The physiology of aggression: towards understanding violence
The research reported in this thesis was carried out at the Department of Behavioral Physiology of the University of Groningen (Biology Center, Haren, The Netherlands). The completion of this thesis was supported by the educational program of the BCN (Graduate School of Behavioral and Cognitive Neurosciences, University of Groningen). Printing this thesis was sponsored by Noldus Information Technology BV, Data Sciences International, BCN, Faculty of Mathematics and Natural Sciences, and the University of Groningen.
Cover image:
Cover design: Lay-out: Printer:
“Cain and Abel” (1542-44), Titian, Oil on canvas Santa Maria della Salute, Venice, Italy Doretta Caramaschi Dick Visser Drukkerij van Denderen BV, Groningen, NL
ISBN: 978-90-367-3985-6
© 2009 by Doretta Caramaschi No part of this book may be reproduced or transmitted in any form or by any means without prior written permission of the author.
RIJKSUNIVERSITEIT GRONINGEN
The physiology of aggression: towards understanding violence
PROEFSCHRIFT
ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. F. Zwarts, in het openbaar te verdedigen op vrijdag 9 oktober 2009 om 14.45 uur
door
Doretta Caramaschi geboren op 23 november 1979 te Suzzara, Italië
Promotor: Copromotor:
Prof. dr. J. M. Koolhaas Dr. S. F. de Boer
Beoordelingscommissie:
Prof. dr. B. Olivier Prof. dr. A. J. W. Scheurink Prof. dr. A. Sgoifo
“Mankind must evolve for all human conflict a method which rejects revenge, aggression, and retaliation. The foundation of such a method is love." Martin Luther King Jr., December 11, 1964
Table of contents CHAPTER 1
General introduction
9
CHAPTER 2
Is there co-selection for aggressiveness, coping strategy and emotionality in mice? 45 CHAPTER 3
Development of violence in mice through repeated victory along with changes in prefrontal cortex neurochemistry
61
CHAPTER 4
Is hyper-aggressiveness associated with physiological hypo-arousal? A comparative study on mouse lines selected for high and low aggressiveness
81
CHAPTER 5
Differential role of the 5-HT1A receptor in aggressive and non-aggressive mice: an across-strain comparison.
99
CHAPTER 6
Changes in serotonin-1A receptor functionality with social experience in mouse lines selected for high and low aggression
119
CHAPTER 7
Tryptophan-free diet lowers fronto-cortical serotonin levels with no effect on mouse aggression
133
CHAPTER 8
Dynamic intracellular distribution of serotonin-1A receptors in mice predisposed to violence
147
CHAPTER 9
General discussion
161
Reference list
175
Nederlandse samenvatting
199
Sintesi della tesi in italiano
205
Acknowledgements
212
1
CHAPTER
General introduction
CHAPTER 1
Aggression is a behaviour expressed in a social context, when two individuals are in conflict with each other. From a biological point of view, aggression is necessary to achieve social dominance and thereby procure resources and the means to reproduce. Thus, a certain level of aggressiveness is generally considered to contribute to higher fitness. In non-human animals, aggression in competition over resources is considered a normal behaviour. In human societies, in contrast, aggression is often unwanted and can reveal itself as a symptom of psychopathologies. Human aggression can entail violent acts that cause suffering and eventually death and that are therefore punishable by law. Despite various political and financial efforts to reduce human violence, little is known about which prevention and intervention programmes are the most effective (Krug et al., 2002). Part of the problem lays in the fact that we still lack a clear mechanistic explanation of violence. Most of the animal models for aggression that have tried to address this issue cannot be translated easily into the violent human phenotype. The aim of this thesis is to identify central and peripheral physiological mechanisms associated with extreme aggressiveness using animal models, with the ultimate aim to contribute to more evidence-based intervention and prevention programs for human violence. Knowledge of these mechanisms may also have implications for the guidance of public and judicial policies.
AGGRESSION AND VIOLENCE Definitions A considerable problem in the study of aggression across species is the confusion between the definitions of aggression and violence. The two constructs overlap, but there are important differences. Both terms were first used to describe human behaviour. In the Oxford English Dictionary, aggression is defined as “Hostile or destructive tendency or behaviour, held to arise from repressed feelings of inferiority, frustration or guilt. In addition, feeling or energy displayed in asserting oneself, in showing drive or initiative, aggressiveness, assertiveness, forcefulness. (Usu. as a positive quality.)”. Violence, on the other hand, is “The exercise of physical force so as to inflict injury on, or cause damage to, persons or property; action or conduct characterized by this; treatment or usage tending to cause bodily injury or forcibly interfering with personal freedom.” (Simpson and Weiner, 2000). It is noticeable that these definitions cannot easily be translated in nonhuman animal behaviours, since we can neither measure objectively “feelings of inferiority”, “drive or initiative” and “assertiveness” nor know the motivation “to inflict injury”. Another point of confusion is the positive or negative connotation that aggression assumes depending on the circumstances. Often, in humans, it is 10
GENERAL
INTRODUCTION
almost overlapping with violence, indicating an unwanted behaviour. Alternatively, it is sometimes used to describe a positive behaviour when it overlaps with assertiveness and drive. Since none of these concepts is applicable to nonhuman animal behaviour, translational (across-species) research on aggression and violence needs operational definitions that consist of objectively measurable behaviours both in human and non-human studies. Non-human ethological research has recognized the function of agonistic interactions in survival and reproduction for decades. Aggression is primarily an adaptive behaviour and is part of a more general proactive coping strategy aimed at mastering challenging environmental situations. Due to its important biological function, a certain degree of aggression is evolutionarily conserved. This functional aggression (Natarajan et al., 2009), as described in non-human animals, shows moderate intensity, context-dependency and sensitivity to inhibitory cues. Violence, as a more extreme form of this behavioural expression, goes beyond these contextual inhibitory characteristics; it often results in maladaptive outcomes such as injuries to partner and progeny. The pathological characteristics of this exaggerated behaviour make it more comparable to human violence. A growing number of studies report examples of non-human animal violent behaviour, even among invertebrates. The distinction between functional aggression and violence is an important issue to consider in the study of violence, since the physiological determinants of aggression might not overlap completely with the pathological determinants of violence, and the interpretation of data obtained in such studies might differ drastically. Therefore, when studying violence in a non-human animal model it is important to think carefully about definitions. Can we define violence in a rodent model? What should the criteria be to allow us to compare it to human violence? I touch upon this issue particularly in Chapter 3, although it is a major theme throughout the thesis. Types of aggression Another confusing point in the study of aggression is the fact that the term “aggression” refers both to the life-stable behavioural trait and to the discrete actual aggressive act or state. Aggression, as a behavioural trait, is the propensity to engage in agonistic interactions. As a behavioural state, aggression, or the aggressive act, can be categorized in different types. Most of the research on aggression in non-human species is carried out in rodents. Depending on the stimulus eliciting the aggressive act, different types have been described in rodents, namely predatory, intermale, fear-induced, territorial, irritable, sex-related, maternal and instrumental (Moyer, 1968). Except for predatory aggression, which is typically interspecies, from a predator to its prey, the other types are 11
CHAPTER 1
expressed in intraspecific interactions. Mothers defending their progeny exhibit maternal aggression, while the other types are usually found in males. Fearinduced and instrumental aggression differ mainly in the accompanying arousal level, which is higher in the former than the latter. Further research in rodents identifies a major distinction between offensive and defensive aggression, where the difference lies in the presence or absence of a perceived threat immediately before the expression of aggression (Blanchard and Blanchard 1977; Blanchard and Blanchard 2003). Similarly, in humans, there is a consensus about two distinct types of aggression, namely reactive aggression (affective, emotional or hostile aggression) and the premeditated, instrumental or proactive aggression (Vitiello and Stoff 1997). It seems that most aggressive children display mainly reactive aggression, while a smaller proportion shows both reactive and instrumental aggression. Reactive aggression is conceptually close to impulsive aggression, yet individuals that engage in instrumental aggression are very often also highly impulsive in other contexts; yet they form a separate group from the impulsive aggressive individuals. Reactive and proactive types of aggression are often described in children and are thought to be the result of the activation of distinct neural pathways and to correlate with different levels of physiological reactivity. In this thesis I focus on male offensive aggression, although the characterization of female aggression and non-offensive subtypes of aggression in rodents needs to be explored further.
PERIPHERAL PHYSIOLOGY OF AGGRESSION Stress hormones There is a strong intuitive connection between aggression and stress. In certain individuals, aggression is displayed as part of the effort to cope with a social challenge or stressor (Benus et al., 1991b; Koolhaas et al., 1999). At the peripheral level, any perceived stressor activates two main physiological pathways: the sympatho-adrenomedullary (SAM) system and the hypothalamic-pituitaryadrenocortical (HPA) axis. Aggressive individuals tend to respond physiologically to stress with a high activation of the SAM system, while low-aggressive or reactive individuals show a higher HPA-axis activation (Benus et al., 1991b; Koolhaas et al., 1999). In mice, both offensive and defensive aggression give rise to high HPA and SAM activation during the social encounter (Bartolomucci et al., 2003). In humans, individuals with a Type A personality, e.g. those who are hostile and irritable, are more prone to develop cardiovascular problems associated with high sympathetic or poor parasympathetic cardiac control (Ward et al., 1986; Lee and Watanuki 2007). Often, display of aggression is associated with the feeling of 12
GENERAL
INTRODUCTION
anger and a high emotional response. In contrast to this reactive, hostile, emotional aggression, instrumental or cold-blooded aggression is executed in the absence of emotional arousal. Individuals that frequently engage in antisocial behaviour using instrumental aggression show a reduced autonomic baseline tone and low HPA axis tone and activation (Virkkunen, 1985; Raine et al., 1990; Scarpa and Raine 1997; Raine et al., 2000; Raine, 2002a; Raine, 2002b; Popma et al., 2006). A long-lasting physiological activation in response to stress can also be one of the causes of the development of aggressive personality. Glucocorticoids and noradrenaline released in the blood stream during a stressful situation may affect neural substrates by binding to certain receptors, specifically those on brain circuits involved in neuroplasticity, i.e. the hippocampus, and in the regulation of social behaviour, i.e. the serotonin and vasopressin systems. In vulnerable individuals, stress can lead to a subtype of depression characterized by anxiety, anger and aggressive outbursts (van Praag, 2004). Moreover, adverse early life experiences such as maltreatment or neglect can lead to the development of aggressive temperamental disturbances (Caspi et al., 2002; Van Goozen et al., 2007). Recent studies in rats have also demonstrated pro-aggressive effects in juveniles and adults that are maternally separated or early socially deprived (Veenema et al., 2006; Toth et al., 2008). In adolescent hamsters, social stress can lead to accelerated development of adult forms of agonistic behaviour via activation of the HPA axis (Wommack and Delville 2007a; Wommack and Delville 2007b). However, reduced HPA-axis functioning in adult rats causes an increase in pathological aggression, expressed as a shift to a predominance of injurious attacks on vulnerable body regions rather than ritualized, less injurious ones (Haller et al., 1998; Haller et al., 2000; Haller et al., 2001). It has been proposed that pronounced stress activation in the early phases of life tunes or primes the functioning of the neuronal fear circuit, leading to desensitization in individuals that develop antisocial behaviour (Van Goozen and Fairchild 2008). Stress-related physiological mechanisms might be a good indicator of different types of aggression and of the severity of the risk of developing violence. But are aggressive and violent mice physiologically similar to violent humans? I investigate this topic in Chapter 4. Gonadal hormones Testosterone, the male sex hormone, exerts its main function in determining the male sex during prenatal life and at puberty. Since in many mammalian societies, aggressive behaviour is typically expressed in male intrasexual competition for females and social dominance, it is logical to think of a role for testosterone and/or its metabolites in the development of an aggressive behavioural phenotype. With this in mind, previous studies have looked for an association and a 13
CHAPTER 1
causal relationship between testosterone and aggression. Basically, one of the conceptual sources for studying the neurochemical basis of social behaviour derives from the endocrine depletion-repletion research paradigm that was introduced in 1849 by Arnold Berthold. He removed the gland of interest (testis) that secreted the source of the endogenous substance (testosterone) that he suspected to be necessary for the display of a specific behaviour (aggressive displays) and consequently that behaviour disappeared. Thereafter he replaced the glandular material and recorded the return of the original behaviour (Berthold, 1849). The near-total decline in aggressive displays and fighting after castration and the effective restoration of this behaviour after administration of testosterone has been replicated in several invertebrate, fish and avian species thereby indicating the obligatory role of androgens in aggressive behaviour. However, in mammalian species like rodents and primates, instead of being obligatory in their function, androgens exert a modulatory effect on aggressive behaviour. Castrated mice and rats without prior to aggressive experience rarely fight when confronted by a male conspecific. However, when aggressive behaviour is fully established in the behavioural repertoire, castration gradually reduces but does not prevent aggression against a conspecific male (De Bold and Miczek, 1981). A positive correlation between aggression and testosterone has been shown in humans e.g. see for review (Archer, 2006) and non-human primates (Higley et al., 1996). However, the correlation is not always found, leading some to propose that testosterone is more related to dominance, which can also be expressed without aggression e.g. see for review (Archer, 2006). A cross-strain genetic analysis in mice also revealed a positive correlation between testosterone and aggression (Roubertoux et al., 2005). In line with a causal relationship is the finding that a reduction of testosterone levels in rodents leads to a decrease, and enhancement to an increase, of aggression e.g. see for review (Nelson and Trainor 2007). This effect could occur through different mechanisms. First, testosterone may act either directly or as one of its active metabolites, dihydrotestosterone (DHT), dihydro-epiandrosterone (DHEA) and estradiol or a combined androgenic/estrogenic action (see for review (Simon et al., 1996)). Second, testosterone might stimulate vasopressin synthesis and in this way enhance aggression (Ferris et al., 1989; Ferris and Delville 1994; de Vries, 2008). Vasopressin itself can also promote aggression indirectly by reducing serotonin neurotransmission in the hypothalamus (Ferris and Delville 1994). Similarly, androgens might modulate 5-HT and GABA and therefore affect aggression (Cologer-Clifford et al., 1997; Cologer-Clifford et al., 1999; Miczek et al., 2002; Mitchell et al., 2008). Another testosterone-dependent mechanism which might exert its effects through the serotonin pathway involves nitric oxide (Nelson and Trainor 2007). 14
GENERAL
INTRODUCTION
NEUROBIOLOGY OF AGGRESSION Neuroanatomy and neurophysiology A schematic overview of a putative neural network involved in aggression is presented in figure 1.1 (Koolhaas et al., in press). As an emotional behaviour, aggression involves the activation of brain structures involved in emotion control and appraisal (Dalgleish, 2004). The expression of an aggressive state, studied in detail in cats, is associated with modulation by the midbrain on sympathetic activation (Bard, 1928). Electric stimulation of the hypothalamus in cats, rats and monkeys is sufficient to induce aggression together with neuroendocrine and autonomic activation. During an aggressive state, brain areas related to the organization of general stress responses like the stria terminalis, mediocentral amygdala, hypothalamus, pituitary, septum, locus coeruleus, dorsal periacqueductal grey and orbitofrontal cortex become activated e.g. see (Gregg and Siegel 2001). Recently, using modern brain imaging techniques (fMRI) in awake rats that were triggered to aggressive motivation, this suspected putative neural circuitry of aggression was indeed activated together with the unexpected intense activation of anterior thalamic nuclei (Ferris et al., 2008).
Gonadal steroids
Olfactory input
Sensory input
VNO
Thalamus
Aob Bnst
Aha
Poa
Mea
LS
PFC
Motor cortex
Basal ganglia
PAG Neuroendocrine output
Autonomic output Nociception
Motor output
Figure 1.1 Schematic diagram of a putative neural network involved in offensive aggression (Koolhaas et al., in press).VNO= vomeronasal organ, Aob= anterior olfactory bulb, Mea= medial amygdala, LS= lateral septum, PAG= periacqueductal gray, Bnst= bed nucleus of stria terminalis, Poa= hypothalamic preoptic area, Aha= anterior hypothalamic area, PFC= prefrontal cortex'
15
CHAPTER 1
Instrumental aggression is not associated with emotional arousal and might be more related to developmental impairments in amygdala-based aversive conditioning, as well as dysfunctions in error-monitoring and emotional appraisal in the anterior cingulate cortex, e.g. see for review (Crowe and Blair 2008). Instrumental aggression, in particular its unemotional and callous characteristic, might also be related to low empathy and brain dysfunctions in areas involved in empathy (Patrick and Zempolich 1998). In general, prefrontal cortical brain regions are responsible for behavioural planning, execution, inhibition and emotional and sensorimotor integration. Therefore my thesis focuses on this anatomical area. Neurochemistry: serotonin The best-known neurochemical mechanism associated with violence is the brain serotonergic system. Initially the association was found in isolated mice that became aggressive while reducing their serotonin turnover, which is an indicator of serotonin neurotransmission and consequent degradation (Garattini, 1967; Giacalone et al., 1968; Modigh, 1973). Similar studies were then performed in human and non-human primates, confirming a negative correlation between serotonin metabolite levels in the cerebrospinal fluid and high, impulsive aggression levels (Brown et al., 1979; Brown et al., 1982; Linnoila et al., 1983; Virkkunen et al., 1989; Limson et al., 1991; Mehlman et al., 1994; Higley et al., 1996; Fairbanks et al., 2001). The negative correlation between serotonin turnover and aggression became known as the serotonin-deficiency hypothesis of aggression. A proper functioning of the serotonergic system involves the integration of several processes: serotonin synthesis, release, receptor activation, re-uptake and degradation. From the serotonin-deficiency hypothesis as a starting-point, followup research has pursued the idea that changes in one or more of these elements might explain the low serotonin metabolite levels in aggressive/violent individuals. The functioning of the enzyme tryptophan hydroxylase (TPH) tightly regulates serotonin synthesis. Manipulating serotonin synthesis by reducing the availability of tryptophan, the essential precursor of serotonin, increases aggressiveness in various species and contexts (Gibbons et al., 1979; Chamberlain et al., 1987; Kawai et al., 1994; Cleare and Bond 1995; Bjork et al., 2000; Bond et al., 2001; Bell et al., 2001). Early studies on the effects of the TPH inhibitor parachlorophenylalanine documented increases in aggression in rats (Sewell et al., 1982; Vergnes et al., 1986). Moreover, a single-nucleotide polymorphism in the coding region of the TPH gene has been associated with anger and aggression in healthy human subjects (Rujescu et al., 2002; Hennig et al., 2005). Serotonin release is reduced in the prefrontal cortex of rats after fighting against an intruder (van Erp and Miczek 2000), while stimulation of serotonin 16
GENERAL
INTRODUCTION
release with fenfluramine reduces aggression in human subjects, especially those with a history of aggression (Coccaro et al., 1996; Cherek and Lane 1999; Cherek and Lane 2001). Serotonin release is under the control of the inhibitory autoreceptors 5-HT1A and 5-HT1B, located distally and proximally, respectively, to the synapses (Pineyro and Blier 1999). Activation of 5-HT1A or 5-HT1B receptors on the serotonergic neurons reduces aggression in rats and mice (de Boer et al., 2000; Bannai et al., 2007). 5-HT1A receptors are described in more detail in the box “The 5-HT1A receptor: functions and regulation” and figure 1.2. Serotonin release is triggered by action potentials in the serotonergic cells and is modulated by heteroreceptors, for example NMDA, GABAA/B, α1/2 and D2, which are activated by release of their specific neurotransmitters by neuronal projections originating from different brain areas (Pineyro and Blier 1999). It is not yet clear how these heteroreceptors on serotonergic neurons are involved in aggressive behaviour. Once released, serotonin is transiently available in the synaptic cleft for neurotransmission. The extracellular availability of serotonin depends on the functionality of the serotonin transporter (5-HTT) and the monoamino-oxidase (MAO) enzyme, responsible for reuptake and degradation, respectively. Acutely, inhibition of 5-HTT with SSRIs leads to reduction of aggression in several animal models (Olivier et al., 1989; Ferris and Delville 1994; Dodman et al., 1996), and mice and rats lacking 5-HTT show reduced aggressiveness (Holmes et al., 2002;
Figure 1.2 Schematic representation of a serotonin neuron and the position of the inhibitory 5HT1A receptors on it (autoreceptors) and on a non-serotonergic neuron (heteroreceptors).
17
CHAPTER 1
Homberg et al., 2007). It therefore seems that the longer serotonin is available at the synaptic cleft, the lower the aggressiveness. However, during development, low activity of the MAO-A enzyme due to genetic polymorphisms is associated with a highly aggressive personality and an increased predisposition to develop antisocial behaviour and conduct disorder (Brunner et al., 1993; Manuck et al., 2000; Caspi et al., 2002; Foley et al., 2004). Accordingly, deletion of the MAO-A gene in mice results in increased aggressiveness (Cases et al., 1995). These studies suggest that serotonin availability at the synapses might have differential roles during different developmental stages. Serotonin binds to at least 14 subtypes of specific postsynaptic membranebound receptors, each with a different location and function. In vivo pharmacological studies have provided evidence for decreased functionality of poststynaptic 5-HT2A/C and 5-HT1A heteroreceptors, probably hypothalamic, in aggressive individuals (Coccaro et al., 1989; O'Keane et al., 1992; Coccaro et al., 1995; Moeller et al., 1998; Cherek et al., 1999; Cleare and Bond 2000). Similarly, decreased 5-HT2-induced activation of the prefrontal cortex and the anterior cingulate and a decreased amount of 5-HT1A receptors in the anterior cingulated, prefrontal cortex, amygdala and dorsal raphe have been found in aggressive human subjects (Parsey et al., 2002; New et al., 2002). Although there seems to be a consensus that low serotonergic neurotransmission is linked to high aggressiveness, other studies show a positive correlation or no correlation between serotonin neurotransmission and aggressiveness. In most of the cases, contradictory reports can be explained by examining in detail the different definitions of aggression used and the choice of subjects. Typically, research on aggression in non-human animals and in healthy humans has generated results that are in apparent conflict with data on pathologically violent human offenders. The latter suggests that the role of the serotonergic system might differ between the different subtypes of aggression and it is therefore important to take into account the specific type of aggressive behaviour studied. The paradigms used to assess components of the serotonergic system also seem to matter in adding complexity to the interpretation. Indeed, the serotonergic system is a highly complex and dynamic network that is not restricted to a modulatory function in the central nervous system, but has also a key role in the periphery, where serotonin serves physiologically essential functions. Is serotonin related specifically to violence and therefore the most pathological cases of aggression, or to aggressive tendency in general? How is serotonergic neurotransmission characterized in aggressive and violent individuals? What are the role and the dynamics of the inhibitory 5-HT1A receptors in violent individuals? Chapters 5 , 6, 7, and 8 provide some answers to these questions.
18
GENERAL
INTRODUCTION
Other neurochemical correlates Other neurotransmitter systems have been linked to aggression, such as dopamine, GABA, glutamate and noradrenaline (Nelson and Trainor 2007). Dopaminergic D2 antagonists are used clinically to treat violent outbursts. D2 receptor agonists (but not D1) induce defensive behaviour in cats. Elevated frontocortical dopamine precedes attacks and defensive behaviour in rats and amphetamine-induced dopamine release can increase aggressive behaviour in mice and rats. However, the specificity of dopamine in the control of aggression remains speculative since many of the dopaminergic compounds enhance or reduce general activity (agitation, sedation, fatigue)(see for review DE Almeida et al., 2005). GABA is generally found to be at a low level in aggressive individuals. However, compounds that modulate the activity of GABAA receptors, including alcohol, may have opposite effects on aggression, also depending on type of aggression. The anti-aggressive effects of benzodiazepines seem to be due to sedation, except for escalated/pathological forms of aggression in mice and rats in which there is a more selective antiaggressive effect (see for review De Almeida et al., 2005). Glutamate excitation of the PAG elicits defensive rage behaviour in cats and is involved in exaggerated emotional reactivity leading to seizures (Gregg and Siegel 2001). Again, the specificity of the antiaggressive effect of compounds acting on glutamate NMDA receptors is dependent on the aggression model used. Belozertseva and Bespalov (1999) reported that in isolation-induced mouse aggression it was mainly due to sedation, whereas it seemed to be more specific for aggression in the morphine-heightened aggression model (Belozertseva and Bespalov 1999). Since all these neurotransmitters exert modulatory actions on the serotonergic neurons and vice versa, their effects on aggression might due to the indirect interaction with components of the serotonergic system. Convergence on serotonin as unifying mechanism The aforementioned neural and endocrinological mechanisms proposed to explain aggression in mammals share a common feature: the involvement of serotonin along their upstream or downstream pathways. A summarized overview is depicted in Figure 1.3.
19
CHAPTER 1
Figure 1.3 Schematic diagram of hormones and neurotransmitters found to be related to aggression and serotonin. See text for abbreviations and details.
20
GENERAL
INTRODUCTION
Box 1.1: THE 5-HT1A RECEPTOR: FUNCTIONS AND REGULATION The serotonin-1A receptor (5-HT1A-R) is the most studied of the serotonin receptors, since it was the first to be visualized in tissue, cloned and manipulated pharmacologically. The 5-HT1A protein is coded by an intronless gene, Htr1a, located on chromosome 13 in the mouse genome and on chromosome 5 in the human genome. Htr1a codes for a protein of 421-422 aminoacids, with high homology between mouse and human (Sundaresan et al., 1989). Upstream non-coding regions have been identified as domains for transcriptional regulation for 5-HT1A (Le Francois et al., 2008). Although there is no high-resolution image of the protein, its predicted structure consists of seven putative hydrophobic transmembrane domains, with the amino terminus oriented facing the extracellular space, three hydrophilic intracellular loops and three extracellular loops. N-glycosylation sites at the extracellular terminus suggest that the protein is transferred from the ER, where it is produced, to the Golgi, where it is glycosylated (Raymond et al., 1999). Due to a specific motif in the C-terminus and to interaction with other proteins, 5-HT1A-R is then translocated to specialized regions of the cell membrane (Carrel et al., 2006; Carrel et al., 2008). 5-HT1A-R tends to be localized in specific membrane microdomains (Kalipatnapu and Chattopadhyay 2005), where it exerts its function of G-protein couple receptor, GPCR. 5-HT1A-R is expressed in several regions of the central nervous system, specifically the lateral septum, hippocampus, frontal and enthorinal cortices, anterior raphe nuclei, neocortex, several thalamic and hypothalamic nuclei, nucleus of the solitary tract, dorsal tegmentum, nucleus of the spinal tract and of the trigeminal nerve, and in the spinal cord (Palacios et al., 1990; Chalmers and Watson 1991; Kia et al., 1996b). It is mainly localized on the soma and dendrites of neurons, and possibly also on glial cells. Peripherally, 5-HT1A-R is also expressed in the skin, gut and blood mononuclear cells (Gershon et al., 1990; Yang et al., 2006; Nordlind et al., 2008). The receptor has a high affinity for serotonin. As GPCR, the 5-HT1A receptor exerts its main cellular function through activation of G-proteins upon agonist ligand stimulation. The main signalling mechanisms in which 5-HT1A-R is involved are: (i) inhibition of adenylyl
21
CHAPTER 1
cyclase, and consequently decrease of cyclic AMP, through coupling with Gi/o proteins; and (ii) activation of phospholipase C and other phospholipases, and consequently PKC, through G-protein βγ-subunits. Through these mechanisms, and depending on the cells in which it is expressed, 5-HT1A-R stimulates: (i) inward current and intracellular mobilization of Ca2+; (ii) Erk, Akt and nuclear factor κB, and consequently acts on proliferation and cell survival pathways; (iii) production of reactive oxygen species and nitric oxide; and (iv) active ion transport. 5-HT1A-R also activates G-protein-gated inwardly rectifying K+ channels (GIRK channels), with consequent neuronal hyperpolarization, see for review (Raymond et al., 1999). Differences in functions exerted through different brain regions might depend on a differential coupling to G proteins. Functional mutations in the second intracellular loop might influence the coupling itself (Mannoury la Cour et al., 2006; Kushwaha et al., 2006). In the brain, 5-HT1A-R exerts important functions, such as the regulation of circadian rhythms, mood, eating, fear conditioning and social behaviour (de Boer et al., 2000; Albert and Lemonde 2004; Horikawa and Shibata 2004; Ebenezer et al., 2007; Shields and King 2008). At the physiological level, the 5-HT1A receptor is involved mainly in temperature and cardiovascular regulation (Hjorth, 1985; Nalivaiko and Sgoifo 2008; Audero et al., 2008). As mentioned above (see Serotonin deficiency), 5-HT1A-R is involved in these functions by acting as a postsynaptic heteroreceptor in non-serotonergic cells and as a presynaptic autoreceptor in serotonergic cells. The intracellular functions in which the 5-HT1A autoreceptor is implicated have important consequences for the serotonin-producing neurons and are therefore a candidate pathway for the research on aggression and violence. The functionality of 5-HT1A-R is dependent, as mentioned earlier, on its expression levels, i.e. transcriptional regulation. Moreover, 5-HT1A-R can be phosphorylated and therefore desensitized by PKA, PKC and GRK (G-proteincoupled receptor kinase) (Raymond et al., 1999). Finally, 5-HT1A-R functionality is attenuated by regulators of G-protein signaling (RGS) proteins (Raymond et al., 1999). The functional desensitization of 5-HT1A-R is also obtained through agonist-mediated endocytosis/internalization (Riad et al., 2001; Riad et al., 2004). In relation to aggression research, malfunctioning at any of these regulation levels may have important consequences for the functioning of the serotonergic system.
22
GENERAL
INTRODUCTION
INDIVIDUAL VARIATION Individuals showing different personality types may differ largely in aggressive traits. Human personalities are generally described in terms of five independent factors (Digman, 1990). Several behavioural variables are correlated with each other and cluster within one of these factors. These five dimensions are Neuroticism, Extraversion, Conscientiousness, Agreeableness and Openness to Experience. Aggression seems to be associated with both high Neuroticism and low Agreeableness (or high Antagonism). The neurotic aggressive type is characterized by emotional instability, anger and irritability, while the antagonist aggressive type is hostile and antisocial. A recent meta-analysis (Bettencourt et al., 2006) found positive correlations between high laboratory aggression and high self-reported trait aggressiveness and irritability both under unprovoking and provoking conditions. Other personality traits such as trait anger, type A personality, narcissism and impulsivity were only associated with aggression under provocation. The authors concluded in the first instance that there is a strong association between certain personality dimensions and physical aggression and, secondly, that trait aggressiveness and trait irritability include a mixture of proactive and reactive aggression, while trait anger, type A personality, narcissism and impulsivity relate to reactive aggression. Longitudinal studies confirm the idea that individuals show life-stable aggressive traits from early life and that some individuals show a propensity to commit violent crimes from youth (Nagin and Tremblay 1999; Loeber and Pardini 2008). Although it is clear that certain personality types are more prone to exhibit aggression, the question is whether these individuals are at risk of showing violence, defined as an extreme and injurious form of aggressive behaviour usually not acceptable in our society. From prison settings and criminal records, it seems that individuals with antisocial personality disorder and/or psychopathic personality are the most common violent types. Antisocial personality disorder is defined by the Diagnostic and Statistical Manual of Mental Disorder (fourth edition) as a “pervasive pattern of disregard for, and violation of, the rights of others that begins in childhood or early adolescence and continues into adulthood”. It is characterized by child conduct problems, impulsivity, irresponsibility, absence of long-term goals and poor behavioural control. Psychopathy partially overlaps with these behavioural deviances, but adds an extra dimension of emotional detachment and has various physiological correlates. A low proportion of individuals with antisocial personality disorder expresses signs of psychopathy (antisocial psychopaths). Individuals with antisocial personality disorder are partly defined by their criminal behaviour and therefore they are typically violent. In prisons, high psychopathy is associated with a high incidence of violent crimes and recidivism, especially in 23
CHAPTER 1
low-IQ and socially withdrawn psychopaths (Patrick and Zempolich 1998). However, in the population there is also a proportion of individuals characterized by unemotional behaviour that is not responsible for violent crimes or is not imprisoned. Interestingly, the psychopathic personality, or at least some of its most salient features, can be assessed already in children. Therefore, these features could potentially be used to identify a risk of becoming a violent offender. Given the difficulty in defining and quantifying behavioural traits that would predict a risk for the development of a violent personality, it is necessary to identify stable individual differences in physiological and neurochemical traits that correlate with stable individual differences in aggression and violence. I try to elucidate this association throughout the thesis.
ANIMAL MODEL: MOUSE SELECTION LINES FOR HIGH AND LOW AGGRESSIVENESS In order to address the issues mentioned above, I used a unique tool: three pairs of mouse lines selected for high and low aggression. A few decades ago, three independent laboratories bred aggressive mouse lines through artificial selection for certain aggression parameters. This resulted in three different animal models for aggression, each of them represented by a highly aggressive and a low/nonaggressive line. The details of the selections are summarized in Table 1.1. In Turku, Finland, Lagerspetz and Sandnabba developed the Turku-Aggressive (TA) and the Turku-Nonaggressive (TNA) lines from an outbred colony of Swiss albino mice (Lagerspetz, 1961). The mice from each line were kept in different rooms, weaned at 21 days of age and individually housed. At 60 days of age, males were tested for aggression in a standard 7-minute dyadic test in a neutral arena (clean glass container) against standard opponents (pretested nonaggressive animals). The aggressive behaviour was rated on a 7-point scale as follows: 1) The animal shows no interest in its test partner except occasional nosing. Tries to escape, squeaks, is immobile, if attacked by the partner. 2) Frequent nosing. Escapes, but tries occasionally to protect itself from the attacks of the partner. 3) Frequent vigorous nosing. The animal assumes occasionally a position of readiness for fight. The animal does not attack the test partner but protects himself when attacked. 4) Tail rattling, vigorous nosing. The animal follows and occasionally attacks the test partner. 5) Slight wrestling and occasional powerful attacks. The animal attacks its partner and bites. Tail rattling. 24
Wild-type mice captured in Groningen, The Netherlands
Van Oortmerssen and Bakker, 1981
Lagerspetz, 1961 Lagerspetz and Lagerspetz, 1971
Cairns et al., 1983
SAL LAL
TA TNA
NC900 NC100
Out-bred ICR (Institute of Cancer Research)
Out-bred colony of Swiss albino mice
Background strain
Obtained Reference lines
Table 1.1 Overview of the three selection programs.
Housed in relative isolation. At 45 days of age. In a neutral partition cage, against a group-reared test partner. 5 minutes with slide closed and 10 minutes with slide open (video recording).
Housed individually. At the age of 60 days. In a neutral cage against a non-aggressive mouse pre-tested.
Housed with a female. At the age of 14 weeks. On the edge of the home cage against a docile albino intruder (MAS-Gro). Resident-intruder paradigm repeated on three consecutive days.
Aggression test
Attack-frequency/ Attack latency score up to 900
7-point scale ratings (see text)
Average of the three Attack Latency Tests= Attack Latency Score (ALS)
Parameters scored
Brothers-sisters mating avoided
Brothers-sisters mating avoided
Brother-sisters mating only in the first four generations
Breeding
GENERAL INTRODUCTION
25
CHAPTER 1
6) Fierce wrestling and biting during most of the period. 7) Fierce wrestling. The animal bites the partner hard enough to draw blood. Males with high aggression scores were used for breeding together with females, sisters of high-scoring males, avoiding brother-sister mating (Lagerspetz and Lagerspetz 1971). The selection produced an aggressive line in the 11th generation, while the other line was already non-aggressive in the first generation. However, mice that showed some aggression were always present in the TNA line. Breeding pairs of both lines (10 of each, 77th generation) were transferred to Groningen, The Netherlands in 2003 and are currently in their 83th generation of selection. In Groningen, the Netherlands, the selection was carried out in a colony randomly bred from wild house mice trapped in a mansion near Groningen (van Oortmerssen and Bakker 1981). To breed the Short Attack Latency (SAL) and Long Attack Latency (LAL) lines, mice were kept after weaning (3–4 weeks) in unisexual litters until sexual maturity (7–9 weeks), at which point they were male-female paired in small cages. At the age of 14 weeks, each male mouse was tested for aggression in a resident-intruder test carried out in a large cage divided into 4 compartments (A, B, C and D). The resident experimental animal was housed in A and B, while the compartment C was the test arena. After an hour of exploration of compartment C, the experimental animal was tested with a naive albino intruder (MAS-Gro), previously confined to compartment D (sensory precontact allowed by a perforated Plexiglas slide). From the point of slide removal onwards, the attack latency was recorded. Later studies confirmed that this parameter correlates negatively with other parameters of aggression (Benus et al., 1991b). The final score used to define a SAL or a LAL mouse was the Attack Latency Score (ALS), i.e. the average of the attack latency times measured over three consecutive days of testing. When an animal did not attack in 600 sec, it was considered a non-attacking mouse, and the test was stopped. While the SAL line appeared in the 4th generation already and remained quite stable, the selection of a docile LAL line failed at the first few attempts and was obtained only at the 23rd generation of the control line. In North Carolina, U.S., Cairns et al. selected a base population of out-bred NCR mice on a dual criterion, namely increased attack and heightened reactivity to stimulation (Cairns et al., 1983). In this way, a highly aggressive line and an immobile line were created (I-lines). A second attempt focused only on aggression as a selection criterion and generated the NC900 and NC100 lines. The testing procedure was as follows: after weaning at 21 days, the mice were housed in relative isolation. At 45 days of age, aggressiveness was measured in a standard 10min dyadic test carried out in a neutral Plexiglas box, after 5-min of adaptation in which physical contact was not allowed. The attack frequency (in 5-sec intervals) 26
GENERAL
INTRODUCTION
and the attack latency were measured and combined together with other 31 variables in a scoring system up to 900. The result was the selection of an aggressive line, NC900, and a non-aggressive line, NC100. Breeding pairs of both lines (10 of each, 48th generation) were transferred to Groningen, The Netherlands, in 2004 and are currently in the 52th generation of selection. Behavioural phenotypes Aggression as a behavioural trait is only part of a more general suite of correlated behaviours that an individual performs in order to cope actively with the environment, i.e. the proactive coping strategy. The presence of alternative coping strategies and of correlations between behavioural traits has been extensively investigated in the SAL/LAL animal model, while in the other two models studies have addressed certain behaviours in relation to depression/anxiety etc. A summary of the literature concerning this issue is presented in Table 1.2. The six mouse lines were developed based on high and low intermale aggression levels. However, in view of our current distinction between aggression and violence, one may wonder which of these selection lines show signs of violence. Although the attack latency of SAL and LAL is highly correlated with other measures of aggressiveness, it does not determine dominance. Body weight appears to be an important determinant in the acquisition of dominance (Van Zegeren, 1980; van Oortmerssen et al., 1985). Recently, the aggressiveness of SAL and LAL mice was compared in terms of pathological/maladaptive aspects. SAL mice were found to be more violent than LAL, since SAL attacked more intensely the opponent in vulnerable regions and caused more wounds to the opponents (Haller et al., 2006). In TA and TNA mice, aggression has been mainly studied by examining the role of odour/olfaction communication in the induction of aggressiveness. The territorial marking pattern of urination differs in these two lines. In reaction, male mice tend to avoid the TA-marked places, while females choose those of TNA males (Sandnabba, 1985). TA odour on castrates has been found to elicit aggression from NMRI and Swiss albino mice of the parental strain used for the selection, but to a higher extent when the TA mice have been trained to high aggressiveness compared to TA-defeated mice (Sandnabba, 1986a; Sandnabba, 1986b). Male and female odour preferences of the two lines are associated with aggressiveness and dominant/subordinate social position (Sandnabba, 1986c). TA and TNA mice differ in their territoriality when tested in colonies, where TA mice fight excessively and often became dominant, while TNA mice stay in their original territories with less fighting (Sandnabba, 1997). These experiments suggest some correlation between dominance, female preference and TA aggressiveness. Recently, using a more detailed statistical approach, the social behaviour of TA mice was described as high in consummate aggression (e.g. box, bite, threat, chase), and that of TNA 27
28
Reference
Veenema et al., 2003
Hogg et al., 2000
Sluyter et al., 1995 Sluyter et al., 1996
Benus et al., 1988
Benus et al., 1990
Benus et al., 1990
Benus, 1988 Benus et al., 1989
Van Oortmerssen et al., 1985 Compaan et al., 1993 Benus et al., 1987
SAL/LAL Van Zegeren, 1980
Lines
Experiment Isolated SAL and LAL, no effect of attack latency in dominance acquisition, but effect of body weight Socially housed SAL and LAL, effect of attack latency rather than body weight in the acquisition of dominance TP aggressive effect on SAL; E no effect
Results
SAL learned earlier than LAL; SAL latency to explore was not affected by an extra object in the maze, while LAL explored; SAL learned slower when a different maze was presented every day. Social stress Defeat SAL flee more and LAL freeze more Controllable Two way active shock avoidance SAL escaping more, but within LAL three groups of non-social stress different speed and learning, one group similar to SAL Uncontrollable Inescapable shock LAL activity is suppressed, while SAL are hardly non-social stress influenced Reversal learning Y-maze SAL more difficulties in changing their motor behavioural patterns, more routine-formation Circadian rhythmicity Light-dark inversion SAL adjusted their rhythm to the inversion much slower Dark-Dark than LAL; SAL had a free-running period similar to the solar period, while the one of LAL was shorter Coping strategy Nest-building SAL showed more nest building than LAL Coping strategy Shock-probe/Defensive burying LAL showed more immobility than SAL in both situations; in home cage and fresh sawdust LAL did not adopt defensive burying in fresh sawdust, while SAL actively did it. Anxiety Hexagonal-tunnel maze SAL less anxious than LAL in the hexagonal-tunnel maze, Light-dark box while no differences were found in the LD box Anxiety and social -Elevated plus maze LAL less active than SAL in all the tests and showed more stress -Sudden silence test freezing and immobility; -Open field Increase in grooming in LAL and decrease in digging -Forced swim test in two paradigms: in SAL in the sensory contact paradigm; -Sensory contact Decrease in immobility, decreases in activity after -Sensory contact + daily defeat defeat paradigm in both lines.
Aggressive behaviour Population studies in semi-natural conditions Aggressive behaviour Population studies in semi-natural conditions Female aggression Testosterone propionate or estradiol neonatal treatment Learning Standard maze
Behaviour studied
Table 1.2 Overview of the behavioural characteristics (endophenotypes) of the aggressive and non-aggressive lines.
CHAPTER 1
Sex related coping styles Anxiety
Kvist et al., 1997
Metabolism Maze learning Open field Elevated plus-maze Staircase Light-dark box 7-min dyadic interaction
Cricket intruder Colony
NC900/ Hood & Quigley, 2007 Exploration NC100
Novel arena Light/dark box Elevated plus maze Neohypophagia
Aggression RTS (reactivity tactile stimulation) Social and Non-social Dyadic interactions Reactivity Immobility NC900/ Bauer & Gariépy, 2000 Freezing behaviour in 10-min dyadic tests in juvenile mice NC100 dyadic interaction
NC900/ Gariépy et al., 1988 NC100
Vekovischeva et al., 2007
Aggression
Predatory aggression Territorial behaviour
Nyberg et al., 2003
Resident-intruder test
Aggression test Sexuality test Marking behaviour Investigation of TA/TNA marked behaviour by NMRI mice Urinary marking pattern analysis Odor discrimination Open field with odours of other mice Learning Maze learning Passive avoidance Open field ambulation Base-line open field After maze learning open field
Sexual behaviour
Aggressive behaviour/violence
Sandnabba, 1995 Sandnabba, 1997
Selander & Kvist, 1991
Kvist, 1989
Sandnabba, 1986
TA/TNA Lagerspetz & Hautojärvi, 1967 Sandnabba, 1985
Haller et al., 2006
NC900 more reactive and less freezing than NC100. Freezing as a conservation/withdrawal response that reduces emotional arousal NC900 slower approach and lower levels of exploratory behaviours. NC900 slowest to enter light box, remain longer in dark box. NC900 less drinking in novel environment
NC900 and NC100 same reactivity NC100 more immobility
TA more consummatory aggression TNA more non-aggressive social interaction
TA marked territory was avoided by NMRI mice TA mice marked the whole bottom of the cage, while TNA urinated in fewer places with bigger spots TNA avoid TA odor; TA search for TA odor, but with loosing experience they tend to avoid it TA better in maze learning (active learning) TNA better in passive avoidance (passive learning) TA more locomotor activity and shorter latency to move than TNA. TA less emotionality (defecation) than TNA All these measures were suppressed together with suppression of aggression in the TA and returned to initial levels after rest. TA show more predatory aggression than TNA TA established hierarchies quickly through fights TNA remained in their original territories Differences in coping styles are found inter- and intrastrains between males and females TA less anxious than TNA
No difference in sexual behaviour between the strains
Attacks at vulnerable regions in SAL more than in LAL
GENERAL INTRODUCTION
29
CHAPTER 1
as socially explorative (e.g. sniffing, investigation, non-aggressive contact) (Vekovischeva et al., 2007b). TNA were found to be slightly aggressive, although their aggression was low in intensity (Vekovischeva et al., 2007b). Intermale aggression is associated with other forms of aggression. TA aggressive mice show more predatory aggression than TNA mice when exposed to a cricket intruder (Sandnabba, 1995). TA and TNA maternal and testosteroneinduced aggression levels differ in accordance with TA and TNA male-male aggressiveness, suggesting a similar genetic background for female and male aggression (Lagerspetz and Lagerspetz 1975; Lagerspetz and Lagerspetz 1983; Sandnabba, 1993b; Sandnabba et al., 1994). Similar findings have been obtained in the NC lines. Maternal aggression and inter-female aggression were found to be higher in NC900 than in NC100 females, similarly to the difference observed in males, even though this difference was expressed later in life (Cairns et al., 1983; Hood and Cairns 1988). During the SAL/LAL selection, female aggression resulted in a similar pattern to male aggression in the first generations but did not show any differences later on, suggesting a major role of environmental factors. From this it was concluded that in SAL and LAL mice, female aggression had a different genetic basis to that of male aggression (van Oortmerssen and Bakker 1981). However, when treated with testosterone in adulthood, SAL females show greater aggression than LAL females, with a major role played by a difference in the conversion of testosterone to estradiol by the enzyme aromatase (Compaan et al., 1993b). In the T lines, low aggressiveness is also associated with high sexuality (Lagerspetz and Hautojarvi 1967; Hautojarvi and Lagerspetz 1968; Lagerspetz and Lagerspetz 1971; Lagerspetz and Lagerspetz 1975), but contradictory results have been found (Sandnabba, 1993a; Sandnabba and Korpela 1994), suggesting a more complex relationship between sex and aggression. In the SAL/LAL model, several studies have investigated whether selection for aggression reflects a more general dichotomy in coping strategies by investigating behavioural strategies in non-social challenging situations. SAL mice show overall a proactive style, i.e. aimed at mastering the stressors, while LAL mice show a reactive/passive strategy. When exposed to social or non-social escapable stress, SAL males tend to flee whereas LAL males freeze more, with more variability among the LAL group (Benus et al., 1989). When exposed to inescapable nonsocial stress, LAL mice suppress their activity, while SAL mice are hardly affected (Benus et al., 1990a). SAL mice, in general, cope actively with their home environment, as shown by defensive burying of a shock-probe in the home cage (Sluyter et al., 1996a) and nest building (Sluyter et al., 1995b). On the other hand, SAL mice cope worse than LAL in a variable environment, showing more rigid and routine-like behaviour and less flexibility (Benus et al., 1988; Benus et al., 1990b). TA and TNA mice have been tested for coping strategies in terms of 30
GENERAL
INTRODUCTION
learning performances. TA mice perform better in solving a maze (active learning), while TNA mice perform better in a shuttle box (passive learning) (Kvist, 1989; Ewalds-Kvist et al., 1997). NC900 and NC100 have been tested for reactivity to stimuli and immobility/freezing behaviour, as indicative of active and passive strategies, respectively. NC100 mice are highly immobile in social and non-social contests, but no difference in reactivity to stimulation from the experimenter or from an intruder mouse has been documented (Gariepy et al., 1988). Freezing during social contests is more pronounced in NC100 juveniles, and it helps in reducing emotionality and enhancing affiliative behaviour (Bauer and Gariepy 2001). Regarding exploratory behaviour, the results are contradictory, since high motor activity and open-field exploration are associated with high aggressiveness in the TA line and with low aggressiveness in the LAL line (Lagerspetz, 1961; Selander and Kvist 1991; Veenema et al., 2003a). When tested for emotionality, mice from the Groningen and the Turku aggressive lines show less anxious behaviour (or defecation in case of the Turku lines) than their non-aggressive counterparts (Lagerspetz, 1961; Selander and Kvist 1991; Hogg et al., 2000; Nyberg et al., 2003; Veenema et al., 2003a). An early study showed that NC900 were considered similar to NC100 in terms of behavioural reactivity and defecation as measures for emotionality, although their urination response was higher (Gariepy et al., 1988). More recently, NC900 were found to be less explorative in a novel arena and in the light compartment of the light-dark box, and drank less milk in a novel environment compared to NC100 (Hood and Quigley 2008). In general, it seems that NC900 are more emotional than NC100. We can conclude that there are some similarities between the aggressive lines in their proactive strategy and between the non-aggressive lines in their reactive/passive coping style. However, the paradigms used are very different from each other and the heterogeneity of studies does not allow a solid conclusion. Is it possible that the selections generated individuals with different types of aggression? Are mouse personality types related to aggression and violence? Which mouse line is exhibits a form of aggression more similar to human violence? In Chapters 2 and 3 I attempt to answer these questions by subjecting all the six mouse lines to the same behavioural paradigms, and look for similarities and differences. Genetics The three models prove that aggression has a hereditary component. Realized heritability has been calculated at 0.34 for the TA and 0.30 for the SAL line at the 11th generation. Since aggressive behaviour was reported in SAL males but not in 31
CHAPTER 1
females, it was hypothesized that genes on the Y chromosome are associated with aggression in the SAL/LAL lines. Using F1 hybrid lines and congenic lines, it has been shown that the non-pseudo-autosomal region of the Y chromosome is involved in determining aggressiveness in terms of attack latency, although the effect may depend on the combination with the pseudo-autosomic region and the autosomal background (van Oortmerssen and Sluyter 1994; Sluyter et al., 1994b). The Y chromosome does not affect behavioural flexibility, nest-building behaviour or the dopaminergic system in terms of sensitivity to apomorphine (Sluyter et al., 1995a; Sluyter et al., 1996b; Sluyter et al., 1997). An effect of the non-pseudoautosomal region of the Y chromosome has been found in the defensive burying behaviour, but the effect is somehow masked by the autosomal background (Sluyter et al., 1999). Although correlated with aggression, the nonpseudoautosomal region of the Y chromosome is not correlated with adult testosterone levels and anatomical differences in the hippocampus (Van Oortmerssen et al., 1992; Hensbroek et al., 1995). In conclusion, genes on the Y chromosome do play a role in the determination of the aggressive behavioural phenotype of male SAL mice, although the effect is not associated with other components of the SAL proactive coping style, nor with some of their characteristic neuroendocrine and neurochemical features. Beside these classic genetic studies, a molecular approach has been used to investigate the genetic characteristics that underlie the aggressive phenotype of SAL mice and differentiate it from LAL mice. Since the project was part of a major research line on depression, the brain region of choice was the hippocampus. Both genome-expression profiles generated by Serial Analysis of Gene Expression and GeneChip analysis have identified a general down-regulation in SAL mice, compared to LAL, of transcripts related to the cytoskeleton, members of a specific calcium/clamodulin signal transduction cascade (e.g. ERK2, raf-related oncogene), and several MAPK related genes involved in learning and memory (Feldker et al., 2003a; Feldker et al., 2003b). Interestingly, the only up-regulated transcript identified is gas5 (growth-arrest-specific-5), a gene coding for small nuclear RNAs (Feldker et al., 2003b). In conclusion, genes present on the Y chromosome play a role in determining individual differences in aggressiveness. However, the genetic basis of the more general suite of behavioural and physiological traits, in which the aggressive behaviour is embedded, seems to lie in the genes related to structural changes in brain structures, as shown in the hippocampus. Phenotypic plasticity Environmental manipulations carried out in mice from the SAL/LAL, TA/TNA and NC900/NC100 lines are summarized in Table 1.3. 32
Reference
Period
Exposure to aggression in a glass container Exposure to aggression through a wire mesh Encounter with more aggressive and less aggressive individuals Housing with male siblings or isolation Ethanol injection to TNA and to inhibited TA Repeated defeat Only visual, olfactory and auditory stimuli of the defeat After repeated defeat, animals in isolation or in sensory contact with the residents Isolation rearing or female-paired rearing Intruder-resident, neutral cage and resident-intruder female
Post-weaning
Pre-weaning Post-weaning
Post-weaning
Post-weaning
Post-weaning
Cairns et al., 1983
Post-weaning
Isolation or group-rearing
Cross-fostering Cross-fostering + endotoxin exposure
Testosterone injection Isolation-rearing Cross-fostering
Pre-weaning Post-weaning Pre-weaning
Post-weaning
Cross-fostering
Embryo transfer Cross-fostering
Experiment
Pre-weaning
NC900/ Cairns et al., 1983 NC100 Hood and Cairns, 1989 Pre-weaning Granger et al., 2001 Pre-weaning
Nyberg et al., 2004
Benus and Rondings, 1990 Compaan et al., 1992 Van Oortmerssen and TA/TNA Lagerspetz and Wuorinen, 1965 Sandnabba et al., 1993 Lagerspetz, 1961, 1964 Lagerspetz and Lagerspetz, 1971 Lagerspetz and Ekqvist, 1978 Lagerspetz and Sandnabba, 1982
SAL/LAL Van Oortmerssen et al., 1985 Sluyter et al., 1996 Prenatal Sluyter et al., 1995 Pre-weaning
Lines
No effect Endotoxin diminished aggression in NC900, no effect of cross-fostering Elicited aggression in isolated mice
Isolation elicits aggression also against a female Previous experience affects aggressive behaviour
Only after isolation aggression levels restored
Inhibited aggression No inhibition of aggression
Alcohol does not change aggression
Isolated mice showed higher aggressiveness
No effect Enhanced aggression Temporary change
No effect No effect on attack latency Effect on the evolution of attack latency No influence on pup growth of the different mother’s behaviour Reduction in aggression in LAL No effect Marginal effect, reduced aggression in both lines
Effect on aggression
Table 1.3 Overview of the environmental effects on the aggressive behaviour in the three selection programs.
