VU Research Portal

Genetic architecture and behavioral analysis of attention and impulsivity Loos, M.

2012

document version Publisher's PDF, also known as Version of record

Link to publication in VU Research Portal

citation for published version (APA) Loos, M. (2012). Genetic architecture and behavioral analysis of attention and impulsivity.

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?

Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

E-mail address: [email protected]

Download date: 28. Sep. 2021

Genetic architecture and behavioral analysis of attention and impulsivity

Maarten Loos

1

About the thesis The work described in this thesis was performed at the Department of Molecular and Cellular Neurobiology, Center for Neurogenomics and Cognitive Research, Neuroscience Campus Amsterdam, VU University, Amsterdam, The Netherlands. This work was in part funded by the Dutch Neuro-Bsik Mouse Phenomics consortium. The Neuro-Bsik Mouse Phenomics consortium was supported by grant BSIK 03053 from SenterNovem (The Netherlands).

Publication of this thesis was financially supported by: Vrije Universiteit Amsterdam, The Netherlands Sylics, services in mouse phenomics, cellomics and bioinformatics. Trade name of Synaptologics B.V., Amsterdam, The Netherlands

About the cover The cover illustrates the emergence of complex phenomena from simple rules. The image was adapted from a painting of an unknown artist exhibited in Centre national d'art et de culture Georges-Pompidou, Paris, France in August 2008.

2 VRIJE UNIVERSITEIT

Genetic architecture and behavioral analysis of attention and impulsivity

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. L.M. Bouter, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de faculteit der Aard- en Levenswetenschappen op donderdag 10 mei 2012 om 15.45 uur in het auditorium van de universiteit, De Boelelaan 1105

door Maarten Loos geboren te Roosendaal

promotor: prof.dr. A.B. Smit copromotoren: prof.dr. S. Spijker dr. T. Pattij

leescommissie: dr. A. Palmer dr. M.M. van Gaalen prof.dr. C.M.A. Pennartz dr. M.J.H. Kas prof.dr. E.J. de Geus dr. L. Diergaarde dr. R.O. Stiedl

Voor mijn ouders

Contents

Chapter 1 General introduction 9

Chapter 2 Dopamine receptor D1/D5 gene expression in the medial 21 prefrontal cortex predicts impulsive choice in rats

Chapter 3 Inhibitory control and response latency differences between 37 C57BL/6J and DBA/2J mice in a Go/No-Go and 5-choice serial reaction time task and strain-specific responsivity to amphetamine

Chapter 4 Activity and impulsive action are controlled by different 59 genetic and environmental factors

Chapter 5 Independent genetic loci for sensorimotor gating and 87 attentional performance in BXD recombinant inbred strains

Chapter 6 Insights in the genetic architecture of impulsivity in mice 111

Chapter 7 Enhanced alcohol self-administration and relapse in a highly 127 impulsive inbred mouse strain

Chapter 8 General discussion 143

References 151

Nederlandse samenvatting 169

Dankwoord 175

Curriculum Vitae 178

Peer-reviewed publications 179

7

8 Chapter 1

General Introduction

9

10 Introduction

Complex genetic traits With his laws of inheritance presented in 1865, Gregor Mendel explained how biological characteristics, such as pea shape and flower color, were inherited (Mendel, 1866). The presence of only one genetic factor appeared to be necessary and sufficient to determine flower color, and the inheritance of this genetic trait therefore could be explained in simple terms. In the same year Sir Francis Galton, a biometrician devoted to quantifying human characteristics such as reaction time, described that many characteristics do not follow a dichotomous distribution but appear in several intermediate quantifiable flavors (Galton, 1865). He compared biological characteristics of parents and children, and established the degree of correlation between relatives as index of the heritability. The rivalry between the schools of dichotomous Mendelian geneticists and the quantitative biometricians was resolved in the 1918 by Sir Ronald Fisher. He statistically demonstrated that continuous characters as measured by biometricians could be governed by the action of a large number of Mendelian factors (Fisher, 1918), and the ‘complex genetic trait’ was born. The molecular inheritance of these factors and a functional mechanism by which they were transmitted to the offspring were not defined until 1953, when James Watson and Francis Crick published and discussed the structure of DNA (Watson & Crick, 1953) and suggested a DNA replication mechanism. Over the years, variation in most human characteristics (even traits such as IQ) has turned out to be, albeit to different extent, genetic in origin and complex in nature. Many genes underlie human behavior and their unique combinatorial inheritance includes the risk to develop various prevalent psychiatric and neurological diseases, such as attention deficit hyperactivity disorder (ADHD), schizophrenia and drug addiction (Faraone & Biederman, 1998; Greenwood et al., 2007; Kreek et al., 2005). By identifying genetic variants in the human population that predispose individuals to disease, quantitative genetics can yield insights into the contribution of multiple genes to complex disease traits. This is of great importance given the fact that the molecular basis of most psychiatric and neurological diseases is at best poorly understood.

Dissecting complex traits into core elements will facilitate gene finding In genetic linkage studies and genome wide association studies, the aim is to relate the presence of a specific genetic variant to the observation of a specific phenotype. Over the last decade it has become clear that for prevalent human psychiatric and neurologic diseases, with overall substantial heritability, it is difficult to identify the causal genetic variants. To date, association studies have cumulated evidence for a limited number of genes associated with the clinical diagnosis of psychiatric disease, e.g. ADHD (DRD4, DRD5, SLC6A3, SNAP-25 and HTR1B; Faraone & Mick, 2010) and Schizophrenia (NRG1 and DTNBP1; Munafo et al., 2006 and Riley & Kendler, 2006). Together these risk alleles explain only a fraction of total risk contributed by genetic factors to these

11 Chapter 1 diseases despite tremendous advances in SNP genotyping technology and increases in samples sizes owing to large collaborative projects. Two explanations for this ‘missing heritability’ are being discussed at the moment. First, many genetic variants may act together, each with a small effect (< 1.0 %) that can barely be detected in even the largest collaborative studies. Secondly, each affected person may carry a different variant with a modest effect size that would go unnoticed in current genotyping assays focusing on common genetic variants and would not be picked up by genetic linkage studies due to too low effect size. Most psychiatric diseases are heterogeneous in nature and patients may have varying degrees of severity of each of the identified symptoms. For instance, in ADHD three subtypes are being recognized, characterized by symptoms of inattention, of hyperactivity/impulsivity, and a combined subtype. In fact, it is under debate whether symptoms of hyperactivity/impulsivity and inattention in ADHD reflect the same or separate disorders (Derefinko et al., 2008; Diamond, 2005; Milich et al., 2001). It is conceivable that quantitative genetic studies are facilitated when the disease phenotype is partitioned better into its core elements, to each of which a limited number of genetic variants contribute (Fig. 1; de Geus, 2002; Kas & Van Ree, 2004).

Figure 1 | The genetic architecture of complex traits. (a) Many genetic polymorphisms on the genome contribute to higher (+) or lower (-) phenotypic values of brain phenotypes (b; grey ovals) and behavioral phenotypes (c; white rectangles). These phenotypes are core elements that are intermediate to genome and disease (d; grey rectangles). They are genetically less complex than the disease itself and are therefore a suitable starting point for a genetic mapping study to identify disease-contributing genes. Schematic modified after De Geus (2002).

12 Introduction

Translation of psychiatric disease symptoms into rodent paradigms Executive functions are a diverse collection of cognitive processes such as rule acquisition, inhibiting inappropriate actions, planning, cognitive flexibility and selective attention. These executive functions have in common that they require an intact prefrontal cortex. Because these functions are highly demanding, they are sensitive to deterioration by e.g. ageing, chronic stress and occur under certain neurological conditions, such as in Alzheimer’s and Parkinson’s disease (Perry & Hodges, 1999; Rodriguez-Oroz et al., 2009) and are affected in many psychiatric diseases, such as schizophrenia, ADHD and addiction. In humans, deficits in executive functions can objectively be quantified in computerized response tasks. Heritability estimates clearly indicate that measures obtained in these tasks are sensitive to genetic effects (Bidwell et al., 2007). Recently, the first linkage studies reported successfully employed neuropsychological measures to identify chromosomal locations contributing to deficits in executive functions that are observed in multiple psychiatric disorders (Table 1).

Table 1 | Mapping studies employing executive functions as intermediate phenotypes. Chr. Clinical Neuropsychological measure location indication Patients with Sustained attention: undegraded CPT hit rate (Lien et 12q24.32 schizophrenia al., 2010) Patients with Wide range of neurocognitive traits associated with 3q13 ADHD ADHD (Doyle et al., 2008) Patients with False alarm rate on the undegraded and degraded CPT 2q22.1 schizophrenia (Lien et al., 2011) Patients with Variability in reaction time in a motor timing task 2q21.1 ADHD (Rommelse et al., 2008) Patients with Verbal working memory: Digit span backwards 13q12.11 ADHD (Rommelse et al., 2008)

These heritable core elements of complex human disorders play a pivotal role in the translation into genetic animal models of the disorder, for which a large suite of genetic, molecular and physiological techniques is more readily available than for human subjects. Indeed, recently a QTL was reported in a mapping study of a translation of a neuropsychological measure of behavioral flexibility (Laughlin et al., 2011). Two classes of executive functions are well studied in both humans and rodents, i.e., the capacity to selectively attend to stimuli (attention) and impulse control (i.e. preventing impulsive behavior). Searching for the underlying genetic components of these two aspects of behavior in rodents is the primary focus of this thesis.

13 Chapter 1

Attention The continuous performance task (CPT) of attentional processing (Beck et al., 1956) has been adapted for rodents into the widely used 5-choice serial reaction time task (5-CSRTT; Robbins, 2002). In this task, rodents are trained to respond to a brief, usually 1 s light stimulus in one of five response holes. During one 5- CSRTT session that typically lasts for 60 to 100 trials during which rodents can obtain a reward by responding into the illuminated stimulus hole. Response accuracy is the primary measure of attention, and is defined by the percentage of correct responses into an activated response hole over the total number of correct and incorrect responses. In addition, intra-individual variability in correct response latencies (hereafter referred to as response variability) has been recognized as measure of lapses in attention (Castellanos & Tannock, 2002; Leth-Steensen et al., 2000; Loos et al., 2010b; Sabol et al., 2003; Sergeant & van der Meere, 1990). The interest in intra- individual variability in response latencies in human studies has grown over the last decade following the observation of increased intra-individual variability in ADHD (Bidwell et al., 2007; Castellanos et al., 2006; Castellanos & Tannock, 2002; Klein et al., 2006; Leth-Steensen et al., 2000; Spencer et al., 2009). Whereas the peak of a response latency distribution could be viewed as an index of processing speed, larger intra-individual variability in this distribution has been interpreted as lapses in attention (Klein et al., 2006; Sergeant & van der Meere, 1990). A failure to respond into any stimulus hole after the presentation of the light stimulus, i.e. an error of omission, is a third measure of attention in the 5- CSRTT. This measure may reflect the inability to pay attention to the task in every trial. However, disruptions of locomotor behavior, the motivation to perform or the ability to execute a cognitive task in general can affect this parameter as well. For the interpretation of errors of omission in terms of deficits in attention it is therefore imperative to rule out alternative explanations, for instance by analyzing latency to collect rewards from the magazine in the 5- CSRTT or a test of locomotor function in another apparatus.

Impulsivity Several operant tasks of impulsivity exist (Evenden, 1999; Winstanley et al., 2006a), broadly separated into those that measure impulsive action and impulsive choice (Winstanley et al., 2006a). Whereas impulsive action involves the loss of inhibitory control over pre-potent response patterns, impulsive choice reflects less beneficial decision-making processes. One frequently employed index to measure impulsive choice is the inability to tolerate delay of reinforcement. In such tasks, ADHD patients show increased preference for immediate small monetary over delayed larger – and therefore more beneficial – rewards (Barkley et al., 2001; Solanto et al., 2001; Sonuga-Barke, 2002; Sonuga-Barke et al., 1992). Methylphenidate (Ritalin), currently the most commonly prescribed drug

14 Introduction for the treatment of ADHD, reduces impulsive choice in a delayed reward task in healthy human volunteers (Pietras et al., 2003) indicating that this task relates to an important clinical symptom, i.e., an intermediate phenotype of ADHD. The rodent analogue of this task measures the preference for immediate small (1 food pellet) over delayed larger reward (4 food pellets). With respect to impulsive action, currently the 5-CSRTT is the most frequently employed paradigm (Eagle & Baunez, 2010), which is similar to errors of commission in the CPT (Beck et al., 1956). In the 5-CSRTT, a premature response to a food predictive stimulus is a measure of impulsive action. Other tasks of impulsive action require rodents to signal situations in which responding should be inhibited such as in a Go/No-Go task, or in which an already initiated response should be terminated such as in a stop signal reaction time task.

Dissecting complex traits by exploiting genetic and environmental variation In laboratory animals, such as outbred Wistar rats, individual differences in impulsive action and impulsive choice have been documented that are stable over many days and weeks (Diergaarde et al., 2008). As such, impulsivity is a quantitative trait and should have its molecular and cellular underpinning in frontocortical and striatal brain regions critically involved in impulsivity (for reviews see Cardinal, 2006 and Winstanley et al., 2006a). However, because these strains of rats are bred to maintain a certain degree of genetic variation among individuals, the individual differences can arise from genetic and/or environmental differences among individual rats (Crabbe & Cunningham, 2007). Using such outbred populations of rodents, it is therefore possible to identify molecular correlates of attention and impulsivity, but more difficult to determine the percentage of contribution of genetic variation to a given trait. In particular, genetically identical (isogenic) inbred strains of mice provide the opportunity to separate genetic and environmental factors (Fig. 2). When kept under highly controlled conditions, differences between inbred mouse strains result from additive genetic effects and gene-by-environment interactions (Crabbe et al., 1999). In contrast, behavioral differences among isogenic mice of the same strain result from environmental effects idiosyncratic to each

Figure 2 | Generation of BXD recombinant inbred strains of mice. The commonly used laboratory inbred strains C57BL/6J and DBA/2J were crossed (B  D, hence BXD) to obtain an isogenic F1 population. During F1  F1 matings, genetic recombination occurs randomly at a rate of about 2 events per chromosome, yielding a genetically diverse F2 population. Recombination events continue to accumulate during inbreeding of BXD lines, until fully isogenic recombinant inbred lines are produced.

15 Chapter 1 individual. Comparing within-strain (environmental) with between-strain (genetic) variation therefore indicates the degree of heritability of any given trait. Furthermore, panels of recombinant inbred strains of mice, derived from an intercross of common inbred strains can be used to identify quantitative genetic loci (QTL) underlying phenotypes such as impulsivity (see Fig. 3). Currently, the largest and therefore most widely employed panel of recombinant inbred strains was derived from C57BL6/J and DBA2/J mice (Fig. 2; Peirce et al., 2004), which are renowned for their difference in alcohol consumption behavior (Crabbe et al., 1983). Furthermore, mice of these two strains show profound differences in their brain dopaminergic system (Cabib et al., 2002; D'Este et al., 2007) that may be relevant for research on dopaminergic dysfunction in relation to psychiatric disease. Because these panels of inbred mouse strains are retrievable genetic resources, they facilitate the systematic integration of data obtained in different studies. For instance, the available databases of brain gene expression of these strains aid to efficiently identify candidate genes within QTL regions (Chesler et al., 2005). Finally, genetic reference panels of inbred strains can be used to investigate the relation between any pair of phenotypes measured, from expression levels of proteins in the synapse to behavioral parameters. High correlation between two phenotypes across strains would indicate that these measures are controlled by common genetic or gene-by-environment interaction effects (genetic correlation), albeit such genetic correlations inherently contain some residual

Figure 3 | Mapping a locus on the genome (i.e. QTL) harboring a quantitative trait gene (QTG) affecting the trait. (a).The genotype, i.e. whether the locus originates from founder C57BL/6J or DBA/2J, of each strain is determined at regular intervals across the genome (black denotes genotype C57BL/6J, white denotes genotype DBA/2J). The change in color within the genome of a strain reflects the occurrence of recombination between stretches of C57BL/6J and DBA/2J origin during generation of the strain. (b) Recombinant inbred strains can quantitatively differ in the expression of a trait, such as the level of impulsivity. By statistically evaluating the difference in phenotype between carriers of a C57BL/6J versus DBA/2J allele at each locus, the probability can be determined whether a given locus is influencing the trait. The size of the locus is determined by the number and proximity of recombination events surrounding the QTG. The dashed lines indicate upstream and downstream genomic regions surrounding the locus with the highest probability for which – due to linkage – still a significant influence on phenotypes is detected. Due to this effect, a QTL region can become large and can contain up to a few hundred genes.

16 Introduction environmental variation if they are derived from experimentally estimated strain means (Crusio, 2006). In contrast, high within-strain correlations would indicate that these behaviors are controlled by common environmental factors (environmental correlation).

Aim and outline of this thesis The aim of the research projects described in this thesis was to elucidate genetic factors responsible for individual variation in attention and impulsive behavior, in the context of the psychiatric diseases ADHD, drug addiction and schizophrenia. In addition to the symptoms of impulsivity and deficits in attention that these psychiatric diseases have in common, each of them is accompanied by the well-documented disease-specific symptoms of hyperactivity (ADHD), deficits in sensorimotor gating (Schizophrenia) and drug seeking (addiction). To investigate each of these symptoms separately, I therefore employed tailored assays in my thesis. The aim in chapter 2 was to understand the role of genes that have been strongly associated with ADHD in impulsive choice behavior. Rats are capable of performing a delayed reward paradigm and show individual variation in impulsive choice behavior (Diergaarde et al., 2008). Correlation analysis between behavior and gene expression suggested a contribution of dopamine D1- like receptors located in the mPFC to impulsive choice behavior. The functional role of these receptors was demonstrated by the disrupting effect of infusing dopamine D1-like receptor agonists and antagonists into a sub region of the medial prefrontal cortex (mPFC) on choice behavior. In contrast to inflexible and unstable performance in a delayed reward paradigm, mice were capable of acquiring tasks that rely on sustained attention and inhibitory control, such as the 5-CSRTT and Go/No-Go task (chapter 3). Moreover, I observed differences in the expression of dopamine D1, 4 and 5 receptors in the mPFC of these strains, possibly related to the strain differences in behavior. Differences in inhibitory control between common inbred strains appeared large and reproducible (chapter 4). However, both the locomotor response in a novel open field (Piazza et al., 1989) and deficits in inhibitory control (Belin et al., 2008; Dalley et al., 2007) are predictive of addictive behavior, questioning to what extent inhibitory control and the response to novelty are genetically independent. Therefore, I measured the response to novelty in a range of different assays and identified the lack of a genetic relation between this response and inhibitory control in the 5-CSRTT, indicating that QTL studies into both phenotypes were valuable. In chapters 5 and 6, I analyzed the QTL that (i) underlie attentional performance and inhibitory control in the 5-CSRTT, and (ii) sensorimotor gating in a test of prepulse inhibition. The identified QTL did not overlap for the two behavioral phenomena. I also analyzed the genetic correlation among any of these

17 Chapter 1 phenotypes and did not identify a genetic correlation. Using data on single nucleotide polymorphisms (SNPs) and brain gene expression of these strains, as well as the association of QTL genes with diagnosis of ADHD and/or schizophrenia I selected candidate genes potentially involved in the respective behaviors. In humans and rodents, increased impulsivity scores are predictive of enhanced addictive behavior (Belin et al., 2008; Caspi et al., 1996; Dawes et al., 1997; Diergaarde et al., 2008). Therefore, in chapter 7 I investigated whether reduced inhibitory control, as observed in the highest impulsive BXD strain, would predict enhanced addictive behavior in a paradigm of alcohol-seeking behavior. Finally, chapter 8 summarizes the findings on intermediate phenotypes and discusses these in the context of the psychiatric diseases of ADHD, addiction and schizophrenia. Furthermore, several avenues into new research lines were opened during my thesis, and I discussed the most promising ones in this final chapter.

18

19

20

Chapter 2

Dopamine receptor D1/D5 gene expression in the medial prefrontal cortex predicts impulsive choice in rats

Maarten Loos, Tommy Pattij, Mieke C. W. Janssen, Danielle S. Counotte, Anton N. M. Schoffelmeer, August B. Smit, Sabine Spijker, Marcel M. van Gaalen

Cerebral Cortex, 2010;20(5):1064–70

21

22 mPFC D1/D5 receptors and impulsive choice

Abstract A neuropsychological hallmark of attention-deficit/hyperactivity disorder (ADHD) is the reduced ability to tolerate delay of reinforcement, leading to impulsive choice. Genetic association studies have implicated several genes involved in dopaminergic neurotransmission in ADHD. In this study, we investigated whether differences in the expression level of these dopamine- related genes of rats predict the individual level of impulsive choice. Among all frontostriatal brain regions tested, only in the medial prefrontal cortex (mPFC) we observed significant positive correlations between impulsive choice and transcript levels of the dopamine receptor D1, the dopamine receptor D5 and calcyon. Local mPFC infusions of the dopamine D1/D5 antagonist SCH 23390 and agonist SKF 38393 resulted in increased impulsive choice, in agreement with the idea that endogenous dopamine receptor D1/D5 stimulation in the mPFC promotes the choice of large delayed rewards. Together, these data indicate that this class of dopamine receptors in the mPFC plays a pivotal role in impulsive choice and aberrancies thereof might contribute to ADHD symptomatology.

Introduction Attention-deficit/hyperactivity disorder (ADHD) is a highly disruptive, disabling disorder clinically characterized by inattention, hyperactivity and impulsiveness. ADHD patients show deficits on different neuropsychological tasks. Current theories suggest that ADHD is the result of dysfunctions in a spectrum of neuropsychological processes (Nigg et al., 2005; Sergeant et al., 2003; Sonuga- Barke, 2002, 2005). One of these dysfunctional processes is the inability to tolerate delay of reinforcement, as ADHD patients show increased preference for immediate small over delayed larger – and therefore more beneficial – rewards (Barkley et al., 2001; Solanto et al., 2001; Sonuga-Barke, 2002; Sonuga-Barke et al., 1992). This results in a so-called ‘impulsive choice’ for the small reward. Methylphenidate (Ritalin), currently the most commonly prescribed drug for the treatment of ADHD, reduces impulsive choice in a delayed reward task in humans (Pietras et al., 2003) indicating that this task relates to an important clinical symptom of ADHD. Manipulation of dopamine, noradrenalin and serotonin neurotransmission is known to modulate impulsive choice (Robinson et al., 2008; van Gaalen et al., 2006b; Winstanley et al., 2005; for review see Pattij & Vanderschuren, 2008). Dopaminergic neurotransmission has received particular attention in this respect, since treatment with drugs that specifically (GBR 12909) or less specifically (methylphenidate, amphetamine and cocaine) inhibit dopamine reuptake decrease impulsive choice as shown in both human (Pietras et al., 2003) and rodent studies (van Gaalen et al., 2006b; Wade et al., 2000). Recently, genetic studies have identified associations between the occurrence of ADHD and genes involved in dopaminergic neurotransmission (for review see Waldman & Gizer, 2006),

23 Chapter 2 including genes encoding the dopamine receptor D1 (DRD1; Misener et al., 2004), the dopamine receptor D4 (DRD4; Faraone et al., 2001), the dopamine receptor D5 (DRD5; Lowe et al., 2004), the dopamine transporter (SLC6A3 or DAT1; Faraone et al., 2005), dopamine beta hydroxylase (DBH; Daly et al., 1999) and calcyon (CALY; Laurin et al., 2005). Interestingly, the latter gene encodes a protein product that is involved in regulation of the affinity state of D1 receptors for agonists (Lidow et al., 2001). Thus far, polymorphisms within the protein-coding sequence of these dopamine-related genes have not been identified, except for the 7-repeat polymorphism in the gene encoding the dopamine receptor D4 that alters affinity to certain antipsychotics (Faraone et al., 2001; Van Tol et al., 1992). This may suggest that polymorphisms in genes causally related to ADHD reside in regulatory sequences instead, and exert their effect by altering the expression levels of these genes. As yet, it is unknown whether differences in gene expression of dopamine-related genes are associated with specific deficits observed in ADHD, such as impulsive choice. Previously, we have shown that rats vary in their individual level of impulsive choice and that this phenotype is stable over time (Diergaarde et al., 2008). In the present study, we investigated whether the individual level of impulsive choice could be predicted by the expression levels of the aforementioned dopamine- related genes in frontocortical and striatal brain regions, areas known to be critically involved in impulsive decision making (for reviews see Cardinal, 2006; Winstanley et al., 2006a). The gene expression of two dopamine receptors (D1 and D5), and the protein calcyon significantly correlated with impulsive choice, notably exclusively in the infra- and prelimbic regions of the medial prefrontal cortex (mPFC). By intracranial infusions of the D1/D5 antagonist SCH 23390 and agonist SKF 38393 into the mPFC we examined whether D1/D5 receptors in the mPFC are functionally involved in impulsive choice. To rule out possible agonistic effects of SCH 23390 on 5-HT2c receptors (Millan et al., 2001), SCH 23390 was also co-infused with the 5-HT2a/2c receptor antagonist ketanserin.

Materials and Methods

Subjects Male Wistar rats (200-250 g, Harlan CPB, Horst, The Netherlands) were housed two per cage (lights on from 7 PM to 7 AM) with water available ad libitum during the entire experiment. After a habituation period of 2 weeks, animals were food restricted to 85%–90% of their free feeding weight before training of the operant delayed reward task started. The Animal Care Committee of the VU University Amsterdam approved all experiments.

24 mPFC D1/D5 receptors and impulsive choice

Delayed reward paradigm Operant test chambers (Med Associates, St. Albans, VT) were equipped with five nose poke holes with stimulus lights and infrared detectors and a pellet dispenser (45 mg, Noyes Precision Pellets, New Brunswick, NJ) at the opposite wall. Five sessions were scheduled per week, one session each day. More details on the procedure of training the animals to perform the operant delayed reward paradigm have previously been described (van Gaalen et al., 2006b). Briefly, after training for ~12 weeks, the task required animals to respond to a 10 second light stimulus in the middle hole at the beginning of each trial (initiation period), which after a response, was followed by 10 second light stimulus in the left and right hole immediately adjacent to the middle hole (choice period). A response in one hole extinguished both stimuli and delivered a small reward (one pellet) into the magazine, while a response in the other hole extinguished both stimuli and started the delay period after which a large reward (four pellets) was delivered into the magazine. For an individual rat the same hole was always associated with large reward and left and right associations were counterbalanced across rats. A session consisted of 60 trials in total, with 5 blocks of 12 trials. The first 2 trials of each block were forced trials during which once a small reward and once a large reward could be chosen (randomly presented). In the subsequent 10 trials a rat was free to choose. After each block of 12 trials, the delay period between a response into the hole associated with the large reward and delivery of the large reward was programmed to increase. No conditioned stimuli were presented during the delay period. The inter-trial interval was programmed such that the duration of each trial was 100 s. Failures to respond within the initiation period were registered as errors of omission. Rats that persistently omitted the large reward during all forced and choice trials were excluded. Percentage choice for the large reward was calculated for each block of 10 trials; for each delay [number of large rewards earned * 100 / (number of large rewards earned + number of small rewards earned)]. In this paradigm, an enhanced level of impulsive choice is defined as a reduced percentage of choice for the large reward (e.g. % impulsive choice = 100 – % choice for the large reward).

Experiment 1: Analysis of correlations between impulsive choice and the expression of dopamine-related genes Delayed reward paradigm: After 21 training sessions a group of 24 rats learned to perform the delayed reward paradigm at the full range of delay periods (0, 10, 20, 40 and 60 s). Rats were trained at this range of delay periods for 5 additional sessions, before their preference for the large reward was monitored for 17 baseline sessions. Within each of these sessions, the individual level of impulsive choice was defined as the percentage choice for the large reward during trials with intermediate delays (10 and 20 s; a total of 20 trials). For correlation analysis, the average individual level of impulsive choice was calculated across these 17 baseline sessions.

25 Chapter 2

Locomotor response to novelty: On the day between the 14th and 15th baseline session rats were transferred to a dimly lit testing room and introduced in square grey plastic boxes (50  50 cm) where their behavior was analyzed using a camera connected to a computer equipped with EthoVision (version 3.1, Noldus, Wageningen, The Netherlands). The total distance moved during 60 min was taken as measure of their locomotor response to novelty. Gene expression measurements: Immediately after the 17th baseline session, rats were decapitated and brains rapidly frozen using –80 ºC isopentane. In a cryostat, the brains were sliced into 200 μm coronal sections and the following brain regions were isolated according Paxinos and Watson (1998): orbitofrontal cortex (OFC), agranular insular cortex (AI), prelimbic together with infralimbic cortex (mPFC), anterior cingulated cortex (Ac), nucleus accumbens core (NAcC) and shell (NAcS) and the medial and lateral caudate putamen (mCPU and lCPU; medial to lateral coordinates ±1.0 to ±2.0 and ±2.0 to +/-4.0 respectively). From these tissues, RNA was isolated that was DNAse-treated and reverse transcribed to cDNA using random hexanucleotide primers. Gene expression was then analyzed by real-time quantitative PCR (ABI Prism® SDS 7900, Applied Biosystems Inc., Foster City, CA, USA) using a SYBR Green approach (300 nM gene specific primers and the cDNA equivalent of 40 ng RNA in a total volume of 10 µl) and normalized to the geometric mean of 3 housekeeping genes (- actin, Hprt and Nse). Normalized gene expression levels (GEnorm) were calculated from detection cycle threshold values (Ct) as follows: [GEnorm = – (Ctgenex – CtHKgenes)]. The fold difference in gene expression level between high impulsive (HI) and low impulsive (LI) groups was calculated as follows: [2(GEnorm HI – GEnorm LI)]. In addition to the aforementioned dopamine related genes (dopamine receptors D1 (Drd1), D5 (Drd5), D4 (Drd4), Dbh, Slc6a3 and Caly) also the expression of dopamine receptor D2 (Drd2s and Drd2l; the short and long variant), dopamine receptor D3 (Drd3) and the catechol-o-methyl- transferase (Comt) gene were measured. Primers were selected for a low probability of folding into loop structures, a low probability of forming primer dimers and were blasted against the NCBI rat ESTs database to check for specificity. We determined the melting curves for all primer sets used, in order to check for possible primer-dimer formation. All primers yielded a specific product of the correct melting temperature. The amplification efficiency of all primers was checked and found to be between 1.9–2.0. Nucleotide sequences of the primers are provided in Supplementary Table 1.

Experiment 2: Infusion of a dopamine D1/D5 receptor agonist or antagonist into the mPFC Surgery: Before surgery, rats were trained to perform the operant delayed reward paradigm at delay periods ranging from 0 to 40 s (0, 5, 10, 20, 40) for 17 sessions. Five days before surgery operant training was stopped and food was available ad libitum. Rats were anaesthetized using a combination of xylazine

26 mPFC D1/D5 receptors and impulsive choice

(Rompun; Bayer AG, Leverkusen, Germany; 7 mg/kg, i.p.) and ketamine (Alfasan; Woerden, The Netherlands; 100 mg/kg, i.m.). A double guide cannula (Plastics One, Roanoke, VA, USA) was placed above the mPFC according to coordinates derived from Paxinos and Watson (Paxinos G and C Watson 1998); anteroposterior +3.2 mm from bregma, lateral +/–0.75 mm from midline. After surgery rats were housed individually and fed ad libitum for one week after which they received 12 retraining sessions. Rats were ranked on the average impulsive choice of the last 3 retraining sessions, then divided into pairs, and from each pair one rat was randomly assigned to the group that would receive bilateral infralimbic infusions and the other to the group that would receive bilateral prelimbic infusions. Drug infusions: All drugs (SCH 23390 hydrochloride (Sigma Aldrich, St. Louis, MO, USA), SKF 38393 hydrochloride (Sigma Aldrich, St. Louis, MO, USA) and ketanserin (Janssen Pharmaceutica, Beerse, Belgium) were freshly dissolved in sterile saline on the day of infusion. The drug solutions were infused through an injector that was inserted in the guide canula and protruded to the pre- or infralimbic area (3.7 or 4.9 mm below the surface of the skull). A volume of 0.5 l was delivered at a flow rate of 0.25 l/min using 10 μl Hamilton syringes driven by a syringe infusion pump (Harvard Apparatus, South Natick, MA, USA). Injectors were left in place for an additional 2 min to allow diffusion. Rats were placed back in their home cage and were placed in the operant chamber 10 min later. At least 2 training days separated infusion days. For all rats the order of infusions was the same; sham infusion, saline1, 0.5 g SCH 233900, 1 g SCH 23390, saline2, 0.05 g SKF 38393, 0.1 g SKF 38393, saline3, 0.1 g ketanserin and finally 0.1 g ketanserin followed after 10 min by 1 g SCH 23390.

Statistical analysis Differences between HI and LI rats were analyzed using one-way analysis of variance (ANOVA). Using the QVALUE package (Storey & Tibshirani, 2003) in the statistical software program R (version 2.7.2) the false discovery rate (FDR) was calculated for the list of p-values of the dopamine-related genes implicated in ADHD. Pearson correlations coefficients and probabilities were used to analyze correlations between impulsive choice and normalized gene expression levels. The effects of local drug infusions were analyzed using repeated-measures ANOVA, with two within-subjects factors (dose and delay) and one between- subject factor (area: infusion in either infra- or prelimbic area). If Mauchly’s test for sphericity of data was significant, more conservative Greenhouse Geisser corrected probability values were reported. Where appropriate, a post hoc analysis (Fisher least square difference) was used for pair-wise comparisons of doses. All statistical analyses, except calculation of the FDR, were performed using Statistical Package for Social Sciences version 14.0 (SPSS Inc, Chicago, IL, USA).

27 Chapter 2

Results

Experiment 1: Correlations between of impulsive choice and expression of dopamine-related genes

Stable individual differences in impulsive choice The individual level of impulsive choice was calculated for all 17 baseline sessions and reliability analysis on all 23 included rats indicated that this individual level was stable across baseline sessions (Cronbach’s  = 0.966). The individual level of impulsive choice between the upper quartile (low percentage choice for large reward; HI rats) and the lower quartile (high percentage choice for large reward; LI rats) is shown in Figure 1a. The average impulsive choice across all baseline sessions of HI and LI rats differed significantly (F(1,10) = 97.126, P < 0.0001, Fig. 1b).

Impulsive choice does not correlate with errors of omission or the locomotor response to novelty No significant differences between HI and LI rats were observed in distance moved during one hour in a novel box (F(1,10) = 0.415, P = 0.534) and in errors of omission on trials used to calculate the individual level of impulsivity (F(1,10) = 0.530, P = 0.483). Similarly, when calculated across all rats in the experiment, the individual level of impulsive choice did not correlate with errors of omission (Pearson r = 0.209, P = 0.338; not shown) or the distance moved (Pearson r = - 0.04, P = 0.874; Fig. 2a).

