Mutant mouse models of depression: Candidate genes and current mouse lines

Abstract

Depression is a multifactorial and multigenetic disease. At present, three main theories try to conceptualize its molecular and biochemical mechanisms, namely the monoamine-, the hypothalamus-pituitary-adrenal- (HPA-) system- and the neurotrophin-hypotheses. One way to explore, validate or falsify these hypotheses is to alter the expression of genes that are involved in these systems and study their respective role in animal behavior and neuroendocrinological parameters. Following an introduction in which we briefly describe each hypothesis, we review here the different mouse lines generated to study the respective molecular pathways. Among the many mutant lines generated, only a few can be regarded as genetic depression models or as models of predisposition for a depressive syndrome after stress exposure. However, this is likely to reflect the human situation where depressive syndromes are complex, can vary to a great extent with respect to their symptomatology, and may be influenced by a variety of environmental factors. Mice with mutations of candidate genes showing depression-like features on behavioral or neurochemical levels may help to define a complex molecular framework underlying depressive syndromes. Because it is conceivable that manipulation of one single genetic function may be necessary but not sufficient to cause complex behavioral alterations, strategies for improving genetic modeling of depression-like syndromes in animals possibly require a simultaneous targeted dysregulation of several genes involved in the pathogenesis of depression. This approach would correspond to the new concept of ‘endophenotypes’ in human depression research trying to identify behavioral traits which are thought to be encoded by a limited set of genes.

Introduction

Affective disorders are thought to be caused by a complex interplay of genetic vulnerabilities and unfavorable environmental events (Kessler, 1997, Sullivan et al., 2000). According to this view, depressive episodes develop if a single or several pathogenetic factors in combination reach critical dimensions or occur at the same time. Modern biological concepts indicate that these factors induce transient or even persistent dysfunctional changes in several brain regions and systems. Different molecular theories have been put forward to define the cause and pathogenesis of depression at the genetic and biochemical level. Accordingly, depression is regarded as a multifactorial disorder in which multiple genes are involved.

Genetically, modified mice represent a powerful tool to study candidate genes thought to participate in a particular disease. Molecular genetic manipulations allow to generate ‘transgenic’ and ‘knockout’ lines. Classical transgenic mice have additional copies of normal or abnormal genes in their genome, which usually results in a so-called ‘gain of function’. However, transgenes can also be used to induce a ‘loss of function’ when the transgene inserted is an antisense to a targeted gene. In classical knockout mice, specific target genes are disrupted leading to a ‘loss of function’. However, knockin strategies, in which genetic material is knocked into a particular locus, e.g. knocking in tTA downstream of a particular promotor, can also cause a gain of function. The main distinction between knockin/knockout techniques and transgenic techniques is that the former targets a specific locus, while the latter acts by random integration into the genome.

The sophisticated combination of these techniques allows the generation of mutant mouse lines in which the genetic manipulation occurs only in selected cell populations and at a specific developmental stage, or where the over- or under-expression of genes can even be turned on and switched-off artificially at a specific time point determined by the researcher. The hypothesis-orientated generation of genetically modified mice should result in animal models that imitate specific genetic, biochemical or behavioral features of human depression. These models can be used to study the pathogenesis of depressive episodes, which may lead to improvements in diagnosis and prevention. Furthermore, a better understanding of the molecular mechanisms could allow the design of better treatment strategies with specific molecular target sites and less side effects.

The first drugs used to treat depression were discovered by serendipity. Tricyclic compounds (TCA) were initially used as antihistaminergic agents. The first monoamine oxidase inhibitors (MAOI) were antitubercular drugs. Due to careful clinical observations they were found to improve the symptoms of depression. Their common biological mechanism is their potential influence on monoaminergic systems. Since this discovery, depression has been explained by an impairment of serotonergic and noradrenergic, and more recently also dopaminergic neurotransmission. This monoaminergic deficiency can result from several mechanisms: (i) decreased synthesis or early degradation of the neurotransmitters, (ii) altered expression or function of the neurotransmitter receptors and (iii) impairment of signal transduction systems activated by post-synaptic neurotransmitter receptors (Fig. 1).