GENERAL INTRODUCTION
33
CHAPTER 1
It is known that prenatal events can affect the behavioural phenotype in adulthood. Unfortunately, only in the SAL/LAL mice has this possibility been explored. Experiments in which embryos of SAL, LAL and their reciprocal F1’s were transferred to NMRI females have shown that the genotype, but not the prenatal environment, influences intermale aggression (Van Oortmerssen et al., 1992; Sluyter et al., 1994b; Sluyter et al., 1996c). Postnatal environmental stimuli before and after weaning can affect the adult behavioural phenotype. Preweaning experiments have been performed in all the selection lines using cross-fostering and other manipulations. In SAL and LAL parental lines, no effect of cross-fostering was observed on the attack latency (van Oortmerssen et al., 1985), whereas in their reciprocal F1’s only the evolution of attack latency over three days is affected (Sluyter et al., 1995c). In general, the genotype is the main determinant of the behavioural difference between SAL and LAL, even though SAL and LAL mothers show significantly different behaviour toward the pups (Benus and Rondigs 1996). Similarly, differences in the maternal behaviour of TA and TNA mothers only marginally affects the aggressive behaviour of TA and TNA males in adulthood, without masking the genetic factor (Lagerspetz and Wuorinen 1965). Cross-fostering does not have any effect on the aggressiveness of NC900 and NC100 and their replicate lines I900 and I100 (Cairns et al., 1983; Hood and Cairns 1989). Other preweaning manipulations have been carried out in the Dutch and North Carolina mice. Artificially enhanced testosterone on the day of birth reduces aggression in LAL males, but does not affect aggression levels in SAL males (Compaan et al., 1992), suggesting a higher neonatal sensitivity to testosterone in the more aggressive line. Exposure to E. coli endotoxin at a young age diminishes aggression in NC900 aggressive mice, and the effect is associated with enhanced HPA-axis activity in adulthood (Granger et al., 1996; Granger and Hood 1997; Granger et al., 2001). This has been explained as a higher immune reactivity and sensitivity to early stressors in the aggressive line, presumably through an effect on the mother’s maternal behaviour (Hood et al., 2003). Handling for a period of 3 weeks increases aggression in NC900 mice and the change is not directly associated with changes in plasma corticosterone and dopamine1 receptors (Gariepy et al., 2002). As with preweaning experience, postweaning manipulations have been performed in the aggressive and non-aggressive mouse lines. Isolation- versus group-rearing is a major focus for all the three mouse models for aggressiveness, since isolation-induced aggression is a widely used model for aggression research. Van Oortmerssen and Bakker’s (1981) selection of SAL aggressive mice was performed on animals reared with siblings until sexual maturation. After weaning, SAL and LAL mice were paired-housed with a female throughout the whole experiment. Social isolation at weaning or 1 month before testing did not 34
GENERAL
INTRODUCTION
have any effect on the aggressive behaviour of the SAL mice (van Oortmerssen and Bakker 1981). In contrast, TA and TNA mice were selected using an isolation period before the test for aggressiveness. When reared with brothers, neither TA nor TNA males showed any aggressiveness at all. The line difference appeared again after further isolation, and the aggressiveness of TA mice increased proportionally to the length of the isolation period (Lagerspetz and Lagerspetz 1971). Despite the high dependence on genotype and early life events, TA and TNA behavioural phenotypes were demonstrated to be highly influenced by isolationrearing conditions. When socially housed, experience of victory and defeat play a role, at least temporarily, in determining the aggression level of TA and TNA mice (Lagerspetz, 1961). TA mice were found to be particularly sensitive to the type of experience. Exposure to aggression in youth results in increased adult aggression, to a greater extent in TA than TNA mice (Sandnabba, 1993b). TA mice become non-aggressive after repeated defeats, whereas ethanol injection does not increase aggressive behaviour in TNA mice (Lagerspetz and Ekqvist 1978). Continuous physical contact with aggressive mice, rather than mere sensory contact, reduces the aggressiveness of TA mice, especially when the physical contact precedes the sensory contact period rather than vice versa (Lagerspetz and Sandnabba 1982). A more recent study showed that TA mice are always more aggressive than TNA in different social tests and rearing conditions, suggesting a strong genetic component. However, previous male-male interactions and housing with a female instead of group- or isolation-rearing interacts with genotype in a complex mechanism influencing the aggressive behaviour of the two lines (Nyberg et al., 2004). Group-rearing reduces the difference in aggression duration and frequency between NC900 and NC100 mice, compared to when the animals are reared in isolation (Cairns et al., 1983; Hood and Cairns 1989). When the rearing conditions are kept constant, repeated testing attenuates the difference in latency to attack between the two lines (Cairns et al., 1983). In general, despite the heterogeneity of paradigms and conditions, it can be concluded that animals reared in groups adjust their behavioural phenotype probably because of their previous experiences of victory and/or defeat. However, a strong genetic component plays a stable role in determining an aggressive phenotype in all three genetic selection lines. How does repeated social experience influence the aggression of these mouse lines? Can it contribute to the escalation to violence? I examine these issues in Chapter 3. Physiological and neurochemical correlates Bidirectional artificial selection for aggression results in the selection for alternative behavioural phenotypes, which reveal also distinct physiological and neurobiological mechanisms. A summary of the data on these aspects is in Table 1.4. 35
36
Sex hormones Serotonergic
Compaan et al., 1994 Korte et al., 1996
Serotonergic HPA axis
HPA axis
HPA axis and serotonergic
Serotonergic
Van Riel et al., 2002
Veenema et al., 2003
Veenema et al., 2003
Feldker et al., 2003
Van der Vegt et al., 2001 Serotonergic
Sexl hormones
Dopaminergic Serotonergic
SAL/LAL Benus et al., 1991 Olivier et al., 1990
Compaan et al., 1993
System studied
Reference
Lines
5-HT1A Northern blot analysis on hippocampus
Body weight and organ weight Corticosterone and ACTH levels, GR and MR mRNA in hippocampus and CRH mRNA in PVN (hypothalamus) after forced swimming test After acute and chronic social stress: Body weight and organ weight Corticosterone and ACTH plasma levels MR, GR and CRH in situ hybridization 5-HT1A receptor binding
5-HT1A in vivo functionality (alnespirone-induced hypothermia) Electrophysiological recordings after serotonin stimulation of hippocampal slices; 5-HT1A, MR, GR in situ hybridization; Plasma corticosterone levels
In vitro brain aromatase activity 5-HT1A in situ hybridization and ligand binding
Apomorphine stereotyped behaviour Catecholamine levels (HPLC) in the whole brain Testosterone in adult and young; Vasopressin immunoreactivity
Experiment
SAL more T in adulthood but less neonatally than LAL; SAL less VP density in LS and BNST than LAL POA aromatase activity lower in SAL than in LAL SAL more 5-HT1A mRNA in hippocampus and more ligand binding in hippocampus, lateral septum and frontal cortex SAL higher hypothermia than LAL (higher sensitivity of the receptor) In SAL higher hyperpolarization of CA1 cells after serotonin; SAL less corticosterone than LAL after novelty stress; SAL more 5HT1A mRNA in CA1; SAL less MR mRNA in DG SAL smaller thymus and spleen than LAL; SAL higher corticosterone (light phase) and ACTH levels (light and dark phase); LAL more MR mRNA production in CA2 after stress and CRH in PVN Prolonged body weight in LAL but not in SALdecrease after social stress; Higher corticosterone and ACTH stress response in LAL; Higher 5-HT1A binding in SAL hippocampus than in LAL SAL more mRNA than LAL
SAL are more sensitive to apomorphine than LAL SAL less serotonin than LAL
Results
Table 1.4 Overview of the neurochemical, endocrinological and immunological characteristics of the aggressive and non-aggressive lines.
CHAPTER 1
GABA-ergic
Immune Tumor susceptibility
Petitto et al., 1993
Dopaminergic
NC900/ Lewis et al., 1988 NC100 DeVaud et al., 1989
Weerts et al., 1992
Glutamatergic
Vekovischeva et al., 2007
Serotonergic Noradrenergic Testis and seminal vesicles
Neuronal activation
Haller et al., 2005
TA/TNA Lagerspetz et al., 1968
Serotonergic
Veenema et al., 2005
Reduction of aggression (biting), in particular the non-competitive one reduces boxing
Serotonin: TA
TNA in the brain stem Testis: TA>TNA Adrenalin in adrenals: TA>TNA
5-HT1A mRNA more in SAL than in LAL in CA1; 5-HT1A ligand binding more in SAL than in LAL in CA1 and DG; Less 5-HT in SAL than in LAL in brain stem; Lower 5-HIAA/5-HT in SAL than in LAL in striatum; Effects on behaviour of 5-HT1A agonists in SAL and in LAL In SAL pattern of activation different than LAL and different from territorial aggression models; activation of central amygdale and ventrolateral periaqueductal grey, similar to models of violence
HPLC nucleus accumbens and caudate ; NC100 lower dopamine and metabolites than nucleus NC900 in both regions; D1 and D2 receptors binding autoradiography Higher D1 and D2 density in NC100 in nucleus accumbens and caudate nucleus. Benzodiazepine effect on behaviour in a Benzodiazepine at high doses reduced aggressive dyadic interaction behaviour in NC900 and motor behaviour in Benzodiazepine binding in vivo NC100; GABA-dependent chloride uptake assay NC900 less benzodiazepine binding in cortex, hypothalamus and hippocampus; no difference in pons and medulla; NC900 less GABA-dependent chloride uptake Tumor induction NC100 more tumor than NC900 NK cell function NK cell function lower in NC100 than in NC900 Serum corticosterone assay Trend higher serum corticosterone level in NC100
Chemical detection of serotonin and noradrenaline in the forebrain and in the brain stem; adrenals, testis and seminal vesicles weight; adrenaline content in the adrenals. Effects of glutamatergic competitive and non-competitive antagonists on aggression
c-fos after aggressive encounter
HPLC for serotonin and 5-HIAA 5-HT1A in situ hybridization and autoradiography ligand binding Behavioural effects of 5-HT1A agonists in forced-swim test
GENERAL INTRODUCTION
37
38
Lines
Dopaminergic
Dopaminergic
Immune system and HPA axis Immune system and HPA axis
Dopaminergic
Immune system
Immune system and maternal behaviour Acute tolerance to ethanol
Lewis et al., 1994
Gariépy et al., 1995
Granger et al., 1996
Gariépy et al., 1998
Petitto et al., 1999
Granger et al., 2001
Reed et al., 2001
Granger et al., 1997
System studied
Immune
Reference
Petitto et al., 1993
Table 1.4 Continued. Experiment
Repeated intragastric ethanol injection and progressive recovery from motor impairment on isolated and group-reared mice
Dihydrexidine (full D1-agonist) effects in social interaction test and D1 binding in striatum in continuously isolated and isolated-grouped mice; NK cell activity in cells from mice with different post-weaning social experience Endotoxin to pups; Cross-fostering
Dihydrexidine (full D1-agonist) effects in social interaction test in group- and isolation-reared mice Perinatal endotoxin exposure effects on HPA axis and social behaviour Endotoxin effects on physiological parameters, social behaviour and HPA axis
In cells from isolated and group-reared mice: -Mitogen assay -Interleukin-2 assay -γ-interferon assay -NK cell activity Dihydrexidine (full D1- agonist) effects in social interaction test; D1-induced behaviour and effects of D1and D2-antagonist pretreatment
Results
NC100 less NK activity than NC900; No effect of the post-weaning experience Endotoxin decreased aggressive behaviour and social reactivity in NC900; no interaction effect with the fostering condition Genetic-environment interaction effect on females, but not on males.
Endotoxin affected social behaviour in both lines and diminished hypothalamic CRF in NC900 Lower threshold of temperature, body weight and corticosterone in NC900; decreased social reactivity and aggressiveness in NC900 and increased social reactivity in NC100 DHX effects on behaviour and D1 binding higher in continuously isolated mice
D1 agonist reduces aggression in NC900 and non-agonistic approach in NC100; no effect on freezing behaviour; D1-antagonist, not D2-antagonist, antagonized D1-agonist effects on social interaction. D1 agonist enhances social reactivity especially in the isolated; NC900 more reactive to social stimuli
NC100 less mitogen stimulation effect, IL-2 production, γ-interferon and NK activity than NC900; no housing by line interaction effect
CHAPTER 1
GENERAL
INTRODUCTION
In accordance with other models of proactive/reactive coping strategies, SAL mice show less corticosterone reactivity to corticotrophin-releasing hormone (CRH) challenge than LAL, as well as lower plasma corticosterone baseline levels, but more fluctuation in corticosterone levels across light-dark cycle (Korte et al., 1996; Veenema et al., 2003b). Furthermore, the ACTH response to stress is higher in SAL mice compared to LAL, while corticosterone response is much less pronounced and shorter, suggesting a reduced adrenocortical sensitivity in the former line (Veenema et al., 2003b; Veenema et al., 2004). The higher and longer-lasting stress response from the LAL mice is associated with higher mineralocorticoid receptor (MR) mRNA expression in the hippocampus and higher CRH in the hypothalamic paraventricular nucleus (Veenema et al., 2003b; Veenema et al., 2004). In LAL mice, more than SAL, the HPA hyper-responsivity after a chronic social stress paradigm results in a more pronounced body weight loss and an increase in hippocampal MR mRNA, with correlated increase in passive behavioural responses in anxiety tests (Veenema et al., 2003a). In NC mice, baseline and endotoxin-stimulated corticosterone activation during a dyadic interaction are higher in the aggressive than the non-aggressive line, while hypothalamic CRF does not differ (Granger et al., 1996; Granger and Hood 1997). These results indicate that proneness to aggression is associated with different HPA-axis functioning, although to what extent is difficult to extrapolate. It may be that this system is more generally related to the proactive coping style of SAL mice and is not well represented in the NC900 line, due to the lack of a distinct proactive phenotype. While stress research has described SAL mice as more resistant to stress than LAL, immunological findings have shown that NC900 mice are more resistant to the development of immune diseases and tumours (Petitto et al., 1993; Petitto et al., 1994; Petitto et al., 1999). The discrepancy in immune functioning could not be ascribed to a baseline difference in plasma corticosterone levels, but indications of a different corticosterone response to infections reveals an important role of the HPA axis in differentiating these lines (Granger et al., 1996; Granger and Hood 1997; Granger et al., 2001). A more consistent finding in the stress physiology of the three mouse models of aggression concerns the sympathetic-adrenomedullary system. It seems that aggression is related to high sympathetic reactivity to stress, as revealed by high adrenaline content in the adrenals of TA mice (Lagerspetz et al., 1968). In this thesis, I investigate the peripheral physiology of the six mouse lines for the first time in the same experiments (see Chapters 4 and 5). Since testosterone is generally known to correlate positively with aggressiveness, researchers have investigated the neuroendocrine circuit in which testosterone is involved. An early study reported that the selection for SAL mice corre39
CHAPTER 1
lated with high plasma testosterone levels due to high gonadal production (van Oortmerssen et al., 1987). The secretory capacity of Leydig cells is highest in SAL mice during pre-puberty and in adulthood, whereas in LAL it is highest neonatally (de Ruiter et al., 1993). Prenatal testosterone exposure is higher in SAL than in LAL, with consequent sensitization of the adult SAL male to testosterone, leading to an increased capacity to display aggressive behaviour (Compaan et al., 1992; de Ruiter et al., 1993). High circulating testosterone levels increase brain aromatase activity from the day of birth, while the two phenomena are not strictly associated prenatally (Compaan et al., 1994a; Compaan et al., 1994b). In the SAL/LAL mice, the vasopressin content of the lateral septum is testosteronedependent, with SAL having the lowest fibre number and vasopressin content (Compaan et al., 1993a). In contrast, testosterone does not seem to be associated with male aggression in the T lines (Sandnabba et al., 1994), although testes of TA mice weigh more than those of TNA mice (Lagerspetz et al., 1968) and not much more research has been done on this topic in these lines. Testosterone is also not associated with male aggression in the NC-lines (Gariepy JL et al., 1996). Typical neurochemical systems involved in the control of aggression are serotonin, dopamine, γ-aminobutiric acid (GABA) and noradrenaline. The tissue level of forebrain serotonin and its metabolite is lower in the aggressive lines than the non-aggressive ones (Lagerspetz et al., 1968; Olivier et al., 1990; Veenema et al., 2005a). The serotonin-deficiency in SAL mice is associated with enhanced sensitivity of the presynaptic and postsynaptic 5-HT1A receptor and its expression in cortico-limbic structures. In SAL mice, only the postsynaptic 5-HT1A receptor is more sensitive than that in LAL mice (Korte et al., 1996; van der Vegt et al., 2001; Feldker et al., 2003a; Veenema et al., 2005a; Veenema et al., 2005b). SAL and LAL show different behavioural sensitivity to 5-HT1A agonists regarding aggressive behaviour, but also different characteristics of their alternative coping strategies (Veenema et al., 2005a). Because of the differential HPA-axis and serotonergic stimulation converging on the hippocampus, and consequent hippocampal remodelling, SAL and LAL may develop different behavioural phenotypes in order to cope successfully with environmental stimuli (Van Riel et al., 2002; Veenema et al., 2004). Is this mechanism involved in trait aggression or more generally in the coping behavioural response? Chapters 3 and 5 compare the prefrontal serotonin levels and the 5-HT1A neurotransmission of the six mouse lines and give some evidence related to this issue. Following the results of the previous chapters, in Chapters 6, 7 and 8 I choose to focus on the serotonergic system of violence, rather than more generally aggression. Therefore I restricted my focus to the mouse lines that were more suitable for answering this question. Pharmacological and psychiatric studies show an involvement of dopamine in the initiation and execution of aggressive behaviour (Miczek et al., 2002). SAL 40
GENERAL
INTRODUCTION
mice show higher sensitivity than LAL to the dopaminergic agonist apomorphine, in terms of the increase in stereotyped behaviour caused by the drug (Benus et al., 1991a). In the NC lines, both agonistic and non-agonistic social behaviours are mediated by D1 receptors, with no difference between the two lines, but more so in isolated than group-reared animals (Lewis et al., 1994; Gariepy et al., 1998). It seems that a differential dopaminergic activity reflects a different coping style, and differential proneness to social behaviours. Preclinical data suggest a controversial function of GABA and the GABAA modulator benzodiazepine in the control of aggressive behaviour. Only one study has been performed on the GABAergic system in the selection lines. Benzodiazepine reduces motor behaviour in NC100 mice, while it produces a shift in the NC900 mice from aggressive behaviour to more social behaviour. The differential effect on behaviour is reflected by a lower benzodiazepine binding in corticolimbic regions and reduced cortical GABA uptake in the aggressive line, compared to the non-aggressive and medium-aggressive lines (Weerts et al., 1992). These results suggest a difference in the sensitivity of the GABAergic system in individuals with opposite personalities. Glutamatergic modulation of aggression has only recently been investigated in the selected mouse lines. Selective AMPA-type glutamate receptor antagonists acutely reduce offensive aggression of TA mice, with the non-competitive GYKI 52466 suppressing all aggressive behaviours. The competitive NBQX also increases social behaviour and threat in TNA mice, suggesting that its effect may depend on differential sensitivity to the drug (Vekovischeva et al., 2007a). Several neuroanatomical localizations have been proposed for the regulation of aggressive behaviour. In SAL mice, the hippocampal intra- and infrapyramidal mossy fibre (IIPMF) distribution differs from that in LAL mice (Sluyter et al., 1994a). However, these differences do not completely correlate with the aggressiveness displayed and the involvement of the Y chromosome (see Genetic studies). The neuronal activation pattern in SAL males during an aggressive encounter, measured with c-fos immunostaining, shows strong activation of amygdala and periaqueductal grey matter (Haller et al., 2006). Since this differs from the pattern of activation in LAL males and shows similarities with the pattern in violent humans, it is more related to the violent temperament of SAL mice than to mouse-typical territorial aggressiveness.
OUTLINE OF THIS THESIS The main aim of this thesis is to understand the physiological correlates of aggression and the development of violence. Under this theme, I chose to study the 41
CHAPTER 1
three pairs of mouse lines genetically selected for high and low aggressiveness. Due to the heterogeneity of the data generated in these mouse lines and the consequent difficulty to generalize them through comparative literature analysis, I investigated them together in the same experiments using the same approach. I studied their behavioural phenotype, their peripheral stress physiology and their central serotonergic system with emphasis on the serotonin-1A receptor, trying to answer the following research questions: – are there different mouse aggressive characteristics that could be related to different groups of aggressive individuals in humans? – are violent mouse types physiologically similar to violent humans? – in which specific type of mouse aggression is serotonin involved, if it is involved at all? – is the serotonin-1A receptor involved in aggression and violence? In Chapter 2 I investigate the behavioural phenotype of the three pairs of mouse selection lines with respect to non-social proactive behaviours and activity/exploration. In Chapter 3 I investigate the possibility of identifying a violent behavioural phenotype in the mouse lines selected for high and low aggressiveness. To assess mouse violence I consider aggression against females and lack of sensitivity to the cues of the mouse opponent. In Chapter 4 I explore baseline and stress-related autonomic correlates of aggression and violence. In particular, I ask whether the association between low autonomic arousal and violence can be replicated in the different mouse lines, in view of the idea that they represent different types of aggressive personality. In Chapter 5 I investigate whether differences in the serotonergic system, and particularly in the functionality of the serotonin-1A receptor, correlate with the high and low aggressiveness of the mouse lines. In Chapter 6 I test the hypothesis that the development of violence is associated with a change in the functionality of the serotonin-1A receptor. In Chapter 7 I explore the causal relationship between serotonin and violent aggression, through manipulation of dietary levels of the serotonin precursor and measurement of the behavioural effects. In Chapter 8 I investigate the idea that the difference in the functionality of the serotonin-1A receptor between violent and docile mice is related to differences in the receptor ultracellular distribution.
42
GENERAL
INTRODUCTION
43
2
CHAPTER
Is there co-selection for aggressiveness, coping strategy and emotionality in mice?
Doretta Caramaschi, Sietse F. de Boer and Jaap M. Koolhaas
ABSTRACT Personality, as a suite of correlated behavioural traits, has a genetic basis. Therefore, it is likely that selection for one trait leads to co-selection for other traits. We tested the association between aggressiveness and emotionality by measuring proactive/reactive coping with a nonsocial environmental challenge and exploration of a novel environment, in lines of mice (Mus musculus) selected for high and low aggression, namely SAL, LAL, TA, TNA, NC900, and NC100. We expected highly aggressive lines to show high levels of boldness, active coping and low levels of exploration, since these individuals would be better adapted to becoming dominants through fighting in their deme rather than to dispersal to found new colonies, and vice versa for the low-aggressive lines. The results show that, overall, high aggressiveness was related to greater mobility in the Novel Object and Shock Prod tests, while no association was found between aggressiveness and coping in the Forced-Swim test. Exploration levels in an Open Field were associated with low aggressiveness in the SAL-LAL lines, and with high aggressiveness in the TA, TNA, NC900 and NC100 lines. In conclusion, selection for high aggressiveness led to co-selection for boldness, while selection for low aggressiveness led to co-selection for fearfulness and, depending on the original strain, for high and low exploration levels.
CHAPTER 2
INTRODUCTION Mouse populations, like those of other animals, show individual differences in behavioural and physiological traits (Benus et al., 1991b; Koolhaas et al., 2007). These traits have a genetic basis and are apparently the product of natural selection. However, there are differing views on the adaptive nature of behavioural traits. Some behavioural traits may be co-selected and result in adaptive ‘packages’, consistent across contexts and stable throughout the life of an individual (Sih et al., 2004a; Wolf et al., 2007). Alternatively, selection of adaptive traits might result in the co-selection of less adaptive traits as by-products of genetic correlations that act as constraints (Arnqvist and Henriksson 1997). Furthermore, constraints might be found already at the physiological level (Koolhaas et al., 1999; Stamps, 2007). This paper will further analyze the general pattern of behavioural traits that are co-selected with aggression, using a comparison of three mouse strains genetically selected for high and low levels of aggressive behaviour. A growing body of evidence has pointed to the existence of behavioural ‘packages’, or personalities, that constitute the basis for individual differences. Although there is growing consensus on the existence of similar behavioural phenotypes across species and across populations, some behavioural traits might vary along distinct axes. For example, rodent personalities have been described in terms of a coping axis and an emotionality axis (Steimer and Driscoll 2003; Koolhaas et al., 2007). Moreover, suites of correlated traits have been described using different terms with very similar meanings. The term ‘behavioural syndromes’ has been proposed to represent collections of stable correlated traits (Sih et al., 2004b), while ‘coping styles’ gives the functional adaptive connotation of being able to cope with challenging situations (Koolhaas et al., 1999). ‘Temperament’ is a more general term that takes into account the repeatability of objectively measurable behaviours and their early onset in the life of an individual (Reale et al., 2007). Beside these more technical terms, ‘personality’ is the preferred one in the human literature, where it was first used. Laboratory studies in biomedical research have often neglected the existence of personalities in their mouse populations. However, recent studies show that there are individual differences in the vulnerability to diseases and in the response to pharmacological treatments, and that in several cases these differences are related to different personalities (Mehta and Gosling 2008). Moreover, in the study of human aggression, it is worth mentioning that aggressiveness may be typical of certain personalities (Ramirez and Andreu 2006) and differential neurobiological mechanisms may be behind these relationships (Siever, 2008). In nonhuman populations, aggressiveness is often correlated with proactive coping, boldness, risk-taking behaviours, fearlessness and low levels of activity/explo46
AGGRESSIVENESS,
COPING STRATEGY AND EMOTIONALITY
ration (Koolhaas et al., 1999; Groothuis and Carere 2005; Dingemanse et al., 2007). In human populations, aggressiveness is expressed either with an impulsive/affective connotation or within a social-cognitive/instrumental dimension. In the five-factor model of personalities (Digman, 1990), aggressiveness is particularly associated with high neuroticism/low agreeableness. Aggressive behaviour is expressed in a cold-blooded manner in people with high trait aggression/irritability, while it is expressed under provocation in people with high trait anger, Type A personality, rumination traits, emotional susceptibility, narcissism and impulsivity (Bettencourt et al., 2006). In conclusion, studying correlations between behavioural traits in animal models, and therefore aiming at a better behavioural characterization of individual differences, may have important translational implications in human research. Pairs of genetic mouse lines artificially selected for high and low aggressiveness from different original populations give a unique opportunity for investigating the functional association between aggression and other behavioural traits. The highly aggressive SAL (Short Attack Latency) and the low-aggressive LAL (Long Attack Latency) lines were selected on the basis of attack latency against a conspecific male intruder from a colony of wild-trapped house mice in the area of Groningen, the Netherlands (van Oortmerssen and Bakker 1981). The TA (Turku Aggressive) and TNA (Turku Non-Aggressive) lines were obtained on the basis of ratings of aggressiveness against a male conspecific from outbred Swiss Webster mice at the University of Turku, Finland (Lagerspetz, 1961). The NC900 and NC100 high- and low-aggression lines were selected based on aggressiveness ratings in male-male competition from ICR mice at the University of North Carolina, USA (Cairns et al., 1983). The SAL-LAL lines have been characterized extensively in their proactive/reactive alternative coping styles, with links to their neurobiological and neuroendocrine profiles (Koolhaas et al., 1999; Veenema et al., 2005a). In the other lines, a behavioural characterization based on alternative active/passive coping, without attempting to describe alternative personalities, seems to point in the same direction observed in the SAL-LAL lines (Gariepy et al., 1988; Kvist, 1989; Ewalds-Kvist et al., 1997; Bauer and Gariepy 2001; Vekovischeva et al., 2007b). Exploratory behaviour and motor activity showed a negative association with aggressiveness in the SAL-LAL lines (Veenema et al., 2003a) and a positive association in the TA-TNA lines and NC-lines (Selander and Kvist 1991; Hood and Quigley 2008). This study was designed to unravel the association between trait aggression, boldness, fearfulness, proactive/reactive coping and exploration in male mice. We defined trait aggression as the propensity of a male to attack an age-matched male intruder in his own territory. Boldness was the propensity to approach and explore a novel object. Proactive coping was defined as the capacity to avoid a 47
CHAPTER 2
shock from an electrified probe by covering it with bedding material, as reviewed in de Boer and Koolhaas (2003), as opposed to reactive/passive coping, which is characterized mainly by immobility. Across all these tests, immobility was taken as an index of fearfulness. Emotionality was defined as the exploration levels or inactivity in a large novel arena, which was intended to mimic the situation of a novel and potentially risky environment to explore in order to found a new colony. We aimed to replicate findings previously obtained in the SAL-LAL model of proactive/reactive coping and extend those to the TA-TNA and NC900-NC100 models of aggression. Based on previous research and on the idea that docile mice are better suited to leave their original colonies and disperse, we expected the highly aggressive lines SAL, TA, and NC900 to be bolder and proactive, and the low-aggressive lines LAL, TNA, and NC100 to be more cautious.
METHODS Animals Male mice (n=30) from the SAL, LAL, TA, TNA, NC900 and NC100 lines were used in this experiment. The mice were bred in our laboratory at the Biological Center, University of Groningen, Haren, the Netherlands, and weaned at 21 days of age. At around 40 days, each male mouse was housed with a sister, to avoid social isolation and intrasexual aggression, in Makrolon® Type II cages furnished with food shavings as bedding material, shredded paper EnviroDry® (BMI, Helmond, Netherlands) for nesting and a cardboard tube as cage enrichment. The mice had ad libitum access to food (AMII, ABdiets, Worden, The Netherlands) and water and were kept under standard laboratory conditions, at a temperature of 22 ± 2 °C and on a 12:12 light-dark cycle (lights on at 21.00). The tests started when the mice were 3 months old and consisted of the following behavioural tests: Attack Latency/Resident-Intruder, Novel Object, Shock Prod, Forced-Swim, Open Field (see below for detailed explanation of each test). The tests were conducted in the dark phase in the aforementioned order. At the end of the experimental session the mice were euthanized with a mixture of CO2/O2. All the procedures were carried out at the Biological Center, University of Groningen, Haren, the Netherlands, under approval of the Institutional Animal Care Committee of the University of Groningen (licence D4540D) and in compliance with the Dutch law on animal experimentation and the European Communities Council Directive of 24 November 1986 (86/609/EEC). Attack Latency/Resident-Intruder The resident-intruder test was performed according to standard procedures. 48
AGGRESSIVENESS,
COPING STRATEGY AND EMOTIONALITY
Briefly, two days prior to the test, the mice were housed together with their female partners in cages for testing aggression. On the third day, when the mice were habituated to the cage and had marked it as their own territory, the females were removed and an unknown, age-matched, socially naïve male intruder from the A/J strain was introduced in each cage. The time it took for the resident experimental mouse to attack was recorded as Attack Latency. The intruder was then removed and placed back in its home cage, while the female was returned to the testing cage. The test was repeated on the two following days, using the same procedure. Testing was conducted in such a way that each resident mouse always encountered a new opponent in its home cage. On the third day of testing, the intruder was left in the testing cage for 5 minutes to allow the social interaction to develop fully. The whole of this interaction was videotaped for later behavioural analyses. Novel Object and Shock Prod Three days after the last Resident-Intruder interaction, the Novel Object paradigm was performed, followed by the Shock Prod test one day later. In the Novel Object and the Shock Prod tests, each male mouse was confronted with the introduction in their home cage of a prod capable of giving electric shocks. In the Novel Object test the prod was switched off and the mice experienced the presence of an unknown object in their home cage, while in the Shock Prod test they received an electric shock of 0.7 mA intensity every time they touched the prod (de Boer and Koolhaas 2003). The Novel Object test lasted for 5 minutes after the first approach to the prod, while the Shock Prod lasted for 5 minutes after the first shock was received. Both paradigms were given in absence of the females and in fresh sawdust, and were video recorded for later behavioural analyses. Forced Swim Two days after the Shock Prod test, the Forced Swim test was performed. For this test each experimental mouse was introduced to a cylinder (diameter = 14 cm, height = 20 cm) containing tap water maintained at a temperature of 25 °C. The water level reached the height of 15 cm. Each mouse was left for 5 minutes in the water, during which time its behaviour was videotaped for later analyses. After removal from the water, each mouse was wrapped in a towel where it was kept for a few minutes to dry and then placed in its home cage. Open field A week after the Forced Swim test, each animal was tested in an Open Field. Briefly, the mouse was placed in the centre of a large, round arena (120 cm diameter) and was left undisturbed for 30 minutes, during which a video camera 49
CHAPTER 2
connected to a PC recorded its movement trajectory from above. The arena was washed thoroughly with water at the end of each test. Behavioural analyses The behaviours in the Resident-Intruder, Novel Object, Shock Prod and Forced Swim tests were analysed using The Observer 5.0 software (Noldus Information Technology bv, Wageningen, the Netherlands). From the Resident-Intruder videos, the following behaviours were scored as continuous behavioural states: Attack, Threat, Chase, Social exploration, Non-social exploration, Immobility and Body care. From the Novel Object and Shock Prod videos, Prod burying, Prod sniffing, Cage explore, Body care, Immobility and Tail rattle were scored as behavioural states. From the Forced Swim videos, Swimming, Climbing and Immobility were scored as behavioural states. The behaviour of the mice in the Open Field was analysed using the Ethovision 3 tracking system (Noldus Information Technology bv, Wageningen, the Netherlands). The distance moved in the whole arena during each 5-minute interval was extracted as a measure of emotionality levels. Statistical analyses Statistical tests were performed using SPSS 14.0 for Windows (SPSS Inc., Chicago, Illinois, USA). Attack latencies were analysed statistically with ANOVA for repeated measurements, including ‘day’ (3 levels: day1, day2, day3) as a within-subject factor and ‘type’ (2 levels: aggressive line and non-aggressive line) and ‘selection’ (3 levels: Groningen, Turku, North Carolina) as between-subjects factors. All the interaction effects were included in the analysis. The duration of the behaviours scored in each observation of the Resident-Intruder, Novel Object, Shock Prod and Forced Swim tests was considered for statistical analyses. Group means for the duration of each behaviour were analysed using a two-factor general linear model with ‘type’ (2 levels: aggressive line and non-aggressive line) and ‘selection’ (3 levels: Groningen, Turku, North Carolina) and a ‘type*selection’ interaction (6 levels: SAL, LAL, TA, TNA, NC900, NC100) as between-subjects effects. Post-hoc analyses were computed using Tukey’s multiple comparisons. The distance moved in the Open Field was analysed using a general linear model for repeated measurements with similar between-subjects effects as in the other analyses and ‘interval’ (6 levels: 0-5, 5-10, 10-15, 15-20, 20-25, 25-30) as a within-subject effect. Post-hoc analyses were performed using Tukey’s multiple comparisons in each interval.
50
AGGRESSIVENESS,
COPING STRATEGY AND EMOTIONALITY
RESULTS Attack latency/Resident-Intruder The propensity to engage in inter-male aggression was tested in the attack latency test. The attack latency (figure 2.1A) decreased significantly across the three days of testing (day effect: F(2,48)=6.65, p=0.003), as expected from previous studies. On average, the mice from the aggressive lines displayed significantly lower attack latencies than low-aggression lines (type effect: F(1,24)=128.61, p<0.001). However, a significant ‘type*selection’ interaction effect (F(1,24)= 3.45, p=0.048) showed that the ‘type’ effect was less pronounced in the Turku lines. The results of the post-hoc analyses are indicated in figure 2.1A.
attack latency (sec)
300
A
b,c c
250 200
b
150 a
100
SAL LAL TA TNA NC900 NC100
a
50 a
0 day1
duration (% of observation)
80
B
day2
day3
resident-intruder
SAL LAL TA TNA NC900 NC100
60
40
20
0
a b a b bb
attack
threat
chgase
social exp.
a b
non-social immobility body care exp.
Figure 2.1 A) Means and standard errors of attack latencies during the attack-latency test. B) Means and standard errors of total duration of behaviours exhibited by the residents during the 5-min Resident-Intruder test. a, b, c are homogeneous subsets after Tukey’s multiple post-hoc comparisons of the main ‘type*selection’ effect.
51
CHAPTER 2
A more detailed analysis of the aggressive interaction is presented in figure 2.1B. The high-aggression lines spent significantly more time in attack (F(1,24)=80.59, p<0.001) , threat (F(1,24)=27.85, p<0.001), chase (F(1,24)= 7.77, p=0.010) and immobility (F(1,24)=10.66, p=0.003) and less time in social exploration (F(1,24)=47.66, p<0.001), non-social exploration (F(1,24)=21.71, p<0.001) and body care (F(1,24)=4.48, p=0.045) than the low-aggression lines. Immobility levels significantly differed also according to the ‘type*selection’ interaction effect (F(2,24)=5.69, p=0.009), since NC900 mice showed greater mobility than the other high-aggression lines, almost significantly compared to SAL (Tukey’s HSD= -8.15, p=0.054) and significantly compared to TA (Tukey’s HSD= –10.07, p=0.017), and similarly low to its own counterpart, NC100, and to the low-aggression lines. Novel Object and Shock Prod The propensity to approach and explore an unfamiliar object was tested in the Novel Object test (figure 2.2A). Immobility levels were significantly lower in the aggressive lines (F(1,24)=6.69, p=0.016). There were no other significant differences across the groups in any other behaviour observed. Interestingly, mice belonging to all lines except for LAL did bury the prod in this test, although no shock was applied. The ability to cope proactively with a noxious object was investigated in the Shock Prod test (figure 2.2B). Immobility was significantly lower in the aggressive lines (F(1,24)=5.57, p=0.027) and in the North Carolina selection lines (F(1,24)=4.63, p=0.020). The proactive behaviour of defensive burying, as well as the other behaviours observed, did not show significant differences across groups. Interestingly, the burying levels in this test were not different from the burying levels in the Novel Object test, as shown using ANOVA for repeated measurements (F(1,24)=3.06, p=0.093, not significant). Again, none of the LAL mice displayed this behaviour. Forced Swim The ability to cope proactively was tested further in the Forced Swim test (figure 2.3). The way the mice responded to the challenge was affected solely by the selection pair they belonged to. ANOVA revealed a highly significant effect of ‘selection’ in the amount of climbing (F(2,23)=7.93, p=0.002) and immobility (F(2,23)=22.08, p<0.001). In particular, the Turku lines showed less climbing (almost significantly vs. Groningen: Tukey’s HSD= –12.79, p=0.066; significantly vs. NC: Tukey’s HSD= –21.20, p=0.002) and more immobility than the other selection lines (Turku vs. Groningen: Tukey’s HSD= 29.38, p<0.001; Turku vs. NC: Tukey’s HSD= 26.90, p<0.001). 52
AGGRESSIVENESS,
duration (% of observation)
80
COPING STRATEGY AND EMOTIONALITY
novel-object
A
SAL LAL TA TNA NC900 NC100
60
* 40
20
0
duration (% of observation)
80
shock-prod
B
60
40
20
0
* immobility
cage explore
prod sniff prod bury body care tail rattle
Figure 2.2 A) Means and standard errors of total duration of behaviours exhibited during the Novel Object test. B) Means and standard errors of total duration of behaviours exhibited during the Shock Prod test.
forced swim
duration (sec)
80
60 #
**
*** ***
SAL LAL TA TNA NC900 NC100
40
20
0
swimming
climbing
immobility
Figure 2.3 Means and standard errors of total duration of behaviours exhibited during the Forced Swim test. Post-hoc test of the ‘selection’ effect, Tukey’s multiple comparisons: # 0.05
53
CHAPTER 2
total distance travelled (cm)
Open field Emotionality levels were investigated in the Open Field test (figure 2.4A and 2.4B). Overall, mice from the Groningen selection lines travelled significantly more than mice from the Turku and North Carolina selection pairs, as revealed by a significant main effect of ‘selection’ on the total distance travelled (F(2,24)=17.96, p<0.001) and subsequent post-hoc tests (Groningen vs. Turku: Tukey’s HSD= 14884, p<0.001; Groningen vs. North Carolina: Tukey’s HSD= 10681, p=0.003). There was also a significant ‘type*selection’ interaction effect (F(2,24)=4.98, p=0.016) that was due to values being higher in the LAL line compared to TA (Tukey’s HSD= 12925, p=0.017), TNA (Tukey’s HSD= 22809, p<0.001), NC900 (Tukey’s HSD= 11322, p=0.046) and NC100 (Tukey’s HSD= 16008, p=0.002), but lower in the TNA line compared to SAL (Tukey’s HSD= –16842, p=0.001), LAL (Tukey’s HSD= -22809, p<0.001)and TA (Tukey’s HSD= –11487, p=0.042). A more detailed analysis of the time course of the exploration
40000
a,b
30000
b
b,c
b,c
20000 c
10000 0
8000
distance travelled (cm)
open field
a
A
SAL
LAL
TA
TNA NC900 NC100
open field
B
6000
4000 SAL LAL TA TNA NC900 NC100
2000
0
0–5
5–10
10–15 15–20 20–25 25–30
Figure 2.4 A) Means and standard errors of total distance travelled during the 30-min Open Field test. a,b,c are homogeneous subsets after Tukey’s multiple comparisons. B) Means and standard errors of distance travelled at 5-min intervals during the Open Field test.
54
AGGRESSIVENESS,
COPING STRATEGY AND EMOTIONALITY
of the open field was conducted on data obtained every 5 minutes, analysed with ANOVA for repeated measurements. Overall, the distance travelled significantly changed across the time span (F(5,120)=8.41, p<0.001), and the change was significantly different according to ‘measurement*type’ (F(5,120)=3.79, p=0.003), ‘measurement*selection’ (F(5,120)=2.065, p=0.032) and ‘measurement*type*selection’ interaction effects (F(5,120)=4.37, p<0.001). There were also significant main ‘selection’ (F(2,24)=17.955, p=0.016) and ‘type*selection’ effects (F(2,24)=4.98, p<0.001), similar to the results from the ANOVA on the total distance travelled. Post-hoc analyses performed as ANOVA for repeated measurements within the Groningen selection pair revealed a significant ‘measurement*type’ effect (F(1,8)=8.17, p=0.021). Further post-hoc tests revealed that SAL mice reduced significantly their exploration levels in the last part of the test (paired t tests relative to t5, uncorrected for multiple comparisons: tt25(4)=5.45, p=0.006; tt25(4)=5.77, p=0.004), while LAL maintained their levels at similarly high values from the start to the end. Repeated-measurements ANOVA performed on the Turku selection lines showed significant main effects of ‘measurement’ (F(1,8)=6.82, p=0.031) and ‘type’ (F(1,8)=13.72, p=0.006), which indicates that there was a significant decrease in the exploration levels overall and that TNA mice travelled significantly smaller distances. Similarly to the Turku lines, significant effects of ‘measurement’ (F(1,8)=61.94, p<0.001) and ‘type’ (F(1,8)=5.52, p=0.047) were found in the North Carolina selection pair, indicating that there was a significant decrease overall in the distance travelled across the measurements and that the low-aggressive NC100 line travelled significantly shorter distances.
DISCUSSION This study shows that genetic selection of mice for high and low aggression does not always lead to the same general behavioural phenotypes. In the SAL-LAL selection, aggression is negatively, but weakly, associated with reactive/passive coping and exploration of a novel environment. In the TA-TNA and NC900-NC100 selections, aggression is associated with low emotionality in a large novel environment. Possible reasons for the different phenotypes obtained through selection for aggression are differences between the parental strains and the selection criteria used. Differences between the parental strains are a very plausible explanation. The SAL-LAL model is somewhat distinct from the other two pairs of lines, which are more similar to each other. Considering that the SAL and LAL lines originated from a wild population, while the Turku and NC lines were selected from albino 55
CHAPTER 2
laboratory strains, it is possible that the latter pairs of lines are more genetically related to each other through a common ancestor. Furthermore, the range of genotypes and phenotypes in the original populations of the Turku and NC lines is probably narrower than that of a wild house mouse population. A similar phenomenon has been shown by comparing Wistar rats with WTG (Wild-Type Groningen) rats (de Boer et al., 2003). The criteria used in the selection might have some relevance too. While SAL-LAL mice were selected based on attack latency alone, i.e. on the readiness to attack a conspecific intruder in the homecage, the Turku lines were selected on the basis of an observer rating, in which the lowest score was given to a very immobile animal and the highest one was given to a fiercely aggressive animal, and the NC lines were selected mainly on the basis of behavioural frequencies. Furthermore, the Turku and NC lines were tested in a neutral arena, and therefore the context in which they displayed aggression was of different ecological relevance compared to the home-cage of SAL-LAL mice. Some results obtained in this study confirm and extend previous research. The higher levels of immobility in the low-aggressive lines in the Novel Object test confirm previous studies in which LAL mice were more affected than SAL by changes in the environment (Benus et al., 1987). Similarly, TNA mice showed more immobility than TA mice in an elevated plus maze, and NC100 exhibited more freezing behaviour in a social interaction (Gariepy et al., 1988). The higher level of immobility behaviour in the LAL males can be considered as an adaptive, reactive way of coping with the challenge, as opposed to a passive way to react to the challenge driven by emotionality and fear. This interpretation is in line with the previous literature (Benus et al., 1989; Benus et al., 1990a; Koolhaas et al., 1999; Veenema et al., 2003a) and is also consistent with the present data on the low levels of defensive burying behaviour in both the Novel Object and Shock Prod tests, and with the high activity levels in the Open Field. In the other pairs of lines, it is more likely that differences in fear play a role, since in our study the TNA and NC100 non-aggressive lines are less active in a novel environment. In the Turku lines, a high emotionality component shown by high limmobility levels in the Forced Swim test might also be involved. In these lines, therefore, an augmented fearfulness seems to be the most plausible explanation. Interestingly, LAL mice did not show more immobility during the Forced Swim test, contrary to previous reports (Veenema et al., 2003b). One explanation for this discrepancy could be the order of the tests performed in the present study and the fact that previous tests might have interfered with the outcome in the Forced Swim test. In support of this reasoning, it should be noted that LAL show a very flexible behavioural phenotype consistent with their capability to adapt to novel situations. The Forced Swim test is quite a strong stressor for mice and therefore pushes the 56
AGGRESSIVENESS,
COPING STRATEGY AND EMOTIONALITY
behavioural response to ceiling levels that might be beyond the functional coping responses, while activating a slightly different catecholaminergic response (Koolhaas et al., 1997). Perhaps in this study the mice were emotionally affected by the previous Shock Prod test and their behavioural response to the swimming test might have been altered by a higher emotional activation that goes beyond the adaptive coping response. The defensive burying behaviour typically expressed in reaction to a Shock Prod is not correlated with the aggression levels in our study. This result does not replicate earlier findings in SAL and LAL mice (Sluyter et al., 1996a; Sluyter et al., 1999) or previous data on the wild-type Groningen rats (de Boer et al., 2003). This may be because, for unknown reasons, the time spent burying in our study is much lower on average than that in previous studies. Nevertheless, our data indicate a similar difference, since the LAL line was the only one that did not show burying behaviour. Moreover, the strategy of LAL mice was clearly different from SAL in this test, since they were more immobile, not burying at all, and doing considerable exploring of the cage. The latter was clearly an attempt to escape from the cage (personal observation). We could not see such differences in the other two pairs of lines, suggesting that if there is a positive association between coping style and aggression, this is best represented in the SAL-LAL lines. Earlier studies indicate that a similar association could have been present in the Turku lines, since the aggressive line TA was described as a better maze learner, while the nonaggressive TNA line was better passive learner (Kvist, 1989; Sandnabba, 1996). However, in our data there is no clear association. In order to understand the association between aggression and exploration in the open field we need to consider the meaning of explorative behaviour in a large neutral arena. This test may reveal the tendency to explore a new environment, related to a lack of fear. However, it might also relate more generally to activity levels. In our data on SAL and LAL mice, aggression is correlated with lower distance travelled. The activity in the open field is somewhat related to the activity levels in the home cage (Benus et al., 1988; Caramaschi et al., 2008b) and in an exposed area of a tunnel maze (Hogg et al., 2000). LAL nonaggressive mice seem to have a tendency for high spontaneous home-cage activity and new territory exploration. The other lines show different scenarios, which are perhaps more related to emotional aspects. In the Turku and North Carolina lines, the differences in activity levels in the open field are not supported by differences in the spontaneous cage activity. Our data confirm previous findings in the Turku lines (Selander and Kvist 1991; Ewalds-Kvist et al., 1997; Nyberg et al., 2003), while they are in contrast with findings in the NC lines (Hood and Quigley 2008). A complete agreement between those studies and ours would be difficult to achieve in any case, since the procedures applied differ substantially. A lack of 57
CHAPTER 2
association between aggressiveness and activity levels was reported in a study of mouse lines selected on high running wheel behaviour (Gammie et al., 2003). This confirms that the low exploration levels in the low-aggressive lines TNA and NC100 were associated with higher emotionality and fearfulness in the open field. Overall, these and the previous behavioural characterizations indicate that selection for aggression traits might act on different axes, the coping strategy and the emotionality axes, but these can be somehow correlated. Selection for high aggressiveness might result in proactive coping, as in the SAL-LAL selection. Alternatively, it might co-select for low emotionality/fearfulness, as in the Turku and North-Carolina lines. Acknowledgements We would like to thank Regina Scharma and Marlon van der Wal for extensive and careful help during the experiments and data collection.
58
AGGRESSIVENESS,
COPING STRATEGY AND EMOTIONALITY
59
CHAPTER
3
Development of violence in mice through repeated victory along with changes in prefrontal cortex neurochemistry Doretta Caramaschi, Sietse F. de Boer, Han de Vries and Jaap M. Koolhaas
Published in Behavioral Brain Research (2008) 189:263-72
ABSTRACT Recent reviews on the validity of rodent aggression models for human violence have addressed the dimension of pathological, maladaptive, violent forms of aggression in male rodent aggressive behaviour. Among the neurobiological mechanisms proposed for the regulation of aggressive behaviour in its normal and pathological forms, serotonin plays a major role. However, the results on the detailed mechanism are still confusing and controversial, mainly because of difficulties in extrapolating from rodent to human psychopathological behaviour. Our aim was to investigate the involvement of serotonin in pathological aggression. We subjected mice genetically selected for high (SAL, TA, NC900 lines) and low (LAL, TNA, NC100) aggression levels to a repeated resident-intruder experience (RRI mice) or to handling as a control (CTR mice). Pathological aggression parameters we recorded were aggression towards females and lack of communication between the resident and its opponent. In the same mice, we measured the monoamine levels in the prefrontal cortex, a brain region strongly involved in the regulation of motivated behaviour. Our results show that SAL mice augmented their proneness to attack and showed the most pathological type, with disregard of the opponent’s sex, high territorial behavioural patterns, and low sensitivity to signals of subordination. In contrast, TA and NC900 augmented their proneness to attack and low discrimination of the opponent’s signals, without showing offence towards females. After repeated resident-intruder experience, serotonin levels in the prefrontal cortex were significantly lower in SAL than in LAL whereas dopamine turnover was significantly higher, compared to CTR mice. Serotonin turnover was significantly reduced in all RRI mice, with no strain differences. Noradrenaline was significantly lower in aggressive mice of the TA and NC900 lines compared to their low-aggressive counterparts, with no effect of the repeated resident-intruder experience. We conclude that social experience changes prefrontal cortex neurochemistry differentially in individuals of different genetic background, and elicits pathologically aggressive phenotypes.
