shown to be an antidepressant, the use of T ST and FST in mice may be useful provided that three strains of mice (Swiss, NMRI and C57Bl/6 Rj) were used concomitantly. Some antidepressants with different mechanisms of action such as tricyclics or S SRIs induce a neighbouring behavioural response (with the exception of some small, difficult to quantify differences) that can be identified by the eye of an experienced experimenter. This is comparable to a psychiatrist with a confirmed clinical experience who can detect rough clinical signs of depression. For these antidepressants, in order to clarify more precisely their mechanism of action, it is useful to associate them with more or less specific compounds at sub-active doses and to practise the FST (Redrobe and Bourin 1999a). According to all results, a decision tree was established to help the screening and to give an indication on the mechanism of action (Bourin et al. 2005a). In the same research, it could be simple to discover new antidepressants as well as to have an idea about their mechanism of action.

4 Role of the FST in Evaluating Mood Stabilisers

It is difficult to model bipolar disorder animal models because of the complexity to have in the same animal very different syndromes as mania and depression. Bipolarity is a restrictive disease, affecting everyday life. It can affect different domains such as cognitive faculties (by disturbing the memory, attention or the executive functions of the patients) and sleeping (insomnia without fatigue can be the sign of a manic episode) or manifest through excessive fatigue. It is also characterised by the impossibility of being able to manage its emotions, and this emotional hyperreactivity is incarnated in irritable, angry behaviour. It can also give rise to anxiety disorders. Yet some animal models are used but they are very far from the clinic complexity of the disorder. The amphetamine-induced behaviour is the pivotal test of the disorder (Machado-Viera et al. 2004). Under these conditions, it was interesting to use the FST to better understand their antidepressive activity (Bourin and Prica 2007). This is all the more important since antidepressants are not recommended in the treatment of bipolar disorder ; some mood stabilisers are fortunately poor antidepressants, the reason why they escape to switch to mania or hypomania. Lithium is considered an antimanic more than an antidepressant; however, our team was able to show some antidepressant-like effects in the FST in mice (Hascoet et al. 1994). Moreover, it was possible to potentiate the action of some S SRIs (Nixon et al. 1994; Bourin et al. 1996a); these results match with clinical reports. On the other hand, combination studies of SSRI antidepressants using different drugs as clonidine, lithium and quinine précised the role of 5HT receptor subtypes in the effects of antidepressant action (Redrobe and Bourin 1999b). Lithium did not show antidepressant activity in the rat on FST, sometimes it showed contrary effects (Mague et al. 2003; Carlezon et al. 2006; O’Donnell and Gould 2007), it is very often in behavioural research that rats respond differently with mice, so we must be careful when comparing the interspecies results. 5-HT1A receptors are involved in the mechanism of action of sodium valproate and carbamazepine, yet both drugs are inactive alone on the FST. That suggests that other neurotransmitters than 5HT are involved. Carbamazepine and sodium valproate have complex mechanism of action regarding their anticonvulsive activity, and we know that they affect GABA, dopamine (DA) and noradrenalin (NA) (Post et al. 1992). Interesting results were obtained with lamotrigine, topiramate and phenytoin (Bourin et al. 2005b). Lamotrigine which is an atypical antiepileptic is now for years used as mood stabiliser mainly in depressed bipolar patients, decreasing immobility time in the mice FST. Topiramate and phenytoin as well decrease immobility time in the FST following i.p. administration (unpublished data). We can conclude that lamotrigine, topiramate and phenytoin have a marked activity on FST. Lamotrigine was designed at the early stage of development to become an antidepressant, but its poor antidepressant action leads researchers to move to look at an anticonvulsant action. This antiepileptic action has been suggested because of its action on the inhibition of glutamate release, by an effect on voltage-sensitive sodium channels (Leach et al. 1986; Kuo and Lu 1997). Veratrin, a Na+ channel activator that increases glutamate release was used to study the role of ion channels in the mechanism of action of drugs on the FST (Lizasoain et al. 1995). The co-administration of veratrin, with lamotrigine, topiramate and phenytoin mice, was studied compared with “true” antidepressants of different mechanisms of action as paroxetine, imipramine and desipramine . Veratrin was antagonising the effect of phenytoin, lamotrigine and topiramate suggesting that sodium channels underlie their action in the forced swimming test. In contrast, veratrin did not affect antidepressant activity of the studied antidepressant drugs (Prica et al. 2008). Thus, the neurobiological mechanisms underlying the processes of swimming or immobility are more complex than envisaged by the discoverers of the FST. So the core idea is that the mechanism underlying the anti-immobility effect of mood stabilisers on their “antidepressant activity” is related to sodium channel and that FST is sensible to sodium channel mechanism and in a way to glutamatergic mechanism (van Enkhuizen et al. 2015).

Conclusion

It is difficult to compare all the animal models of depression as they vary widely in the manner of inducing abnormal behaviour, in the aspects of behaviour chosen for study and in the time course of antidepressant action. This difficulty can be problematic in exporting data from the various laboratories. Other factors hinder the comparison of these models such as strain, age, seasonal variations, light cycles utilised, etc. (Bourin et al. 1998; David et al. 2001a, b). These different parameters can lead to observational differences between laboratories for the same drugs.

The perfect animal model of depression as yet does not exist. No single animal model reviewed here is a precise paragon of depression as seen in humans and questions concerning the utilisation of a battery of tests and/or instead of a single model to determine antidepressant activity have been raised. As different aspects of depression are measured in each model and the different models possibly represent a different category of depression, the question remains whether a true comparison between models of a compound’s antidepressant activity is possible (Bourin et al. 1996b). However, the screening of drugs in these paradigms allows for a better understanding of the mode of action of antidepressants, the neurobiology of depression (Remus and Dantzer 2016) as well as the discovery of new and more effective antidepressants (Wang et al. 2017). The progress in knowledge of these animal models is a way leading to translational psychopharmacology (Bourin 2010). That means the researchers are able to understand over the models the clinical features and make the synthesis we need to discover new drugs and even to understand better the mood disorders. Animal models of depression have probably not expressed their full capacity, mainly because there are not enough intellectual links between the preclinical and clinical researchers. They can still be very useful in the understanding of depressive illness (Wong and Licinio 2002).