Brain region-specific correlations of impulsive choice with Drd1, Drd5 and Caly Significant differences (P < 0.05, FDR < 0.33) between HI and LI rats were detected in the mPFC for Drd1, Drd5 and Caly, and in the NAcS for Drd5 (Table 1). Correlation analysis across all 23 rats in the experiment indicated that impulsive choice correlated significantly (P < 0.05) with the expression of Drd1 (r = 0.52), Drd5 (r = 0.56) and Caly (r = 0.45) in the mPFC (Fig. 2b-2d), but not with Drd5 in the NAcS (r = 0.237, P = 0.314).

Figure 1 | Temporally stable differences in impulsive choice between rats. (a) The level of impulsive choice (100 – % choice large reward) of LI (white circles) and HI rats (black circles) across 17 baseline sessions. (b) LI and HI rats differed in their average percentage choice for large reward calculated from 17 baseline sessions (B = baseline). *** P < 0.0001.

28 mPFC D1/D5 receptors and impulsive choice

Figure 2 | Scatter plots of correlations between the level of impulsive choice (100 - % choice large reward) and (a) the level of locomotor activity as measured as distance moved and the level of expression in the mPFC of (b) Drd1, (c) Drd5 and (d) Caly. Normalized levels of gene expression are displayed as deviation from the group mean expression level (GEnorm; log2 scale). * P < 0.05. Colors of the circles represent LI (white circles), HI rats (black circles) and intermediate impulsive rats (grey circles).

Table 1 | Differences in expression level of dopamine-related genes between HI and LI rats.

mPFC Ac OFC AI Drd1 1.51 (1.26–1.8)* 1.24 (1.13–1.36) -1.03 (-1.21–0.88) 1.16 (0.88–1.53) Drd2s 1.28 (1.08–1.51) 1.23 (1.03–1.46) -1.39 (-1.68–1.15) -1.05 (-1.62–0.68) Drd2l 1.12 (0.95–1.31) 1.11 (1.02–1.21) -1.18 (-1.36–1.03) -1.54 (-1.99–1.19) Drd3 na na na na Drd4 1.13 (0.99–1.29) -1.08 (-1.49–-0.78) 1.20 (0.39–3.75) 1.24 (0.19–8.05) Drd5 2.32 (1.34–4.03)* 1.19 (0.88–1.61) -1.13 (-1.38–0.92) 1.32 (1.15–1.52) Dat1 na na na na Dbh -1.42 (-1.77–-1.13) -1.06 (-1.11–-1.02) 1.33 (0.72–2.48) 1.17 (0.92–1.48) Caly 1.23 (1.19–1.27)* -1.07 (-1.11–-1.03) -1.06 (-1.1–1.02) -1.28 (-1.34–1.22)

NAcS NAcC lCPU mCPU Drd1 1.13 (1.02–1.26) 1.03 (1.02–1.05) 1.05 (1.04–1.05) -1.03 (-1.08–0.99) Drd2s 1.12 (1.02–1.23) -1.02 (-1.05–1.00) 1.00 (0.99–1.01) 1.01 (0.95–1.08) Drd2l 1.00 (0.95–1.06) 1.05 (1.02–1.08) 1.10 (1.09–1.11) 1.01 (0.98–1.04) Drd3 -1.03 (-1.14–0.94) -1.01 (-1.11–0.92) na 1.88 (0.76–4.67) Drd4 na na na na Drd5 1.22 (1.19–1.25)* -1.36 (-1.63–1.14) 1.11 (1.03–1.20) -1.11 (-1.30–0.95) Dat1 na na na -1.20 (-1.45–1.00) Dbh 1.10 (0.89–1.37) -1.25 (-1.54–1.01) 1.06 (0.27–4.18) -1.00 (-1.14–0.88) Caly -1.12 (-1.3–0.97) 1.01 (0.99–1.04) 1.01 (0.31–3.32) -1.02 (-1.03–1.00) Differences are reported as fold difference where positive fold changes correspond to higher expression in HI rats. The respective 95% confidence interval of the fold difference is reported between brackets. In some brain regions reliable quantification of Dat1, Drd3 and Drd4 expression was not possible because more than 35 PCR cycles were required to reach the detection threshold, and no fold difference is reported (na). *Significant difference between HI and LI rats (bold).

29 Chapter 2

Experiment 2: Infusion of a dopamine D1/D5 receptor agonist and antagonist into the mPFC In all results described below, the between-subject factor ‘area’ never had a significant effect on the percentage choice for the large reward, indicating that there was no significant difference in the response to infusions between the groups receiving infralimbic and prelimbic mPFC infusions. None of the saline infusions changed the percentage choice for the large reward when compared to the next day (saline1 F(1,9) = 0.288, P = 0.604; saline2 F(1,12) = 0.762, P = 0.400; saline3 F(1,8) = 0.009, P = 0.929). Furthermore, the percentage choice for the large reward after the first, second and third saline infusion did not differ from each other (F(2,14) = 0.036, P = 0.964). Therefore, all reported drug effects below are compared to the average percentage choice for the large reward calculated over all three saline infusions. mPFC infusions of D1/D5 antagonist SCH 23390 alone and in combination with an infusion of ketanserin The effect of mPFC infusions of two doses of the D1/D5 antagonist SCH 23390, as well as the effect of SCH 23390 after a preceding infusion of the serotonin 5- HT2a/c receptor antagonist ketanserin were statistically evaluated together (Fig. 3a). These infusions affected the percentage choice for the large reward significantly (F(3,33) = 7.020; P < 0.001), an effect that was not delay-dependent (dose  delay interaction F(12,132) = 1.153; P = 0.324). Post hoc analyses revealed that the percentage choice for the large reward was significantly

Figure 3 | Agonist and antagonist infusions of D1/D5 receptors in the mPFC increased impulsive choice. (a) SCH 23390 increased impulsive choice (decreased % choice large reward) compared to saline, an effect which was not blocked by co- infusion of ketanserin. (b) SKF 38393 increased impulsive choice compared to saline. * P < 0.05, ** P < 0.01 compared to saline. Data are expressed as mean ± SEM.

30 mPFC D1/D5 receptors and impulsive choice increased after an infusion of 1 g SCH 23390 (P < 0.001). When an infusion of 1 g SCH 23390 was preceded by an infusion 0.1 g ketanserin, post hoc analyses indicated that SCH 23390 still significantly increased the percentage choice for the large reward compared to saline (P < 0.001). Moreover, this combination of SCH 23390 and ketanserin did not change the percentage choice for the large reward compared with a single infusion of 1 g SCH 23390 (P = 0.366). Infusions of SCH 23390 alone and in combination with ketanserin also significantly increased the errors of omission (Fig. 4) (F(3,33) = 3.246; P = 0.034), an effect that was delay-dependent (dose  delay interaction F(12,132) = 1.880; P = 0.042). Post hoc analyses revealed that errors of omission increased significantly after 1 g SCH 23390 when the delivery to the large reward was delayed for 20 s (P = 0.015) and 40 s (P = 0.006), an effect that was blocked by co-infusion of 0.1 g ketanserin (Fig. 4). mPFC infusions of the D1/D5 receptor agonist SKF 38393 Infusion of SKF 38393 increased the percentage choice for the large reward (Fig. 3b; F(2,22) = 4.762; P = 0.019), an effect that was not delay-dependent (dose  delay interaction F(8,88) = 1.954; P = 0.062). Post hoc analysis revealed that the percentage choice for the large reward compared to saline increased after infusion of both 0.05 g (P = 0.026) and 0.1 g (P = 0.019) SKF 38393. No effect of SKF 38393 on errors of omission was found (F(2,22) = 0.239, P = 0.789). mPFC infusions of a single dose of ketanserin There was no significant effect of infusion of 0.1 g ketanserin on the percentage choice for the large reward (F(1,9) = 1.307, P = 0.282), its interaction with delay (dose  delay interaction F(8,72) = 1.559, P = 0.153) or errors of omission (F(1,9) = 2.766; P = 0.131).

Figure 4 | Effect of drug treatment on the number of omissions. The increase in errors of omission after intra-mPFC infusion of SCH 23390 was blocked by co-infusion of ketanserin. SCH 23390 increased errors of omission (decreased % choice large reward) compared to saline, an effect which was not blocked by co-infusion of ketanserin. * P < 0.05, ** P < 0.01 compared to saline. Data are expressed as mean ± SEM.

31 Chapter 2

Discussion In this study we observed stable differences between rats in their inability to tolerate delay of reinforcement, which is a prominent neuropsychological deficit observed in ADHD (Barkley et al., 2001; Solanto et al., 2001; Sonuga-Barke, 2002; Sonuga-Barke et al., 1992). We correlated the individual level of impulsive choice in a rodent delayed reward paradigm with the expression level of six dopamine-related genes previously associated with ADHD, and observed significant positive correlations between the expression level of three genes (Drd1, Drd5 and Caly) and impulsive choice only in the mPFC, and not in the other corticostriatal brain areas investigated. In line with previous studies (Perry et al., 2005; Wilhelm & Mitchell, 2009), we found that the level of locomotor activity did not correlate with individual levels of impulsive choice. This indicates that the observed differences in impulsive choice and gene expression levels do not result from individual differences in activity. Subsequent intracranial infusions of the D1/D5 agonist SKF 38393 and antagonist SCH 23390 into the mPFC confirmed the role of D1/D5 receptor signaling in this brain area in impulsive choice. With regard to the neuroanatomical circuitry underlying impulsive choice, different regions are thought to promote the choice of immediate small rewards and delayed large rewards (Cardinal, 2006; McClure et al., 2004). A human fMRI study, demonstrated that limbic regions (including the OFC) are activated during a choice for a small immediate reward, while regions of the lateral cortex (including the dorsolateral PFC) are activated by choice irrespective of delay (McClure et al., 2004). In line with this, a lesion study in rats showed that the OFC indeed promotes the choice for a small reward (Winstanley et al., 2004). Lesions of the mPFC in rodents, encompassing the prelimbic region which is functionally homologous to the primate dorsolateral PFC (Uylings et al., 2003), were less conclusive with respect to the role of the mPFC in impulsive choice. Rather than promoting choice for the immediate small reward, lesions to the mPFC appeared to produce some form of insensitivity to the contingencies or stimuli present in the delayed reward task (Cardinal et al., 2001; Dalley et al., 2004). The current findings are consistent with the aforementioned human neuroimaging data and extend previous lesioning data, as we found that modulating the dopamine D1/D5 tone in the mPFC increased impulsive choice. The findings with both SCH 23390 and SKF 38393 suggest that endogenous tonically- and perhaps phasically-activated dopamine D1/D5 receptors in the mPFC might promote the choice of large delayed rewards. Dopamine D1-like receptors are involved in potentiating and tuning delay-period activity of prefrontal cortical neurons (Goldman-Rakic et al., 2000; Seamans & Yang, 2004) and concomitantly prevent behavioral distraction across delays between sample and retention (Robbins, 2005). The mechanisms by which mPFC D1/D5 receptor occupancy promotes the choice for delayed large rewards in the current study may possibly be explained by its role in maintaining a correct

32 mPFC D1/D5 receptors and impulsive choice representation of specific information gathered during the task. Previous studies indicated that blockade or stimulation of mPFC D1/D5 receptors by intracranial infusion of high doses of D1/D5 receptor agonists and antagonists disrupt delay period activity (Williams & Goldman-Rakic, 1995) and impair performance in delayed response tasks (Sawaguchi & Goldman-Rakic, 1991; Seamans et al., 1998; Zahrt et al., 1997). In the current study, infusion of the high doses of SKF 38393 and SCH 23390 may likewise have interfered with representation of task- relevant information, such as the length of the delay in a previous trial or the time that has passed with respect to the start of the session. It is conceivable that uncertain or altered representation of this information, for instance, uncertainty regarding the delay duration in the previous trial or altered session-wide timing, contributed to an accelerated discounting of the large reward. In contrast to high doses of D1/D5 agents, low doses of D1/D5 agents have been shown to potentiate the delay-period activity of prefrontal cortical neurons (Goldman-Rakic et al., 2000) and, moreover, to enhance attention in rats with low baseline performance (Granon et al., 2000). Two reasons may explain why low doses of SKF 38393 and SCH 2339 did not decrease impulsive choice in the current study. First, endogenous D1/D5 stimulation may have been sufficient to bridge the delay times within a trial or the required session-wide temporal discrimination. In support of this argumentation, beneficial effects D1/D5 agonists have been observed for delays in the magnitude of 12 h, and not within the range of delays relevant for the current study, i.e. in the order of minutes and seconds (Floresco & Phillips, 2001). Second, a beneficial effect of low doses of the agonist may only have been visible in rats with a high baseline impulsive choice and presumably low dopamine D1/D5 receptor occupancy. Upon further categorizing rats into high, intermediate and low impulsive groups, we did not find a significant interaction between baseline impulsivity and the effect of the low doses of SKF 38393 (data not shown). However, the number of rats in this study may have been insufficient to detect small improvements, or alternatively, the employed doses of SCH 23390 and SKF 38393 were still too high to exert beneficial effects. It should be emphasized that besides affinity for dopamine D1/D5 receptors, SCH 23390 may also act as an agonist of the serotonin 5-HT2c receptor (Millan et al., 2001). In fact, a systemic injection of the serotonin 5-HT2a/2c agonist 2,5- dimethoxy-4-iodoamphetamine (DOI) has been reported to enhance impulsive choice (Evenden & Ryan, 1999). In order to disentangle D1/D5 and 5-HT2c receptor-mediated effects, we co-infused the serotonin 5-HT2a/c receptor antagonist ketanserin with SCH 23390. Interestingly, the dose of ketanserin used was sufficient to attenuate the increase in omissions by SCH 23390, indicating that effects of SCH 23390 on omissions may be mediated by a non-specific effect at 5-HT2c receptors. More importantly, ketanserin did not alter the increase in impulsive choice by SCH 23390, indicating that the effect of SCH 23390 on

33 Chapter 2 impulsive choice is D1/D5 receptor-mediated and not due to its affinity for 5-HT2c receptors. Taken together, our observation that an endogenous D1/D5 tone in the mPFC is involved in impulsive choice is supported by the observation that dopamine release in the mPFC is enhanced during performance in the delayed reward paradigm (Winstanley et al., 2006b). The current data extend this observation and suggest that the increase in dopamine during task performance may be related to the optimization of an endogenous dopamine D1/D5 tone to optimize or maintain a representation of task-relevant information. In addition to the relationship between Drd1 and Drd5 gene expression in the mPFC and impulsive choice, we found a similar relationship between Caly gene expression and impulsive choice. Calcyon mRNA is abundantly expressed in regions of prefrontal cortex expressing Drd1 or Drd5 (Zelenin et al., 2002) and its protein product, calycon, is involved in regulation of the affinity state of D1 receptors for agonists (Lidow et al., 2001). Although gene expression differences not necessarily predict differences in protein abundance or receptor density, collectively these data suggest the existence of a pathway through D1/D5 receptors and calcyon in the mPFC that is involved in impulsive choice. In fact, our observation of a positive correlation between Caly gene expression and impulsive choice is in line with a recent study showing that spontaneous hypertensive rats show both enhanced Caly gene expression (Heijtz et al., 2007) and increased impulsive choice (Fox et al., 2008). The dopamine transporter has low expression in the frontal lobes (Ciliax et al., 1999) and catechol-o-methyl-transferase (COMT), which degrades dopamine, has been shown to play an important role in regulating dopamine levels in this brain region (Hong et al., 1998). Although COMT polymorphisms have been associated with prefrontal dopamine levels and executive functioning (for review see Tunbridge et al., 2006), several studies failed to confirm the initial report (Eisenberg et al., 1999) of an association between COMT and ADHD (for review see Waldman & Gizer, 2006). In the current study we also measured COMT gene expression in the mPFC, but failed to detect a significant difference between high and low impulsive rats (data not shown). Previous genetic studies have revealed an association of human DRD1, DRD5 and CALY with ADHD. By and large, in these studies no variation in the coding region has been identified for DRD1 and CALY (Laurin et al., 2005; Misener et al., 2004) and, in addition, no causal polymorphism in the coding region of DRD5 has been identified thus far (Hawi et al., 2003; Lowe et al., 2004). These observations suggest that polymorphisms causally related to ADHD reside in regulatory sequences of these genes, thereby altering their expression levels. In this respect, two conclusions can be drawn from the current data. First, our findings clearly demonstrate that variation in the expression levels of these three aforementioned genes is correlated with impulsive choice in rats, particularly within the mPFC. Second, pharmacological manipulations of D1/D5 receptors

34 mPFC D1/D5 receptors and impulsive choice within the mPFC increase impulsive choice, further stressing the involvement of prefrontal D1/D5 receptors in impulsive decision making. Together, these data indicate that a pathway involving D1/D5 receptors and possibly calcyon in the mPFC plays a pivotal role in impulsive choice and may constitute a valuable target for further research and possible treatment of ADHD.

Acknowledgements This work was supported by the Center for Neurogenomics and Cognitive Research, and Neuro-Bsik Mouse Phenomics grant by SenterNovem (BSIK03053).

Supplementary information

Supplementary Table 1 | Sequences of the primers used for real-time quantitative PCR. Forward Reverse Drd1 GGCTCCATCTCCAAGGACTGTA AGCTTCTCCAGTGGCTTAGCTATTC Drd2 short CAACACCAAGCGCAGCAGT TGGGAAACTCCCATTAGACTTCA Drd2 long CAACACCAAGCGCAGCAGT GCGGGCAGCATCCTTGA Drd3 GCCCTCTCCTCTTTGGTTTCA GGAACGTAGAAGGACACCACTGA Drd4 CGCCTCTGTGACACCCTCAT CACAAACCTGTCCACGCTGAT Drd5 GGGCCTTTCGATCACATGTCT AAGGAAACCTCTTCCTCACAGTCA Dat1 ATTGCATGGGCACTGCACTA ATGGGAGGTCCATGGTGAAG Dbh CCCTGCGACTCCAAGATGA CACGTGGCGACAGTAGTTGAG Caly GCTGCGATGACTGTGTCCAC TCTCTCATGTCCAAAGGCTTCC Comt CCTGGAGGCCATCGACAC TGCATCCATGATTTGGCCTT -actin TGACCCAGATCATGTTTGAGACC AGTGGTACGACCAGAGGCATACA Hprt ATGGGAGGCCATCACATTGT ATGTAATCCAGCAGGTCAGCAA Nse ACGTGGTTCCATTTCAAGATGAC CGAGCTTCAGTTAGTGCACCAA

35

36

Chapter 3

Inhibitory control and response latency differences between C57BL/6J and DBA/2J mice in a Go/No-Go and 5-choice serial reaction time task and strain-specific responsivity to amphetamine

Maarten Loos, Jorn Staal, Anton N.M. Schoffelmeer, August B. Smit, Sabine Spijker, Tommy Pattij

Behavioral Brain Research, 2010;214(2):216–24

37

38 Comparison of Go/No-Go and 5-CSRTT

Abstract Among the best-replicated and most heritable endophenotypes of attention deficit/ hyperactivity disorder (ADHD) are deficits in attention, inhibitory response control and larger intra-individual variability in response latencies. Here, we explored the presence of these heritable ADHD endophenotypes in two commonly used inbred mouse strains, C57BL/6J and DBA/2J, and investigated whether treatment with the stimulant amphetamine affected these phenotypes. Both in an operant Go/No-Go task and 5-choice serial reaction time (5-CSRT) task, DBA/2J mice showed reduced inhibitory response control compared with C57BL/6J mice. Mean correct response latencies of DBA/2J mice were slower in both tasks. Analysis of the distribution of correct response latencies suggested similar processing speed, but DBA/2J mice displayed larger intra-individual variability. Amphetamine did not affect inhibition in the Go/No-Go task but increased omission errors. In contrast, in the 5-CSRTT, amphetamine did not affect omission errors but impaired inhibitory response control, specifically in C57BL/6J mice. The dopamine uptake inhibitor, GBR 12909, mimicked this effect and decreased accurate choice, specifically in C57BL/6J mice, indicating that dopamine modulates inhibitory response control and attention in the murine 5-CSRTT. Amphetamine did not affect response distributions in either task. Furthermore, we extended previous reports on differences in the brain dopamine system of DBA/2J and C57BL/6J mice, by showing differential gene expression levels of three dopamine receptors (Drd1, Drd4 and Drd5) in the mPFC. In conclusion, genetic differences between DBA/2J and C57BL/6J mice translate into multiple ADHD-related phenotypes, indicating that these strains are valuable resources to understand genetic mechanisms underlying ADHD-relevant phenotypes.

Introduction Attention-deficit/ hyperactivity disorder (ADHD) is a highly heritable psychiatric disorder (Faraone & Biederman, 1998) with a complex genetic architecture. Current theories suggest that multiple neuropsychological deficits can lead to ADHD, such as mild executive dysfunction and/or increased aversion to delay of reinforcement (Castellanos et al., 2006; Solanto et al., 2001). In order to facilitate studies of this complex genetic architecture and neuropsychological heterogeneity, the use of endophenotypes has been proposed (Gottesman & Gould, 2003) that may be more proximal to the genes involved in ADHD. Among the best-replicated (Castellanos & Tannock, 2002) and most promising ADHD endophenotypes for genetic studies (Bidwell et al., 2007) are various executive functions, such as attentional processing and inhibitory response control, and possibly measures of inattention such as intra-individual variability in response latencies. Stimulant treatment (i.e. methylphenidate and amphetamine) improves attentional processing and inhibitory response control,

39 Chapter 3 and decreases intra-individual variability in response speed in ADHD patients (Pietrzak et al., 2006; Spencer et al., 2009), underscoring the clinical relevance of these endophenotypes. Furthermore, the beneficial effect of stimulant treatment is in line with the hypothesis of a dopaminergic dysfunction in ADHD. Heritable endophenotypes play a pivotal role in the translation of complex human disorders into genetically valid animal models of the disorder, and vice versa. Among other genetic animal models (Sagvolden et al., 2005), the comparison of C57BL/6J versus DBA/2J mouse strains seems promising in this respect. These commonly used inbred strains of mice show differences in sustained visual attention and inhibitory response control (Loos et al., 2009; Patel et al., 2006; Pattij et al., 2007; Young et al., 2009a) and increased aversion to delay of reinforcement (Helms et al., 2006). Furthermore, mice of these strains show profound differences in their brain dopaminergic system (Cabib et al., 2002; D'Este et al., 2007) that may be relevant for research on dopaminergic dysfunction in ADHD. In further support of selecting these two inbred strains, the existing panel of recombinant inbred strains derived from crossbreeding C57BL/6J  DBA/2J mice (BXD; Peirce et al., 2004) potentially provides the opportunity to genetically dissect ADHD-relevant phenotypes and identify genetic loci and/or genes that contribute to these phenotypes. In the present study, we investigated putative ADHD-related phenotypes in C57BL/6J and DBA/2J mice, namely inhibitory response control, visuospatial attention, and intra-individual variability in response latencies. These measures were assessed in an operant Go/No-Go and 5-choice serial reaction time (5- CSRT) task. Although both tasks measure inhibitory control, the inhibition of commission errors in the Go/No-Go task and the inhibition of premature responses in the 5-CSRTT may not require identical control mechanisms (Eagle & Baunez, 2010). Therefore these tasks might act complementary to describe putative differences in inhibitory control between C57BL/6J and DBA/2J mice. The interest in intra-individual variability in response latencies in human studies has grown over the last decade (e.g. Bidwell et al., 2007; Castellanos et al., 2006; Castellanos & Tannock, 2002; Klein et al., 2006; Spencer et al., 2009) following the observation of increased intra-individual variability in ADHD (Leth-Steensen et al., 2000; Sergeant & van der Meere, 1990). Whereas the peak of a response latency distribution could be viewed as an index of processing speed, larger intra- individual variability has been interpreted as lapses in attention (Leth-Steensen et al., 2000; Sabol et al., 2003). In the present study, we employed the number of correct response latencies in both the Go/No-Go tasks and 5-CSRTTs to analyze response distributions and to investigate potential strain differences in processing speed and lapses in attention. Stimulant treatment modulates inhibitory control (Eagle et al., 2008; Pietrzak et al., 2006), primarily via dopaminergic mechanisms (Pattij & Vanderschuren, 2008), and decreases intra-individual variability in response latencies in humans and rats (Pietrzak et al., 2006; Sabol et al., 2003; Spencer et al., 2009). To

40 Comparison of Go/No-Go and 5-CSRTT investigate dopaminergic modulation of these cognitive processes and possible translational validity, we here investigated the effects of amphetamine in the murine Go/No-Go and 5-CSRTT. Because of the aforementioned differences in the brain monoamine systems between C57BL/6J and DBA/2J mice, we anticipated that these strains could respond in a strain-specific manner. Finally, several genetic association studies implicated dopamine receptors in the etiology of ADHD (Faraone et al., 2001; Lowe et al., 2004; Misener et al., 2004) and interfering with these receptors in the medial prefrontal cortex (mPFC) is known to affect various executive functions, such as attention, correct response time and working memory (Granon et al., 2000; Sawaguchi & Goldman-Rakic, 1991). Therefore, to explore putative differences between C57BL/6J and DBA/2J in dopamine receptor signaling that might underlie observed behavioral differences, we compared gene expression levels of dopamine receptors in the medial prefrontal cortex (mPFC).

Materials and Methods

Subjects and housing Male seven-week-old mice were obtained from Charles River (Go/No-Go task: n = 12 C57BL/6J and n = 11 DBA/2J mice; gene expression measurement: n = 16 C57BL/6J and n = 16 DBA/2J mice) or our own breeding colony at Harlan Germany (5-CSRT: n = 16 C57BL/6J and n = 10 DBA/2J mice). Mice were singly housed on sawdust in standard Makrolon type II cages (26.5 cm long, 20.5 cm wide and 14.5 cm high) enriched with cardboard nesting material. After a habituation period of minimally 1 week, body weights were recorded. Mice subjected to behavioral testing were food-restricted to gradually decrease their body weight to 90% of the initial body weight the following 7 days. Mice were placed in the operant cages for 5 days each week for testing.

Operant chambers Operant chambers (MEDNPW-5M, Med Associates Inc., St Albans, VT, USA) were equipped with five response holes, a food magazine with a reward dispenser (12 mg pellet, Formula P; Research Diets Inc., New Brunswick, NJ, USA) at the opposite wall, a house light, a 4500 Hz tone generator (Mallory Sonalert Products Inc., Indianapolis, IN, USA), and placed in sound-attenuating ventilated cubicles. Both response holes and the food magazine contained yellow LED stimulus lights and infrared response detectors.

Drug injections (+)–Amphetamine sulfate (O.P.G., Utrecht, The Netherlands) and GBR 12909 (Sigma, St. Louis, MO, USA) were dissolved in sterile saline (0.9 g/l NaCl) and aliquots were stored at –20ºC. Drugs were injected intraperitoneally in a volume

41 Chapter 3 of 10 ml/kg body weight, 10 min before testing. Different doses of the same drug (Go/No-Go task: amphetamine 0.5 and 1.0 mg/kg; 5-CSRTT: amphetamine 0.25, 0.5 and 1.0 mg/kg and GBR12909 2.5, 5 and 10 mg/kg) and vehicle were administered in a Latin square design on Tuesday and Fridays, resulting in at least two wash out days in between each dose. The dose ranges were chosen based on the reported effectiveness in operant procedures taxing impulsivity (Isles et al., 2003; van Gaalen et al., 2006a).

Go/No-Go task In this paradigm, the cue light in the middle of the five response holes (i.e. hole 3) was designated as the “start stimulus” and the cue light in the response hole immediately to the left or right (i.e. hole 2 or 4, counterbalanced across mice and strains) was designated as the “Go stimulus”. All sessions ended if a subject completed 100 trials or after 30 min, whichever came first. Shaping of Go-response step 1. In the first session, both the start stimulus and Go stimulus were switched on, and after a response into one of these two holes a reward was delivered and both cue lights were turned off. Subsequently, following an inter-trial interval (ITI) of 10 s both the start stimulus and Go stimulus were switched on. After 7 sessions all animals completed more than 40 trials in 30 min. Shaping of Go-response step 2. In the next 5 sessions, the start of a trial was signaled by the illumination of the start stimulus. A response in this hole switched off the start stimulus and was immediately followed by the presentation of the Go stimulus. A response into the Go stimulus hole resulted in the delivery of a reward, turned off the stimulus light and after an ITI of 10 s the next trial started. Premature Go-responses in an inactive Go stimulus hole resulted in a 5 s time out (TO) during which both house light and stimulus lights were turned off. Shaping of Go-response step 3. In the subsequent 9 sessions, responding at a VR3 schedule in the illuminated stimulus-hole was required to initiate the Go stimulus. The latency between the onset of the Go stimulus and a response into the Go stimulus hole was recorded for each trial and reflected the Go-response latency (GoRT). Shaping of Go-response step 4. In the next 10 sessions, the Go stimulus was only switched on for the duration of an individually-titrated Limited Hold period (LH). A Go-response during the LH time resulted in the delivery of a reward. An omission of a Go-response during the LH time was followed by a 5 s time out, during which both house light and stimulus lights were turned off. In the first five sessions, the GoRTs of the previous session for each mouse were ranked, and the LH time was set to the 90th percentile of the GoRTs of the previous session. In the last 5 sessions, the LH times were manually titrated for each mouse individually such that in 30% of Go trials the mouse did not have a chance to respond within the LH time (30% of Go omissions).

42 Comparison of Go/No-Go and 5-CSRTT

Go/No-Go task. In the final Go/No-Go paradigm, an auditory No-Go signal (4500 Hz) was presented randomly in 20% of the Go trials. The No-Go-signal was switched on and off simultaneous with the illumination of the cue light in the Go-stimulus hole for the duration of the LH period. A response (Go) during a No-Go trial resulted in a TO, whereas inhibiting a response during these trials resulted in the delivery of a reward. For two sessions, the LH during No-Go trials was set to the individual mean GoRT to facilitate the acquisition of inhibition. Subsequently, during the first until the third Go/No-Go session, the Go and No- Go LH were both set to the individual titrated level as assessed during the last session of shaping step 4. Several mice of the DBA/2J strain showed difficulties in learning to inhibit responding during No-Go trials. Therefore, all DBA/2J mice received 3 additional Go/No-Go training sessions. Subsequently, 7 badly performing DBA/2J mice (Pinhibition < 0.1; see below) received 2 additional Go/No-Go training sessions during which the auditory No-Go signal was presented randomly in 33 and 50% of Go-trials. Mice were trained for 15 standard Go/No-Go sessions, during which their individual LH was again titrated such that the number of omissions during Go trials was 30% to force mice to respond as quickly as possible. The average performance over the two following sessions was taken as baseline performance. In this task, inhibitory control (Pinhibition) was calculated, similar to previous studies [3, 32], from PNo-Go, (number of omitted No-Go trials / number No-Go trials) by taking Pomission (number of omitted Go trials / number of Go trials) into account: Pinhibition = [(PNo-Go – Pomission) / (1 – Pomission)]. When the percentage of omissions during Go-trials was 100 % (Pomission = 1; observed only after drug infusions), Pinhibition was set to 1.

5-CSRTT Mice were trained on a individually-paced schedule to perform the 5-CSRTT, as described previously (Loos et al., 2009). Mice were habituated to the operant cages for one session of 20 min, during which the house light was switched off, and the stimulus lights in the magazine and stimulus holes turned off. In the subsequent 4 sessions, rewards were distributed into the magazine at variable intervals (ITI; 4, 8, 16 or 32 s), which coincided with switching on the magazine stimulus light. An ITI was only initiated when the previous reward had been collected as indicated by a magazine response, after which the magazine stimulus light was off. A session ended after 25 min or sooner when the criterion of 50 rewards was reached. In the next sessions, a trial started by the illumination of all five stimulus holes. An instrumental response into any of these holes switched off the light in all five stimulus holes, switched on the stimulus light in the magazine and delivered reward into the magazine. Sessions lasted for 25 min or 60 earned rewards. As soon as mice earned 50 or more rewards in two sessions they commenced the next training phase, or after 17 sessions. Subsequent trials started by the illumination of only one stimulus hole. Responses into the non- illuminated holes were without consequence. As soon as mice earned 50 or more

43 Chapter 3 rewards in two sessions they commenced to the next training phase, or after 10 sessions. Finally, mice reached 5-CSRTT training. A trial started with a response of the subject into the illuminated magazine, which switched off magazine light and after an ITI of 5 s a stimulus in one of the five stimulus holes was presented for a limited duration (stimulus duration). A response in the correct stimulus hole within the LH of 4 s after termination of the stimulus switched on the magazine light and delivered a food pellet. Both an incorrect response into a non- illuminated stimulus hole and an omission of a correct response resulted in a time-out period, during which all stimulus lights and the house light were turned off. When the time-out period ended, both the house light and the magazine light were switched on, and the subject could start the next trial. A premature response into a non-illuminated stimulus hole during the delay period also resulted in a time-out period, but a subsequent response into the illuminated magazine restarted the same trial. The percentage of omissions was defined as [100  (omissions) / (omissions + number of correct and incorrect responses)]. Accurate choice or accuracy, the commonly used index of visuospatial attention in the 5- CSRTT, was defined as [100  (number of correct responses) / (number of correct and incorrect responses)]. Inhibitory response control in this task, in terms of the percentage of premature responses, was defined as [100  (number of premature responses) / (number of omissions + correct + incorrect responses)]. In the first 5-CSRTT session the stimulus duration was set at 16 s, which was decreased in subsequent sessions to 8, 4, 2, 1.5 and 1 s if the subject reached criterion performance (omissions < 30%, accuracy > 60%, started trials > 50) or after 10 sessions. Baseline 5-CSRTT performance was calculated from the 6th until 10th session at SD of one second.

Intra-individual variability in response latencies in Go/No-Go and 5-CSRTT Several methods have been proposed to derive indices of response speed and intra-individual variability from the distribution of response latencies (Leth- Steensen et al., 2000; Sabol et al., 2003) that converge in the assumption that the peak of the distribution measures sensory-motor processing speed while the tail (positive skew) of the distribution measures intra-individual variability, also known as lapses in attention (Leth-Steensen et al., 2000). Both for Go-responses in the Go/NoGo task and correct responses in the 5-CSRTT we used the mode of the response latencies as a measure of sensory-motor processing speed. The mode was calculated according to the half rang method (HRM; Hedges & Shah, 2003). HRM uses the densest half of samples in an iterative fashion to estimate the mode. The difference between the mode and the mean (devmode) of the response latencies was used as a measure of the intra-individual variability.

Gene expression measurements An independent set of mice (4 pools of mice, n = 8 per pool) that had not been subjected to food restriction was used for real time quantitative PCR

44 Comparison of Go/No-Go and 5-CSRTT measurements (RT-qPCR). The mPFC was freshly dissected on ice and frozen (– 80° C) until further use. RNA was isolated (Loos et al., 2009), DNAse-treated and reverse transcribed to cDNA using random hexanucleotide primers. Gene expression levels of dopamine receptor 1 through 5 (Drd1 – Drd5) were analyzed (ABI Prism® SDS 7900, Applied Biosystems Inc., Foster City, CA, USA) using a SYBR Green approach (cDNA equivalent of 40 ng RNA in a total volume of 10 µl) with 300 nM gene specific primers. Expression levels were normalized to the geometric mean of 2 housekeeping genes (-actin and HPRT).