CHAPTER 3
INTRODUCTION Despite a large number of studies, the neurobiological determinants for the development of pathological aggression and violence are still far from clear. Usually in the context of resource competition, ritualized forms of aggressive behaviour are displayed that are under tight inhibitory reconciliation and appeasement mechanisms and hence do not frequently result in serious harm and injury (Benus et al., 1991b; Ferrari et al., 2005). However, in certain individuals under particular conditions, the motivation for aggression may escape control and escalate into violent and indiscriminate forms which inflict a considerable burden on society. Animal models for escalated aggression or violence often focus merely on the intensity factor as a parameter to delineate the escalation, for example measuring number of attacks or attack duration and frequency. Recently, it has been suggested that in order to mimic the human psychopathology, other dimensions of rodent aggressive behaviour should be measured, such as the loss of discrimination revealed by attacks towards females, attacks on vulnerable regions and/or insensitivity towards the social submission signals of the opponent (Haller and Kruk 2006). In humans, the loss of discrimination is often expressed by violence against women, particularly expressed in a domestic context. Sexual abuse and domestic violence have severe physical and psychological effects on the victims and represent a major problem in our society (Campbell, 2002; Pico-Alfonso, 2005). In laboratory conditions, intensively aggressive mice have been obtained through bidirectional artificial selection for aggression. Using this method, several genetic lines of mice selected for high (SAL, TA, NC900) or low (LAL, TNA, NC100) aggressiveness were generated (Lagerspetz and Lagerspetz 1971; van Oortmerssen and Bakker 1981; Cairns et al., 1983). In mice, a highly aggressive behavioral phenotype is present in a semi-natural population and is selected under certain environmental conditions because of high fitness due to better access to/defense of food and females (van Oortmerssen and Bakker 1981). Beyond a clear genetic predisposition to aggression, escalated aggression can also be achieved through frustration, instigation, alcohol consumption (De Almeida and Miczek 2002) or repeated social victory experiences (Fish et al., 2001; de Boer et al., 2003; Kudryavtseva et al., 2004). In particular, repeated victories may reinforce the use of aggressive behavior in order to achieve better position in social hierarchies. In such experiments, aggression increases in terms of duration and frequency, while attack latency decreases (Oyegbile and Marler 2005). It has been suggested that these escalations mimic the development of psychopathology, although it is not known to what extent the genetic aggressive predisposition is necessary for this process and therefore a risk factor for violence. Aggression by males towards females has been observed in the SAL and TA aggressive selection lines. SAL males attacked females 62
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
more than LAL and this behaviour was enhanced with a period of repeated daily winning experiences against males (Benus et al., 1990b). TA male mice attacked females more than TNA when they had previously been isolated (Nyberg et al., 2004). To our knowledge, aggression by males against females has never been studied in the NC lines. Other studies on psychopathological elements of intermale aggressive behaviour portrayed SAL mice as antisocial and violent compared to non-aggressive LAL, and TA as insensitive to social cues compared to TNA (Sluyter et al., 2003; Haller et al., 2006; Vekovischeva et al., 2007b). From clinical and preclinical studies it is known that, among the various central neurotransmitters and neuromodulators, serotonin plays a major role in the control of aggression and more generally of motivated behaviours, whereas monoamines in general play a crucial role in mood regulation (Millan, 2004). Brain serotonin is produced by neurons whose cell bodies form the raphe nuclei and whose projections reach virtually all brain areas, including the prefrontal cortex where the innervation is considerable. Among the brain areas involved in the regulation of aggression, prefrontal cortex is particularly interesting, since it is associated with aggressive psychopathologies. Reduced activity of the prefrontal cortex, in particular its medial and orbitofrontal portions, has been associated with violent/antisocial aggression (Raine et al., 2000; Blair, 2004). In laboratory rats, serotonin dynamics in the prefrontal cortex was associated to the execution of aggression (van Erp and Miczek 2000). Male mice of the SAL and TA aggressive lines had significantly lower serotonin tissue levels in the prefrontal cortex than the low-aggressive LAL and TNA lines, while the difference was not so pronounced in the NC lines (Caramaschi et al., 2007). These lower serotonin levels are associated with a higher inhibitory activity of the major short- and longlooped feedback regulatory mechanism of serotonin cells, the 5-HT1A receptor, as an autoreceptor in the case of SAL mice and as a postsynaptic receptor in the case of the TA mice (van der Vegt et al., 2001; Caramaschi et al., 2007). The first objective of our study was to escalate aggression levels from normal to pathological in mice genetically selected for high and low aggressiveness. We subjected SAL, LAL, TA, TNA, NC900 and NC100 male mice to repeated daily resident-intruder experience and tested aggression against females as a criterion of the development of pathological aggression. To confirm the interpretation of the results in terms of the possible development of a lack of social communication skills in highly aggressive mice, the sequential structure of behaviour during the last resident-intruder interaction was analysed in detail. Our second aim was to elucidate the involvement of prefrontal cortex serotonin levels in the development of pathological aggression. We measured serotonin and monoamine levels in the prefrontal cortex of the same mice and compared them with those of control mice of the same selection lines that never experienced any male-male interaction. 63
CHAPTER 3
MATERIALS AND METHODS Animals Male mice (n=60) from six different lines (SAL, LAL, TA, TNA, NC900 and NC100) obtained through three independent selection breeding programs (SAL, LAL = Groningen; TA, TNA = Turku, Finland; NC900, NC100 = North Carolina) were used as experimental subjects. They were kept from weaning (3–4 weeks of age) in familiar unisexual groups in Makrolon Type II cages, and subsequently (68 weeks of age) housed in pairs with a familiar female to avoid social isolation. Female mice (n=60) from the same lines were used as female intruders. MAS-Gro male mice were used as male intruders, since they exhibit a neutrally docile phenotype in a male-male confrontation. Male and female intruders were housed in unisexual groups of four animals. All the mice were at least 3–4 months old at the beginning of the experiment. Rodent food pellets (AMII, ABDiets, Woerden, The Netherlands) and water with a low chloride content were accessible ad libitum during the whole experiment. All the mice were kept under controlled 12/12 hrs light/dark cycles and a constant temperature of 22 ±2 ºC. The experiment consisted of a series of behavioural tests in the following order: female attack novel cage test 1, female attack home-cage test 1, repeated resident-intruder test, female attack home-cage test 2, and female attack novel cage test 2. All tests were performed with the approval of the Institutional Animal Care and Use Committee of the University of Groningen (DEC n. 4540A).
Repeated resident-intruder test To obtain pathologically aggressive mice, 30 male mice underwent a repeated resident-intruder treatment (RRI group). This consisted of nine male-male resident-intruder (RI) (van Oortmerssen and Bakker 1981; de Boer et al., 2000) experiences, one each day, carried out at the same time of the day (at the beginning of the dark period) in a test cage (75 x 29 x 27), where each male had previously been housed with a female, in presence of food and water. Each day, 1 hour before the RI experience, the female partner was removed from the cage. Subsequently, a naive male intruder was placed in the cage and the attack latency, i.e., the time it took the resident to attack the intruder, was scored. The intruder was removed from the experimental cage immediately after the first attack from the resident. If there was no attack, the test was stopped after 10 minutes and a score of 600 seconds was given. In the rare event of an attack from the intruder on the resident (observed in less than five cases, in the LAL and NC100 lines), the test was stopped immediately in order to avoid any defeat experience, and a score of 600 seconds was assigned. In the last resident-intruder experience (RI9), the 64
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
intruder was left for 5 minutes in the resident cage and a video recording was made for subsequent behavioural analysis. The remaining 30 experimental male mice were used as a control group (CTR). They were not subjected to the RRI paradigm, but instead were briefly handled and their female partners removed for the same duration every day as for the RRI group. Aggression against females As one criterion of pathological aggression, offensive behaviour by males towards females was examined before and after the RRI experience, both in a novel cage against a familiar female and in the home cage against an unfamiliar female. The first test measures the effects of a mild stressor (a novel environment) on the aggressive behaviour against females whereas the latter tests the situation of an unknown female in the own territory. Female attack novel cage (FN) At the beginning of the experimental session, each male-female pair was housed in a novel test cage provided with new bedding and nesting material, food and water. Any attack from the male mouse towards the female cage-mate within 30 minutes was scored. At the end of the RRI experimental session, a second test was performed when the male-female pairs were housed in standard cages. Female attack home cage (FH) This test was a modified version of a previously described resident-intruder test in which the intruder was an unfamiliar female mouse instead of a male. Two days after being housed together with a female in a test cage in which he could establish his territory, the resident male underwent a first FH test (FH1). Briefly, at the beginning of the dark phase, the familiar female was removed. One hour later, an unfamiliar female of the same line as the resident male was introduced in the cage containing the male. The interaction between the male resident and the female intruder lasted 5 minutes and was video-recorded for subsequent behavioural analysis. Immediately after the test, the familiar female was reintroduced after the removal of the unfamiliar one, which was returned to its home cage. The FH test was repeated after 10 days (FH2). In this test the same females were used as in FH1 but in a different order, so that each male would never encounter the same female intruder. In both FH1 and FH2, each female was checked during handling for oestrus or non-oestrus (dioestrus, metaoestrus or anoestrus). A small brush was gently inserted 0.5 cm into the vagina and rotated gently. The material obtained was smeared on a microscope slide, with three samples taken for each animal. A drop of methylene blue was added to each 65
CHAPTER 3
sample and allowed to dry overnight at room temperature. The next day, the samples were examined under a light microscope. Oestrus was inferred from a predominance of keratinized epithelial cells, and non-oestrus (dioestrus, prooestrus, metoestrus and anoestrus) inferred otherwise. Behavioural analysis of RI9 To further characterise pathological aggression, the interaction between a male resident and a male intruder during the last confrontation was examined. Behavioural analysis was performed using The Observer 5.0 (Noldus Information Technology b.v.), using low speeds (5x and 20x) when required. The behavioural states of the resident and the intruder in the last male-male interaction (RI9), were scored. Since the attack latencies were different among the mice, and they represented the first part of the total 5 minutes of interaction, in order to analyze the patterns in the aggressive behaviour between all the animals in a consistent amount of time, we scored one minute of interaction. The first minute after the first attack was chosen because it has the highest frequency of aggressive behaviour. The behavioural elements “attack”, “chase”, “threat”, “social exploration”, “mounting”, “non-social behaviours” and “inactivity” were scored for the resident animal, while “submission”, “move away”, “social exploration”, “non-social behaviours” and “inactivity” were scored for the intruder. The interactions were scored with a two-subject configuration, on separate occasions for the resident and the intruder, at low speed and with time synchronization in order to precisely identify the start and end of each behaviour. For each interaction, two channels of simultaneous behaviour sequences were thereby obtained to be analysed separately for the resident and intruder. The two channels were synchronized and considered as one sequence for an event-lag based sequential analysis, in order to identify predictable interaction patterns between the resident and the intruder (see data analysis and statistics for details). Brain-tissue preparation and HPLC In order to avoid any acute stress effect, all the animals were sacrificed under CO2 anaesthesia 24 hours after the last behavioural test, the female attack novel cage test 2. The male residents were weighed and decapitated and their brains removed. The prefrontal cortex was dissected from each brain, frozen in liquid nitrogen and stored at –80ºC. The PFC samples were homogenised in 1 ml 0.1 M perchloric acid for 60 seconds and centrifuged at 14,000 rpm for 10 min at 4ºC. The supernatant was removed and stored for 1-2 days at -80°C in order to avoid serotonin degradation, since -80 °C storage, even for a few weeks, was shown to yield comparable results (unpublished observations). 100 µl of supernatant were subsequently injected into a HPLC (High-Performance Liquid Chromatography) 66
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
column (Gemini C18 110A, 150 x 4.60 mm, 5 u, Bester) connected to a detector (analytical cell: ESA model 5011, 0.34 V). The mobile phase consisted of 62.7 mM Na2HPO4, 40.0 mM citric acid, 0.27 mM EDTA, 4.94 mM HSA and 10% MeOH (pH 4.1). Known amounts of monoamines were run in parallel for standardisation. Monoamine levels were calculated as ng/g tissue. Data analysis and statistics Attacks towards females in the FN and in the FH tests were expressed as ratio of attacking to non-attacking animals in each line and tested for significant line differences using an exact test of independence for a 2 x 6 contingency table (Mehta and Patel 1983; Kirkman, 2007). Since the SAL mice already attacked females in FH1, their attack/threat ratio was calculated and analysed using ANOVA for repeated measures, with “test” as a within-subject factor (female attack in home cage 1 vs. 2) and “group” as a between-subjects factor (RRI vs. CTR). For the RRI data, attack latency data were analysed using ANOVA for repeated measures with “day” (9 levels) as a within-subject factor and “type” and “selection” as between-subjects factors. The duration and frequency of the behaviours scored in the RI9 test were analysed within the aggressive and low-aggressive lines using a two-way MANOVA, with “selection” as a between-subjects factor. Post-hoc analyses were performed using a t test for independent samples in the case of two samples, Tukey's test in the case of multiple comparisons and a paired t test in the case of repeated measurements. Furthermore, an analysis of the sequential structure in RI9 behavioural sequences was performed according to the first-order Markov chain analysis model (Van Hooff, 1982) . Briefly, after grouping together all the animals of each line, matrices with first-order transition numbers across behaviours and subjects were obtained using the event-lag sequential analysis module from The Observer (Noldus Technology b.v.). The matrices were subsequently tested using MatMan (MfW version 1.1: Noldus Information Technology 2003; earlier version described in De Vries et al., 1993) for independence through calculation of adjusted residuals and the χ2 test. Residuals were considered statistically significant when p<0.05 after a sharper Bonferroni correction (Hochberg, 1988). High positive (negative) significant residuals indicate behavioural transitions that occur more (less) frequently than expected by chance on the basis of the row and column totals. Kinetograms were constructed to show which behaviours enhance the probability of occurrence of other behaviours (high positive residuals) or inhibit the occurrence of other behaviours (high negative residuals), relative to the overall occurrence of these behaviours. Differences between the lines were analysed qualitatively through comparison of the diagrams. 67
CHAPTER 3
The HPLC data on the monoamine amounts and their turnover in the prefrontal cortex were analysed using a two-way ANOVA within each selection using “group” (2 levels: control and RRI-treated) and “type” (2 levels: aggressive and low-aggressive) as between-subjects factors.
RESULTS Repeated resident-intruder (RRI) test The changes in attack latency during the RRI experience are shown in Fig. 3.1. Aggressive and low-aggressive mice from different selection lines were affected in a significantly different manner by the daily experience, as found in the “day*selection*type” interaction effect (multivariate test, Hotelling’s trace: F(16,32)=2.22, p<0.05). Therefore separate analyses within each selection program were performed. Since the data did not meet the sphericity assumption, the degrees of freedom were corrected using Huynh-Feldt epsilon. In the Dutch mice, SAL had significantly lower attack latencies than LAL throughout the whole experiment (F(1,8)=13.48, p<0.01), as expected. The repeated resident-intruder procedure reduced the attack latency in SAL and LAL mice (F(5,47)=2.34, p<0.05), but the overall reduction was not significantly different between the lines. In contrast, the overall attack latency of TA did not differ from that of TNA mice, but as in SAL-LAL mice it was significantly reduced during the 9-day procedure (F(4,36)=6.97, p<0.001), with no significant difference in the overall change. In the NC lines, NC900 had lower attack latencies than NC100 (F(1,8)=14.54, p<0.01) and a general reduction was observed throughout the 9day experiment (F(5,39)=5.83, p<0.001). However, the change was not significantly different between the two lines. Aggression against females The number of males that attacked females is expressed in Table 3.1 as percentages of the total number of mice of each sub-group. Female attack novel cage (FN) When housed in a new cage, SAL male mice exhibit offensive aggression toward their female partners within 30 minutes (p<0.001, with SAL having the highest residual). The phenomenon is already present in the first test and remains consistent after the RRI or control experience. SAL mice also attacked oestrous females. The attack can be so violent as to cause the death of the female, as observed in one cage the day after the FN. None of the males of the other lines exhibited attack behaviour towards their females within 30 minutes of the novel situation. 68
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
SAL LAL
600 500 400 300 200 100 0
TA TNA
attack latency (sec)
600 500 400 300 200 100 0
NC900 NC100
600 500 400 300 200 100 0
RI1
RI2
RI3
RI4
RI5
RI6 RI7
RI8
Figure 3.1 Change in attack latency of aggressive (SAL, TA, NC900) and low-aggressive (LAL, TNA, NC100) mice during the repeated resident-intruder paradigm. Data are expressed as mean attack latency ± S.E.M. observed on each day of testing (from RI1 to RI9).
RI9
Female attack home cage (FH) SAL males attacked non-familiar females both before and after the treatment. After the RRI or the control experience, some mice of the TA and NC900 lines were triggered to attack their females, suggesting a non-specificity of the treatment. Attacks from the NC100 mice were observed after the RRI period. Mice from the SAL and NC100 lines showed offensive aggression also towards oestrous females. SAL mice attacked females significantly more than mice from the other lines in the first test both in the control (p<0.001, with SAL having the highest residual) and in the experienced group (p=0.036, with SAL having the highest residual), whereas in the second test SAL attacked significantly more only in the control group (p=0.011, with SAL having the highest residual), probably because of the increase in attacks after the experience in the other lines. 69
CHAPTER 3
Table 3.1 Percentage of males that attacked a female in each group and in each test performed. Line
Group
Novel cage1
Novel cage2
Home cage1
Home cage2
SAL
control RRI control RRI control RRI control RRI control RRI control RRI
80 80 0 0 0 0 0 0 0 0 0 0
100 80 0 0 0 0 0 0 0 0 0 0
80 100 0 0 20 0 0 0 17 0 25 0
80 60 0 0 60 20 0 0 33 20 0 40
LAL TA TNA NC900 NC100
Since SAL mice fiercely attacked their females already in FH1, we analysed them separately using ANOVA. A specific effect of the RRI treatment was a significant increase in the attack/threat ratio (Fig. 3.2) in the SAL males that attacked females in FH1 (test*group interaction effect: F(1,14)=10.72, p=0.007; Tukey's post-hoc test: pre-RRI vs. post-RRI, p<0.05). Analysis of social communication in RI9 Fig. 3.3 shows the total duration of the behaviours of resident and intruder mice in the first minute of resident-intruder test 9. Multivariate ANOVA found overall a highly significant “type” effect (F(5,22)= 274.7, p=0.006), a marginally non-significant “selection” effect (F(5,22)= 29.39, p=0.057) and a significant “type*selection” interaction effect (F(5,22)= 69.31 , p=0.012). As expected, the aggressive mice attacked (F(1,29)= 34.97, p<0.001), threatened (F(1,29)= 8.99, p<0.01) and chased (F(1,29)= 4.6, p<0.05) the intruders more than low-aggressive mice. The
*
attack/threat ratio
6 5 4
RRI CTR
3 2 1 0
70
*
pre
post
Figure 3.2 Attack/threat ratio observed in SAL males that attacked females at FH1. Data are expressed as mean + S.E.M. of the RRI and CTR groups, in FH1 (pre) and FH2 (post), i.e. before and after the 9-day experience respectively. Post hoc analyses: * p<0.05.
DEVELOPMENT
duration (sec)
40
SAL TA NC900
resident non-aggressive lines
LAL TNA NC100
30 20 10 0
50 40
duration (sec)
resident aggressive lines
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
attack
threat
chase
social non- inactivity exploration social
intruder aggressive lines
SAL TA NC900
attack
threat
chase
social non- inactivity exploration social
intruder non-aggressive lines
*
#
LAL TNA NC100
30 20 10 0
social non-social inactivity submission moveexploration away
social non-social inactivity submission moveexploration away
Figure 3.3 Duration and frequency of the resident’s and intruder’s behaviour during the last resident-intruder test (RI9). Data are expressed as group mean + S.E.M. See text for statistical details and explanation. Post hoc analyses: * p<0.05, # 0.05
mice from the low-aggressive lines spent more time in social exploration (F(1,29)= 13.75, p< 0.01). Significant differences were also found in the intruders’ behaviour. When exposed to aggressive mice, the intruders showed more submission and move-away behaviours (F(1,29)= 53.06, p<0.001). When exposed to lowaggressive mice, the intruders exhibited more social and non-social behaviours (F(1,29)= 4.66, p<0.05, F(1,29)= 21.61, p<0.001). Within the low-aggressive lines (“type*selection” interaction effect: F(2,29)=3.77, p<0.05), the intruders’ nonsocial behaviour was significantly reduced when confronted by the TNA residents, compared to LAL and NC100. The within and between resident-intruder behavioural transitions are depicted in Fig. 3.4. Due to the low transition frequencies in the low-aggressive lines, the analysis was performed only in the aggressive lines. When tested for independence, the transition matrices all showed dependence across subjects and behaviours (SAL: χ2(71)= 411.38, p<0.001; TA: χ2(131)= 458.42, p<0.001; 71
CHAPTER 3
SAL
TA RESIDENT
INTRUDER 24.6%
threat 69
RESIDENT
submission 68 threat 68
52.2
47.8%
24.4%
move-away 48 attack 84
INTRUDER
submission 48
35.5%
34.3%
48.4%
move-away 67
attack 78
33.7%
35.5% 16.7%
75%
chase 32
16.7%
social exploration 21
47.6% 25.8% 8.7%
non-social behaviors 31
25.8%
inactivity 18
52.6%
social exploration 3 non-social behaviors 4
social exploration 24
inactivity 19
non-social behaviors 22
47.1%
chase 20
31.8%
social exploration 7 non-social behaviors 10 inactivity 15
inactivity 10
NC900 RESIDENT
INTRUDER
submission 77
threat 143
59%
attack 135
30.5%
chase 14
non-social behaviors 59
31.5%
move-away 81
7.1%
social exploration 3 non-social behaviors 12
20.4%
inactivity 18
11.5%
35.7%
social exploration 45
46.6%
46.6%
23.5%
Figure 3.4 Kinetogram of behavioural transitions performed in the first minute after the first attack of the last male-male resident-intruder confrontation (RI9) by male mice from the aggressive lines (SAL, TA and NC900). The areas of circles are proportional to the overall observed frequency of each behavioural event (frequencies shown). For NC900 only, the size of the boxes is reduced by factor 2, since the frequencies were much higher than in the other lines. Continuous arrows indicate the transitions that occur significantly more than by chance, whereas dashed arrows indicate the transitions that occur significantly less than by chance. The widths of continuous arrows are proportional to the number of transitions. On each arrow, the observed percentage of transitions is indicated.
72
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
NC900: χ2(71)= 306.02, p<0.001). The analysis of the residuals showed different within- and between-individuals dependency patterns for each line. The behaviour of SAL residents is defined by a fairly strict sequence of behaviours. SAL residents enter offensive behaviour from social exploration, reaching threat, and from threat and chase reaching attack, while occasionally they exit from threat to non-social behaviours. The communication between the resident and the intruder is well represented, with resident’s threat followed by submission and resident’s attack followed by intruder’s move-away. The intruder’s behaviour is somehow inhibiting the resident’s offence, since move-away inhibits attack and submission enhances the resident’s non-social behaviours. Like SAL, TA residents show some dependence within their behaviours (chase and threat enhance attack, which in turn enhances threat, and non-social behaviours enhance social exploration), although to a much lower degree and without significant entry or exit transitions to and from the offensive behaviours. The interaction with the intruders is similar to that of SAL residents, although less striking. As in SAL, TA attack enhances the intruder’s move-away, but in contrast to SAL, submission does not inhibit or enhance any of the resident’s behaviours in TA. The only inhibiting transition that involves the intruder’s behaviour is from move-away to attack. NC900 residents show clear within-resident dependence. The loop between threat and attack previously described in SAL and TA is also present in NC900, but in this case it is more prominent, with both behaviours also represented at high frequencies. The shift from offence to inoffensive behaviour happens as a transition from chase to non-social behaviours, while there is no evidence for an entry to offence. The interaction pattern with the intruder is more noticeable than the within-individual dependence. Similarly to SAL and TA, the resident’s attack enhances the intruder’s move-away, but in contrast to SAL and TA it inhibits the intruder’s submission, non-social behaviours and inactivity. Differently from SAL and TA, the NC900 resident's threat inhibits inactivity. Similarly to SAL and TA, the intruder’s move-away behaviour inhibits attack, but in the NC900 mice it also inhibits threat. Biochemical data The concentrations of noradrenaline, serotonin, dopamine and their metabolites in the PFC are reported in Fig. 3.5. Noradrenaline was significantly lower in the prefrontal cortex of the TA (F(1,16)=6.89, p<0.05) and NC900 (F(1,16)=8.42, p<0.05) aggressive lines compared to their less-aggressive counterparts. TA and TNA mice that underwent the repeated resident-intruder experience had a significantly lower amount of noradrenaline than control mice (F=13.88, p<0.01). No significant effects were found in the SAL–LAL model. 73
CHAPTER 3
NA
ng/g tissue
dopamine
*
600
600
*
*
400
400
200
200
0
0
1000
** ***
800
ng/g tissue
control rri
*
5-HT
500 400
*
*
600
200
200
100
1.0
**
* **
300
400
0
5-HIAA
*
0
5-HIAA/5-HT
(HVA+dopac)/dopa
*
**
3
0.8
**
0.6
*
*
*
0.4
* **
* *
2
1
0.2 0.0
SAL
LAL
TA
TNA NC900 NC100
0
SAL
LAL
TA
TNA NC900 NC100
Figure 3.5 Amounts of noradrenalinee (NA), dopamine, serotonin (5-HT), 5-HIAA, serotonin turnover ratio (5-HIAA/5-HT) and dopamine turnover ratio ((HVA+dopac)/dopa) in the prefrontal cortex of mice of the control and repeated-resident-intruder (RRI) groups. Data were obtained using HPLC and are expressed as mean + S.E.M. for each aggressive (SAL, TA, NC900) and low-aggressive (LAL, TNA, NC100) mouse line. * Significantly different at p<0.05. ** Significantly different at p<0.01. *** Significantly different at p<0.001. # 0.05 < p < 0.10.
No significant effects were found for dopamine levels in the prefrontal cortex of any of the selection models studied. However, the dopamine turnover was affected by the RRI experience in the SAL–LAL mice in an opposite way (type*treatment effect: F(1,16)=14.61, p<0.01), with a marginally non significant increase in SAL-RRI mice (p=0.057) and a significant decrease in LAL-RRI mice (p=0.027) compared to their respective controls. 74
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
A highly significant “type*treatment” effect in the SAL-LAL serotonin content (F(1,16)=18.75, p=0.001) is represented by a significant increase in both lines after RRI treatment (F(1,16)=166.22, p<0.001), which was much more pronounced in the low-aggressive line. Indeed, SAL and LAL mice had similar serotonin levels in the control group, while after RRI experience SAL had significantly lower serotonin levels than LAL (p<0.01). Serotonin metabolite 5-HIAA was significantly reduced in the SAL, LAL, TA and TNA mice after RRI experience (SAL-LAL: F(1,16)=96.78, p<0.001; TA-TNA: F(1,16)=12.77, p<0.01). Serotonin turnover was decreased in RRI mice from all the lines (SAL-LAL: F(1,16)=96.78, p<0.001; TA-TNA: F(1,16)=18.50, p=0.001; NC900-NC100: F(1,16)=6.6, p<0.05) compared to control groups, although the effect was more pronounced in SAL and LAL mice.
DISCUSSION The present study shows that differences in prefrontal cortex serotonin levels are associated with a particular type of pathological aggression, induced by subjecting wild-derived mice genetically predisposed to aggression to a repeated winning experience. This is the first study, to our knowledge, relating prefrontal neurochemical changes to a rodent’s pathological aggression and genetic predisposition for aggressiveness. Effects of social experience on aggression and violence Repeated social experience, in the form of resident-intruder interactions, escalates aggression levels in aggressive lines, as represented by the decrease in attack latency. We could not exclude that the low-aggressive lines also escalated their aggression levels, since TNA mice reach very low attack-latency values. However, while the attack latency usually correlates with the level of expressed offensive aggression, it is more an index of motivation to engage in aggressive behavior. As a general conclusion, although all the lines were genetically selected for high versus low aggressiveness, there are differences between the three models in the way they are affected by social experience and in the type of escalated aggression that the aggressive lines exhibit. SAL, TA and NC900 mice are all highly aggressive and their aggression levels escalate according to the “winner effect”, namely the increased probability of winning an aggressive encounter following previous victories (Oyegbile and Marler 2005; Hsu et al., 2006). However, SAL mice show high levels of pathological aggression, measured as offensive behaviour toward familiar and unfamiliar females, both in the home-cage and in a novel environment, even before any male-male experience. This study is not the first to report 75
CHAPTER 3
aggressive acts from males toward females in the wild house mouse. In a wild house-mouse population observed in semi-natural conditions, natural selection favoured the evolution of a highly aggressive mouse phenotype that exhibited violence against males and females indiscriminately and against pups and juveniles as well as adults (van Oortmerssen and Busser 1989). In a laboratory setting, the percentage of SAL males that attacked familiar females after 9 days of male-male resident-intruder training was significantly higher than that of LAL mice, in which this behaviour was almost absent (Benus et al., 1990b). In our experiment, most of the SAL mice attacked familiar and unfamiliar females before and after the nine male-intruder tests. In line with the previous experiment, the extremely aggressive phenotype of the SAL mice suggests a violent component comparable to that of a highly aggressive human personality. As it was described by Sluyter and colleagues (Sluyter et al., 2003), SAL mice have this and other characteristics of violent men that persistently displayed antisocial behaviour in a human longitudinal study. Aggressive animals experienced the repeated malemale resident-intruder paradigm as a repeated winning experience, since at the first attack the intruders were showing submissive postures that indicated the establishment of a dominant-subordinate hierarchy. This experience exacerbated the aggressive phenotype of SAL individuals, enhancing their attack/threat ratio against females. It is tempting to consider the aggressive behaviour towards females as an indication of a lack of social communication skills. If this interpretation is correct, one might expect to see a lack of social communication in a male-male interaction as well. A detailed analysis of the sequential structure of the social interaction was used to study the sensitivity of the resident to the opponent’s signals. The aggressive behaviour of the SAL male clearly depends on the behaviour of the opponent. However, this sensitivity is able to inhibit SAL’s aggression only temporarily. SAL behaviour also showed intrinsic regulation, as seen in the high degree of intraindividual behavioural dependence, although the behavioural pattern comprises the rodent-typical sequence of behaviours shown by dominant males (from social exploration to threat and attack, or from threat to non-social behaviours and inactivity). Different pictures are revealed by the TA and NC900 aggressive lines, obtained from mouse laboratory strains, where no attacks on familiar females were observed and very little offensive aggression was shown towards unfamiliar females. However, the latter was enhanced by daily handling and separation from the female partner (control group) and it was also seen in TNA and NC100 mice, even though they were less prone to aggressive behaviour towards males, suggesting that this effect was specific neither to repeated victory experience, nor to individuals genetically predisposed to aggressiveness. In the male-male interac76
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
tion, TA mice did not show clear within- and between-individual dependence, therefore the cues for determining their behavioural shifts may not lie in intrinsic motivation or in intruder’s signals. The NC900 were characterized by a higher behavioural turnover, especially in the attack/threat shifts, suggesting some uncertainty of motivation of an approach/withdrawal type. NC900 discriminate opponent’s signals better than TA, although the type of communication seems to differ from that in SAL interactions. In general, both lines show less typical territorial patterns compared to SAL. To interpret this result it is important to remember that the strains from which these lines originated consisted of albino animals that had been already selected for being not so active, easy to handle, not aggressive with their cage-mates and easily bred in laboratory cages. It may be that components of the dominant phenotype were lost during this procedure. A similar trend has been well documented (de Boer et al., 2003) in rats, in which escalated aggressive traits typical of a high proportion of wild rats are completely missing from the standard laboratory Wistar rat strain. TA aggression may be more sensitive to environmental cues outside the cage or highly insensitive to cues in general, whereas NC900 aggression may be largely determined by the motility/immobility of the opponent in the cage since the behavioural dependence is higher after move-away/inactivity than after submission. There is also a possibility that these two lines, having originated from albino strains, show some sensory-motor impairments compared to the dominant wild-derived mice, so their sensitivity to external cues may be disrupted by indirect factors (Balkema and Dräger 1991; Adams et al., 2002). In conclusion, we consider these three escalated forms of aggression as violence, because of the high intensity, persistence, and poor social communication. It seems that the three highly aggressive lines we studied may represent different types of pathological aggressive behaviours in humans, for example the aim-focused, hypoarousally driven, persistent aggression of psychopaths (represented in SAL mice) and highly emotional, reactive forms of pathological aggression (represented in TA and NC900) (Vitiello and Stoff 1997; Blair, 2004). More research on the physiology and neurochemistry of these animals is needed to elucidate this concept. Monoamine levels and types of aggression Prefrontal cortex neurochemistry seems to vary according to the different types of aggression observed. Serotonin is the neurotransmitter mostly involved in differentiating the violent SAL type from the docile LAL, since its prefrontal cortex level is differentially changed by the male-male repeated resident-intruder paradigm in SAL and LAL mice. The fact that serotonin level was higher in the SAL-LAL animals that underwent the repeated social experience compared to the controls seems to contradict the “serotonin-deficiency” hypothesis, which states that highly 77
CHAPTER 3
aggressive/impulsive individuals show diminished serotonergic transmission activity (Valzelli, 1982; Lee and Coccaro 2001). However, the serotonergic system is also activated during a resident-intruder interaction, as shown previously by measuring raphe neuronal activation in rats (van der Vegt et al., 2003b) and extracellular serotonin levels by microdialysis in lizards (Summers et al., 2003a; Summers et al., 2003b). Hence, the higher need for serotonin in social interactions may have enhanced the baseline tissue content in SAL and LAL mice. During a social encounter, serotonin is rapidly but transiently released and, consequently, feedback mechanisms decrease the neuronal firing of the raphe nuclei. This may result in a long-term increase of serotonin synthesis by increasing the activity of tryptophan hydroxylase (Boadle-Biber, 1993), the rate-limiting enzyme for the production of serotonin. The serotonin levels in the prefrontal cortex of the experienced SAL and LAL mice are similar to the data obtained in our previous study where SAL and LAL mice were previously tested for aggression several times throughout an experiment (Caramaschi et al., 2007). The lower amount of serotonin in the prefrontal cortex of SAL mice compared to LAL mice observed in that experiment might be due to the social experience the mice had during the aggression test that was performed before the experiment to screen their behavioural phenotype. The higher sensitivity of the inhibitory 5-HT1A autoreceptor (Caramaschi et al., 2007), a major feedback mechanism for the serotonergic nuclei, is a possible mechanism for the resulting lower serotonin change in the SAL mice compared to that in the LAL. The serotonergic system was affected by the social experience in terms of a strong reduction of serotonin turnover. However, since the change was found in both the aggressive and low-aggressive lines, there is no association with victory, aggression and violence. Perhaps this change is reflected in other behavioural characteristics such as impulsivity or anxiety that we did not explore in this study. Dopamine turnover was enhanced in SAL and lowered in LAL after the repeated social test. An earlier study showed higher nigrostriatal dopaminergic activity in SAL than in LAL mice, when all the mice were previously screened for aggressive behaviour (Benus et al., 1991a). It may be that the difference observed was due to the social experience, as shown in this experiment. An association between violent/impulsive behaviour and dopamine has previously been suggested, although the mechanism is far from clear (Retz et al., 2003). Our data support the idea that a difference in the reduction of serotonin availability is associated with differences in dopamine neurotransmission, according to an inverse relationship. In TA mice characterised by high aggression levels but little intrinsic and intruder-based regulation, aggressive behaviour is associated with low noradrenaline tissue levels in the prefrontal cortex. This result is in apparent contrast with 78
DEVELOPMENT
OF VIOLENCE AND PREFRONTAL NEUROCHEMISTRY
an early study in TA and TNA mice, in which the aggressive line showed higher brainstem noradrenaline levels than the low-aggressive one (Lagerspetz et al., 1968). The difference in brain region investigated may underlie this discrepancy. However our results suggest a negative correlation between noradrenaline levels in the prefrontal cortex and aggressiveness, and are in line with previous findings on aggressive patients with Alzheimer’s dementia (Matthews et al., 2002). In conclusion, this study shows that violence can be engendered in wildderived mice genetically selected for aggression and that in these mice the genotype interacts with social experience, resulting in low increase in prefrontal cortex serotonin levels and dopamine neurochemistry and leading to the reinforcement of the dominant status towards a psychopathological condition. The behavioural analysis and the neurochemical data show that the genetically selected lines develop distinct violent behaviour that is associated with differential prefrontal cortex dynamics. Acknowledgements The authors would like to thank Ramon A. Granneman for the HPLC analysis, Auke Meinema for the animal care, and Tim W. Fawcett for proofreading the manuscript. All the animal experimental procedures were performed accordingly to the Dutch Law for Animal Experiments and approved by the Animal Experiment Committee (DEC) of the University of Groningen (D4050).
79
CHAPTER
4
Is hyper-aggressiveness associated with physiological hypo-arousal? A comparative study on mouse lines selected for high and low aggressiveness Doretta Caramaschi, Sietse F. de Boer and Jaap M. Koolhaas
Published in Physiology & Behaviour (2008) 95:591-8
ABSTRACT Aggressiveness is often considered a life-long, persistent personality trait and is therefore expected to have a consistent neurobiological basis. Recent meta-analyses on physiological correlates of aggression and violence suggest that certain aggression-related psychopathologies are associated with low functioning of the hypothalamo-pituitary-adrenal (HPA) axis and autonomic nervous system (ANS). We tested this hypothesis in mice selected for high and low aggressiveness by measuring baseline plasma corticosterone levels and, via radiotelemetry, heart rate and core body temperature. The radiotelemetric recordings were made for 48 hours under baseline undisturbed conditions and for 90 minutes after a handling stressor. Consistent with the hypoarousal hypothesis of violence, we found lower resting heart rates in two out of the three highly aggressive selection lines. In contrast, body temperature during the active phase, as another ANS-regulated physiological parameter, was higher in two out of three highly aggressive lines. The handling-induced tachycardiac and hyperthermic responses were similar across the six mouse lines except for the most docile and obese line, which showed a blunted reactivity. Besides significant differences between strains, no differences in plasma corticosterone levels were found between the high- and low-aggressive phenotypes. These results are discussed in relation to the different types of aggression (normal versus pathological) exhibited by the three highly aggressive lines. We conclude that while high trait-like aggressiveness is generally associated with a higher active-phase core body temperature, only animals that express pathological forms of aggression are characterized by a low resting heart rate.
CHAPTER 4
INTRODUCTION Aggressiveness is a behavioural trait used by individuals in competition with each other for vital resources such as food, territory, and mates, and to communicate social status (Koolhaas et al., 1999; Koolhaas et al., 2007). When appropriately displayed, aggressive behaviour serves an important biological function in securing these resources to reproduce successfully and transmit genes, and thus becomes evolutionarily conserved (Sih et al., 2004a). However, when expressed in an escalated manner, it is intense, frequent, injurious, and may lose its function in social competition and communication. In humans, this intense aggressiveness is referred to as violence, and is recognized as a pathological condition that requires treatment and prevention (Krug et al., 2002; Moffitt et al., 2008). In order to develop successful intervention programs, it is necessary to understand the neurobiological causes of violence and their generality with respect to different forms of aggression and different environmental backgrounds (Miczek et al., 2007). It has been established that pathological aggressiveness in humans has a heritable component and that certain individuals have a predisposition to engage in violent acts (Raine, 2002b). These individuals are aggressive and antisocial already in early childhood, their violent behaviour progresses during adolescence, and it persists during adulthood. These so-called persisters differ from individuals that abstain from violence after adolescence (Broidy et al., 2003). The best-replicated biological correlate of life-persisting extreme aggressiveness so far is low resting heart rate (Raine, 2002a; Lorber, 2004). Further characteristics include a low baseline activity of the hypothalamic-pituitary axis in delinquent adolescent humans and in violent rats (Haller et al., 2004; Popma et al., 2006), and deficits in prefrontal cortex functioning in antisocial psychopaths and violent rats (Raine et al., 2000; Blair, 2004; Halasz et al., 2006). Together, these data have led to the formulation of the “hypoarousal theory of pathological aggression”, which states that individuals with low physiological arousal engage more in violent acts throughout their lives as a form of stimulation-seeking behaviour, without fear of punishment or aversive outcome (Raine, 2002a; Haller and Kruk 2006). While the literature concerning low resting heart rate and violence is quite consistent, autonomic reactivity to stressors seems to be considerably more variable. Hostile/aggressive Type A individuals show a positive correlation between trait aggression and emotionality measures (Contrada et al., 1982; Ward et al., 1986). Similarly, resident rats that readily attack male conspecifics have higher sympatho-adrenomedullary activation (Summers and Greenberg 1994; Sgoifo et al., 1996; Sgoifo et al., 2005). Similarly, in pigs, aggression levels and sympathetic 82
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
activation during social confrontation are positively correlated (Fernandez et al., 1994). Lines of mice selected for heightened aggressiveness, but originating from different strains, have been characterized in a comparative way in terms of their more or less violent aggressive phenotypes (Sluyter et al., 2003; Haller et al., 2006; Caramaschi et al., 2007; Caramaschi et al., 2008a; Natarajan et al., 2009), and may therefore represent a useful model to investigate the physiological correlates of pathological aggression. The high-aggressive Short Attack Latency (SAL), Turku Aggressive (TA), and North Carolina 900 (NC900) male mice show similarly high frequencies and long durations of offensive aggression and short attack latencies to male docile intruders in their home cage. However, only SAL males show no discrimination based on the opponent’s sex, fiercely attacking unfamiliar and familiar females, and very little discrimination in response to the opponent’s inhibitory cues (Caramaschi et al., 2008a). TA and NC900, although showing high levels of offensive aggression, do not attack females. In addition, while TA, like SAL mice, show a high motivation to maintain an offensive interaction, NC900 show a more fragmented pattern of social interactions with the intruder (Caramaschi et al., 2008a; Natarajan et al., 2009). To date, information on the autonomic and neuroendocrine physiological phenotype of these lines is scarce. Regarding autonomic sympathoadrenal functioning, adrenaline content in the adrenals and in the brain stem of TA mice is higher than that of the corresponding low-aggressive Turku Non-Aggressive (TNA) mice (Veenema et al., 2003b). When exposed to injection stress, core body temperature in SAL, TA, and NC900 male mice responds more than in the corresponding low-aggressive lines (Caramaschi et al., 2007). Regarding the HPA axis, baseline corticosterone levels are similar between SAL mice and their low-aggressive LAL (Long Attack Latency) counterparts, but SAL males show a substantially reduced reactivity to stress (Veenema et al., 2003b). Studies on the immune system and aggression have revealed a lower sensitivity of the HPA axis to neonatal endotoxin exposure in NC900 compared to the low-aggressive NC100 mice, as well as lower serum corticosterone levels during handling (Lagerspetz and Lagerspetz 1971; Granger and Hood 1997). The aim of the present research is to characterize the physiological phenotypes associated with these three different types of highly aggressive mice and to examine the generality of the hypo-arousal hypothesis of violence. In accordance with the literature cited above, we expect low resting and low stress reactivity values for the physiological parameters under study in all three high-aggressiveness lines.
83
CHAPTER 4
MATERIALS AND METHODS Animals and experimental design For this experiment, we used male mice of the aforementioned lines obtained through three different breeding programs for the selection of high aggressiveness (SAL, TA, NC900) and low aggressiveness (LAL, TNA, NC100) (Lagerspetz and Lagerspetz 1971; van Oortmerssen and Bakker 1981; Cairns et al., 1983). The mice were bred in our own laboratory at the University of Groningen, The Netherlands, and kept after weaning (at 21 days of age) in unisexual familiar groups in Makrolon cages (type II). At 50 days of age, each male mouse was paired and caged with a female of the same line, to avoid male-male competition and social isolation. All the males were tested for attack latency using the standard procedure previously described (van Oortmerssen and Bakker 1981). Agematched docile albino laboratory mice of the MAS-Gro strain were used as opponents. For the corticosterone assay, n=5 mice of each line were decapitated at the beginning of the dark phase under CO2 anesthesia, and their trunk blood collected (for further details, see Corticosterone assay). For the telemetry recordings, n=5 or 6 male mice were implanted with a radio transmitter (for details, see Transmitter implantation and biotelemetry setup). Heart rate, core body temperature and activity were recorded around the clock to assess baseline physiology, whereas stress physiology was assessed from the heart rate, temperature and activity response to a brief handling challenge. During the entire experiment, the animals were kept under standard laboratory conditions, at a temperature of 22 ± 2 °C and on a 12:12 light-dark cycle (light on at 00.30), with food (AMII, ABdiets, Worden, The Netherlands) and water available ad libitum. The experiments were approved by the Institutional Animal Care and Use Committee of the University of Groningen, the Netherlands, in compliance with the European Communities Council Directive of 24 November 1986 (86/609/EEC). Corticosterone assay Trunk blood was collected in chilled tubes containing EDTA for determination of corticosterone levels. Blood samples were centrifuged at 2600g for 10 min at 4°C. Plasma samples were stored at –20°C until assayed. Plasma corticosterone was determined in duplicate using ImmuChemTM Mouse Double-antibody Corticosterone 125I RIA Kit, MP Biomedicals, LLC, Diagnostics division, Orangebourg, NY, US. The minimum detectable dose of corticosterone using this assay was 7.7 ng/ml, with an intra-assay variation coefficient of 4.4% and an inter-assay variation coefficient of 6.5%. 84
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
Transmitter implantation and biotelemetry setup For the biotelemetry recordings, a TA10ETA-F20 mouse transmitter (DSI, St. Paul, MN, US) was implanted surgically in the intraperitoneal cavity of each male mouse (n=5 or 6 for each line) after reaching at least 20 g of body weight. During the surgery, the mice were anaesthetized with a mixture of isoflurane (5% to induce and 3% to maintain) and NO2/O2, and placed on a Harvard heating pad to avoid hypothermia. The surgery was performed as previously described in Caramaschi et al., 2007. Immediately after surgery, each mouse was placed in a clean cage under a heating lamp to avoid post-operative hypothermia. Cages were placed on a platform receiver connected via a matrix to a computer running Dataquest Labpro software for data collection. The mice were then allowed to recover undisturbed in their home cage for 4–6 days alone, and for the following week with their female partner. During this period, body weight and physiological parameters were monitored to ensure complete recovery and the presence of physiological circadian pattern. Data collection and analysis Plasma corticosterone levels were compared using a two-way ANOVA with type (two levels: aggressive and non-aggressive), strain (three levels: SAL/LAL, TA/TNA, NC900/NC100), and type*strain interaction as between-subject effects. Post-hoc Tukey’s pairwise comparisons were performed to decompose significant effects of factors with more than two levels. Baseline recordings of heart rate, temperature and activity were obtained by sampling segments of 10 seconds every 5 minutes for a period of 48 hours in which the animals were left undisturbed in their home cages. For heart rate and temperature, the 24-hr best-fitted curve was obtained using a linear harmonic regression fit that describes the data by adding harmonics to the principal wave function (Oster et al., 2006). Averages of maximum, minimum, and amplitude were computed and compared within each pair of selected lines using t tests for independent samples. Stress data were collected as response to handling, which consisted of lifting the experimental animal and placing it on a scale in order to measure body weight and immediately afterwards putting it back in its home cage. All handling on a given day took place in a 15-minute time window between 10.00 and 14.00. Before and after handling, heart rate and temperature data were sampled every 5 minutes and logged automatically by a computer. The 60 minutes before the experimenter entered the room was considered a 'pre-stress' period, which was averaged in the analysis to give one datapoint (pre-handling). After handling, stress response was monitored for 90 minutes (post-handling), which was sufficient time to see a return to pre-stress values. Within each strain, the response 85
CHAPTER 4
curve of the aggressive line was compared with that of the low-aggressive line using a repeated-measures ANOVA with time (19 levels) and type (aggressive and non-aggressive) as within- and between-subject factors, respectively. From the heart rate and temperature time-response curves, the area under the curve (AUC) was computed to obtain an overall response measure. For the overall response in activity, the sum of the 60-min baseline (pre-handling) activity was subtracted from the sum of the 90-min response (post-handling) activity. AUCs and total activity group averages were compared within each pair of selected lines using t tests for independent samples.
RESULTS Attack latency test The behavioural test for aggressiveness, in which the animals had to face the challenge of a male intruder in their home cage, confirmed the highly aggressive phenotype of the SAL, TA, and NC900 lines and the low-aggressive phenotype of LAL, TNA, and NC100. All mice of the three high-aggressiveness lines quickly attacked the male intruders (attack latency in seconds, mean ± S.E.M.: SAL=15.54±3.81; TA=77.04±25.49; NC900=83.53±30.44). As previously shown (Caramaschi et al., 2007; Caramaschi et al., 2008a), among the lowaggressive lines, TNA mice showed a considerable amount of aggressiveness in terms of the number of attacking mice per line (LAL=0/11, TNA=6/11, NC100=1/11) and average attack latency (LAL=300±0, TNA=209.94±31.51, NC100=295±5). However, attacking TNA mice spent significantly less time in offensive behaviours than the highly aggressive lines (one-way ANOVA on all the attacking mice: F4,32=8.65, p<0.001; meanTNA=16.45 vs. meanSAL=46.51, t13= –4.5, p=0.001; meanTNA vs. meanTA=51.96, t13=–4.59, p=0.001; meanTNA vs. meanNC900 = 44.22, t11= –3.42, p=0.006). Corticosterone data As shown in figure 4.1, plasma corticosterone levels were not associated with aggression. A two-way ANOVA on log-transformed values revealed a significant strain effect (F2,28=8.44, p=0.002), which is due to the fact that TA/TNA mice had lower values than the other mice (compared with SAL/LAL, at p=0.001; compared with the NC lines, at p=0.046). Baseline physiology around the clock As shown in figure 4.2 and summarized in table 4.1, SAL mice had very low heart rates during their resting period. The minimum value was significantly lower in 86
HYPER-AGGRESSIVENESS
corticosterone (ng/ml)
160
**
AND HYPO -AROUSAL?
*
120 80 40 0
SAL
LAL
TA
TNA NC900 NC100
Figure 4.1 Plasma corticosterone levels at the beginning of dark phase in SAL, LAL, TA, TNA, NC900 and NC100 mice (n=6 for each line). * p<0.05, ** p<0.01 with Tukey’s pairwise comparisons.
SAL than in LAL mice (t10=–2.71, p=0.022) and in TA compared to TNA mice (t9=–5.98, p<0.001). The circadian wave in heart rate of the aggressive lines had a larger amplitude than that of low-aggressive lines, although significantly so in only TA compared to TNA (SAL/LAL: t10=2.2, p=0.052, TA/TNA: t9=3.88, p=0.004, NC900/NC100: t10=1.94, p=0.084). As shown in figure 4.3 and in table 4.2, active phase temperature values were significantly higher in the TA and NC900 lines compared to their low-aggressive counterparts (TA/TNA: t10=2.46, p=0.034, NC900/NC100: t10=2.39, p=0.038), contributing to a difference in amplitude, with TA and NC900 having significantly bigger amplitudes than TNA and NC100, respectively (TA/TNA: t10=3.93, p=0.003, NC900/NC100: t10=4.15, p=0.002). Furthermore, in NC900, the resting-phase temperature was significantly lower than in NC100 (t10=–2.43, p=0.035). The general activity levels (figure 4.4 and table 4.3) were zero for all the mice at the minimum point of the cycle, which was during the light phase. The values for the dark phase were log-transformed in order to render the distribution normal and reduce the mean-to-variance relationship. No significant differences were revealed by t tests between the high-aggressive and low-aggressive lines (SAL vs. LAL: t10=–1.82, p=0.1; TA vs. TNA: t10=1.65, p=0.1; NC900 vs. NC100: t9=–1.87, p=0.09). However, after removal of three outliers, the logtransformed activity levels were significantly higher in the LAL mice than in SAL (t8=–2.56, p=0.02). In the other lines, no significant differences were found.
87
CHAPTER 4
900 800 700 600 500 400
SAL LAL
300 900
heart rate (bpm)
800 700 600 500 400
TA TNA
300 900 800 700 600 500 400
NC900 NC100
300 0
12
24
36
48
hours
Figure 4.2 Circadian rhythm of average heart rate (bpm = beats per minute) in SAL, LAL, TA, TNA, NC900 and NC100 mice (n=5 or 6 for each line) measured every 5 minutes. Each panel shows two alternative mouse lines obtained through artificial selection for high (solid line) and low (dotted line) aggression. Standard errors were omitted for clarity.
88
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
40 39 38 37 36 35
SAL LAL
34 40
temperature (°C)
39 38 37 36 35
TA TNA
34 40 39 38 37 36 35
NC900 NC100
34 0
12
24
36
48
hours
Figure 4.3 Circadian rhythm of average body temperature (°C) in SAL, LAL, TA, TNA, NC900 and NC100 mice (n=5 or 6 for each line) measured every 5 minutes. Each panel shows two alternative mouse lines obtained through artificial selection for high (solid line) and low (dotted line). Standard errors were omitted for clarity.
89
CHAPTER 4
500 400
SAL LAL
300 200 100 0
500 400
TA
activity (counts)
TNA
300 200 100 0
500 400 NC900
300
NC100
200 100 0
0
12
24
36
48
hours
Figure 4.4 Circadian rhythm of average activity (counts) in SAL, LAL, TA, TNA, NC900 and NC100 mice (n=5 or 6 for each line) measured every 5 minutes. Each panel shows two alternative mouse lines obtained through artificial selection for high (solid line) and low (dotted line) aggression. Standard errors were omitted for clarity.
90
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
Table 4.1 Heart rate (bpm) during baseline conditions. Mouse line SAL LAL TA TNA NC900 NC100
Max
Min
755.0 ± 18.1 728.8 ± 30.6 636.0 ± 13.2 649.2 ± 9.0 653.7 ± 22.8 632.8 ± 41.3
401.4 ± 26.8 * 498.3 ± 23.8 475.6 ± 3.9 ** 550.4 ± 10.8 423.0 ± 9.9 498.1 ± 46.9
Amplitude 353.6 ± 23.0 # 230.5 ± 51.0 160.4 ± 10.4 *** 98.8 ± 10.9 230.7 ± 30.3 # 134.7 ± 39.0
t-test comparison with corresponding low-aggressive line * p<0.05, ** p<0.01, *** p<0.001, # 0.05
Table 4.2 Temperature (°C) during baseline conditions. Mouse line SAL LAL TA TNA NC900 NC100
Max 38.3 ± 0.08 38.1 ± 0.17 38.0 ± 0.07 * 37.7 ± 0.09 37.9 ± 0.17 * 37.4 ± 0.13
Min 35.5 ± 0.21 35.5 ± 0.38 35.7 ± 0.15 36.0 ± 0.10 35.5 ± 0.12 * 36.0 ± 0.15
Amplitude 2.76 ± 0.21 2.54 ± 0.44 2.31 ± 0.15 ** 1.68 ± 0.06 2.43 ± 0.18 ** 1.44 ± 0.16
t-test comparison with corresponding low-aggressive line * p<0.05, ** p<0.01
Table 4.3 Activity (counts) during baseline conditions. Mouse line SAL LAL TA TNA NC900 NC100
Max
Min
Amplitude
391 ± 57.43 652.3 ± 133.5 553.5 ± 226.8 244.2 ± 29.4 344.8 ± 42.8 447.3 ± 38.3
0 0 0 0 0 0
391 ± 57.43 652.3 ± 133.5 553.5 ± 226.8 244.2 ± 29.4 344.8 ± 42.8 447.3 ± 38.3
91
CHAPTER 4
900
900
800
800
heart rate (bpm)
heart rate (bpm)
Physiological response to handling Heart rate (figure 4.5) significantly increased in all lines due to the handling procedure (SAL/LAL: F18,180=10.19, p<0.001, TA/TNA: F18,180=3.21, p<0.001, NC900/NC100: F18,180=5.63, p<0.001). A repeated-measures ANOVA did not show a differential change between aggressive and non-aggressive lines, although the overall response relative to baseline, measured as AUC, was lower in NC100 than in NC900 (t10=2.88, p=0.016). Temperature (figure 4.6) significantly increased in all lines due to the handling procedure (SAL/LAL: F18,180=26.43, p<0.001, TA/TNA: F18,180=22.27, p<0.001, NC900/NC100: F18,180=26.14, p<0.001). All the aggressive lines had significantly higher values of temperature during the whole period (SAL/LAL: F1,10=1.94, p=0.012, TA/TNA: F1,10=7.65, p=0.012, NC900/NC100: F1,10=16.23, p=0.002). However, the change was significantly different only
700 600 500 SAL LAL
400
700 600 500 TA TNA
400
300
300 0
20
40
60
80
0
20
min
40
60
80
min 30000
900
AUC (beats)
heart rate (bpm)
800 700 600 500
20000
10000
NC900 NC100
400 300
0 0
20
40
60
80
SAL
LAL
TA
TNA NC900 NC100
min
Figure 4.5 Heart rate (bpm = beats per minute) in response to handling stress in SAL, LAL, TA, TNA, NC900, and NC100 mice (n=6 for each line), depicted as group means ± standard error. Each panel shows two alternative mouse lines obtained through artificial selection for high (black) and low (white) aggression. The bottom-right panel shows the overall response calculated as AUC (=Area Under the Curve) for all the lines.