The discovery of the molecular targets of first generation antidepressants led to the design of second and third generation drugs such as selective serotonin reuptake inhibitors (SSRI) and selective norepinephrine reuptake inhibitors (SNRI). Clinically, the new compounds are not more efficient than the older agents but have fewer side effects. They exert their effect by increasing the availability of monoamines in the synaptic cleft by either (i) blocking their reuptake by influencing directly or indirectly presynaptic transporter systems or (ii) inhibiting the degradation of the neurotransmitters. Many preclinical and clinical studies have clearly implicated the serotonin (5HT) and norepinephrine (NE) neurotransmitter systems in antidepressant action (for review, see: Mann, 1999). Furthermore, several subtypes of 5HT receptors (in particular, 5HT_1A, 5HT_1B, 5HT_1D, 5HT_2A, 5HT_2C, 5HT₆ and 5HT₇) are thought to participate in the pathophysiology of depression (Mann, 1999, Pauwels, 2000).

The monoaminergic theory of depression can explain some features of the disorder and has constituted treatment regimens used for more than 50 years, but did not become a clear headstone for the illustration of pathogenesis. The monoaminergic concept neither provides an explanation for the number of patients not responding to current therapeutic agents; nor for the lag period of the therapeutic response to antidepressant treatment. Whereas enhancement of serotonergic or noradrenergic transmission occurs already within hours after drug administration, only chronic antidepressant treatment has mood-elevating effects. These observations suggest that alterations in monoaminergic transmission only represent the initiation of more slowly developing plasticity changes, which may underlie the onset and reversal of depressive episodes. Recent hypotheses try to conceptualize second messenger pathways that cause long-term changes of gene expression, which may represent the molecular correlate of plastic changes underlying the pathogenesis and therapy of depression.

One important second messenger system activated by antidepressant treatment is the cyclic adenosine monophosphate (cAMP) pathway (Vaidya and Duman, 2001). The generation of cAMP results in the activation of cAMP-dependent protein kinase (PKA), which in turn activates the transcription factor cAMP response element binding protein (CREB) via phosphorylation. Activated CREB enhances the transcription of many target genes, including brain-derived neurotrophic factor (BDNF). BDNF exerts its effects mainly by activation of its specific receptor: tyrosine receptor kinase B (TrkB) (Fig. 2). BDNF and other neurotrophins influence structural plasticity and have trophic effects on neurons. Clinical and experimental observations have led to the hypothesis that a deficiency in BDNF contributes to the pathophysiology of depression (Duman et al., 1997, Altar, 1999).

Chronic stress, and a subsequent rise in plasma corticosteroid levels, is regarded as a major risk factor for the development of a depressive episode (see below: 1.3.) (Holsboer, 2000, Holsboer, 2001). Stress or treatment with corticosteroids causes a decrease in BDNF mRNA in the hippocampus and other brain regions thought to be involved in the pathogenesis of depression, most likely by activation of glucocorticoid receptors (Nibuya et al., 1995, Smith et al., 1995, Ueyama et al., 1997, Rasmusson et al., 2002). Moreover, in a genetic rat model of depression, the Flinders Sensitive Line, BDNF is expressed in lower levels in several brain regions compared to controls, the Flinders Resistant Line (Angelucci et al., 2000). In rodents, the stress-induced downregulation of BDNF can be reversed by antidepressants or electroconvulsive therapy (ECT) (Nibuya et al., 1995). This effect is enhanced by combination with voluntary physical activity, which is also known to have antidepressant effects in humans (Russo-Neustadt et al., 2001). Even without prior stress, hippocampal BDNF expression is induced by chronic antidepressant treatment (Nibuya et al., 1995, Smith et al., 1997) or ECT (Nibuya et al., 1995, Zetterstrom et al., 1998). Moreover, chronic administration of several distinct classes of antidepressants upregulates CREB mRNA expression (Nibuya et al., 1996) and CREB phosphorylation (Thome et al., 2000, Saarelainen et al., 2003) within the hippocampus and other brain areas. Regions in which upregulation of BDNF in response to antidepressant treatment occurs, correspond to those regions where CREB is activated. This correlation suggests that CREB may contribute to the antidepressant-induced increase in hippocampal BDNF expression. Taken together these results indicate that stress and depression coincide with a decreased activity of the CREB-BDNF-TrkB pathway, while antidepressants evoke an activation of this signaling cascade.

At the clinical level, postmortem investigations have shown that untreated depressive patients have lower levels of CREB and phosphorylated CREB in the cortex than healthy controls (Dowlatshahi et al., 1998, Chen et al., 2001, Yamada et al., 2003). Moreover, reduced BDNF serum levels have been correlated to the depressive state of patients (Karege et al., 2002). Clinical studies detected higher concentrations of CREB (Dowlatshahi et al., 1998), BDNF (Chen et al., 2001) and TrkB (Bayer et al., 2000) in patients under antidepressive medication than in untreated patients. In summary, these data support the assumption that the CREB-BDNF-TrkB pathway is involved in the pathophysiology and therapy of depression.