Statistical analysis Differences between strains were analyzed using one-way analysis of variance (ANOVA), the effects of treatment or session were evaluated using repeated measures ANOVA with strain as between-subjects variable (in case of drug administration) and treatment as within-subjects variable. When Mauchly’s test for sphericity of data was significant, more conservative Huynh–Feldt corrected degrees of freedom and subsequent probability values were reported. In case of a statistically significant effect of treatment, paired-sample t-tests were used to assess task manipulation or dose effects within strains. In case of statistically significant strain  treatment interaction effects, independent sample t-tests were used to evaluate the differences between strains. All data are depicted as means +/- standard error of the mean (SEM), and the level of significance was set at P < 0.05. All analyses were performed using the Statistical Package for the Social Sciences for Windows version 15.0 (SPSS, Chicago, IL, USA).

Results

Go/No-Go task Acquisition and baseline responding. The limited hold period was individually titrated such that each mouse omitted around 30% of the Go trials (shaping step 4). By the 10th session, the limited hold was titrated such that the number of omissions of Go stimuli did not differ between strains (F(1,21) = 1.56, non- significant (n.s.)). DBA/2J mice required longer limited hold times (F(1,21) = 23.29, P <0.001) compared with C57BL/6J mice. After introduction of the No- Go signal, from the 1st until the 3rd Go/No-Go session, DBA/2J mice consistently showed a lower Pinhibition than C57BL/6J mice (strain: F(1,21) = 36.49, P < 0.001) without an interaction with session number (session  strain: F(2,42) = 0.043, n.s.). By the third session (Fig. 1a), 92% of the C57BL/6J mice (11 out of 12) showed inhibition of Go-responses during No-Go trials (Pinhibition > 0) whereas only 36% (7 out of 11) of the DBA/2J mice did. Even when only mice with Pinhibition > 0 were evaluated, C57BL/6J mice showed a higher Pinhibition than DBA/2J mice (strain: F(1,13) = 4.85, P < 0.05) (Fig. 1a).

45 Chapter 3

After prolonged training in the Go/No-Go task, all C57BL/6J mice (12 out 12) and ten DBA/2J mice (10 out of 11) consistently showed a higher percentage of inhibition during No-Go trials than omissions during Go trials (i.e. Pinhibition > 0). Similar to the first three sessions, DBA/2J mice displayed a significant lower Pinhibition than C57BL/6J mice (F(1,21) = 5.00, P < 0.05). Because the lack of inhibition during No-Go trials can be the result of poor inhibitory control, or failure to properly acquire the Go/No-Go task, only mice with Pinhibition > 0 were considered for analyses of baseline performance and the effects of amphetamine. Even under these conditions, a trend towards a lower Pinhibition in DBA/2J mice was observed compared with C57BL/6J mice (F(1,20) = 4.06, P = 0.057) in the remaining mice (Fig. 1a). During the prolonged training phase the LH times were titrated for each individual mouse to reach a 30% omission rate during Go-trials. Indeed, strains did not differ in their omission rate of Go trials (F(1,20) = 2.60, n.s.), but significantly differed in the required LH to achieve this level of performance (F(1,20) = 41.91, P < 0.001). Furthermore, C57BL/6J mice and DBA/2J mice differed significantly in their mean GoRT (F(1,20) = 6.40, P < 0.05) and LH (F(1,20) = 41.90, P < 0.001).

Figure 1 | Baseline Go/No-Go task performance. (a) During the third Go/No-Go session (session 3), DBA/2J mice showed impaired inhibitory response control compared with C57BL/6J mice. Only mice were included that showed a Pinhibition > 0, as indicated by numbers above the bars. After prolonged training (Baseline), most mice successfully acquired the task and a trend toward lower inhibitory response control in DBA/2J mice was observed. (b) Distribution of GoRTs for each strain (bin size 0.1 s). The frequency was calculated for each mouse, by dividing the number of correct responses within a time bin by the total number of correct responses of the respective mouse. (c) Representation of the mean GoRT, the mode of the Go-response distribution and the intra-individual variability in Go- response latencies (i.e. devmode). * P < 0.05.

46 Comparison of Go/No-Go and 5-CSRTT

DBA/2J mice showed a significantly larger devmode (F(1,20) = 8.92, P < 0.01) with no significant difference in mode (F(1,20) = 0.50, n.s.) (Fig. 1b-c), suggesting that the difference in mean GoRT was likely due to a larger intra-individual variability in GoRTs in DBA/2J mice. The two strains did not differ in number of premature Go- responses (F(1,20) = 0.40, n.s.).

Effects of amphetamine on Go/No-Go performance. At the highest dose, one C57BL/6J mouse omitted to respond during all Go trials and was therefore excluded from further analyses. Amphetamine (Fig. 2) did not significantly affect Pinhibition (dose: F(2,40) = 1.15, ns; dose  strain: F(2,40) = 0.17, ns; strain: F(1,20) = 2.27, n.s.) (Fig. 2a), number of premature Go- responses (dose: F(2,40) = 1.41, ns; dose  strain: F(2,40) = 0.91; strain: F(1,20) = 0.60, n.s.), mean GoRT (dose: F(2,38) = 0.47, ns; dose  strain: F(2,38) = 1.47, ns; strain: F(1,19) = 18.05, P < 0.001) (Fig. 2b), mode of GoRTs (dose: F(1.70,33.89) = 0.18, ns; dose  strain: F(1.70,33.89) = 1.75, ns; strain: F(1,20) = 5.18, P < 0.05) or intra-individual deviation from the mode (dose: F(2,38) = 0.83, ns; dose  strain: F(2,38) = 0.72, ns; strain: F(1,19) = 13.26, P < 0.01). In contrast, amphetamine significantly increased the omission rate during Go trials (Fig. 2c; dose: F(2,40) = 40.28, P < 0.001) in both strains (dose  strain interaction: F(2,40) = 1.77, ns; strain: F(1,20) = 4.71, P < 0.05) and post hoc testing revealed a significant higher percentage of omissions after 0.5 mg/kg (P < 0.01) and 1.0 mg/kg (P < 0.001). Figure 2 | Effects of amphetamine on Go/No-Go task performance. No effects of different doses of amphetamine (0.5 and 1 mg/kg) were observed on (a) inhibitory response control and (b) mean GoRTs between strains. (c) Amphetamine significantly increased omissions of Go-responses. ** P < 0.01, *** P < 0.001, V = Vehicle.

47 Chapter 3

5-CSRT Acquisition and baseline responding. All mice completed the standard (Loos et al., 2009) shaping phase including the first five sessions at a baseline SD of 1 s in approximately 40 training days with no difference between strains (strain: F(1,24) = 1.94, n.s.). In the subsequent 5 sessions, significant strain differences were observed in the percentage of premature responses (Fig. 3a; strain: F(1,24) = 5.08, P < 0.05) and the mean correct response time (strain: F(1,24) = 9.59, P < 0.01). Analysis of response latencies (Fig. 3b-c) suggested that the slower correct response latencies of DBA/2J mice were the result of a different mode of the response latencies (F(1,24) = 7.47, P < 0.05) and not of intra-individual variability in the deviation of the mode of response latencies (F(1,24) = 0.16, n.s.). No strain differences were observed in the number of omissions (strain: F(1,24) = 0.18, n.s.) and accurate choice (strain: F(1,24) = 0.14, n.s.).

Effects of amphetamine on 5-CSRTT performance. Amphetamine (Fig. 4) increased the percentage of premature responses (dose: F(2.74,60.20) = 3.95, P < 0.05) in a strain-dependent fashion (dose  strain: F(2.74,60.20) = 3.67, P < 0.05; strain: F(1,22) = 1.27, n.s.). Post hoc testing showed no significant effect in

Figure 3 | Baseline 5-CSRTT performance. (a) DBA/2J mice made more premature responses than C57BL/6J mice. (b) Distribution of GoRTs for each strain (bin size 0.1 s). The frequency was calculated for each mouse, by dividing the number of correct responses within a time bin by the total number of correct responses of the respective mouse. (c) Representation of the mean correct response time, the mode of the correct response distribution and the intra-individual variability in GoRTs (i.e. devmode). * P < 0.05.

48 Comparison of Go/No-Go and 5-CSRTT

DBA/2J mice (dose: F(1.60,11.18) = 2.68, n.s.), but only in C57BL/6J mice amphetamine increased the percentage of premature responses (dose: F(1.31,19.66) = 6.20, P < 0.05) at the highest dose compared with saline (P < 0.05). The apparent, but non-significant effect of 0.5 mg/kg amphetamine on the percentage of premature responses in DBA/2J mice was due to one individual with 48.7 % premature responses, which did not show any outlying behavior on other parameters during that same session. Amphetamine did not significantly affect mean correct response time (dose: F(1.24,27.18) = 0.30, ns; dose  strain: F(1.24,27.18) = 0.39, ns; strain: F(1,22) = 0.42, n.s.), mode (dose: F(1.53,33.65) = 0.45, ns; dose  strain: F(1.53,33.65) = 0.68, ns; strain: F(1,22) = 0.28, n.s.), intra-individual variability in the deviation of the mode of the response latencies (dose: F(3,66) = 0.53, ns; dose  strain: F(3,66) = 0.73, ns; strain: F(1,22) = 0.05, n.s.), errors of omission (dose: F(1.43,31.51) = 2.20, ns; dose  strain: F(1.43,31.51) = 0.68, ns; strain: F(1,22) = 1.39, n.s.) or accurate choice (dose: F(3,66) = 0.13, ns; dose  strain: F(3,66) = 0.50, ns; strain: F(1,22) = 2.05, n.s.).

Effects of GBR 12909 on 5-CSRTT performance. The selective dopamine uptake inhibitor GBR 12909 (Fig. 4) significantly increased the percentage of premature responses (dose: F(1.92,44.11) = 5.26, P < 0.01), an effect which was dependent upon strain (dose  strain: F(1.92,44.11) = 4.38, P < 0.05; strain: F(1,23) = 1.99, n.s.). Post hoc testing revealed no significant effect in DBA/2J mice (dose: F(3,24) = 0.22, n.s.), whereas in C57BL/6J mice the percentage of premature responses was increased (dose: F(1.60,24.02) = 11.05, P < 0.001) both after 5 and 10 mg/kg GBR 12909 (P < 0.05 and P < 0.001, respectively). GBR 12909 did not affect the mean correct response latencies (dose: F(2.67,61.23) = 0.58, ns; dose  strain: F(2.67,61.23) = 1.76, ns; strain: F(1,23) = 14.75, P < 0.001), mode (dose: F(3,69) = 1.41, ns; dose  strain: F(3,69) = 0.17, ns; strain: F(1,23) = 6.02, P < 0.05), intra-individual deviation from the mode (dose: F(2.18,50.14) = 1.56, ns; dose  strain: F(2.18,50.14) = 0.52, ns; strain: F(1,23) = 3.15, n.s.) and omissions (dose: F(3,69) = 1.15, ns; dose  strain: F(3,96) = 1.07, ns; strain: F(1,23) = 4.81, P < 0.05). With respect to accurate choice (Fig. 4f), no significant main effect of GBR 12909 was observed (dose: F(3,69) = 0.88, n.s.), but an interaction with strain was observed (dose  strain: F(3,69) = 3.07, P <0.05; strain: F(1,23) = 2.91, n.s.). Post hoc testing revealed a significant decrease in accurate choice in C57BL/6J mice at the highest dose of GBR 12909 compared with saline (P < 0.05). A more detailed analysis revealed that the number of incorrect responses increased significantly (P < 0.05) from 2.1 ± 0.6 under saline to 4.6 ± 0.8 after 10 mg/kg GBR 12909, whereas the number of correct responses decreased significantly (P < 0.05) from 30.1 ± 1.5 to 26.8 ± 1.4.

49 Chapter 3

Figure 4 | Effects of amphetamine and GBR 12909 on 5-CSRTT performance. Amphetamine and GBR 12909 (a, b) increased the percentage of premature (impulsive) responses, but did not affect (b, e) the mean correct response time or (e, f) the accuracy. * P < 0.05, ** P < 0.01, V = Vehicle.

50 Comparison of Go/No-Go and 5-CSRTT mPFC gene expression levels of dopamine receptors DBA/2J mice showed a significantly higher gene expression level of dopamine D1, D4 and D5 receptors in the mPFC compared with C57BL/6J mice (Fig. 5; Drd1: 1.37 fold, F(1,7) = 8.71, P < 0.05; Drd4: 2.09 fold, F(1,7) = 17.45, P < 0.01; Drd5: 1.31 fold, F(1,7) = 8.96, P < 0.05). In addition, dopamine receptor D2 and D3 could not reliably be quantified due to low gene expression levels; more than 35 PCR cycles were required to reach the detection threshold.

Discussion The present study demonstrates that two commonly used inbred strains of mice differ in inhibitory control, response speed and response variability, which are among the best replicated endophenotypes of ADHD (Castellanos & Tannock, 2002). In line with the earlier suggestion that both DBA/2J mice and C57BL/6J mice harbor ADHD-related phenotypes (Cabib et al., 2002), we observed strain differences in inhibitory control, response latencies and the tail of the response time distribution as measured in two different cognitive paradigms. Furthermore, we showed that inhibitory control can be modulated by amphetamine and GBR12909 in a strain-specific manner. Finally, in addition to previously reported differences in the brain dopaminergic system between these two strains (Cabib et al., 2002; D'Este et al., 2007), we observed a higher gene expression level of Drd1, Drd4 and Drd5 in the mPFC of DBA/2J compared with C57BL/6J mice.

ADHD-related phenotypes in C57BL/6J and DBA/2J mice One of the principle reasons to select C57BL/6J and DBA/2J strains of mice in the present study is that these strains are the founders of the BXD recombinant inbred strain of mice (Peirce et al., 2004). Employing this panel of BXD strains in search for ADHD-related phenotypes, future studies may be able to identify genetic loci and/or genes that contribute to these phenotypes or specific BXD strains that may actually capture multiple ADHD-related phenotypes within one strain. Thus, in view of this, observing strain differences in various ADHD

Figure 5 | Gene expression levels for Drd1, Drd4 and Drd5 in DAB/2J and C57BL/6J mice. DBA/2J mice have a significantly higher transcript levels of Drd1, Drd4 and Drd5 in the mPFC compared with C57BL/6J mice, as measured by RT-qPCR. * P < 0.05.

51 Chapter 3 related phenotypes in these founder strains would be a first indicator of the fruitfulness of this approach.

Inhibitory response control – Both in the Go/No-Go and 5-CSRTT DBA/2J mice showed poorer inhibitory response control capacities. DBA/2J mice had a reduced ability to inhibit responding during No-Go trials when an auditory No- Go signal was presented, although this difference was not maintained after prolonged baseline training (Fig. 1). This reduced ability to inhibit responding was probably not due to the reported hearing difficulties of DBA/2J mice (Erway et al., 1993), as a pilot study indicated that the auditory signal used in the current study was sufficient to elicit conditioned approach behavior to a food predictive stimulus in this strain (unpublished observations). In the 5-CSRT, reduced inhibitory control of DBA/2J mice was apparent in terms of the ability to withhold premature responses to a food predictive stimulus. Whereas, to our knowledge, this is the first study to compare C57BL/6J and DBA/2J mice in the Go/No-Go task, several previous studies have compared these strains in the 5- CSRTT. In line with these studies and despite differences in equipment and protocols (Loos et al., 2009; Patel et al., 2006; Pattij et al., 2007), DBA/2J mice showed poorer inhibitory control than C57BL/6J mice, indicating that this ADHD-related phenotype is robustly differentiating these strains.

Visuospatial attention – With respect to visuospatial attention that was only assessed in the 5-CSRTT in terms of accuracy of responses both strains showed a similar behavioral performance. In contrast, one previous study reported better performance of DBA/2J mice (Pattij et al., 2007), whereas other studies (Greco et al., 2005; Patel et al., 2006) reported that C57BL/6 mice outperformed DBA/2 mice. It is possible that the short LH of 2 s we employed previously (Pattij et al., 2007) partly explains this discrepancy, as the study by Patel and colleagues (2006) and the current study have used a longer LH period (5 s and 4 s, respectively). In support of this, mean incorrect response latencies of both C57BL/6J (2.0 s) and DBA/2J (2.4 s) mice were much longer than their correct response latencies, and the strain difference for incorrect response latencies nearly reached significance (P = 0.056). Hence, the shorter LH in our previous work (Pattij et al., 2007) might have lowered the number of incorrect responses and increased omission rate in particular in DBA/2J mice, thereby resulting in higher accurate choice in this strain.

Correct response latencies and intra-individual variability – Although DBA/2J mice had lower mean correct response latencies in both the Go/No-Go and 5- CSRTT, analyses of the response time distributions indicated task-specific differences. In the Go/No-Go task, the mode of the response latencies (peak of response distribution) task did not differ, suggesting that both strains have similar sensory- and motor-processing times (Leth-Steensen et al., 2000; Sabol et al.,

52 Comparison of Go/No-Go and 5-CSRTT

2003). However, the intra-individual variability in correct response latencies (devmode) of DBA/2J mice was significantly larger, suggesting that this strain might have had more lapses in attention towards the Go stimulus. Upon individual analysis, it appeared that this delayed response peak was present in the histogram of the majority of DBA/2J mice and not resulting from a few mice with slower modes. Together, the observations in the Go/No-Go test suggest that sensorimotor processing speed in DBA/2J mice is not different from C57BL/6J mice, but the former strain might display lapses in attention, an important endophenotype of ADHD. In contrast to the Go/No-Go task, the slower mean response time of DBA/2J mice in the 5-CSRTT appeared to be due to a slower mode of the response latencies in this strain instead of a slower devmode. One critical difference between the Go/No-Go and 5-CSRTT paradigm is the temporal predictability of the presentation of a cue light in the 5-CSRTT paradigm that we employed. Whereas in the Go/No-Go task the Go stimulus was presented randomly after one, two or three responses (variable ratio 2 schedule) in the start-stimulus hole in order to prevent prepotent response patterns, in the 5-CSRTT the visual stimulus onset always occurred 5 s following a magazine response to start a trial. Hence, the significantly slower mode of response latencies of DBA/2J mice in the 5-CSRTT is probably not explained by sensorimotor processing speed, but rather by the reduced ability of DBA/2J mice to use this temporal information to focus their attention to the visual stimulus 5 s later. This notion is further supported by previous studies (Patel et al., 2006; Pattij et al., 2007; Young et al., 2009a) in which mice were either not required to start a trial or the inter-trial interval was unpredictable such that neither of the strains could effectively use a timing strategy. Indeed, in these studies correct response latencies did not differ between C57BL/6J and DBA/2J mice (Patel et al., 2006; Pattij et al., 2007; Young et al., 2009a). Unfortunately, to date, there is no data available on the timing ability of DBA/2J mice to further support this idea. In fact, a study on strain differences measuring timing ability might be worthwhile, as disturbances in timing ability have been linked to ADHD (Toplak et al., 2006).

Pharmacological effects on Go/No-Go and 5-CSRTT performance The availability of transgenic mouse lines has led to the development of a vast number of operant tasks to assess cognitive function in mice over the last decade. Nonetheless, the pharmacological evaluation of most of these tasks is lacking. For example, to date, there is no study describing pharmacological manipulation of Go/No-Go performance in mice. With respect to the 5-CSRTT, recent studies show that visuospatial attention in mice is enhanced by nicotine (Pattij et al., 2007; Young et al., 2004), in line with the effects in humans and rats (Levin et al., 1998; Levin et al., 1996). In addition, inhibitory response control in mice measured in the same paradigm has recently been demonstrated to be modulated by GABAA, NMDA and serotonin 5-HT2 receptor ligands (Fletcher et al., 2007;

53 Chapter 3

Oliver et al., 2009). Nonetheless, little is known with respect to dopaminergic modulation of visuospatial attention and inhibitory response control in mice. Therefore, in addition to studying phenotypic differences between C57BL/6J and DBA/2J mice in more detail, we evaluated the effects of the stimulant amphetamine in both operant tasks, as the behavioral effects of this compound in rat versions of these tasks largely seem to depend on dopamine transmission (Pattij & Vanderschuren, 2008).

Go/No-Go task – In the current study, amphetamine did not affect Pinhibition, the principal measure of inhibitory control in our Go/No-Go task. Pinhibition is an index of inhibition of commission errors during No-Go trials, corrected for the number of omission errors during Go-trials. Therefore, the lack of effect of amphetamine on Pinhibition may be due to the observed robust increase in omission errors during Go trials in both strains. This was not paralleled by an increase in GoRT, suggesting a specific effect of amphetamine on omission errors rather than a general disruption of task performance. With respect to human Go/No-Go studies, stimulants have been found to increase inhibition (Trommer et al., 1991; Vaidya et al., 1998) or, similar to the current results, not affect inhibition (Kratz et al., 2009; van der Meere et al., 1999). However, changes in omission errors have not been reported in human studies. This could be due to a more general procedural difference, as omission errors are scarce in human Go/No-Go studies inhibition (Kratz et al., 2009; van der Meere et al., 1999) in contrast to the current murine version, in which the number of omission errors was titrated to 30% in both strains. Taken together, the unprecedented increase in omission errors in our murine Go/No-Go task after amphetamine administration could be indicative of an enhanced inhibition of Go-responses that was not specific to No- Go trials. In DBA/2J and C57BL/6J mice, amphetamine did not affect mean GoRTs and intra-individual variability. In human Go/No-Go tasks, stimulants have various effects on these parameters and were found to either decrease (Kratz et al., 2009) or unaltered GoRTs (Vaidya et al., 1998; van der Meere et al., 1999), and moreover, did not affect (Kratz et al., 2009) intra-individual variability. In contrast, results obtained in simple discrimination and Stop Signal paradigms in humans and rats appear less unequivocal, with reports of decreased GoRT and intra-individual variability (Pietrzak et al., 2006; Sabol et al., 2003; Spencer et al., 2009). Together, these findings suggest that the murine Go/No-Go task may not be the most well-suited paradigm to investigate mechanisms by which stimulants affect response time distributions and for this purpose it might be more promising to develop a murine version of the Stop Signal paradigm.

5-CSRTT – In contrast to the Go/No-Go task, we observed that amphetamine increased the number of premature responses in mice in the 5-CSRTT, consistent with findings in rats (Cole & Robbins, 1987; Harrison et al., 1997; van Gaalen et

54 Comparison of Go/No-Go and 5-CSRTT al., 2006a). This effect was only observed in C57BL/6J mice, clearly showing that genotype interacts with drug effects. The lack of effect in DBA/2J mice is not simply explained by an altered metabolism or uptake into the brain, since behavioral effects have been shown with comparable doses in the Go/No-Go task (1.0 mg/kg; present study) and in a conditioned place preference paradigm (Cabib et al., 2000). Importantly, large functional differences of the dopaminergic system of both strains have been reported (Cabib et al., 2002; Puglisi-Allegra & Cabib, 1997; Ventura et al., 2004) that likely affect the behavioral responses to the psychostimulant amphetamine. In further support of this notion we tested the effects of the selective dopamine transporter inhibitor GBR12909 in the 5-CSRTT. Similar to amphetamine, GBR 12909 strongly increased premature responding in C57BL/6J mice suggesting that the effects of amphetamine were dopamine-dependent and mediated by its inhibition of the dopamine transporter rather than norepinephrine or serotonin transporters (Rothman et al., 2001). Amphetamine and GBR 12909 did not affect mean correct response latencies and intra-individual variability therein in the 5-CSRTT. This is in line with previous studies in rats, in which these compounds did not significantly affect mean response latencies (Cole & Robbins, 1987; van Gaalen et al., 2006a but see Harrison et al., 1997). Intra-individual variability in correct response latencies in this task has not been studied in detail before. In contrast, in a simple discrimination task, amphetamine was found to decrease mean response latencies and intra-individual variability (Sabol et al., 2003). Collectively, similar to our murine version of the Go/No-Go task, stimulant treatment does not clearly affect response time distributions in the 5-CSRTT. Visuospatial attention was affected by the highest dose of GBR 12909, specifically in the C57BL/6J strain, whereas the doses of amphetamine might have been too low to observe similar effects. This observed decrement in accurate choice may partly be resulting from the increase in premature responses observed at this dose, as these measures have been demonstrated to correlate inversely (Dalley et al., 2008). Tardy premature responses might also be recorded as incorrect responses and thereby lead to a decrease in accurate choice. However, the observed concomitant significant decrease in the total number of correct responses indicates that GBR 12909 did not only decrease inhibitory control, but also visuospatial attention. These data are in line with the effects of GBR 12909 in rats (van Gaalen et al., 2006a) and imply the involvement of dopaminergic neurotransmission in visuospatial attention in mice.

Differences in dopamine receptor gene expression in the mPFC Given the involvement of the dopamine system in ADHD, we explored the level of expression of three dopamine receptors (Drd1, Drd4 and Drd5) in the mPFC of DBA/2J mice. The higher levels of expression of these receptors extend earlier findings of differences in the mPFC dopaminergic system between these strains

55 Chapter 3 phenotypes in these strains. Furthermore, additional experiments are required to establish a functional relation between dopamine receptor expression and the strain-specific response to amphetamine.

Taken together, this study provides evidence for a generalized reduction in inhibitory control of DBA/2J mice in comparison with C57BL/6J mice that is evident in two tasks measuring aspects of inhibitory control that, according to the present pharmacological data and earlier suggestions, do not depend on identical control mechanisms (Eagle & Baunez, 2010). Most prominently, we showed that these strains also differ in intra-individual variability in response latencies in the Go/No-Go task, suggesting more lapses in attention in DBA/2J mice. Based on these strain differences we anticipate that the panel of BXD recombinant inbred strains, derived from crossbreeding C57BL/6J and DBA/2J (Peirce et al., 2004), will be of use in further developing our understanding of ADHD by genetically dissecting ADHD-relevant phenotypes and identifying the loci/genes involved. Furthermore, despite the reduction in inhibitory response control and visuospatial attention in the 5-CSRTT by amphetamine and/or GBR 12909, which is opposite to the effect required to treat ADHD, this study clearly indicated that the dopaminergic system modulates these executive functions in the murine version of the 5-CSRTT. From a mechanistic point of view these initial findings warrant further investigation of dopamine transmission in modulating these executive functions in mice. On the other hand, from a clinical point of view, developing other operant paradigms, taxing different aspects of inhibitory control, such as response inhibition in a stop signal paradigm, might be more promising to detect beneficial effects of stimulant treatment (Pattij & Vanderschuren, 2008). Regardless, the strain-specific effects of amphetamine and GBR 12909 clearly indicate that genetic background might influence the response to stimulants. In conclusion, the genetic differences between the two employed inbred strains might contribute to multiple distinguishable ADHD-related phenotypes, suggesting that DBA/2J and C57BL/6J mice and recombinant inbred strains derived thereof likely constitute valuable animal models to study ADHD-related phenotypes.

Acknowledgements We thank M. C. W. Janssen for her contribution to the operant tests and R. van der Loo for coordinating animal deliveries. This study was in part funded by the Dutch Neuro-Bsik Mouse Phenomics consortium, supported by grant BSIK 03053 from SenterNovem (The Netherlands).

56

57

58

Chapter 4

Activity and impulsive action are controlled by different genetic and environmental factors

Maarten Loos, Sophie van der Sluis, Zoltán Bochdanovits, Iris J. van Zutphen, Tommy Pattij, Oliver Stiedl, Neuro-BSIK Mouse Phenomics consortium, August B. Smit, Sabine Spijker

Genes Brain and Behavior, 2009;8(8):817–28

59

60 Impulsivity and novelty exploration

Abstract Both impulsivity in operant tasks and locomotor activity in a novel open field are known to predict the development of addiction-related behavior in rodents. Here, we investigated to what extent impulsivity in the 5-choice serial reaction time task and various measures of novelty exploration are controlled by shared genetic and environmental factors in twelve different inbred mouse strains. No genetic correlation was observed between the level of impulsivity and levels of activity, a low correlation was observed with traditional measures of anxiety-like behavior (impulsive strains tend to be less anxious) and a highly significant correlation was found between impulsivity and specific aspects of movement. Furthermore, we found that impulsivity and all measures of novelty exploration were under control of different environmental factors. Interestingly, in the dorsal medial prefrontal cortex, a brain region involved in impulsivity and activity in novelty exploration tests, these behavioral measures correlated with the expression of different genes (respectively Frzb, Snx5, BC056474 and the previously identified Glo1). Taken together our study shows that impulsivity and activity in novelty exploration tests are genetically and environmentally distinct, suggesting that mouse models of these behaviors provide complementary insights into the development of substance abuse disorder.

Introduction Impulsivity and novelty seeking are core aspects of human personality (Bouchard & McGue, 2003) that received a great deal of attention, not the least because they have been associated with substance abuse disorder (Kreek et al., 2005; Wills et al., 1994). In rats, individual differences in impulsivity and the locomotor response to a novel open field, a putative measure of novelty seeking are predictive of aspects of drug self-administration behavior (Belin et al., 2008; Diergaarde et al., 2008; Ellenbroek & Cools, 2002; Piazza et al., 1989). These individual differences can arise from genetic differences among individual rats and/or from environmental factors. Given the relevance of animal models of predictive factors of addiction-related behavior, we examined whether shared, or alternatively, distinct genetic and/or environmental factors control impulsivity and various measures of novelty exploration. In particular inbred strains of mice provide the opportunity to separate genetic and environmental factors. When kept under highly controlled conditions, differences between inbred mouse strains result from additive genetic effects and gene-by-environment interactions (Crabbe et al., 1999). Thus, high correlation between measures of impulsivity and novelty exploration across strain means would indicate that these measures are controlled by common genetic or gene- by-environment interaction effects (genetic correlation), albeit such correlations contain some residual environmental influence due to estimation of strain means (Crusio, 2006). In contrast, behavioral differences between isogenic mice of the

61 Chapter 4 same strain result from environmental effects idiosyncratic to each individual, and hence, high within-strain correlations of impulsivity and measures of novelty exploration would indicate that these are controlled by common environmental factors (environmental correlation). Several operant tasks of impulsivity exist (Evenden, 1999), broadly categorized into those that measure impulsive action or impulsive choice (Winstanley et al., 2006a). Here, the murine version of the 5-choice serial reaction time task (5- CSRTT; Pattij et al., 2007; Robbins, 2002) was used, in which premature responses to a food predictive stimulus are a measure of impulsive action, previously shown to be predictive of increased motivation to self-administer addictive drugs in rats (Belin et al., 2008; Diergaarde et al., 2008). The locomotor response in a novel open field, predictive of amphetamine self- administration (Piazza et al., 1989), is the net result of several interacting behavioral constructs (most notably anxiety, novelty seeking and general activity). Therefore in this study novelty exploration was also measured in tests that differ in the anxiogenic nature of novel stimuli, the motivational salience of novel stimuli and the required activity to cover the novel stimulus. In addition to total activity, novelty exploration was also measured in terms of the locomotor strategy and location of activity. Finally, because the dorsal part of the rodent medial prefrontal cortex (mPFC) is involved in both impulsivity in the 5-CSRTT (Muir et al., 1996) and activity in a novel open field (Deacon et al., 2003; Holmes & Wellman, 2008), we subsequently analyzed gene expression data of this brain region (Hovatta et al., 2005) to identify genes whose expression was specifically correlating with either impulsivity or activity in a novel open field.

Materials and Methods

Subjects Six-week-old male mice (n = 12 per strain) were obtained from Jackson Laboratory (A/J, C3H/HeJ, C57BL/6J, DBA/2J, FVB/NJ), from our own breeding colony (BXD2, BXD6, BXD11, BXD21, BXD32, and BXD40) or Taconic Farms (129S6/SvEvTac). Mice arrived in the facility in 6 different batches in a period spanning 18 weeks. Batches 1 through 5 consisted of BXD mice of different strains, batch 6 entirely consisted of the remaining 6 strains. Mice were housed individually in Macrolon cages on sawdust bedding, which were, for the purpose of animal welfare, enriched with cardboard nesting material and a curved PVC tube. A metal food cup was placed in the home cage (for the purpose of habituation; used in subsequent hypophagia test). Food (Harlan Teklad) and water was provided ad libitum, except food supply during the weeks of operant testing. All mice were habituated to the facility for 14 days before testing started. During habituation mice were familiarized 4 times, on separate

62 Impulsivity and novelty exploration days, to a highly palatable snack (crumbles of cream cracker) placed in a metal food cup. Housing rooms were controlled for temperature, humidity and light- dark cycle (7 AM lights on, 7 PM lights off). All experimental procedures were approved by the local animal research committee and complied with the European Council Directive (86/609/EEC).

General procedure of behavioral testing All mice were subjected to all behavioral tests, and the order of the tests was identical for all mice. Below, days are numbered with respect to the first testing day. All testing occurred exclusively between 8-12 AM. The housing and testing rooms (4 m apart) were separated by two sound attenuating doors. On testing days, mice were transferred one by one from the housing room to the testing room and immediately introduced into the test apparatus. The same experimenter performed all tests, except for daily training in the 5-CSRTT.

Home cage behavior During the first 3 test days, the familiar snack was introduced into the familiar metal food cup in the home cage and the latency to start eating was scored (maximum duration 240 s). If the subject did not start eating within 240 s, the maximum time was assigned. The mean latency to eat was calculated over 3 days. On days 7 and 9, novel objects (metal spout and a blue plastic bottle cap respectively) were introduced into the home cage. Both latency until first touch of the object and cumulative time touching the object during 4 min were recorded manually. If a subject did not touch the object during the test, the latency was set to 240 s. The mean of both novel object sessions was used for analysis. On day 8, the familiar snack was introduced into the home cage in an unfamiliar glass cup. Latency until first touch of the unfamiliar glass cup, total duration of exploring the unfamiliar glass cup before eating, and latency to eat the snack were recorded for a maximum of 10 min. If a subject did not touch the object, the latency to explore the object was set to 600 s. If the subject did not eat the snack, the latency to eat was set to 600 s.

Novel cage On days 4 and 11, mice were transferred to a novel clean cage with fresh bedding containing the metal cup with the familiar snack. Both the latency until the first touch of the cup and latency to start eating the snack were recorded manually. The mean of latencies of both days was used for analysis. If a subject did not eat within 720 s, the maximum time was assigned. After testing, both nesting material and tube were transferred to the novel home cage.