92
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
40
40
39
39
temperature (°C)
temperature (°C)
between the NC900 and NC100 lines, as shown by the repeated-measures ANOVA (time*type interaction effect: F18,180=1.67, p=0.048) and analysis of the AUCs (t10=3.98, p=0.003). General locomotor activity (figure 4.7) increased after handling in all lines (SAL/LAL: F18,180=5.37, p<0.001, TA/TNA: F18,180=6.85, p<0.001, NC900/ NC100: F18,180=10.74, p<0.001). The increase in activity due to handling was lower in the SAL mice compared to LAL (time*type interaction effect: F18,180=1.78, p=0.031), and lower in the NC100 mice compared to NC900 (time*type interaction effect: F18,162=1.81, p=0.028). However, the total activity corrected for the baseline showed a significant difference only between NC900 and NC100 mice (t9.=3.98, p=0.027; analysis performed on log-transformed data to correct for non-normality of the distribution).
38 37 36 35
SAL LAL
0
20
40
60
38 37 36 35
80
TA TNA
0
20
min
60
80
min 300
40 39
AUC (°C*min)
temperature (°C)
40
38 37 36
200
100
NC900 NC100
35
0 0
20
40
60
80
SAL
LAL
TA
TNA NC900 NC100
min
Figure 4.6 Core body temperature (°C) in response to handling stress in SAL, LAL, TA, TNA, NC900, and NC100 mice (n=6 for each line), depicted as group means ± standard error. Each panel shows two alternative mouse lines obtained through artificial selection for high (black) and low (white) aggression. The bottom-right panel shows the overall response calculated as AUC (=Area Under the Curve) for all the lines.
93
CHAPTER 4
500 SAL LAL
400 300 200 100
activity (counts)
activity (counts)
500
0
TA TNA
400 300 200 100 0
0
20
40
60
80
0
20
min
60
80
min
500
5000 NC900 NC100
400 300 200 100 0
total activity (counts)
activity (counts)
40
4000 3000 2000 1000 0
0
20
40
60
80
SAL
LAL
TA
TNA NC900 NC100
min
Figure 4.7 General activity (counts) in response to handling stress in SAL, LAL, TA, TNA, NC900, and NC100 mice (n=5 or 6 for each line), depicted as group means ± standard error. Each panel shows two alternative mouse lines obtained through artificial selection for high (black) and low (white) aggression. The bottom-right panel shows the overall response calculated as activity after handling minus activity before handling, for all the lines.
DISCUSSION Consistent with the hypoarousal theory of violence, low resting heart rates were observed in highly aggressive SAL and TA. These selection lines have previously been shown to be violent in terms of the sequential structure of agonistic interactions and insensitivity to inhibitory cues from the opponent (Caramaschi et al., 2008a; Natarajan et al., 2009). High peak temperatures, as seen in TA and NC900 mice, seem to be associated with less violent forms of aggression. Aggression in these selection lines is still context-dependent, since it has been shown previously that they do not attack familiar females in a novel cage or immobilized opponents (Caramaschi et al., 2008a; Natarajan et al., 2009). 94
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
Heart rate As previously reported in other mouse strains, our lines showed clear circadian rhythms in heart rate, with the highest activity in the dark phase. The lower resting heart rate in the aggressive lines is in agreement with another mouse study, in which 5-HT1B knock-out mice displayed an aggressive phenotype and a low resting heart rate (Bouwknecht et al., 2001). Both our study and Bouwnecht’s reported resting heart rate values lower than 500 bpm, while in other laboratory strains that typically display low levels of aggression the values remain around 500 bpm on average (van Bogaert et al., 2006; Depino and Gross 2007). Overall, we can conclude that low resting heart rate is associated with high levels of trait aggression in mice. The difficulty in extrapolating these data to humans is that the various studies performed on human populations are very heterogeneous in their operational definitions of aggression, as well as in the characteristics of the samples in terms of age, sex, socio-economic status, substance abuse, conviction, and so on. However, recent reviews support an association in humans between low resting heart rate and high injurious aggression (Arnett, 1997; Raine, 2002a; Lorber, 2004), high antisocial/aggressive behaviour (Arnett, 1997), and antisocial psychopathy (Raine, 2002a). Recently, this association has been confirmed by Popma and co-workers (Popma et al., 2006) and Raine (Raine, 2003). In contrast, an association between aggression and resting heart rate was not found in our NC900/NC100 lines. We can interpret this one of in two ways. One possibility is that the high levels of aggression exhibited by the NC900 males are representative of normal mouse aggression, since male mice, in contrast to other rodent species and humans, display high levels of physical aggression (Ferrari et al., 2005). Alternatively, the high levels of aggression in the NC900 mice could represent a reactive form of aggression, according to the distinction between proactive/instrumental and reactive/emotional behaviour emphasized by (Scarpa and Raine 1997). In support of the first interpretation, NC900 mice showed no offensive behaviour toward females, and higher sensitivity to the opponent’s cues than the other aggressive lines (Caramaschi et al., 2008a). Moreover, their 5HT1A/serotonin system was not associated with their high aggression levels, in contrast to the other two more violent aggressive lines (Caramaschi et al., 2007). The most consistent finding regarding heart rate in this study is the larger circadian amplitude in the high-aggressive lines, compared to the low-aggressive lines. This may represent a superior physical fitness in the more aggressive lines (Atkinson et al., 1993). Low resting heart rates in the most violent individuals strengthen this interpretation of superior physical fitness, since low resting rates are a typical characteristic of well-trained athletes in humans (Shin et al., 1997).
95
CHAPTER 4
Temperature and activity As with heart rate, temperature showed daily variations with the highest peak occurring during dark phase. Both TA and NC900 mouse lines showed an association between high aggression and high peak temperature. Again, we can view the absence of this association in the SAL and LAL lines in one of two possible ways. First, one has to take into account the activity levels, since activity generates heat (Refinetti, 1994). Higher average activity levels in the LAL line might raise their body temperature slightly during the active phase, bringing it up to values comparable to those of SAL mice. Higher activity levels in LAL were only partially seen in this study, but they have been shown in a previous study (Benus et al., 1988). Alternatively, the high peak temperature/aggression pattern might be a characteristic of less violent forms of aggression, mainly of the reactive type. Indeed, activity in the cage does not seem to have an effect on circadian rhythms in temperature, as previously shown in a rat telemetry experiment (Strijkstra et al., 1999). As with heart rate, the amplitude of daily temperature variation was greater in the TA and NC900 aggressive mice than their low-aggressive counterparts, though this pattern did not extend to the SAL/LAL lines. In general, we can conclude that aggressive mice display bigger amplitudes in their circadian rhythms. Autonomic reactivity to stress The handling experiment showed that only in the NC900/NC100 lines was aggression positively associated with higher heart rate and temperature reactivity to the stressor. As discussed earlier, selection for aggression may include a coselection for higher reactivity to stress only in the case of discriminative aggression, and not in the most pathological selection conditions. In that case, the behavioural and physiological phenotype of these mice might then be more similar to the human reactive/emotional/hostile heightened aggressive type, rather than a mixture of instrumental/proactive and emotional/reactive types, which might be represented by the other two lines. Unfortunately this is difficult to determine, since the resident-intruder test for aggression is a typical challenge for reactive aggression, since the mice are threatened by the intrusion of an unfamiliar mouse in the cage. We have no information about the levels of instrumental aggression in these highly aggressive mouse lines. To our knowledge, a test for instrumental aggression in rodents has not yet been developed. Alternatively, following a more careful interpretation, we should acknowledge that the stress-induced increase in temperature and heart rate was significantly lower in the NC100 non-aggressive line than in all the other lines, aggressive and non-aggressive (Temperature AUC: contrast30=3.22, p=0.003; Heart rate AUC: contrast30=61.2, p=0.002). This lower autonomic responsiveness might be more directly related to the fact that these mice may have developed some form of 96
HYPER-AGGRESSIVENESS
AND HYPO -AROUSAL?
metabolic dysfunction related to obesity. In support of this hypothesis, the NC100 mice have a significantly higher body weight (means (g)±S.E.M: SAL=21.64 ±0.66, LAL= 21.24±0.59, TA= 34.1±0.6, TNA= 30.06±0.52, NC900=37.78 ±1.33, NC100=48.82±2.5; NC100 significantly different from the other lines with t10=–7.82, p<0.001, planned contrasts corrected for inequality of variances), fairly low baseline activity levels (this paper), and higher leptin plasma concentration (means (ng/ml)±S.E.M: SAL=2.13±0.35, LAL=2.78±0.57, TA= 3.02±0.44, TNA=2.34±0.54, NC900=3.58±0.45, NC100=5.8 ± 0.72; NC100 significantly different from the other lines with t5=–4.01, p=0.012, planned contrasts corrected for inequality of variances). In line with these findings, it has been shown that diet-induced obese rats are hypo-responsive to stress (Levin et al., 2000). This line of reasoning should be further explored in the obesity-prone NC100 mouse line. The discrepancy between the three pairs of selected lines with regard to stress reactivity could also be explained in terms of specificity of the stressor. Handling is a routine procedure in the lab, a mild stressor that might be perceived with a similar degree of threat by mice with alternative behavioural phenotypes and that does not allow them to exhibit their alternative behavioural repertoires aimed to cope with the situation. It would be interesting to examine further the autonomic responsiveness in a situation where the mice can exhibit their higher or lower aggressiveness. Indeed, in rats, autonomic reactivity is positively associated with high levels of aggression (Sgoifo et al., 1996), and the response is dependent on the stressor applied (Koolhaas et al., 1997). In humans, too, the higher autonomic reactivity to a stressor in type A individuals seems to depend on the stressor applied (Ward et al., 1986; Lee and Watanuki 2007). Hence, in line with previous research on mice (van Bogaert et al., 2006), such a routine stressor as handling was stressful enough to elicit in mice a pronounced physiological response, consisting of hyperthermia, tachycardia and hyperactivity, and is therefore relevant for understanding stress-related autonomic activation. In conclusion, our study shows that different types of aggressive behaviour are associated with different physiological traits. The selection lines that exhibit aggression/violence in its most indiscriminate form are characterized by autonomic hypo-arousal in the resting phase. Less pathological forms of aggression are related to hyper-arousal during the active period. Acknowledgements The authors would like to thank Auke Meinema for breeding the mice, Izabella Jonas, Deepa Natarajan and Dr. Tim Fawcett for help during the preparation of the manuscript.
97
CHAPTER
5
Differential role of the 5-HT1A receptor in aggressive and non-aggressive mice: an across-strain comparison Doretta Caramaschi, Sietse F. de Boer and Jaap M. Koolhaas
Published in Physiology & Behaviour (2007) 16:590-601
ABSTRACT According to the serotonin (5-HT)-deficiency hypothesis of aggression, highly aggressive individuals are characterized by low brain 5-HT neurotransmission. Key regulatory mechanisms acting on the serotonergic neuron involve the activation of the somatodendritic inhibitory 5HT1A autoreceptor (short feedback loop) and/or the activation of postsynaptic 5-HT1A receptors expressed on neurons in cortico-limbic areas (long feedback loop). In this study, we examined whether low serotonin neurotransmission is associated with enhanced 5-HT1A (auto)receptor activity in highly aggressive animals. Male mice (SAL-LAL, TA-TNA, NC900-NC100) obtained through different artificial-selection breeding programs for aggression were observed in a resident-intruder test. The prefrontal cortex level of 5-HT and its metabolite 5-HIAA were determined by means of HPLC. The activity of the 5-HT1A receptors was assessed by means of the hypothermic response to the selective 5-HT1A agonists S-15535 (preferential autoreceptor agonist) and 8-OHDPAT (full pre- and postsynaptic receptor agonist). Highly aggressive mice had lower serotonin levels in the prefrontal cortex and two out of three aggressive strains had higher 5-HT1A (auto)receptor sensitivity. The results strengthen the validity of the serotonindeficiency hypothesis of aggression and suggest that chronic exaggerated activity of the 5-HT1A receptor may be a causative link in the neural cascade of events leading to 5-HT hypofunction in aggressive individuals.
CHAPTER 5
INTRODUCTION The neurobiology of aggressive behaviour includes a central role for brain serotonin (5-HT) neurotransmission. The most favored idea is the “5-HT deficiency” hypothesis of aggression (Giacalone et al., 1968; Brown et al., 1982; Mehlman et al., 1994; Berman et al., 1997), stating that high trait-like levels of aggressive behaviour are associated with low 5-HT neurotransmission activity. This hypothesis implies that individuals of high and low aggressiveness might somehow differ in the regulation of 5-HT neurotransmission. An important regulatory mechanism of the serotonergic system is represented by 5-HT1A receptors (Blier et al., 1998; Pineyro and Blier 1999). These are located pre-synaptically as autoreceptors on the soma and the dendrites of serotonergic neurons, as well as postsynaptically on non-serotonergic neurons in several corticolimbic areas that receive 5-HT terminals (Azmitia et al., 1996; Kia et al., 1996a). Activation of somatodendritic 5-HT1A autoreceptors by 5-HT or 5-HT1A agonists potently decreases the firing rate of 5-HT neurons and consequently leads to the suppression of 5-HT synthesis, 5-HT turnover and 5-HT release in the diverse projection areas (Hamon et al., 1988; Hjorth et al., 1995; Pineyro and Blier 1999). In addition to this local somatodendritic autoreceptor short negativefeedback loop, a long feedback loop also exists, which entails postsynaptic 5-HT1A receptor activation to inhibit 5-HT neuron-firing activity via reducing excitatory afferent input (Pineyro and Blier 1999; Hajos et al., 1999; Celada et al., 2001). Furthermore, it has been shown that 5-HT1A agonists exert potent anti-aggressive properties in rodents (Olivier et al., 1995; Pruus et al., 2000; de Boer and Koolhaas 2005), and in particular in animals that show high and/or escalated levels of aggressive behaviour (Miczek et al., 1998; Stork et al., 1999; de Boer et al., 2000). Therefore, it can be hypothesized that an impairment of serotonergic neurotransmission in high aggressive animals may be caused by excessive inhibitory 5-HT1A receptor activity. Several years ago, three independent artificial selection programs in the Netherlands, Finland and North Carolina were carried out in mice. This has lead to the generation of three highly aggressive lines (SAL= Short Attack Latency, TA= Turku Aggressive, NC900= North Carolina 900) and three “docile” lines (LAL= Long Attack Latency, TNA= Turku Non Aggressive, NC100= North Carolina 100) ((van Oortmerssen and Bakker 1981; Gariepy JL et al., 1996; Sandnabba, 1996); for review see (Miczek et al., 2001)). Several studies suggest that these pairs of highly and weakly aggressive lines show similar differences in the serotonergic system. The overall brain content of serotonin is lower in SAL than in LAL mice (Olivier et al., 1990). In another study (Veenema et al., 2005a) the amounts of 5-HT, 5-hydroxyindoleacetic acid (5-HIAA) and the 5-HIAA/5-HT 100
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
ratio were examined in different brain regions and in different phases of the daily light/dark cycle. The 5-HT baseline level in the brain stem of SAL mice was lower than that for LAL during the light phase, while the metabolic ratio was lower in the striatum and in the amygdala. In a second experiment, the 5-HT content of saline-treated SAL mice was lower in the prefrontal cortex, striatum, hippocampus and amygdala than of saline-treated LAL mice (Veenema et al., 2005a). Neurochemical studies in the Finnish selected mice showed similar results. Mice from the aggressive line (TA) had 19 % lower levels of serotonin in the prefrontal cortex than mice from the non-aggressive line (TNA) (Lagerspetz et al., 1968). An initial study on the regulation of the serotonergic system found higher 5-HT1A receptor ligand binding in SAL than LAL in the dentate gyrus, CA1, lateral septum and frontal cortex, and higher 5-HT1A mRNA in the CA1 and dentate gyrus of the hippocampus (Korte et al., 1996). Further analysis (Veenema et al., 2005a) replicated some of these findings. However, no difference was found in 5-HT1A mRNA expression and binding capacity in the prefrontal cortex and lateral septum, probably due to a circadian fluctuation affecting the results. Moreover, the behavioural response tested in the forced-swimming test to 8-OH-DPAT, a selective full preand postsynaptic 5-HT1A agonist, and to S-15535, a preferential pre-synaptic 5HT1A agonist, was different between the two lines, suggesting a different sensitivity of this receptor. Indeed, the hypothermic response to the full 5-HT1A receptor agonist alnespirone, as measured with a rectal probe, was higher in SAL than in LAL (van der Vegt et al., 2001), indicating a more sensitive 5-HT1A receptor functioning in the aggressive line. In line with these findings, electrophysiology experiments on hippocampal slices showed that the serotonin-induced membrane hyperpolarization was more pronounced in SAL mice than in LAL mice (Van Riel et al., 2002). To our knowledge, no investigation of the serotonergic system in the NC900 and NC100 lines has been performed. This study focuses on the involvement of the 5-HT1A autoreceptor regulation of serotonergic activity in aggressive behaviour. The three genetic selection lines of high and low aggressive behaviour were used to study the general validity of this putative autoreceptor-mediated control in aggression. According to the “serotonin-deficiency” hypothesis of impulsive and violent aggression, we would expect high aggression to correlate with low serotonin and/or low serotonin metabolism. To relate aggressiveness levels to serotonin, the behaviour of the mice was observed in a resident-intruder paradigm and the tissue concentrations of 5-HT, 5-HIAA and the 5-HIAA/5-HT ratio in the prefrontal cortex were determined. In order to assess the functionality of the preand postsynaptic 5-HT1A receptors, highly aggressive and non-aggressive mice were injected with the selective 5-HT1A agonists S-15535 (Millan et al., 1993b) and 8-OH-DPAT (Bjork et al., 1989), and a decrease in body temperature due to 101
CHAPTER 5
the drugs was the readout for the sensitivity of the receptor. In rats, the novel ligand S-15535 has been described as a suppressor of the stress-induced hyperthermia (SIH) due to the injection (de Boer et al., 2000), while 8-OH-DPAT is well known to cause hypothermia (Hjorth, 1985; Hillegaart, 1991; de Boer et al., 2000). A tonically higher functional sensitivity of the 5-HT1A receptor would explain a stronger inhibition of the serotonergic output from the raphe nuclei.
MATERIALS AND METHODS Animals Male mice (Mus musculus domesticus) from six different strains obtained by means of three selective breeding programs for offensive aggression were used. Short Attack Latency (SAL) and Long Attack Latency (LAL) were inbred strains derived from a wild population in Groningen, the Netherlands (van Oortmerssen and Bakker 1981). Turku aggressive (TA) and non-aggressive (TNA) were outbred strains obtained through artificial selection on Swiss albino mice in Turku, Finland (Sandnabba, 1996). NC900 (aggressive) and NC100 (non-aggressive) were outbred strains derived from selection on ICR mice in North Carolina (Gariepy JL et al., 1996). The animals were bred in our laboratory and kept in unisexual familiar groups until weaning in perspex cages (17 x 11 x 13 cm). The mice were weaned at 3-4 weeks of age and housed with a female of the same line at the age of 6-8 weeks, in order to avoid social isolation and inter-male competition. During all experiments each male-female pair was housed in a Makrolon Type II cage (375 cm2) with sawdust as bedding material, shredded paper (envirodry) as nesting material and a cardboard tube as cage enrichment. Food in the form of rodent pellets (AMII, ABDiets, Woerden, The Netherlands), and water with a low chloride content were provided ad libitum. The animals were kept in a room with a 12:12 light/dark cycle and constant temperature (22 ± 2ºC). Biotelemetry At 4-6 months of age, a group of male mice of each line was implanted with a biotelemetry transmitter for chronic core body temperature recordings. Two types were used: 3000XM-FH (Mini-Mitter, USA) and TA10ETA-F20 (DSI, St Paul, Minnesota, USA). During surgery, the animals were anesthetized with 5% isoflurane/O2/N2O, placed on a Harvard homoeothermic heating pad in order to prevent hypothermia due to the anesthetics, and maintained under anesthesia with 2.5% isoflurane/O2/N2O. A transmitter was placed inside the abdominal cavity, and in the case of TA10ETA-F20, one lead was fixed to the xyphoid process of the sternum and the other to the pectoral muscular layer in the right medi102
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
astinum. Internal suturing was performed by means of reabsorbable surgical thread, while skin was sutured with silk with a reverse-knot method, in order to prevent chewing by the animal. Natrium penicillin was injected as post-surgery antibiotic treatment. The signal produced by each transmitter was received by an antenna/receiver board (RPC-1, Data Sciences Int., 32 x 22 x 3 cm), placed underneath the animal’s cage. Each receiver was connected to a consolidation matrix (BCM-100), which was in turn connected to a PC (IBM Pentium-compatible) data-acquisition and analysis system (Dataquest LabproTM, Data Sciences). The data-acquisition parameters were set for 10 seconds' sampling every 5 minutes on a 24 h basis. The animals were allowed to recover for at least 14 days after the surgery, during which time their temperature and locomotor activity were monitored. The re-establishment and stability of circadian rhythmicity was a prerequisite for the start of the experiments. Pharmacological challenges Prior to the start of and one week after the pharmacological challenge session, each male mouse implanted for telemetry was tested in the Attack Latency Time (ALT) test according to the procedure described by Van Oortmerssen and Bakker (van Oortmerssen and Bakker 1981), in order to confirm the aggressive or nonaggressive behavioural phenotype, which could have been affected by the surgical procedure. Behavioural results showed that the aggression trait did not change due to the pharmacological challenges. Drug challenge tests were performed during different phases of the light/dark cycle. A wash-out period of 1–3 days between treatments was used, since 24 hours are considered enough to obtain complete clearance of S-15535 and 8OHDPAT (Zuideveld et al., 2001; Vis et al., 2001). Each experimental animal received a series of subcutaneous injections in randomized order, which consisted of distilled water (vehicle, 5 ml/kg of body weight), S-15535 (10 mg/kg) and 8OH-DPAT (0.25 or 0.5 mg/kg). The volume injected was 5-10 ml/kg of body weight. The animals were distributed to different cohorts for logistic reasons. The first and the second cohorts consisted of SAL and LAL mice implanted with MiniMitter transmitters, whereas the third cohorts included also mice from the TA, TNA, NC900 and NC100 lines, all implanted with DSI transmitters. Due to the different baseline body temperature of SAL and LAL mice between the light (first and second cohorts) and the dark phase (third cohort), these data were analyzed separately. For TA, TNA, NC900 and NC100, the data were averaged and analyzed together. S-15535-3 methanesulfonate [(4-benzodioxan-5-yl)-1-(indan-2-yl)piperazin, lot n. EI798] was provided by Institut de Recherches Internationales Servier, 103
CHAPTER 5
France. 8-OH-DPAT [(±)-8-Hydroxy-2-(dipropylamino) tetralin hydrobromide] was obtained from Sigma-Aldrich, the Netherlands. Drugs were dissolved in distilled water (vehicle) at room temperature. Behavioural testing: the resident-intruder paradigm At least one week after the last injection, all the animals were tested for aggressive behaviour using a modified version of the ALT test (van Oortmerssen and Bakker 1981). One day prior to the test, the males were housed in one compartment of an 80 x 30 x 30 cm partition cage with their female partners. One hour before the first test, the females were removed. The test, which was performed at the beginning of the dark phase in dim light, consisted of placing an albino male naive intruder (MAS-Gro) in one side of the cage, physically separated from the resident male by a perforated sliding partition that allowed sensory contact. When the resident showed interest in the intruder (sniffing, approaching), or after 5 minutes if no interest was detected, the partition was removed to allow a physical interaction. At the first attack, the Attack Latency Time (ALT) was recorded. The intruders were not removed after the first attack, as in the original version of the test; instead, a 5-minute videotape recording of the interaction for each animal was made. If there was no attack from the resident, the test was continued for a maximum of 10 minutes, since an ALT of 600 seconds would arbitrarily indicate a true non-attacking phenotype. Immediately after the test, all animals (males and females) were put back in their home cages. From the videotape recordings, and using The Observer software (Noldus Information Technology bv), the following behaviours were scored, as described by Koolhaas et al. (Koolhaas et al., 1980) in rats and Brain (Brain, 1981) in mice: digging, non-social investigation (explore, rear, supported rear, scan), social investigation (approach, crawl over, crawl under, follow, groom, head groom, investigate, nose sniffing), immobility, resting, body care (self-grooming, wash, shake, scratch), drinking/ eating, attack (charge, lunge, attack, chase), threat (aggressive groom, sideways offensive, upright offensive, tail rattle), withdrawal (retreat), and defense. An ethogram was created based on the following behavioural classes, which were subsequently analyzed: (1) non-social exploration (digging + non-social investigation); (2) social investigation; (3) inactivity (immobility + resting); (4) body care; (5) drinking/eating; (6) offense (attack + threat); (7) withdrawal; (8) defense. Biochemical assay In order to determine the 5-HT and 5-HIAA contents, all the animals were anesthetized with CO2 and decapitated at least two days after the behavioural test. The brains were rapidly removed from the skull and the prefrontal cortex (PFC) removed, frozen in liquid nitrogen and stored at –80 º C. The PFC samples were 104
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
homogenized in 1 ml 0.1 M perchloric acid and centrifuged at 14000 rpm for 10 min at 4ºC. The supernatant was removed and 100 µl were injected into a HPLC (High-Performance Liquid Chromatography) column (Gemini C18 110A, 150 x 4.60 mm, 5 u, Bester) connected to a detector (analytical cell: ESA model 5011, 0.34 V). The mobile phase consisted of 62.7 mM Na2HPO4, 40.0 mM citric acid, 0.27 mM EDTA, 4.94 mM HSA and 10% MeOH (pH 4.1). Known amounts of 5HT and 5-HIAA were run in parallel for standardization. Monoamine levels were calculated as ng/g tissue. Data analysis Temperature telemetry data were collected from 60 minutes before the injection until 120 minutes after the vehicle or drug injection respectively. For each animal, the average of the values before each challenge was considered as a baseline. After qualitative examination of the data, the end of the response was set at 90 minutes after the injection of vehicle and 60 minutes after S-15535 or 8-OHDPAT. The analysis of the vehicle response was carried out using a Repeated-Measures ANOVA with “Time” (19 levels) as within-subject factor, and “Type” (2 levels: aggressive and non-aggressive) as between-subject factor, in order to examine possible differences that could mask the drug responses. The time-response data for each drug were then divided by the corresponding time-response data for the vehicle, for each animal. A repeated-measures ANOVA with “Time” (13 levels) as within-subject factor and “Type” (2 levels: aggressive and non-aggressive) as between-subject factor was performed on the corrected data within each selection program. All post hoc analyses were performed using a t-test for independent samples. The ALT and the behavioural data obtained from the videotape scoring were analyzed using a two-way ANOVA with “type” (2 levels: aggressive, non-aggressive) and “selection” (3 levels: Netherlands, Finland, North Carolina) as betweensubject factors. A similar two-way ANOVA was performed on the HPLC data for 5-HT, 5-HIAA and the 5-HT/5-HIAA ratio. Post hoc analyses were carried out by means of t tests and Tukey tests for multiple comparisons. Moreover, for each attacking mouse (from aggressive and non-aggressive lines), AUC (“Area Under the Curve”, namely the area comprised between the baseline and the response curve) and Dmax (maximum difference from baseline) values were calculated from the drug/vehicle ratios for S-15535 challenge and used to compute a correlation matrix together with the following variables: 5-HT, 5-HIAA, 5-HIAA/5-HT, ALT and Offense. All the statistical analyses were carried out using SPSS version 14.0. 105
CHAPTER 5
RESULTS Behavioural type confirmation The ALT test performed on the mice that underwent a surgical procedure confirmed the aggressive and non-aggressive phenotype of almost all the animals. Only one SAL did not attack in 10 minutes of confrontation. This animal was excluded from the data analysis. The other mice of the aggressive lines had ALT <100 s, with only four mice with ALT between 100 and 150 seconds. All the attacking mice from the non-aggressive lines (<10% of the LAL, 75% of the TNA) were considered in the data analsysis, since their attack latency was > 300 sec. Behavioural data The average attack latency times for each line are shown in Fig. 5.1. As expected, the mice selected for aggression attacked much faster than the non-aggressive ones, as revealed by a highly significant “type” effect (F(1,65)=364.31, p<0.001). The effect was not of the same magnitude in all the strains, as shown by a “selection” effect (F(2,65)=10.07 p<0.001) and a “selection*type” interaction effect (F(2,65)=12.39, p<0.001). However, post hoc analyses show that in all the three selection breeding programs, the aggressive line was significantly faster than the non-aggressive line (SAL vs. LAL, t=–38.44, p<0.001; TA vs. TNA, t=–3.51, p<0.01; NC900 vs. NC100, t =–56.04, p<0.001). From all the scored behaviours, eight categories were analyzed and the results are summarized in Fig. 5.1. A two-way ANOVA revealed significant effects of “type” on non-social exploration (F(1,64)=9.21, p<0.01), social exploration (F(1,65)=76.85, p<0.001), offense (F(1,65)=279.55, p<0.001), withdrawal (F(1,65)=5.69, p<0.05) and defense (F(1,65)=5.09, p<0.05). The aggressive mice derived from the three different selection breeding programs exhibited similar behavioural characteristics when confronted with an intruder in their home-cage. They spent on average 40-50 % of the total test-time in offensive aggression, 20–30% in non-social exploration and much less time in body care, inactivity and withdrawal. The non-aggressive animals spent on average 40% of the confrontation in social exploration, 40% in non-social exploration and little time in body care, inactivity and defense. 5-HT and 5-HIAA concentration in PFC The results of the HPLC on the PFC samples are shown in Fig. 5.2. Aggressive mice had significantly lower levels of 5-HT and 5-HIAA in the prefrontal cortex, compared to the non-aggressive ones, as shown by the two-way ANOVA (“type” effect: F(1,65)=18.84, p<0.001). A highly significant “selection” effect (F(2,65)= 10.93, p<0.001) performed on the 5-HT and on the 5-HIAA data. Between the 106
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
100
total observation time (%)
80
ALT (sec)
600
SAL TA NC900 LAL TNA NC100
400 200
60
0
SAL TA NC LAL TNA NC 900 100
40
20
0
non social social body care drink/eat inactivity offence withdrawal defence exp. exp.
Figure 5.1 Behavioural phenotype (main figure) and average attack latency time (top-inset) of SAL (n=17), LAL (n=14), TA (n=8), TNA (n=9), NC900 (n=8) and NC100 (n=9) mice during a 5-minute resident-intruder confrontation (see text for explanation). Data are represented as means ± SEM. ** type effect: aggressive vs. non-aggressive different at p<0.001, ** type effect: aggressive vs. non-aggressive different at p<0.01, * type effect: aggressive vs. nonaggressive different at p<0.05.
selection programs, the Dutch mice had lower serotonin than the Finnish (Tukey HSD=–200.96, p<0.001) and the American mice (Tukey HSD=–288.42, p<0.001) and lower 5-HIAA than the American mice (Tukey HSD=–81.71, p<0.001). Since no significant “selection*type” interaction effect was found neither in the 5-HT nor in the 5-HIAA data, we conclude that the differences between aggressive and non-aggressive mice were robust independently on the selection breeding program. No significant effects were found in the 5-HIAA/5-HT data. Pharmacological challenges VEHICLE RESPONSE As shown in Fig. 5.3 and Fig. 5.4, an injection of distilled water provoked in all the animals a pronounced stress-induced hyperthermia, revealed as a time effect in the ANOVA (SAL vs. LAL in light phase: F(12,180)=167.3, p<0.001; SAL vs. LAL in dark phase: F(12,180)=15.08, p<001; TA vs. TNA: F(18,180)=9.3, p<0.001; NC900 vs. NC100: F(18,180)=3.7, p<0.001). Within the Dutch selection program, in the light phase SAL mice showed an enhanced hyperthermia compared to LAL mice (type effect: F(1,57)=16.77, p<0.001) with a peak body temperature after 15 107
CHAPTER 5
5-HT (ng/g tissue)
120
800
400
0
5-HIAA (ng/g tissue)
400 300 200 100 0
5-HIAA/5-HT
0.4 0.3 0.2 0.1 0.0 SAL
TA
NC900 LAL
TNA NC100
Figure 5.2 Amounts of serotonin (5-HT), its metabolite (5-HIAA) and its metabolic ratio (5HIAA/5-HT) in the prefrontal cortex of mice selected for high (SAL, n=17; TA, n=8; NC900, n=8) and low aggressiveness (LAL, n=14; TNA, n=9, NC100, n=9). * type effect: aggressive vs. non-aggressive different at p<0.001.
minutes, with SAL temperature higher than LAL (t=–3.7, p<0.001). The same result is seen in the dark phase (type effect: F(1,15)=11.65, p<0.01), with SAL responding more than LAL (peak temperature after 15 minutes: t=–3.6, p<0.01). TA mice responded more than TNA (type effect: F(1,10)=7.62, p<0.05). NorthCarolina mice showed a similar result, with a higher hyperthermia in the aggressive animals (NC900) compared to the non-aggressive ones (NC100) (type effect: F(1,10)=14.39, p<0.01). 108
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
39 38 37 36 35
SAL (light phase)
LAL (light phase)
SAL (dark phase)
LAL (dark phase)
TA
TNA
NC900
NC100
34 39 38 37
body temperature (°C)
36 35 34 39 38 37 36 35 34 39 38 37 36 35 34 -60 -40 -20
0
20
40
60
80 100 120 -60 -40 -20
0
20
40
60
80 100 120
time (min)
Figure 5.3 Body temperature (observed data) of SAL, LAL, TA, TNA, NC900, NC100 mice in response to vehicle (distilled water), S-15535 and 8-OH-DPAT. Data were obtained by means of a telemetry system. Plotted values on the left of the line indicate the measurements before the injection.
109
CHAPTER 5
39
SAL
A
SAL
B
LAL
LAL
38
body temperature (°C)
37 36 35 39
TA
C
NC900
D
TNA
NC100
38 37 36 35 -60
-40 -20
0
20
40
60
80
-60
-40 -20
0
20
40
60
80
time (min)
Figure 5.4 Body temperature response of mice (closed symbol: aggressive line, open symbol: non-aggressive line) injected with distilled water (5ml/kg). Graphs represent: A) SAL (n=30) and LAL (n=29) (light phase); B) SAL (n=9) and LAL (n=8) (dark phase); C) TA (n=6) and TNA (n=6); D) NC900 (n=6) and NC100 (n=6). Data were obtained by means of a telemetry system. Plotted values on the left of the line indicate the measurements before the injection.
S-15535 RESPONSE The response to an injection of S-15535 is illustrated in Fig. 5.3 (observed values) and Fig. 5.5, in which body temperature was expressed relative to the vehicle. A decrease in the drug/vehicle ratio was observed in all the lines (time effect: SAL vs. LAL, light phase: F(12,264)=12.62, p<0.001; SAL vs. LAL dark phase: F(12,180)=4.57, p<0.001; TA vs. TNA= F(12,120)=3.86, p<0.001; NC900 vs. NC100: F(12,120)=5.17, p<0.001). A repeated measures ANOVA did not show significant effects on SAL/LAL values in the light phase, but a tendency to a more rapid decrease in SAL than LAL mice was observed, which led to lower values in SAL than in LAL at t=10 (p=0.06). In the dark phase, a “type*time” interaction effect was found (F(12,180)=2.8, p<0.01), and SAL had lower values than LAL already after 5 minutes (t=2.59, p<0.05). No significant difference was found between TA and TNA. No significant difference was found in the NC-lines. 110
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
1.02
1.01 1.00
1.00
0.99 0.98
drug/vehicle
0.98 0.97
SAL LAL
A
0.96
0.96
SAL LAL
B
1.01
1.01
1.00
1.00
0.99 0.99
0.98
0.98
TA TNA
C 0.97 0
10
20
30
40
50
0.97 60
0.96
NC900 NC100
D 0
10
20
30
40
50
60
time after injection (min)
Figure 5.5 Body temperature response of mice (closed symbol: aggressive line, open symbol: non-aggressive line) injected with S-15535 (10 mg/kg). Graphs represent: A) SAL (n=30) and LAL (n=29) (light phase); B) SAL (n=9) and LAL (n=8) (dark phase); C) TA (n=6) and TNA (n=6); D) NC900 (n=6) and NC100 (n=6). Data were obtained by means of a telemetry system and corrected for the vehicle response. t=0 is the average value of the 60-min recordings prior to the injection.
8-OH-DPAT RESPONSE The response to an injection of 8-OH-DPAT is illustrated in Fig. 5.3 (observed values) and Fig. 5.6, in which body temperature was expressed relative to the vehicle. A marked decrease in drug/vehicle ratio was observed in all the animals (time effect: SAL/LAL, light phase, 0.25 mg/kg: F(12,480)=51.34, p<0.001; SAL/LAL, light phase, 0.50 mg/kg: F(12.252)=31.03, p<0.001; SAL/LAL, dark phase, 0.25 mg/kg: F(12,180)=12.82, p<0.001; TA/TNA, 0.25 mg/kg: F(12,120)= 17.2, p<0.001; CA/CNA, 0.25 mg/kg: F(12,120)=14.46, p<0.001). In the light phase, the response to the 8-OH-DPAT (0.25 mg/kg) was faster in SAL than in LAL when the dosage was higher (0.50 mg/kg) (“type*time” effect: F(12,252)= 2.8, p=0.001). The effect was consistent in the dark phase at lower dose (0.25 mg/kg) (“type*time” effect: F(12,180)=1.9, p<0.05). Regarding the TA/TNA selection, no difference in the speed of the response was found, but an overall higher 111
CHAPTER 5
response in the aggressive line, compared to the non-aggressive one (“type” effect: F(1,10)= 4.8, p=0.05), reaching the maximum difference after 20 minutes (t= –3.2, p=0.01). The response of NC900 mice did not differ significantly from the response of NC100 mice.
1.02
1.04
8-OH-DPAT (0.25 mg/kg) light phase
8-OH-DPAT (0.25 mg/kg) dark phase
1.02
1.00
1.00
0.98
0.98 0.96
0.96
0.94
SAL LAL
A
0.94
SAL LAL
B
0.92
0.92 8-OH-DPAT (0.50 mg/kg) ilight phase
drug/vehicle
1.00 0.98 0.96 0.94 0.92
SAL LAL
C
0.90
1.04
8-OH-DPAT (0.25 mg/kg)
1.02
1.02
1.00
1.00
0.98
0.98
0.96
0.96
0.94 0.92 0
0.94
TA TNA
D 10
20
30
40
50
8-OH-DPAT (0.25 mg/kg)
60
0.92
NC900 NC100
E 0
10
20
30
40
50
60
time after injection (min) Figure 5.6 Body temperature response of mice (closed symbol: aggressive line, open symbol: non-aggressive line) injected with 8-OH-DPAT (0.25-0.50 mg/kg). Graphs represent: a) SAL (n=22) and LAL (n=20) (0.25 mg/kg light phase); b) SAL (n=11) and LAL (n=12) (0.25 mg/kg dark phase); c) SAL (n=9) and LAL (n=8) (0.50 mg/kg, light phase); d) TA (n=6) and TNA (n=6) (0.25 mg/kg); e) NC900 (n=6) and NC100 (n=6) (0.25 mg/kg). Data were obtained by means of a telemetry system and corrected for the vehicle response. t=0 is the average value of the 60-min recordings prior to the injection.
112
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
DISCUSSION Compared to non-aggressive mice, excessively aggressive mice were found to have: 1) enhanced 5-HT1A receptor sensitivity; 2) low amounts of serotonin (5HT) and its main metabolite, 5-HIAA, in the prefrontal cortex. When considering all the attacking mice, independent the type of selection program and the strain used for the breeding, serotonin in the prefrontal cortex, the response to the 5HT1A autoreceptor agonist and offensive/impulsive offensive/impulsive aggression are correlated, as shown in Table 5.1. The amount of serotonin in the prefrontal cortex is low in the aggressive mice with higher sensitivity of the 5HT1A autoreceptor. This suggests that the 5-HT1A autoreceptor inhibition of serotonergic neurons at the level of the raphe nuclei is a major trait characteristic in highly aggressive individuals. To measure the sensitivity of the 5-HT1A receptor we used pharmacological tools, namely the selective 5-HT1A agonists S-15535 and 8-OH-DPAT. The reference values used in this analysis were the data obtained in response to a vehicle injection. The mice used in our study already showed a pronounced strain difference in the hyperthermic response to the vehicle. Interestingly, within each selecTable 5.1 Correlation matrix between the variables: Dmax (S15535), AUC (S15535), 5-HT, 5HIAA, 5-HIAA/5-HT ratio, ALT, and Offense, in the attacking mice from all the lines (n=25). Dmax
Dmax AUC 5-HIAA 5-HT 5-HIAA/5-HT ALT Offense
r p r p r p r p r p r p r p
AUC
5-HIAA
5-HT
5-HIAA/ 5-HT
ALT
1 .857** .000 .002 .991 .314 .126 -.539** .005 .291 .158 -.379 .062
1 -.003 .988 .306 .137 -.526** .007 .384 .058 -.411* .041
1 .819** .000 .211 .312 .539** .005 -.523** .007
1 -.378 .063 .536** .006 -.507** .010
1 -.086 .683 .026 .901
1 -.680** .000
** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).
113
CHAPTER 5
tion program, all the highly aggressive mice showed higher stress-induced hyperthermia than the low aggressive ones. A rise in body temperature is typically observed in response to stress and is primarily mediated by the autonomic nervous system, particularly with activation of the sympatho-adrenal-medullary branch (Olivier et al., 2003). In rodents, the release of noradrenaline from sympathetic terminals in the brown adipose tissue and the sympathetic-driven cutaneous vasoconstriction mediate, respectively, enhanced heat generation and reduced heat dissipation, resulting in increased body temperature (Morrison, 2004). Thus, based on the thermal response to stress, it can be concluded that all three aggressive lines of mice show higher sympathetic activation during stress. This corroborates findings in other animals such as rats and humans, in which a higher sympathetic activation during stress response was observed in aggressive individuals (Arnett, 1997; Sgoifo et al., 2005). Furthermore, this result is consistent with the characteristics of the hostile-impulsive-uncontrolled human type of aggression as defined by Ramirez and Andreu (Ramirez and Andreu 2006), and may be used to experimentally explore this human condition. As a measure of presynaptic 5-HT1A autoreceptor sensitivity, we used the S15535 attenuation of stress-induced hyperthermia, a method described previously in rats (de Boer et al., 2000). S-15535 behaves as a full agonist at the presynaptic 5-HT1A autoreceptor and as a weak partial agonist/antagonist at postsynaptic 5HT1A receptors (Millan et al., 1993b; de Boer et al., 2000). In line with its action on somatodendritic 5-HT1A receptors, it exhibits anxiolytic (blockade of StressInduced Hyperthermia (SIH)) and anti-aggressive properties without compromising motor and defensive behaviours (Millan et al., 1997; de Boer et al., 2000). The overall anti-SIH response to this drug was more pronounced in SAL than in LAL mice, particularly in the dark phase, indicating a higher sensitivity of the 5HT1A autoreceptors in this line of aggressive mice. No significant differences were found in the other two genetic selection lines. In order to study the sensitivity of the 5-HT1A postsynaptic receptors, we used 8-OH-DPAT, a selective 5-HT1A full agonist, which causes not only an attenuation of the SIH, but also a dose-dependent pronounced hypothermia below baseline levels (Millan et al., 1993a). However, in our animals, hypothermia appears only after the correction for the vehicle response. The 8-OH-DPAT-induced hypothermia measured by means of biotelemetry has been described in detail in the rat (Zuideveld et al., 2002). However similar telemetry studies in mice are scarce and usually employ rectal temperature measurements that could mask the drug response with the handling/injection stress response, and/or slightly different protocols and/or mice strains (Goodwin et al., 1985; Matsuda et al., 1990; Bill et al., 1991; Li et al., 1999; Gardier et al., 2001; Hedlund et al., 2004). Our baseline temperature values, obtained when the animals were left undis114
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
turbed in their home cage, were much lower than the ones reported in those studies in which a biotelemetry system was not used. After the correction for the vehicle, the hypothermia due to the full 5-HT1A agonist 8-OH-DPAT was more pronounced in SAL vs. LAL and in TA vs. TNA indicating a higher sensitivity of this receptor in vivo in two out of three aggressive lines. Previous studies in rats and mice similarly showed a higher sensitivity of 5-HT1A receptors in aggressive and impulsive individuals (van der Vegt et al., 2001; Schiller et al., 2006). In humans contrasting results have been obtained (Coccaro et al., 1995; Netter et al., 1999; Cleare and Bond 2000; Minzenberg et al., 2006). It has been postulated in the past decades that in rats and humans the 8-OHDPAT-induced hypothermia is a postsynaptically mediated response in rats and humans (O'Connell et al., 1992; Millan et al., 1993a; Blier et al., 2002), while in mice it is presynaptic (Goodwin et al., 1987; Bill et al., 1991; Martin et al., 1992). This is still controversial and the problem remains unsolved. However, in our study we used S-15535 as a powerful tool to elucidate the contribution of the presynaptic 5-HT1A receptor in the regulation of aggressive behaviour. The fact that the decrease in temperature due to the S-15535 was not as pronounced as the one caused by 8-OH-DPAT indicates that the 8-OH-DPAT hypothermia is not merely a presynaptic effect, but probably an additive effect of the pre- and postsynaptic 5-HT1A receptors. In conclusion, 5-HT1A autoreceptors seem to be very important in modulating the behaviour of the Dutch mice. In the Finnish selection program, the postsynaptic 5-HT1A receptor seems to be more important than the presynaptic one in modulating aggressive behaviour. We could not replicate these results in the NC900/NC100 mice, but some considerations about these animals should be made. First, NC100 mice were significantly heavier than NC900 (mean body weight (g) ± s.e.m.: SAL= 21.7 ± 0.5; LAL= 21.6 ± 0.6; TA= 31.6 ± 0.6; TNA= 31.9 ± 0.4; NC900= 36.2 ± 0.9; NC100= 49.3 ± 1.9; Tukey’s test NC900 vs. NC100 different at p<0.001), which seems to be due to more pronounced fat deposits. If this is a sign of some kind of metabolic dysfunction, it could generate a discrepancy in the pharmacodynamics-pharmacokinetics of the drugs injected and it would lead us to erroneous interpretation of the data. Moreover, since we injected the drugs according to the body weight of the animals, the NC100 mice received a higher absolute dose compared to the NC900, which could explain the more pronounced response to both challenges. This topic requires further examination, for instance various metabolic parameters should be checked in these animals, in order to unravel possible confounding variables that might have affected our data. From the behavioural analysis, we can conclude that all the mice from aggressive lines showed high levels of aggressive behaviour (in terms of offensive 115
CHAPTER 5
aggression time) and impulsivity (expressed by attack latency time). The aggressive behaviour in the Turku and North Carolina aggressive mice, which were selected with testing and breeding procedures different from the Dutch ones and from different founder strains (van Oortmerssen and Bakker 1981; Gariepy JL et al., 1996; Sandnabba, 1996), is therefore very stable through generations. It does not require isolation and is comparable to that of the Dutch aggressive mice in the same testing conditions. Surprisingly, the involvement of serotonin differs slightly between strains, not only in terms of the involvement of the pre- and postsynaptic 5-HT1A receptor, but also regarding the amount of serotonin and its metabolite. From the biochemical assay data, the Dutch mice had lower serotonin concentration in the PFC than the Finnish mice and the North-Carolina mice, and lower metabolite concentration than the North-Carolina mice. As a general conclusion, these results indicate that there are strain differences in the relationship between aggressive behaviour and 5-HT1A-receptor system. Aggressive behaviour is a functionally important and adaptive form of social behaviour in many mammalian species, including humans. While human studies generally concentrate on violence as the pathological expression of aggressive behaviour, animal studies are generally based on non-pathological, functional forms of aggression. This discrepancy between normal and pathological aggression may explain the strain differences in our study and the contradictory results on serotonergic functioning in aggression obtained in animal and human studies. Acknowledgements The authors would like to thank Ramon Granneman for the HPLC analysis. All the experiments were approved by the Committee for the Use of Experimental Animals of the University of Groningen (DEC no. 4083).
116
5-HT1A
RECEPTOR SENSITIVITY AND AGGRESSION
117
6
CHAPTER
Changes in serotonin-1A receptor functionality with social experience in mouse lines selected for high and low aggression Doretta Caramaschi, Cees A. Mulder, Sietse F. de Boer, Jaap M. Koolhaas
Published in Physiology & Behaviour (2007) 16:590-601
ABSTRACT Despite the growing evidence confirming a negative relationship between violence and serotonin neurotransmission in the brain, the development of this relationship is still poorly understood. Among the regulatory molecules in the serotonergic neuron, the 5-HT1A auto-receptor have been proposed as key players in this phenomenon. In this study we aimed at the characterization of functional changes in the 5-HT1A receptors along with the development of aggression into violence. We used male mice genetically selected for high and low, i. e. the ShortAttack Latency (SAL) and Long-Attack Latency (LAL), and Turku-Aggressive (TA) and TurkuNon-Aggressive (TNA) mouse lines. Each mouse was surgically implanted with a transmitter for biotelemetry recordings of core body temperature and the functionality of 5-HT1A receptors was assessed injecting S-15535, the preferential pre-synaptic agonist and post-synaptic antagonist for 5-HT1A receptors. The reduction of stress-induced hyperthermia was recorded and used as functional read-out of receptor activation. The pharmacological challenges were performed before and after a 9-trial repeated aggression-test in which the mice from the aggressive lines were expected to reinforce their genetic aggressive predisposition to more violent behaviour. The results show that the repeated aggression-test lead to escalation of aggression levels particularly in the aggressive lines, and, overall, desensitization of 5-HT1A autoreceptors. At the end of the treatment, but not at the beginning, the most violent mice were the ones that responded more to the S-15535 challenge. In summary, repeated aggression tests reinforce aggressive behavioural predispositions while affecting 5-HT1A autoreceptor functionality. In particular, the association between 5-HT1A autoreceptor enhanced sensitivity and violence appears only after its development in individuals genetically predisposed to high aggressiveness.