One more indication that neurotrophins could mediate the effect of monoaminergic antidepressants is the antidepressant effect of BDNF in rodents. When infused near the raphe nucleus, the main source of serotonergic innervation of the hippocampus, BDNF has an antidepressant-like effect in the learned helplessness model of depression (Siuciak et al., 1997). Bilateral infusion of BDNF or neurotrophin 3 into the dentate gyrus of the hippocampus also produces an antidepressant-like effect in this paradigm as well as in the forced swim test (Shirayama et al., 2002). With respect to clinical employment, however, one has to consider the size of the BDNF molecule, ruling out the blood brain barrier passage. Thus, it cannot be administered effectively in patients.

Stress is regarded as an important pathogenetic factor for the development of depressive episodes. Therefore, a dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis, the major endocrinological stress system in mammals, has been postulated to play a role in human depression. Under physiological conditions, the HPA system is activated by external and internal stressors in order to regulate the increased energy needs, e.g. by release of glucose, increase of blood flow, increase of attention etc. in situations that are critical or even life-threatening for the organism. The hypothalamus produces corticotropin-releasing factor (CRF) which stimulates the production of adrenocorticotropic hormone (ACTH) in the pituitary, consecutively inducing the synthesis and release of glucocorticoids (cortisol in humans, corticosterone in rodents) by the adrenal glands (Fig. 3). The released cortisol exerts a negative feedback on the hypothalamus to keep the system in balance.

Many patients with severe major depressive episodes have a dysregulated circadian cortisol secretion with significantly higher blood levels than healthy control subjects (Strickland et al., 2002, Gold et al., 2002, Gibbons, 1964, Dolan et al., 1985, Rubin et al., 1987, Deakin et al., 1990, Wong et al., 2000). Elevated cortisol levels of depressive patients usually return to normal under successful antidepressant treatment (Bhagwagar et al., 2002) or once depression disappears spontaneously (Steckler et al., 1999). Moreover, cortisol synthesis inhibitors such as ketoconazole have also shown positive effects in the treatment of major depression (Wolkowitz et al., 1993, Wolkowitz et al., 1999). Altogether these data suggest an overactivity of the HPA system in the etiology of mood disorders (Stokes, 1995). So far, the mechanisms of the HPA axis disinhibition in depression have not been elucidated and could involve any anatomical level of regulation, from higher brain centers to the adrenals.

Corticosteroid receptors play a key role in HPA system regulation, because they mediate the negative feedback of corticosteroids in higher centers such as hypothalamus and hippocampus. A dysfunction of these receptors would explain the HPA axis over-activation observed in depressive patients. Alternatively, a primary upregulation of CRF could lead to a disinhibition of the HPA system and a secondary corticosteroid receptor downregulation, which would also—by a vicious circle—result in excessive plasma cortisol levels. An open clinical trial suggested that a GR antagonist (mifepristone, RU486) has antidepressant effects in psychotic major depression, which represents the type of depression with the most severe abnormalities in the HPA system (Belanoff et al., 2002). This effect could be obtained within a few days, in contrast to the few weeks that conventional antidepressants usually need to become effective (Belanoff et al., 2002). A large number of animal studies support the view that unrestrained secretion of CRF in the central nervous system produces several signs of depression-like and anxiety-like disorders through continuous activation of CRF receptors (Stout et al., 2002). Thus, intracerebroventricular or site-specific injection of CRF to rats or mice results in arousal and anxiety-like behavior (Weiss et al., 1986a, Buwalda et al., 1997, Kagamiishi et al., 2003).

Despite the obvious difficulties to imitate in animals psychiatric diseases that are very specifically related to the human condition (with symptoms such as thoughts of guilt, suicide and death), numerous models have been developed and proposed to mimic essential features of depression. Because such animal models have been reviewed in other articles of this issue, we would like to point out the genetic models among them. Confusion should be avoided by the distinction between a model and a test. A model can be defined as an (non-human) organism or a particular state of an organism that reproduces aspects of the human pathology, providing a certain degree of predictive validity (Geyer and Markou, 2000). A test, on the other hand, provides only an end-point behavioral or physiological measure (read-out) designed to assess the effect of a genetic, pharmacological or environmental manipulation. In this respect, models of depression are different from tests used to monitor the effects of antidepressants. So-called ‘behavioral despair’ paradigms such as the Porsolt et al. (1977) forced swim test and the tail suspension test (Steru et al., 1985), therefore, do not represent depression models for several reasons. These reasons include the question whether immobility, the major read-out in these tests, really reflects despair. Immobility could also reflect energy conservation. The major argument, why these tests cannot be considered etiological models of depression is that the behavioral read-out is altered by acute rather than chronic drug treatment. Nevertheless, their efficiency to screen for antidepressant-like activity is, generally, recognized because only few false positives have been detected in these tests. Furthermore, increased immobility time in these tests can still be regarded as a part, i.e. a single sign, of a more complex depression-like syndrome. Such a syndrome would ideally comprise several features of depression-like behavior and characteristic biological or endocrinological alterations in addition, representing a model reconciling with the complex human condition. However, this strategy holds the danger of anthropomorphic transfer of emotions from humans to mice (Crawley, 2000).