Open field (OF) On day 14, mice were introduced into a corner of the white square open field (50  50 cm, walls 35 cm high) illuminated with a single white fluorescent light bulb

63 Chapter 4 from above (130 lx), and exploration was tracked for 10 min (12.5 frames/s; EthoVision 3.0, Noldus Information Technology). Time spent in, and number of entries into the center square area (20  20 cm) was measured using EthoVision. The SEE software (Strategy for the Exploration of Exploration; Kafkafi et al., 2005) was used to smoothen path shape to calculate the total distance moved. Furthermore, SEE uses the distribution of speed peaks to parse the locomotor data into slow local movements (lingering episodes) and progression segments, which together constitute all distance traveled. In addition to the traditional measures in the open field, describing the animal tendency to engage in exploratory behavior, SEE was used to calculate the number of progression segments and the median duration of a lingering episode. SEE also enables the calculation of measures that describe the strategy of movement once exploration has been initiated: the median distance traveled per progression segment, the median duration of a progression segment, the number of stops per distance and the median acceleration during a progression.

Dark-light box (DLB) On day 15, mice were introduced into the dark compartment (<10 lx; 25  30  30 cm, length  width  height) of a dark-light box (TSE-Systems; Baarendse et al., 2008), 15 s later the motorized black door to the identically-sized brightly lit compartment (320 lx) was opened and behavior was tracked for 5 min using infrared beams (1.4 cm apart). Complete transitions between dark and light compartment yielded the latency to enter the light, the number of entries into the light, and time spent in the light compartment.

Elevated plus maze (EPM) On days 16 or 17, all mice were introduced into the same closed arm of an EPM (arms 30  6 cm, walls 35 cm high, elevated 50 cm above the ground), facing the closed end of the arm. The EPM was illuminated with a single white fluorescent light bulb from above (130 lx) and exploratory behavior was video tracked for 5 min (12.5 frames/s, EthoVision 3.0, Noldus Information Technology). The border between center and arm entries was defined at 2 cm into each arm, producing the number of entries into the open arms, into the closed arms, onto the center platform, and time spent on the open arms. In addition, latency to explore was defined by the time between introduction onto the maze and the first appearance in the maze center.

5-CSRTT Between days 23 and 28, mice were food-restricted to gradually decrease their body weight to 90% of their free feeding weight. Operant chambers were equipped with 5 response holes, a food magazine at the opposite wall and a house light (MEDNPW-5M, Med Associates, St. Albans, VT, USA), and placed in sound attenuating ventilated cubicles. Both response holes and food magazine

64 Impulsivity and novelty exploration contain yellow LED stimulus lights and infrared response detectors. On day 28, mice were placed into the operant chambers for a 20 min habituation session during which none of the stimulus lights was switched on. On days 29, 30 and 31, purified precision pellets (12 mg, Formula P, Research Diets, New Brunswick, NJ, USA) were distributed into the magazine at random fixed inter trial intervals (ITI; 4, 8, 16 and 32 s), which coincided with switching on the magazine stimulus light. An ITI was only initiated when the previous pellet had been collected as indicated by a magazine response, during which the magazine stimulus light was off. A session ended after 25 min or sooner when the criterion of 50 pellets was reached. During the 3rd session the mean number of responses into the empty non-illuminated magazine was recorded for further analyses. In the next sessions, a trial started by the illumination of all 5 stimulus holes. An instrumental response into any of these holes switched off the light in all 5 stimulus holes, switched on the stimulus light in the magazine and delivered a food pellet into the magazine. An ITI was only initiated when the previous pellet had been collected as indicated by a magazine response, after which the magazine stimulus light was off. Sessions lasted for 25 min or 60 earned food pellets. As soon as mice earned 50 or more pellets in two sessions, or after 17 sessions, they commenced to the next phase. In this phase, trials started by the illumination of only one stimulus hole. Responses into the non-illuminated holes were without consequence. As soon as mice earned 50 or more pellets in two sessions, or after 10 sessions, they commenced to the actual 5-CSRTT procedure. In a 5-CSRTT session, a trial started with a response of the subject into the illuminated magazine, which switched off the magazine light. After a delay of 5 s (ITI) a stimulus was switched on in one of the 5 stimulus holes for a limited duration (stimulus duration). A response in the correct stimulus hole, during stimulus presentation or within the limited hold of 4 s after termination of the stimulus, switched on the magazine light and delivered a food pellet. Both incorrect responses into non-illuminated stimulus holes and omissions of a response resulted in a 5 s time-out period, during which all stimulus lights and the house light were switched off. When the time-out period ended, both the house light and the magazine light were switched on and the subject could start the next trial. A premature response into a non-illuminated stimulus hole during the ITI also resulted in a time-out period, and a subsequent response into the illuminated magazine restarted the same trial. The percentage of omissions was defined as [100  (omissions) / (omissions + number of correct and incorrect responses)]. The accuracy was defined as [100  (number of correct responses) / (number of correct and incorrect responses)]. The percentage of premature responses (impulsivity) was defined as [100  (number of premature responses) / (number of omissions + correct + incorrect responses)]. In the first 5-CSRTT session the stimulus duration was set to 16 s, which was gradually decreased in subsequent sessions to 8, 4, 2, 1.5 and 1 s if the subject reached the criterion (omissions < 30%, accuracy > 60%, started trials > 50) or after 10 sessions at the

65 Chapter 4 same stimulus duration. Baseline 5-CSRTT performance was calculated from the 6th until 10th session at stimulus duration of one second. During these baseline sessions, the average percentage of premature responses was recorded and used for further analyses together with the average latency to collect the reward from the magazine after a correct response (calculated only if a subject responded correctly for 5 times during one or more baseline sessions). During the 14th session at stimulus duration of 1 second, the ITI was programmed to vary between three fixed intervals (5, 7.5 and 12.5 s), with each interval occurring an equal number of times for all mice.

Statistical analyses of behavioral measures Before statistical analyses and graphical representation, all data were transformed to a normal distribution by normal scores transformation according to Blom's method (see Altman, 1991), except for Figure 2 in which data is displayed in the unit of measurement Estimates of the genetic effect size (narrow sense heritability) were calculated according to Hegmann and Possidente (1981) using a custom function (Microsoft Excel 2003), which takes the differences in the number of animals per group into account when estimating the within and between-strain variance (Lynch & Walsh, 1998) as previously implemented by others (Heimel et al., 2008). Environmental correlations between measures were calculated from the individual performance of all 144 individual mice after subtraction of their respective strain mean. Genetic correlations between measures were calculated using mean strain performances and using bootstrap analysis, 95% confidence intervals were calculated from the genetic correlations of 2000 resampled inbred strain means. Principal component analysis, discriminant analysis, and correlation analyses were performed with the Statistical Package for Social Sciences version 14.0 (SPSS, Chicago, IL, USA). For PCA, cases were excluded pair-wise in case of missing values. The PCA solution was promax rotated. For discriminant analysis, missing data were replaced by strain means. Discriminant analysis was performed by entering all 29 measures simultaneously into the analysis, followed by a varimax rotation of the structure matrix.

Gene expression Raw gene expression data of 6 of the 12 strains in the study was downloaded (cingulate cortex samples of 2-5 male mice per array, 2-3 arrays per strain, Affymetrix U74Av2 arrays ((Hovatta et al., 2005) NCBI GEO database, #GDS1406) and normalized using Robust Multi-chip Analysis (ArrayAssist version 3.2, Stratagene, La Jolla, CA, USA). The Excel plug-in of Significance Analysis of Microarrays (Tusher et al., 2001) was used to generate a list of probe sets that were differentially expressed between these 6 strains (FDR < 0.01). For real time quantitative PCR measurements (RT-qPCR), brains were rapidly frozen in –60 to –70ºC isopentane. In a cryostat, the brains were sliced into 150

66 Impulsivity and novelty exploration

μm thick coronal sections and the dorsal mPFC, encompassing the cingulate and prelimbic cortices, was dissected (according to Paxinos & Watson (Paxinos & Watson, 1998); Cg1, Cg2 and PrL from +2.68 to +1.34 mm anterior to Bregma). RNA was isolated, DNAse-treated and reverse transcribed to cDNA using random hexanucleotide primers (Spijker et al., 2004). Gene expression was analyzed (ABI Prism® SDS 7900, Applied Biosystems) using a SYBR Green approach (300 nM gene specific primers and the cDNA equivalent of 40 ng RNA in a total volume of 10 µl) and normalized to the geometric mean of 2 housekeeping genes (GAPDH and HPRT). Normalized gene expression levels (GEnorm) were calculated from detection cycle threshold values (Ct) as follows: GEnorm = 2 ^ (- (Ctgene x – CtHKgenes))

Results Strain means of impulsivity in the 5-CSRTT and measures of novelty exploration in five different test settings, i.e. in the home cage, novel cage, open field, dark- light box and on the elevated plus maze are presented in Supplementary Figure 1. Together, the battery of impulsivity and novelty exploration tests yielded a set of 29 behavioral measures. The genetic effect size of these measures varied between 9% and 54% (Fig. 1), indicating that both environmental and genetic variation contribute to these measures. Data were obtained for 144 subjects, but for 9 subjects (of 6 different strains) magazine latencies in the 5-CSRTT were missing (too few correct responses) and for 1 subject open field data was missing. All 12 strains displayed both average and extreme performance (Fig. 1), indicating that all strains contributed to the complex pattern of strain differences and no strain constituted an outlier on all measures.

Stability of between-strain and within-strain differences in impulsivity Reliability analysis revealed that the level of impulsivity was stable across 5 baseline sessions (Cronbach’s  = 0.91). Subsequent reliability analyses on the strain means (Cronbach’s  = 0.99) and on individual performance after subtraction of strain mean (Cronbach’s  = 0.83) indicated that both genetic and environmental factors contribute to the stability of impulsivity across baseline sessions. We observed a significant strain difference (F(11,132) = 13.07, p < 0.001) in the baseline level of impulsivity (Fig. 2) with an estimated genetic effect size of 34%. A within-session variable ITI, which makes the task temporally less predictable, significantly increased impulsivity (ITI; F(1,128) = 378.60, p < 0.001) in all strains (post hoc p < 0.05 for all strains) compared to baseline levels (Fig. 2). The known visual impairment of some strains (Wong & Brown, 2006) did not prevent any of these strains to perform significantly above chance level (20%, see Supplementary Fig. 2). Moreover, all 144 individual mice showed response accuracies above chance level (data not shown).

67 Chapter 4

Figure 1 | Behavioral differences between twelve inbred strains of mice. In the left panel, tests are ordered chronologically and their measures are ordered alphabetically from top to bottom. The genetic effect size was estimated for each measure. The right panel is a graphical representation (heat map) of strain means after normal transformation of the data (the distributions of all measures have a mean of 0 and a standard deviation of 1). Intense colors represent a positive (red) or negative (blue) deviation from the mean (white) equal to 1.96 times the standard deviation as indicated in the legend (top right). The cluster tree represents the average distance (Pearson correlation) between the 12 inbred strains used (n = 12 per strain). Note that BXD strains segregate beyond the parental lines (C57BL/6J and DBA/2J) showing transgression for several phenotypes.

68 Impulsivity and novelty exploration

Figure 2 | Levels of impulsivity of 12 inbred strains during standard and variable inter- trial interval. The level of impulsivity, measured as percentage of premature responses of 12 inbred strains of mice ranked according to baseline performance (ITI of 5 s, white bars) and during a session with variable ITI (black bars).

Lack of genetic and environmental correlation between impulsivity and distance traveled in a novel open field The inbred strains significantly differed in distance traveled in a novel open field (F(11,132) = 21.60, p < 0.001; estimated genetic effect size 54%). No significant correlation was observed between impulsivity and distance traveled in a novel open field when calculated across all 144 individual mice (i.e. no phenotypic correlation; Fig. 3a; Pearson r = 0.01, p = 0.88), nor when calculated across strain means (i.e. no genetic correlation; Fig. 3b; Pearson r = 0.02, p = 0.96). Impulsivity and distance traveled in a novel open field did also not correlate when calculated across individual mice after subtraction of strain means (i.e. no environmental correlation; Fig. 3c; Pearson r = 0.01, p = 0.92).

Lack of genetic correlation between impulsivity and novelty exploration in other tests Inbred strain means were used to calculate genetic correlations between impulsivity and 28 measures of the test battery (Table 1). As inbred strain means calculated from the average of 12 individuals per strain are estimates of the real strain mean, they are not free of environmental variation. Therefore, 95% bootstrap confidence intervals were constructed for each genetic correlation based on resampling (2000 bootstrap draws) from the twelve subjects of each inbred strain (Table 1). No significant genetic correlations were observed between impulsivity and measures describing the activity in the novelty exploration tests. Some genetic correlations with traditional measures of anxiety- like behavior were not statistically significant even though the 95% bootstrap

69 Chapter 4 confidence intervals did not include zero, illustrating that the sampling individuals in the current study may have contributed to the suppression of a possible genetic correlation. Strong significant genetic correlations were observed between impulsivity and measures describing the strategy of movement after initiation of exploration (median locomotor acceleration (Pearson r = 0.81, p < 0.01), the median duration of a progression segment in the open field (Pearson r = -0.70, p < 0.05), and the number of anticipatory magazine responses during the shaping phase of the 5-CSRTT (Pearson r = 0.81, p < 0.01; Table 1). In contrast to impulsivity, total distance traveled in the open field significantly correlated with measures of activity from several other tests (Table 1). More specifically, genetic correlations were observed between total distance traveled in the open field and on the EPM and DLB (frequency of light compartment entries) and with all other measures obtained from the open field, except median locomotor acceleration.

Figure 3 | Impulsivity does not correlate with distance traveled in a novel open field. (a) No significant phenotypic correlation (i.e. across all 144 mice in the experiment) was observed between percentage of premature responses and distance traveled in a novel open field. On both the x- and y-axis, scores of individual mice are displayed as the deviation from the mean of all 144 mice (fold difference). (b) No significant genetic correlation (i.e. across 12 strain means) was observed between percentage of premature responses and distance traveled in a novel open field. On both the x- and y-axis, strain mean scores are displayed as the deviation from the mean of all 12 strains (fold difference). (c) No significant environmental correlation (i.e. across all 144 mice in the experiment after subtraction of the respective strain mean) between percentage of premature responses and distance traveled in a novel open field was detected. On both the x- and y-axis, scores of individual mice are displayed as the deviation from its respective strain mean (fold difference).

70 Impulsivity and novelty exploration

Table 1 | Genetic and environmental correlation of all measures with impulsivity and the distance traveled in a novel open field. Impulsivity Distance traveled in open field rG rE rG rE Home cage Duration exploring novel object Duration exploring novel object with food Latency to eat food from familiar object Latency to eat food from novel object Latency to explore novel object -.38# Latency to explore novel object with food -.54# Novel cage Latency to eat food from familiar object Latency to explore familiar object with food Open field Duration in center .84** .43** Duration of a lingering episode -.71** -.31** Entries into center .90** .71** Number of progressions .92** .54** Total distance traveled NA NA Open field – Strategy of movement Acceleration .81** Distance traveled per progression -.43# .78** .51** Duration of a progression -.70* .58* .34** Number of stops per distance .46# -.67* -.44** Dark light box Duration in the light .31# Entries into the light .77** .35** Latency to enter the light Elevated plus maze Duration on the open arms .43# Entries into center .34# .34** Entries into closed arms .38# Entries into open arms .51# Latency to enter the center -.57# Total distance traveled .63* .37** Operant cage Anticipatory magazine responses .81** Impulsivity NA NA Latency to collect food from magazine

The Pearson correlation coefficients (r) of the genetic correlation (rG) and the environmental correlation (rE) with impulsivity and the exploration of a novel open field. Correlations with |r| < 0.3 are not displayed. Significant Pearson correlation coefficients are indicated in bold (* p < 0.05; ** p < 0.01). #For these non-significant genetic correlations bootstrap analyses showed that the Pearson correlation coefficients were greater than zero in more than 95% of the bootstrapped samples.

71 Chapter 4

Discriminant analysis of impulsivity and novelty exploration Multivariate analyses can be used to get insight in the structure of genetic correlations underlying impulsivity and novelty exploration. Although inbred strain means can be used in a PCA to extract genetic variation common to multiple measures, the limited availability of strain means (i.e. twelve) in this study made PCA inappropriate. Instead, a discriminant analysis was used. This analysis separates the complex pattern of strain differences into dimensions of behavior by using linear combinations (discriminant functions) of the original 29 behavioral measures. Those measures that show large differences between- strains, and hence are more influenced by genotype, receive larger weights in the discriminant functions than measures that show little between-strain differences. The use of twelve strains allowed for the extraction of a maximum of 11 discriminant functions. According to Wilks’ lamba tests, the first 10 functions (p < 0.001) were required to explain the dimensions of the strain differences. Whereas the standardized coefficients for each variable in the discriminant function can be used to determine redundant measures in a test battery, correlation of the discriminant functions with the original measures can be used to interpret the discriminant functions (for details see Stevens, 2002). To interpret the discriminant functions in terms of genetic dimensions, we therefore calculated correlations between the strain means on the discriminant scores and the strain means on all 29 measures (Table 2). Impulsivity correlated strongly with discriminant function 1, together with two measures that describe the strategy of movement once exploration has been initiated and the number of anticipatory magazine responses during the shaping phase of the 5-CSRTT (see also Table 1). Measures that describe the tendency of an animal to engage in exploration of novelty correlated best with function 2 (exploration of OF and DLB), function 5 (exploration of novel object and DLB), function 7 (exploration of OF), function 8 (exploration of EPM), function 9 (exploration of novel object with food and EPM) and function 10 (exploration of novel object with food). Apart from its correlation with function 1, strategy of movement correlated well with function 4. Impulsivity significantly correlated with function 4 as well. The discriminant analysis was also performed after excluding subjects with missing values (i.e. excluding 10 subjects of 6 different strains) to investigate whether substituting missing data by strain means had any undue effects. The exclusion affected the discriminant analysis marginally; the most noteworthy change concerned the increased correlation of impulsivity with discriminant 3 (|r| < 0.6) which in turn correlated with measures describing the strategy of movement (like function 1) and the latency to explore a novel object with and without food. In summary, impulsivity and strategy of movement seem to constitute similar genetic dimensions, whereas there is less overlap between impulsivity and genetic dimensions describing other aspects of novelty exploration.

72 Impulsivity and novelty exploration

Table 2 | Impulsivity and novelty exploration correlate with different genetic dimensions. Discriminant functions 1 2 3 4 5 6 7 8 9 10 Home cage Duration exploration novel object .90 Duration exploration novel object with food .81 Latency to eat food from familiar object Latency to eat food from novel object -.59 Latency to explore novel object -.60 -.65 .65 -.67 Latency to explore novel object with food -.69 .60 -.73 Novel home cage Latency to eat food from familiar object Latency to explore familiar object with food Open field Duration in center .72 .96 Duration of a lingering episode -.84 Entries into center .82 .91 Number of progressions .90 .76 Total distance traveled .93 .67 .70 Open field – Strategy of movement Acceleration .87 -.63 .60 Distance traveled per progression .64 .95 Duration of a progression -.73 .70 .97 Number of stops per distance -.95 Dark light box Duration in the light .89 Entries into the light .80 .65 .61 Latency to enter the light -.67 -.61 Elevated plus maze Duration on the open arms .92 Entries into center .89 Entries into closed arms .85 Entries into open arms .90 Latency to enter the center -.69 Total distance traveled .82 Operant cage Anticipatory magazine responses .89 -.60 Impulsivity .95 -.59 Latency to collect food from magazine -.96 Pearson correlation coefficient of the genetic correlation between the ten discriminant functions and 29 measures were calculated across strain means. Pearson correlation coefficients (r ≥ 0.576, p<0.05 and r ≥ 0.708, p<0.01 in bold) were used to interpret discriminant functions.

73 Chapter 4

Lack of environmental correlation between impulsivity and novelty exploration in other tests In addition to distance traveled in the open field, we analyzed the environmental correlations between impulsivity and 27 other obtained measures (Table 1). No substantial environmental correlations between impulsivity and other behavioral measures were observed (|r| > 0.3). In contrast, significant environmental correlations were observed between the total distance traveled in the open field and several measures of exploration in other tests (Table 1). To better understand the structure of the environmental factors that underlie novelty exploration a PCA was performed on the matrix of environmental correlations among all 29 measures (Table 3). Ten principal components (PCs) were extracted (Eigenvalues > 1), together explaining 76% of the variance. Impulsivity and total distance traveled in the open field loaded onto different principal components (PC 10 vs. PC 1 & 2, respectively) clearly indicating that these two measures are influenced by different environmental factors.

Gene expression correlations with impulsivity and distance traveled in a novel open field Previously acquired microarray data of gene expression in the dorsal mPFC were available for 6 (A/J, C3H/HeJ, C57BL/6J, DBA/2J, FVB/NJ and 129S6/SvEvTac) out of the 12 inbred strains in this study (Hovatta et al., 2005). Out of ~12,000 probe sets on the arrays, 82 probe sets were differentially expressed (FDR < 1%) among these six strains, and correlation of their expression level with impulsivity and total distance traveled in the open field was evaluated. Fourteen probe sets with a Pearson correlation coefficient > 0.73 (p < 0.1) with either impulsivity (7 probe sets) or total distance traveled in the open field (7 probe sets) were selected (Table 4). Using RT-qPCR, the expression level of the respective genes was measured in the dorsal mPFC of individual mice from all 12 strains that were subjected to the test battery in this study. Correlations between expression level of these 14 genes and impulsivity or total distance traveled in the open field were analyzed. Significant correlations (p < 0.05) were observed between 3 genes and impulsivity (Frzb, r = 0.71; Snx5, r = 0.64; BC056474, r = 0.59) and 1 gene and total distance traveled in the open field (Glo1, r = - 0.73; Fig. 4). Thus, alike the different genetic and environmental factors controlling impulsivity and the total distance traveled in a novel open field, also gene expression in the dorsal mPFC mapped differently onto these behaviors.

74 Impulsivity and novelty exploration

Table 3 | Impulsivity and novelty exploration load onto different environmental factors. Principal components 1 2 3 4 5 6 7 8 9 10 Home cage Duration exploration novel object -.78 Duration exploration novel object with food .98 Latency to eat food from familiar object .64 Latency to eat food from novel object .45 .48 Latency to explore novel object .75 Latency to explore novel object with food .46 .65 Novel home cage Latency to eat food from familiar object .58 Latency to explore familiar object with food .53 -.41 Open field Duration in center .84 Duration of a lingering episode .53 -.47 Entries into center .93 Number of progressions .66 Total distance traveled .45 .74 Open field – Strategy of movement Acceleration .80 Distance traveled per progression .99 Duration of a progression .91 Number of stops per distance -.99 Dark light box Duration in the light .82 Entries into the light .74 Latency to enter the light -.82 Elevated plus maze Duration on the open arms .96 Entries into center .78 Entries into closed arms .96 Entries into open arms .94 Latency to enter the center .66 Total distance traveled .40 .58 Operant cage Anticipatory magazine responses -.69 Impulsivity .86 Latency to collect food from magazine .70 The PCA solution was promax rotated and the loadings from the oblique rotated pattern matrix are shown. Loadings < 0.4 are omitted, loadings > 0.7 in bold.

75 Chapter 4

Discussion After separately analyzing genetic and environmental variation, we found no evidence for either a genetic or an environmental correlation between impulsivity and total distance traveled in a novel open field or activity in other novelty exploration tests. In addition, we showed that impulsivity and activity in a novel open field correlated with different genes in terms of gene expression in the dorsal mPFC. Operant procedures, including the 5-CSRTT, allow measurement of the same behavior across multiple test sessions, unlike most novelty-related tests that by their nature can only be novel during the first session. By repeatedly measuring the individual level of impulsivity, we were able to show that impulsivity was stable across multiple test sessions. Reliability analyses on within-strain and between strain differences in impulsivity indicated that both environmental and genetic factors contribute to this stability of impulsivity in mice. The 5-CSRTT employed in the current study requires animals to acquire a complex sequence of behaviors. We have two reasons to believe that strain differences in impulsivity are not the result of strain differences in task

Table 4 | Gene expression correlates of impulsivity and exploration of a novel open field. Probe ID Gene Symbol Correlation with Correlation Correlation Affimetrix Array data between array reproduced U74v2 and RT-qPCR using RT-qPCR

Impulsivity 101579_at Srp9 -0.96# -0.72 No 99849_at 1200016E24Rik 0.89# NA NA1 101516_at Cd59a 0.83# NA NA2 92621_at Pcbp2 -0.79# -0.99 No3 104225_at Snx5 0.74# 0.86 0.65* 104672_at Frzb 0.74# 0.91 0.71* 98516_at BC056474 0.74# 0.56 0.61*

Distance traveled in the open field 93269_at Glo1 -0.92# 0.80 -0.73* 95430_f_at Spg21 0.87# -0.61 No 92995_at Vsnl1 -0.84# 0.17 No 98531_g_at Gas5 0.84# 0.09 No 93964_s_at Ddx6 -0.76# -0.43 No 97710_f_at C530046L02Rik 0.75# -0.79 No 96302_at Sfrs7 -0.75# -0.48 No #Pearson correlation coefficient > 0.729 (p < 0.1) with impulsivity or the total distance traveled. *The correlation between gene expression level and impulsivity or the total distance traveled was reproduced at a more stringent level (p < 0.05) using RT-qPCR. 1Probeset sequence corresponds to a repetitive sequence in the mouse genome of viral origin, no RT-qPCR was performed. 2Primer sequences perfectly matched target, but no product was detected after 40 PCR cycles. 3Correlation with impulsivity was significant (p < 0.05), but of opposite direction compared to micro array data.

76 Impulsivity and novelty exploration acquisition or strategy to solve the task. First, general impairments in the acqui- sition of the task are unlikely, since under baseline task requirements all inbred strains made significantly more correct responses than could be expected only by chance. Secondly, manipula- tion of the temporal predict- tability of the task by varying the ITI (Robbins, 2002) significantly increased impul- sivity in all strains. This indicates that all strains displayed more difficulties in withholding responses par- ticularly when the ITIs were long, suggesting similar inhibitory control mecha- nisms. Figure 4 | Correlation of gene expression with impulsivity In particular the behavior of and distance traveled in a novel open field. Scatter plots of the A/J strain of mice the significant correlations between normal transformed levels of (a - c) impulsivity (% premature responses) and (d) exemplified the absence of a the distance traveled in a novel open field and the level of genetic correlation between expression of four genes in the mPFC. Normalized levels of impulsivity and measures of gene expression are displayed as deviation from the mean activity. The A/J strain of expression level (GEnorm). mice is well known for its docile, inactive appearance, while it turned out to be the second most impulsive strain in this study. Furthermore, we noticed the particular good performance of the 129S6/SvEvTac strain with respect to high accuracy and low impulsivity, in line with the performance of related 129S2/SvHsd strain published previously (Pattij et al., 2007). Because novelty exploration in tests such as the open field does not only depend on the animal’s tendency to explore novelty but also on the induction of a state of anxiety and general levels of activity, we measured novelty exploration in several tests that varied in anxiogenic nature of the novel stimulus, motivation to move, and the required activity to cover the novel stimulus. Yet, no significant environmental or genetic correlations were found between impulsivity and measures of activity in these novelty exploration tests, indicating that impulsivity and activity in novelty exploration tests are influenced by different environmental and genetic factors.

77 Chapter 4

In contrast to general measures of activity, specific measures of movement did genetically correlate with impulsivity, most notably locomotor acceleration. Acceleration was previously reported to be genetically different from other locomotion-related measures as well (Kafkafi et al., 2005). Bootstrap analyses indicated that the presence of some environmental variation in the genetic correlations (see Introduction) may have prevented significant correlations between impulsivity and traditional measures of anxiety-like behavior (e.g. the duration on open arm and duration in the light compartment). This finding suggests that correlations between impulsivity and anxiety-like behavior may reach statistical significance in a different or larger sample of mice. It is well known that prior experience affects behavior in subsequent tests (Henderson, 1969; McIlwain et al., 2001). To minimize carry over effects in our battery, we administered the tests in the order of increasing putative anxiogenic content. However, in order to calculate environmental (within-strain) correlations between multiple measures in the test battery it was necessary to subject each individual mouse to all tests in the battery. With respect to the environmental correlations, we did not control the rearing environment of the subjects in this study. Hence, it could be possible that in the current study not all environmental factors were present that might have affected both impulsivity and novelty exploration. In this study, genetic correlations were calculated from differences between inbred mouse strains. Between-strain differences result from additive genetic effects and gene-by-environment (e.g. laboratory) interactions (Crabbe et al., 1999). Therefore, we note that the absence of genetic correlations between impulsivity and novelty exploration in our study does not entirely rule out the possibility of a correlation when these experiments are repeated with the same strains obtained from a different vendor, or when performed in a different laboratory. Furthermore, 6 out of 12 strains used in this study were recombinant inbred strains derived from a cross between C57BL/6J and DBA/2J (BXD strains). The behavioral phenotypes of these BXD strains transgressed beyond the phenotype of their parents on several measures, likely due to the novel combinations of parental loci in these strains, indicating that the BXD strains contributed to the genetic diversity in the current study. In addition, we calculated genetic correlations between impulsivity and all other measures while omitting the BXD strains, but no significant genetic correlations emerged other than those presented in Table 1. This suggests that the inclusion of recombinant inbred strain with common inbred strains did not specifically influence our results. The dorsal part of the mPFC has been implicated in impulsivity as measured in the 5-CSRTT (Muir et al., 1996) and activity in novelty exploration tests (Deacon et al., 2003; Holmes & Wellman, 2008). Using previously acquired microarray data of the dorsal mPFC, we observed correlations between impulsivity and the expression level of the genes Frzb, Snx5, and BC056474.

78 Impulsivity and novelty exploration

These correlations were confirmed by RT-qPCR in the dorsal mPFC of mice that were subjected to the test battery in this study. The exploration of a novel open field showed a high correlation with the expression of Glo1, a gene involved in glutathione homeostasis. This correlation between expression of Glo1 and behavior in an open field has been observed before (Ditzen et al., 2006; Hovatta et al., 2005). Knockdown and overexpression of Glo1 in the dorsal mPFC, respectively decreased and increased time spent in the center of the open field, indicating that Glo1 expression levels in the dorsal mPFC are crucial to behavior in the open field. The correlations between impulsivity and Frzb, Snx5 and BC056474 gene expression have not been described before. Frzb (Sfrp3) is a member of the secreted frizzled-related protein family that inhibits the Wnt signaling pathway. Among other functions, Wnt signaling is important in axon path finding (Bovolenta et al., 2006), and synapse structure and function (Ataman et al., 2008) Interestingly, Frzb was downregulated by acute ethanol treatment in the mPFC of mice (Kerns et al., 2005), indicating that the expression of this gene is regulated by at least this type of drug of abuse. From the other two identified genes, Snx5 belongs to the sorting nexin family. Members of this family are involved in intracellular trafficking, and Snx5 may be involved in trafficking of cell surface receptors (Otsuki et al., 1999). The gene BC056474 is an uncharacterized conserved protein (www.informatics.jax.org). The involvement of both Snx5 and BC056474 in behavioral functions is essentially unclear and needs to be established. In conclusion, we provide evidence that two predictors of addiction-related behavior in rodents, namely impulsivity and activity in novelty exploration tests, which may be processed by similar underlying neuronal systems such as the mPFC, are under the control of different genetic and environmental factors. Rodent models of impulsivity and activity in novelty exploration tests will therefore act complementary in our understanding of the mechanisms underlying susceptibility for substance abuse (Belin et al., 2008; Diergaarde et al., 2008; Ellenbroek & Cools, 2002; Fattore et al., 2009; Piazza et al., 1989), and research defining the role of the identified genes may contribute to the understanding of the different molecular mechanisms involved.

Acknowledgements We thank T. van der Vlis and A. Chatzopoulou for their contribution to the 5- CSRTT and the gene expression experiments. We thank R. van der Loo for coordinating animal deliveries and Dr. I. Golani and A. Dvorkin for use of, and help with SEEbasic software. S. van der Sluis was financially supported by NWO/MaGW VENI-016-095-172. This study was in part funded by the Dutch Neuro-Bsik Mouse Phenomics Consortium, supported by grant BSIK 03053 from SenterNovem (The Netherlands). The Neuro-Bsik Mouse Phenomics consortium is composed of the laboratories of A.B. Brussaard, J.G. Borst, Y. Elgersma, N.

79 Chapter 4

Galjart, G.T. van der Horst, C.N. Levelt, C.M. Pennartz, A.B. Smit, B.M. Spruijt, M. Verhage and C.I. de Zeeuw, and the companies Noldus Information Technology B.V. (www.noldus.com) and Synaptologics B.V. (sylics.com).

80 Supplementary data

Supplementary data

81 Chapter 4

82 Supplementary data

Supplementary Figure 1 | Strain means of measures in the home cage (a – c), novel cage (d), open field (e – i), dark light box (j – k), elevated plus maze (l – n) and operant cage (o).

83 Chapter 4

Supplementary Figure 2 | The accuracy in the 5-CSRTT is defined by the number of correct responses divided by the sum of correct and incorrect responses. If a random nose-poke response is made in one of the 5 response holes, the chance level of a correct response is 20%. All strains, including the strains with early onset retinal degeneration (black bars), perform well above chance level (dashed line, one-sample t-test; *p<0.001, **p<0.0001).

84

85

86

Chapter 5

Independent genetic loci for sensorimotor gating and attentional performance in BXD recombinant inbred strains

Maarten Loos, Jorn Staal, Tommy Pattij, Neuro-BSIK Mouse Phenomics consortium, August B. Smit, Sabine Spijker

Genes Brain and Behavior, In Press

87

88 Sensorimotor gating and attention

Abstract A startle reflex in response to an intense acoustic stimulus is inhibited when a barely detectable pulse precedes the startle stimulus by 30 – 500 ms. It has been theorized that this phenomenon, named prepulse inhibition (PPI) of a startle response, is an automatic early-stage gating process contributing to the ability to focus attention. Deficits in PPI may therefore contribute to deficits in attentional processing. Both deficits are observed in schizophrenia spectrum disorders. Here, we investigated whether there is overlap in genetic control of PPI and attentional processing phenotypes in the panel of BXD recombinant inbred strains of mice. Using an individually-titrated prepulse intensity to handle differences in perceived prepulse intensities among strains, we identified a significant QTL for PPI at the mid-distal end of chromosome 17. A measure of attentional processing in the 5-CSRTT, response variability, mapped to a different locus on proximal- mid chromosome 16. In addition, the estimated genetic and environmental correlations between PPI and several attentional phenotypes were low and not significant. Taken together, the observation of separate genetic loci for PPI and attention and the absence of genetic and environmental correlations indicate that differences in sensorimotor gating do not contribute to differences in attentional performance. Therefore, it is worth pursuing the causative genes residing in both attention and PPI QTL as these may contribute to separate molecular pathways implicated in neuropsychiatric diseases, such as schizophrenia.