CHAPTER 6
INTRODUCTION Individuals with aggressive temperament are at a higher risk to engage in violent behaviours than others, particularly if exposed to socioeconomical conditions that reinforce competitive and dominant behaviours (Loeber and Pardini 2008). Gene X environment interactions seem to shape neurobiological substrates that may activate exaggerated aggression and violence (Raine, 2002b; Van Goozen and Fairchild 2008). Towards a neurobiological explanation of violence, low baseline serotonin neurotransmission has been found in individuals with a history of pathological forms of aggressive behaviour, in particular those that lack of inhibitory control, e.g. impulsive aggression (Linnoila et al., 1983; Coccaro, 1989; Mehlman et al., 1994; Higley et al., 1996). Dynamic changes in the serotonin system seem to occur related to social experience that may differ depending on a more or less aggressive temperament (Haller et al., 2006; Caramaschi et al., 2008a). Serotonin (5-HT) neurotransmission is self-regulated by inhibitory feedback mechanisms, i.e. the activation of 5-HT1A, 5-HT1B autoreceptors on the serotonergic neurons and heteroreceptors (Pineyro and Blier 1999). The activation of 5HT1A and 5-HT1B receptors acutely inhibits selectively aggressive behaviours (Olivier et al., 1989; Miczek et al., 1998; de Boer et al., 2000; DE Almeida and Miczek 2002; van der Vegt et al., 2003b), suggesting an important inhibitory role for these receptors in the motivation for and/or execution of aggressive acts. While 5-HT1B receptors have a developmental role in protecting from adult trait aggression, as shown in knock-out mice (Bouwknecht et al., 2001), the developmental role of 5-HT1A receptors still remains controversial and may be more generally related to anxiety (Gross et al., 2002). A causal role for 5-HT1A receptors in the development of highly aggressive and violent personality is yet to be demonstrated mainly for two reasons. First, auto- and hetero- 5-HT1A receptors in different brain areas and at different stages of life might play contrasting roles in the development of behavioural traits. The possibility to dissect these roles using transgenic strategies has only recently become available. This issue can also be addressed using specific pharmacological tools for auto- and hetero- receptors. Secondly, most animal models focus on low to moderate species-typical aggression yet with no signs of violence and psychopathology. This might give different results than animal models aimed at understanding human violent offenders (Haller and Kruk 2006). Male mice genetically selected for high and low aggression display high and low levels of offensive aggression against a docile male intruder in their home cage, respectively. Mice of the SAL (Short Attack Latency) line are considered to be violent (Caramaschi et al., 2008a; Natarajan et al., 2009) because the exhibit 120
5-HT1A RECEPTOR
SENSITIVITY AND DEVELOPMENT OF VIOLENCE
offensive aggression also against females and anaesthetized (inoffensive) intruders. Moreover, their offensive behaviour is not sensitive to social signals from the opponent. The mice of the TA (Turku Aggressive) line on the other hand do not attack females and anaesthetized intruders, yet showing levels of aggression against males similar to those of SAL mice. The mice of the TNA (Turku NonAggressive) line show some signs of offensive aggression to male intruders in their home cage, although their aggression level is much lower than the SAL and TA lines. Finally, mice of the LAL (Long Attack Latency) line rarely show offensive aggression even against a male intruder (Caramaschi et al., 2008a; Natarajan et al., 2009). Hence, these genetic selection lines of mice can be used to delineate the neurobiology underlying high and low levels of aggression, but also to unravel the neurobiological differences between high levels of aggression and violence. These mouse lines differ also in their serotonergic system. SAL mice showed lower serotonin levels than LAL, after a series of resident-intruder tests through which their pathological aggression against females had increased (Caramaschi et al., 2008a). Subsequent experiments in socially experienced SAL males showed that these lower 5-HT levels may be due to enhanced 5-HT1A autoreceptor functionality. Socially experienced TA mice showed enhanced 5-HT1A heteroreceptor functionality compared to TNA (Caramaschi et al., 2007). These data suggest a link between 5-HT1A autoreceptor functionality, brain serotonin levels and violence. However, the role of the 5-HT1A autoreceptor in the neurobiological dynamics accompanying the escalation of aggression into violence is not understood. The aim of this study is to investigate whether the increased 5-HT1A autoreceptor functionality develops with the transition of aggression into violence or is a stable characteristic of the high aggressive individuals independent of the development of violence. In order to do evaluate 5-HT1A autoreceptor in vivo functionality we injected S15535 in mice previously implanted with biotelemetry transmitters and recorded their core body temperature. As function of its agonistic action on the 5-HT1A presynaptic autoreceptors, S-15535 has been shown to reduce the stress-induced rise in body temperature that follows a subcutaneous injection. A stronger reduction of stress-induced hyperthermia would indicate a higher functionality of the 5HT1A autoreceptor. To measure the change in receptor functionality, coupled with the development of violence, we performed the pharmacological challenge in socially naïve animals and after a training of repeated social interactions, which is in practice a repeated winning experience for the aggressive animals.
121
CHAPTER 6
METHODS Animals Male mice from the SAL, LAL, TA, and TNA genetic selection lines were used for this study. The mice were bred at the Biological Center, Haren, the Netherlands. After weaning at 21 days, the mice were kept with their siblings. Before reaching sexual maturation, the mice were housed in male-female pairs to avoid intermale aggression and social isolation in Makrolon type II cages provided with food shavings as bedding material, Envirodry shredder paper for nesting, and a cardboard tube as cage enrichment. Food in the form of rodent pellets (AMII, ABDiets, Woerden, The Netherlands) and water were available ad libitum. The mice were kept at a 12:12 light-dark schedule at 22 ± 2°C. All the experimental procedure were conducted under approval of the Institutional Animal Care and Committee of the University of Groningen, the Netherlands (licence number D4540C), in compliance with the Dutch law on animal experimentation and the European Communities Council Directive of 24 November 1986 (86/609/EEC). The experimental design consisted of the following tests: pre-treatment drug challenges, repeated resident-intruder experience, post-treatment drug challenges. At the end of the experiment the all the mice were euthanized with a mixture of CO2 and O2. Surgery and telemetry setup At the age of three months the mice were surgically implanted with a transmitter TA10ETA-F20 for biotelemetry chronic recordings of core body temperature (Data Sciences Int., St Paul, Minnesota, USA) following a previously developed surgical procedure (Caramaschi et al., 2007). During surgery, the animals were anesthetized with 5% isoflurane/O2/N2O, placed on a Harvard homoeothermic heating pad in order to prevent hypothermia due to the anaesthetics, and maintained under anaesthesia with 2.5% isoflurane/O2/N2O. A transmitter was placed inside the abdominal cavity and one lead was fixed to the xiphoid process of the sternum and the other to the pectoral muscular layer in the right mediastinum. Internal suturing was performed by means of reabsorbable surgical thread, while skin was sutured with silk with a reverse-knot method, in order to prevent chewing by the animal. Natrium penicillin was injected as post-surgery antibiotic treatment. The signal produced by each transmitter was received by an antenna/receiver board (RPC-1, Data Sciences Int., 32 x 22 x 3 cm), placed underneath the animal’s cage. Each receiver was connected to a consolidation matrix (BCM-100), which was in turn connected to a PC data-acquisition and analysis system (Dataquest A.R.T. silver, Data Sciences). The data-acquisition parameters were set for 10-second sampling every 5 minutes on a 24 h basis. 122
5-HT1A RECEPTOR
SENSITIVITY AND DEVELOPMENT OF VIOLENCE
The animals were allowed to recover for at least 14 days after the surgery, during which time their temperature and locomotor activity were monitored. The re-establishment and stability of circadian rhythmicity was a prerequisite for the start of the experiments. Pharmacological challenges Drug challenge tests were performed within the first three hours of the dark phase. The injections were carried out in two consecutive days. Each experimental animal received two subcutaneous injections, which consisted of distilled water (vehicle, 5 ml/kg of body weight) and S-15535 (10 mg/kg). The volume injected was 5-10 ml/kg of body weight. The animals were undisturbed in their home cage before and immediately after the injections on the day of the experiment to avoid any stress-related disturbance in the data. S-15535-3 methanesulfonate [(4-benzodioxan-5-yl)-1-(indan-2-yl)piperazin, lot n. EI798] was provided by Institut de Recherches Internationales Servier, France, and it was dissolved in distilled water (vehicle) at room temperature on the day of the injection. Repeated Resident-Intruder (RRI) paradigm A group of mice previously implanted with transmitter and tested with pharmacological challenges underwent a repeated resident-intruder treatment (RRI group). This consisted of nine male-male resident-intruder (RI) experiences, one each day, carried out at the same time of the day (at the beginning of the dark phase) in a test cage (75 x 29 x 27), where each male had previously been housed with a female, in presence of food and water. Each day, 1 hour before the RI experience, the female partner was removed from the cage. Subsequently, a naive male intruder was placed in the cage and the attack latency, i.e., the time it took the resident to attack the intruder, was scored. The intruder was removed from the experimental cage immediately after the first attack from the resident. If there was no attack, the test was stopped after 5 minutes and a score of 300 seconds was given. In the first (RI1) and in the last resident-intruder experience (RI9), the intruder was left for 5 minutes in the resident cage and a video recording was made for subsequent behavioural analysis. The videos were analyzed with The Observer 5.0 (Noldus Information Technology bv, Wageningen, the Netherlands). From each 5-minute observation the following behavioural states were scored: approach, attack, threat, chase, withdrawal, social exploration, non-social exploration, walk/tail rattle, body care, and inactivity. Furthermore, the number of open wounds on the opponent was counted as a rough estimate of the injurious dimension of the interaction. Another group of male mice previously implanted with transmitter and tested 123
CHAPTER 6
with pharmacological challenges were used as a control group (CTR). They were not subjected to the RRI paradigm, but instead were briefly handled and their female partners removed for the same duration every day as for the RRI group. Data analysis and statistics The attack latency data were analyzed with ANOVA for repeated measurements with “day” (9 levels) as within-subject factor and “strain” (2 levels: Groningen and Turku) and “type” (2 levels: high-aggressive and low-aggressive) as betweensubject factors. Similarly, total duration of the single behaviours scored in the resident-tests were analyzed with ANOVA for repeated measurements with “day” (2 levels: day1 and day9) as within-subject factor and “strain” (2 levels: Groningen and Turku) and “type” (2 levels: high-aggressive and low-aggressive) as between-subject factors. The number of wounds inflicted to the opponents was analyzed using ANOVA for repeated measurements with “day” (2 levels: day1 and day9) as within-subject factor and “strain” (2 levels: Groningen and Turku) and “type” (2 levels: high-aggressive and low-aggressive) as between-subject factors. Core body temperature measurements were extracted from telemetry system starting one hour before the injection took place until 90 minutes after the injection. The 60 minutes before the injection were averaged and considered as baseline. The temperature response to S-15535 injection was divided by the temperature response to the vehicle injection, since the vehicle alone was previously shown to induce a stress-induced hyperthermia (Caramaschi et al., 2007). From the drug/vehicle ratios, the maximum deviation from baseline, and the area under the curve were calculated. They were then analyzed statistically using ANOVA for repeated measurements, in which “test” (2 levels: pre and post) was the within-subject effect, while “group” (2 levels: rri and ctr), “strain” (2 levels: Groningen and Turku), and “type” (2 levels: high-aggressive and low-aggressive) were the between-subject effects. As post hoc, in case of significant within-subject interaction effects separate analyses were performed in the “rri” and in the “ctr” groups, to identify in which subgroup of subjects the effect was statistically significant. In case of significant “strain*type” interaction effect Tukey’s multiple comparisons were performed. To answer the question of the presence of a relationship between 5-HT1A receptor and aggression/violence in naïve animals or only after the repeated resident-intruder experience, bivariate correlations were investigated between dmax and AUC values pre- and post- treatment and attack duration, attack latency, and number of wounds at day 1 and day 9, respectively.
124
5-HT1A RECEPTOR
SENSITIVITY AND DEVELOPMENT OF VIOLENCE
RESULTS Resident-Intruder tests Overall, aggressive mice had shorter attack latencies (figure 6.1) than low-aggressive mice (type: F(1,21)=39.87, p<0.001), while Turku had almost significantly longer attack latencies than Groningen mice (selection: F(1,21)=4.03, p=0.058). Attack latency was significantly reduced along with the repeated resident-intruder experience (day: F(8,14)=4.19, p=0.010). The change was different depending on the type (day*type: F(8,14)=3.16, p=0.029) and on the selection pair (day*selection: F(8,14)=2.93, p=0.038). Highly aggressive mice had significantly shorter attack latencies than low-aggressive mice from day 3 to day 8. Turku mice had significantly longer attack latencies than Groningen mice only at day 1. The total duration of the offensive behaviours is depicted in figure 6.2. Attack duration increased overall from day 1 to day 9 (F(1,21)=6.0, p=0.023). Throughout the experiment, aggressive mice attacked more than low-aggressive mice (F(1,21)=22.13, p<0.001). Moreover Groningen mice attacked significantly more than Turku (F(1,21)=5.46, p=0.03). Threat duration was significantly higher in high- than low-aggressive lines (F(1,21)=7.79, p=0.011). Chase increased significantly from day 1 to day 9 (F(1,21)=4.60, p=0.044). The total duration of non-offensive behaviours is summarized in table 6.1. Walk/tail rattle was significantly more in the high- than low-aggressive lines (F(1,21)=13.55, p=0.002). Social exploration was significantly higher in the Turku lines (F(1,21)=7.50, p=0.012). Approach did not differ significantly across any
SAL LAL TA TNA
350
attack latency (sec)
300 250 200 150 100 50 0 0
1
2
3
4
5
6
7
8
9
10
day
Figure 6.1 Attack latencies (group means ± SEM) of the SAL, LAL, TA and TNA mice during the repeated resident-intruder paradigm. Statistical details are discussed in the text.
125
CHAPTER 6
attack % observations
% observations
40 30 20 10
threat
16
50
0 4
12 8 4 0 16
3
day 1 day 9
wounds number
% observations
chase
2 1 0
12
opponent's wounds
8 4 0
SAL
LAL
TA
TNA
SAL
LAL
TA
TNA
Figure 6.2 Time spent in offensive behaviour (attack, threat, and chase) and number of wounds inflicted on the opponent by the SAL, LAL, TA and TNA mice during the first and the last resident-intruder tests. All data are expressed as group means ± SEM. Statistical details are discussed in the text.
condition tested. Withdrawal was significantly more in the aggressive lines (F(1,21)=13.44, p=0.001) and in the Groningen lines (F(1,21)=7.08, p=0.015). Non-social exploration was overall lower in the highly aggressive lines (F(1,21)=10.63, p=0.004) and it changed significantly from day 1 to day 9 depending on the type (F(1,21)=4.86, p=0.039), which resulted in a significantly longer duration in low- compared to high-aggressive mice at day 9 (t(24)=–3.9, p=0.001). There were no significant differences in body care and inactivity. The number of wounds inflicted to the opponent was overall significantly higher in the high-aggression lines (F(1,21)=29.16, p<0.001), in the Groningen selection (F(1,21)=21.29, p<0.001), and more specifically in the SAL line (F(1,21)=21.29, p=0.044). The change from day 1 to day 9 was significant depending on the type (ANOVA “day*type”: F(1,21)=18.98, p<0.001) and on the mouse line (ANOVA “day*type*selection”: F(1,21)=6.17, p=0.022). Tukey’s post hoc comparisons show that SAL inflicted significantly more wounds than TA (difference=6.86, p=0.002) and TNA (HSD=8.36, p=0.001) already at the day 1, while LAL inflicted a similar amount of wounds, significantly higher than that of TNA (HSD=5.50, p=0.030). From day 1 to day 9 the amount of wounds inflicted 126
5-HT1A RECEPTOR
SENSITIVITY AND DEVELOPMENT OF VIOLENCE
Table 6.1 Total duration (sec) of non-aggressive behaviours in the 1st and 9th resident-intruder tests. Soc. exp.
RI1 SAL LAL TA TNA
RI2 SAL LAL TA TNA
Approach Withdraw
Walk/ Non social Body care Inactivity tail rattle exp.
Mean SEM Mean SEM Mean SEM Mean SEM
3.12 1.32 15.88 6.93 25.90 4.87 31.88 7.60
3.30 1.17 2.15 0.80 2.71 0.90 0.80 0.52
5.34 1.20 2.16 0.83 1.78 0.69 1.54 0.70
4.28 1.14 1.89 0.95 4.30 1.75 0.61 0.61
31.66 8.44 40.88 10.50 34.74 8.85 40.18 10.34
3.30 1.92 0.72 0.59 6.48 3.15 4.43 0.88
21.59 5.38 26.66 15.03 8.92 2.53 11.44 6.72
Mean SEM Mean SEM Mean SEM Mean SEM
0.11 0.08 22.16 4.16 21.43 12.84 21.07 6.43
2.53 0.56 3.45 1.35 2.04 0.86 1.33 0.80
4.30 0.58 0.93 0.58 2.58 1.24 0.53 0.27
2.59 0.59 0.51 0.42 4.34 1.94 0.74 0.62
12.67 3.01 54.35 8.87 17.39 6.31 46.39 9.70
3.69 2.33 3.03 1.67 6.09 3.51 6.21 2.21
19.29 2.28 7.54 2.12 9.27 3.30 6.72 2.66
increased significantly in the SAL (paired t=-2.89, p=0.028) and the TA lines (paired t=-3.32, p=0.021), while in LAL (paired t=2.8, p=0.038) it decreased significantly and in the TNA it did not change. Consequently at day 9 SAL mice inflicted significantly more wounds than all the other lines (Tukey’s HSD: SAL–LAL=11.38, p<0.001; SAL–TA=9.04, p<0.001; SAL–TNA=12.88, p<0.001). Serotonin-1A receptor functionality Figures 6.3 shows an example of the core body temperature response to injection of vehicle and S-15535. The dmax and AUC are summarized and depicted in figure 6.4. The maximum deviation from baseline, dmax, showed a significant effect of strain (F(1,36)=4.94, p=0.033). The Groningen mice responded more to the challenge overall. Similarly, the overall response measured as area under the curve showed a significant effect of “strain” (F(1,36)=8.83, p=0.005). The Groningen mice responded more to the challenge overall. Moreover, there was a significant effect of “group” (F =5.67, p=0.023). The RRI mice responded less to the challenge overall. There were no significant interaction effects. 127
CHAPTER 6
core body temperature (°C)
40 vehicle S-15535
39
38
37
36 0
20
40
60
80
time after injection (min)
Figure 6.3 Example (mouse nr. 5, SAL, before repeated resident-intruder) of the body-temperature response to a subcutaneous injection of distilled water (stress-induced hyperthermia) and to S-15535 (10 mg/kg).
Correlation between 5-HT1A receptor functionality and aggression Table 6.2 summarizes the correlations between the response to S15535 (expressed as dmax and AUC) and the aggression levels (attack latency, attack duration and wounds). Before the repeated resident-intruder experience, attack duration was negatively correlated with the response to S-15535 while attack latency was positively correlated with the response to S-15535. At the end of the experiment, attack duration and latency remained correlated with the response to S-15535. Additionally, a highly significant negative correlation was found between the amount of wounds and the response to S15535, as shown in Figure 6.5.
DISCUSSION The present study shows that repeated winning experience enhances aggressive behaviour, in particular in the high aggressive males. Using the S15535-induced reduction of stress-hyperthermia as a pharmacological index of 5-HT1A autoreceptor sensitivity, it was demonstrated that a repeated social experience affects the sensitivity of the receptor in the direction of desensitization. However, the mice that developed violent behaviour, beyond increasing the attack duration time, were the ones with more sensitive 5-HT1A receptors after the experience. The present study confirms the relationship between 5-HT1A autoreceptor func128
5-HT1A RECEPTOR
SAL
dmax (drug/vehicle ratio)
0.00
LAL
TA
SENSITIVITY AND DEVELOPMENT OF VIOLENCE
TNA
-0.02
-0.04
-0.06 CTR-PRE CTR-POST RRI-PRE RRI-POST
-0.08
-0.10
AUC (drug/vehicle ratio)
1 0 -1 -2 -3 CTR-PRE CTR-POST RRI-PRE RRI-POST
-4 -5
Figure 6.4 Overall response to S-15535 corrected for vehicle response. Data are expressed as group means ± SEM of the maximum deviation form baseline (dmax) and of the area under the response curve (auc). * significant effect of “strain” or “group” at p<0.05.
Table 6.2 Bivariate correlations (Pearson coefficient) between 5-HT1A receptor sensitivity before the repeated resident-intruder experience and aggression levels at 1st (RI1) and at the 9th (RI9) resident-intruder test. Wounds RI1 RI9 dmax AUC
Pearson p-value Pearson p-value
-0.345 0.116 -0.281 0.206
-0.713 0.000 -0.518 0.014
Attack duration RI1 RI9 -0.492 0.020 -0.298 0.179
-0.471 0.027 -0.338 0.124
Attack latency RI1 RI9 0.430 0.046 0.308 0.163
0.455 0.033 0.283 0.202
129
CHAPTER 6
0.00
dmax (ratio)
-0.02 -0.04 -0.06 SAL LAL TA TNA
-0.08 -0.10 0
5
10
15
20
number of wounds
Figure 6.5 Scatterplot of the maximum deviation from baseline (dmax) after injection of S15535 after the repeated resident-intruder experience plotted against the number of wounds inflicted to the opponent during the ninth confrontation. The line is the best fitted regression line.
tionality and social behaviour. Additionally, it shows that enhanced 5-HT1A autoreceptor functionality enhanced sensitivity is related to pathological rather than species-typical aggression. Overall, the dynamics of 5-HT1A auto-receptors seems to follow that of aggression escalation and depends on the genetic predisposition to high or low aggressive temperament. In line with previous studies (Miczek et al., 1998; van der Vegt et al., 2001; Schiller et al., 2006), our study shows that serotonin-1A autoreceptors play a role in aggressive behaviour. In particular they appear to be sensitized in individuals prone to develop pathological aggression. The same social experience desensitized serotonin-1A autoreceptors of individuals that escalated less their aggression, while it did not affect those with low aggressive genetic background. The differences in the resulting phenotype and the genetic predisposition are therefore related to the serotonin-1A autoreceptor dynamics and might have interesting implications on the fate of serotonin-1A autoreceptors and temperamental differences. As a consequence of this differential dynamics one might also expect differential effects on the serotonergic raphe nuclei. As the main effect of the 5-HT1A autoreceptors is inhibitory, the raphe nuclei would become less activated in terms of production and release of serotonin. However, serotonin is released both tonically and phasically and it is not clear yet which of these release patterns will me most affected. 130
5-HT1A RECEPTOR
SENSITIVITY AND DEVELOPMENT OF VIOLENCE
Previous studies have tried to understand the functionality of the 5-HT1A autoreceptors with the use of 8-OH-DPAT. However, the specificity of the challenge with such a full agonist is not enough to distinguish receptors subpopulations. This full agonist was thought to determine hypothermia in mice through autoreceptors, but unfortunately contradictory reports in mice and other species keep the question still open. Studies on the receptor expression were not able to find a link between the autoreceptors and aggression, or rather they showed a link between 5-HT1A heteroreceptors and high trait-aggression (Korte et al., 1996; van der Vegt et al., 2001; Veenema et al., 2005a). And other studies showed 5HT1A heteroreceptor expression and/or blunted functionality in aggressive individuals presenting different subtypes of aggression, e.g. in Alzheimer patients and reactive/defensive rats (Vitiello and Stoff 1997; Lai et al., 2003; Popova et al., 2005). Although it has been proposed that aggressive individuals differ in the subtypes of aggression they display (Natarajan et al., 2009), detailed information about the central serotonergic neurotransmission in these subtypes has not been studied to our knowledge. The main conclusion of this study is that individuals with differential behavioural characteristics in aggression, whether genetically obtained or as a product of genetic predisposition and social exposure, show distinct changes in 5-HT1A auto-receptors. These findings might have very important consequences for the functioning of raphe nuclei. This study also suggests that a better behavioural characterization of subtypes of aggression into less and more pathological ones might help disentangle otherwise controversial findings.
131
7
CHAPTER
Tryptophan-free diet lowers fronto-cortical serotonin levels with no effect on mouse aggression Doretta Caramaschi, S. F. de Boer, C. A. Oosterhof, J.M. Koolhaas
ABSTRACT The relationship between low serotonergic neurotransmission and high impulsive aggression has been documented in several human and non-humans studies. Yet it is not always replicated and it is limited to correlation relationships rather than causal effects. A major confounding problem is the heterogeneous definition of aggression and the lack of animal models of more violent forms of aggression known in human beings. In this study we investigated the effects on intermale aggression of lowering brain serotonin levels by removing its essential precursor tryptophan from the diet of mice genetically selected for high and low aggression, e.g. Short Attack Latency, SAL, and Long Attack latency, LAL, respectively. Previous studies have shown that the behaviour of the SAL mice is extreme and pathological, while that of the LAL mice is extremely docile in a social interaction and vulnerable to depression/anxiety-like behavioural disturbances. Moreover, the mouse lines differ in their serotonergic system. The results show that, while tryptophan depletion reduced significantly prefrontal cortex serotonin levels and body weight, and increased significantly plasma corticosterone levels in both mouse lines, it did not cause any significant changes in their intermale behavioural repertoire. We conclude that a reduction of brain serotonin levels is not enough to change aggressiveness levels of individuals genetically predisposed for high and low aggressiveness. Instead, the correlation between serotonin levels and aggression could be rather the consequence of an upstream common mechanism that leads to both low serotonin levels and high aggressiveness.
CHAPTER 7
INTRODUCTION Serotonin is indisputably the most widely studied neurotransmitter in relation to aggressive behaviour. The idea of aggressive individuals having low brain serotonergic neurotransmission, i.e. the “serotonin-deficiency hypothesis of aggression”, originated from early studies and has been supported by evidence in human and non-human studies (Maas, 1962; Valzelli, 1982; Lee and Coccaro 2001; Ferrari et al., 2005; Miczek et al., 2007). Low levels of the major serotonin metabolite, 5hydroxyindoleacetic acid or 5-HIAA, which is a measurement of the amount of serotonin released and consequently metabolized, have been found in the cerebrospinal fluid of impulsive aggressive human individuals (Brown and Linnoila 1990). Similarly, serotonergic functioning, measured through fenfluramine challenge, was inversely correlated both with self-reported and laboratory aggression (Coccaro et al., 1996). Studies in mice showed that brain serotonin levels are inversely correlated to aggression (Caramaschi et al., 2007). These data confirm the general idea that high levels of aggression are associated with low levels of brain serotonin. However, it remains to be determined if this serotonin deficiency is a predisposing causal factor for the development of aggression or rather a consequence of other mechanisms that lead both to aggression and to serotonin deficiency. The possible causal relationship between aggression and low serotonin levels has been investigated by damaging the raphe nuclei and by inhibition of serotonin production. Rats in which serotonin neurons were lesioned with 5,7dihydroxytryptamine showed more shock-elicited aggression and muricidal behaviour (Breese and Cooper 1975; Hole et al., 1977; Kantak, 1981; Kantak et al., 1981; Vergnes et al., 1988). Increases in muricide and offensive aggression were also observed when rats or mice were injected with para-chlorophenylalanine, a serotonin-synthesis inhibitor (Miczek et al., 1975; Paxinos et al., 1977; Gibbons et al., 1978; Valzelli et al., 1981; Sewell et al., 1982; Ieni and Thurmond 1985; Albert et al., 1985; Molla-Hosseini, 1985; Vergnes et al., 1988; Keele, 2001). A less invasive method used to inhibit the synthesis of serotonin is through depleting the brain from tryptophan, the essential aminoacid precursor of serotonin (Biggio et al., 1974). In rats, chronic tryptophan depletion increased shockinduced aggression (Kantak et al., 1980b) and mouse-killing behaviour (Gibbons et al., 1979). In humans and non-human primates, tryptophan depletion is obtained giving the subjects a tryptophan-free aminoacid mixture (Bell et al., 2001). Using this method, aggression was acutely increased in women during menstrual phase (Bond et al., 2001), in healthy male subjects (Pihl et al., 1995; Moeller et al., 1996; Bjork et al., 1999) and in male subjects with high trait aggression (Cleare and Bond 1995; Pihl et al., 1995; Bjork et al., 2000; McCloskey et al., 2008). In primates, a tryptophan-free aminoacid mixture 134
TRYPTHOPHAN-FREE
DIET AND AGGRESSION
increased aggression in males (Chamberlain et al., 1987). Although these findings suggest a causative link between lower serotonin levels and aggression, other studies showed an effect in the opposite direction or failed to show a a change in aggression. Reduction of serotonin levels did not affect aggressive behaviour in mice and rats (Conner et al., 1970; Miczek et al., 1975; Kantak et al., 1980a; Vergnes and Kempf 1982; van der Vegt et al., 2003a), or even inhibited aggression (Malick and Barnett 1976; Jones et al., 1976). Similarly, in humans aggression levels did not change or hostility levels even improved following acute tryptophan depletion (Salomon et al., 1994; Pihl et al., 1995; LeMarquand et al., 1999). These conflicting data corroborate the general view of a causal relationship between serotonin and trait aggression. This may be due to the possibility that tryptophan, and consequently serotonin depletion, may be affected only in a certain group of individuals. Indeed, humans scoring high in trait aggression (Pihl et al., 1995; Wingrove et al., 1999; Dougherty et al., 1999a; Dougherty et al., 1999b), showed completely opposite effects compared to individuals with low levels of aggressiveness (Cleare and Bond 1995; Bjork et al., 2000). Moreover, it is still under discussion whether acute tryptophan depletion has an effect at all on the levels of serotonin and on serotonergic neurotransmission (van der Plasse et al., 2007). In addition to the issue of individual variation, it is important to notice that, while serotonin levels are negatively correlated with aggression as a trait characteristic, recent evidence shows that serotonin release is needed for the initiation and/or execution of an aggressive act. In apparent paradox with the serotonindeficiency theory, serotonin seems to be released during a social interaction and aggressive acts (van der Vegt et al., 2003b; Summers et al., 2005). Inhibiting this release through enhanced presynaptic 5-HT1A and 5-HT1B autoreceptors activation reduces rodents’ offensive aggression (Bell and Hobson 1994; Joppa et al., 1997; Miczek et al., 1998; de Boer et al., 2000; DE Almeida and Miczek 2002). If serotonin release is indeed required for the performance of the aggressive act itself, it can be hypothesized that 5-HT depletion in aggressive individuals may lead to reduced serotonin release during a social interaction and consequently reduce their aggression levels. To test the hypothesis that the effects of serotonin depletion on aggression might only be observed in certain individuals, two mouse lines were used that were artificially selected for short and long attack latency, SAL and LAL, respectively, in a resident-intruder context. The mouse lines were generated from a wild population of house mice near Groningen, the Netherlands, and have been shown to represent extreme alternative phenotypes, with SAL being violent and LAL being very docile, although occasionally a few LAL mice attacked very briefly 135
CHAPTER 7
(Caramaschi et al., 2008a; Natarajan et al., 2009). Moreover, these mouse lines differ strongly in serotonergic neurotransmission. SAL and LAL mice that had been exposed to resident-intruder tests, showed inverse relationship between serotonin levels and aggression levels (Caramaschi et al., 2007; Caramaschi et al., 2008a). Furthermore, SAL mice showed increased 5-HT1A auto- and hetero-receptor inhibitory feedback, suggesting an inhibited serotonergic neurotransmission when 5-HT1A receptors are activated (van der Vegt et al., 2001; Caramaschi et al., 2007). In view of this differentiation in both aggression and serotonergic neurotransmission, we subjected socially naïve mice of the SAL and LAL lines to a chronic tryptophan-free diet and studied its effects on their intermale offensive aggression in the resident-intruder test, compared to mice fed with a balanced diet containing tryptophan. If a certain level of serotonin neurotransmission is needed for the execution of aggression, we expected that the serotonin-depleted SAL mice would reduce their aggression levels. If low serotonin tissue levels are a causal mechanism for the execution of aggression, we expected that a reduction of serotonin tissue levels due to tryptophan-depleted diet in LAL mice would enhance their aggressiveness. After completion of the behavioural experiments prefrontal cortex samples were taken and analyzed for 5-HT content. In order to gain more insights in the physiology underlying the hypothetical interaction between chronic trypthophan-deficiency and aggressive behaviour, we examined also the effects of the diet on body weight and plasma corticosterone. Based on previous reports (D'Souza et al., 2004), we expected a decrease in body weight and an increase in corticosterone levels in the tryptophan-depleted mice.
MATERIALS AND METHODS Animals and dietary manipulation The subjects of this study were male mice (n=32) from the SAL and LAL lines obtained from our breeding colonies. The mice were kept in familiar groups until 21 days of age, then they were kept in unisexual groups of siblings until ca. 40 days of age, when each male was co-housed with a female, in order to avoid social isolation and male-male competition. Each male-female pair was housed in a cage in which the bedding was never changed or altered, in order to keep the odours of the male residents. The bedding consisted of wood shavings and envirodry material for nest building. The room conditions were maintained at 22 ± 2 °C on a 12:12 light-dark schedule. The mice had food and water available ad libitum. A few days (3-6) before the start of the experiment, all the mice started being fed with the control diet (Laboratori Piccioni, Segrate, MI, Italy), in order to habituate to the new food. At the start of the experimental manipulation, half of 136
TRYPTHOPHAN-FREE
DIET AND AGGRESSION
the mice started being fed with tryptophan-free diet, TRP-free (Laboratori Piccioni, Segrate, Italy), while the other half continued with a control diet, CTR, containing tryptophan (Laboratori Piccioni, Segrate, Italy). The TRP-free diet (Fadda et al., 2000) had the following composition per 100 g of food: zein (protein extracted from corn) 2%, ammonium citrate 2.34%, gelatine 3%, glycine 3%, lysine HCl 1.06%, histidine 0.42%, methionine 0.53%, phenylalanine 0.71%, leucine 1.42%, isoleucine 0.62%, threonine 0.97%, valine 1.11%, Hegsted salt 4%, choline chloride 0.15%, maize oil 10%, sucrose 68.38%, plus complete vitamin integration 0.25% (thiamine HCl, riboflavin, pyridoxine HCl, nicotinic acid, calcium panthotenate, folic acid, biotin, vit. B12, A, D3, E, K). The CTR diet had identical composition, except that 0.7% of TRP was added in place of an equivalent amount of sucrose (Fadda et al., 2000). The body weight of the mice was checked every 1-2 days to monitor potential physiological changes due to the diet. All the experimental procedures had been approved by the Institutional Animal Care University Committee of the University of Groningen, the Netherlands (DEC protocol n. D4540B), and were in accordance with the Dutch Law on Animal Experimentation of 12 January 1977 (reviewed in 2003) and with the European Communities Council Directive of 24 November 1986 (86/609/ EEC). Resident-intruder tests Prior to the start of the dietary manipulation, the mice were subjected to the Attack Latency Test (ALT). Briefly, the mice underwent resident-intruder tests for three consecutive days, in order to habituate to the test and remove any novelty effect. On the first day, the mice were only presented with an intruder in their home-cage until they attacked it, or for 5 minutes in case of no attack. On the third day, the intruder was left in the cage for 5 minutes for all the mice and a video recording was made. This test accounted as Resident-Intruder 1 (RI1), and was used as baseline. To investigate the effects of the dietary manipulation on aggressive behaviour, a second and a third resident-intruder test was performed after one and two weeks of dietary manipulation, respectively. These RI2 and RI3 tests were performed using the same protocol of RI1. All the resident-intruder tests were performed in the beginning of the dark phase, in temporary absence of the female, and against an unknown docile opponent of the albino strain A/J. From the videotapes of the RI1, RI2, and RI3 the behaviour of the resident mice was scored using The Observer 5.0 (Noldus Information Technology bv, the Netherlands). Mutually exclusive behavioural states scored were attack (bite, keep-down, charge), threat (lateral threat, tail rattling), chase, retreat (withdrawal from attack), social exploration (sniffing, allogrooming), non-social exploration (sniffing, climbing, digging), body-care (drink/eat, self-grooming), inactivity (immobility, rest, freezing). 137
CHAPTER 7
Brain-tissue preparation and monoamine determination with HPLC One day after the RI3 test, the animals were sacrificed under CO2 anaesthesia, decapitated and their brains removed. The prefrontal cortex was dissected from each brain, frozen in liquid nitrogen and stored at –80 ºC. The PFC samples were homogenised in 1 ml 0.1 M perchloric acid for 60 seconds and centrifuged at 14,000 rpm for 10 min at 4ºC. The supernatant was removed and 100 µl were injected into a HPLC (High-Performance Liquid Chromatography) column (Gemini C18 110A, 150 x 4.60 mm, 5 u, Bester) connected to a detector (analytical cell: ESA model 5011, 0.34 V). The mobile phase consisted of 62.7 mM Na2HPO4, 40.0 mM citric acid, 0.27 mM EDTA, 4.94 mM HSA and 10% MeOH (pH 4.1). Known amounts of monoamines were run in parallel for standardisation. Monoamine levels were calculated as ng/g tissue. Corticosterone assay During the decapitation, trunk blood was collected in chilled tubes containing EDTA for determination of corticosterone levels. Blood samples were centrifuged at 2600 g for 10 min at 4°C. Plasma samples were stored at –20°C until assayed. Plasma corticosterone was determined in duplicate using ImmuChemTM Mouse Double-antibody Corticosterone 125I RIA Kit, MP Biomedicals, LLC, Diagnostics division, Orangebourg, NY, US. The minimum detectable dose of corticosterone using this assay was 7.7 ng/ml, with an intra-assay variation coefficient of 4.4% and an inter-assay variation coefficient of 6.5%. Statistical analyses Body weights obtained on day 0, when the diet was started, at day 5, and at day 10 were compared using ANOVA for repeated measurements, with “measurement” as within-subject factor (3 levels), “diet” (trp-free or ctr) and “line” (SAL or LAL) as between-subject factors. Simple contrasts were set a priori to compare the measurements after the diet (level 2 and 3) with the measurement 1. Corticosterone data were analyzed using two-way ANOVA with “line” and “diet” as between-subject effects. A similar analysis was conducted on serotonin levels, 5HIAA levels and 5-HIAA/5-HT ratios. Attack latencies and behavioural total durations were analyzed using ANOVA for repeated measurements in a similar way to the body weights. Post hoc analyses were performed using t test for independent samples in case of two groups and Tukey’s pairwise comparisons in case of more than two groups.
138
TRYPTHOPHAN-FREE
DIET AND AGGRESSION
RESULTS Body weight Statistical analyses were conducted on the raw body weight data. The change in body weight relative to the baseline is depicted in figure 7.1. SAL mice were overall significantly heavier than LAL mice (“line” effect: F1,28=11.08, p=0.002). Body weight was significantly lower in the tryptophan-depleted group (“diet” effect: F1,28=10.72, p=0.003). Body weight changed significantly along with the experiment (“measurement” effect: F2,27=14.14, p<0.001). The change was significantly different between tryptophan-depleted animals and controls (“measurement*diet” interaction effect: F2,27=23.11, p<0.001). Contrasts analyzed separately for trp-free and ctr groups showed that the interaction effect is caused by a significant decrease in the tryptophan-depleted animals (level2 vs. level1: F1,14=32.25, p<0.001; level3 vs. level1: F1,14=32.25, p<0.001), while no significant changes occurred in the body weight of control animals. Moreover, body weight of tryptophan-depleted mice was not lower than controls before the start of the experiment (t30=1.08, p=0.29), but only at day 5 (t30=2.83, p=0.008) and at day 10 (t30=4.53, p<0.001).
change in body weight (% of baseline)
Plasma corticosterone levels Figure 7.2 shows the corticosterone levels of the mice at the end of the experiment. Two-Way ANOVA revealed that plasma corticosterone levels were significantly higher in mice fed with trp-free diet than in controls (“diet” effect: F1,28=30.17, p<0.001). There was no effect of line or line*diet interaction.
110
100
90 LAL controls LAL TRP-free
80
SAL controls SAL TRP-free
day0
day5
day10
Figure 7.1 Group means and standard errors of change in body weight relative to baseline (day 0). Asterisks indicate significant changes in the tryptophan (Trp)-depleted mice at day 5 and day 10 (p<0.001) after planned contrasts on raw data.
139
CHAPTER 7
control
plasma corticosterone (ng/ml)
500
TRP-free
400 300 200 100 0
LAL
SAL
Figure 7.2 Group means and standard errors of plasma corticosterone levels. Asterisks indicate ANOVA main effect of diet (control vs. tryptophan (TRP)-free, p<0.001)
Prefrontal cortex serotonin levels The levels of prefrontal cortical serotonin, its metabolite 5-HIAA, and the 5-HIAA/5-HT turnover ratio measured at the end of the experiment are represented in Figure 7.3. 5-HT and 5-HIAA raw data were transformed to their natural logarithms to obtain a normal distribution and to homogenize the group variances. Two-Way ANOVA revealed a significant line effect (F1,28=4.32, p=0.047) and significant diet (F1,28=8.42, p=0.007) effect in serotonin levels. Overall, SAL mice had lower serotonin levels than LAL, and tryptophan-depleted mice had lower 5-HT levels than control mice. There was no significant interaction effect between diet and line. However, the tryptophan-free diet reduced serotonin levels by 45% in the LAL mice and by only 29% in the SAL mice. Neither 5-HIAA levels nor serotonin turnover ratios were significantly affected by either diet or line. Behaviour Group means of the attack latencies during the experiment are depicted in Figure 7.4. On average, a trend to a significant reduction of attack latency towards the end of the experiment was observed (“measurement” effect: F2,56=3.033, p=0.056), and there was no interaction effect with “line” or diet”. Significant between-subject effects were “diet” (F1,28=4.81, p=0.037), “line” (F1,28=177, p<0.001), and there was a trend to significance in the “diet*line” interaction effect (F1,28=3.77, p=0.062). As expected, SAL mice had significantly lower attack latencies than LAL mice, and, overall, Trp-depleted mice had lower attack latencies than controls. SAL differed overall from LAL in the behavioural repertoire exhibited during the resident-intruder interactions. Significant “line” effects were found in attack 140
TRYPTHOPHAN-FREE
DIET AND AGGRESSION
5-HT (ng/g tissue)
600
400
200
0
5-HIAA (ng/g tissue)
250 200 150 100 50 0 control
5-HIAA/5-HT ratio
0.8
TRP-free
0.6 0.4 0.2 0.0
LAL
SAL
Figure 7.3 Group means and standard errors of 5-HT, 5-HIAA, and 5-HT turnover (5-HIAA/5HT). * indicates ANOVA main effect of line (LAL vs. SAL, p<0.05). ** indicates ANOVA main effect of diet (control vs. tryptophan (TRP)-free, p<0.01). Statistical analyses were performed on data transformed to their natural logarithms.
(F1,27=97.25, p<0.001), threat (F1,27=23.65, p<0.001), chase (F1,27=33.03, p<0.001), social exploration (F1,27=21.25, p<0.001), non-social exploration (F1,27=7.63, p=0.01), retreat (F1,27=60.89, p<0.001), and in the total offence duration (F1,27=136.93, p<0.001). Moreover, social interaction increased almost significantly along the experiment (F2,54=2.8, p=0.07), while inactivity decreased significantly (F2,54=4.35, p=0.018). No significant effects of “diet” or “diet*line” interaction were found in any of the behaviours scored in the resident-intruder test. 141
CHAPTER 7
300
attack latency (sec)
250 200 LAL controls SAL controls LAL Trp-free SAL Trp-free
150 100 50 0
RI1
RI2
RI3
Figure 7.4 Group means and standard errors of attack latencies in the Resident-Intruder (RI) tests. See text for statistical details.
LAL control
SAL control
LAL Trp-free
SAL Trp-free
60
total duration (% observations)
40
20
0 RI1 RI2 RI3
60
40
20
0
offence social non- body in- retreat explore social care activity
offence social non- body in- retreat explore social care activity
Figure 7.5 Group means and standard errors of total duration of offence, social exploration, non-social exploration, body care, inactivity, and retreat. See text for statistical details.
142
TRYPTHOPHAN-FREE
DIET AND AGGRESSION
DISCUSSION This study confirms that SAL mice, while expressing high trait-aggression, have less serotonin in the prefrontal cortex, a key area involved in the regulation of social behaviour. This is in line with several reports that support the low serotonin levels as a trait characteristic of highly aggressive individuals (Maas, 1962; Valzelli, 1982; Brown and Linnoila 1990; Lee and Coccaro 2001; van der Plasse et al., 2007). Similarly, low serotonin brain levels were found in SAL mice that had been exposed to male intruders (Caramaschi et al., 2008a). However, the present study shows that further lowering serotonin levels by tryptophan depletion does not affect aggression significantly. Low serotonin is apparently not a sufficient condition to change the behavioural traits of mice with genetic predisposition to extremely high and low aggression levels. The lack of behavioural changes in both SAL and LAL mice will be discussed separately for the SAL and the LAL line. It was shown that serotonin is released during a social interaction and aggressive acts (van der Vegt et al., 2003b; Summers et al., 2005), and inhibiting this mechanism through enhanced presynaptic 5-HT1A and 5-HT1B autoreceptors activation reduces rodents’ offensive aggression (Bell and Hobson 1994; Joppa et al., 1997; Miczek et al., 1998; de Boer et al., 2000; DE Almeida and Miczek 2002). These reports suggest that an acute reduction of serotonergic neurotransmission inhibits the execution of aggression. In view of the fact that serotonin content in the cortex is severely reduced by the tryptophan-free diet, the lack of behavioural changes in the SAL line suggests that minimal levels of serotonin are sufficient to allow the expression of aggression against another male. The lack of behavioural change might also be due to homeostatic changes in the serotonergic system of the SAL mice, which may keep serotonergic neurotransmission intact. In support of this idea it was shown that both acute and chronic tryptophan depletion causes changes in the 5HT1A and 5-HT2A receptors (Kawai et al., 1994; D'Souza et al., 2004; Cahir et al., 2007). Desensitization of 5-HT1A receptors might reduce the negative feedback control of the 5-HT neuron and may in this way have maintained sufficient levels of serotonin release. The lack of change in serotonin release was also recently shown in rats (van der Plasse et al., 2007). Similarly, in LAL mice, the serotonin depletion alone did not change their low aggressiveness. In line with our previous studies, it is more likely that aggressiveness is related to 5-HT1A autoreceptor sensitizations, therefore serotonin depletion alone is not expected to increase aggressiveness. This study further delineates another potential mechanism underlying the lack of a behavioural effect of tryptophan depletion. Homeostatic changes in the stress physiology might have counteracted the brain changes and consequently 143
CHAPTER 7
maintained intact the behaviour of the mice. Glucocorticoids released via activation of the hypothalamic-pituitary axis during a stressful event are known to act on the central serotonergic system, in particular modulating the expression of somatodendritic 5-HT1A autoreceptors (de Kloet et al., 1986; Chalmers et al., 1993; van Praag, 2004). Since tryptophan depletion caused an enhancement of the stress induced plasma corticosterone response, this might have induced 5HT1A receptor desensitization. It is interesting to note that serotonin levels in the prefrontal cortex were not completely reduced, but remained at minimal levels. Since tryptophan is lacking completely from the diet, and serotonin needs to be produced in the liver and the guts to a higher extent and in a minor amount in the brain, it is likely that protein storages in the body are used to allow tryptophan to be used for serotonin synthesis, therefore assuring a minimal serotonin amount to maintain basic physiological functions. This might explain the reduction in body weight observed in the present experiment. Finally, in this study we explored mouse intermale aggression exhibited in the context of a resident-intruder interaction against a non-offensive opponent. Other rodent studies suggest that the effects of tryptophan depletion might be found in different behavioural components, for example circadian activity, response to amphetamine, and stress-sensitivity (Kawai et al., 1994; Carta et al., 2006; Tanke et al., 2008). The general reducing effect of tryptophan depletion on attack latencies, not coupled with an increase in aggression levels in the LAL line, suggests that behavioural traits as impulsivity and stress coping with novelty might be affected by the treatment, while maintaining the non-aggressive characteristic feature of the LAL mouse line. Human studies suggest that tryptophan depletion might be effective in conditions of serotonin related disorders such as depression, anxiety, sleep, eating disorders and obsessive-compulsive disorder (for review see (Bell et al., 2001)). In conclusion, this study confirms in a mouse model for male violence that brain serotonin levels are associated negatively with high aggressiveness, but it shows also that low brain serotonin level may be a necessary but not a sufficient condition for high aggressiveness. This conclusion will have implications for future studies of human aggression, suggesting that its neurobiological determinants might be found in the upstream molecular pathways that lead to low serotonin levels in frontal cortical areas. Acknowledgements The authors are thankful to Ramon Granneman for performing the HPLC analysis and Jan Bruggink for performing the HPLC and the radioimmuno assay.
144
TRYPTHOPHAN-FREE
DIET AND AGGRESSION
145
CHAPTER
8
Dynamic intracellular distribution of serotonin-1A receptors in mice predisposed to violence Doretta Caramaschi, Han J. L. van der Want, Henk de Weerd, Sietse F. de Boer, Jaap M. Koolhaas
ABSTRACT Violent behaviour has been associated with low baseline brain serotonergic activity in both humans and other animals. In feral house mice, individuals of the Short Attack Latency (SAL) strain are genetically predisposed to develop pathological aggression. Recently we demonstrated that these violent mice have significantly enhanced inhibitory 5-HT1A receptor (5-HT1AR) function that may underlie or lead to their reduced brain 5-HT activity. However, the mechanism of this hypersensitization is not yet clear. Since the functionality of 5-HT1A-R relates to its presence on the neuronal cell membrane, we investigated the subcellular distribution of the 5HT1A-R in mice of the SAL (aggressive) and LAL (non-aggressive) lines, under baseline conditions and after 5-HT1A-agonist stimulation. We visualized both the autoreceptors in the dorsal raphe and the postsynaptic heteroreceptors of the prefrontal cortex ex vivo with immuno-electron microscopy using sensitive gold substituted silver-enhanced peroxidase staining. In nonstimulated, saline-injected control mice, the label distribution showed fewer 5-HT1A-R in intracellular compartment in neurons of SAL mice compared to LAL. After injection of the full 5-HT1A-R agonist 8-OH-DPAT (0.5 mg/kg), a more pronounced intracellular labeling in the prefrontal cortex, but not dorsal raphe, was seen in SAL mice compared to saline-injected controls. This differs from the LAL animals, where an agonist challenge did not change receptor distribution. The results indicate that prefrontal 5-HT1A-R in violent individuals show accelerated/enhanced dynamics. We argue that the subcellular distribution of 5-HT1A-R does not explain the enhanced 5-HT1A-R sensitivity and that a more responsive 5-HT1A-R dynamics in violent individuals may relate to their higher receptor sensitivity.
CHAPTER 8
INTRODUCTION Four decades of preclinical and clinical studies have accumulated an overwhelming body of evidence that brain serotonergic neurotransmission plays a critical role in modulating aggressive behaviour (Valzelli, 1984; Brown and Linnoila 1990; Virkkunen and Linnoila 1990; Miczek et al., 2007). This scientific database strongly favours a negative link between basal measures of serotonin (5-HT) activity and indices of aggressive and violent behaviour, i.e. the 5-HT deficiency hypothesis of aggression (Coccaro, 1989). Indeed, serotonergic agonists, in particular 5-HT1A and 5-HT1B receptor agonists, acutely reduce aggression levels in many animal models (White et al., 1991; Mos et al., 1993; Bell and Hobson 1994; Joppa et al., 1997; Miczek et al., 1998; de Boer et al., 1999; Pruus et al., 2000; Sperry et al., 2003; Clotfelter et al., 2007). These data suggest that the activation of 5-HT1A receptors is strongly involved in restraining aggressive behaviour, and that diminished 5-HT signalling at these receptors is associated with increased aggression. This idea is supported by a number of observations that healthy people scoring high in trait-aggression, Alzheimer’s patients with enhanced aggression, aggressive pre-pubertal pigs and rats genetically selected for defensive aggression show reduced 5-HT1A receptor expression and/or functionality (Netter et al., 1999; Cleare and Bond 2000; Parsey et al., 2002; Lai et al., 2003; D'Eath et al., 2005; Popova et al., 2005; Popova et al., 2007). In sharp contrast, however, a large number of other studies showed a positive association between structural and functional properties of 5-HT1A receptors and trait-like offensive aggressiveness (Korte et al., 1996; van der Vegt et al., 2001; Van Riel et al., 2002; Feldker et al., 2003a; de Boer and Koolhaas 2005; Veenema et al., 2005a; Schiller et al., 2006; Caramaschi et al., 2007). One way to explain this discrepancy is the heterogeneity of the operational definition of aggression and violence used in the various studies. In general, preclinical studies have obtained neurobiological data that were associated with normal species-specific aggression levels and have tried to extrapolate them to the pathological human condition of violence. Only recently, in our and in some other labs, parameters for pathological aggression in rodents have been described (de Boer et al., 2003; Haller and Kruk 2006; Miczek et al., 2007; Cervantes and Delville 2007; Caramaschi et al., 2008a). The Short Attack Latency (SAL) mice genetically selected for high aggressiveness from a wild population (van Oortmerssen and Bakker 1981) show signs of pathological aggression toward docile and/or immobilized opponents in terms of high intensity of attacks, lack of gender discrimination, high dominance, and lack of inhibition from the opponent’s signals (Caramaschi et al., 2008a; Natarajan et al., 2009). These mice were originally selected through a bidirectional selection 148
INTRACELLULAR 5-HT1A
RECEPTORS AND VIOLENCE
for attack latency that generated both the high aggressive line, SAL, and the low aggressive line, LAL (Long Attack Latency) (van Oortmerssen and Bakker 1981). In a resident-intruder test SAL male mice attack a standard docile intruder in less than 50 seconds, spend 50% of the test duration on offensive behaviour (attack, threat, and chase) and their bites cause several wounds on the opponent’s body. A LAL male very rarely attack, and if so, with a long latency (more than 300 seconds) and for a short duration (less than 5% of the total time) and without causing injuries. Compared to other genetically selected aggressive mouse lines, we recently proposed that the offensive aggressive behaviour of SAL mice mimics human violence (Sluyter et al., 2003; Caramaschi et al., 2008a; Natarajan et al., 2009). In line with the 5-HT deficiency hypothesis of aggression, SAL mice showed lower serotonin tissue levels in the prefrontal cortex than LAL mice, in particular after repeated social experience (Caramaschi et al., 2007; Caramaschi et al., 2008a). In addition, they showed increased structural and functional properties of postsynaptic 5-HT1A heteroreceptors (van der Vegt et al., 2001; Van Riel et al., 2002; Feldker et al., 2003a). SAL mice also exhibited enhanced somatodendritic 5-HT1A autoreceptor functionality, suggesting exaggerated raphe 5-HT autoinhibition in the aggressive and violent-prone mouse line (van der Vegt et al., 2001; Caramaschi et al., 2007). However, in contrast to the postsynaptic heteroreceptors, this increased autoreceptor functionality does not seem to be associated with increased 5-HT1A receptor levels in this brain region: 5-HT1A mRNA expression and 8-OH-DPAT radioligand binding showed no quantitative differences between SAL and LAL in the raphe nuclei (Korte et al., 1996; Veenema et al., 2005a). Furthermore, the 5-HT1A receptor is highly dynamic both in its expression and functionality. Hence, the temporal dynamics of 5-HT1A receptors may generate differential results if samples are taken at different time-points relative to age, social experiences, etc. and analyzed with different techniques that look at different spatial-temporal aspects (e.g. mRNA, protein, G-protein activation, temperature/hormone response to agonists, etc.). More details on the 5-HT1A receptor dynamics especially those with respect to responses to pharmacological treatments in animal models that develop antisocial/violent behaviour would help us understanding the neurobiological changes that accompany the development of violence. Although there are no structural differences between auto- and hetero- 5HT1A receptor protein, several studies have shown functional differences that could be related to different signalling systems involved (Mannoury la Cour et al., 2006) and to differential trafficking properties (Riad et al., 2001). 5-HT1A receptors are rapidly internalized upon agonist treatment particularly in the dorsal raphe nucleus, but not in the hippocampus (Riad et al., 2001; Zimmer et al., 2004; Riad et al., 2004; Aznavour et al., 2006). Receptor internalization has been 149
CHAPTER 8
proposed as a fast dynamic regulatory mechanism by which G-protein coupled receptors become rapidly desensitized upon agonist stimulation, i.e. internalization reduces the number of free receptor sites at the cell surface and hence the 5HT signalling function decreases (Koenig and Edwardson 1997; Bloch et al., 1999; Bernard et al., 2006). This adaptive process has been characterized in vivo and in vitro for many neurotransmitter and neuropeptide receptors (Chuang et al., 1980; Keith et al., 1998; Dumartin et al., 1998; Bernard et al., 1999; Haberstock-Debic et al., 2003; Reyes et al., 2006; Boudreau et al., 2007). Furthermore, receptor endocytosis itself activates specific signalling cascades other than those mediated via G-proteins, with possible consequences on cell survival and neuroplasticity (Kang et al., 2005; Lefkowitz and Shenoy 2005; Ma and Pei 2007). Agonist-stimulated endocytosis/internalization of 5-HT1A receptors has been identified in vitro and in vivo (Della Rocca et al., 1999; Riad et al., 2001). Based on this rationale, we investigated the ultrastructural distribution of the 5-HT1A receptors in mice of the SAL (aggressive) and LAL (non-aggressive) lines under baseline conditions and after 5-HT1A stimulation. With this approach we tested whether the subcellular distribution is reflected in the 5-HT1A receptor functionality/sensitivity. If so, we expect less intracellular receptors in SAL mice than in LAL mice, since SAL mice have enhanced 5-HT1A auto- and heteroreceptor sensitivity. Alternatively, if the process of internalization is critically dependent upon 5-HT1A receptor sensitivity (reflecting an adaptive neuronal response), we expect faster/more pronounced receptor accumulation of intracellular receptors upon agonist stimulation in these aggressive SAL animals.