According to Willner and Mitchell (2002), an appropriate animal model for human depression should fulfill the following criteria as best as possible: strong phenomenological similarities and similar pathophysiology (face validity), comparable etiology (etiological validity), and common treatment (predictive validity). The relative importance of the individual criterion is still controversial but will not be discussed here (for review see also other articles of this issue as well as Geyer and Markou, 2000, Willner and Mitchell, 2002, Cryan and Mombereau, 2004, Willner et al., 1992, Willner, 1997, McKinney and Bunney, 1969, McKinney, 2001). As pointed out earlier, stress is an important cause of depression. Chronic mild stress (Willner, 1997) as well as learned helplessness (Seligman and Beagley, 1975), where animals are exposed to a series of stressors, therefore, constitute models with good etiological validity. Many of the behavioral features induced by stress in rodents bear similarities to human depression, in particular anhedonia (i.e. loss of interest in pleasurable things), which is a core symptom of the depressive syndrome. Many of these ‘symptoms’ are reversed by antidepressant treatment. Pharmacological depression models have also been established, including the use of reserpine, which depletes the aminergic pools in the presynaptic nerve terminals, imitating the pathophysiology that is postulated by the monoamine hypothesis, including anhedonia (Bourin, 1990, Almeida et al., 1998, Skalisz et al., 2002). Some effects of reserpine administration are similar to human depression and are reversed by antidepressants or ECT (Redrobe and Bourin, 1999, Vetulani et al., 1986). However, some authors consider this model as inadequate because of its lack of etiological validity and its non-specific effects on all types of monoamine neurotransmission. In their view this model rather mimics Parkinson's disease associated depression (Skalisz et al., 2002). Recently, withdrawal from drugs like amphetamines, which in humans may cause a depressive syndrome, has been shown to increase immobility in the forced swim test and to affect intracranial self-stimulation in rodents (Cryan et al., 2003, Anraku et al., 2001, Noda et al., 2000). Therefore, amphetamine withdrawal has been proposed as a measure for inducing depression in rodents (Cryan et al., 2003, Barr and Phillips, 1999, Lynch and Leonard, 1978).

A major drawback of the previous models is that the pathophysiological changes that are observed may rather reflect the manipulation by which this state was induced, e.g. stress or pharmacological treatment, than being a biological correlate of the animals’ ‘depressive state’. For example, the downregulation of BDNF in a stress paradigm may rather result from the stress procedure itself than represent a molecular correlate of concomitant depression-like behaviors. Therefore, alternative animal models of affective disorders would be endogenous, genetic models, more independent of external factors, which mimic essential aspects of the human disorder and respond to standard regimens of therapy. Such animal models could be designed by introduction and/or disruption of genes. Genetically, modified animals also allow to study whether and how specific genes participate in the pathophysiology of a particular disease and to identify new targets for treatment. Several lines of mice have already been generated to identify the role of genes postulated to be involved in depression according to one of the different theories: the monoamine, the neurotrophin and the HPA axis hypotheses. Recently, also other neuromodulatory systems have been implicated in depression, e.g. substance P, neuropeptide Y and the immune system, including in particular the activation of interleukins.

Depression has a multifactorial and multigenetic pathology and the number of genes involved is still unknown. According to the diathesis-stress concept three major factors determine the individuals risk to develop a depressive episode: genetic predisposition, early life stress and acute or chronic stress preceding the index episode. All three factors will become manifest at the level of gene expression. Therefore, an ideal genetic model would aim to reflect this by manipulating different genes conditionally (temporally and regionally) during ontogenesis. However, so far there are no definite data available, when and where specific genes are dysregulated prior to the onset of a depressive episode. Furthermore, although a differential dysregulation of several genes may soon become possible in mice, currently used mutant animals usually have a single mutation, which is at best region-specific or even inducible. In the following chapters we will, therefore, present a panel of mice with single mutations of genes that have been postulated to be involved in the pathogenesis of depression according to current molecular concepts.