Introduction The expression of a startle reflex in response to an intense acoustic stimulus is markedly reduced when preceded by a non-startling stimulus (Graham, 1975), a phenomenon called prepulse inhibition (PPI). The dominant theory of PPI is that gating mechanisms dampen processing of new disruptive input, thereby protecting processing of the prepulse stimulus (Graham, 1975). This bottom-up control over information processing is thought to contribute to the ability to focus attention on the most salient aspects of the stimulus-laden environment (Braff & Geyer, 1990; Braff et al., 2001). Besides this conceptual relation between PPI and attentional processing, there is evidence from patients with schizophrenia spectrum disorders, in which deficits in PPI (for review see Braff et al., 2001) and attentional performance (Nieuwenstein et al., 2001) co-occur. Given the fact that both PPI and attentional performance are influenced by genetic factors (Anokhin et al., 2003; Greenwood et al., 2007; Groot et al., 2004), this questions to what extent PPI and attentional processing might be influenced by common genetic factors. Human PPI and attention tasks have excellent translational rodent counterparts. Rodent PPI assays show high predictive and construct validity with respect to schizophrenia (Geyer et al., 2001; Powell et al., 2009). The human continuous performance task (CPT) of attentional processing (CPT; Beck et al., 1956) has been adapted for rodents into the 5-choice-CP task (Young et al., 2009a), and the

89 Chapter 5 widely used 5-choice serial reaction time task (5-CSRTT; Robbins, 2002). In the 5-CSRTT, response accuracy is the primary measure of attention. Intra- individual variability in correct response latencies (response variability hereafter) has been recognized as measure of lapses in attention (Castellanos & Tannock, 2002; Leth-Steensen et al., 2000; Loos et al., 2010b; Sabol et al., 2003; Sergeant & van der Meere, 1990) and can also be measured in the 5-CSRTT. To investigate whether PPI and attention phenotypes map to the same genetic loci, we employed a panel of recombinant inbred strains of mice, which are derived from an intercross of C57BL/6J and DBA/2J (BXD strains; Peirce et al., 2004). These strains have previously been used to detect QTL underlying PPI, (Hitzemann et al., 2001; McCaughran et al., 1999; Willott et al., 2003), but not attentional performance. In addition, inbred strains of mice can be used to approximate genetic and environmental correlations between PPI and attentional phenotypes (Loos et al., 2009). To identify candidate genes in mouse QTL potentially influencing PPI and attention, we used brain gene expression data of BXD strains.

Materials and Methods

Animals Parental and BXD lines were received from Jackson Laboratories, or from Oak Ridge Laboratory (BXD43, BXD51, BXD61, BXD62, BXD65, BXD68, BXD69, BXD73, BXD75, BXD87, BXD90), and were bred by the Neuro-BSIK consortium. Male seven-week-old mice were singly housed on sawdust in standard Makrolon type II cages (26.5 cm long, 20.5 cm wide and 14.5 cm high) enriched with cardboard nesting material with food (Teklad 2018, Harlan Laboratories, Horst, The Netherlands) and water ad libitum. After one week of habituation, body weights were recorded and mice were subjected to a single 45 min session in the startle response chambers. Over the subsequent 6 days, food was limited to gradually decrease body weights to 90 – 95% of their initial body weight, before daily training in operant cages commenced (5 days each week, 30 min per session). Due to logistical reasons, not all mice tested in the 5-CSRTT were subjected to PPI testing. All experimental procedures were approved by the local animal research committee and complied with the European Council Directive (86/609/EEC).

Acoustic startle and prepulse inhibition It has been well-documented that prepulse intensity strongly modifies PPI (Braff et al., 2001; Plappert et al., 2004; Varty et al., 2001). This indicates that strain differences in perceived salience of prepulse stimuli could interfere with the calculation of PPI, and hence analyses of correlation with attentional performance. By generating an extensive acoustic startle response (ASR) curve

90 Sensorimotor gating and attention and determining the startle threshold (ST) for each individual mouse we aimed to control for differences in perceived intensity of prepulses when calculating PPI. In addition, a passive PPI procedure with neutral prepulse stimuli and a short prepulse to pulse stimulus onset asynchrony (SOA) was used to reduce (Bohmelt et al., 1999; Dawson et al., 1993; Filion et al., 1993), but not prevent, (Elden & Flaten, 2003; Thorne et al., 2005) the possibility of attentional processes to modify PPI. Acoustic startle and PPI were measured during one 45 min session in four Plexiglas cylinders in ventilated sound-attenuating chambers (Med Associates, St. Albans, VT), placed on separate passive heavy vibration-free tables (Newport Corporation, Irvine, CA). During testing, a separate speaker provided white noise background of 75 dB. The intensity of the startle stimuli was calibrated with a microphone placed inside the Plexiglas cylinders in a closed chamber, while white background noise was switched off. Hence, the reported prepulse and pulse intensities are lower than the actual cumulative sound pressure level during testing (e.g., a 75 db pulse in addition to a 75 dB background noise add up to 78.01 dB). The session started with a habituation period of 5 min, followed by a total of 245 trials with pseudo randomized interval periods (5 – 15 s) consisting of 10 acoustic startle trials with white noise bursts at various intensities (65, 70, 75, 80, 85, 90, 95, 100, 105, 110, and 115 dB) and 15 prepulse inhibition trials with white noise bursts at various prepulse intensities (0, 65, 70, 75, 80, 85, 90, 95, 100 dB; startle intensity always 120 dB) in pseudo randomized order such that all 4 boxes produced startle stimuli at exactly the same time. Onset of white noise prepulse stimuli (20 ms; 1 ms programmed rise/fall time) and startle stimuli (40 ms; no programmed rise/fall time) were separated by a 60 ms interval (i.e., SOA). In each trial, the highest startle intensity peak (in relative machine units) was collected during the 100 ms interval after the startle stimulus, from which the individual mean highest startle intensity peak during the 100 ms null-period prior to prepulse stimuli was subtracted. The startle sensitivity was determined for each mouse by determining the ST at which a statistically significant startle response was measured (one-sample t-test, with average null-period as reference value). The equipment was calibrated to allow for a wide range of startle intensities, however, the force generated by some mice at the highest pulse intensities could exceed the dynamic range of the equipment (maximum of 2047 units) artificially reducing the percentage of PPI in subsequent analyses. Therefore, when the number of such censored 120 dB pulse trials was more than 33% (e.g., more than 5 out of 15) no PPI was calculated. The percentage of PPI was calculated as follows: PPI = 100 * [(mean startle intensity pulse) - (mean startle intensity pulse with prepulse) / (mean startle intensity pulse)].

5-CSRTT Mice were trained to perform the 5-CSRTT on an individually-paced schedule in operant chambers (MEDNPW-5M, MedAssociates, St Albans, VT, USA), as

91 Chapter 5 described previously (Loos et al., 2010b; Loos et al., 2009). During the first week, mice underwent one habituation and 4 magazine training sessions. In the next week, mice were trained to perform an instrumental response into the stimulus holes to earn a reward, and only commenced to 5-CSRTT training when they earned at least 50 rewards within one session. During 5-CSRTT training a trial started with a response of the subject into the illuminated magazine, which switched off magazine light and after an ITI of 5 s a stimulus in one of the five stimulus holes was presented for a limited duration (stimulus duration). A response in the correct stimulus hole within the limited hold period of 4 s after termination of the stimulus, switched on the magazine light and delivered a food pellet. The latency between a correct response and the detection of a head entry into the magazine were recorded (magazine latency). Both an incorrect response into a non-illuminated stimulus hole and an omission of a correct response resulted in a time-out period, during which all stimulus lights and the house light were turned off. When the time-out period ended, both the house light and the magazine light were switched on, and the subject could start the next trial. A premature response into a non-illuminated stimulus hole during the delay period also resulted in a time-out period, and a subsequent response into the illuminated magazine restarted the same trial. The percentage of omission errors was defined as [100  (omissions) / (omissions + number of correct and incorrect responses)]. The response accuracy was defined as [100  (number of correct responses) / (number of correct and incorrect responses)]. In the first 5-CSRTT session the stimulus duration was set at 16 s, and was decreased in subsequent sessions to 8, 4, 2, 1.5 and 1 s when the subject reached criterion performance (omissions < 30%, accuracy > 60%, started trials > 50) or after 10 sessions. Intra-individual variability of correct response latencies was defined by the standard deviation of the correct response latencies. In addition, the mean and the mode of correct response latencies were calculated as described previously (Loos et al., 2010b). Dependent measures were calculated from the 6th until the 10th session at stimulus duration of 1 s, and the average of these sessions was used as standard 5-CSRTT performance. In the week following the 10th session, on Tuesday the stimulus duration was programmed to vary (1, 0.5 and 0.25 s; durations occurring an equal number of times within session). Mice were excluded from analyses when they initiated fewer than 30 trials on average or made no correct or incorrect responses during two or more standard sessions. Strains that completed fewer than 50 trials on average in combination with magazine latencies greater than 4 s, together indicative of reduced motivation, were excluded.

Data analysis For evaluation of strain differences, the statistics of one-way analysis of variance (ANOVA) were reported. The effects of startle stimulus intensity on ASR were analyzed using repeated-measures ANOVA; if Mauchly’s test for sphericity of data was significant, more conservative Greenhouse-Geisser corrected degrees of

92 Sensorimotor gating and attention freedom and resulting probability values were reported. Pearson correlations coefficients and respective probabilities were calculated across strain means, to approximate genetic correlations. Environmental correlations were estimated from scores of individual mice after subtraction of the respective strain mean, as described previously (Loos et al., 2009). Interval mapping analysis was performed in GeneNetwork (www.genenetwork.org) that uses the embedded MapManager software (Manly et al., 2001) to perform Haley–Knott regression. Empirical P-values, derived using 1000 permutations, were used to assess the whether the peak of a QTL was statistically significant (genome-wide P-value <0.05) or suggestive (on average one false positive per genome scan; genome- wide P-value <0.63; Lander & Kruglyak, 1995). A one-LOD drop-off was used to determine the QTL confidence interval of each peak. GeneNetwork’s gene expression databases of whole brain (UCHSC BXD Whole Brain M430 2.0 November 2006 RMA, Accession number: GN123) or prefrontal cortex tissue (Virginia Commonwealth University, BXD Prefrontal Cortex Saline Control M430 2.0 (Dec06) RMA Dataset, Accession number: GN135) were used to calculate Pearson correlations of different probesets of genes with measures of PPI and attention respectively, using GeneNetwork’s online tools. For the analysis of non-synonymous mutations in genes under the QTL peak, GeneNetwork’s comprehensive SNP browser was used (date: June 2010). Behavioral phenotypes of mouse mutants of candidate genes detected in our study were derived from the Mouse Genome Informatics (MGI) database and information on human disorders associated with these candidate genes was derived from the Online Mendelian Inheritance in Man (OMIM) database.

Results

Strain differences in acoustic startle and prepulse inhibition A total of 442 mice, encompassing 39 strains, was tested for ASR and PPI, i.e., 35 BXD strains, the parental lines and the reciprocal F1 (n ≥ 6 per strain, average 11.3 per strain). The mean startle intensity levels during the null period differed significantly among strains (F(38,403) = 4.99, P < 0.001), indicative of differences in activity between strains in the apparatus, and was used to correct startle responses intensities as described in the materials and methods. Startle threshold. The ASR significantly increased when louder startle stimuli were applied (Fig. 1; strain: F(1.91,773.42) = 1005.80, P < 0.001). We observed significant strain differences in the shape of these ASR-curves (strain  startle stimulus intensity: F(72.93, 773.42) = 9.1, P < 0.001), that were at least partly explained by the threshold at which mice started showing a significant startle (ST; Fig. 2a). A significant difference among strains was detected for this individually determined ST (F(38,403) = 7.44, P < 0.001).

93 Chapter 5

Prepulse inhibition. The variation among mice in ST could reflect differences in the perceived salience of the auditory stimuli. To control for differences among mice in perceived salience of prepulse stimuli, data of prepulse trials, in which the prepulse intensity was 10 dB lower than the ST, were taken to calculate the individually titrated PPI. In addition, we used filtering criteria (see Materials and Methods) to exclude mice with extreme startle sensitivity (ST ≤ 75 dB) or startle intensity (> 33% censored pulse trials), together excluding 3 strains for further analyses due to low n (n < 6; BXD 1, 6, 97). Significant strain differences in PPI were detected among the remaining 335 mice of 36 strains at the individually titrated prepulse intensity (Fig 2b; F(35,299) = 2.50, P < 0.001). Furthermore, conventional prepulse inhibition percentages were calculated at prepulse intensities ranging from 65 – 75 dB and significant strain differences were detected (65 dB: F(35,299) = 4.70, P < 0.001; 70 dB: F(35,299) = 5.23, P < 0.001; 75 dB: F(35,299) = 5.44, P < 0.001). Strains with known hearing deficits (BXD16 and BXD24; Willott & Erway, 1998) showed a decreased PPI at these conventional PPI, however, they showed PPI to the same extent as other BXD strains at the individually titrated prepulse intensity.

Figure 1 | Qualitative differences in the shape of ASR curves. For each strain the average ASR curve is presented. Although BXD16 and 24 show a clear startle response at 120 dB, the inset indicates impaired startle at low stimulus intensities, which is reflected in their high ST. BXD 23 shows a clear startle at low stimulus intensities, but remarkably low startle at 120 dB compared with other strains. Some strains (e.g., BXD 1, 6 and 97) show an asymptotic ASR curve due to the censoring of startle response at an upper limit of 2047 units (see M&M for inclusion criteria).

94 Sensorimotor gating and attention

Figure 2 | Strain differences in startle sensitivity. (a) The startle stimulus intensity (dB; ± SEM) at which mice showed significant startle compared with the average null-period, 100 ms prior to a stimulus, was taken as startle threshold (ST). (b) The percentage of PPI (± SEM) was calculated for trials at the individually titrated prepulse intensity.

Strain differences in attentional performance In total, we analyzed 5-CSRTT data of 558 mice, distributed across 41 BXD strains (on average n = 13 per strain, minimum n = 6 per strain) and the parental lines (C57BL/6J, n = 34 and DBA/2J, n = 20). Significant strain differences were detected for measures of attention; response accuracy (Fig. 3a; F(42, 545) = 3.69, P < 0.001), response variability (Fig. 3b; F(42, 545) = 2.71, P < 0.001) and errors of omission (F(42, 545) = 4.48, P < 0.001; not shown). The first two of these measures correlated significantly across strains (rresponse accuracy, response variability = -0.75, n = 43, P < 0.001). When calculated from the scores of individual mice (i.e., after subtraction of the respective strain mean), the estimated environmental correlation between both measures was strong (rresponse accuracy, response variability= -0.64, n = 588, P < 0.001). In contrast, errors of omission, did not correlate with response accuracy (rresponse accuracy, omissions = -0.06, n = 43, n.s.) and response variability (rresponse variability, omissions = -0.01, n = 43, n.s.) across strains, or across individual performance (rresponse accuracy, omissions = 0.04, n = 588, ns; rresponse variability, omissions = 0.00, n = 588, n.s.), indicating that this measure taps into a different aspect of 5-CSRTT performance. Although mice with potentially low motivation or motor impairment, as indicated by number of trials and magazine latencies

95 Chapter 5

Figure 3 | Strain differences in attentional performance of BXD strains. (a) Response accuracy is the traditional measure of attention in the 5-CSRTT. All BXD strains showed response accuracy (± SEM) well above chance level (20%). (b) Displayed is the mean response variability (± SEM), a putative measure of lapses in attention. were removed before analyses, we did observe a correlation between errors of omission and magazine latencies (r omissions, magazine lantencies = 0.47, n = 43, P < 0.001). All included mice showed response accuracies above 40%. Furthermore, we observed a significant decrease in response accuracy after increasing attentional load by randomly shortening the stimulus duration (SD: F(1,534) = 355.46, P < 0.001; SD  strain: F(42, 534) = 1.53, P < 0.05), which did not reach significance in all strains (Supplementary Fig. 1). Together these data indicate that all strains used an attention strategy to solve the task, as opposed to a random response strategy.

QTL analysis and candidate gene analyses Prepulse inhibition. A significant QTL was detected for the percentage of PPI at the individually titrated prepulse intensity (Fig. 4a) at chromosome 17 (one-LOD drop-off confidence interval: 71.10 – 75.47 Mb). The peak additive effect in terms of PPI was 7.63%, with the BL6 allele contributing to higher PPI scores. No significant QTL were detected for ASR or other measures of PPI. Suggestive QTL are presented in Table 1. The PPI QTL region contained 33 genes (Supplementary Table 1). Analyses of publicly available SNP data of

96 Sensorimotor gating and attention

C57BL/6J and DBA/2J mice (www.genenetwork.org) indicated the presence of non-synonymous mutations in 7 of these genes (Fig. 4a; Myom1, Emilin2, Ndc80, Spdya, Clip4, Alk, Tct27). Analyses of adult gene expression data of whole brain tissue indicated that from these 33 genes there were significant correlations between the percentage of PPI at the individually titrated prepulse intensity and probe sets targeting three genes (Fig. 4a; YpeI5, Ehd3, Spast) (Pearson |r| > 0.444, P < 0.05). However, none of these correlations could withstand Bonferroni correction for multiple testing (Supplementary Table 1). Measures of attention. A significant QTL was detected for response variability (Fig. 4b; peak additive effect 0.053 s) at chromosome 16 (one-LOD drop-off confidence interval: 19.88 – 26.55 Mb), with the BL6 allele contributing to reduced response variability (i.e., increased attentional performance). This QTL coincided with a suggestive QTL for response accuracy at approximately the same location (Table 1), in which the BL6 allele contributed to increased response accuracy. Other dependent 5-CSRTT measures mapped significantly (mode of correct response latencies; Chr16 (93.93 – 97.54)) or suggestively (mean of response latencies, errors of omission and magazine latencies; see Table 1) to loci that did not overlap the attention QTL on chromosome 16. Analyses of publically available SNP data of C57BL/6J and DBA/2J mice (www.genenetwork.org) did not indicate non-synonymous mutations within the 79 genes located in the attention QTL. Analyses of adult gene expression data of prefrontal cortex tissue indicated that correlations between response variability and probe sets targeting exons or untranslated regions of four of these genes (Fig. 4b; Abcc5, 2310042E22Rik, Fetub and Cldn16) were significant (Pearson |r| > 0.423, 0.01 < P < 0.05). However, none of these correlations could withstand Bonferroni correction for multiple testing (Supplementary Table 2).

No genetic or environmental correlation between sensorimotor gating and attention Response accuracy and response variability did not correlate significantly with the percentage of PPI at the individually titrated prepulse intensity across strain means (Fig. 5a). PPI at prepulse intensities ranging from 65 – 70 dB did not correlate with attention in the 5-CSRTT either, except for the correlation between accuracy and PPI at the lowest prepulse intensity of 65 dB (rPPI65, accurate responses = 0.36). This correlation was solely due to two BXD strains with known hearing impairment (BXD16 and BXD24; (Johnson et al., 2000; Willott & Erway, 1998)) and high ST (Fig. 2b); when these strains were excluded from correlation analyses, no significant correlation was detected anymore (Fig. 5b). In addition to these correlations across strain means, which approximate the genetic correlation, we calculated environmental correlations from scores of individual mice after subtraction of strain means between PPI at various prepulse intensities (individually titrated, 65, 70 and 75 dB) and attention (response accuracy and response variability). All of these environmental correlations were

97 Chapter 5

Figure 4 | Significant QTL for PPI and attention. The LRS scores (y-axis) quantify the relation between genomic markers (x-axis) and the trait. The threshold for significance (genome-wide P = 0.05) and suggestive significance (genome-wide P = 0.63) is indicated. (a) Genome-wide significance was reached at a locus on chromosome 17 for PPI at individually titrated prepulse intensity and (b) response variability at a locus on chromosome 16. The magnified views display the one-LOD drop off confidence interval of the QTL. Candidate genes potentially influencing both traits are indicated. The 7 genes harboring non-synonymous SNPs are indicated, and the 7 genes of which the expression correlated with the respective trait are indicated in italics. The height of the grey markers placed on the x-axis displays the number of SNPs between C57BL/6J and DBA/2J at each locus as presented in WebQTL (www.genenetwork.org).

98 Sensorimotor gating and attention non-significant (|r| < 0.07, n = 294). Furthermore, we did not detect genetic or environmental correlations between errors of omission and PPI measures.

Discussion

QTL for sensorimotor gating We observed substantial differences in ASR at maximum startle intensity and percentage of PPI among mouse strains, indicative of a genetic contribution to these phenotypes in line with previous studies (Hitzemann et al., 2001; Hitzemann et al., 2008; Leussis et al., 2009; Liu et al., 2003; Petryshen et al., 2005; Samocha et al., 2010; Watanabe et al., 2007). In addition, we noted differences in the shape of ASR curves among inbred strains that we collapsed into a parameter called startle sensitivity (i.e. ST) that showed a genetic effect size of 30%. Using this ST, we were able to identify 1) mice that exhibit a startle response at stimulus intensity levels that are regularly used as prepulse stimulus (i.e., 65 – 75 dB), and 2) mice that do not show a startle response until intensity levels increased up to 115 dB. This is problematic in face of the definition of PPI; the prepulse should not elicit a startle response. Furthermore, this suggests that a prepulse with certain intensity (e.g., 70 dB) may not be equally salient to all mice, and therefore may impact the percentage of PPI. One reason for differences in perceived salience of the prepulse is the segregation of alleles that cause age- related hearing loss in the panel of BXD lines (Johnson et al., 2000; Willott & Erway, 1998). As a result, strain differences in PPI at a specific prepulse intensity may reflect differences in perceived salience in addition to differences in sensorimotor gating function. To counteract potential differences in prepulse perception, we titrated the

Table 1 | Significant and suggestive QTL of PPI and attention.

Measure One LOD drop–off C.I. of QTL (Mb)

ASR at 120 dB Chr3(54.56 – 57.63), Chr10(64.17–71.42), Chr12(15.66–26.75, 36.1–46.20)

Startle Threshold Chr10(67.08–79.09), Chr16(59.7–67.89, 76.38–82.14), Chr18(34.13–39.20)

PPI at 65 dB Chr1(108.76–126.52), Chr12(115.17–12.16), Chr13(104.62–108.47)

PPI at 70 dB Chr1(108.76–126.52), Chr13(105.47–107.93)

PPI at 75 dB Chr1(108.76–126.52), Chr13(106.90–107.82)

PPI at ST – 10 dB Chr17(71.10–75.47) 1 Response accuracy Chr2(137.49–159.19), Chr5(14.16–23.47), Chr16(23.74–27.50)

Response variability Chr5(24.39–59.04), Chr16(19.88–26.55)

Errors of omission Chr14(79.83–84.04), Chr19(57.68–61.21) Response latency mean Chr5(29.34–40.03), Chr8(105.16–113.84), Chr12(48.41–59.05), 1 Chr16(92.95–97.61, 17.66–26.07 , 7.96–13.14)

Response latency mode Chr16(93.93–97.54), Chr8(105.16–113.84), Chr12(48.41–59.05) Magazine latency Chr8(98.23–99.92), Chr13(45.77–52.73), Chr16(30.07–36.61), Chr19(16.50–21.21) For each parameter the one LOD drop-off confidence interval is given for suggestive (genome-wide P < 0.63) and significant (genome-wide P < 0.05) QTL. QTL that reached genome wide significance are indicated in bold. 1This suggestive QTL overlaps with the location of the significant response variability QTL.

99 Chapter 5 prepulse intensity for individual mice to 10 dB below its ST under the assumption that such a prepulse would be perceived, but would not result in a startle response. We used the percentage of PPI elicited by these individually titrated prepulses in a genetic mapping analysis, and detected a significant QTL on mouse chromosome 17. This finding at least indicates the sensitivity of our PPI analysis method to genetic effects. Notwithstanding procedural differences, our PPI data correlated significantly with published and unpublished PPI datasets of intersecting panels of BXD strains in WebQTL (best correlation found: rPPI75 current study, PPI85 Philip et. al. = 0.51, P < 0.05; n = 28; record ID 11428). Despite this consistency in results, there is no overlap in significant or suggestive QTL identified by individually titrated PPI in the current study (Chr 1, 12 (~118mb), 13, 17), or conventional PPI in our study, with significant QTL detected in other studies using intercrosses of C57BL/6J and DBA/2J mice, such as Chr 3, 5, 11 and 16 (Hitzemann et al., 2001; Hitzemann et al., 2008; Liu et al., 2003). Also, no overlap between these QTL and QTL identified using intercrosses of other mouse strains were identified, e.g., Chr 4, 10, 11, 12, 16 (~73 Mb) and Y (Leussis et al., 2009; Petryshen et al., 2005; Samocha et al., 2010; Watanabe et al., 2007). Apart from differences in the experimental set-up between all these studies, the lack of overlap with our study might be due to differences in the programmed interval between prepulse and pulse onset in this study (60 ms) compared to all previous studies (100 ms), a factor that is well-known to modulate PPI (Braff et al., 2001; Plappert et al., 2004; Varty et al., 2001). In addition to the numerous SNPs surrounding each gene (Fig. 4a, x-axis) within the confidence interval of the PPI QTL on chromosome 17 that may affect gene expression, at least 7 genes contained non-synonymous SNPs. Adult whole brain gene expression levels of 3 genes showed marginally significant correlations with the PPI phenotype that did not withstand correction for multiple testing. Nevertheless, significant gene expression correlation may reside in specific brain regions, or at different developmental stages. None of these 10 genes or any other gene in the QTL is implicated in a mouse behavioral phenotype related to sensorimotor gating in the MGI database, or a human inherited disorder involving deficits in sensorimotor gating in the OMIM database.

QTL for attentional processing The human continuous performance task of attentional processing (CPT; Beck et al., 1956) has been adapted for rodents into the widely used 5-CSRTT (Robbins, 2002). Apart from the 5-CSRTT, as used in this study, a 5-choice CPT was developed to increase similarity to the human CPT (Young et al., 2009a), Although no differences in attentional performance were detected between C57BL/6J and DBA/2J, in line with previous reports in the 5-CSRTT and 5- choice CPT (Loos et al., 2010b; Young et al., 2009a), we observed significant differences among BXD recombinant inbred strains that transgressed beyond the

100 Sensorimotor gating and attention phenotypes of the founders. This suggested the contribution of multiple genetic loci to these phenotypes, of which we detected a significant one on chromosome 16 for response variability. We observed a strong correlation between response accuracy - the most frequently used measure of attentional processing (Robbins, 2002) - and a novel putative measure of lapses in attention in the 5-CSRTT, i.e., response variability. This correlation is not due to mathematical dependence, since response accuracy is calculated from numbers of responses, whereas response variability is calculated from latencies of these responses. Moreover, the significant chromosome 16 QTL for response variability coincided with a suggestive QTL for response accuracy. Together this supports the idea that these different measures share underlying mechanisms, which would be in line with the coincidence of reduced CPT performance and increased response variability in humans (Bidwell et al., 2007; Klein et al., 2006). Other 5-CSRTT measures that, in addition to attentional performance, may reflect motor function and/or motivation mapped significantly or suggestively to loci that did not overlap with attention QTL on chromosome 16, emphasizing the specificity of the QTL with respect to attention. Errors of omission did not correlate with response accuracy and response variability, and did not map to the chromosome 16 locus. Errors of omission may capture aspects of attentional processing, provided that other task parameters can exclude motivational deficits or motor impairments (Guillem et al., 2011; Robbins, 2002). Nonetheless, the observed positive correlation between errors of omission and magazine latencies suggests that this measure in the current study may especially reflect motivational/motor aspects of task performance. None of the 79 genes located in the significant QTL for response variability on chromosome 16 harbored non-synonymous SNPs. Although this region shows relatively few SNPs between the C57BL/6J and DBA/2J founder strains (Fig. 4b, x-axis), the ones that are present can affect gene expression. Analysis of correlation with adult gene expression data obtained from the prefrontal cortex, a region implicated in attentional performance in rodents in the 5-CSRTT (Muir et al., 1996) indicated marginally significant correlations with 4 genes, which did not withstand correction for multiple testing. Nevertheless, significant gene expression correlation may reside in other brain regions or at different

Figure 5 | No significant correlation between PPI and attention.( a) Response accuracy and response variability did not correlate with PPI at the individually titrated prepulse intensity, and (b) with conventional PPI at prepulse intensities of 65 – 75 dB. Founder lines are indicated (C57BL/6J in black, DBA/2J in grey).

101 Chapter 5 developmental stages. According to the MGI database only the gene Chrd has been associated with behavioral phenotypes in mice that could remotely be relevant for attentional performance, i.e. spatial memory (Sun et al., 2007). None of the genes in the QTL significantly associated with a human inherited disorder specifically related to attentional performance in the OMIM database.

Sensorimotor gating and attentional processing do not share genetic factors An important outcome of this study was that the significant QTL for response variability in the 5-CSRTT did not map to a locus significantly linked to PPI in our data, nor to a PPI locus in any previously published study (Hitzemann et al., 2001; Leussis et al., 2009; Liu et al., 2003; Petryshen et al., 2005; Samocha et al., 2010; Watanabe et al., 2007) except one (Hitzemann et al., 2008). However, the direction of QTL for PPI in the latter study was of the opposite effect; the B6 allele decreased PPI, while it increased attention performance in the current study (i.e., it decreased response variability). Furthermore, neither of the measures of attention correlated significantly with any measure of PPI in our study. Current theories suggest a bidirectional relation between PPI and attentional performance. Firstly, the gating mechanisms responsible for PPI may enhance attention performance by filtering out excess or irrelevant stimuli in bottom-up flow of information (Braff et al., 2001; Graham, 1975). Secondly, levels of PPI are increased when humans or rodents selectively attend to prepulse stimuli, indicative of cognitive top-down control over sensorimotor gating (Bohmelt et al., 1999; Dawson et al., 1993; Elden & Flaten, 2003; Filion et al., 1993; Thorne et al., 2005). Hence, there is a clear relation between attentional performance and PPI in tasks that employ top-down modulation of PPI (Dawson et al., 1993; Dawson et al., 2000; Scholes & Martin-Iverson, 2009, 2010). In the current study, we investigated the bottom-up relation between PPI and attention. We used passive prepulse stimuli, in contrast to rodent studies that focus on top- down modulation of PPI by pairing foot shocks with prepulse stimuli prior to the actual PPI task (Li et al., 2009). Furthermore, the short prepulse to pulse interval (SOA of 60 ms) may reduce the possibility that attentional processes modify PPI (Braff et al., 2001; Li et al., 2009), as was observed in several studies (Bohmelt et al., 1999; Dawson et al., 1993; Filion et al., 1993), but not all (Elden & Flaten, 2003; Thorne et al., 2005). Because of the use of neutral prepulse stimuli and the short prepulse-pulse interval, PPI in the current study is not expected to be influenced by top-down attentional modification. The lack of correlation between PPI scores and attentional performance therefore suggest that the bottom-up gating mechanisms involved in PPI do not contribute to attention performance in the 5-CSRTT. The absence of both a genetic and environmental correlation between the passive PPI procedure and attention are in line with studies in humans demonstrating no significant relation between measures of attention and PPI scores in passive PPI tasks (Bitsios & Giakoumaki, 2005; Greenwood et al., 2007; Swerdlow et al.,

102 Sensorimotor gating and attention

1995). In further support of this, mutant mice showing deficits in attention did not show altered PPI in passive procedures (i.e. Acnb2: Guillem et al., 2011; Drd4: Young et al., 2011). However, in a panel of 30 C57BL6J mice, Bitanihirwe et al. found a significant correlation between individual attentional performance in a 2-choice task and PPI scores in a PPI protocol that differed in the SOA used (100 ms instead of 60 ms in the current study, Bitanihirwe et al., 2011). As previously proposed (Scholes & Martin-Iverson, 2009), attention to an auditory stimulus is not absent in a passive PPI task, it is just unmeasured, meaning that mice may indeed be actively listening out to the prepulse stimuli. Only when the time interval between prepulse and pulse allows, differences in PPI scores could be explained by differences in attentional processes between individual mice. Hence, whereas our study focused on the bottom-up relation between PPI and attentional performance, the study by Bitanihirwe and coworkers may also have studied top-down attentional modulation of PPI. Furthermore, within-strain variability in inbred strains results from (largely unknown) factors in the environment of individual mice that can be specific to the strain used. Hence, the discrepancy between both studies could result from strain-specific environmental factors present in the latter study that affected C57BL/6J mice but not the large cohort of BXD mice in the current study. Taken together, the coincidence of reduced PPI and deficits in attention may not be due to modification of attentional processes by PPI, but rather by impaired top-down modification of PPI by attentional processes (Braff et al., 2001; Li et al., 2009).

In conclusion, the differences in startle sensitivity among strains in the current study warrant future PPI studies in genetically diverse populations to take potential differences in perceived prepulse salience into account. Additional mapping studies are required to reproduce and fine-map both the PPI and response variability QTL to pinpoint the causative genes. Attentional performance does not appear to be modified by sensorimotor gating. Therefore, it is worth pursuing causative genes residing in both attention and PPI QTL to better understand the separate mechanisms that contribute to diseases such as schizophrenia.

Acknowledgments This study was in part funded by the Dutch Neuro-Bsik Mouse Phenomics Consortium, supported by grant BSIK 03053 from SenterNovem (The Netherlands). The Neuro-Bsik Mouse Phenomics consortium is composed of the laboratories of A.B. Brussaard, J.G. Borst, Y. Elgersma, N. Galjart, G.T. van der Horst, C.N. Levelt, C.M. Pennartz, A.B. Smit, B.M. Spruijt, M. Verhage and C.I. de Zeeuw, and the companies Noldus Information Technology B.V. and Synaptologics B.V.

103 Chapter 5

Supplementary data

Supplementary Figure 1 | Decrease in response accuracy during a session with variable stimulus duration. (a) Decrease in response accuracy (± SEM) during the session with fixed randomly variable stimulus duration (0.25, 0.5 and 1 s) compared with the response accuracy during the baseline sessions. Response accuracy decreased in all strains, but not significantly in BXD 19, 21, 24, 27, 38, 42, 69, 90, 96 and D2B6F1. From left to right, strains are ranked on increasing response accuracy (b), as depicted in Figure 3a in the main text.

104

Supplementary Table 1 | Genes located within the confidence interval of the PPI QTL on chromosome 17.