RESULTS Figure 8.1 shows the light microscopy images of the regions of interest for the electron microscopy screening. The SAL and LAL brain samples did not differ in the morphology of the tissue. Larger cells were seen in the dorsal raphe, compared to those of the prefrontal cortex. The screening of ultra-thin sections using electron microscopy (Fig. 8.2) confirmed the presence of labeling in cell bodies and occasionally in dendrites. In very few areas (n<5) labeling was found in terminals and therefore excluded from the quantification. Within the cell bodies, where the receptor was found mainly, the label was specifically located on organellar membranes, such as ER, Golgi, transport vesicles, and endosomes. Dense labeling was present inside multivesicular bodies and lysosomes. In few cases the label was present on the cell membrane, but in many cases the precipitate was located just underneath the membrane. 150
INTRACELLULAR 5-HT1A
A
B
C
D
RECEPTORS AND VIOLENCE
Figure 8.1 Examples of 50 µm vibratome sections as seen with light microscope of prefrontal cortex (A, B) and dorsal raphe (C, D), which indicate the regions of interest selected for ultrathin sections of prefrontal cortex (A) and dorsal raphe (C) at low magnification, and the specific labeling of 5-HT1A receptor in the cytoplasm and at the cell membrane of neurons in the prefrontal (B) cortex and dorsal raphe (D) at higher magnification (scale bar indicates 100 µm). The digital pictures were obtained with a Leica DMIRB light microscope (Leica, Cambridge, UK) equipped with a Leica DFC350FX camera, magnification 40X, and QWin software (A, C) and Leica SP2 confocal microscope in the wide field transmission mode (Leica, Nussloch, Germany).
The quantification of immunolabeling is shown in Figure 8.3. Two –Way ANOVA identified a significant effect of the “line*treatment” interaction in prefrontal cytoplasmic (ER, Golgi, small trafficking vesicles) labeling (F1,12=8.4, p=0.02) and almost significant “line*treatment” interaction effect in total prefrontal labeling (F1,12=5.16, p=0.053). In control conditions, cytoplasmic labeling in the prefrontal cortex was lower in SAL than in LAL (t4=3.6, p=0.022). In SAL agonist-treated mice, the cytoplasmic proportion of granules was significantly increased compared to SAL controls (t4=4.11, p=0.018). Similarly, the 151
CHAPTER 8
A
B
C
D
E
F
152
INTRACELLULAR 5-HT1A
800
total
RECEPTORS AND VIOLENCE
intracellular
# dots/100 µm2
600 400 200 0
# dots/100 µm2
600
PFC
DR
prefrontal cortex
PFC
DR
dorsal raphe LAL_CTR LAL_DPAT SAL_CTR SAL_DPAT
400
200
0
cytosol
degradation
cell membrane
cytosol
degradation
cell membrane
Figure 8.3 Quantification of gold-silver peroxidase product from electron micrographs. The histogram bars represent averages and their standard errors of LAL (Long Attack latency) and SAL (Short Attack Latency) mice in either CTR (control) or DPAT (agonist-stimulated) condition. PFC=Prefrontal cortex, DR=Dorsal raphe. * p<0.05
nearly significant interaction effect in the total prefrontal labeling was due to enhancement of labeling after agonist injection in the SAL mice (t4=2.94, p=0.044). Close to significance was also the interaction effect in prefrontal intracellular labeling (F1,12 =4.34, p=0.071), which represents a trend in the same direction, with agonist-treated SAL having more labeling than SAL controls (t4=2.8, p=0.049). There were neither significant nor close to significance effects in the labeling of degrading structures and cell membrane. No significant differences were found in the raphe region.
Figure 8.2 (left) Examples of images demonstrating intracellular 5-HT1A receptor labeling. In (A) part of the cell body is shown with a nucleus and diffuse label distributed in ER and small vesicles in the cytoplasm. In (B) the label shows clustering in multivesicular bodies in proximity of the nucleus. In (C) multivesicular bodies with clustered label are in the cytoplasm. In (D) the label is present in a Golgi complex and diffuse in small vesicles in the cytoplasm. In (E) and (F) the label is clustered in dendritic profiles within a multivesicular body and early endosomes close to the membrane, respectively. Open arrows indicate labeled multivesicular bodies and lysosomes, while closed arrows indicate labeled Golgi, ER and small trafficking vesicles. N = nucleus, D = dendrite, G = Golgi complex. Scale bar indicates 1 µm
153
CHAPTER 8
DISCUSSION With the present study, we show that individual differences in the levels of aggression are related to the intracellular trafficking of 5-HT1A receptors. Based on the differential sensitivity of 5-HT1A auto- and heteroreceptors in SAL and LAL mice (van der Vegt et al., 2001; Caramaschi et al., 2007), and based on the fact that internalization of cell membrane receptors is thought to be a mechanism by which cells regulate their exposure and consequently their functionality (Ferguson, 2001), we investigated the receptor localization at the ultrastructural level. In control conditions, we found less intracellular localization of the postsynaptic 5HT1A heteroreceptors in the prefrontal cortex of SAL brain preparations. When stimulated with the agonist, the amount of intracellular receptors increased only in this highly aggressive line, suggesting a more dynamic system in the highly aggressive line. A surprising aspect of this study is that 5-HT1A immunoreactivity was mostly present intracellularly also in the control animals, contrarily to our expectations and to other reports (Riad et al., 2001; Riad et al., 2008). Using the gold substituted silver-enhanced peroxidase technique, we visualized with high resolution and sensitivity the presence of immunolabelled receptors in detailed structures, as discussed in (Morara et al., 2001). For instance, granules that are sometimes located in early endosomes just under the cell membrane could be classified as being surface receptors when visualized as bigger particles using the immunogold technique. Yet, it is difficult to imagine that at baseline conditions, neurons express such a small number of 5-HT1A receptors at the outer cell membrane. It is more likely that the mice in these experiments were affected by handling/injection stress also in control conditions, and most of the surface 5-HT1A receptors were already internalized after the stress-induced serotonin release. Another explanation is that the antibody used was more likely to bind intracellular/non-functional receptors than those exposed on the cell surface. The antibody used in this study gave similar intracellular immunoreactivity patterns as those presented in other previous studies on rodent brain tissue (Collin et al., 2002; Zhang et al., 2004; Luttgen et al., 2005). The epitope recognized by this antibody is located in the third intracellular loop of 5-HT1A molecule, a region highly involved in interactions with other proteins and phosphorylation sites (Raymond et al., 1999). Therefore, one possibility is that the epitope is partially masked when 5-HT1A is present on the cell membrane, and not when it is located intracellularly. Yet, even considering the partial masking of 5-HT1A receptor when exposed to the cell membrane and the possibility of stress-induced internalization, the lower amount of cytoplasmic 5-HT1A in SAL mice prefrontal cortex may indicate a lower tonic serotonergic release in this brain area. This hypothesis is in line 154
INTRACELLULAR 5-HT1A
RECEPTORS AND VIOLENCE
with the previous literature of low serotonergic neurotransmission in impulsive aggressive individuals, see for review (Lee and Coccaro 2001). As in previous studies, 5-HT1A autoreceptors were found intracellularly in the neurons of the dorsal raphe nuclei. Interestingly, in our study 5-HT1A receptors were already intracellularly present in non-stimulated conditions. Although it was shown in rats that 5-HT1A heteroceptors in hippocampal neurons do not internalize (Riad et al., 2001), in our study using mice as subjects a considerable proportion of immunoreactivity was found in the intracellular compartment of prefrontal cortical neurons. These discrepancies might reflect a species-specific difference or a regional difference and they remain to be explored. To understand the functional consequences of our inference we need to consider the receptor dynamics in detail. When activated by an agonist, 5-HT1A receptor undergoes conformational changes that eventually lead to receptor endocytosis (Chini and Parenti 2004; Wolfe and Trejo 2007). The endocytosed receptor can be either recycled to the cell membrane or degraded in lysosomes. The short-term (minutes to hours) desensitization effect temporarily shuts down the receptor from its function. It is possible that a short-term higher endocytosis rate upon agonist-stimulation in SAL mice is counteracted by a long-term enhanced recycling and/or de novo synthesis. In SAL mice the highly sensitive 5HT1A receptors may internalize quickly, as shown by the increase in intracellular labelling, therefore they may be temporarily shut down from their functions (van der Vegt et al., 2001; Caramaschi et al., 2007). However, the increase in intracellular labelling in SAL mice is represented mainly by structures involved in receptor synthesis and or recycling, which might ultimately result in higher number of receptors, as found in previous studies (Korte et al., 1996; Veenema et al., 2005a). In conclusion, these data suggest a less intracellular and more dynamic distribution of 5-HT1A receptors in frontocortical brain areas of individuals predisposed to violence. This important link between behaviour and brain molecular physiology requires further investigation and may open up new hypotheses in the field of the neural and molecular mechanisms of aggression.
EXPERIMENTAL PROCEDURE Animals Male mice (n=12) from the SAL (n=6) and LAL (n=6) selected lines were obtained from our breeding colonies. After weaning at 21 days, they were kept in unisexual groups with siblings. At the age of 40 days, each male mouse was housed with a female in order to avoid social isolation. The mice were housed in 155
CHAPTER 8
Makrolon® Type II cages, with free access to food (AMII, ABdiets, Woerden, The Netherlands) and water, in a room at 22 ± 2°C and 12/12 light/dark cycle (light off 2:30 PM). All the mice were naïve regarding inter-male social experiences in adulthood. This study was carried out in accordance with the Dutch Law for Animal experiments and with the European Communities Council Directive (86/609/EEC), and under approval of the Institutional Animal Care and Use Committee of the University of Groningen (DEC protocol n. D4328A). Brain tissue collection During adulthood (4-7 months), on the day of the experiment, half of the mice were injected subcutaneously with either 0.5 mg/kg of 8-OH-DPAT (± 8-hydroxydipropylamino tetralin hydrobromide) (Tocris, Bristol, UK) or vehicle (saline). The mice were naïve to the injection procedure, since it was shown that injection habituation procedures may have the opposite stress-increasing effect in mice (Hennessy et al., 1977). Twenty minutes after the injection, they were anesthetized with Pentobarbital (30 mg/kg, i.p.). Under anesthesia, they were perfused transcardially with 20 ml rinsing solution (0.1 M phosphate buffer, 2% polyvinylpyrrolidone (MW=40K), 0.4% NaNO2, 1.2 ml/l heparine), and subsequently with fixative containing 2 % paraformaldehyde, 0.05% glutaraldehyde, and 2% polyvinylpyrrolidone; in some cases 0.2% picric acid was added to improve ultrastructural preservation. The brain was removed from the skull and left in the same fixative solution overnight at 4°C. Vibratome sections of prefrontal cortex and dorsal raphe nuclei (50µm thickness) were cut, collected in PBS and stored at 4°C. Gold-silver substituted peroxidase 5-HT1A immunolabeling For each excised brain, three sections of the prefrontal cortex and three sections of the dorsal raphe were quenched with 0.3% H2O2 in PBS and rinsed in Trisbuffered saline (TBS, pH=7.6) on ice. The sections were then preincubated for 30 minutes with 5% normal goat serum (NGS) and 0.05% Triton X-100 in TBS and subsequently incubated overnight with guinea pig anti-5-HT1A receptor (Chemicon International, Inc., 1:1000), 1% NGS and 0.05% Triton X-100 in TBS at room temperature. The sections were then rinsed with TBS on ice and incubated for 90 minutes with biotinylated goat anti-guinea pig (Jackson Immunoresearch laboratories, Inc., 1:500), 1% NGS and 0.05% Triton X-100. They were subsequently rinsed in TBS on ice and incubated for 60 minutes with ABC-reagent (Vectastatin PK 6100, 1:400) and 0.05% Triton X-100. The staining was developed with 3,3'-diaminobenzidine (DAB)/H2O2 reaction (0.025% DAB in TBS and 0.004% H2O2). For orientation purposes and selection of the appropriate neuronal areas, sections were examined under light microscope. 156
INTRACELLULAR 5-HT1A
RECEPTORS AND VIOLENCE
Following the immunolabeling, the sections were treated according to the gold substituted silver-enhanced peroxidase method as in (Morara et al., 2001) washed with 0.1 M sodiumcacodylate buffer (pH 7.4), osmicated in 1% OsO4, 1.5% potassium hexacyanoferrate(II) in 0.1 M cacodylate buffer (pH 7.4) at 4°C, dehydrated in graded series of ethanol and embedded in Epon on silane-coated microscope slides. Sections treated without primary antibody did not show any labeling. Furthermore, using prefrontal cortical tissue samples of both SAL and LAL mice, a preliminary Western blot with the same antibody was performed, which revealed a single immunoreactive band around 37 kDa (Natarajan D., unpublished). Electron microscopy analysis Light microscopy was used to select regions of interest from the prefrontal cortex and the dorsal raphe nuclei and further processed for ultrathin sectioning. The ultrathin sections were counterstained with uranyl acetate and lead citrate, and examined using a Philips 201 and a Philips 208 transmission electron microscopes at 60kV (FEI, Electron Optics, Eindhoven, the Netherlands). Since the reaction product was organized in small round dense black precipitates with defined contours, the density of the labeling reaction product was quantified by counting from electron micrographs, taken at final magnification of 20.000 – 40.000X on the intracellular cytoplasm and on the cell membranes. Localization on organelles was performed by careful observation at the microscope and on the photos. Intracellular labeling in clear large membranesurrounded structures that would fall in the definition of endosomes, lysosomes, and multivesicular bodies was considered as “degradation” since they most likely reflect the intracellular protein degradation pathway. Cytoplasmic labeling was defined as immunolabeling located over organellar membranes of small trafficking vesicles, rough endoplasmic reticulum, tubulo-vesicular smooth endoplasmic reticulum, Golgi cisternae. Labeling in these organelles more likely represents newly synthesized receptors. The total number of gold-dots per region (prefrontal cortex or dorsal raphe) and per type (degradation or cytoplasmic) were divided by the total area examined and expressed as number of gold-dots per 100 _m. Since the quantitative data were not continuous, but discrete (i.e., counts), they were transformed to their natural logarithms. Two-way ANOVA was performed to identify the significance of “line” (2 levels: SAL and LAL), “treatment” (2 levels: control and DPAT) and “line*treatment” interaction effects. Furthermore, post hoc comparisons using t-tests for independent samples were performed to compare the total and the intracellular labeling in SAL controls vs. LAL controls, in order to test the specific hypothesis on differences in sub cellular distribution at baseline. To test the effect of the agonist, within each line, agonist157
CHAPTER 8
treated animals were compared with relative controls using t-test for independent samples. The level of significance was p<0.05. However, post hoc comparisons for the interaction effects were performed also in case of trends (0.05
158
INTRACELLULAR 5-HT1A
RECEPTORS AND VIOLENCE
159
CHAPTER
General discussion
9
CHAPTER 9
Four decades of social and biological research have been spent in search of an explanation for human aggressiveness, since it inflicts such a heavy and costly burden on society. Although only a small fraction of the population escalate their aggressiveness and become criminally violent, violence is highly detrimental for society as a whole. Beyond being a direct cause of death and immediate physical injury, violence also produces less obvious, but profound, long-term physical and emotional disabilities in its victims and witnesses. Most research effort has concentrated on the recipient of an aggressive attack, who may show lasting changes in the brain that lead to severe psychological (anxiety, post-traumatic stress disorder and depression) and physiological (cardiovascular, immunerelated) illnesses. However, it is often overlooked that the offenders are victims themselves, in that somehow they are not able to stop themselves actively from causing injuries and harm. When outbursts of aggression are co-morbid with DSM-IV-defined neuropsychiatric disorders, the offenders are given psychiatric care, but when they appear to belong to the normal healthy population their most likely fate is punishment (prison) by the criminal justice system. In the past, (neuro-)biological factors have not been given sufficient consideration in the search for ways to deal with human aggression and violence. However, with recent advances in pharmacology, molecular genetics and brainimaging techniques, scientists have gathered more detailed knowledge about the physiological factors and molecular events that control aggression and the ways in which social experiences and gene-environment interactions impinge on them. Lifetime-persistent, high-frequency expression of excessive physical aggression can be considered a pathological condition of the neural circuitry and molecular mechanisms of aggression. As such, it should be treatable using the typical biomedical principles of prevention and intervention. Unfortunately, however, in contrast to various other disorders, pathological aggression still lacks an appropriate objective definition, let alone physiological biomarkers that could help to identify individuals at risk. Moreover, proper and effective treatment options have not yet been developed. To initiate evidence-based prevention and intervention programmes we need a solid causative, neurobiological explanation of pathological aggression. I have approached this problem by considering that individuals differ in their tendency to display offensive aggression because of a genetic predisposition. These individuals might show physiological characteristics that render them more vulnerable to becoming violent. In particular, social conflicts that they “win” through aggressive behavioural acts might enhance their propensity to be aggressive in future situations. Therefore it is necessary to study the effect of such social experiences in biologically predisposed individuals and the consequences for brain functioning. In search of a way to study all these issues in a laboratory 162
GENERAL
DISCUSSION
setting, I chose to look for a physiological explanation of mouse violence, with the assumption that we can extrapolate the findings to the human situation, based on the evolutionary relatedness of the two species.
SUMMARY OF THE RESULTS In this thesis I studied three pairs of mouse lines selected for high (SAL, TA, NC900) and low (LAL, TNA, NC100) aggressiveness. I examined their behaviour, their peripheral stress physiology and their central serotonergic system with emphasis on the serotonin-1A receptors. Here I present an overview of the results in relation to the original research questions. Are there behavioural correlates of the aggressive mouse lines that relate to different groups of aggressive people? In Chapter 2 the behavioural phenotypes of mouse lines selected for high and low aggression were studied in several nonsocial contexts. The SAL-LAL pair, originally derived from wild-trapped mice, showed a weak association between aggression and proactive coping. In the two other selection pairs that originated from domesticated albino laboratory strains, high aggression was negatively associated with emotionality. Related to this, which lines develop similar pathologies to human violence? What are the neurochemical correlates of these more or less violent phenotypes? In Chapter 3 the SAL line was found to show signs of violence (=pathological aggression), in attacking females and in not being sensitive to submission signals from a docile intruder. When permitted to dominate physically other conspecifics repeatedly in daily resident-intruder tests, the SAL mice increased their violence levels against females, in terms of increasing the attack/threat ratio, i.e. the actual attacking behaviour relative to the preparatory threatening component. During the training, SAL and LAL mice increased their prefrontal cortex serotonin levels, but the levels of SAL mice increased only slightly, consequently generating a significant difference in the serotonin levels between SAL (lower levels) and LAL at the end of the training. Dopamine metabolism was also affected differentially in the SAL and LAL lines, with the SAL increasing and the LAL decreasing their dopamine metabolism in the prefrontal cortex. The other highly aggressive lines did not show violence and did not change their frontocortical serotonin levels. In contrast, noradrenaline was negatively associated with high aggression and with the repeated resident-intruder experience in these lines. Finally, the social training reduced the serotonin metabolism in all the lines, independent of their aggression levels and genetic background. These results support the current view that the neural network controlling aggression is not static but plastic, as social experiences reshape preand/or postsynaptic neuronal elements to enhance aggressive behaviour. 163
CHAPTER 9
Are violent mice physiologically similar to violent humans? In Chapter 4 the violent SAL mice and the less pathological TA aggressive mice were found to be bradycardic in their resting baseline, while the TA and NC900 aggressive lines were found to have higher maximum core body temperature during their awake baseline. In general, the aggressive lines had also higher circadian amplitudes for heart rate and core body temperature. During a handling stressor, the high aggressive lines did not show autonomic hyperactivation. Interestingly, the NC100 lowaggressive line showed blunted reactivity in combination with a phenotype characterized by low activity levels and high body weight and fat deposits. In which types of mouse aggression is serotonin involved? How involved are 5-HT1A receptors in these more or less violent mice? In Chapter 5 the violent SAL mice were found to have the lowest prefrontal serotonin levels, followed by TA and NC900. Overall, low serotonin levels were associated with high aggression levels. In SAL there was a strong inhibitory feedback from the serotonin-1A autoreceptor, while in TA there was a stronger activation of the serotonin-1A heteroreceptor. In the NC-lines, the low-aggressive NC100 showed higher serotonin-1A heteroreceptor functionality, perhaps related to their obesity-prone phenotype. How do the dynamics of 5-HT1A receptors relate to the escalation of aggression into violence? In Chapter 6 changes in the functionality of the serotonin-1A autoreceptors were associated with the escalation of high aggression into violence. The SAL mice were the ones reaching a more violent phenotype and, at the end of the daily repeated resident-intruder experience, violence levels were correlated with higher 5-HT1A autoreceptor sensitivity. Is serotonin causally involved in determining the aggressive phenotype? In Chapter 7 a reduction in tryptophan availability in the diet significantly reduced the prefrontal levels of serotonin in SAL and LAL mice. While reducing body weight and increasing plasma corticosterone levels, there was no significant change in the high or low aggressiveness of the mice against an intruder. Can the ultracellular distribution of 5-HT1A receptors explain the difference in functionality associated with mouse violence? In Chapter 8 the prefrontal intracellular serotonin-1A heteroreceptors of violent SAL mice were found to be fewer in number than those of LAL mice, and more dynamic in terms of an agonistinduced increase in the intracellular pool, mainly in organelles not involved in the degradation pattern.
VIOLENT MICE In the context of research on aggression and violence, this thesis demonstrates an important novel result that genetically predisposed mice can develop a violent 164
GENERAL
DISCUSSION
phenotypic trait or combination of traits that resembles that of pathologically aggressive human individuals. By combining the results from my studies with other studies in the lab, I found this phenotype specifically in the feral SAL mice, in which it was especially prominent after a repeated social experience that reinforces their tendency to exhibit exaggerated offensive aggression. Research on aggressive behaviour that uses preclinical animal models has always encountered difficulties in translation to humans, because too often the adopted animal models were not showing signs of pathology, but only exhibiting species-typical aggressive traits in the context of adaptive and functional social communication. In resident mice, in particular, offensive agonistic behaviour against an intruding conspecific is a manifestation of territorial control. In humans, this functional aggression is not expressed in terms of agonistic behaviour. Human offensive physical aggression is considered pathological, since healthy individuals have ways to inhibit and control their propensity to engage in aggressive acts. In order to find an animal model that best represents human violence, the Behavioural Physiology research group in Groningen decided to compare three selection pairs for high and low predisposition to aggression. From human research, a number of behavioural parameters that are present in violent individuals can also be objectively analyzed in the mice. The criteria are: – aggressive temperament from youth onwards/genetic predisposition – aggression exhibited across different situations – lack of discrimination in terms of type opponent and/or body-region of attack – lack of inhibition – very little ritualized threatening in comparison to the actual injurious act. This research considers the SAL line as the most valid model for human violence, since it shows all these criteria. Together with their characteristic violent phenotype, SAL mice display certain physiological features that are not typical of the other highly aggressive, less violent lines. At the neurobiological level I have identified differences in the serotonergic system. Low serotonin levels in the prefrontal cortex are characteristic of SAL mice after they have received a social experience and their aggression levels have saturated to the highest limit. The low serotonin levels are not the constitutional cause of the extreme aggressiveness but, rather, they seem to be a consequence of other underlying factors. An enhanced sensitivity of the 5-HT1A autoreceptor seems to be a plausible causative mechanism. While during a repeated social paradigm other mice would desensitize their 5-HT1A autoreceptors, individuals that develop the most violent phenotype of the SAL line are resistant to this process, so that at the end of the paradigm there is a high correlation between violence and 5-HT1A autoreceptor sensitivity. The lower serotonin levels and the enhanced 5-HT1A autoreceptor functionality may eventually result in inhibited 165
CHAPTER 9
tonic release of serotonin. The consequence is probably an impaired serotonergic neuromodulation in control areas of social behaviour, such as the prefrontal cortex. On the other hand, it seems that the 5-HT1A heteroreceptors in the prefrontal cortex of SAL mice are more dynamic, as shown by the higher intracellular responsiveness to a pharmacological challenge. Table 9.1 shows a summarized overview of the neurobiological phenotypes of the six mouse lines studied in this thesis. The brain serotonergic system also importantly modulates the peripheral autonomic and endocrine physiology. The consequences of a lower activity of serotonergic neurons are also seen in a slow resting heart rate and higher hyperthermia during wakefulness and stress. Since the direct link between serotonergic regulation of the autonomic nervous system is not yet well understood, we cannot Table 9.1 Neurobiological phenotypes of the six mouse lines selected for high and low aggressiveness (= indicates no difference).
SAL TA NC900 TNA NC100 LAL
PFC 5-HT
PFC NE
PFC DOPA metab.
Auto5-HT1A-R function.
Hetero 5-HT1A-R function.
DRN 5-HT1A intracell. dynamics
PFC5-HT1A intracell. dynamics
Low Low Low = = High
= Low = High = =
High = = = = Low
High = = Low = Low
High High Low Low High Low
=
High
=
Low
Table 9.2 Physiological phenotypes of the six mouse lines selected for high and low aggressiveness (= indicates no difference).).
SAL TA NC900 TNA NC100 LAL
Rest HR
Day Temp.
Auton. react.
CORT
Leptin
BW
Anti-oxidant*
Low Low = = = =
= High High = = =
High High High = Low =
= Low = Low = =
= = = = High =
= = = = High =
Low
* from (Costantini et al., 2008)
166
High
GENERAL
DISCUSSION
make direct hypotheses about anatomical structures and their functioning. However, it is interesting that the peripheral physiological traits we have identified are potential biomarkers for violence. Table 9.2 shows an overview of the physiological phenotypes of the six mouse lines studied in this thesis.
HOW VALID IS THE SAL LINE AS A MODEL FOR HUMAN VIOLENCE? As for every animal model of psychopathology, one has to evaluate the validity of the genetic selection lines for aggression as a model for human violence. I start with the face validity, i.e. the resemblance of the behavioural symptoms to the human psychopathology. By looking at the type of aggression displayed, and its more or less pathological aspects, I identified that SAL mice are more similar to violent people than the TA and NC900 mice. I also looked at behavioural endophenotypes, i.e. quantifiable heritable and stable traits associated with aggression. I found violence in the aggressive individuals to be associated with a proactive behavioural coping style. Similarly, in humans, a subgroup of individuals with antisocial personality, the psychopaths, can have a successful life characterized by a series of social victories and a considerable amount of instrumental, manipulative and goal-directed behaviour. Secondly, the model also has predictive validity for violence, in the sense that individuals of the SAL line that are exposed to a series of these social victories are predicted to enter a routine of chronic physical aggression and recidivism, as humans would do in such a situation. This is the case for SAL mice, but less evidently for the non-violent lines. Finally, construct validity, i.e. the similarity between the neurobiological mechanisms involved, is also present in the SAL line. Low resting heart rate and other indicators of autonomic regulation are the best biomarkers so far for recognizing individuals at risk of violence and psychopathy. Low serotonin is also consistent with the violent offenders, and the evidence about 5-HT1A is slowly appearing in the human literature, due to the newly available imaging methods. So which individuals did I model in my mouse studies? In longitudinal studies, individuals that show life trajectories of chronic physical aggression comprise roughly 4-5% of the population (Nagin and Tremblay 1999). These individuals commit most of the violent crimes. A similar percentage of early-onset aggressive individuals with chronically high aggressive tendencies has been found in monkeys and feral rodents (van Oortmerssen and Busser 1989; Suomi, 2003; de Boer et al., 2003). The artificial selection in the wild mice that generated the SAL violent mice increased this percentage, making the physiological studies statistically more powerful. However, the phenotype of the SAL mice is present in 167
CHAPTER 9
only a small number of individuals in the original feral population. The presence of this violent phenotype in the population is suggested by the number of casualties after wounding in the females and in the juveniles after the increase in population density (de Boer et al., 2003; Koolhaas et al., 2007). The SAL line is therefore very suitable as an animal model for human violence and should be studied in more detail in terms of early prevention and intervention strategies based on its characteristic physiological phenotype. A comparison of SAL mice with antisocial/violent human offenders is summarized in Table 9.3.
PARENTAL STRAIN MATTERS An interesting result of this thesis is the failure to identify a violent phenotype in the other two highly aggressive lines. To understand why this happened, it is necessary to consider how the selection was originally made. First, the parental Table 9.3 SAL mice as an animal model for human antisocial/violent behaviour (adapted from Sluyter et al., 2003). Humans
SAL mice
High High High High
Aggression impulsive/short attack latency Aggressive personality frequency of violence/violent attacks domestic violence/attacks females
High High High High
High High High
Behavioural signs/disorders reproduction/littersize alcohol dependence/preference prevalence anxiety and depressive symptoms
High High ?
Low High Low
Autonomic/neuroendocrine HPA-axis (re)activity HPG-axis (re)activity neurosympathetic tone
Low High Low
Low High(?) Low
168
Measure
Neurobiological Serotonin Vasopressin Prefrontal cortex functioning
Low High Low(?)
GENERAL
DISCUSSION
strain was different in the three pairs of lines. Only the SAL and LAL mice were obtained from a wild population, while the other two pairs were selected from Swiss-Webster and ICR laboratory strains of mice. The strong artificial selection that was imposed on the laboratory mice in a non-systematic way over centuries might have caused the violent phenotype to disappear. Mice from laboratory strains are typically selected for being docile, easy to handle and good at reproducing in captivity. The same concept applies for rats, in which the production of laboratory strains led to the disappearance of the violent phenotype (de Boer et al., 2003). The consequences of this for the study of the neurobiology and physiology of violence are huge: it suggests that most of the studies conducted on standard laboratory strains describe the neurobiological and physiological characteristics of aggressive animals that do not fall in the violent percentage of a feral population. Although the ultimate evolutionary explanation is still unclear, biologists recognize the existence of maladaptive aggression in several animal species (Sih et al., 2004a). In the original population of house mice used for the selection of SAL and LAL, the presence of an extremely aggressive phenotype was related to the increase of population density in the population cycles (Koolhaas et al., 1999). These aggressive individuals easily obtain social dominance and territorial control, pushing away the less aggressive ones. Since they are quite territorial, they do not need to evolve mechanisms to explore new environments. Nonaggressive individuals, in contrast, are more easily defeated and need to disperse and migrate in order to find new territories to establish their colonies. Hence they benefit from being more explorative. As previously shown, the phenotypes of both SAL and LAL mice have high adaptive value in different contexts (Digman, 1990). The high aggression of the SAL mice is also a part of a more general coping strategy aimed at proactively changing the surrounding environment in search of comfort and dominance. The strategy of LAL mice is that of non-aggressive/reactive individuals, aimed at flexibly reacting to environmental conditions with behavioural plasticity. The same adaptive explanation cannot be advanced for the T and NC lines because of the high degree of artificial selection that preceded the formation of these lines. Their degree of aggressiveness seems to vary with a certain degree of emotionality.
MULTIDIMENSIONAL NATURE OF THE SELECTION: WHAT HAVE WE SELECTED FOR? Different selection criteria used in the selection process possibly contributed to the absence of a violent phenotype in the TA and NC900. The SAL mice were selected on attack latency, i.e. the time it takes to attack a cage intruder in the home cage. 169
CHAPTER 9
Attack latency is correlated with time spent in offensive behaviour, but primarily it measures the readiness to attack and might relate to impulsivity and lack of behavioural control. In contrast, the TA and NC900 mice were selected in a neutral cage. Agonistic interactions in the neutral cage might function differently in the mouse social system. A subordinate mouse in its home territory that is pushed away from its deme might still be able to obtain social dominance in a neutral territory. Moreover, the NC900 mice were selected on the basis of the frequency, rather than the duration, of aggressive behaviour. These mice seem to show a high frequency of behavioural shifts, but their total offensive behaviour may not be extremely high. The generally high behavioural frequency of these mice might still reveal a very cautious and communicative behavioural predisposition, very different from the SAL’s resilience to give up the fight. Aggression is one manifestation of a broader behavioural characteristic. This line of reasoning leads to an important conclusion. Genetic predisposition for aggressive personality may not always lead to a pathological behavioural phenotype, even after a reinforcing winning social experience. On the contrary, individuals with a genetic predisposition for being very proactive and inflexible, if exposed to aggression-escalating routines, might be more at risk to engage in violent acts, i.e. injurious, uninhibited, exaggerated aggressions, whereas individuals with genetic predisposition for low emotionality might show aggressive tendencies to achieve dominant status without incurring the pathology. Table 9.4 shows an overview of the behavioural phenotypes of the six mouse lines studied.
Table 9.4 Behavioural phenotypes of the six mouse lines selected for high and low aggressiveness (= indicates no difference).
SAL TA NC900 TNA NC100 LAL
170
Aggression
Non-social coping
Activity
High/Pathological High High Medium Low Low
proactive = = anxious anxious reactive
= = = = Low High
GENERAL
DISCUSSION
TWO-TIER MODEL Individual variation in behavioural traits is measured through behavioural responses to certain tests or challenges. Many of the behaviours exhibited by one individual are correlated with each other and they cluster in higher-order independent factors. Human personality is, for example, described by three, five or more factors according to different authors (Digman, 1990). Rodent studies have identified at least two independent axes that are hierarchically related, each representing a different level of variation (Steimer et al., 1997; Bronikowski et al., 2001; Koolhaas et al., 2007; Veenema and Neumann 2007). Along one axis there is the variation in coping style, from an extremely reactive to an extremely proactive strategy. The coping strategy is a way to react to challenges of different nature, a basic rule that an individual applies in different contexts. It determines the type of response, whether aimed more at mastering the situation by changing the environment to make it more comfortable, or at adjusting one’s behaviour and physiology in a flexible way, or a response intermediate between the two. This axis seems to be a fundamental one in defining an individual’s personality. A second level of variation is the appraisal of the challenge as safe or dangerous and it determines the intensity of the response. A challenge can be perceived as more or less of a threat, depending on a certain threshold. Individuals with a lower threshold perceive almost anything as challenging and consequently respond more emotionally with their proactive or reactive behavioural strategy. Individuals can be placed in a two-dimensional space described by these two orthogonal axes. From the comparative analysis of the high- and low-aggressive lines, it can be concluded that aggressiveness can be part of different components of one’s personality. Aggressiveness that arises from the highly proactive individuals might be more likely to transform into violence, whereas another tendency to high aggression might be just a part of a low-anxiety behavioural profile. This is well represented in our mouse lines, in which the SAL represents the highly proactive and highly aggressive phenotype and is the most likely to develop violence. In contrast, the selection criteria that generated the highly aggressive TA and NC900 lines might have acted along the appraisal axis, since it is the emotionality response rather than the coping strategy that differentiates the high- from the low-aggressive lines, particularly in the T selection. The neurobiology and physiology described in the mouse selection lines might help explain this two-tier model. Serotonin levels, in particular 5-HT1A auto-receptors, seem to be involved in the general differentiation in coping styles, while noradrenaline seems to characterize the aggressive/low-anxiety phenotype. Dopamine may be involved in the proactive phenotype when it has been pushed to pathological aggression, through routine formation. In the peripheral physi171
CHAPTER 9
ology, aggression is related to low resting heart rate and hyperthermia. Daily fluctuations in heart rate and temperature are greater in the selection for proactivereactive coping style, suggesting that proactive aggressive behavioural traits correlate with a bodily physiological make-up well suited to exerting/sustaining future potential fights. It could be hypothesized that GABA-ergic neuromodulation is involved in the variation along the appraisal axis, since most of the anxiolytics are effective on the GABA-ergic system. This hypothesis remains to be tested. A further interesting result of this research is that general activity levels and consequent metabolic physiological characteristics might represent an independent behavioural trait, adding an extra dimension to this two-tier model. Maybe this variation was included in the selection of the NC lines, also based on the presence of selection criteria that related to activity levels. Some studies have already suggested that aggression, degree of emotionality and activity are selected independently (Sluyter et al., 1995c; Sluyter et al., 1996c; Benus and Rondigs 1997).
DYNAMICS OF THE SYSTEM: PHENOTYPIC PLASTICITY In this thesis, a dynamic picture of the physiology of aggression has emerged. Often, studies on aggression or other behaviours hunt for fixed correlations between the behavioural and physiological traits observed. These studies often result in contrasting results or no correlation at all. The presence of contradictory data or the failure to replicate previous results can be explained by differentiating between state and trait. A state is the sum of temporary individual characteristics in a certain moment of time, while a trait includes more life-stable individual features. However, life-stable individual traits are difficult to observe, since individuals often adapt to environmental changes. In this thesis, it is clear that individuals are supported by highly dynamic physiological processes and that they show a certain degree of phenotypic plasticity at several levels. Individuals are able to adapt their behavioural strategies to the environmental challenges based on their predisposition (genetic, epigenetic) and their previous experiences. Meanwhile, at the central and peripheral level a certain number of molecular rearrangements must occur. Thus, this high dynamicity has a remarkable impact on the experimental designs. We have shown how adaptive and normal forms of aggression can escalate into a more violent phenotype, depending on previous experience and genetic predisposition. We have also shown how the serotonin system adapts to these changes in the behaviour, even when the experiences occur in adulthood. Experiences during early life, before or after birth and/or during adolescence might have an even stronger impact on these neurochemical systems. 172
GENERAL
DISCUSSION
PLASTICITY VS. RIGIDITY: DEFINITION OF PATHOLOGY? Environmental challenges might have a very different impact on different individuals. In the SAL line, a strong genetic component is responsible for their aggressiveness and the development of violence. Embryo-transfer and cross-fostering studies show that the perinatal environment is not very relevant in determining their aggressive phenotype (Sluyter et al., 1996c). An aggression-reinforcing repeated social experience exaggerates the aggressive behaviour towards a routine-like saturation level and changes the molecular machinery of the SAL mice in a way that might reduce its adaptive capacity. One might expect that when the adaptive capacity is reduced to its minimum, the individual cannot cope with a challenge and might develop pathology. The next challenge for neurobiologists is to unravel the molecular mechanisms that are involved in determining this behavioural rigidity.
CONCLUDING REMARKS AND PERSPECTIVES In this thesis I have provided for the first time a valid mouse model for human violence, which is represented by the SAL genetic selection line obtained from wild house mice from Groningen, the Netherlands, and is corroborated by repeated social winning experience. I advise preclinical studies to use the SAL line, together with its counterpart, the LAL line, in genetic, pharmacological and neuro-developmental studies to investigate new methods for prevention and intervention of human violence. I suggest first a replication of the mouse lines from a new wild population of house mice, to confirm the results of the selection process. If successful, the SAL and LAL lines should be genotyped for candidate genes, including candidate genes for proteins of neurochemical systems that play a role in aggression and violence. Further, at the neuro-developmental level, studies should aim at early environmental manipulation paradigms such as communal nesting, early adverse stressors and psychopharmacological treatments, including serotonergic drugs. Such information would be highly valuable not only for developing intervention strategies to prevent/suppress human violence but also for the guidance of public and judicial policies to deal with aggression and violence.
173
References
REFERENCES
A Adams, B., Fitch, T., Chaney, S., and Gerlai, R., 2002. Altered performance characteristics in cognitive tasks: comparison of the albino ICR and CD1 mouse strains. Behav Brain Res. 133, 351. Albert, D.J., Walsh, M.L., and White, R., 1985. Mouse killing induced by para-chlorophenylalanine injections or septal lesions but not olfactory bulb lesions is similar to that of fooddeprived spontaneous killers. Behavioural Neuroscience. 99, 546-554. Albert, P.R. and Lemonde, S., 2004. 5-HT1A receptors, gene repression, and depression: guilt by association. Neuroscientist. 10, 575-593. Archer, J., 2006. Testosterone and human aggression: an evaluation of the challenge hypothesis. Neurosci Biobehav Rev. 30, 319-345. Arnett, P.A., 1997. Autonomic responsivity in psychopaths: a critical review and theoretical proposal. Clinical Psychology Review. 17, 903-936. Arnqvist, G. and Henriksson, S., 1997. Sexual cannibalism in the fishing spider and a model for the evolution of sexual cannibalism based on genetic constraints. Evolutionary Ecology. 11, 255-273. Atkinson, G., Coldwells, A., Reilly, T., and Waterhouse, J., 1993. A comparison of circadian rhythms in work performance between physically active and inactive subjects. Ergonomics. 36, 273-281. Audero, E., Coppi, E., Mlinar, B., Rossetti, T., Caprioli, A., Banchaabouchi, M.A., Corradetti, R., and Gross, C., 2008. Sporadic autonomic dysregulation and death associated with excessive serotonin autoinhibition. Science. 321, 130-133. Azmitia, E.C., Gannon, P.J., Kheck, N.M., and Whitaker-Azmitia, P.M., 1996. Cellular localization of the 5-HT1A receptor in primate brain neurons and glial cells. Neuropsychopharmacology. 14, 35-46. Aznavour, N., Rbah, L., Riad, M., Reilhac, A., Costes, N., Descarries, L., and Zimmer, L., 2006. A PET imaging study of 5-HT(1A) receptors in cat brain after acute and chronic fluoxetine treatment. Neuroimage. 33, 834-842. B Balkema, G.W. and Dräger, U.C., 1991. Impaired visual thresholds in hypopigmented animals. Vis neurosci. 6, 577-585. Bannai, M., Fish, E.W., Faccidomo, F., and Miczek, K.A., 2007. Anti-aggressive effects of agonists at 5-HT1B receptors in the dorsal raphe nucleus of mice. Psychopharmacology. 193, 295304. Bard, P., 1928. A diencephalic mechanism for the expression of rage with special reference to the sympathetic nervous system. Am J Physiol. 84, 490-515. Bartolomucci, A., Palanza, P., Costoli, T., Savani, E., Laviola, G., Parmigiani, S., and Sgoifo, A., 2003. Chronic psychosocial stress persistently alters autonomic function and physical activity in mice. Physiol Behav. 80, 57-67. Bauer, D. and Gariepy, J.L., 2001. The functions of freezing in the social interactions of juvenile high- and low-aggressive mice. Aggressive Behaviour. 27, 463-475. Bell, C., Abrams, J., and Nutt, D., 2001. Tryptophan depletion and its implications for psychiatry. Br J Psychiatry. 178, 399-405. Bell, R. and Hobson, H., 1994. 5-HT1A receptor influences on rodent social and agonistic behaviour: a review and empirical study. Neurosci Biobehav Rev. 18, 325-338. Belozertseva, I.V. and Bespalov, A.Y., 1999. Effects of NMDA receptor channel blockade on aggression in isolated male mice. Aggressive Behaviour. 25, 381-396. Benus, R.F., Koolhaas, J.M., and van Oortmerssen, G.A., 1987. Individual-differences in behavioural reaction to a changing environment in mice and rats. Behaviour. 100, 105-122. Benus, R.F., Koolhaas, J.M., and van Oortmerssen, G.A., 1988. Aggression and adaptation to the light-dark cycle: role of intrinsic and extrinsic control. Physiol Behav. 43, 131-137.
176
REFERENCES
Benus, R.F., Bohus, B., Koolhaas, J., and Van Oortmerssen G., 1989. Behavioural strategies of aggressive and non-aggressive male mice in active shock avoidance. Behavioural Processes. 20, 1-12. Benus, R.F., Bohus, B., Koolhaas, J., and van Oortmerssen, G.A., 1990a. Behavioural strategies of aggressive and non-aggressive male mice in response to inescapable shock. Behavioural Processes. 21, 127-141. Benus, R.F., Den Daas, S.J., Koolhaas, J.M., and van Oortmerssen, G.A., 1990b. Routine formation and flexibility in social and nonsocial behaviour of aggressive and nonaggressive malemice. Behaviour. 112, 176-193. Benus, R.F., Bohus, B., Koolhaas, J.M., and van Oortmerssen, G.A., 1991a. Behavioural differences between artificially selected aggressive and non-aggressive mice: response to apomorphine. Behav Brain Res. 43, 203-208. Benus, R.F., Bohus, B., Koolhaas, J.M., and van Oortmerssen, G.A., 1991b. Heritable variation for aggression as a reflection of individual coping strategies. Experientia. 47, 1008-1019. Benus, R.F. and Rondigs, M., 1996. Patterns of maternal effort in mouse lines bidirectionally selected for aggression. Animal Behaviour. 51, 67-75. Benus, R.F. and Rondigs, M., 1997. The influence of the postnatal maternal environment in accounting for differences in aggression and behavioural strategies in Mus domesticus. Behaviour. 134, 623-641. Berman, M.E., Tracy, J.I., and Coccaro, E.F., 1997. The serotonin hypothesis of aggression revisited. Clin Psychol Rev. 17, 651-665. Bernard, V., Levey, A.I., and Bloch, B., 1999. Regulation of the subcellular distribution of m4 muscarinic acetylcholine receptors in striatal neurons in vivo by the cholinergic environment: evidence for regulation of cell surface receptors by endogenous and exogenous stimulation. J Neurosci. 19, 10237-10249. Bernard, V., Decossas, M., Liste, I., and Bloch, B., 2006. Intraneuronal trafficking of G-proteincoupled receptors in vivo. Trends Neurosci. 29, 140-147. Berthold, A.A., 1849. Transplantation der hoden. Archiv fuer Anatomie, Physiologie und Wissenschaftliche Medicin, 16:42-46. Bettencourt, B.A., Talley, A., Benjamin, A.J., and Valentine, J., 2006. Personality and aggressive behaviour under provoking and neutral conditions: A meta-analytic review. Psychological Bulletin. 132, 751-777. Biggio, G., Fadda, F., Fanni, P., Tagliamonte, A., and Gessa, G.L., 1974. Rapid depletion of serum tryptophan, brain tryptophan, serotonin and 5-hydroxyindoleacetic acid by a tryptophan-free diet. Life Sci. 14, 1321-1329. Bill, D.J., Knight, M., Forster, E.A., and Fletcher, A., 1991. Direct evidence for an important species difference in the mechanism of 8-OH-DPAT-induced hypothermia. Br J Pharmacol. 103, 1857-1864. Bjork, L., Hook, B.B., Nelson, D.L., Anden, N.E., and Hacksell, U., 1989. Resolved N,N-dialkylated 2-amino-8-hydroxytetralins: stereoselective interactions with 5-HT1A receptors in the brain. J Med Chem. 32, 779-783. Bjork, J.M., Dougherty, D.M., Moeller, F.G., Cherek, D.R., and Swann A.C., 1999. The effects of tryptophan depletion and loading on laboratory aggression in men: time course and a food-restricted control. Psychopharmacology. 142, 24-30. Bjork, J.M., Dougherty, D.M., Moeller, F.G., and Swann, A.C., 2000. Differential behavioural effects of plasma tryptophan depletion and loading in aggressive and nonaggressive men. Neuropsychopharmacology. 22, 357-369. Blair, R.J., 2004. The roles of orbital frontal cortex in the modulation of antisocial behaviour. Brain Cogn. 55, 198-208. Blanchard, D.C. and Blanchard, R.J., 2003. What can animal aggression research tell us about human aggression? Hormones and Behaviour. 44, 171-177.
177
REFERENCES
Blanchard, R.J. and Blanchard, D.C., 1977. Aggressive behaviour in the rat. Behav Biol. 21, 197-224. Blier, P., Pineyro, G., el, M.M., Bergeron, R., and de, M.C., 1998. Role of somatodendritic 5-HT autoreceptors in modulating 5-HT neurotransmission. Ann N Y Acad Sci. 861, 204-216. Blier, P., Seletti, B., Gilbert, F., Young, S.N., and Benkelfat, C., 2002. Serotonin 1A receptor activation and hypothermia in humans: lack of evidence for a presynaptic mediation. Neuropsychopharmacology. 27, 301-308. Bloch, B., Dumartin, B., and Bernard, V., 1999. In vivo regulation of intraneuronal trafficking of G protein-coupled receptors for neurotransmitters. Trends Pharmacol Sci. 20, 315-319. Boadle-Biber, M.C., 1993. Regulation of serotonin synthesis. Prog Biophys Mol Biol. 60, 1-15. Bond, A.J., Wingrove, J., and Critchlow, D.G., 2001. Tryptophan depletion increases aggression in women during the premenstrual phase. Psychopharmacology (Berl). 156, 477-480. Boudreau, A.C., Reimers, J.M., Milovanovic, M., and Wolf, M.E., 2007. Cell surface AMPA receptors in the rat nucleus accumbens increase during cocaine withdrawal but internalize after cocaine challenge in association with altered activation of mitogen-activated protein kinases. J Neurosci. 27, 10621-10635. Bouwknecht, J.A., Hijzen, T.H., van der Gugten J., Maes, R.A., Hen, R., and Olivier, B., 2001. Absence of 5-HT(1B) receptors is associated with impaired impulse control in male 5HT(1B) knockout mice. Biol Psychiatry. 49, 557-568. Brain, P.F. Differentiating types of attack and defense in rodents. In: P.F.Brain, D.Benton (Eds.), Multidisciplinary approaches to aggression research., Elsevier/North Holland Biomedical Press, 1981, pp. 53-77. Breese, G.R. and Cooper, B.R., 1975. Behavioural and biochemical interactions of 5,7-dihydroxy-tryptamine with various drugs when administered intracisternally to adult and developing rats. Brain Research. 98, 517-527. Broidy, L.M., Nagin, D.S., Tremblay, R.E., Bates, J.E., Brame, B., Dodge, K.A., Fergusson, D., Horwood, J.L., Loeber, R., Laird, R., Lynam, D.R., Moffitt, T.E., Pettit, G.S., and Vitaro, F., 2003. Developmental trajectories of childhood disruptive behaviours and adolescent delinquency: A six-site, cross-national study. Developmental Psychology. 39, 222-245. Bronikowski, A.M., Carter, P.A., Swallow, J.G., Girard, I.A., Rhodes, J.S., and Garland, T.J., 2001. Open-field behaviour of house mice selectively bred for high voluntary wheelrunning. Behaviour Genetics. 31, 309-316. Brown, G.L., Goodwin, F.K., Ballenger, J.C., Goyer, P.F., and Major, L.F., 1979. Aggression in humans correlates with cerebrospinal fluid amine metabolites. Psychiatry Res. 1, 131-139. Brown, G.L., Ebert, M.H., Goyer, P.F., Jimerson, D.C., Klein, W.J., Bunney, W.E., and Goodwin, F.K., 1982. Aggression, suicide, and serotonin: relationships to CSF amine metabolites. Am J Psychiatry. 139, 741-746. Brown, G.L. and Linnoila, M.I., 1990. CSF serotonin metabolite (5-HIAA) studies in depression, impulsivity, and violence. J Clin Psychiatry. 51 Suppl, 31-41. Brunner, H.G., Nelen, M., Breakefield, X.O., Ropers, H.H., 1993. Abnormal behaviour associated with a point mutation in the structural genefor monoaminoxidase A. Science. 262: 578-80. C Cahir, M., Ardis, T., Reynolds, G.P., and Cooper, S.J., 2007. Acute and chronic tryptophan depletion differentially regulate central 5-HT1A and 5-HT2A receptor binding in the rat. Psychopharmacology (Berl). 190, 497-506. Cairns, R., MacCombie, D., and Hood, K., 1983. A developmental-genetic analysis of aggressive behaviour in mice: I. Behavioural outcomes. J Comp Psychol. 97, 69-89. Campbell, J.C., 2002. Health consequences of intimate partner violence. Lancet. 359, 1331-1336. Caramaschi, D., de Boer, S.F., and Koolhaas, J.M., 2007. Differential role of the 5-HT1A receptor in aggressive and non-aggressive mice: an across-strain comparison. Physiology & behaviour. 90, 590-601.