GeneSymbol Description Non-synonymous Probe set Pearson R mutations with highest (n = 20) correlation Dlgap1 disks large-associated protein 1 isoform 2 1438098_at 0.403 mKIAA4162 Mus musculus adult male testis cDNA, RIKEN full-length enriched library, (Dlgap1) clone:4933422O14 product:DISKS LARGE-ASSOCIATED PROTEIN 1 (DAP-1) (GUANYLATE KINASE-ASSOCIATED PROTEIN) (SAP90/PSD-95-ASSOCIATED PROTEIN 1) (SAPAP1) (PSD-95/SAP90 BINDING PROTEIN 1) (FRAGMENT) homolog [Mus musculus], full insert sequence. Tgif1 homeobox protein TGIF1 isoform c 1444752_at -0.131 Mylc2b myosin regulatory light chain 12B 1428609_at 0.403 2900073G15Rik myosin light chain, regulatory B-like 1450013_at -0.269 (Myl12a) Myom1 myomesin-1 isoform 2 MRS11808222 1420693_at 0.214 MRS11808398 Lpin2 phosphatidate phosphatase LPIN2 isoform 2 1460290_at -0.223 Emilin2 EMILIN-2 precursor rs33256687 1435264_at 0.163 MRS11808658 MRS11808663 rs33550769 rs33068573 MRS11808782 mKIAA0650 Mus musculus adult male testis cDNA, RIKEN full-length enriched library, (Smchd1) clone:4931400A14 product:hypothetical protein, full insert sequence. Smchd1 structural maintenance of chromosomes flexible 1442497_at 0.348 Ndc80 kinetochore protein NDC80 homolog rs33625632 Spdya speedy homolog A (Xenopus laevis) rs33645336 mKIAA0007 Mus musculus 10 days embryo whole body cDNA, RIKEN full-length enriched library, (Wdr43) clone:2610318G08 product:KIAA0007 PROTEIN (FRAGMENT) homolog [Homo sapiens], full insert sequence. Wdr43 WD repeat-containing protein 43 1428390_at 0.025 4632412N22Rik hypothetical protein LOC320159; family with sequence similarity 179, member A 1440898_at -0.144 (Fam179a) Clip4 Mus musculus adult male thymus cDNA, RIKEN full-length enriched library, MRS11809565 1431382_a_at -0.058 clone:5830409B12 product:weakly similar to CLIP-170-RELATED PROTEIN [Homo MRS11809573 sapiens], full insert sequence.

105

Supplementary Table 1. (Continued) Alk ALK tyrosine kinase receptor precursor MRS11811254 1449987_at 0.023 MRS11811903 MRS11811905 MRS11811906 Ypel5 protein yippee-like 5 1451196_at -0.602# Lbh protein LBH 1451629_at -0.216 Lycat lysocardiolipin acyltransferase 1 1434690_at -0.004 Capn13 calpain-13 1437960_at -0.065 Galnt14 polypeptide N-acetylgalactosaminyltransferase 1453630_at -0.155 Ehd3 EH domain-containing protein 3 1417235_at 0.468# Xdh SubName: Full=Putative uncharacterized protein; 1451006_at 0.289 Srd5a2 3-oxo-5-alpha-steroid 4-dehydrogenase 2 1422960_at -0.021 Memo1 protein MEMO1 1449821_a_at 0.276 2810410M20Rik protein dpy-30 homolog (Dpy30) Spast spastin isoform 1 1443074_at 0.452# Slc30a6 zinc transporter 6 1424241_at 0.047 Nlrc4 NLR family CARD domain-containing protein 4 Yipf4 protein YIPF4 1426417_at 0.381 Birc6 baculoviral IAP repeat-containing 6 1437789_at 0.390 Ttc27 tetratricopeptide repeat protein 27 MRS11814684 Indicated is the mouse gene symbol, the description, the presence of non-synonymous SNPs, the probe set identifier on the affymetrix (UCHSC BXD Whole Brain M430 2.0 November 2006 RMA, Accession number: GN123) with the highest Pearson correlation and its correlation coefficient. #Correlation between expression in brain and the percentage of PPI is significant (P < 0.05), but does not withstand correction for multiple testing (74 correlations tested between 74 probe sets targeting the 33 genes and PPI score, Bonferroni corrected P < 6.67 * 10-4).

106

Supplementary Table 2 | Genes located within the confidence interval of response variability QTL on chromosome 16. GeneSymbol Description Probe set Pearson R with highest (n = 22) correlation A930003A15Rik Mus musculus adult retina cDNA, RIKEN full-length enriched library, hypothetical protein, full insert sequence. Klhl6 kelch-like protein 6 1437886_at 0.082 Klhl24 kelch-like protein 24 1429351_at -0.109 Yeats2 YEATS domain-containing protein 2 isoform 3 Map6d1 MAP6 domain-containing protein 1 1433630_at -0.041 Parl presenilins-associated rhomboid-like protein, 1433478_at 0.072 Cyp2ab1 cytochrome P450, family 2, subfamily ab, Abcc5 multidrug resistance-associated protein 5 1435683_a_at 0.444# Eif2b5 translation initiation factor eIF-2B subunit 1433886_at 0.066 Dvl3 segment polarity protein dishevelled homolog 1442006_at -0.278 Ap2m1 AP-2 complex subunit mu 1450894_a_at -0.215 Abcf3 ATP-binding cassette sub-family F member 3 1426748_s_at -0.252 BC052055 von Willebrand factor A domain-containing 1443369_at -0.018 (Vwa5b2) EG328644 predicted gene, EG328644 (Vwa5b2) Alg3 dolichyl-P-Man:Man(5)GlcNAc(2)-PP-dolichyl 1460742_at 0.292 Ece2 endothelin-converting enzyme 2 isoform b 1424561_at 0.096 Camk2n2 calcium/calmodulin-dependent protein kinase II 1429204_at 0.069 Ece2 endothelin-converting enzyme 2 isoform d 1424561_at 0.096 Psmd2 26S proteasome non-ATPase regulatory subunit 2 1415831_at -0.087 Eif4g1 eukaryotic translation initiation factor 4 gamma 1427036_a_at -0.272 2900046G09Rik hypothetical protein LOC78408 precursor; family with sequence similarity 131, member A 1426729_at -0.115 (Fam131a) Clcn2 chloride channel 2 1449248_at 0.078 Polr2h DNA-directed RNA polymerases I, II, and III 1424473_at -0.05 Thpo thrombopoietin precursor 1449569_at 0.269 Chrd chordin precursor 1417304_at 0.068 2310042E22Rik hypothetical protein LOC66561 1423494_at 0.443# Ephb3 ephrin type-B receptor 3 precursor 1451550_at 0.335 A830060N17 Mus musculus 10 days neonate cortex cDNA, RIKEN full-length enriched library, unclassifiable Vps8 vacuolar protein sorting-associated protein 8 1445796_at 0.125 mKIAA0804 Mus musculus 10 days neonate skin cDNA, RIKEN full-length enriched library, clone:4732480N14 product: (Vsp8) CDNA FLJ12883 FIS, CLONE NT2RP2003981 107

Supplementary Table 2. (Continued) 2510009E07Rik hypothetical protein LOC72190 1428097_at -0.117 Ehhadh peroxisomal bifunctional enzyme 1448382_at 0.260 1300002E11Rik Mus musculus adult male liver cDNA, RIKEN full-length enriched library, clone:1300002E11 product:unclassifiable, full insert sequence. AK041461 Mus musculus 3 days neonate thymus cDNA, RIKEN full-length enriched library, clone:A630012F06 product:unclassifiable, full insert sequence. Map3k13 mitogen-activated protein kinase kinase kinase Tmem41a Mus musculus 10 day old male pancreas cDNA, RIKEN full-length enriched library, clone:1810033N04 1424191_a_at -0.016 product:hypothetical Y71A12C.2 CG8408 D2013.10 5730578N08RIK T07F10.4 POSSIBLE containing protein, full insert sequence. Liph lipase member H isoform 1 1456619_at 0.357 Senp2 sentrin-specific protease 2 1451717_s_at -0.288 Smt3ip2 sentrin-specific protease 2 Igf2bp2 insulin-like growth factor 2 mRNA-binding Sfrs10 transformer-2 protein homolog beta 1419543_a_at -0.220 Etv5 ETS translocation variant 5 1420998_at -0.100 Dgkg diacylglycerol kinase gamma 1445910_at 0.272 9230117E06Rik Mus musculus adult male epididymis cDNA, RIKEN full-length enriched library, clone:9230117E06 product:unclassifiable, full insert sequence. Crygs beta-crystallin S 1420453_at -0.245 Tbccd1 TBCC domain-containing protein 1 1436752_at 0.077 Dnajb11 dnaJ homolog subfamily B member 11 precursor 1423151_at -0.099 AK149453 Mus musculus adult male liver tumor cDNA, RIKEN full-length enriched library, clone:C730014P15 product:unclassifiable, full insert sequence. Ahsg alpha-2-HS-glycoprotein precursor 1455093_a_at -0.239 Fetub fetuin-B isoform 3 1449555_a_at 0.428# Hrg histidine-rich glycoprotein 1449242_s_at 0.097 Kng2 kininogen 2 isoform 2 Kng1 kininogen-1 isoform 2 1426045_at 0.165 Eif4a2 eukaryotic initiation factor 4A-II isoform a 1450934_at -0.043 Rfc4 replication factor C subunit 4 1424321_at -0.136 Adipoq adiponectin 1447582_x_at 0.389 BC098222 Mus musculus 4 days neonate thymus cDNA, RIKEN full-length enriched library, clone:B630019A10 product:hypothetical protein, full insert sequence. C730014E05Rik Mus musculus adult male liver tumor cDNA, RIKEN full-length enriched library, clone:C730040G15 product:unclassifiable, full insert sequence. B630019A10Rik SubName: Full=BC106179 protein; SubName: Full=Putative uncharacterized protein;

108

Supplementary Table 2. (Continued) St6gal1 beta-galactoside alpha-2,6-sialyltransferase 1 1420927_at 0.404 Rtp1 receptor-transporting protein 1 Masp1 mannan-binding lectin serine protease 1 1438602_s_at 0.220 Masp3 mannan-binding lectin serine protease 1 AK144897 Mus musculus lung RCB-0558 LLC cDNA, RIKEN full-length enriched library, clone:G730049K06 product:unclassifiable, full insert sequence. Rtp4 receptor-transporting protein 4 1418580_at 0.246 Sst somatostatin precursor 1417954_at -0.076 Rtp2 receptor-transporting protein 2 Bcl6 B-cell lymphoma 6 protein homolog 1450381_a_at -0.223 1110054M08Rik Mus musculus 18-day embryo whole body cDNA, RIKEN full-length enriched library, clone:1110054M08 product:unclassifiable, full insert sequence. Lpp lipoma-preferred partner homolog isoform 1 1436714_at 0.345 A230028O05Rik Mus musculus adult male hypothalamus cDNA, RIKEN full-length enriched library, clone:A230028O05 product:unclassifiable, full insert sequence. 5430420C16Rik Tprg transformation related protein 63 regulated 1430114_at 0.147 (Tprg) Tp73l tumor protein p63 isoform Trp63 transformation related protein 63 1418158_at 0.354 p73H (Tcp1) tumor protein p63 isoform Leprel1 prolyl 3-hydroxylase 2 precursor 1442486_at 0.184 Cldn1 claudin-1 1437932_a_at 0.337 Cldn16 claudin-16 1420434_at 0.483# Tmem207 transmembrane protein 207 Indicated is the mouse genesymbol, the description, the probe set (Virginia Commonwealth University, BXD Prefrontal Cortex Saline Control M430 2.0 (Dec06) RMA Dataset, Accession number: GN135) with the highest Pearson correlation and its correlation coefficient. #Correlation between expression in prefrontal cortex and impulsive action is significant (0.01 < P < 0.05), but does not withstand correction for multiple testing (146 correlations tested between 146 probesets targeting the 79 genes and PPI score, Bonferroni corrected P < 3.4 * 10-4). None of the genes harbors non-synonymous SNPs.

109

110

Chapter 6

Insights in the genetic architecture of impulsivity in mice

Maarten Loos, Jorn Staal, Rolinka J. van der Loo, Neuro-BSIK Mouse Phenomics consortium, Danielle Posthuma August B. Smit, Sabine Spijker

In preparation

111

112

Genetic architecture of impulsivity

Abstract Impulsivity as observed in tasks of inhibitory response control is a heritable endophenotype of attention deficit hyperactivity disorder (ADHD). To identify genes underlying this variety of impulsivity we assessed inhibitory control of 43 BXD recombinant inbred strains in a five-choice serial reaction time task (5- CSRTT). QTL analysis indicated that impulsivity significantly mapped to a locus on chromosome 14. Candidate genes within QTL interval were identified by the presence of non-synonymous mutations (Ppyr1, Mmrn2, Wapal and 4930474N05Rik) and analysis of the genetic association of syntenic human genes with ADHD diagnosis (Nrg3). Impulsivity in the BXD strains did not correlate with response variability, which is a novel measure of lapses of attention and a well-replicated ADHD endophenotype. This was in line with the hypothesis that separate molecular pathways contribute to attention deficit and hyperactivity- impulsivity in this heterogeneous disorder. Future studies that identify the causative genes within the impulsivity QTL will help to understand the pathway contributing to symptoms of hyperactivity-impulsivity in ADHD.

Introduction The multifaceted construct of impulsivity is associated with several psychiatric disorders, most notably attention deficit hyperactivity disorder (ADHD; DSM- IV, 1994). Numerous studies have indicated that symptomatic levels of impulsive behavior translate into deficits in tasks of inhibitory response control (Aron & Poldrack, 2005; Bidwell et al., 2007), and that this relation is primarily genetic in origin (Young et al., 2009b). Indeed, genetic factors influence inhibitory control, with heritability estimates in the range of 18 – 50% (Kuntsi et al., 2006; Schachar et al., 2010). As such, studies that aim to identify genetic variants underlying inhibitory control are instrumental in understanding parts of the etiology of ADHD. Moreover, since it is under debate whether symptoms of hyperactivity- impulsivity and inattention in ADHD reflect the same or separate disorders (Derefinko et al., 2008; Diamond, 2005; Milich et al., 2001), it is of interest to investigate whether genetic variants underlying inhibitory control affect attentional performance. To identify quantitative genetic loci (QTLs) underlying impulsivity, we employed the panel of BXD recombinant inbred strains, which are derived from an intercross of C57BL/6J and DBA/2J (Peirce et al., 2004). Previous studies have shown that these founder strains differ in impulsivity phenotypes, suggesting that the BXD recombinant inbred strains derived thereof are powerful resources to genetically dissect impulsivity (Helms et al., 2006; Loos et al., 2010b; Loos et al., 2009; Pattij et al., 2007; Young et al., 2009a). Available databases of brain gene expression and single nucleotide polymorphisms (SNPs) of these strains aid to efficiently identify candidate genes within QTL regions (Chesler et al., 2005).

113

Chapter 6

In rodent research, several operant tasks of impulsivity exist (Evenden, 1999; Winstanley et al., 2006a). Of the subset of tasks that assesses inhibitory control, the five-choice serial reaction time task (5-CSRTT) is the most widely used paradigm (Eagle & Baunez, 2010). In this task, premature responses to a food predictive stimulus are a measure of reduced response inhibition (Robbins, 2002). In line with human studies, inhibitory control in the 5-CSRTT is influenced by genetic variation, with an estimated heritability of 34% in a genetically diverse panel of inbred stains (Loos et al., 2009). The primary measure of attention in the 5-CSRTT is the accuracy of responding to a brief light stimulus in one of five response holes (response accuracy hereafter; Robbins, 2002). In addition, this task allows extracting intra-individual variability in correct response latencies (response variability hereafter), which is a novel measure of lapses of attention and one of the best-replicated ADHD endophenotypes (Castellanos & Tannock, 2002; Leth-Steensen et al., 2000; Loos et al., 2010b; Sabol et al., 2003). Here, we investigated the relation between impulsivity (inhibitory control) and measures of attention in the same task. Correlation analyses across the recombinant inbred BXD strains (genetic correlation) with minimal influence of environmental factors (Crusio, 2006; Loos et al., 2009) were used to evaluate the genetic relation between impulsivity and attention.

Materials and Methods

Animals Parental and BXD lines were received from Jackson Lab, or from Oak Ridge Laboratory (BXD43, BXD51, BXD61, BXD62, BXD65, BXD68, BXD69, BXD73, BXD75, BXD87, BXD90), and were bred in the facility of the Neuro- Bsik consortium of the VU University Amsterdam. Seven-week-old male mice were singly housed on sawdust in standard Makrolon type II cages (26.5 cm long, 20.5 cm wide and 14.5 cm high) enriched with cardboard nesting material. After a habituation period of minimally 1 week, body weights were recorded and mice were subjected to one test of prepulse inhibition that lasted 40 min. In the subsequent week, mice were food-restricted to gradually decrease their body weight to 90 – 95% of their initial body weight before daily training in operant cages commenced (5 days each week). Water was available ad libitum throughout the experiment.

5-CSRTT Mice were trained to perform the 5-CSRTT on an individually paced schedule, as described previously (Loos et al., 2010b; Loos et al., 2009). During the first week, mice underwent 1 habituation and 4 magazine training sessions. In the next week, mice were trained to perform an instrumental response into the stimulus

114

Genetic architecture of impulsivity holes to earn a reward, and only commenced to 5-CSRTT training when they earned at least 50 rewards within one session. During 5-CSRTT training a trial started with a response of the subject into the illuminated magazine, which switched off magazine light and after an ITI of 5 s a stimulus in 1 of the 5 stimulus holes was presented for a limited duration (stimulus duration). A response in the correct stimulus hole within the limited hold time of 4 s after termination of the stimulus switched on the magazine light and delivered a food pellet. Both an incorrect response into a non-illuminated stimulus hole and an omission of a correct response resulted in a time-out period, during which all stimulus lights and the house light were turned off. When the time-out period ended, both the house light and the magazine light were switched on, and the subject could start the next trial. A premature response into a non-illuminated stimulus hole during the delay period also resulted in a time-out period, but a subsequent response into the illuminated magazine restarted the same trial. The percentage of omission errors was defined as [100  (omissions) / (omissions + number of correct and incorrect responses)]. Response accuracy was defined as [100  (number of correct responses) / (number of correct and incorrect responses)]. Impulsivity in terms of the percentage premature responses was defined as [100  (number of premature responses) / (number of omissions + correct + incorrect responses)]. In the first 5-CSRTT session, the stimulus duration was set at 16 s, which was decreased in subsequent sessions to 8, 4, 2, 1.5 and 1 s as soon as the subject reached criterion performance (omissions < 30%, response accuracy > 60%, started trials > 50) or after 10 sessions. Intra- individual variability in correct response latencies (response variability in short) was defined by the standard deviation of the correct response latencies. The total number of sessions required to reach the stimulus duration of 1 s was used as measure of required training sessions. Dependent measures were calculated from the 6th until the 10th session at stimulus duration of 1 s, and the average of these sessions was used as standard 5-CSRTT performance. In the week following the 10th session, the ITI was programmed to vary (5, 7.5 and 12.5 s), with each interval occurring an equal number of times within session. Mice were excluded from analyses when they initiated fewer than 30 trials on average or made no correct or incorrect responses during two or more standard sessions. Strains that completed fewer than 50 trials on average in combination with magazine latencies greater than 4 s, together indicative of reduced motivation, were excluded.

Data analysis For evaluation of strain differences and task manipulation, analysis of variance (ANOVA) or ANOVA with repeated measures were used, with paired Student's t-tests for post hoc testing. Estimates of the genetic effect size (narrow sense heritability) were calculated according to Hegmann and Possidente (Hegmann & Possidente, 1981) using a custom function (Microsoft Excel 2003) that takes the

115

Chapter 6 differences in the number of animals per group into account when estimating the within- and between-strain variance (Lynch & Walsh, 1998) as previously implemented (Heimel et al., 2008; Loos et al., 2009). Interval mapping analysis was performed in GeneNetwork (www.genenetwork.org) that uses the embedded MapManager software (Manly et al., 2001) to perform Haley–Knott regression. Empirical P-values, derived from 1000 permutations, were used to assess whether the peak of a QTL was statistically significant (P-value <0.05) or suggestive (on average one false positive per genome scan; genome-wide P-value <0.63; (Lander & Kruglyak, 1995). A one-LOD drop-off was used to determine the QTL confidence interval of each peak. GeneNetwork’s gene expression database of prefrontal cortex tissue (Virginia Commonwealth University, BXD Prefrontal Cortex Saline Control M430 2.0 (Dec06) RMA Dataset, Accession number: GN135) was used to calculate Pearson correlations using GeneNetwork’s online tools. For the analysis of non-synonymous mutations in genes under a QTL peak, GeneNetwork’s comprehensive SNP browser was used (date: June 2010). NCBI’s homologene datafiles (version July 13th 2009) were used to convert mouse gene coordinates (UCSC genome build 9) to syntenic human genome coordinates (NCBI human genome build 37.1), with an additional 2000 base pairs before and after each human gene. The online LiftOver tool in the USCS genome browser (http://genome.ucsc.edu/cgi-bin/hgLiftOver) was used to convert human genome coordinates from build 37 to build 36, to allow selection of SNPs in the syntenic human loci. Confirmation of gene-based association in human data was carried out using VEGAS software (version 0.8.27; Liu et al., 2010). VEGAS applies a test that incorporates information from a set of markers within a gene (or region) and accounts for linkage disequilibrium (LD) between markers by using adaptive simulations from the multivariate normal distribution. An empirical P value of 0 from 100000 simulations can be interpreted as P < 10-5, which exceeds a Bonferroni-corrected threshold at [0.05/(number of tested autosomal genes)]. The initial test for marker association with a trait was carried out using Plink software (Purcell et al., 2007) using logistic regression or a transmission disequilibrium test on the family based sample. The results from Plink were used as input for the VEGAS gene-based test analysis. Human genetic datasets informative for ADHD was retrieved from the Database of Genotyped and Phenotypes (accession number phs000016.v2.p2). The ADHD data is from the International Multicentre ADHD genetics project (IMAGE), and consists of a family-based dataset of 913 ADHD cases and 1378 healthy controls. Standard quality control procedures were carried out as described previously (Lips et al., 2011; Rizzi et al., 2011).

116

Genetic architecture of impulsivity

Results

In total, we obtained 5-CSRTT data of 588 mice (Supplementary Table 1), distributed across 39 BXD strains (on average n = 13.1 per strain, minimum n = 6 per strain), the parental lines (C57BL/6J, n = 35; DBA/2J, n = 20) and reciprocal F1 lines (B6D2F1, n = 12; D2B6F1, n = 10). For technical reasons, 6 mice were not subjected to a variable ITI.

Impulsivity of BXD mice in the 5-CSRTT Strains differed significantly in their level of impulsivity (Fig. 1a; percentage premature responses; F(42, 545) = 5.00, P < 0.001). The estimate of the genetic effect size for all 5-CSRTT parameters, as well as body weight-related parameters are shown in Table 1. Increasing the demand for inhibitory control by randomly increasing the ITI enhanced impulsivity (ITI: F(1, 539) = 884.26, P < 0.001; ITI  strain: F(42,539) = 2.39, P < 0.001) in all strains significantly (one- sample Student's t-test P < 0.05), except for BXD38 (Fig. 1b), indicating that all strains used a similar strategy to solve the task.

Figure 1 | (a) Strain differences in impulsivity, in terms of the number of premature responses in the 5-CSRTT. (b) During a session with an extended variable ITI (5, 7.5 and 12.5 s) the number of premature responses increased in all strains compared to baseline sessions with a fixed ITI of 5 s. Except for BXD38, this increase was significant for all strains.

117

Chapter 6

QTL analysis of impulsivity and identification of candidate genes A significant QTL was detected for impulsivity (Fig. 2a), with an additive effect in terms of premature responses of 2.9 %. A one-LOD drop-off confidence interval implicated 36 genes within the impulsivity QTL (Supplementary Table 2). Analysis of publicly available SNP data indicated the presence of non- synonymous mutations in 4 genes in the QTL (Fig. 2b; Ppyr1, Mmrn2, Wapal and 4930474N05Rik). In addition, analysis of adult gene expression data of prefrontal cortex tissue indicated that the correlation with a probe set for Grid1 was significant (Pearson |r| > 0.423, 0.01 < P < 0.05). However, this correlation did not withstand Bonferroni correction for multiple testing (Supplementary Table 2). We used the genes in the mouse QTL to guide a SNP analysis in the human data set. A total of 438 SNPs was selected from the GAIN dataset that were located in or around the human homologues of murine genes located within the one-LOD drop-off confidence interval of the impulsivity QTL. The SNP with the lowest P- value for genetic association with ADHD diagnosis is reported for each gene in Supplementary Table 2. After Bonferroni correction for multiple testing, the threshold for significance was set at P < 1.14 * 10-4, indicating that none of the observed association signals could withstand correction for multiple testing indicating that even the most significant SNP in Nrg3 (P = 0.0009) should be considered suggestive.

Genetic correlation of impulsivity with measures of attention Impulsivity significantly correlated (Table 1) with response accuracy (P < 0.001) but not with response variability (Fig. 3). No significant genetic correlations with

Table 1 | Genetic effect size and relation to impulsivity of other 5-CSRTT-related measures. Genetic Pearson correlation effect size with impulsivity (r) Acquisition Number of required training sessions 0.14 -0.19 Attention and impulsivity Response accuracy (%) 0.08 -0.57** Response variability (s) 0.07 0.27 Premature responses (%) 0.12 Na Other 5-CSRTT parameters Number of correct responses 0.11 -0.16 Number of incorrect responses 0.05 0.50** Mean of correct latencies (s) 0.11 -0.09 Omission errors (%) 0.11 -0.16 Mean of magazine latencies (s) 0.05 -0.19 Body weight Body weight before food restriction (g) 0.30 0.07 Relative body weight during task (%) 0.03 0.07 **Significant correlation (bold), P < 0.001. 118

Genetic architecture of impulsivity other frequently reported 5-CSRTT task parameters were detected with impulsivity, except for the number of incorrect responses (Table 1). Initial body weight before food restriction and the relative body weight (93.4 ± 0.04 % of initial body weight) during the weeks of basal 5-CSRTT training did not correlate with impulsivity, suggesting that the food regime was well-standardized across strains and did not confound observed strain differences in impulsivity.

Discussion In this study, we observed a gradual distribution of significant differences in impulsivity among BXD recombinant inbred strains that transgressed beyond the phenotypes of the founder strains, C57BL/6J and DBA/2J. This suggested the contribution of multiple genetic loci to this phenotype, of which we detected one significant locus one on chromosome 14.

Figure 2 | Significant QTL for impulsivity. The LRS score (y-axis) quantifies the relation between genomic markers (x-axis) and the trait. The threshold for significance (P = 0.05) and suggestive significance (P = 0.63) is indicated. (a) Genome-wide significance was reached at a locus on chromosome 14 for impulsivity. (b) The magnified view displays the one-LOD drop off confidence interval of the QTL. Genes harboring non-synonymous SNPs (regular font), genes of which the expression (www.Genenetwork.org) correlated with the respective trait (italics) and the gene harboring the SNP with the strongest association with ADHD diagnosis (bold) are indicated. The height of the grey markers placed on the x-axis displays the number of SNPs between C57BL/6J and DBA/2J at each locus as presented in WebQTL (www.Genenetwork.org).

119

Chapter 6

Candidate gene approach This novel QTL for impulsivity did not harbor genes that have been previously associated with impulsivity or with ADHD in general. In order to narrow down the number of candidate genes in the QTL region we took three steps. First, we searched for non-synonymous mutations in these genes in existing databases. At least three genes (Ppyr1, Mmrn2 and 4930474N05Rik) harbor mutations that result in the exchange with amino acids of different polarity or charge, potentially altering the activity of the protein. In depth analysis of the effect of these mutations could indicate whether these mutations are causal to the observed behavioral phenotype. Second, we investigated the correlation between the respective trait and the expression level of these genes in mouse adult prefrontal cortex, a brain region involved in impulsivity (Muir et al., 1996). A few marginally significant correlations were observed that did not withstand correction for multiple testing. This obviously does not exclude the possibility that the causal mutation in the QTL region has a causal effect through changing the expression level of a gene. The effect on gene expression could be located in a different brain region, a subregion of the mPFC, may be cell type specific, or exert its effect during development. Third, since increased impulsivity is a core symptom in ADHD, we looked for association of SNPs in homologous human genes with ADHD diagnosis. The SNP with the lowest probability was located in Nrg3, and, resulting from bootstrap analysis, this association should be interpreted as suggestive. Interestingly, this gene has been associated with schizophrenia (Morar et al., 2010), and recent evidence indicates that patients with schizophrenia show inhibitory control deficits in a stop signal reaction time task (Lipszyc & Schachar, 2010; Nolan et al., 2010).

Figure 3 | Impulsivity and attention (a) Impulsivity (premature responses) correlated significantly with response accuracy, the traditional measure of attention in the 5-CSRTT. (b) The coefficient of correlation between impulsivity and response variability was low and not significant (P = 0.07). 120

Genetic architecture of impulsivity

Genetic correlations among measures impulsivity and attention The negative correlation observed between impulsivity and response accuracy in the 5-CSRTT has been reported before in rats (Blondeau & Dellu-Hagedorn, 2007). Our analysis indicated that impulsivity correlated significantly with the number of incorrect responses but not with the number of correct responses, indicating that the relation between response accuracy and impulsivity depends on the propensity to make incorrect responses. This may be interpreted along a mechanism of inhibitory control. In trials in which the location of the 1 s stimulus was not detected properly, low impulsive strains may refrain from responding, whereas impulsive strains tend to make an incorrect response. As such, response accuracy does not only depend on attentional processes, but also on inhibitory control mechanisms. This may explain why impulsivity did not correlate with the other measure of attention, response variability, since the latter only takes into account correct responses and not incorrect responses. The genetic independence of impulsivity (chromosome 14), and accuracy and response variability (chromosome 16; see chapter 5) indicated that these measures are instrumental in disentangling inhibitory control and attentional processes, and symptomatic levels of impulsivity and response variability may result from different molecular pathways. Taken together, despite various bioinformatics strategies, additional experiments are required to narrow down the size of the loci (fine mapping) or that manipulate the function of the candidate genes to unequivocally identify the causative genes. Future studies, identifying these genes within the QTL will help to understand the molecular pathway that contributes to impulsivity.

Acknowledgements This study was in part funded by the Dutch Neuro-Bsik Mouse Phenomics consortium. The Neuro-Bsik Mouse Phenomics consortium was supported by grant BSIK 03053 from SenterNovem (The Netherlands). Funding support for the International Multisite ADHD Genetics (IMAGE) project was provided by NIH grants R01MH62873 and R01MH081803 to S.V. Faraone and the genotyping of samples was provided through the Genetic Association Information Network (GAIN). The dataset(s) used for the analyses described in this manuscript were obtained from the database of Genotypes and Phenotypes (dbGaP) found at www.ncbi.nlm.nih.gov/gap through dbGaP accession number phs000016.v2.p2. Samples and associated phenotype data for the International Multi-Center ADHD Genetics Project were provided by the following investigators: S. Faraone, R. Anney, P. Asherson, J. Sergeant, R. Ebstein, B. Franke, M. Gill, A. Miranda, F. Mulas, R. Oades, H. Roeyers, A. Rothenberger, T. Banaschewski, J. Buitelaar, E. Sonuga-Barke, M. Daly, C. Lange, N. Laird, J. Su, and B. Neale.

121

Chapter 4

Supplementary data

Supplementary Table 1 | Number of mice tested per strain.

Tested Excluded Baseline Variable ITI BXD1 11 0 11 10 BXD2 27 -8 19 19 BXD5 10 all BXD6 25 -1 24 24 BXD8 14 all BXD9 9 0 9 9 BXD11 14 0 14 14 BXD13 14 -3 11 10 BXD14 11 0 11 11 BXD15 23 -1 22 21 BXD16 23 -2 21 18 BXD19 7 0 7 7 BXD21 12 0 12 12 BXD23 16 0 16 16 BXD24 13 -1 12 12 BXD27 9 -3 6 6 BXD28 12 -3 9 9 BXD29 14 0 14 14 BXD31 12 0 12 12 BXD32 27 -3 24 24 BXD33 14 0 14 14 BXD34 13 0 13 13 BXD38 9 0 9 9 BXD39 14 0 14 14 BXD40 26 -5 21 21 BXD42 12 0 12 12 BXD43 15 0 15 15 BXD51 12 0 12 12 BXD55 19 -1 18 18 BXD61 10 0 10 10 BXD62 12 0 12 12 BXD65 12 -3 9 9 BXD68 12 0 12 12 BXD69 12 0 12 12 BXD73 14 0 14 14 BXD75 12 0 12 12 BXD86 13 -1 12 12 BXD87 12 0 12 12 BXD90 12 -5 7 7 BXD96 12 0 12 12 BXD97 6 0 6 6 C57BL/6J 35 -1 34 34 B6D2F1 12 0 12 12 D2B6F1 11 -1 10 10 DBA/2J 20 0 20 20 Total 654 -66 588 582 Indicated is the number of mice for each strain that entered testing in 5-CSRTT protocol, the number of mice that was excluded according to criteria, the number of mice that were used to calculate the strain mean baseline performance (SD of 1.0 s and an ITI of 5 s), and the number of mice that were used to calculate the strain performance during the session with a variable ITI (5, 7.5 and 12.5 s).

122

Supplementary Table 2 | Genes located within the confidence interval of the impulsivity QTL on chromosome 14.