178
REFERENCES
Caramaschi, D., de Boer, S.F., De Vries, H., and Koolhaas, J.M., 2008a. Development of violence in mice through repeated victory along with changes in prefrontal cortex neurochemistry. Behav Brain Res. 189, 263-272. Caramaschi, D., de Boer, S.F., and Koolhaas, J.M., 2008b. Is hyper-aggressiveness associated with physiological hypoarousal? A comparative study on mouse lines selected for high and low aggressiveness. Physiol Behav. 95, 591-598. Carrel, D., Hamon, M., and Darmon, M., 2006. Role of the C-terminal di-leucine motif of 5HT1A and 5-HT1B serotonin receptors in plasma membrane targeting. J Cell Sci. 119, 42764284. Carrel, D., Masson, J., Al, A.S., Capra, C.B., Lenkei, Z., Hamon, M., Emerit, M.B., and Darmon, M., 2008. Targeting of the 5-HT1A serotonin receptor to neuronal dendrites is mediated by Yif1B. J Neurosci. 28, 8063-8073. Carta, M., Fadda, F., and Stancampiano, R., 2006. Tryptophan-deficient diet increases the neurochemical and behavioural response to amphetamine. Brain Res. 1094, 86-91. Cases, O., Seif, I., Grimsby, J., Gaspar, P., Chen, K., Pournin, S., Muller, U., Aguet, M., Babinet, C., Shih, J.C., and ., 1995. Aggressive behaviour and altered amounts of brain serotonin and norepinephrine in mice lacking MAOA. Science. 268, 1763-1766. Caspi, A., McClay, J., Moffitt, T.E., Mill, J., Martin, J., Craig, I.W., Taylor, A., and Poulton, R., 2002. Role of genotype in the cycle of violence in maltreated children. Science. 297, 851854. Celada, P., Puig, M.V., Casanovas, J.M., Guillazo, G., and Artigas, F., 2001. Control of dorsal raphe serotonergic neurons by the medial prefrontal cortex: Involvement of serotonin-1A, GABA(A), and glutamate receptors. J Neurosci. 21, 9917-9929. Cervantes, M.C. and Delville, Y., 2007. Individual differences in offensive aggression in golden hamsters: a model of reactive and impulsive aggression? Neuroscience. 150, 511-521. Chalmers, D.T., Kwak, S.P., Mansour, A., Akil, H., and Watson, S.J., 1993. Corticosteroids regulate brain hippocampal 5-HT1A receptor mRNA expression. The journal of neuroscience. 13, 914-923. Chalmers, D.T. and Watson, S.J., 1991. Comparative anatomical distribution of 5-HT1A receptor mRNA and 5-HT1A binding in rat brain--a combined in situ hybridisation/in vitro receptor autoradiographic study. Brain Res. 561, 51-60. Chamberlain, B., Ervin, F.R., Pihl, R.O., and Young, S.N., 1987. The effect of raising or lowering tryptophan levels on aggression in vervet monkeys. Pharmacol Biochem Behav. 28, 503510. Cherek, D.R. and Lane, S.D., 1999. Effects of d,l-fenfluramine on aggressive and impulsive responding in adult males with a history of conduct disorder. Psychopharmacology (Berl). 146, 473-481. Cherek, D.R., Moeller, F.G., Khan-Dawood, F., Swann, A., and Lane, S.D., 1999. Prolactin response to buspirone was reduced in violent compared to nonviolent parolees. Psychopharmacology (Berl). 142, 144-148. Cherek, D.R. and Lane, S.D., 2001. Acute effects of D-fenfluramine on simultaneous measures of aggressive escape and impulsive responses of adult males with and without a history of conduct disorder. Psychopharmacology (Berl). 157, 221-227. Chini, B. and Parenti, M., 2004. G-protein coupled receptors in lipid rafts and caveolae: how, when and why do they go there? J Mol Endocrinol. 32, 325-338. Chuang, D.M., Kinnier, W.J., Farber, L., and Costa, E., 1980. A biochemical study of receptor internalization during beta-adrenergic receptor desensitization in frog erythrocytes. Mol Pharmacol. 18, 348-355. Cleare, A.J. and Bond, A.J., 1995. The effect of tryptophan depletion and enhancement on subjective and behavioural aggression in normal male subjects. Psychopharmacology (Berl). 118, 72-81.
179
REFERENCES
Cleare, A.J. and Bond, A.J., 2000. Ipsapirone challenge in aggressive men shows an inverse correlation between 5-HT1A receptor function and aggression. Psychopharmacology (Berl). 148, 344-349. Clotfelter, E.D., O'Hare, E.P., McNitt, M.M., Carpenter, R.E., and Summers, C.H., 2007. Serotonin decreases aggression via 5-HT1A receptors in the fighting fish Betta splendens. Pharmacol Biochem Behav. 87, 222-231. Coccaro, E.F., 1989. Central serotonin and impulsive aggression. Br J Psychiatry Suppl. 52-62. Coccaro, E.F., Kavoussi, R.J., and Hauger, R.L., 1995. Physiological responses to d-fenfluramine and ipsapirone challenge correlate with indices of aggression in males with personality disorder. Int Clin Psychopharmacol. 10, 177-179. Coccaro, E.F., Siever, L.J., Klar, H.M., Maurer, G., Cochrane, K., Cooper, T.B., Mohs, R.C., and Davis, K.L., 1989. Serotonergic studies in patients with affective and personality disorders. Correlates with suicidal and impulsive aggressive behaviour. Arch Gen Psychiatry. 46, 587599. Coccaro, E.F., Berman, M.E., Kavoussi, R.J., and Hauger, R.L., 1996. Relationship of prolactin response to d-fenfluramine to behavioural and questionnaire assessments of aggression in personality-disordered men. Biol Psychiatry. 40, 157-164. Collin, M., Backberg, M., Onnestam, K., and Meister, B., 2002. 5-HT1A receptor immunoreactivity in hypothalamic neurons involved in body weight control. Neuroreport. 13, 945-951. Cologer-Clifford, A., Simon, N.G., Lu, S.F., and Smoluk, S.A., 1997. Serotonin agonist-induced decreases in intermale aggression are dependent on brain region and receptor subtype. Pharmacol Biochem Behav. 58, 425-430. Cologer-Clifford, A., Simon, N.G., Richter, M.L., Smoluk, S.A., and Lu, S., 1999. Androgens and estrogens modulate 5-HT1A and 5-HT1B agonist effects on aggression. Physiol Behav. 65, 823-828. Compaan, J.C., de Ruiter, A.J., Koolhaas, J.M., van Oortmerssen, G.A., and Bohus, B., 1992. Differential effects of neonatal testosterone treatment on aggression in two selection lines of mice. Physiol Behav. 51, 7-10. Compaan, J.C., Buijs, R.M., Pool, C.W., de Ruiter, A.J., and Koolhaas, J.M., 1993a. Differential lateral septal vasopressin innervation in aggressive and nonaggressive male mice. Brain Res Bull. 30, 1-6. Compaan, J.C., van Wattum, G., de Ruiter, A.J., van Oortmerssen, G.A., Koolhaas, J.M., and Bohus, B., 1993b. Genetic differences in female house mice in aggressive response to sex steroid hormone treatment. Physiol Behav. 54, 899-902. Compaan, J.C., Hutchison, J.B., Wozniak, A., de Ruiter, A.J., and Koolhaas, J.M., 1994a. Brain aromatase activity and plasma testosterone levels are elevated in aggressive male mice during early ontogeny. Brain Res Dev Brain Res. 82, 185-192. Compaan, J.C., Wozniak, A., de Ruiter, A.J., Koolhaas, J.M., and Hutchison, J.B., 1994b. Aromatase activity in the preoptic area differs between aggressive and nonaggressive male house mice. Brain Res Bull. 35, 1-7. Conner, R.L., Stolk, J.M., Barchas, J.D., Dement, W.C., and Levine, S., 1970. The effect of parachlorophenylalanine (PCPA) on shock-induced fighting behaviour in rats. Physiology & behaviour. 5, 1221-1224. Contrada, R.J., Glass, D.C., Krakoff, L.R., Krantz, D.S., Kehoe, K., Isecke, W., Collins, C., and Elting, E., 1982. Effects of control over aversive stimulation and type A behaviour on cardiovascular and plasma catecholamine responses. Psychophysiology. 19, 408-419. Costantini, D., Carere, C., Caramaschi, D., and Koolhaas, J.M., 2008. Aggressive and nonaggressive personalities differ in oxidative status in selected lines of mice (Mus musculus). Biol Lett. 4, 119-122. Crowe, S.L. and Blair, R.J., 2008. The development of antisocial behaviour: what can we learn from functional neuroimaging studies? Dev Psychopathol. 20, 1145-1159.
180
REFERENCES
D D'Eath, R.B., Ormandy, E., Lawrence, A.B., Sumner, B.E., and Meddle, S.L., 2005. Residentintruder trait aggression is associated with differences in lysine vasopressin and serotonin receptor 1A (5-HT1A) mRNA expression in the brain of pre-pubertal female domestic pigs (Sus scrofa). J Neuroendocrinol. 17, 679-686. D'Souza, D.N., Zhang, Y., Garcia, F., Battaglia, G., and Van de Kar, L.D., 2004. Fluoxetineinduced changes in body weight and 5-HT1A receptor-mediated hormone secretion in rats on a tryptophan-deficient diet. Am J Physiol Regul Integr Comp Physiol. 286, R390-R397. Dalgleish, T., 2004. The emotional brain. Nat Rev Neurosci. 5, 583-589. De Almeida, R.M. and Miczek, K.A., 2002. Aggression escalated by social instigation or by discontinuation of reinforcement ("frustration") in mice: inhibition by anpirtoline: a 5-HT1B receptor agonist. Neuropsychopharmacology. 27, 171-181. De Almeida, R.M., Ferrari, P.F., Parmigiani, S., and Miczek, K.A., 2005. Escalated aggressive behaviour: dopamine, serotonin and GABA. Eur J Pharmacol. 526, 51-64. De Boer, S.F., Lesourd, M., Mocaer, E., and Koolhaas, J.M., 1999. Selective antiaggressive effects of alnespirone in resident-intruder test are mediated via 5-hydroxytryptamine1A receptors: A comparative pharmacological study with 8-hydroxy-2-dipropylaminotetralin, ipsapirone, buspirone, eltoprazine, and WAY-100635. J Pharmacol Exp Ther. 288, 1125-1133. De Boer, S.F., Lesourd, M., Mocaer, E., and Koolhaas, J.M., 2000. Somatodendritic 5-HT(1A) autoreceptors mediate the anti-aggressive actions of 5-HT(1A) receptor agonists in rats: an ethopharmacological study with S-15535, alnespirone, and WAY-100635. Neuropsychopharmacology. 23, 20-33. De Boer, S.F., van der Vegt, B., and Koolhaas, J.M., 2003. Individual variation in aggression of feral rodent strains: a standard for the genetics of aggression and violence? Behav Genet. 33, 485-501. De Boer, S.F. and Koolhaas, J.M., 2003. Defensive burying in rodents: ethology, neurobiology and psychopharmacology. Eur J Pharmacol. 463, 145-161. De Boer, S.F. and Koolhaas, J.M., 2005. 5-HT1A and 5-HT1B receptor agonists and aggression: a pharmacological challenge of the serotonin deficiency hypothesis. Eur J Pharmacol. 526, 125-139. De Bold, J.F. and Miczek, K.A., 1981. Sexual dimorphism in the hormonal control of aggressive behaviour in rats. Pharmacol Biochem Behav. 14 Suppl 1:89-93. De Kloet, E.R., Sybesma, H., and Reul, H.M., 1986. Selective control by corticosterone of serotonin1 receptor capacity in raphe-hippocampal system. Neuroendocrinology. 42, 513-521. De Ruiter, A.J., Feitsma, L.E., Keijser, J.N., Koolhaas, J.M., van Oortmerssen, G.A., and Bohus, B., 1993. Differential perinatal testosterone secretory capacity of wild house mice testes is related to aggressiveness in adulthood. Horm Behav. 27, 231-239. De Vries, G.J., 2008. Sex differences in vasopressin and oxytocin innervation of the brain. Prog Brain Res. 170, 17-27. De Vries, H., Netto, W.J., and Hanegraaf, P.L.H., 1993. Matman - A Program for the Analysis of Sociometric Matrices and Behavioural Transition Matrices. Behaviour. 125, 157-175. Della Rocca, G.J., Mukhin, Y.V., Garnovskaya, M.N., Daaka, Y., Clark, G.J., Luttrell, L.M., Lefkowitz, R.J., and Raymond, J.R., 1999. Serotonin 5-HT1A receptor-mediated Erk activation requires calcium/calmodulin-dependent receptor endocytosis. J Biol Chem. 274, 4749-4753. Depino, A.M. and Gross, C., 2007. Simultaneous assessment of autonomic function and anxiety-related behaviour in BALB/c and C57BL/6 mice. Behav Brain Res. 177, 254-260. Digman, J.M., 1990. Personality Structure - Emergence of the 5-Factor Model. Annual Review of Psychology. 41, 417-440. Dingemanse, N.J., Wright, J., Kazem, A.J.N., Thomas, D.K., Hickling, R., and Dawnay, N., 2007. Behavioural syndromes differ predictably between 12 populations of three-spined stickleback. Journal of Animal Ecology. 76, 1128-1138.
181
REFERENCES
Dodman, N.H., Donnelly, R., Shuster, L., Mertens, P., Rand, W., and Miczek, K., 1996. Use of fluoxetine to treat dominance aggression in dogs. J Am Vet Med Assoc. 209, 1585-1587. Dougherty, D.M., Moeller, F.G., Bjork, J.M., and Marsh, D.M., 1999a. Plasma L-tryptophan depletion and aggression. Adv Exp Med Biol. 467, 57-65. Dougherty, D.M., Bjork, J.M., Marsh, D.M., and Moeller, F.G., 1999b. Influence of trait hostility on tryptophan depletion-induced laboratory aggression. Psychiatry Research. 88, 227-232. Dumartin, B., Caille, I., Gonon, F., and Bloch, B., 1998. Internalization of D1 dopamine receptor in striatal neurons in vivo as evidence of activation by dopamine agonists. J Neurosci. 18, 1650-1661. E Ebenezer, I.S., Arkle, M.J., and Tite, R.M., 2007. 8-Hydroxy-2-(di-n-propylamino)-tetralin inhibits food intake in fasted rats by an action at 5-HT1A receptors. Methods Find Exp Clin Pharmacol. 29, 269-272. Ewalds-Kvist, S.B., Selander, R.K., and Sandnabba, N.K., 1997. Sex-related coping responses in mice selectively bred for aggression. Percept Mot Skills. 84, 911-914. F Fadda, F., Cocco, S., and Stancampiano, R., 2000. A physiological method to selectively decrease brain serotonin release. Brain Res Brain Res Protoc. 5, 219-222. Fairbanks, L.A., Melega, W.P., Jorgensen, M.J., Kaplan, J.R., and McGuire, M.T., 2001. Social impulsivity inversely associated with CSF 5-HIAA and fluoxetine exposure in vervet monkeys. Neuropsychopharmacology. 24, 370-378. Feldker, D.E., Datson, N.A., Veenema, A.H., Meulmeester, E., de Kloet, E.R., and Vreugdenhil, E., 2003a. Serial analysis of gene expression predicts structural differences in hippocampus of long attack latency and short attack latency mice. Eur J Neurosci. 17, 379-387. Feldker, D.E., Datson, N.A., Veenema, A.H., Proutski, V., Lathouwers, D., de Kloet, E.R., and Vreugdenhil, E., 2003b. GeneChip analysis of hippocampal gene expression profiles of short- and long-attack-latency mice: technical and biological implications. J Neurosci Res. 74, 701-716. Ferguson, S.S., 2001. Evolving concepts in G protein-coupled receptor endocytosis: the role in receptor desensitization and signaling. Pharmacol Rev. 53, 1-24. Fernandez, X., Meunier-Salaun, M.C., and Mormede, P., 1994. Agonistic behaviour, plasma stress hormones, and metabolites in response to dyadic encounters in domestic pigs: interrelationships and effect of dominance status. Physiol Behav. 56, 841-847. Ferrari, P.F., Palanza, P., Parmigiani, S., DE Almeida, R.M., and Miczek, K.A., 2005. Serotonin and aggressive behaviour in rodents and nonhuman primates: predispositions and plasticity. Eur J Pharmacol. 526, 259-273. Ferris, C.F., Axelson, J.F., Martin, A.M., and Roberge, L.F., 1989. Vasopressin immunoreactivity in the anterior hypothalamus is altered during the establishment of dominant/subordinate relationships between hamsters. Neuroscience. 29, 675-683. Ferris, C.F. and Delville, Y., 1994. Vasopressin and serotonin interactions in the control of agonistic behaviour. Psychoneuroendocrinology. 19, 593-601. Ferris, C.F., Stolberg, T., Kulkarni, P., Murugavel, M., Blanchard, R., Blanchard, D.C., Febo, M., Brevard, M., Simon, N.G., 2008. Imaging the neural circuitry and chemical control of aggressive motivation. BMC Neurosci. Nov 13: 9-111. Fish, E.W., Faccidomo, F., De Bold, J.F., and Miczek, K.A., 2001. Alcohol, allopregnanolone and aggression in mice. Psychopharmacology. 153, 473-483. Foley, D.L., Eaves, L.J., Wormley, B., Silberg, J.L., Maes, H.H., Kuhn, J., and Riley, B., 2004. Childhood adversity, monoamine oxidase a genotype, and risk for conduct disorder. Arch Gen Psychiatry. 61, 738-744.
182
REFERENCES
G Gammie, S.C., Hasen, N.S., Rhodes, J.S., Girard, I., and Garland, T., 2003. Predatory aggression, but not maternal or intermale aggression, is associated with high voluntary wheelrunning behaviour in mice. Hormones and Behaviour. 44, 209-221. Garattini, S., 1967. Isolation, aggressiveness and brain 5-hydroxytryptamine turnover. Journal of Pharmacy and Pharmacology. 19, 338-339. Gardier, A.M., Gruwez, B., Trillat, A.C., Jacquot, C., Hen, R., and Bourin, M., 2001. Interaction between 5-HT(1A) and 5-HT(1B) receptors: effects of 8-OH-DPAT-induced hypothermia in 5-HT(1B) receptor knockout mice. Eur J Pharmacol. 421, 171-175. Gariepy, J.L., Hood, K., and Cairns, R., 1988. A developmental-genetic analysis of aggressive behaviour in mice (Mus musculus): III. Behavioural mediation by heightened reactivity or immobility? Journal of Comparative Psychology. 102, 392-399. Gariepy JL, Lewis MH, and Cairns RB Genes, neurobiology and aggression: Time frames and functions of social behaviour in adaptation. In: Stoff DM, Cairns RB (Eds.), Aggression and violence: neurobiological, biosocial, and genetic perspectives., Erlbaum L, New York, 1996, pp. 41-63. Gariepy, J.L., Gendreau, P.L., Cairns, R.B., and Lewis, M.H., 1998. D-1 dopamine receptors and the reversal of isolation-induced behaviours in mice. Behavioural Brain Research. 95, 103111. Gariepy, J.L., Rodriguiz, R.M., and Jones, B.C., 2002. Handling, genetic and housing effects on the mouse stress system, dopamine function, and behaviour. Pharmacol Biochem Behav. 73, 7-17. Gershon, M.D., Wade, P.R., Kirchgessner, A.L., and Tamir, H., 1990. 5-HT receptor subtypes outside the central nervous system. Roles in the physiology of the gut. Neuropsychopharmacology. 3, 385-395. Giacalone, E., Tansella, M., Valzelli, L., and Garattini, S., 1968. Brain serotonin metabolism in isolated aggressive mice. Biochem Pharmacol. 17, 1315-1327. Gibbons, J.L., Barr, G.A., Bridger, W.H., and Fryer Leibowitz, S., 1978. Effects of parachlorophenylalanine and 5-hydroxytryptophan on mouse killing behaviour in killer rats. Pharmacology Biochemistry and Behaviour. 9, 91-98. Gibbons, J.L., Barr, G.A., Bridger, W.H., and Leibowitz, S.F., 1979. Manipulations of dietary tryptophan: effects on mouse killing and brain serotonin in the rat. Brain Res. 169, 139-153. Goodwin, G.M., De Souza, R.J., and Green, A.R., 1985. The pharmacology of the hypothermic response in mice to 8-hydroxy-2-(di-n-propylamino)tetralin (8-OH-DPAT). A model of presynaptic 5-HT1 function. Neuropharmacology. 24, 1187-1194. Goodwin, G.M., De Souza, R.J., Green, A.R., and Heal, D.J., 1987. The pharmacology of the behavioural and hypothermic responses of rats to 8-hydroxy-2-(di-n-propylamino)tetralin (8-OH-DPAT). Psychopharmacology (Berl). 91, 506-511. Granger, D. and Hood, K., 1997. Effects of peripheral immune activation on social behaviour and adrenocortical activity in.. Aggressive Behaviour. 23, 93-105. Granger, D.A., Hood, K.E., Ikeda, S.C., Reed, C.L., and Block, M.L., 1996. Neonatal endotoxin exposure alters the development of social behaviour and the hypothalamic-pituitaryadrenal axis in selectively bred mice. Brain Behav Immun. 10, 249-259. Granger, D.A., Hood, K.E., Dreschel, N.A., Sergeant, E., and Likos, A., 2001. Developmental effects of early immune stress on aggressive, socially reactive, and inhibited behaviours. Dev Psychopathol. 13, 599-610. Gregg, T.R. and Siegel, A., 2001. Brain structures and neurotransmitters regulating aggression in cats: implications for human aggression. Prog Neuropsychopharmacol Biol Psychiatry. 25, 91-140. Groothuis, T.G.G. and Carere, C., 2005. Avian personalities: characterization and epigenesis. Neuroscience and Biobehavioural Reviews. 29, 137-150.
183
REFERENCES
Gross, C., Zhuang, X., Stark, K., Ramboz, S., Oosting, R., Kirby, L., Santarelli, L., Beck, S., and Hen, R., 2002. Serotonin1A receptor acts during development to establish normal anxietylike behaviour in the adult. Nature. 416, 396-400. H Haberstock-Debic, H., Wein, M., Barrot, M., Colago, E.E., Rahman, Z., Neve, R.L., Pickel, V.M., Nestler, E.J., von Zastrow M., and Svingos, A.L., 2003. Morphine acutely regulates opioid receptor trafficking selectively in dendrites of nucleus accumbens neurons. J Neurosci. 23, 4324-4332. Hajos, M., Hajos-Korcsok, E., and Sharp, T., 1999. Role of the medial prefrontal cortex in 5HT1A receptor-induced inhibition of 5-HT neuronal activity in the rat. Br J Pharmacol. 126, 1741-1750. Halasz, J., Toth, M., Kallo, I., Liposits, Z., and Haller, J., 2006. The activation of prefrontal cortical neurons in aggression--a double labeling study. Behav Brain Res. 175, 166-175. Haller, J., Millar, S., and Kruk, M.R., 1998. Mineralocorticoid receptor blockade inhibits aggressive behaviour in male rats. Stress. 2, 201-207. Haller, J., Millar, S., van de Schraaf, J., de Kloet, R.E., and Kruk, M.R., 2000. The active phaserelated increase in corticosterone and aggression are linked. J Neuroendocrinol. 12, 431436. Haller, J., van de Schraaf, J., and Kruk, M.R., 2001. Deviant forms of aggression in glucocorticoid hyporeactive rats: a model for 'pathological' aggression? J Neuroendocrinol. 13, 102-107. Haller, J., Halasz, J., Mikics, E., and Kruk, M.R., 2004. Chronic glucocorticoid deficiencyinduced abnormal aggression, autonomic hypoarousal, and social deficit in rats. Journal of neuroendocrinology. 16, 550-557. Haller, J. and Kruk, M.R., 2006. Normal and abnormal aggression: human disorders and novel laboratory models. Neurosci Biobehav Rev. 30, 292-303. Haller, J., Toth, M., Halasz, J., and de Boer, S.F., 2006. Patterns of violent aggression-induced brain c-fos expression in male mice selected for aggressiveness. Physiol Behav. 88, 173-182. Hamon, M., Fattaccini, C.M., Adrien, J., Gallissot, M.C., Martin, P., and Gozlan, H., 1988. Alterations of central serotonin and dopamine turnover in rats treated with ipsapirone and other 5-hydroxytryptamine1A agonists with potential anxiolytic properties. J Pharmacol Exp Ther. 246, 745-752. Hautojarvi, S. and Lagerspetz, K., 1968. The effects of socially-induced aggressiveness or nonaggressiveness on the sexual behaviour of inexperienced male mice. Scand J Psychol. 9, 45-49. Hedlund, P.B., Kelly, L., Mazur, C., Lovenberg, T., Sutcliffe, J.G., and Bonaventure, P., 2004. 8OH-DPAT acts on both 5-HT1A and 5-HT7 receptors to induce hypothermia in rodents. Eur J Pharmacol. 487, 125-132. Hennig, J., Reuter, M., Netter, P., Burk, C., and Landt, O., 2005. Two types of aggression are differentially related to serotonergic activity and the A779C TPH polymorphism. Behav Neurosci. 119, 16-25. Hensbroek, R.A., Sluyter, F., Guillot, P.V., van Oortmerssen, G.A., and Crusio, W.E., 1995. Y chromosomal effects on hippocampal mossy fiber distributions in mice selected for aggression. Brain Res. 682, 203-206. Higley, J.D., Mehlman, P.T., Poland, R.E., Taub, D.M., Vickers, J., Suomi, S.J., and Linnoila, M., 1996. CSF testosterone and 5-HIAA correlate with different types of aggressive behaviours. Biol Psychiatry. 40, 1067-1082. Hillegaart, V., 1991. Effects of local application of 5-HT and 8-OH-DPAT into the dorsal and median raphe nuclei on core temperature in the rat. Psychopharmacology (Berl). 103, 291296. Hjorth, S., 1985. Hypothermia in the rat induced by the potent serotoninergic agent 8-OHDPAT. Journal of neural transmission. 61, 131-135.
184
REFERENCES
Hjorth, S., Suchowski, C.S., and Galloway, M.P., 1995. Evidence for 5-HT autoreceptor-mediated, nerve impulse-independent, control of 5-HT synthesis in the rat brain. Synapse. 19, 170-176. Hochberg, Y., 1988. A Sharper Bonferroni Procedure for Multiple Tests of Significance. Biometrika. 75, 800-802. Hogg, S., Hof, M., Wurbel, H., Steimer, T., de Ruiter, A., Koolhaas, J., and Sluyter, F., 2000. Behavioural profiles of genetically selected aggressive and nonaggressive male wild house mice in two anxiety tests. Behav Genet. 30, 439-446. Hole, K., Johnson, G.E., and Berge, O.G., 1977. 5,7-Dihydroxytryptamine lesions of the ascending 5-hydroxytryptamine pathways: habituation, motor activity and agonistic behaviour. Pharmacol Biochem Behav. 7, 205-210. Holmes, A., Murphy, D.L., and Crawley, J.N., 2002. Reduced aggression in mice lacking the serotonin transporter. Psychopharmacology (Berl). 161, 160-167. Homberg, J.R., Pattij, T., Janssen, M.C., Ronken, E., de Boer, S.F., Schoffelmeer, A.N., and Cuppen, E., 2007. Serotonin transporter deficiency in rats improves inhibitory control but not behavioural flexibility. Eur J Neurosci. 26, 2066-2073. Hood, K. and Cairns, R., 1989. A Developmental-Genetic Analysis of Aggressive Behaviour in Mice: IV. Aggressive Behaviour. 15, 361-380. Hood, K.E. and Cairns, R.B., 1988. A developmental-genetic analysis of aggressive behaviour in mice. II. Cross-sex inheritance. Behav Genet. 18, 605-619. Hood, K.E., Dreschel, N.A., and Granger, D.A., 2003. Maternal behaviour changes after immune challenge of neonates with developmental effects on adult social behaviour. Dev Psychobiol. 42, 17-34. Hood, K.E. and Quigley, K.S., 2008. Exploratory behaviour in mice selectively bred for developmental differences in aggressive behaviour. Dev Psychobiol. 50, 32-47. Horikawa, K. and Shibata, S., 2004. Phase-resetting response to (+)8-OH-DPAT, a serotonin 1A/7 receptor agonist, in the mouse in vivo. Neurosci Lett. 368, 130-134. Hsu, Y., Earley, R.L., and Wolf, L.L., 2006. Modulation of aggressive behaviour by fighting experience: mechanisms and contest outcomes. Biological reviews of the Cambridge Philosophical Society. 81, 33-74. I Ieni, J.R. and Thurmond, J.B., 1985. Maternal aggression in mice: Effects of treatments with PCPA, 5-HTP and 5-HT receptor antagonists. European Journal of Pharmacology. 111, 211220. J Jones, A.B., Barchas, J.D., and Eichelman, B., 1976. Taming effects of p-chlorophenylalanine on the aggressive behaviour of septal rats. Pharmacology Biochemistry and Behaviour. 4, 397400. Joppa, M.A., Rowe, R.K., and Meisel, R.L., 1997. Effects of serotonin 1A or 1B receptor agonists on social aggression in male and female Syrian hamsters. Pharmacol Biochem Behav. 58, 349-353. K Kalipatnapu, S. and Chattopadhyay, A., 2005. Membrane organization of the human serotonin(1A) receptor monitored by detergent insolubility using GFP fluorescence. Mol Membr Biol. 22, 539-547. Kang, J., Shi, Y., Xiang, B., Qu, B., Su, W., Zhu, M., Zhang, M., Bao, G., Wang, F., Zhang, X., Yang, R., Fan, F., Chen, X., Pei, G., and Ma, L., 2005. A nuclear function of beta-arrestin1 in GPCR signaling: regulation of histone acetylation and gene transcription. Cell. 123, 833847. Kantak, K.M., Hegstrand, L.R., and Eichelman, B., 1980a. Dietary tryptophan modulation and aggressive behaviour in mice. Pharmacol Biochem Behav. 12, 675-679.
185
REFERENCES
Kantak, K.M., Hegstrand, L.R., Whitman, J., and Eichelman, B., 1980b. Effects of dietary supplements and a tryptophan-free diet on aggressive behaviour in rats. Pharmacol Biochem Behav. 12, 173-179. Kantak, K.M., Hegstrand, L.R., and Eichelman, B., 1981. Dietary tryptophan reversal of septal lesion and 5,7-DHT lesion elicited shock-induced fighting. Pharmacology Biochemistry and Behaviour. 15, 343-350. Kantak, K.M., 1981. Facilitation of shock-induced fighting following intraventricular 5,7-dihydroxytryptamine and 6-hydroxydopa. Psychopharmacology. 74, 157-160. Kawai, K., Yokota, N., and Yamawaki, S., 1994. Effect of chronic tryptophan depletion on the circadian rhythm of wheel-running activity in rats. Physiol Behav. 55, 1005-1013. Keele, N.B., 2001. Phenytoin inhibits isolation-induced aggression specifically in rats with low serotonin. Neuroreport. 12, 1107-1112. Keith, D.E., Anton, B., Murray, S.R., Zaki, P.A., Chu, P.C., Lissin, D.V., Monteillet-Agius, G., Stewart, P.L., Evans, C.J., and von Zastrow M., 1998. mu-Opioid receptor internalization: opiate drugs have differential effects on a conserved endocytic mechanism in vitro and in the mammalian brain. Mol Pharmacol. 53, 377-384. Kia, H.K., Brisorgueil, M.J., Hamon, M., Calas, A., and Verge, D., 1996a. Ultrastructural localization of 5-hydroxytryptamine1A receptors in the rat brain. J Neurosci Res. 46, 697-708. Kia, H.K., Miquel, M.C., Brisorgueil, M.J., Daval, G., Riad, M., El, M.S., Hamon, M., and Verge, D., 1996b. Immunocytochemical localization of serotonin1A receptors in the rat central nervous system. J Comp Neurol. 365, 289-305. Kirkman, T. W. Statistics to use. http://www.physics.csbsju.edu/stats/ (19-11-2007) Koenig, J.A. and Edwardson, J.M., 1997. Endocytosis and recycling of G protein-coupled receptors. Trends Pharmacol Sci. 18, 276-287. Koolhaas, J.M., Schuurman, T., and Wiepkema, P.R., 1980. The organization of intraspecific agonistic behaviour in the rat. Prog Neurobiol. 15, 247-268. Koolhaas, J.M., de Boer, S.F., De Rutter, A.J., Meerlo, P., and Sgoifo, A., 1997. Social stress in rats and mice. Acta Physiol Scand Suppl. 640, 69-72. Koolhaas, J.M., Korte, S.M., de Boer, S.F., van der Vegt, B., Van Reenen, C.G., Hopster, H., de Jong, I., Ruis, M.A., and Blokhuis, H.J., 1999. Coping styles in animals: current status in behaviour and stress-physiology. Neurosci Biobehav Rev. 23, 925-935. Koolhaas, J.M., de Boer, S.F., Buwalda, B., and van, R.K., 2007. Individual variation in coping with stress: a multidimensional approach of ultimate and proximate mechanisms. Brain Behav Evol. 70, 218-226. Koolhaas, J.M., de Boer S., F., and Buwalda, B. Ch. 86: Neurobiology of offensive aggression. In: Encyclopedia of behavioural neuroscience, Elsevier, in press. Korte, S.M., Meijer, O.C., de Kloet, E.R., Buwalda, B., Keijser, J., Sluyter, F., Van Oortmerssen G., and Bohus, B., 1996. Enhanced 5-HT1A receptor expression in forebrain regions of aggressive house mice. Brain Res. 736, 338-343. Krug, E.G., Dalberg, L.L., Mercy, J.A., Zwi, A.B., and Lozano, R., World report on violence and health. In: Vol. Geneva, World Health Organization, Geneva, 2002. Kudryavtseva, N.N., Bondar, N.P., and Avgustinovich, D.F., 2004. Effects of repeated experience of aggression on the aggressive motivation and development of anxiety in male mice. Neurosci Behav Physiol. 34, 721-730. Kushwaha, N., Harwood, S.C., Wilson, A.M., Berger, M., Tecott, L.H., Roth, B.L., and Albert, P.R., 2006. Molecular determinants in the second intracellular loop of the 5-hydroxytryptamine-1A receptor for G-protein coupling. Mol Pharmacol. 69, 1518-1526. Kvist, B., 1989. Learning in mice selectively bred for high and low aggressiveness. Psychol Rep. 64, 127-130.
186
REFERENCES
L Lagerspetz, K., Genetic and social causes of aggressive behaviour in mice. In: 1961, pp. 167173. Lagerspetz, K.M.J. and Wuorinen, K., 1965. A cross-fostering experiment with mice selectively bred for aggressiveness and non-aggressiveness . Reports from the institute of psychology University Turku. 17, 1-6. Lagerspetz, K. and Hautojarvi, S., 1967. The effect of prior aggressive or sexual arousal on subsequent aggressive or sexual reactions in male mice. Scand J Psychol. 8, 1-6. Lagerspetz, K.Y., Tirri, R., and Lagerspetz, K.M., 1968. Neurochemical and endocrinological studies of mice selectively bred for aggressiveness. Scand J Psychol. 9, 157-160. Lagerspetz, K.M. and Lagerspetz, K.Y., 1971. Changes in the aggressiveness of mice resulting from selective breeding, learning and social isolation. Scand J Psychol. 12, 241-248. Lagerspetz, K.M.I. and Lagerspetz, K.Y.H., 1975. Expression of genes of aggressiveness in mice: effect of androgen on aggression and sexual behaviour in females. Aggressive Behaviour. 1, 291-296. Lagerspetz, K. and Ekqvist, K., 1978. Failure to induce aggression in inhibited and in genetically non-aggressive mice through injections of ethyl alcohol. Aggressive Behaviour. 4, 105-113. Lagerspetz, K. and Sandnabba, K., 1982. The decline of aggressiveness in male mice during group caging as determined by punishment delivered by the cage mates. Aggressive Behaviour. 8, 319-334. Lagerspetz, K.M.J. and Lagerspetz, K.Y. Genes and aggression. In: E.C.Simmel, M.Hahn (Eds.), Aggressive behaviour: Genetic and neural approach. NJ, 1983, pp. 89-101. Lai, M.K., Tsang, S.W., Francis, P.T., Esiri, M.M., Keene, J., Hope, T., and Chen, C.P., 2003. Reduced serotonin 5-HT1A receptor binding in the temporal cortex correlates with aggressive behaviour in Alzheimer disease. Brain Res. 974, 82-87. Le Francois B., Czesak, M., Steubl, D., and Albert, P.R., 2008. Transcriptional regulation at a HTR1A polymorphism associated with mental illness. Neuropharmacology. Lee, J.M. and Watanuki, S., 2007. Cardiovascular responses of Type A and Type B behaviour patterns to visual stimulation during rest, stress and recovery. J Physiol Anthropol. 26, 1-8. Lee, R. and Coccaro, E., 2001. The neuropsychopharmacology of criminality and aggression. Can J Psychiatry. 46, 35-44. Lefkowitz, R.J. and Shenoy, S.K., 2005. Transduction of receptor signals by beta-arrestins. Science. 308, 512-517. Le Marquand, D.G., Benkelfat C., Pihl, R.O., Palmour R.M., and Young, S.N., 1999. Behavioural disinhibition induced by tryptophan depletion in nonalcoholic young men with multigenerational family histories of paternal alcoholism. The American journal of psychiatry. 156, 1771-1779. Levin, B.E., Richard, D., Michel, C., and Servatius, R., 2000. Differential stress responsivity in diet-induced obese and resistant rats. Am J Physiol Regul Integr Comp Physiol. 279, R1357-R1364. Lewis, M.H., Gariepy, J.L., Gendreau, P., Nichols, D.E., and Mailman, R.B., 1994. Social reactivity and D1 dopamine receptors: studies in mice selectively bred for high and low levels of aggression. Neuropsychopharmacology. 10, 115-122. Li, Q., Wichems, C., Heils, A., Van de Kar, L.D., Lesch, K.P., and Murphy, D.L., 1999. Reduction of 5-hydroxytryptamine (5-HT)(1A)-mediated temperature and neuroendocrine responses and 5-HT(1A) binding sites in 5-HT transporter knockout mice. J Pharmacol Exp Ther. 291, 999-1007. Limson, R., Goldman, D., Roy, A., Lamparski, D., Ravitz, B., Adinoff, B., and Linnoila, M., 1991. Personality and cerebrospinal fluid monoamine metabolites in alcoholics and controls. Arch Gen Psychiatry. 48, 437-441.
187
REFERENCES
Linnoila, M., Virkkunen, M., Scheinin, M., Nuutila, A., Rimon, R., and Goodwin, F.K., 1983. Low cerebrospinal fluid 5-hydroxyindoleacetic acid concentration differentiates impulsive from nonimpulsive violent behaviour. Life Sci. 33, 2609-2614. Loeber, R. and Pardini, D., 2008. Neurobiology and the development of violence: common assumptions and controversies. Philos Trans R Soc Lond B Biol Sci. 363, 2491-2503. Lorber, M.F., 2004. Psychophysiology of aggression, psychopathy, and conduct problems: a meta-analysis. Psychol Bull. 130, 531-552. Luttgen, M., Ogren, S.O., and Meister, B., 2005. 5-HT1A receptor mRNA and immunoreactivity in the rat medial septum/diagonal band of Broca-relationships to GABAergic and cholinergic neurons. J Chem Neuroanat. 29, 93-111. M Ma, L. and Pei, G., 2007. Beta-arrestin signaling and regulation of transcription. J Cell Sci. 120, 213-218. Maas, J.W., 1962. Neurochemical differences between two strains of mice. Science. 137, 621-622. Malick, J.B. and Barnett, A., 1976. The role of serotonergic pathways in isolation-induced aggression in mice. Pharmacology Biochemistry and Behaviour. 5, 55-61. Mannoury la Cour, C., El Mestikawy, S., Hanoun, N., Hamon, M., and Lanfumey, L., 2006. Regional differences in the coupling of 5-hydroxytryptamine-1A receptors to G proteins in the rat brain. Mol Pharmacol. 70, 1013-1021. Manuck, S.B., Flory, J.D., Ferrell, R.E., Mann, J.J., and Muldoon, M.F., 2000. A regulatory polymorphism of the monoamine oxidase-A gene may be associated with variability in aggression, impulsivity, and central nervous system serotonergic responsivity. Psychiatry Res. 95, 9-23. Martin, K.F., Phillips, I., Hearson, M., Prow, M.R., and Heal, D.J., 1992. Characterization of 8OH-DPAT-induced hypothermia in mice as a 5-HT1A autoreceptor response and its evaluation as a model to selectively identify antidepressants. Br J Pharmacol. 107, 15-21. Matsuda, T., Kanda, T., Seong, Y.H., Baba, A., and Iwata, H., 1990. p-chlorophenylalanine attenuates the pituitary-adrenocortical response to 5-HT1A receptor agonists in mice. Eur J Pharmacol. 181, 295-297. Matthews, K.L., Chen, C.P., Esiri, M.M., Keene, J., Minger, S.L., and Francis, P.T., 2002. Noradrenergic changes, aggressive behaviour, and cognition in patients with dementia. Biol Psychiatry. 51, 407-416. McCloskey, M.S., Ben-Zeev, D., Lee, R., Berman, M.E., and Coccaro, E.F., 2008. Acute tryptophan depletion and self-injurious behaviour in aggressive patients and healthy volunteers. Psychopharmacology (Berl). Mehlman, P.T., Higley, J.D., Faucher, I., Lilly, A.A., Taub, D.M., Vickers, J., Suomi, S.J., and Linnoila, M., 1994. Low CSF 5-HIAA concentrations and severe aggression and impaired impulse control in nonhuman primates. Am J Psychiatry. 151, 1485-1491. Mehta, C.R. and Patel, N.R., 1983. A network algorithm for performing Fisher's exact test in r x c contingency tables. J Am Stat Assoc. 78, 427-434. Mehta, P.H. and Gosling, S.D., 2008. Bridging human and animal research: A comparative approach to studies of personality and health. Brain, Behaviour, and Immunity. 22, 651-661. Miczek, K.A., Altman, J.L., Appel, J.B., and Boggan, W.O., 1975. Para-chlorophenylalanine, serotonin and killing behaviour. Pharmacology Biochemistry and Behaviour. 3, 355-361. Miczek, K.A., Hussain, S., and Faccidomo, S., 1998. Alcohol-heightened aggression in mice: attenuation by 5-HT1A receptor agonists. Psychopharmacology (Berl). 139, 160-168. Miczek, K.A., Maxson, S.C., Fish, E.W., and Faccidomo, S., 2001. Aggressive behavioural phenotypes in mice. Behav Brain Res. 125, 167-181. Miczek, K.A., Fish, E.W., De Bold, J.F., and DE Almeida, R.M., 2002. Social and neural determinants of aggressive behaviour: pharmacotherapeutic targets at serotonin, dopamine and gamma-aminobutyric acid systems. Psychopharmacology (Berl). 163, 434-458.
188
REFERENCES
Miczek, K.A., DE Almeida, R.M., Kravitz, E.A., Rissman, E.F., de Boer, S.F., and Raine, A., 2007. Neurobiology of escalated aggression and violence. J Neurosci. 27, 11803-11806. Millan, M.J., Rivet, J.M., Canton, H., Le Marouille-Girardon, S., and Gobert, A., 1993a. Induction of hypothermia as a model of 5-hydroxytryptamine1A receptor-mediated activity in the rat: a pharmacological characterization of the actions of novel agonists and antagonists. J Pharmacol Exp Ther. 264, 1364-1376. Millan, M.J., Rivet, J.M., Canton, H., Lejeune, F., Gobert, A., Widdowson, P., Bervoets, K., Brocco, M., and Peglion, J.L., 1993b. S 15535: a highly selective benzodioxopiperazine 5HT1A receptor ligand which acts as an agonist and an antagonist at presynaptic and postsynaptic sites respectively. Eur J Pharmacol. 230, 99-102. Millan, M.J., Hjorth, S., Samanin, R., Schreiber, R., Jaffard, R., De Ladonchamps, B., Veiga, S., Goument, B., Peglion, J.L., Spedding, M., and Brocco, M., 1997. S 15535, a novel benzodioxopiperazine ligand of serotonin (5-HT)1A receptors: II. Modulation of hippocampal serotonin release in relation to potential anxiolytic properties. J Pharmacol Exp Ther. 282, 148-161. Millan, M.J., 2004. The role of monoamines in the actions of established and "novel" antidepressant agents: a critical review. Eur J Pharmacol. 500, 371-384. Minzenberg, M.J., Grossman, R., New, A.S., Mitropoulou, V., Yehuda, R., Goodman, M., Reynolds, D.A., Silverman, J.M., Coccaro, E.F., Marcus, S., and Siever, L.J., 2006. Blunted hormone responses to Ipsapirone are associated with trait impulsivity in personality disorder patients. Neuropsychopharmacology. 31, 197-203. Mitchell, E.A., Herd, M.B., Gunn, B.G., Lambert, J.J., and Belelli, D., 2008. Neurosteroid modulation of GABAA receptors: molecular determinants and significance in health and disease. Neurochem Int. 52, 588-595. Modigh, K., 1973. Effects of isolation and fighting in mice on the rate of synthesis of noradrenaline, dopamine and 5-hydroxytryptamine in the brain. Psychopharmacologia. 33, 1-17. Moeller, F.G., Dougherty, D.M., Swann, A.C., Collins, D., Davis, C.M., and Cherek, D.R., 1996. Tryptophan depletion and aggressive responding in healthy males. Psychopharmacology (Berl). 126, 97-103. Moeller, F.G., Allen, T., Cherek, D.R., Dougherty, D.M., Lane, S., and Swann, A.C., 1998. Ipsapirone neuroendocrine challenge: relationship to aggression as measured in the human laboratory. Psychiatry Res. 81, 31-38. Moffitt, T.E., Arseneault, L., Jaffee, S.R., Kim-Cohen, J., Koenen, K.C., Odgers, C.L., Slutske, W.S., and Viding, E., 2008. Research Review: DSM-V conduct disorder: research needs for an evidence base. J Child Psychol Psychiatry. 49, 3-33. Molla-Hosseini, C., 1985. Modulation of action of kainic acid on the behaviour of rats by pchlorophenylalanine and by gaba-mimetic drugs. Biochemical correlation between behaviour and treatments. Journal de pharmacologie. 16, 299-312. Morara, S., van der Want, J.J.L., de Weerd, H., Provini, L., and Rosina, A., 2001. Ultrastructural analysis of climbing fiber-Purkinje cell synaptogenesis in the rat cerebellum. Neuroscience. 108, 655-671. Morrison, S.F., 2004. Central pathways controlling brown adipose tissue thermogenesis. News Physiol Sci. 19, 67-74. Mos, J., Olivier, B., Poth, M., van Oorschot, R., and van Aken, H., 1993. The effects of dorsal raphe administration of eltoprazine, TFMPP and 8-OH-DPAT on resident intruder aggression in the rat. European Journal of Pharmacology. 238, 411-415. Moyer, K.E., 1968. Kinds of aggression and their physiological basis. Communications in Behavioural Biology. 2, 65-87. N Nagin, D. and Tremblay, R.E., 1999. Trajectories of boys' physical aggression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Dev. 70, 1181-1196.
189
REFERENCES
Nalivaiko, E. and Sgoifo, A., 2008. Central 5-HT receptors in cardiovascular control during stress. Neurosci Biobehav Rev. Natarajan, D., De Vries, H., Saaltink, D.J., de Boer, S.F., and Koolhaas, J.M., 2009. Delineation of violence from functional aggression in mice: an ethological approach. Behav Genet. 39, 73-90. Nelson, R.J. and Trainor, B.C., 2007. Neural mechanisms of aggression. Nature reviews Neuroscience. 8, 536-546. Netter, P., Hennig, J., and Rohrmann, S., 1999. Psychobiological differences between the aggression and psychoticism dimension. Pharmacopsychiatry. 32, 5-12. New, A.S., Hazlett, E.A., Buchsbaum, M.S., Goodman, M., Reynolds, D., Mitropoulou, V., Sprung, L., Shaw, R.B., Jr., Koenigsberg, H., Platholi, J., Silverman, J., and Siever, L.J., 2002. Blunted prefrontal cortical 18fluorodeoxyglucose positron emission tomography response to meta-chlorophenylpiperazine in impulsive aggression. Arch Gen Psychiatry. 59, 621-629. Nordlind, K., Azmitia, E.C., and Slominski, A., 2008. The skin as a mirror of the soul: exploring the possible roles of serotonin. Exp Dermatol. 17, 301-311. Nyberg, J., Sandnabba, K., Schalkwyk, L., and Sluyter, F., 2004. Genetic and environmental (inter)actions in male mouse lines selected for aggressive and nonaggressive behaviour. Genes Brain Behav. 3, 101-109. Nyberg, J.M., Vekovischeva, O., and Sandnabba, N.K., 2003. Anxiety profiles of mice selectively bred for intermale aggression. Behav Genet. 33, 503-511. O O'Connell, M.T., Sarna, G.S., and Curzon, G., 1992. Evidence for postsynaptic mediation of the hypothermic effect of 5-HT1A receptor activation. Br J Pharmacol. 106, 603-609. O'Keane, V., Moloney, E., O'Neill, H., O'Connor, A., Smith, C., and Dinan, T.G., 1992. Blunted prolactin responses to d-fenfluramine in sociopathy. Evidence for subsensitivity of central serotonergic function. Br J Psychiatry. 160, 643-646. Olivier, B., Mos, J., Tulp, M., Schipper, J., Den Daas, S.J., and van Oortmerssen, G.A. Serotonergic involvment in aggressive behaviour in animals. In: H.M.van Praag, R.Plutchik, A.Apter (Eds.), Violence and suicidality: Perspectives in clinical and psychological research, Brunner/Mazel, New York, 1990, pp. 79-84. Olivier, B., Mos, J., van der Heyden, J., and Hartog, J., 1989. Serotonergic modulation of social interactions in isolated male mice. Psychopharmacology (Berl). 97, 154-156. Olivier, B., Mos, J., van, O.R., and Hen, R., 1995. Serotonin receptors and animal models of aggressive behaviour. Pharmacopsychiatry. 28 Suppl 2, 80-90. Olivier, B., Zethof, T., Pattij, T., van, B.M., van, O.R., Leahy, C., Oosting, R., Bouwknecht, A., Veening, J., van der, G.J., and Groenink, L., 2003. Stress-induced hyperthermia and anxiety: pharmacological validation. Eur J Pharmacol. 463, 117-132. Oster, H., Damerow, S., Hut, R.A., and Eichele, G., 2006. Transcriptional profiling in the adrenal gland reveals circadian regulation of hormone biosynthesis genes and nucleosome assembly genes. J Biol Rhythms. 21, 350-361. Oyegbile, T.O. and Marler, C.A., 2005. Winning fights elevates testosterone levels in California mice and enhances future ability to win fights. Hormones and Behaviour. 48, 259-267. P Palacios, J.M., Waeber, C., Hoyer, D., and Mengod, G., 1990. Distribution of serotonin receptors. Ann N Y Acad Sci. 600, 36-52. Parsey, R.V., Oquendo, M.A., Simpson, N.R., Ogden, R.T., Van Heertum, R., Arango, V., and Mann, J.J., 2002. Effects of sex, age, and aggressive traits in man on brain serotonin 5HT1A receptor binding potential measured by PET using [C-11]WAY-100635. Brain Research. 954, 173-182. Patrick, C.J. and Zempolich, K.A., 1998. Emotion and aggression in the psychopathic personality. Aggression and Violent Behaviour. 3, 303-338.
190
REFERENCES
Paxinos, G., Burt, J., Atrens, D.M., and Jackson, D.M., 1977. 5-Hydroxytryptamine depletion with para-chlorophenylalanine: Effects on eating, drinking, irritability, muricide, and copulation. Pharmacology Biochemistry and Behaviour. 6, 439-447. Petitto, J.M., Gariepy, J.L., Gendreau, P.L., Rodriguiz, R., Lewis, M.H., and Lysle, D.T., 1999. Differences in NK cell function in mice bred for high and low aggression: Genetic linkage between complex behavioural and immunological traits? Brain Behaviour and Immunity. 13, 175-186. Petitto, J.M., Lysle, D.T., Gariepy, J.L., Clubb, P.H., Cairns, R.B., and Lewis, M.H., 1993. GeneticDifferences in Social-Behaviour - Relation to Natural-Killer-Cell Function and Susceptibility to Tumor-Development. Neuropsychopharmacology. 8, 35-43. Petitto, J.M., Lysle, D.T., Gariepy, J.L., and Lewis, M.H., 1994. Association of GeneticDifferences in Social-Behaviour and Cellular Immune Responsiveness - Effects of Social Experience. Brain Behaviour and Immunity. 8, 111-122. Pico-Alfonso, M.A., 2005. Psychological intimate partner violence: the major predictor of posttraumatic stress disorder in abused women. Neurosci Biobehav Rev. 29, 181-193. Pihl, R.O., Young S.N., Harden P., Plotnick S., Chamberlain B., and Ervin F.R., 1995. Acute effect of altered tryptophan levels and alcohol on aggression in normal human males. Psychopharmacology. 119, 353-360. Pineyro, G. and Blier, P., 1999. Autoregulation of serotonin neurons: role in antidepressant drug action. Pharmacol Rev. 51, 533-591. Popma, A., Jansen, L.M., Vermeiren, R., Steiner, H., Raine, A., Van Goozen, S.H., van Engeland, H., and Doreleijers, T.A., 2006. Hypothalamus pituitary adrenal axis and autonomic activity during stress in delinquent male adolescents and controls. Psychoneuroendocrinology. 31, 948-957. Popova, N.K., Naumenko, V.S., Plyusnina, I.Z., and Kulikov, A.V., 2005. Reduction in 5-HT1A receptor density, 5-HT1A mRNA expression, and functional correlates for 5-HT1A receptors in genetically defined aggressive rats. J Neurosci Res. 80, 286-292. Popova, N.K., Naumenko, V.S., and Plyusnina, I.Z., 2007. Involvement of brain serotonin 5-HT1A receptors in genetic predisposition to aggressive behaviour. Neurosci Behav Physiol. 37, 631-635. Pruus, K., Skrebuhhova-Malmros, T., Rudissaar, R., Matto, V., and Allikmets, L., 2000. 5-HT1A receptor agonists buspirone and gepirone attenuate apomorphine-induced aggressive behaviour in adult male Wistar rats. J Physiol Pharmacol. 51, 833-846. R Raine, A., Venables, P.H., and Williams, M., 1990. Autonomic orienting responses in 15-year-old male subjects and criminal behaviour at age 24. The American journal of psychiatry. 147, 933-937. Raine, A., Lencz, T., Bihrle, S., LaCasse, L., and Colletti, P., 2000. Reduced prefrontal gray matter volume and reduced autonomic activity in antisocial personality disorder. Arch Gen Psychiatry. 57, 119-127. Raine, A., 2002a. Annotation: the role of prefrontal deficits, low autonomic arousal, and early health factors in the development of antisocial and aggressive behaviour in children. J Child Psychol Psychiatry. 43, 417-434. Raine, A., 2002b. Biosocial studies of antisocial and violent behaviour in children and adults: a review. J Abnorm Child Psychol. 30, 311-326. Raine, A., 2003. Corpus callosum abnormalities in psychopathic antisocial individuals. Archives of general psychiatry. 60, 1134-1142. Ramirez, J.M. and Andreu, J.M., 2006. Aggression, and some related psychological constructs (anger, hostility, and impulsivity); some comments from a research project. Neurosci Biobehav Rev. 30, 276-291. Raymond, J.R., Mukhin, Y.V., Gettys, T.W., and Garnovskaya, M.N., 1999. The recombinant 5HT1A receptor: G protein coupling and signalling pathways. Br J Pharmacol. 127, 1751-1764.