GeneSymbol Transcription start site SNP id (amino acid Probe set Pearson R Human Human p-value of substitution) with highest (n = 22) Homologue SNP ADHD correlation association AK076784 Mus musculus adult male testis cDNA, RIKEN full-length enriched library, clone:4930503F20 product:unclassifiable, full insert sequence. AK019667 Mus musculus adult male testis cDNA, RIKEN full-length enriched library, clone:4930503F20 product:unclassifiable, full insert sequence. Gdf10 bone morphogenetic protein 3B precursor 1424007_at -0.215 GDF10 rs7093975 0.0278 Gdf2 growth/differentiation factor 2 precursor 1425882_at 0.215 GDF2 rs11204215 0.1010 Rbp3 retinol-binding protein 3 precursor 1457855_at 0.368 RBP3 rs2070706 0.3144 Zfp488 zinc finger protein 488 ZNF488 rs7071684 0.2128 EG432838 hypothetical protein LOC432838 AK041587 Mus musculus 3 days neonate thymus cDNA, RIKEN full- length enriched library, clone:A630023A22 product:unclassifiable, full insert sequence. Antxrl anthrax toxin receptor-like precursor Anxa8 annexin A8 1417732_at 0.345 ANXA8L1 Ppyr1 neuropeptide Y receptor type 4 rs13482131 (R/Q)# 1422271_at 0.186 PPYR1 rs51067289 (P/S)# Syt15 synaptotagmin-15 SYT15 3110001K24Rik Fam35a family with sequence similarity 35, member A 1437654_at 0.176 FAM35A rs11202332 0.3661 Glud1 glutamate dehydrogenase 1, mitochondrial 1416209_at -0.114 GLUD1 rs2296063 0.4236 2200001I15Rik Fam25c family with sequence similarity 25, member C 1437019_at 0.014 FAM25B Sncg gamma-synuclein 1417788_at -0.156 SNCG rs9420407 0.2691 Mmrn2 multimerin-2 precursor rs47039688 (G/S)# 1437123_at -0.280 MMRN2 rs34587013 0.6573 rs50165120 (L/I)# rs51174312 (D/E) Bmpr1a bone morphogenetic protein receptor type-1A 1451729_at 0.312 BMPR1A rs4934265 0.1299 9230112D13Rik 9230112D13Rik RIKEN cDNA 9230112D13 gene 1431657_at 0.212 Ldb3 LIM domain-binding protein 3 isoform e 1445121_at -0.193 LDB3 rs6586023 0.0012 Opn4 melanopsin isoform 1 1421584_at 0.335 OPN4 rs2736689 0.3252 Wapal wings apart-like protein homolog rs30252573 (A/V) 1434835_at 0.100 WAPAL rs17422850 0.0569 Grid1 glutamate receptor delta-1 subunit precursor 1445779_at 0.429* GRID1 rs2352461 0.0056 4930596D02Rik hypothetical protein LOC239036

123

Supplementary Table 2. (Continued) 4930474N05Rik hypothetical protein LOC218921 rs46605863 (H/P)# Gcap14 granule cell antiserum positive 14 isoform 2 1431972_a_at 0.388 KIAA1128 Rgr RPE-retinal G protein-coupled receptor 1422832_at 0.303 RGR rs4244948 0.0136 Lrrc21 leucine-rich repeat, immunoglobulin-like domain 1441110_at -0.152 LRIT1 rs6585848 0.1625 Lrrc22 leucine-rich repeat, immunoglobulin-like domain 1443940_at -0.025 LRIT2 rs11200925 0.2038 Pcdh21 cadherin-related family member 1 precursor 1418304_at -0.359 PCDH21 rs7099098 0.0039 AK012157 Mus musculus 10 days embryo whole body cDNA, RIKEN full-length enriched library, clone:2610528A11 product:unclassifiable, full insert sequence. BC049685 Mus musculus 10 days embryo whole body cDNA, RIKEN full-length enriched library, clone:2610528A11 product:unclassifiable, full insert sequence. Ghitm growth hormone-inducible transmembrane protein 1447504_at -0.347 GHITM rs2306321 0.0117 Nrg3 pro-neuregulin-3, membrane-bound isoform isoform 1442950_at 0.140 NRG3 rs12358612 0.0009 AK133078 Mus musculus adult male testis cDNA, RIKEN full-length enriched library, clone:4930568D09 product:unclassifiable, full insert sequence. AK007151 Mus musculus adult male testis cDNA, RIKEN full-length enriched library, clone:1700109I08 product:unclassifiable, full insert sequence. Indicated is the mouse gene symbol, the transcription start site according to genome build 37 (mm9), the presence and identity of non-synonymous SNPs, the probe set identifier on the affymatrix microarray with the highest Pearson correlation and its correlation coefficient for the 36 genes within the impulsivity confidence interval. Human SNPs were selected in homologous human genes, and for each human gene the SNP with the strongest association with ADHD diagnosis its identifier and respective p-value are displayed. Indicated are genes with SNPs that lead amino acid substitutions with different polarity or charge (#), and with significant (*) correlation between expression in prefrontal cortex and the percentage premature responses (0.01 < P < 0.05). However, the indicated expression correlations did not withstand correction for multiple testing (36 correlations tested, Bonferroni corrected P < 1.4  10-3).

124

125

126

Chapter 7

Enhanced alcohol self-administration and relapse in a highly impulsive inbred mouse strain

Maarten Loos, Jorn Staal, August B. Smit, Taco J. de Vries, Sabine Spijker

In preparation

127

128

Impulsivity, attention and alcohol self administration of BXD16

Abstract Impulsivity has frequently been associated with alcohol abuse disorder. However, it is unknown to what extent pre-existing levels of impulsive behavior predict the motivation to take alcohol or the vulnerability to relapse to alcohol seeking. BXD16, a recombinant inbred strain derived from a cross between high alcohol-drinking C57BL/6J and alcohol-avoiding DBA/2J mice, was identified in the 5-CSRTT as the strain with the highest level of impulsive action. We compared BXD16 with C57BL/6J mice in a simple choice reaction time task (SCRTT) and confirmed the impulsive and inattentive phenotype of BXD16. In a sucrose/alcohol fading self-administration procedure C57BL/6J and BXD16 did not differ in fixed ratio (FR1) responding for 10% sucrose. However, BXD16 showed increased motivation to earn 10% alcohol solution, both under FR1 and progressive ratio (PR2) schedules of reinforcement. Responding readily decreased in both strains during extinction training. Upon re-exposure to alcohol- associated cues after extinction, alcohol seeking was reinstated to a larger extent in BXD16 than C57BL/6J mice. Although additional studies are required to dissect the genetic relation between impulsivity and addiction-related behavior, this is the first study in mice suggesting that impulsivity predicts the motivation to consume alcohol as well as relapse vulnerability.

Introduction The multifaceted construct of impulsivity has frequently been associated with substance use disorders, including alcohol abuse (Verdejo-Garcia et al., 2008). Accumulating evidence from human prospective studies suggests that pathological levels of impulsivity may not only be a consequence of alcohol use, but could pre-date the development of alcohol abuse and dependence (Caspi et al., 1996; Dawes et al., 1997). In line with these observations in humans, increased levels of impulsive behavior have been observed in alcohol-naïve individuals of rat (Steinmetz et al., 2000; Wilhelm & Mitchell, 2008), as well as mouse lines (Gubner et al., 2010; Logue et al., 1998; Oberlin & Grahame, 2009) that are genetically predisposed to show a high preference/consumption of alcohol. This suggests that genetic factors that contribute to impulsive behavior may facilitate the preference or consumption of alcohol, and thereby the initiation and maintenance of alcohol use. However, whether impulsivity predicts other stages of alcohol abuse, such as the motivation and persistence to take and seek alcohol has not been investigated. Using a 5-CSRTT, an operant task of impulsive action (Winstanley et al., 2006a), we previously identified a highly impulsive mouse inbred strain among the panel of BXD recombinant inbred strains of mice (BXD16; see chapter 6; Peirce et al., 2004). This strain was substantially more impulsive than either of its founders, the DBA/2J and C57BL/6J strains of mice (chapter 5 and 6). In the current study, we investigated whether we could reproduce and extent the

129

Chapter 7 impulsive phenotype of BXD16 compared with C57BL/6J mice in another task of impulsive action, a simple choice reaction time task (SCRTT) that requires the animals to wait an unpredictable number of responses rather than a predictable amount of time. In addition to this, we investigated to what extent the SCRTT could detect differences in attentional performance, as detected previously in the 5-CSRTT for these strains (chapter 6). Subsequently, we compared BXD16 and C57BL/6J mice in an operant sucrose-fading alcohol self-administration protocol to address the question whether differences measured in impulsivity between strains are predictive of the motivation and persistence to take and seek alcohol. Specifically, we investigated the motivation to self-administer alcohol under a schedule of progressive ratio responding, as well as relapse of alcohol seeking behavior after extinction in a cue-induced reinstatement protocol (De Vries et al., 2001).

Materials and Methods

Animals Male seven-week-old mice were bred in the facility of the Neuro-Bsik consortium from the VU University Amsterdam, after receiving breeding pairs from Jackson Laboratory. Mice were singly housed on sawdust in standard Makrolon type II cages (26.5 cm long, 20.5 cm wide and 14.5 cm high) enriched with cardboard nesting material and habituated for minimally 1 week before behavioral testing commenced. In the week prior to operant training, body weights were recorded and mice were food-restricted to gradually decrease their body weight to 90 - 95% of their initial body weight before daily training in operant cages commenced (5 days each week, 30 min. per session).

5-CSRTT and SCRTT Mice (C57BL/6J n = 25; BXD16 n = 12) were trained to perform the 5-CSRTT on an individually-paced schedule, as described previously (Loos et al., 2010b; Loos et al., 2009). In a 5-CSRTT, mice were required to respond to a 1 s stimulus in one of 5 stimulus holes to earn a food pellet dispensed in the magazine at the opposite side of the cage. Impulsivity (impulsive action) was defined in terms of the percentage premature responses: [100  (number of premature responses) / (number of omissions + correct + incorrect responses)]. The two measures of attentional performance in the 5-CSRTT were response accuracy, defined as [100  (number of correct responses) / (number of correct and incorrect responses)] and response variability, defined by the standard deviation of the correct response latencies. These measures, as well as the mean and mode of correct latencies (using the half range method, Hedges & Shah, 2003) were calculated from the 6th until the 10th session at stimulus duration of 1 s, and the average of these sessions was used as standard 5-CSRTT performance.

130

Impulsivity, attention and alcohol self administration of BXD16

An independent group of mice (C57BL/6J n = 10; BXD16 n = 9) that acquired the 1 s stimulus duration limit in 5-CSRTT training were trained to perform the SCRTT. In this paradigm, the cue light in the middle of the five response holes (response hole 3) was designated as start-stimulus, and the cue light in the response hole immediately to the left or right (response hole 2 or 4, counterbalanced across strains) was designated go-stimulus. During the first 4 training sessions, a trial started with the illumination of the start-stimulus. A response into the start-stimulus hole extinguished the start-stimulus and switched on the go-stimulus. A Go-response into the go-stimulus hole switched off the stimulus and was immediately followed by distribution of a reward into the magazine. After an interval of 5 s the next trial commenced. Premature start and Go-responses in non-illuminated response holes, as well as perseverative start- responses after presentation of the go-stimulus were not rewarded but followed by a 5 s time out period during which house light and stimulus light were switched off. In subsequent sessions, responding at a variable 3 ratio (VR3) schedule into the illuminated start-stimulus hole was required to ignite the go-stimulus. The go- stimulus was only switched on for the duration of an individually-titrated limited hold (LH) period, which was set to 5 s during the first session. A Go-response during the LH period resulted in the delivery of a reward, whereas an omission of a Go-response was followed by a 5 s time out period. The percentage of omissions of Go-responses in each session was defined as follows: 100 * [omissions of Go-response / (omissions of Go-response + correct Go-responses + perseverative start-responses)]. To titrate the percentage of omissions to 30% for each subject, in subsequent sessions LH periods were decreased 0.7 fold if the percentage of omissions during the previous session was less than 25%, and increased by 1.25 fold if the percentage of omissions during the previous session was larger than 35%. Impulsivity in the SCRT was defined as the percentage of premature Go-responses, calculated as follows: 100 * [number of trials with premature Go-response / number of started trials]. Furthermore, we recorded the latencies between the onset of the go-stimulus and a Go-response into the go- stimulus hole (GoRT). Prior experiments indicated that GoRTs larger than 1.7 s are observed when mice travel to the magazine in between start and Go- responses. Therefore, GoRTs larger than 1.7 s were excluded in the calculation of mean GoRT, the mode of the GoRTs (using the half range method, Hedges & Shah, 2003) and the variability of the GoRTs (standard deviation of the GoRTs). Performance was considered stable when the mean GoRT and percentage of premature responses did not change significantly for either strain for three consequent sessions.

Operant alcohol-self administration A third group of BXD16 (n = 10) and C57BL/6J (n = 9) mice was trained in operant conditioning cages in sound attenuating chambers (TSE Systems, Bad

131

Chapter 7

Homburg, Germany) 5 days per week. Food and water were available ad libitum throughout the experiment, except during the testing session, which lasted 1 h. During the entire session a red house light provided dim illumination of the chamber. Two levers, of which one was active, were located at opposite walls of the cage. A predefined number of responses (see below) onto the active lever resulted in the delivery of 10 µl liquid reward into the receptacle located to the left of the left lever, and switched on a white stimulus light located above the receptacle for 2 s. After a time out period of 15 s, during which lever presses were without consequence but which were recorded, a red cue light above the receptacle switched on indicating the availability of the next reward. In addition to the number of rewards and active lever responses, the preference of responding onto the active lever was used to quantify behavior, calculated as follows: [number active lever responses / (number active lever responses + number inactive lever responses)]. Sucrose fading, FR1 and PR2 responding for alcohol. Mice were trained to respond for a 10% alcohol solution using a sucrose fading protocol at a fixed ratio 1 (FR1) schedule of reinforcement, in which reward consisted of the following solutions (wt/vol in tap water): 10% sucrose (S1-S9), 10% sucrose and 2% alcohol (S10-S13) 10% sucrose and 4% alcohol (S14-S15), 10% sucrose and 6% alcohol (S16-S17), 10% sucrose and 8% alcohol (S18-S20), 10% sucrose 10% alcohol (S21-S25), 5% sucrose and 10% alcohol (S27-S32) and finally 10% alcohol (S33-S38). In the subsequent 5 sessions, mice were subjected to a progressive ratio (PR2) schedule. During these sessions the response requirement is increased by 2 after each reward delivery. The breakpoint was determined in the 5th en last session, and defined as the last completed ratio. Extinction and reinstatement of alcohol seeking behavior. After PR2 sessions, mice received 6 sessions of FR1 training for 10% alcohol. During the subsequent 20 sessions of extinction training responding on the previously active lever was without programmed consequences. During the last training session reinstatement of alcohol seeking behavior was determined in response to presentation of alcohol conditioned cues: cue lights in response to active lever responses and a droplet of 10% alcohol placed in the receptacle at the start of the session.

Data analysis For the evaluation of strain differences, analysis of variance (ANOVA) was used with ‘strain’ as between-subjects factor. The effects of task manipulations across different sessions were analyzed using ANOVA with repeated measures with ‘session’ as within-subjects factor. When Mauchly’s test for sphericity of data was significant, more conservative Huynh–Feldt corrected degrees of freedom and related probability values were reported. Where appropriate, post hoc tests were performed with Student's t-tests for ‘strain’ effects and paired Student's t- tests for ‘session’ effects. All data are depicted as means ± standard error of the

132

Impulsivity, attention and alcohol self administration of BXD16 mean (SEM), and the level of significance was set at P < 0.05. All analyses were performed using the Statistical Package for the Social Sciences for Windows version 15.0 (SPSS, Chicago, IL, USA).

Results

5-CSRTT After reaching the baseline stimulus duration of 1 s in our standard 5-CSRTT training protocol, BXD16 mice were found to exhibit a higher impulsivity in terms of the percentage of premature responses (Fig. 1a; strain: F(1,35) = 29.87, P < 0.001) compared with C57BL/6J mice. Attentional performance of C57BL/6J mice was superior compared to BXD16 mice, with a significantly higher response accuracy (strain: F(1,35) = 13.17, P < 0.001; data not shown) and lower response variability (Fig. 1c; strain: F(1,35) = 7.30, P < 0.05). Analysis of the distribution of correct response latencies (Fig. 1b) indicated that BXD16 mice were slower in terms of the mean (strain: F(1,35) = 12.20, P < 0.01) and mode (strain: F(1,35) = 12.50, P < 0.01) of the correct response latencies (Fig. 1c).

SCRTT An independent group of C57BL/6J and BXD16 mice was trained to perform the SCRTT. Performance stabilized after 10 training sessions, and data of the subsequent sessions was taken for average SCRT performance. BXD16 mice were more impulsive in terms of percentage of premature Go-responses during presentation of the start stimulus compared with C57BL/6J mice (Fig. 1d; strain: F(1,17) = 6.87, P < 0.05). Although the required limited hold time was significantly longer for BXD16 mice (strain: F(1,17) = 8.85, P < 0.01), their mean GoRTs were not different from C57BL/6J mice (Fig. 1e; strain: F(1,16) = 0.48, n.s.). This was explained by the observation that BXD16 mice had a faster mode of GoRTs (Fig. 1f; strain: F(1,16) = 4.72, P < 0.05) but also a larger GoRT variability (strain: F(1,16) = 8.73, P < 0.01). No significant differences were found in the number of initiated trials (strain: F(1,17) = 3.40, n.s.) and the percentage of omission of Go-responses (strain: F(1,17) = 3.87, n.s.).

Alcohol self administration Sucrose fading, FR1 and PR2 responding for alcohol. There was a significant effect of strain on the number of earned rewards across the first 37 sessions (session: F(36,612) = 21.72, P < 0.001; strain  session: F(36,612) = 5.4, P < 0.001; strain: F(1,17) = 7.53, P < 0.05). Post hoc testing revealed a significant difference during the first session of sucrose acquisition. Analysis of the first 9 sucrose sessions revealed only a trend for session  strain interaction (P = 0.08), and development of strain differences starting from the fading step of 10% sucrose with 8% alcohol and onwards (Fig. 2a). There was a significant

133

Chapter 7

Figure 1 | Consistent differences in impulsivity and attention between BXD16 and C57BL/6J mice in two operant tasks. (a, d) BXD16 mice were more impulsive in terms of the number of premature responses in the 5-CSRTT (A) and SCRTT (d). (b, e) Analysis of the distribution of responses in the 5-CSRTT (b) and SCRTT (e) indicated that (c, f) C57BL6J mice showed lower response variability than BXD16, indicative of fewer lapses in attention in both tasks. * P < 0.05.

134

Impulsivity, attention and alcohol self administration of BXD16 difference in the development of the preference of the active over the inactive lever between stains (session: F(36,612) = 4.87, P < 0.001; strain  session: F(36,612) = 2.18, P < 0.001; strain: F(1,17) = 0.01, n.s.). Post hoc testing revealed an increased preference in BXD16 mice only during the first session of sucrose acquisition (Fig. 2b), suggesting a faster development of the dissociation between active and inactive lever in these mice. Similar effects were found on the number of active lever responses, which include non-rewarding active lever responses during the time out (not shown; session: F(36,612) = 11.02, P < 0.001; session  strain: F(36,612) = 4.06, P < 0.001; strain: F(1,17) = 7.07, P < 0.05). Again, post hoc testing revealed a significant difference during the first session, and the development of strain differences starting from the fading step of 10% sucrose with 8% alcohol and onwards (not shown). There was a significant session  strain effect with respect to the number of inactive responses (session: F(36,612) = 1.68, P <0.01; session  strain F(36,612) = 2.03, P < 0.001). However, post hoc testing indicated that BXD16 mice only showed increased inactive lever responses during two fading steps (5% sucrose with 10% alcohol, 5% sucrose with 10% alcohol), but not during the FR1 sessions with 10% alcohol. To test the motivation to receive a 10% alcohol reward, mice were tested in a PR2 schedule. During the 5th session of PR2 training, BXD16 mice reached a higher breakpoint (i.e., last completed ratio) than C57BL/6J mice (Fig. 3a; strain: F(1,17) = 20.01, P < 0.001), in line with a significantly elevated number of active lever responses (F(1,17) = 16.85, P < 0.01). No difference was detected in the preference of the active versus inactive lever (Fig. 3b; strain: F(1,17) = 0.71, n.s.) and in inactive lever responding (F(1,17) = 2.84, ns; not shown). During the six FR1 sessions following PR2 training, BXD16 mice again displayed higher alcohol self-administration behavior in terms of earned rewards compared with C57BL/6J mice (strain: F(1,17) = 9.18, P < 0.01).

Extinction and reinstatement of alcohol seeking behavior. Over the course of the last FR1 session and 20 subsequent extinction sessions, alcohol-seeking behavior in terms of the number of responses on the previously active lever decreased (Fig. 4a; session: F(20,320) = 8.57, P <0.001). No difference in the rate of extinction or preference of the active lever between groups was detected (session  strain: F(20,320) = 1.54, n.s.). The absolute number of previously active lever responses remained significantly higher in BXD16 (strain: F(1,16) = 17.3, P < 0.001). Similar to extinction of the active lever responses, there was a significant decrease in the preference of the active over the inactive lever in both strains (Fig. 4b; session: F(20,320) = 3.36, P < 0.001; session  strain: F(20,320) = 1.06, n.s.). No difference between strains was detected in the preference for the previously active lever (strain: F(1,16) = 1.64, n.s.). The number of responses on the previously inactive lever showed an interaction effect (session: F(20,320) = 2.31, P < 0.01; session  strain: F(20,320) = 3.10, P < 0.001). Post hoc testing

135

Chapter 7

Figure 2 | Enhanced alcohol seeking in BXD16 mice at an FR1 schedule. During the first sessions an active lever response was rewarded with 10 l sucrose solution (10%). This reward was faded towards an alcohol solution (10%) after 32 sessions. (a) A strain difference in obtained rewards developed during fading in of alcohol that was persistent with a 10% alcohol solution. (b) Except for the first session, strains did not differ in the preference for the active lever over the inactive lever, indicating a similar capacity to dissociate the active from the inactive lever. * P < 0.05.

136

Impulsivity, attention and alcohol self administration of BXD16 indicated that, whereas the number of these responses did not differ during the last FR1 session, inactive lever responding increased in BXD16 from the second extinction session onwards (not shown). Next, sensitivity to alcohol-conditioned cues was tested in a reinstatement session. In comparison to the last extinction session, the number of responses on the previously active lever increased significantly (session: F(1,16) = 25.18, P < 0.001) upon presentation of alcohol- conditioned cues during the reinstatement session (C57BL/6J: from 52.9 ± 9.5 to 85.8 ± 13.4 responses; BXD16: from 106.7 ± 17.0 to 207.2 ± 37.1 responses). This effect was strain dependent (session  strain: F(1,16) = 6.47, P < 0.05), indicating that the relative increase in number of responses on the previously active lever in BXD16 (one-way Student's t-test P < 0.01) was larger than in C57BL/6J (one-way Student's t-test P < 0.05; Fig. 4c). During the reinstatement session, the preference of the previously active lever over the inactive lever increased in both strains (Fig. 4d; session: F(1,16) = 4.64, P < 0.05; session  strain: F(1,16) = 0.47, n.s.). The number of previously inactive lever responses also increased (session: F(1,16) = 9.30, P < 0.01), but in contrast to active lever responses, no significant session  strain interaction was detected (session  strain: F(1,16) = 0.34, n.s.; data not shown).

Discussion

Consistent strain differences in impulsivity and attention Our previous studies (chapters 5 and 6) indicated that BXD16 is the most impulsive and least attentive strain among the panel of BXD recombinant inbred strain as assessed in a 5-CSRTT. In the current study, we compared the performance of C57BL/6J and BXD16 mice in a novel SCRTT. In this SCRTT, the onset of a go-stimulus was rendered unpredictable by randomly varying the

Figure 3 | Enhanced motivation of alcohol self-administration in BXD16 mice at a PR2 schedule. (a) BXD16 mice reached a higher last completed ratio of responding to obtain 10 l alcohol solution (10%), indicative of enhanced motivation to self-administer alcohol. (b) After adaptation to the procedure, strains did not differ in the preference for the active lever during the 5th session at PR2 schedule of reinforcement, indicating a similar capacity to dissociate the active from the inactive lever, as observed during the sucrose-fading paradigm. * P < 0.05. 137

Chapter 7 number of start-responses required to initiate a go-stimulus. This required mice to inhibit prepotent Go-responses for an unpredictable amount of time. Compared with C57BL/6J, BXD16 mice made more premature Go-responses. This extends the observation that BXD16 mice make more premature responses in the 5- CSRTT, in which mice need to inhibit responding during fixed inter-trial interval time of 5 s. As such, we can conclude that BXD16 mice have a reduced ability to inhibit responding in situations, in which waiting is required, whether or not the duration of this waiting period is predictable. In comparison to C57BL/6J mice, BXD16 mice showed increased intra- individual variability in Go-response latencies. This can be explained by an increased frequency or duration of brief lapses in attention during which the onset of a go-stimulus is not detected by BXD16 mice. This observation is in line with previous observations of low attentional performance in the 5-CSRTT, in terms of low choice accuracy and large intra-individual variability (chapter 5). Together, these data clearly indicate reduced attentional performance of BXD16 mice compared with C57BL/6J mice. We detected a faster mode of response latencies in BXD16 mice in the SCRT in the current study. It has been suggested that the mode of response latencies can be taken as measure of sensorimotor processing speed (Leth-Steensen et al., 2000; Sabol et al., 2003), suggesting that BXD16 mice have faster sensorimotor processing speed. The contrasting finding that C57BL/6J mice have a faster mode of response latencies in the 5-CSRTT (chapter 5) may be explained by two procedural differences between 5-CSRTT and SCRTT that have an impact on movement time and timing ability. In the 5-CSRTT, mice have to travel to a stimulus in one of the 5 holes spaced along one side of the operant cage, while in the SCRTT only a small movement from the middle hole to the adjacent hole is required. This is reflected by the markedly lower mode of the response latencies in the SCRTT in the current study compared to the 5-CSRTT. Hence, strain difference in movement speed would affect performance in the 5-CSRTT to a larger extent, which could explain inconsistent strain difference in the mode of response latencies between 5-CSRTT and SCRTT. Secondly, only in the 5- CSRTT mice can use a timing strategy to anticipate the onset of the stimulus after the 5 s ITI. It is possible that C57BL/6J mice have a superior timing ability that would contribute to a shorter mode of responding in the 5-CSRTT, but not in the SCRTT. Taken together, the comparison of the mode of response latencies in the SCRT and 5-CSRTTs suggests that C57BL/6J mice may have faster movement times or better timing ability, while BXD16 mice have faster processing speed.

Strain differences in alcohol-seeking behavior During the first self-administration session, when reward consisted of a 10% sucrose solution, BXD16 mice made more responses on the active lever, earned more rewards and had a higher preference for the active lever than C57BL/6J

138

Impulsivity, attention and alcohol self administration of BXD16

Figure 4 | Stronger reinstatement of alcohol-seeking in BXD16 mice. Compared to the last FR1 session, responding at the previously active lever (a) and the preference for the previously active lever (b) decreased during extinction training in both strains (paired t-tests for comparison between last FR1 and each subsequent extinction session). Upon exposure to alcohol-conditioned cues (c), a larger increase in alcohol-seeking behavior in BXD16 than in C57BL/6J mice in terms of the increase in the number of responses on the previously active lever compared with the last extinction session. (d) Exposure to alcohol-conditioned cues after extinction increased the preference of responding on the previously active lever indicative of alcohol-seeking. * P < 0.05, ** P < 0.01.

139

Chapter 7

mice. This suggests that BXD16 mice were slightly faster in acquiring instrumental responding. No strain difference in the number of rewards was detected in each of the following 8 sessions with 10% sucrose reward, suggesting that BXD16 and C57BL/6J did not differ in motivation to consume sucrose. Also, no differences in number of inactive lever responses were detected, indicating that there was no evidence for a higher general activity of BXD16 mice. During the sucrose fading procedure, a significant strain difference in number of active lever responses and earned rewards developed, starting when reward consisted of 10% sucrose mixed with 8% alcohol. The absence of a concomitant change in the preference for the active lever, and the absence of differences in number of inactive lever responses during the sessions with 10% alcohol solution indicate that the higher number of active lever responses of BXD16 was not merely a higher general activity. The FR1 data suggested that BXD16 mice were more motivated to self- administer alcohol. The latter was confirmed when animals were subjected to the progressive ratio (PR2) procedure for 10% alcohol reward. PR schedules provide a measure of motivational strength of the drug (Mendrek et al., 1998). Compared with C57BL/6J mice, the BXD16 strain reached higher breakpoints in the absence of a difference in the preference for the active lever, indicative of a stronger motivation to self-administer 10% alcohol reward. In the subsequent 20 extinction sessions, the strains did not differ in rate at which responding on the previously active lever decreased, suggesting a similar extinction rate of alcohol-seeking behavior. Similarly, the preference for the previously active lever decreased at the same rate in both strains. However, the number of active lever responses remained significantly higher in BXD16 during extinction sessions and the number of inactive lever responses reached significantly higher levels than C57BL/6J mice from the second extinction session onwards. Such a general activity difference between C57BL/6J and BXD16 was not observed earlier in the operant protocol, and has not been observed before in tests of locomotion in novel environments (Yang et al., 2008). Therefore it is possible that this general activity difference between strains only appears after prolonged periods of habituation, i.e., 50 sessions of training in the operant chambers in the current experiment. Both strains showed reinstatement of alcohol-seeking behavior upon exposure to alcohol-associated stimuli, reflected in responding on, and preference for, the previously active lever. Importantly, statistical analysis indicated that this increase was strain-specific, and a more robust increase was detected in BXD16 than in C57BL/6J mice. C57BL/6J mice are known to show voluntary alcohol intake and alcohol metabolism similar to, or even higher than BXD16 (Crabbe et al., 1983; Grisel et al., 2002; Phillips et al., 1994; Rodriguez et al., 1994). When placed in the context of the concept that different factors predict different stages in the

140

Impulsivity, attention and alcohol self administration of BXD16 development of addiction (Kreek et al., 2005), these observations do not contradict the observation of enhanced alcohol-seeking in BXD16 mice in the current study. A sucrose fading protocol may overcome an initial aversive effect of alcohol in low-alcohol drinking rodents, and enhance subsequent voluntary alcohol intake in the home cage (Tolliver et al., 1988). Therefore, a fading protocol may facilitate the intake of alcohol to levels that provide enough reinforcement to initiate and maintain alcohol-seeking behavior. Thus, in comparison to C57BL/6J, high impulsive BXD16 mice have an equal or lower propensity to initiate alcohol self-administration, but once alcohol self- administration is established, they show enhanced alcohol-seeking behavior, even after a prolonged period of extinction training. The oral consumption of alcohol is accompanied by chemosensory perception of its flavor, which plays an important role in its acceptance and rejection. Like humans, rodents depend on the possibility to detect the sweet (sucrose-like) and bitter (quinine-like) taste of alcohol. BXD16 mice showed the least voluntary consumption of a bitter quinine solution of a panel of BXD strain including the parental lines (Phillips et al., 1991), indicating a good perception of bitterness, as well as a low-moderate voluntary consumption of saccharine (Lush, 1989). This indicates that enhanced alcohol self-administration cannot be explained by a reduced perception of bitterness or increased perception of sweetness of alcohol in the BXD16 strain. In conclusion, this is the first study in rodents linking impulsive action to the enhanced motivation to self-administer alcohol, as well as to the stronger reinstatement of alcohol-seeking upon exposure to alcohol-associated stimuli after extinction. Our results extend previous studies in rodents indicating that impulsive action measured in Go-NoGo tasks (Gubner et al., 2010; Logue et al., 1998) and impulsive choice in a delayed reward paradigm (Oberlin & Grahame, 2009; Poulos et al., 1995; Wilhelm & Mitchell, 2008) coincide with the consumption, preference and sensitizing effects of alcohol. In general, these data are in line with the association between impulsivity and drug self-administration behavior in rats and substance abuse in humans (Belin et al., 2008; Diergaarde et al., 2008; Verdejo-Garcia et al., 2008; Winstanley et al., 2010). Although the current data point towards a common mechanism underlying impulsivity and addiction-related behavior in mice, we underscore that the association of impulsivity and addiction-related behavior may depend on separate genetic loci that are both present among BXD16 and C57BL/6J mice. Studies that identify genetic loci underlying impulsivity and determine their effect on addiction- related behavior are currently underway.

Acknowledgments This study was in part funded by the Dutch Neuro-Bsik Mouse Phenomics consortium, supported by grant BSIK 03053 (SenterNovem, The Netherlands).

141

Chapter 7

.

142

Chapter 8

General discussion

143

Chapter 8

1. Discussion

1.1. Background Of most psychiatric diseases the etiology is at best poorly understood, which hampers the development of adequate treatments. Psychiatric diseases are clinically diagnosed based on the presentation of a collection of behavioral symptoms. Among these symptoms, deficits in inhibitory control and attention are prominent features and are observed in multiple psychiatric diseases, such as schizophrenia, ADHD and drug addiction. The aim of this thesis was to identify genes and molecular processes that contribute to inhibitory control and attention, to provide entry points for understanding and treatment of the frequently observed deficits they underlie. Because inhibitory control and attention are complex cognitive functions, they rely on various brain systems and circuitries therein. They contain a substantial genetic component, as is clear from human genetic studies (Bidwell et al., 2007). Due to this overall complexity they are considered ‘complex traits’. In order to identify genes underlying these traits, adequate measures of these functions need to be available. As described in more detail in chapter 1, to study inhibitory control and attention in humans, specific computerized quantitative response tasks have been developed. These well- defined behavioral measures of brain function represent phenotypes that are intermediate to genome and the individual’s behavior, and may therefore be influenced by only a limited number of genetic factors. As such, they facilitate gene hunting (de Geus, 2002). Importantly, several of these response tasks are readily adaptable for use with rodents, providing an entry into molecular (genetic) dissection, as described in this thesis.

1.2 Exploiting individual variation to specify the role of dopamine-related genes in impulsive choice behavior Human genetic studies have identified a number dopamine-related genes associated with psychiatric disorders, most notably ADHD (Kreek et al., 2005; Waldman & Gizer, 2006). Nonetheless, genetic studies in humans do not easily provide insight into the temporal and spatial aspects of aberrant gene function; this requires studies in model organisms. Genetically modified mice are powerful tools to investigate the involvement of specific genes in a behavioral phenotype, such as impulsivity or attention, based on a priori knowledge on the function of the respective gene. By testing mutant mice in the 5-CSRTT, several genes have been implicated in impulsivity and/or attention (Bailey et al., 2010; Davies et al., 2007; Guillem et al., 2011; Hoyle et al., 2006; Lambourne et al., 2007; Relkovic et al., 2010; Romberg et al., 2011; Young et al., 2007; Young et al., 2004). However, finding novel genes, not previously implicated in impulsivity or attention requires other strategies than targeted mutagenesis. In this thesis individual variation in cohorts of outbred rats

144

General discussion

(chapter 2) and variation among (recombinant) inbred mouse strains was exploited for gene hunting (chapters 4–6). In chapter 2, we aimed to further specify the role of dopamine related genes in impulsive choice, an important intermediate phenotype in ADHD (Barkley et al., 2001; Solanto et al., 2001; Sonuga-Barke, 2002; Sonuga-Barke et al., 1992). Individual differences in levels of impulsive choice among rats were shown to correlate with variation in gene expression of Drd1, Drd5 and Calcyon, specifically in the mPFC. Subsequent pharmacological intervention showed for the first time that signaling through dopamine D1/D5 receptors was functionally involved in impulsive choice behavior. In conclusion, this rodent study identified Drd1, Drd5 in the mPFC to be involved in impulsive choice, and in addition it specified the brain region in which these genes may lead to impulsive choice deficits in ADHD. However, implicating novel genes that have not been associated with impulsivity and/or attention before requires another approach, such as the mapping of QTL in this study.