191
REFERENCES
Reale, D., Reader, S.M., Sol, D., McDougall, P.T., and Dingemanse, N.J., 2007. Integrating animal temperament within ecology and evolution. Biological Reviews. 82, 291-318. Refinetti, R., 1994. Contribution of locomotor activity to the generation of the daily rhythm of body temperature in golden hamsters. Physiol Behav. 56, 829-831. Retz, W., Rosler, M., Supprian, T., Retz-Junginger, P., and Thome, J., 2003. Dopamine D3 receptor gene polymorphism and violent behaviour: relation to impulsiveness and ADHDrelated psychopathology. J Neural Transm. 110, 561-572. Reyes, B.A., Fox, K., Valentino, R.J., and Van Bockstaele, E.J., 2006. Agonist-induced internalization of corticotropin-releasing factor receptors in noradrenergic neurons of the rat locus coeruleus. Eur J Neurosci. 23, 2991-2998. Riad, M., Watkins, K.C., Doucet, E., Hamon, M., and Descarries, L., 2001. Agonist-induced internalization of serotonin-1a receptors in the dorsal raphe nucleus (autoreceptors) but not hippocampus (heteroreceptors). J Neurosci. 21, 8378-8386. Riad, M., Zimmer, L., Rbah, L., Watkins, K.C., Hamon, M., and Descarries, L., 2004. Acute treatment with the antidepressant fluoxetine internalizes 5-HT1A autoreceptors and reduces the in vivo binding of the PET radioligand [18F]MPPF in the nucleus raphe dorsalis of rat. J Neurosci. 24, 5420-5426. Riad, M., Rbah, L., Verdurand, M., Aznavour, N., Zimmer, L., and Descarries, L., 2008. Unchanged density of 5-HT(1A) autoreceptors on the plasma membrane of nucleus raphe dorsalis neurons in rats chronically treated with fluoxetine. Neuroscience. 151, 692-700. Roubertoux, P.L., Guillot, P.V., Mortaud, S., Pratte, M., Jamon, M., Cohen-Salmon, C., and Tordjman, S., 2005. Attack behaviours in mice: from factorial structure to quantitative trait loci mapping. Eur J Pharmacol. 526, 172-185. Rujescu, D., Giegling, I., Bondy, B., Gietl, A., Zill, P., and Moller, H.J., 2002. Association of anger-related traits with SNPs in the TPH gene. Mol Psychiatry. 7, 1023-1029. S Salomon, R.M., Mazure, C.M., Delgado, P.L., Mendia, P., and Charney, D.S., 1994. Serotonin function in aggression: The effect of acute plasma tryptophan depletion in aggressive patients. Biological Psychiatry. 35, 570-572. Sandnabba, N.K., 1985. Differences in the capacity of male odours to affect investigatory behaviour and different urinary marking patterns in two strains of mice, selectively bred for high and low aggressiveness. Behavioural Processes. 11, 257-267. Sandnabba, N.K., 1993a. Effects of early exposure to intermale aggression on the aggressiveness of adult male mice varying in their genetic disposition for aggressive behaviour. Aggressive Behaviour. 19, 435-445. Sandnabba, N.K., 1995. Predatory Aggression in Male-Mice Selectively Bred for IsolationInduced Intermale Aggression. Behaviour Genetics. 25, 361-366. Sandnabba, N.K. and Korpela, S., 1994. Effects of early exposure to mating on adult sexual behaviour in male mice varying in their genetic disposition for aggressive behaviour. Aggressive Behaviour. 20, 429-439. Sandnabba, N.K., 1996. Selective breeding for isolation-induced intermale aggression in mice: associated responses and environmental influences. Behav Genet. 26, 477-488. Sandnabba, N.K., 1997. Territorial behaviour and social organisation as a function of the level of aggressiveness in male mice. Ethology. 103, 566-577. Sandnabba, N.K., Lagerspetz, K.M., and Jensen, E., 1994. Effects of testosterone exposure and fighting experience on the aggressive behaviour of female and male mice selectively bred for intermale aggression. Horm Behav. 28, 219-231. Sandnabba, N.K., 1986a. Changes in male odours and urinary marking patterns due to inhibition of aggression in male mice. Behavioural Processes. 12, 349-361. Sandnabba, N.K., 1986b. Differences Between Two Strains of Mice, Selectively Bred for High and Low Aggressiveness, in the Capacity of Male Odors to Affect Aggressive Behaviour.
192
REFERENCES
Aggressive Behaviour. 12, 103-110. Sandnabba, N.K., 1986c. Effects of Selective Breeding for High and Low Aggressiveness and of Fighting Experience on Odor Discrimination in Mice. Aggressive Behaviour. 12, 359-366. Sandnabba, N.K., 1993b. Female aggression during gestation and lactation in two strains of mice selected for isolation-induced intermale aggression. Behavioural Processes. 30, 157-164. Scarpa, A. and Raine, A., 1997. Psychophysiology of anger and violent behaviour. Psychiatr Clin North Am. 20, 375-394. Schiller, L., Donix, M., Jahkel, M., and Oehler, J., 2006. Serotonin 1A and 2A receptor densities, neurochemical and behavioural characteristics in two closely related mice strains after long-term isolation. Prog Neuropsychopharmacol Biol Psychiatry. 30, 492-503. Selander, R.K. and Kvist, S.B.M., 1991. Open-field parameters and maze-learning in aggressive and nonaggressive male-mice. Perceptual and Motor Skills. 73, 811-824. Sewell, R.G., Gallus, J.A., Gault, F.P., and Cleary, J.P., 1982. p-Chlorophenylalanine effects on shock-induced attack and pressing responses in rats. Pharmacol Biochem Behav. 17, 945-950. Sgoifo, A., de Boer, S.F., Haller, J., and Koolhaas, J.M., 1996. Individual differences in plasma catecholamine and corticosterone stress responses of wild-type rats: relationship with aggression. Physiol Behav. 60, 1403-1407. Sgoifo, A., Costoli, T., Meerlo, P., Buwalda, B., Pico'-Alfonso, M.A., De, B.S., Musso, E., and Koolhaas, J., 2005. Individual differences in cardiovascular response to social challenge. Neurosci Biobehav Rev. 29, 59-66. Shields, J. and King, J.A., 2008. The role of 5-HT1A receptors in the behavioural responses associated with innate fear. Behav Neurosci. 122, 611-617. Shin, K., Minamitani, H., Onishi, S., Yamazaki, H., and Lee, M., 1997. Autonomic differences between athletes and nonathletes: spectral analysis approach. Med Sci Sports Exerc. 29, 1482-1490. Siever, L.J., 2008. Neurobiology of aggression and violence. Am J Psychiatry. 165, 429-442. Sih, A., Bell, A., and Johnson, J.C., 2004a. Behavioural syndromes: an ecological and evolutionary overview. Trends in ecology & evolution. 19, 372-378. Sih, A., Bell, A.M., Johnson, J.C., and Ziemba, R.E., 2004b. Behavioural syndromes: an integrative overview. Quarterly Review of Biology. 79, 241-277. Simon, N.G., McKenna, S.E., Lu, S.F., and Cologer-Clifford, A., 1996. Development and expression of hormonal systems regulating aggression. Ann N Y Acad Sci. 794, 8-17. Simpson, J. and Weiner, E. The Oxford English Dictionary. 2nd ed. 1989. OED Online. http://dictionary.oed.com (2000) Sluyter, F., Bohus, B., Beldhuis, H.J., and van Oortmerssen, G.A., 1995a. Autosomal and Y chromosomal effects on the stereotyped response to apomorphine in wild house mice. Pharmacol Biochem Behav. 52, 17-22. Sluyter, F., Bult, A., Lynch, C.B., Meeter, F., and van Oortmerssen, G.A., 1997. No evidence for a Y chromosomal effect on alternative behavioural strategies in mice. Behav Genet. 27, 477482. Sluyter, F., Bult, A., Lynch, C.B., van Oortmerssen, G.A., and Koolhaas, J.M., 1995b. A comparison between house mouse lines selected for attack latency or nest-building: evidence for a genetic basis of alternative behavioural strategies. Behav Genet. 25, 247-252. Sluyter, F., Jamot, L., van Oortmerssen, G.A., and Crusio, W.E., 1994a. Hippocampal mossy fiber distributions in mice selected for aggression. Brain Res. 646, 145-148. Sluyter, F., Korte, S.M., Bohus, B., and van Oortmerssen, G.A., 1996a. Behavioural stress response of genetically selected aggressive and nonaggressive wild house mice in the shockprobe/defensive burying test. Pharmacol Biochem Behav. 54, 113-116. Sluyter, F., Korte, S.M., Van Baal, G.C., de Ruiter, A.J., and van Oortmerssen, G.A., 1999. Y chromosomal and sex effects on the behavioural stress response in the defensive burying test in wild house mice. Physiol Behav. 67, 579-585.
193
REFERENCES
Sluyter, F., Meijeringh, B.J., van Oortmerssen, G.A., and Koolhaas, J.M., 1995c. Studies on wild house mice (VIII): Postnatal maternal influences on intermale aggression in reciprocal F1's. Behav Genet. 25, 367-370. Sluyter, F., van Oortmerssen, G.A., and Koolhaas, J.M., 1994b. Studies on Wild House Mice VI: Differential effects of the Y chromosome on intermale aggression. Aggressive Behaviour. 20, 379-386. Sluyter, F., van Oortmerssen, G.A., and Koolhaas, J.M., 1996b. Genetic influences on coping behaviour in house mouse lines selected for aggression: Effects of the Y chromosome. Behaviour. 133, 117-128. Sluyter, F., van, d., V, van Oortmerssen, G.A., Koolhaas, J.M., van der, H.F., and de, B.P., 1996c. Studies on wild house mice. VII. Prenatal maternal environment and aggression. Behav Genet. 26, 513-518. Sluyter, F., Arseneault, L., Moffitt, T.E., Veenema, A.H., de Boer S.F., and Koolhaas, J.M., 2003. Toward an animal model for antisocial behaviour: parallels between mice and humans. Behav Genet. 33, 563-574. Sperry, T.S., Thompson, C.K., and Wingfield, J.C., 2003. Effects of acute treatment with 8-OHDPAT and fluoxetine on aggressive behaviour in male song sparrows (Melospiza melodia morphna). J Neuroendocrinol. 15, 150-160. Stamps, J.A., 2007. Growth-mortality tradeoffs and 'personality traits' in animals. Ecology Letters. 10, 355-363. Steimer, T. and Driscoll, P., 2003. Divergent stress responses and coping styles in psychogenetically selected Roman high-(RHA) and low-(RLA) avoidance rats: behavioural, neuroendocrine and developmental aspects. Stress. 6, 87-100. Steimer, T., la, F.S., and Schulz, P.E., 1997. Neuroendocrine correlates of emotional reactivity and coping in male rats from the Roman high (RHA/Verh)- and low (RLA/Verh)-avoidance lines. Behav Genet. 27, 503-512. Stork, O., Welzl, H., Wotjak, C.T., Hoyer, D., Delling, M., Cremer, H., and Schachner, M., 1999. Anxiety and increased 5-HT1A receptor response in NCAM null mutant mice. J Neurobiol. 40, 343-355. Strijkstra, A.M., Meerlo, P., and Beersma, D.G., 1999. Forced desynchrony of circadian rhythms of body temperature and activity in rats. Chronobiol Int. 16, 431-440. Summers, C.H. and Greenberg, N., 1994. Somatic correlates of adrenergic activity during aggression in the lizard, Anolis carolinensis. Horm Behav. 28, 29-40. Summers, C.H., Summers, T.R., Moore, M.C., Korzan, W.J., Woodley, S.K., Ronan, P.J., Hoglund, E., Watt, M.J., and Greenberg, N., 2003a. Temporal patterns of limbic monoamine and plasma corticosterone response during social stress. Neuroscience. 116, 553-563. Summers, T.R., Matter, J.M., McKay, J.M., Ronan, P.J., Larson, E.T., Renner, K.J., and Summers, C.H., 2003b. Rapid glucocorticoid stimulation and GABAergic inhibition of hippocampal serotonergic response: in vivo dialysis in the lizard anolis carolinensis. Hormones and Behaviour. 43, 245-253. Summers, C.H., Korzan, W.J., Lukkes, J.L., Watt, M.J., Forster, G.L., Overli, O., Hoglund, E., Larson, E.T., Ronan, P.J., Matter, J.M., Summers, T.R., Renner, K.J., and Greenberg, N., 2005. Does serotonin influence aggression? comparing regional activity before and during social interaction. Physiol Biochem Zool. 78, 679-694. Sundaresan, S., Yang-Feng, T.L., and Francke, U., 1989. Genes for HMG-CoA reductase and serotonin 1a receptor are on mouse chromosome 13. Somat Cell Mol Genet. 15, 465-469. Suomi, S.J., 2003. Gene-environment interactions and the neurobiology of social conflict. Ann N Y Acad Sci. 1008, 132-139.
194
REFERENCES
T Tanke, M.A., Alserda, E., Doornbos, B., van der Most, P.J., Goeman, K., Postema, F., and Korf, J., 2008. Low tryptophan diet increases stress-sensitivity, but does not affect habituation in rats. Neurochem Int. 52, 272-281. Tóth, M., Halász, J., Mikics, E., Barsy, B., and Haller, J., 2008. Early social deprivation induces disturbed social communication and violent aggression in adulthood. Behav Neurosci. 122: 849-54 V Valzelli, L., Bernasconi, S., and Garattini, S., 1981. p-Chlorophenylalanine-induced muricidal aggression in male and female laboratory rats. Neuropsychobiology. 7, 315-320. Valzelli, L., 1982. Serotonergic inhibitory control of experimental aggression. Pharmacological research communications. 14, 1-13. Valzelli, L., 1984. Reflections on experimental and human pathology of aggression. Progress in neuro-psychopharmacology & biological psychiatry. 8, 311-325. Van Bogaert, M.J., Groenink, L., Oosting, R.S., Westphal, K.G., van der Gugten J., and Olivier, B., 2006. Mouse strain differences in autonomic responses to stress. Genes Brain Behav. 5, 139-149. Van der Plasse, G., Meerkerk, D.T., Lieben, C.K., Blokland, A., and Feenstra, M.G., 2007. Lack of evidence for reduced prefrontal cortical serotonin and dopamine efflux after acute tryptophan depletion. Psychopharmacology (Berl). 195, 377-385. Van der Vegt, B., de Boer, S.F., Buwalda, B., de Ruiter, A.J., de Jong, J.G., and Koolhaas, J.M., 2001. Enhanced sensitivity of postsynaptic serotonin-1A receptors in rats and mice with high trait aggression. Physiol Behav. 74, 205-211. Van der Vegt, B., Lieuwes, N., Cremers, T.I., de Boer, S.F., and Koolhaas, J.M., 2003a. Cerebrospinal fluid monoamine and metabolite concentrations and aggression in rats. Horm Behav. 44, 199-208. Van der Vegt, B., Lieuwes, N., van de Wall, E.H., Kato, K., Moya-Albiol, L., Martinez-Sanchis, S., de Boer, S.F., and Koolhaas, J.M., 2003b. Activation of serotonergic neurotransmission during the performance of aggressive behaviour in rats. Behav Neurosci. 117, 667-674. Van Erp, A.M. and Miczek, K.A., 2000. Aggressive behaviour, increased accumbal dopamine, and decreased cortical serotonin in rats. J Neurosci. 20, 9320-9325. Van Goozen, S.H., Fairchild, G., Snoek, H., and Harold, G.T., 2007. The evidence for a neurobiological model of childhood antisocial behaviour. Psychol Bull. 133, 149-182. Van Goozen, S.H. and Fairchild, G., 2008. How can the study of biological processes help design new interventions for children with severe antisocial behaviour? Dev Psychopathol. 20, 941-973. Van Hooff, J.A.R.A.M. Categories and sequences of behaviour: methods of description and analysis. In: K.Scherer, P.Ekman (Eds.), Handbook of mehtods in non verbal behaviour research., Cambridge Press, Cambridge, 1982, pp. 362-439. Van Oortmerssen, G.A. and Bakker, T.C., 1981. Artificial selection for short and long attack latencies in wild Mus musculus domesticus. Behav Genet. 11, 115-126. Van Oortmerssen, G.A., Benus, I., and Dijk, D.J., 1985. Studies in wild house mice: genotypeenvironment interactions for attack latency. Netherlands Journal of Zoology. 35, 155-169. Van Oortmerssen, G.A. and Busser, J. Studies in wild house mice 3: Disruptive selection on aggression as a possible force in evolution. In: P.F.Brain, D.Mainardi, S.Parmigiani (Eds.), House mouse aggression. A model for understanding the evolution of social behaviour, Harwood Aacademic Publishers, Chur, 1989, pp. 87-118. Van Oortmerssen, G.A., Dijk, D.J., and Schuurman, T., 1987. Studies in wild house mice. II. Testosterone and aggression. Horm Behav. 21, 139-152. Van Oortmerssen, G., Benus, R., and Sluyter, F., 1992. Studies on Wild House Mice IV: On the Heredity of Testosterone and Readiness to Attack. Aggressive Behaviour. 18, 143-148.
195
REFERENCES
Van Oortmerssen, G.A. and Sluyter, F., 1994. Studies on wild house mice. V. Aggression in lines selected for attack latency and their Y-chromosomal congenics. Behav Genet. 24, 73-78. Van Praag, H.M., 2004. Can stress cause depression? Prog Neuropsychopharmacol Biol Psychiatry. 28, 891-907. Van Riel, E., Meijer, O.C., Veenema, A.H., and Joels, M., 2002. Hippocampal serotonin responses in short and long attack latency mice. J Neuroendocrinol. 14, 234-239. Van Zegeren, K., 1980. Variation in Aggressiveness and the Regulation of Numbers in House Mouse-Populations. Netherlands Journal of Zoology. 30, 635-&. Veenema, A.H., Koolhaas, J.M., and de Kloet, E.R., 2004. Basal and stress-induced differences in HPA axis, 5-HT responsiveness, and hippocampal cell proliferation in two mouse lines. Ann N Y Acad Sci. 1018, 255-265. Veenema, A.H., Meijer, O.C., de Kloet, E.R., and Koolhaas, J.M., 2003a. Genetic selection for coping style predicts stressor susceptibility. J Neuroendocrinol. 15, 256-267. Veenema, A.H., Meijer, O.C., de Kloet, E.R., Koolhaas, J.M., and Bohus, B.G., 2003b. Differences in basal and stress-induced HPA regulation of wild house mice selected for high and low aggression. Horm Behav. 43, 197-204. Veenema, A.H., Cremers, T.I., Jongsma, M.E., Steenbergen, P.J., de Boer, S.F., and Koolhaas, J.M., 2005. Differences in the effects of 5-HT(1A) receptor agonists on forced swimming behaviour and brain 5-HT metabolism between low and high aggressive mice. Psychopharmacology (Berl). 178, 151-160. Veenema, A.H., Blume, A., Niederle, D., Buwalda, B., and Neumann, I.D., 2006. Effects of early life stress on adult male aggression and hypothalamic vasopressin and serotonin. Eur J Neurosci. 24:1711-20. Veenema, A.H. and Neumann, I.D., 2007. Neurobiological mechanisms of aggression and stress coping: a comparative study in mouse and rat selection lines. Brain Behav Evol. 70, 274285. Vekovischeva, O.Y., Aitta-aho, T., Verbitskaya, E., Sandnabba, K., and Korpi, E.R., 2007a. Acute effects of AMPA-type glutamate receptor antagonists on intermale social behaviour in two mouse lines bidirectionally selected for offensive aggression. Pharmacology, biochemistry and behaviour. 87, 241-249. Vekovischeva, O.Y., Verbitskaya, E.V., Aitta-aho, T., Sandnabba, K., and Korpi, E.R., 2007b. Multimetric statistical analysis of behaviour in mice selected for high and low levels of isolation-induced male aggression. Behavioural Processes. 75, 23-32. Vergnes, M. and Kempf, E., 1982. Effect of hypothalamic injections of 5,7-dihydroxytryptamine on elicitation of mouse-killing in rats. Behavioural Brain Research. 5, 387-397. Vergnes, M., Depaulis, A., and Boehrer, A., 1986. Parachlorophenylalanine-induced serotonin depletion increases offensive but not defensive aggression in male rats. Physiol Behav. 36, 653-658. Vergnes, M., Depaulis, A., Boehrer, A., and Kempf, E., 1988. Selective increase of offensive behaviour in the rat following intrahypothalamic 5,7-DHT-induced serotonin depletion. Behav Brain Res. 29, 85-91. Virkkunen, M., 1985. Urinary free cortisol secretion in habitually violent offenders. Acta Psychiatr Scand. 72, 40-44. Virkkunen, M., De, J.J., Bartko, J., and Linnoila, M., 1989. Psychobiological concomitants of history of suicide attempts among violent offenders and impulsive fire setters. Arch Gen Psychiatry. 46, 604-606. Virkkunen, M. and Linnoila, M., 1990. Serotonin in early onset, male alcoholics with violent behaviour. Ann Med. 22, 327-331. Vis, P., Della, P.O., Kruk, M., Martin, D., Mocaer, E., Danhof, M., and Jochemsen, R., 2001. Population pharmacokinetic-pharmacodynamic modelling of S 15535, a 5-HT(1A) receptor agonist, using a behavioural model in rats. Eur J Pharmacol. 414, 233-243.
196
REFERENCES
Vitiello, B. and Stoff, D.M., 1997. Subtypes of aggression and their relevance to child psychiatry. J Am Acad Child Adolesc Psychiatry. 36, 307-315. W Ward, M.M., Chesney, M.A., Swan, G.E., Black, G.W., Parker, S.D., and Rosenman, R.H., 1986. Cardiovascular responses in type A and type B men to a series of stressors. J Behav Med. 9, 43-49. Weerts, E.M., Miller, L.G., Hood, K.E., and Miczek, K.A., 1992. Increased Gaba(A)-Dependent Chloride Uptake in Mice Selectively Bred for Low Aggressive-Behaviour. Psychopharmacology. 108, 196-204. White, S.M., Kucharik, R.F., and Moyer, J.A., 1991. Effects of serotonergic agents on isolationinduced aggression. Pharmacol Biochem Behav. 39, 729-736. Wingrove, J., Bond, A.J., and Cleare, A.J., 1999. Tryptophan enhancement/depletion and reactions to failure on a cooperative computer game. Neuropsychopharmacology. 21, 755-764. Wolf, M., van Doorn, G.S., Leimar, O., and Weissing, F.J., 2007. Life-history trade-offs favour the evolution of animal personalities. Nature. 447, 581-584. Wolfe, B.L. and Trejo, J., 2007. Clathrin-dependent mechanisms of G protein-coupled receptor endocytosis. Traffic. 8, 462-470. Wommack, J.C. and Delville, Y., 2007a. Cortisol controls the pubertal development of agonistic behaviour in male golden hamsters via type II corticosteroid receptors. Horm Behav. 51, 306-312. Wommack, J.C. and Delville, Y., 2007b. Stress, aggression, and puberty: neuroendocrine correlates of the development of agonistic behaviour in golden hamsters. Brain Behav Evol. 70, 267-273. Y Yang, G.B., Qiu, C.L., Zhao, H., Liu, Q., and Shao, Y., 2006. Expression of mRNA for multiple serotonin (5-HT) receptor types/subtypes by the peripheral blood mononuclear cells of rhesus macaques. J Neuroimmunol. 178, 24-29. Z Zhang, Y., Gray, T.S., D'Souza, D.N., Carrasco, G.A., Damjanoska, K.J., Dudas, B., Garcia, F., Zainelli, G.M., Sullivan Hanley, N.R., Battaglia, G., Muma, N.A., and Van de Kar, L.D., 2004. Desensitization of 5-HT1A receptors by 5-HT2A receptors in neuroendocrine neurons in vivo. J Pharmacol Exp Ther. 310, 59-66. Zimmer, L., Riad, M., Rbah, L., Belkacem-Kahlouli, A., Le Bars D., Renaud, B., and Descarries, L., 2004. Toward brain imaging of serotonin 5-HT1A autoreceptor internalization. Neuroimage. 22, 1421-1426. Zuideveld, K.P., Maas, H.J., Treijtel, N., Hulshof, J., van Der Graaf, P.H., Peletier, L.A., and Danhof, M., 2001. A set-point model with oscillatory behaviour predicts the time course of 8-OH-DPAT-induced hypothermia. Am J Physiol Regul Integr Comp Physiol. 281, R2059R2071. Zuideveld, K.P., Treijtel, N., Maas, H.J., Gubbens-Stibbe, J.M., Peletier, L.A., van Der Graaf, P.H., and Danhof, M., 2002. A Competitive Interaction Model Predicts the Effect of WAY-100,635 on the Time Course of R-(+)-8-Hydroxy-2-(di-n-propylamino)tetralin-Induced Hypothermia. J Pharmacol Exp Ther. 300, 330-338.
197
Nederlandse samenvatting
NEDERLANDSE
SAMENVATTING
DE AANLEIDING VOOR DIT ONDERZOEK Ondanks decennialang intensief sociaal en biologisch onderzoek naar de oorzaken van agressief gedrag, wordt onze huidige Westerse samenleving in toenemende mate geconfronteerd met de zeer ernstige en nadelige gevolgen ervan. Hoewel slechts een gering percentage van onze populatie buitensporigeen gewelddadige vormen van agressie vertoont, wordt de impact van dit abnormale gedrag op de maatschappij als zeer ernstig en ontwrichtend ervaren. Naast de mogelijke directe fysieke (verwonding en doodslag) consecuenties, leiden gewelddadige conflicten bij veel slachtoffers (waaronder ook getuigen van geweld) op termijn tot invaliderende psychische stoornissen en gezondheidsklachten. Het overgrote deel van het psychologisch- en biomedisch onderzoek is daarom ook gericht op het in kaart brengen van de veranderingen die dergelijke stressvolle ervaringen opwekken in de hersenen, en hoe deze neurale veranderingen vervolgens beïnvloed kunnen worden. Merkwaardigerwijs is een aanzienlijk geringere onderzoeksinspanning gewijd aan de hersenmechanismen die ten grondslag liggen aan de ontregeling van agressief gedrag zelf. Neurobiologisch gezien kunnen veel geweldplegers beschouwd worden als slachtoffer van een falende hersencontrole over eigen gedrag, aangezien hun agressieve gevoelens en uitingen kennelijk niet correct onderdrukt en beteugeld kunnen worden. Omdat agressie, net als alle andere gedragingen, primair door de hersenen wordt opgewekt en bestuurd, kan de ontregeling ervan in het geval van geweld beschouwd worden als een incorrect en/of pathologisch functioneren van het neurale circuit voor de uitvoering van normaal agressief gedrag. Dit agressiecircuit heeft een grote overlap met het bekende angst- en emotiecircuit van het brein. Het is daarom ook niet verbazingwekkend dat bij veel neuropsychiatrische aandoeningen voortkomend uit een ontregeld emotiecircuit (depressie, PTSD, ADHD, etc) agressieve gedragstoornissen veelvuldig als symptoom voorkomt. Omgekeerd heeft recent onderzoek aangetoond dat tussen de 80 – 90% van de gedetineeerden in penitentiaire inrichtingen een psychiatrische stoornis heeft of heeft gehad. Voorts mag aangenomen worden dat de structurele en functionele eigenschappen van dit neurale “agressie netwerk” niet statisch en dus onomkeerbaar vastgelegd zijn maar juist plastisch en veranderbaar, waardoor er dan ook in principe mogelijkheden bestaan om het therapeutisch te kunnen manipuleren. Tot op heden bestaan er helaas nog geen selectieve en effectieve farmaco- of gedragstherapieën voor pathologische vormen van agressie en geweld. Een van de belangrijkste oorzaken hiervoor is het ontbreken van voldoende neurobiologische kennis over de sturing van agressief gedrag, en dan vooral van de fysiologische processen en mechanismen die ten grondslag liggen aan de ontregeling en escalatie ervan in geweld. Onze huidige kennis voor wat betreft de oorzaken, 200
NEDERLANDSE
SAMENVATTING
gevolgen en beïnvloeding van humaan agressief gedrag is namelijk grotendeels gebaseerd op dierexperimentele laboratorium modellen voor normale adaptieve en uiterst functionele vormen van agressie. Het overgrote deel van dit preklinische agressieonderzoek wordt uitgevoerd door sterk gedomesticeerde (en als gevolg nauwelijks agressieve) ratten of muizen te confronteren met een onbekende soortgenoot in competitie om territorium of dominantie (de zgn. residentintruder conflict test). De in deze situatie optredende normale en uiterst functionele vormen van agressie hebben slechts weinig of geen relevantie met het excessieve, pathologische en gewelddadige humaan agressief gedrag. Helaas is er binnen het dierexperimentele agressieonderzoek tot op heden nog maar nauwelijks aandacht geweest voor het ontwikkelen van pathologische agressie. Het onderzoek beschreven in dit proefschrift maakt gebruik van het natuurlijk voorkomende verschil tussen individuen in de geneigdheid om offensief agressief gedrag te vertonen. Dit verschil, dat grotendeels genetisch wordt bepaald, kan ertoe leiden dat sommige individuen onder bepaalde omstandigheden gevoelig zijn om extreme en geëscaleerde vormen van agressie te ontwikkelen. Bij de experimenten van dit onderzoek is daarom gebruik gemaakt van drie verschillende muizenstammen die, in het laboratorium, gefokt en systematisch geselecteerd zijn op hoge (SAL, TA, NC900) en lage (LAL, TNA, NC100) aanvallende aggressiviteit. Als situatie is gekozen voor herhaalde sociale interacties waarbij het experimentele aanvallende dier de optredende conflicten steeds wint. Deze herhaalde winnaarervaringen vergroten namelijk de kans dat een agressief dier bij een volgend conflict een hogere intensiteit van agressie laat zien en uiteindelijk de controle over agressie kan verliezen en gewelddadig wordt.
HET GEWELDDADIGE FENOTYPE: DE SAL MUIS Gebaseerd op de definitie dat geweld een pathologische vorm van normaal agressief gedrag is dat buitenproportioneel is qua intensiteit (out of control) en situatie (out of context), met als gevolg dat de tegenstander ernstig wordt verwond of zelfs wordt gedood, zijn in eerste instantie de objectief meetbare indicatoren en gedragsparameters opgesteld om muizen als gewelddadig te bestempelen. Deze criteria en indicatoren zijn: - Een genetische predispositie voor, en het op jonge leeftijd al aanwezig zijn van een hoge mate aan agressief gedrag - Hoge intensiteit van aanvallende agressie in verschillende test situaties - Verlies van het vermogen om te discrimineren tussen reële en neutrale sociale bedreigingen. - Verlies van inhibities over het uitvoerende gedrag en het negeren van 201
NEDERLANDSE
-
SAMENVATTING
signalen van de tegenstander. Het achterwege laten van signalen (geritualiseerde dreighoudingen) die bijtaanvallen aankondigen evenals het richten van deze aanvallen op vitale lichaamsdelen van de tegenstander, met verwonding als gevolg.
De resultaten van de diverse gedragsanalyses in verschillende conflicttest situaties (resident thuiskooi, neutraal terrein en indringers thuiskooi) laten duidelijk zien dat alleen de mannelijke SAL muizen aan alle bovenstaande criteria voor gewelddadig gedrag voldoen. Bij de twee andere hoogagressieve selectie lijnen, de TA en NC900, werden er geen duidelijke pathologische agressie kenmerken waargenomen. Kennelijk bezitten wilde huismuizen (de stam waaruit de SAL en LAL zijn geselecteerd) bepaalde (pathogene) eigenschappen die door het langdurige en vergaande domesticatieproces bij gebruikelijk laboratorium muizenstammen (waaruit de TA en NC900 uit voortgekomen zijn) verloren zijn gegaan. Deze bevinding onderstreept het belang van de keuze voor de juiste stam of het soort dier bij de ontwikkeling van een proefdiermodel voor een humane pathologie. Voorts is in dit onderzoek bevestigd dat de op hoge agressie geselecteerde muizen gekarakteriseerd worden door een globale proactieve gedragsstrategie (copingstyle of persoonlijkheid). Naast deze karakteristieke gedragskenmerken voor een gewelddadig fenotype, vertonen de SAL muizen ook specifieke neurobiologische verschijnselen die niet bij de andere hoog- en laagagressieve muizenlijnen voorkomen. Een van de meest kenmerkende neurobiologische verschijnselen die bij pathologisch agressieve mensen frequent wordt gevonden is een lage serotonine neurotransmitter activiteit in ondermeer hersengebieden die een sterke controle over (agressief) gedrag hebben. Gebleken is dat de zeer agressieve SAL muizen, in het bijzonder na herhaalde sociale winnaarervaringen, lage serotonine concentraties in de prefrontale cortex hebben. Voorts bleek dat een laag serotonine nivo niet persee verantwoordelijk of noodzakelijk hoeft te zijn voor het vertoonde pathologische gedrag maar dat vooral homeostatische regelfactoren, die de dynamiek van de serotonerge neurotransmissie activiteit nauwgezet controleren, hier een belangrijkere rol lijken te spelen. De zeer agressieve SAL muizen hebben namelijk een sterk verhoogde serotonine auto-inhibitie via de serotonine 5-HT1A autoreceptor. Deze autoreceptoren bevinden zich op het cellichaam van de serotonine neuronen en oefenen, na activatie door serotonine zelf, een sterk remmende invloed uit op de serotonine neurotransmissie aktiviteit. Een voortdurende (te) sterke remfunctie kan uiteindelijke leiden tot een sterke afname in serotonine spiegels. Sommige serotonerge neuronen in bepaalde hersengebieden (m.n. de Raphe Pallides) oefenen een sterke controle uit op de regulatie van de (re)activiteit van 202
NEDERLANDSE
SAMENVATTING
het perifere autonome zenuwstelsel. Gebleken is dat SAL muizen, mogelijk vanwege hun sterke 5-HT1A-gemedieerde remfunctie en daardoor laag functionerend serotonine system, ook een tragere hartslagfrequentie hebben evenals een hogere lichaamstemperatuur stijging tijdens de aktieve fase van hun 24 uurs ritme en na blootstelling aan stressvolle omstandigheden. Deze bevinding versterkt de al geruime tijd geleden geopperde speculatie dat deze perifere fysiologische kenmerken ook als biomarker zou kunnen dienen voor gewelddadige gedragskenmerken.
CONCLUSIE EN AANBEVELING De studies die in dit proefschrift zijn beschreven leiden tot de conclusie dat de van wilde huismuizen afkomstige en op snelle aanvallende agressie geselecteerde SAL muizenlijn een valide en bruikbaar diermodel is voor gewelddadige en antisociale gedragsstoornissen. Het gebruik van deze unieke muizenlijn in het noodzakelijke preklinische agressieonderzoek naar de ontwikkeling van nieuwe en/of betere preventie- en interventiemethoden van deze gedragsstoornissen in mensen strekt dan ook tot aanbeveling.
203
Sintesi della tesi in italiano
SINTESI
DELLA TESI IN ITALIANO
LA FISIOLOGIA DELL’AGGRESSIVITÀ: PER COMPRENDERE LA VIOLENZA Aggressività, violenza e scopo della tesi Decenni di ricerca biologica e sociologica sono stati spesi nel cercare una spiegazione all’aggressività umana, considerato il pesante e costoso fardello che essa infligge sulla nostra società. Sebbene soltanto in una piccola frazione della popolazione umana l’aggressività sia così intensa da trasformarsi in violenza, si tratta di un comportamento ad elevato impatto sociale che può portare non solo al danno fisico e alla morte diretta ma anche a profonde invalidità fisiche ed emotive nelle vittime e negli spettatori. Nella maggior parte dei casi, la ricerca si è concentrata sui destinatari dell’aggressione, dato che questi ultimi spesso mostrano durevoli alterazioni a livello cerebrale con conseguenti marcati disturbi psicologici (ansietà, depressione) e fisici (patologie cardiovascolari e immunitarie). Allo stesso tempo, gli aggressori sono vittime loro stessi, non essendo in grado di autocontrollarsi dal causare sofferenza fisica e psichica. Nel caso in cui gli episodi di aggressività rientrino nei sintomi di altri disordini neuropsichiatrici come indicato nel DSM-IV (Manuale Diagnostico e Statistico dei Disturbi Mentali, quarta edizione), l’aggressore viene sottoposto a trattamento psichiatrico, ma quando è considerato mentalmente sano e appartenente alla popolazione normale il suo unico destino è di essere punito secondo il sistema guidiziario penale (carcere). Il continuo ed eccessivo ricorrere all’aggressività fisica come caratteristica costante nel corso della vita può essere considerato una condizione patologica dei circuiti neurali e dei meccanismi molecolari implicati nella regolazione del comportamento aggressivo. Come tale, l’eccessiva aggressività dovrebbe essere trattata utilizzando i parametri biomedici di prevenzione e intervento. Tuttavia, a differenza di altre patologie, l’aggressività patologica manca ancora di una definizione appropriata e di biomarcatori fisiologici che potrebbero aiutare nell’identificazione di persone a rischio. Inoltre, non esiste ancora un appropriato trattamento che sia efficace nella cura di questa patologia. Per dare il via a programmi di prevenzione ed intervento basati sull’evidenza scientifica è necessario innanzitutto trovare una solida spiegazione delle cause nonché ai meccanismi neurobiologici dell’aggressività patologica. La presente tesi ha affrontato questo problema partendo dal presupposto che gli individui di una popolazione differiscono nella loro tendenza a mostrare aggressività a scopo offensivo in base ad una loro predisposizione genetica. Tali individui potrebbero avere caratteristiche fisiologiche che li renderebbero più inclini a diventare violenti. In particolar modo, “vincere” conflitti sociali attraverso l’utilizzo del comportamento aggressivo potrebbe innalzare la loro propensione ad 206
SINTESI
DELLA TESI IN ITALIANO
essere aggressivi in situazioni future. Perciò è necessario studiare gli effetti di tali esperienze in individui biologicamente predisposti e le conseguenze sul loro funzionamento cerebrale. Cercando un modo per studiare questi argomenti in laboratorio, in questa tesi si è deciso di trovare una possibile spiegazione fisiologica della violenza nel topo, assumendo che sia possibile estrapolare i risultati della ricerca alla situazione umana sulla base della vicinanza evolutiva delle due specie. Più specificatamente, in questa tesi sono state studiate tre linee di selezione genetica nel topo per elevati (SAL, TA, NC900) e bassi (LAL, TNA, NC100) livelli di aggressività. Il fenotipo violento: I topi SAL Nel gruppo di ricerca di Fisiologia Comportamentale dell’Università di Groningen, in Olanda, si è lavorato ad una definizione operazionale di violenza basata sul presupposto logico che sia possibile analizzare oggettivamente una serie di parametri comportamentali sia nelle persone violente che nel topo. I criteri utilizzati sono stati i seguenti: – temperamento aggressivo dall’infanzia/predisposizione genetica – aggressività mostrata in differenti situazioni – mancanza di discriminazione del contesto e della pericolosità dell’avversario – mancanza di auto-inibizione – scarsa componente preparatoria di minaccia ritualizzata rispetto all’attuale aggressione fisica. I risultati raccolti in questa tesi hanno evidenziato che la linea SAL, mostrando tutti i criteri elencati, sia il miglior modello per la violenza umana tra quelli proposti. Oltretutto, risulta che la violenza è associata ad uno stile comportamentale “proattivo” negli individui aggressivi. Similmente nell’uomo, tra le persone violente con personalità antisociale si ritrovano gli psicopatici. Essi possono condurre una vita di successo caratterizzata da una serie di vittorie sociali e da un considerevole uso di comportamento manipolativo, strumentale, orientato ad un fine prefissato, i cui tratti sono paragonabili a quelli riscontrati nella linea SAL. Oltre al loro caratteristico fenotipo comportamentale violento, i topi SAL mostrano determinate caratteristiche fisiologiche che non si riscontrano nelle altre linee meno violente ma altamente aggressive. Nell’uomo, la violenza (intesa come aggressività patologica in generale) è stata più volte associata a bassi livelli o ridotta neurotrasmissione di serotonina in aree cerebrali coinvolte nel controllo del comportamento. Risultati di questo studio hanno mostrato come bassi livelli di serotonina nella corteccia prefrontale siano riscontrabili nei topi SAL successivamente all’esposizione ad un’esperienza sociale durante la quale i loro livelli di aggressività hanno raggiunto la massima saturazione. Inoltre, un basso livello di serotonina non sembra essere la causa fisica dell’aggressività estrema ma piuttosto 207
SINTESI
DELLA TESI IN ITALIANO
la conseguenza di altri fattori. Un’eccessiva inibizione del sistema della serotonina dovuta a recettori 5-HT1A ipersensibili sembra essere una causa plausibile. Nel corso di un ripetuto paradigma di confronto sociale, gli autorecettori 5-HT1A dei topi aggressivi meno violenti sembrano andare incontro a desensibilizzazione. Diversamente, i topi della linea SAL che mostrano il fenotipo più violento sembrano essere resistenti a questo fenomeno, cosicché alla fine del paradigma sociale si formerebbe una stretta correlazione tra livelli di violenza e sensibilità degli autorecettori 5-HT1A. I bassi livelli di serotonina, soprattutto a livello intracellulare, e l’elevata attività degli autorecettori 5-HT1A potrebbero risultare in un’inibizione tonica del rilascio di serotonina. Conseguentemente tale processo risulterà in un’alterata neuromodulazione serotoninergica in aree-chiave per il controllo del comportamento, inclusa la corteccia prefrontale. D’altra parte, sembra che gli eterocettori 5-HT1A della corteccia prefrontale dei topi SAL siano più dinamici, come mostrato dall’elevata responsività intracellulare alla prova farmacologica. Il sistema della serotonina è anche coinvolto nella modulazione fisiologica a livello periferico. Come possibile conseguenza della bassa attività dei neuroni produttori di serotonina, i topi SAL mostrano inoltre una bassa frequenza del battito cardiaco in fase di riposo ed un’elevata temperatura corporea in fase di veglia e durante la risposta di stress. Non essendo stata ancora completamente decifrata la regolazione serotoninergica del sistema nervoso autonomo, non è possibile formulare ipotesi dirette sulle implicazioni a livello anatomico e funzionale. Tuttavia, è interessante notare che tali tratti fisiologici periferici sono potenziali biomarcatori della violenza. Da studi longitudinali condotti sull’uomo si è visto che gli individui della popolazione che seguono un percorso evolutivo di aggressività fisica cronica rappresentano il 4-5% della popolazione (Nagin and Tremblay 1999). Queste persone commettono la maggior parte dei crimini violenti. Una percentuale simile di individui precocemente aggressivi che sviluppano tendenze aggressive croniche è stata osservata in popolazioni di scimmie e roditori (van Oortmerssen and Busser 1989; Suomi, 2003; de Boer et al., 2003). Dagli studi presentati in questa tesi sembra che i topi SAL rappresentino questa parte della popolazione. Assenza di tratti violenti nei topi TA e NC900 Un risultato interessante della presente tesi è l’incapacità ad identificare un fenotipo violento nelle altre due linee aggressive. Innanzitutto, occorre considerare che il ceppo parentale fosse diverso per le tre coppie di linee selezionate. Solamente i topi SAL e LAL sono stati ottenuti da una popolazione selvatica; al contrario, le altre due coppie di linee sono state selezionate a partire dai ceppi di laboratorio Swiss-Websters e ICR. La forte secolare selezione fortuitamente esercitata sui topi 208
SINTESI
DELLA TESI IN ITALIANO
da laboratorio in maniera non sistematica per la loro docilità, facilità alla manipolazione e buona capacità riproduttiva in cattività potrebbe aver causato la scomparsa del fenotipo violento. Ciò potrebbe svelare un errore sistematico negli innumerevoli studi condotti su ceppi standard da laboratorio che tentano di descrivere le caratteristiche neurobiologiche e fisiologiche di animali aggressivi che non appartengono alla percentuale violenta della popolazione naturale. Secondariamente, l’eterogeneità dei criteri di selezione potrebbe aver contribuito all’assenza del fenotipo violento nelle linee TA e NC900. I topi SAL sono stati selezionati in base al tempo di latenza all’attacco nella loro gabbia territoriale, una misura della prontezza all’attacco che potrebbe includere impulsività e controllo comportamentale. Al contrario, i topi TA e NC900 sono stati selezionati per l’intensità dei loro combattimenti in gabbia neutrale. L’interazione agonistica in gabbia neutrale potrebbe avere una funzione diversa nel sistema sociale del topo. Un topo che è subordinato nel suo territorio ma che viene cacciato fuori dal suo deme potrebbe ancora ottenere la dominanza sociale in un territorio neutrale e perciò comportarsi in maniera aggressiva. Inoltre, i topi NC900 sono stati selezionati in base alla frequenza degli attacchi, piuttosto che alla durata. Questi topi, infatti, mostrano una veloce alternanza di brevi stati comportamentali, inclusi gli stati aggressivi. Tuttavia, la durata totale del loro comportamento offensivo potrebbe non essere estremamente elevata. L’elevata frequenza di cambiamenti di comportamento di questi topi potrebbe significare una predisposizione comportamentale indecisa, molto cauta e comunicativa, diversamente dall’insistenza dei topi SAL nel continuare il combattimento. L’aggressività è soltanto una manifestazione di una più vasta caratteristica comportamentale. Seguendo questa premessa si giunge ad un’importante conclusione. La predisposizione genetica all’aggressività non sempre porta alla patologia comportamentale, anche dopo esacerbanti esperienze di vittoria sociale. Al contrario, gli individui predisposti geneticamente ad essere molto proattivi e rigidi, se esposti a routine di intensificazione dell’aggressività, potrebbero essere a più alto rischio di compiere atti di violenza (intesa come aggressività fisica dannosa, non-inibita, e sproporzionata); invece, gli individui con predisposizione genetica alla bassa emotività potrebbero mostrare tendenze aggressive ad ottenere dominanza sociale senza incorrere nella patologia. Le linee T- e NC- dimostrano che l’aggressività può essere parte di diverse componenti del carattere. Diversamente dall’aggressività mostrata dagli individui proattivi e più tendente a trasformarsi in violenza (SAL), un’altra tendenza all’elevata aggressività potrebbe essere la mera appartenenza ad un profilo comportamentale caratterizzato da bassi livelli di ansia (TA e NC900) e a alti livelli di metabolismo e /o attività fisica (NC900). A livello di sistema nervoso centrale, i livelli di serotonina, e più in particolare gli autorecettori 5-HT1A, sembrano essere 209
SINTESI
DELLA TESI IN ITALIANO
coinvolti nella distinzione tra stili alternativi di risposta allo stress, mentre la noradrenalina sembra caratterizzare il fenotipo aggressivo/poco ansioso. La dopamina sembra essere coinvolta nel fenotipo proattivo quando si spinge a diventare aggressività patologica, attraverso la formazione di comportamenti di routine. A livello di fisiologia periferica, l’aggressività è correlata ad un battito cardiaco più lento a riposo e a ipertermia in fase attiva. Le fluttuazioni giornaliere in frequenza cardiaca e temperatura sono più elevate nelle linee selezionate per stile di risposta allo stress proattivo/reattivo, il che suggerisce che i tratti comportamentali del fenotipo proattivo-aggressivo siano associati ad un make-up fisiologico adeguato ad eseguire ed a perseverare in eventuali combattimenti futuri. Conclusione e prospettive per il futuro In questa tesi viene fornito per la prima volta un modello di topo valido per la violenza umana. Tale modello è rappresentato dalla linea di selezione genetica SAL, ottenuta da topi selvatici catturati a Groningen (Paesi Bassi) in seguito a ripetute esperienze di vittoria sociale. In conclusione si suggeriscono ulteriori studi preclinici che includano l’utilizzo della linea SAL, insieme alla sua contrapposta linea LAL, in campo genetico, farmacologico e di sviluppo neuro comportamentale, al fine di esplorare nuovi metodi per la prevenzione e l’intervento contro la violenza umana.
210
SINTESI
DELLA TESI IN ITALIANO
211
ACKNOWLEDGEMENTS
Thanks!!!! Did you enjoy reading this book? Then you should know that I could have not achieved any part of it without other people. Some of them are co-authors of the chapters, others are acknowledged at the end of each chapter. Finally, others are never mentioned throughout the chapters. I will start with the biggest emotional support, the one that usually you see at the end. I would like to thank my fiancée, Dr Tim W. Fawcett. His contribution to this book was essential. He has been there with me from even before the start of the project, when I had to choose whether to embark on it or not, to the middle, with ups and downs, till the very end, making the .tiff figures on Saturday evening, and advising me on what to wear at the defense. Let alone all the corrections to the English and to the statistical analyses. Thank you, Tim! Knowing that you were there all the time made it possible. I am really thankful to the thesis promotor, Prof Jaap M. Koolhaas, and the copromotor, Dr Sietse F. de Boer. I was very lucky to work with them, and I hope I have at least gained a minimum part of their knowledge and “scientific wisdom”. A big thank to Dr Claudio Carere, my “mentor”. We don’t really have one at the University of Groningen, but I think if I had to choose one, he would be my mentor. His collaborative and enthusiastic attitude resulted in extra projects that are not mentioned in my thesis in the field of neuroendocrinology of personality. Thanks to the paranymfen: Deepa Natarajan and Ilir Haxhi for their great help and friendship through these years. With Deepa I shared the lab and the project of mouse aggression. Many concepts and ideas mentioned in this thesis come from discussions with her and her thesis, which I strongly recommend, is complementary to this one. Ilir is a great friend and his social skills and expertise in business and economics make him an ideal support in going through the organization of the defense. Thanks to the whole Animal Physiology department, which, I know, it does not exist officially, as everybody keeps saying, but it exists in the feeling of the people of the department. It is a really vibrant corridor, the D-first floor. In particular, Bauke, Anne, Caroline, Deepa, José, Giulia, Alinde, Ramon, Auke, Riejanne, have been great colleagues in the Behavioural Physiology group. I am really thankful to the old PhD crew: Amalia, Anghel, Robbert, Viktor, Angelique, Izabella, Aniek. As well as to the new one: Arianna, Caroline, Nikoletta, Girstaute, Ivi, Marcelo, Roelina, Henriette, Timur, Paulien, Simon, Gretha, Stefano and Peter. I am really
212
ACKNOWLEDGEMENTS
thankful to the technical personnel: Jan K. and Jan B., especially. I am very thankful to the administrative personnel: Henk, Joke and Pleunie. I am very thankful to the senior academic staff of the department: Anton Scheurink, GertJan van Dijk, Jan Strubbe, Ingrid Nijholt, Paul Luiten, Uli Eisel, Eddy van der Zee and Peter Meerlo. I enjoyed their welcoming attitude, their will to discuss and help interpreting data in any moment of the day. Many thanks also to the Electron Microscopy unit at the Medical Faculty. I am really thankful to Han van der Want, Henk de Weerd and everybody that was so nice and helpful with me in their lab, leading to the realization of chapter 8. Thanks to Han de Vries, from University Utrecht. His contribution has been so quick and mainly through email, but it led up to a major part of the thesis (see chapter 3). Thanks to the reading committee and to the defense committee for taking their time and showing such interest in this thesis. Thanks for Bachelor and Master Students that participated at this project. In particular, a big thank to Cees, Chris, Dirk -Jan, and Jan S. Thanks to Dick Visser for the last-moment layout. Thanks to the BCN for funding all those courses, activities and conferences, in particular thanks to Diana Koopmans and Britta Küst. Thanks to the RuG and to the FWN for funding the scholarship and in particular thanks to Anneke Toxopeus for sorting out all the tax problems derived from the confusion between the two different types of PhD students: bursalen and AIO’s. Thanks to the Theoretical Biology group, led by Prof. Franjo Weissing, for being so welcoming and scientifically stimulating and thanks to all his students and colleagues, in particular Daan, Ido, Max, Bram, Ana, and Liliana. Thanks to Irene Mateo-Leach, Magdalena Kozielska, Aitana Peire, Antonella Pellicoro, Marian Comas, Nikolaus von Engelhardt, Delphine Theard, Teresa Gonzalo; Barbara Feldmeyer and Giorgio Barbareschi for being such good friends and for teaching me how to get through a PhD successfully. Thanks to all the other friends and colleagues from the Biology center, and to all the friends outside biology, from the music ensembles and from Italy. Un ringraziamento particolare alla mia famiglia. Grazie papà, Luisa, Stefania e Riccardo per essere stati così comprensivi e generosi in questi lunghi quattro anni lontano da casa.
Grazie!!!!
213
ACKNOWLEDGEMENTS
214