1.3 Evaluating mouse tasks of impulsivity and attention for QTL analyses QTL mapping studies require large cohorts of subjects - in our case BXD strains - that are typically tested in serial batches across multiple months. In some behavioral assays, differences between strains of mice are poorly reproducible, especially in cases where strain differences are subtle (i.e., when the genetic effect size is low; Crabbe et al., 1999). For the purpose of a QTL study that requires multiple months of experiments, it is important to select a robust task that leads to reproducible results. Operant tasks have the advantage that training can be performed systematically according to objective criteria programmed into an automatic analysis of training results, which enhances the consistency of training across multiple months. Furthermore, handling of animals, which is thought to contribute to idiosyncratic results, is limited to introduction and removal of animals in and from operant boxes thereby increasing reproducibility. Two operant tasks measuring aspects of impulsivity and attention analogous to computerized response tasks for humans were reported for use with mice, i.e., a Go/No-Go task (Logue et al., 1998; McDonald et al., 1998) and 5-CSRTT (Humby et al., 1999). We therefore compared the 5-CSRTT with Go/No-Go task in the context of a QTL study. We detected promising strain differences between the founder strains of the BXD lines (chapter 3) in both 5-CSRTT and Go/No-Go tasks. In the 5-CSRTT, we described behavioral effects of amphetamine in mice similar to those observed in rats (Cole & Robbins, 1987; Harrison et al., 1997; van Gaalen et al., 2006a) indicating that dopamine modulates inhibitory response control in this task as expected. Other neurotransmitter systems involved in inhibitory response control in this paradigm are those using GABAA, NMDA and serotonin 5-HT2 receptor ligands (Fletcher et al. 2007, Oliver et al. 2009). In contrast, although

145

Chapter 8 psychostimulants are known to affect inhibitory control in the Go/No-Go task in human studies (Trommer et al., 1991; Vaidya et al., 1998), we did not find such an effect in the Go/No-Go task, meaning that this murine task may be less well- suited as translational assay. In further support of the 5-CSRTT, several studies genetically validated this task using mouse models of neurological disease that are characterized by deficits in attention and/or inhibitory control (Davies et al., 2007; Lambourne et al., 2007; Relkovic et al., 2010; Romberg et al., 2011). Furthermore, the nicotinergic system is critically involved in attentional processes (Mansvelder et al., 2006) and mice lacking subunits of nicotinic receptors show deficits in attention (Bailey et al., 2010) (Guillem et al., 2011; Hoyle et al., 2006; Young et al., 2007; Young et al., 2004). The differences in impulsivity between C57BL/6J and DBA/2J mice in the 5- CSRTT detected in chapter 3 could be replicated in chapter 4. In addition, we detected substantial heritability for 5-CSRTT parameters, suggesting that alleles segregate in the panel of BXD recombinant inbred lines that affect 5-CSRTT performance. Taken together, the 5-CSRTT appeared to be a good starting point for a QTL study for impulsivity and attention phenotypes in the BXD RI resource.

1.4 Discovering genes for impulsivity and attention using the BXD resource In chapters 5 and 6, we observed a strong transgression in terms of impulsivity and attention phenotypes among BXD lines outside the behavioral range covered by the two parental lines. Together with the gradual distribution of these phenotype, this suggested the action of multiple genetic loci to these phenotypes, indicating that even these well-dissected behavioral phenotypes are genetically complex. Nonetheless, we detected QTL for attention (response variability) and impulsivity (premature responses), indicating that dissection of complex traits into translational behavioral phenotypes (chapter 1, Fig. 1) was a successful approach. These QTL showed striking specificity. Impulsivity and attention, of which deficits co-occur in ADHD, mapped to different loci. This is underscoring the relevance of intermediate phenotypes in genetic studies, in which a combined score of impulsivity and poor attention may have obscured the detection of these distinct QTL. The advantage of intermediate phenotypes is once more illustrated by the fragmentation of the response latency measure into its mode (also interpreted as sensorimotor processing speed; Sabol et al., 2003) and standard deviation (i.e., response variability). Respective traits mapped to QTL at different locations on chromosome 16, showing that even relatively simple measures, such as response latency, are complex traits that can be further dissected in order to identify underlying genetic loci. Prepulse inhibition is a presumed pre-attentive phenomenon that is affected in Schizophrenia (for review see Braff et al., 2001), and hence was of interest in the context of studies on attentional performance in this thesis. Nonetheless,

146

General discussion measures of PPI were genetically independent from measures of attention in the 5-CSRTT, as shown by lack of genetic correlation and separate QTL, indicating that lower levels of attentional performance are not caused by deficits in PPI in the BXD panel. The QTL for PPI therefore harbors genetic variants specifically influencing PPI and not attentional performance in general, and for that reason is of relevance to Schizophrenia but not to ADHD and addiction. Promisingly, none of these QTL harbored genes previously implicated in attention and impulsivity, suggesting that these QTL contain one or more novel genes that may help us to better understand these phenotypes. The power of previous human genome wide association studies (GWAS) on ADHD and Schizophrenia was insufficient to detect significant SNPs. However, the synergism between human GWAS and mouse QTL analyses pinpointed a limited set of genes in the QTL that are the most promising ones for follow up studies (i.e., Nrg3, Ppyr1, Mmrn2, Wapal and 4930474N05Rik). Each of the detected QTL only explained a fraction of the variation observed among strains. This indicates there were multiple genetic loci with presumable lower effect size that we did not detect, hence confirming the idea that impulsivity and attention phenotypes are complex traits. At least two factors explain why we did not detect those QTL. First, the lower the additive effect of a QTL, the larger the sample size required to detect a significant QTL. For reasons of logistics, we were able to test approximately 40 BXD lines in this study. According to simulation studies this results in a power (power = 0.80, p < 0.0001) to reliably detect QTLs accounting for ~40% of the strain variance (Belknap et al., 1996; Peirce et al., 2004). Second, another complicating issue is the effect of genetic interactions, or epistasis. Testing for genetic interactions requires testing of epistatic interactions between all pairs of loci in the genome. This inflates the number of statistical tests and therefore requires even higher numbers of strains than used for single QTL. Obviously, the number of strains in the current studies was insufficient to detect these interactions.

2. Perspectives

2.1. Novel behavioral assays to measure mouse cognition at higher throughput The power to detect QTL, let alone epistatic interaction between QTL, was limited in the current study. For reasons of limited recombination rates, it is unlikely that all genes contributing to a trait (quantitative trait genes; QTG) can be detected in the BXD panel. The current panel of BXD strains contains approximately 4100 recombinations (Peirce et al., 2004). If these recombination events would be distributed evenly across the 2.5 Gb mouse genome (Waterston et al., 2002), then every recombination would occur approximately every 600 Kb. In case a region in between two recombination events contains two QTG

147

Chapter 8 with an opposite additive effect, those QTG would not be detected. Recombinant inbred strains from an eight-way cross between genetically distant founders (the collaborative cross; CC) will yield higher mapping precision, enabling the detection of these neighboring QTG (Chesler et al., 2008). However, assessing well-dissected phenotypes, such as impulsivity and attention in operant tasks, requires a vast amount of labor as compared to for instance anxiety and certain learning and memory phenotypes. Therefore, increasing power using more BXD strains or extending to CC lines is a challenging project unless novel less-labor intensive assays will be developed. Another challenge would be to delineate the underlying genes related to the decision-making counterpart of impulsivity, namely impulsive choice. One frequently employed index to measure impulsive choice is the inability to tolerate delay of reinforcement. In contrast to available assays to study this aspect of impulsivity in rats, to date only one laboratory succeeded in setting up a paradigm for impulsive choice in mice using a within-session increase in delay to reinforcement (Isles et al., 2004; Isles et al., 2003) similar to our paradigm in rats (Loos et al., 2010a). Despite effort, we did not succeed in setting up a within session protocol in mice, as well as three different other labs (personal communication with A. Holmes, T. Kalenscher and F.S. van den Bergh). In our hands, mice were not able to flexibly shift choice preference between large or small reward within one session, but habitually chose one of the two throughout a session. Numerous studies reported delayed reward tasks in mice that use an across-session increase in delay (Adriani & Laviola, 2003; Brunner & Hen, 1997; Oberlin & Grahame, 2009; Wilhelm & Mitchell, 2008). It is possible that mice solve these across session tasks by slowly shifting from one habit (responding for large reward) to another (responding for small reward), which not necessarily reflects the same flexibility displayed by rats, when shifting preference within one session. This may suggest that a task requiring this type of flexibility is cognitively challenging for mice. Yet, this type of executive function, cognitive flexibility, is essential for adequate behavior and is impaired in many psychiatric and neurological diseases. It may take a more potent reward (e.g. sweet condensed milk, Isles et al., 2003) to successfully implement delayed reward tasks, or other inventive assays to assess cognitive flexibility, tailored to the species under investigation. One direction to improve throughput and develop novel tasks of impulsivity, attention and cognitive flexibility is by exploiting the home cage environment of mice. The training intensity can be increased from 25 min in classical operant tasks to theoretically all awake hours in automated home cages to intensify the training. Automatic control over food and water availability would enable the experimenter to administer the tasks at the time of day when animals are most willing to earn food or liquid reward to enhance the motivation to acquire even the most cognitively demanding tasks. Initial steps in this direction have been taken by administering a simple reaction time task to group-housed female mice

148

General discussion in an automated home cage (Vannoni et al., 2010). This task measured aspects of impulsivity and attention and was acquired by mice in less than 2 weeks time, exemplifying the improvement in phenotyping efficiency by applying tasks in a home cage environment.

2.2. Towards a better molecular and cellular understanding of impulsivity and attention using new mouse lines Drawing parallels between behavioral phenotypes and brain phenotypes within one individual or strain is a powerful and frequently employed method to understand behavior and the etiology of associated disease. In this thesis, cohorts of individual outbred rats with extreme phenotypes (chapter 2) and panels of (recombinant) inbred strains of mice (chapters 3–7) have been instrumental in this respect. A third method, exploiting two strains of rodents with extreme phenotypes provides a convenient alternative. In chapter 7, we identified a BXD strain (BXD16) that was more impulsive in both the 5-CSRTT and a simple choice reaction time task, which also showed enhanced alcohol-seeking behavior in terms of higher motivation to lever press for alcohol and an enhanced cue- induced reinstatement of lever pressing after extinction. However, in terms of genetic interpretation, the coexistence of two phenotypes within these lines not necessarily depends on the same genetic factor, since these lines and their respective controls are polymorphic at multiple loci. It is imperative to identify and isolate the causative genetic variant to unequivocally relate two phenotypes to the same genetic factor. So, the ultimate aim of the QTL studies in chapters 5 and 6 was to end with a mouse line that, in comparison with its wild type counterpart, possesses one single genetic polymorphism causally affecting the trait. This can be achieved by, rigorous backcrossing, identifying the QTG and creating a genetically modified line and by viral over-expression to rescue the behavioral trait. Together, the QTL and genes identified in this thesis may generate insight into molecular mechanisms contributing to deficits in inhibitory control and attention, which may provide entry points for further research into treatment of these deficits in ADHD, schizophrenia and drug addiction.

149

150

References

Adriani, W., & Laviola, G. 2003. Elevated levels of impulsivity and reduced place conditioning with d-amphetamine: two behavioral features of adolescence in mice. Behav Neurosci, 117(4): 695-703. Altman, D. G. 1991. Practical statistics for medical research. London: Chapman & Hall. Anokhin, A. P., Heath, A. C., Myers, E., Ralano, A., & Wood, S. 2003. Genetic influences on prepulse inhibition of startle reflex in humans. Neurosci Lett, 353(1): 45-48. Aron, A. R., & Poldrack, R. A. 2005. The cognitive neuroscience of response inhibition: relevance for genetic research in attention-deficit/hyperactivity disorder. Biol Psychiatry, 57(11): 1285-1292. Ataman, B., Ashley, J., Gorczyca, M., Ramachandran, P., Fouquet, W., Sigrist, S. J., & Budnik, V. 2008. Rapid activity-dependent modifications in synaptic structure and function require bidirectional Wnt signaling. Neuron, 57(5): 705-718. Baarendse, P. J., van Grootheest, G., Jansen, R. F., Pieneman, A. W., Ogren, S. O., Verhage, M., & Stiedl, O. 2008. Differential involvement of the dorsal hippocampus in passive avoidance in C57bl/6J and DBA/2J mice. Hippocampus, 18(1): 11-19. Bailey, C. D., De Biasi, M., Fletcher, P. J., & Lambe, E. K. 2010. The nicotinic acetylcholine receptor alpha5 subunit plays a key role in attention circuitry and accuracy. J Neurosci, 30(27): 9241-9252. Barkley, R. A., Edwards, G., Laneri, M., Fletcher, K., & Metevia, L. 2001. Executive functioning, temporal discounting, and sense of time in adolescents with attention deficit hyperactivity disorder (ADHD) and oppositional defiant disorder (ODD). J Abnorm Child Psychol, 29(6): 541-556. Beck, L. H., Bransome, E. D., Jr., Mirsky, A. F., Rosvold, H. E., & Sarason, I. 1956. A continuous performance test of brain damage. J Consult Psychol, 20(5): 343- 350. Belin, D., Mar, A. C., Dalley, J. W., Robbins, T. W., & Everitt, B. J. 2008. High impulsivity predicts the switch to compulsive cocaine-taking. Science, 320(5881): 1352-1355. Belknap, J. K., Mitchell, S. R., O'Toole, L. A., Helms, M. L., & Crabbe, J. C. 1996. Type I and type II error rates for quantitative trait loci (QTL) mapping studies using recombinant inbred mouse strains. Behav Genet, 26(2): 149-160. Bidwell, L. C., Willcutt, E. G., Defries, J. C., & Pennington, B. F. 2007. Testing for neuropsychological endophenotypes in siblings discordant for attention- deficit/hyperactivity disorder. Biol Psychiatry, 62(9): 991-998. Bitanihirwe, B. K., Dubroqua, S., Singer, P., Feldon, J., & Yee, B. K. 2011. Sensorimotor gating and vigilance-dependent choice accuracy: a within-subject correlative analysis in wild-type C57BL/6 mice. Behav Brain Res, 217(1): 178-187.

151

References

Bitsios, P., & Giakoumaki, S. G. 2005. Relationship of prepulse inhibition of the startle reflex to attentional and executive mechanisms in man. Int J Psychophysiol, 55(2): 229-241. Blondeau, C., & Dellu-Hagedorn, F. 2007. Dimensional analysis of ADHD subtypes in rats. Biol Psychiatry, 61(12): 1340-1350. Bohmelt, A. H., Schell, A. M., & Dawson, M. E. 1999. Attentional modulation of short- and long-lead-interval modification of the acoustic startle eyeblink response: comparing auditory and visual prestimuli. Int J Psychophysiol, 32(3): 239-250. Bouchard, T. J., Jr., & McGue, M. 2003. Genetic and environmental influences on human psychological differences. J Neurobiol, 54(1): 4-45. Bovolenta, P., Rodriguez, J., & Esteve, P. 2006. Frizzled/RYK mediated signalling in axon guidance. Development, 133(22): 4399-4408. Braff, D. L., & Geyer, M. A. 1990. Sensorimotor gating and schizophrenia. Human and animal model studies. Arch Gen Psychiatry, 47(2): 181-188. Braff, D. L., Geyer, M. A., & Swerdlow, N. R. 2001. Human studies of prepulse inhibition of startle: normal subjects, patient groups, and pharmacological studies. Psychopharmacology (Berl), 156(2-3): 234-258. Brunner, D., & Hen, R. 1997. Insights into the neurobiology of impulsive behavior from serotonin receptor knockout mice. Ann N Y Acad Sci, 836: 81-105. Cabib, S., Orsini, C., Le Moal, M., & Piazza, P. V. 2000. Abolition and reversal of strain differences in behavioral responses to drugs of abuse after a brief experience. Science, 289(5478): 463-465. Cabib, S., Puglisi-Allegra, S., & Ventura, R. 2002. The contribution of comparative studies in inbred strains of mice to the understanding of the hyperactive phenotype. Behav Brain Res, 130(1-2): 103-109. Cardinal, R. N. 2006. Neural systems implicated in delayed and probabilistic reinforcement. Neural Netw, 19(8): 1277-1301. Cardinal, R. N., Pennicott, D. R., Sugathapala, C. L., Robbins, T. W., & Everitt, B. J. 2001. Impulsive choice induced in rats by lesions of the nucleus accumbens core. Science, 292(5526): 2499-2501. Caspi, A., Moffitt, T. E., Newman, D. L., & Silva, P. A. 1996. Behavioral observations at age 3 years predict adult psychiatric disorders. Longitudinal evidence from a birth cohort. Arch Gen Psychiatry, 53(11): 1033-1039. Castellanos, F. X., Sonuga-Barke, E. J., Milham, M. P., & Tannock, R. 2006. Characterizing cognition in ADHD: beyond executive dysfunction. Trends Cogn Sci, 10(3): 117-123. Castellanos, F. X., & Tannock, R. 2002. Neuroscience of attention-deficit/hyperactivity disorder: the search for endophenotypes. Nat Rev Neurosci, 3(8): 617-628. Chesler, E. J., Lu, L., Shou, S., Qu, Y., Gu, J., Wang, J., Hsu, H. C., Mountz, J. D., Baldwin, N. E., Langston, M. A., Threadgill, D. W., Manly, K. F., & Williams, R. W. 2005. Complex trait analysis of gene expression uncovers polygenic and pleiotropic networks that modulate nervous system function. Nat Genet, 37(3): 233-242. Chesler, E. J., Miller, D. R., Branstetter, L. R., Galloway, L. D., Jackson, B. L., Philip, V. M., Voy, B. H., Culiat, C. T., Threadgill, D. W., Williams, R. W., Churchill, G. A., Johnson, D. K., & Manly, K. F. 2008. The Collaborative Cross at Oak Ridge National Laboratory: developing a powerful resource for systems genetics. Mamm Genome, 19(6): 382-389. 152

References

Ciliax, B. J., Drash, G. W., Staley, J. K., Haber, S., Mobley, C. J., Miller, G. W., Mufson, E. J., Mash, D. C., & Levey, A. I. 1999. Immunocytochemical localization of the dopamine transporter in human brain. J Comp Neurol, 409(1): 38-56. Cole, B. J., & Robbins, T. W. 1987. Amphetamine impairs the discriminative performance of rats with dorsal noradrenergic bundle lesions on a 5-choice serial reaction time task: new evidence for central dopaminergic-noradrenergic interactions. Psychopharmacology (Berl), 91(4): 458-466. Crabbe, J. C., & Cunningham, C. L. 2007. Trait or state? Science, 317(5841): 1033-1035; author reply 1033-1035. Crabbe, J. C., Kosobud, A., Young, E. R., & Janowsky, J. S. 1983. Polygenic and single- gene determination of responses to ethanol in BXD/Ty recombinant inbred mouse strains. Neurobehav Toxicol Teratol, 5(2): 181-187. Crabbe, J. C., Wahlsten, D., & Dudek, B. C. 1999. Genetics of mouse behavior: interactions with laboratory environment. Science, 284(5420): 1670-1672. Crusio, W. E. 2006. An Introduction to Quantitative Genetics. In B. C. Jones, & P. Mormède (Eds.), Neurobehavioral Genetics: Methods and Applications., second ed. Boca Raton, FL, USA.: CRC Press. D'Este, L., Casini, A., Puglisi-Allegra, S., Cabib, S., & Renda, T. G. 2007. Comparative immunohistochemical study of the dopaminergic systems in two inbred mouse strains (C57BL/6J and DBA/2J). J Chem Neuroanat, 33(2): 67-74. Dalley, J. W., Cardinal, R. N., & Robbins, T. W. 2004. Prefrontal executive and cognitive functions in rodents: neural and neurochemical substrates. Neurosci Biobehav Rev, 28(7): 771-784. Dalley, J. W., Fryer, T. D., Brichard, L., Robinson, E. S., Theobald, D. E., Laane, K., Pena, Y., Murphy, E. R., Shah, Y., Probst, K., Abakumova, I., Aigbirhio, F. I., Richards, H. K., Hong, Y., Baron, J. C., Everitt, B. J., & Robbins, T. W. 2007. Nucleus accumbens D2/3 receptors predict trait impulsivity and cocaine reinforcement. Science, 315(5816): 1267-1270. Dalley, J. W., Mar, A. C., Economidou, D., & Robbins, T. W. 2008. Neurobehavioral mechanisms of impulsivity: fronto-striatal systems and functional neurochemistry. Pharmacol Biochem Behav, 90(2): 250-260. Daly, G., Hawi, Z., Fitzgerald, M., & Gill, M. 1999. Mapping susceptibility loci in attention deficit hyperactivity disorder: preferential transmission of parental alleles at DAT1, DBH and DRD5 to affected children. Mol Psychiatry, 4(2): 192-196. Davies, W., Humby, T., Isles, A. R., Burgoyne, P. S., & Wilkinson, L. S. 2007. X- monosomy effects on visuospatial attention in mice: a candidate gene and implications for Turner syndrome and attention deficit hyperactivity disorder. Biol Psychiatry, 61(12): 1351-1360. Dawes, M. A., Tarter, R. E., & Kirisci, L. 1997. Behavioral self-regulation: correlates and 2 year follow-ups for boys at risk for substance abuse. Drug Alcohol Depend, 45(3): 165-176. Dawson, M. E., Hazlett, E. A., Filion, D. L., Nuechterlein, K. H., & Schell, A. M. 1993. Attention and schizophrenia: impaired modulation of the startle reflex. J Abnorm Psychol, 102(4): 633-641. Dawson, M. E., Schell, A. M., Hazlett, E. A., Nuechterlein, K. H., & Filion, D. L. 2000. On the clinical and cognitive meaning of impaired sensorimotor gating in schizophrenia. Psychiatry Res, 96(3): 187-197. 153

References de Geus, E. J. 2002. Introducing genetic psychophysiology. Biol Psychol, 61(1-2): 1-10. De Vries, T. J., Shaham, Y., Homberg, J. R., Crombag, H., Schuurman, K., Dieben, J., Vanderschuren, L. J., & Schoffelmeer, A. N. 2001. A cannabinoid mechanism in relapse to cocaine seeking. Nat Med, 7(10): 1151-1154. Deacon, R. M., Penny, C., & Rawlins, J. N. 2003. Effects of medial prefrontal cortex cytotoxic lesions in mice. Behav Brain Res, 139(1-2): 139-155. Derefinko, K. J., Adams, Z. W., Milich, R., Fillmore, M. T., Lorch, E. P., & Lynam, D. R. 2008. Response style differences in the inattentive and combined subtypes of attention-deficit/hyperactivity disorder. J Abnorm Child Psychol, 36(5): 745- 758. Diamond, A. 2005. Attention-deficit disorder (attention-deficit/ hyperactivity disorder without hyperactivity): a neurobiologically and behaviorally distinct disorder from attention-deficit/hyperactivity disorder (with hyperactivity). Dev Psychopathol, 17(3): 807-825. Diergaarde, L., Pattij, T., Poortvliet, I., Hogenboom, F., de Vries, W., Schoffelmeer, A. N., & De Vries, T. J. 2008. Impulsive choice and impulsive action predict vulnerability to distinct stages of nicotine seeking in rats. Biol Psychiatry, 63(3): 301-308. Ditzen, C., Jastorff, A. M., Kessler, M. S., Bunck, M., Teplytska, L., Erhardt, A., Kromer, S. A., Varadarajulu, J., Targosz, B. S., Sayan-Ayata, E. F., Holsboer, F., Landgraf, R., & Turck, C. W. 2006. Protein biomarkers in a mouse model of extremes in trait anxiety. Mol Cell Proteomics, 5(10): 1914-1920. Doyle, A. E., Ferreira, M. A., Sklar, P. B., Lasky-Su, J., Petty, C., Fusillo, S. J., Seidman, L. J., Willcutt, E. G., Smoller, J. W., Purcell, S., Biederman, J., & Faraone, S. V. 2008. Multivariate genomewide linkage scan of neurocognitive traits and ADHD symptoms: suggestive linkage to 3q13. Am J Med Genet B Neuropsychiatr Genet, 147B(8): 1399-1411. DSM-IV. 1994. Diagnostic and statistical manual of mental disorders. (4th ed.). Washington, DC. Eagle, D. M., Bari, A., & Robbins, T. W. 2008. The neuropsychopharmacology of action inhibition: cross-species translation of the stop-signal and go/no-go tasks. Psychopharmacology (Berl), 199(3): 439-456. Eagle, D. M., & Baunez, C. 2010. Is there an inhibitory-response-control system in the rat? Evidence from anatomical and pharmacological studies of behavioral inhibition. Neurosci Biobehav Rev, 34(1): 50-72. Eisenberg, J., Mei-Tal, G., Steinberg, A., Tartakovsky, E., Zohar, A., Gritsenko, I., Nemanov, L., & Ebstein, R. P. 1999. Haplotype relative risk study of catechol- O-methyltransferase (COMT) and attention deficit hyperactivity disorder (ADHD): association of the high-enzyme activity Val allele with ADHD impulsive-hyperactive phenotype. Am J Med Genet, 88(5): 497-502. Elden, A., & Flaten, M. A. 2003. Similar effects of attention directed to acoustic and tactile stimuli on prepulse inhibition of acoustic startle. Scand J Psychol, 44(4): 363-372. Ellenbroek, B. A., & Cools, A. R. 2002. Apomorphine susceptibility and animal models for psychopathology: genes and environment. Behav Genet, 32(5): 349-361. Erway, L. C., Willott, J. F., Archer, J. R., & Harrison, D. E. 1993. Genetics of age-related hearing loss in mice: I. Inbred and F1 hybrid strains. Hear Res, 65(1-2): 125- 132. 154

References

Evenden, J. L. 1999. Varieties of impulsivity. Psychopharmacology (Berl), 146(4): 348- 361. Evenden, J. L., & Ryan, C. N. 1999. The pharmacology of impulsive behaviour in rats VI: the effects of ethanol and selective serotonergic drugs on response choice with varying delays of reinforcement. Psychopharmacology (Berl), 146(4): 413-421. Faraone, S. V., & Biederman, J. 1998. Neurobiology of attention-deficit hyperactivity disorder. Biol Psychiatry, 44(10): 951-958. Faraone, S. V., Doyle, A. E., Mick, E., & Biederman, J. 2001. Meta-analysis of the association between the 7-repeat allele of the dopamine D(4) receptor gene and attention deficit hyperactivity disorder. Am J Psychiatry, 158(7): 1052-1057. Faraone, S. V., & Mick, E. 2010. Molecular genetics of attention deficit hyperactivity disorder. Psychiatr Clin North Am, 33(1): 159-180. Faraone, S. V., Perlis, R. H., Doyle, A. E., Smoller, J. W., Goralnick, J. J., Holmgren, M. A., & Sklar, P. 2005. Molecular genetics of attention-deficit/hyperactivity disorder. Biol Psychiatry, 57(11): 1313-1323. Fattore, L., Piras, G., Corda, M. G., & Giorgi, O. 2009. The Roman high- and low- avoidance rat lines differ in the acquisition, maintenance, extinction, and reinstatement of intravenous cocaine self-administration. Neuropsychopharmacology, 34(5): 1091-1101. Filion, D. L., Dawson, M. E., & Schell, A. M. 1993. Modification of the acoustic startle- reflex eyeblink: a tool for investigating early and late attentional processes. Biol Psychol, 35(3): 185-200. Fisher, R. A. 1918. The correlation between relatives under the supposition of mendelian inheritance. Trans. Roy. Soc., 52: 399–433. Fletcher, P. J., Tampakeras, M., Sinyard, J., & Higgins, G. A. 2007. Opposing effects of 5-HT(2A) and 5-HT(2C) receptor antagonists in the rat and mouse on premature responding in the five-choice serial reaction time test. Psychopharmacology (Berl), 195(2): 223-234. Floresco, S. B., & Phillips, A. G. 2001. Delay-dependent modulation of memory retrieval by infusion of a dopamine D1 agonist into the rat medial prefrontal cortex. Behav Neurosci, 115(4): 934-939. Fox, A. T., Hand, D. J., & Reilly, M. P. 2008. Impulsive choice in a rodent model of attention-deficit/hyperactivity disorder. Behav Brain Res, 187(1): 146-152. Galton, F. 1865. Hereditary talent and character. Macmillan's Magazine, 12: 157-166, 318-327. Geyer, M. A., Krebs-Thomson, K., Braff, D. L., & Swerdlow, N. R. 2001. Pharmacological studies of prepulse inhibition models of sensorimotor gating deficits in schizophrenia: a decade in review. Psychopharmacology (Berl), 156(2-3): 117-154. Goldman-Rakic, P. S., Muly, E. C., 3rd, & Williams, G. V. 2000. D(1) receptors in prefrontal cells and circuits. Brain Res Brain Res Rev, 31(2-3): 295-301. Gottesman, II, & Gould, T. D. 2003. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry, 160(4): 636-645. Graham, F. K. 1975. Presidential Address, 1974. The more or less startling effects of weak prestimulation. Psychophysiology, 12(3): 238-248. Granon, S., Passetti, F., Thomas, K. L., Dalley, J. W., Everitt, B. J., & Robbins, T. W. 2000. Enhanced and impaired attentional performance after infusion of D1 155

References

dopaminergic receptor agents into rat prefrontal cortex. J Neurosci, 20(3): 1208-1215. Greco, B., Invernizzi, R. W., & Carli, M. 2005. Phencyclidine-induced impairment in attention and response control depends on the background genotype of mice: reversal by the mGLU(2/3) receptor agonist LY379268. Psychopharmacology (Berl), 179(1): 68-76. Greenwood, T. A., Braff, D. L., Light, G. A., Cadenhead, K. S., Calkins, M. E., Dobie, D. J., Freedman, R., Green, M. F., Gur, R. E., Gur, R. C., Mintz, J., Nuechterlein, K. H., Olincy, A., Radant, A. D., Seidman, L. J., Siever, L. J., Silverman, J. M., Stone, W. S., Swerdlow, N. R., Tsuang, D. W., Tsuang, M. T., Turetsky, B. I., & Schork, N. J. 2007. Initial heritability analyses of endophenotypic measures for schizophrenia: the consortium on the genetics of schizophrenia. Arch Gen Psychiatry, 64(11): 1242-1250. Grisel, J. E., Metten, P., Wenger, C. D., Merrill, C. M., & Crabbe, J. C. 2002. Mapping of quantitative trait loci underlying ethanol metabolism in BXD recombinant inbred mouse strains. Alcohol Clin Exp Res, 26(5): 610-616. Groot, A. S., de Sonneville, L. M., Stins, J. F., & Boomsma, D. I. 2004. Familial influences on sustained attention and inhibition in preschoolers. J Child Psychol Psychiatry, 45(2): 306-314. Gubner, N. R., Wilhelm, C. J., Phillips, T. J., & Mitchell, S. H. 2010. Strain differences in behavioral inhibition in a Go/No-go task demonstrated using 15 inbred mouse strains. Alcohol Clin Exp Res, 34(8): 1353-1362. Guillem, K., Bloem, B., Poorthuis, R. B., Loos, M., Smit, A. B., Maskos, U., Spijker, S., & Mansvelder, H. D. 2011. Nicotinic acetylcholine receptor beta2 subunits in the medial prefrontal cortex control attention. Science, 333(6044): 888-891. Harrison, A. A., Everitt, B. J., & Robbins, T. W. 1997. Central 5-HT depletion enhances impulsive responding without affecting the accuracy of attentional performance: interactions with dopaminergic mechanisms. Psychopharmacology (Berl), 133(4): 329-342. Hawi, Z., Lowe, N., Kirley, A., Gruenhage, F., Nothen, M., Greenwood, T., Kelsoe, J., Fitzgerald, M., & Gill, M. 2003. Linkage disequilibrium mapping at DAT1, DRD5 and DBH narrows the search for ADHD susceptibility alleles at these loci. Mol Psychiatry, 8(3): 299-308. Hedges, S. B., & Shah, P. 2003. Comparison of mode estimation methods and application in molecular clock analysis. BMC Bioinformatics, 4: 31. Hegmann, J. P., & Possidente, B. 1981. Estimating genetic correlations from inbred strains. Behav Genet, 11(2): 103-114. Heijtz, R. D., Alexeyenko, A., & Castellanos, F. X. 2007. Calcyon mRNA expression in the frontal-striatal circuitry and its relationship to vesicular processes and ADHD. Behav Brain Funct, 3: 33. Heimel, J. A., Hermans, J. M., Sommeijer, J. P., & Levelt, C. N. 2008. Genetic control of experience-dependent plasticity in the visual cortex. Genes Brain Behav, 7(8): 915-923. Helms, C. M., Reeves, J. M., & Mitchell, S. H. 2006. Impact of strain and D- amphetamine on impulsivity (delay discounting) in inbred mice. Psychopharmacology (Berl), 188(2): 144-151. Henderson, N. D. 1969. Prior-treatment effects on open-field emotionality: the need for representative design. Ann N Y Acad Sci, 159(3): 860-868. 156

References

Hitzemann, R., Bell, J., Rasmussen, E., & McCaughran, J. 2001. Mapping the genes for the acoustic startle response (ASR) and prepulse inhibition of the ASR in the BXD recombinant inbred series: effect of high-frequency hearing loss and cochlear pathology. In J. F. Willott (Ed.), Handbook of mouse auditory research: From behavior to molecular biology.: 441–455. New York: CRC Press. Hitzemann, R., Malmanger, B., Belknap, J., Darakjian, P., & McWeeney, S. 2008. Short- term selective breeding for high and low prepulse inhibition of the acoustic startle response; pharmacological characterization and QTL mapping in the selected lines. Pharmacol Biochem Behav, 90(4): 525-533. Holmes, A., & Wellman, C. L. 2008. Stress-induced prefrontal reorganization and executive dysfunction in rodents. Neurosci Biobehav Rev. Hong, J., Shu-Leong, H., Tao, X., & Lap-Ping, Y. 1998. Distribution of catechol-O- methyltransferase expression in human central nervous system. Neuroreport, 9(12): 2861-2864. Hovatta, I., Tennant, R. S., Helton, R., Marr, R. A., Singer, O., Redwine, J. M., Ellison, J. A., Schadt, E. E., Verma, I. M., Lockhart, D. J., & Barlow, C. 2005. Glyoxalase 1 and glutathione reductase 1 regulate anxiety in mice. Nature, 438(7068): 662- 666. Hoyle, E., Genn, R. F., Fernandes, C., & Stolerman, I. P. 2006. Impaired performance of alpha7 nicotinic receptor knockout mice in the five-choice serial reaction time task. Psychopharmacology (Berl), 189(2): 211-223. Humby, T., Laird, F. M., Davies, W., & Wilkinson, L. S. 1999. Visuospatial attentional functioning in mice: interactions between cholinergic manipulations and genotype. Eur J Neurosci, 11(8): 2813-2823. Isles, A. R., Humby, T., Walters, E., & Wilkinson, L. S. 2004. Common genetic effects on variation in impulsivity and activity in mice. J Neurosci, 24(30): 6733-6740. Isles, A. R., Humby, T., & Wilkinson, L. S. 2003. Measuring impulsivity in mice using a novel operant delayed reinforcement task: effects of behavioural manipulations and d-amphetamine. Psychopharmacology (Berl), 170(4): 376-382. Johnson, K. R., Zheng, Q. Y., & Erway, L. C. 2000. A major gene affecting age-related hearing loss is common to at least ten inbred strains of mice. Genomics, 70(2): 171-180. Kafkafi, N., Benjamini, Y., Sakov, A., Elmer, G. I., & Golani, I. 2005. Genotype- environment interactions in mouse behavior: a way out of the problem. Proc Natl Acad Sci U S A, 102(12): 4619-4624. Kas, M. J., & Van Ree, J. M. 2004. Dissecting complex behaviours in the post-genomic era. Trends Neurosci, 27(7): 366-369. Kerns, R. T., Ravindranathan, A., Hassan, S., Cage, M. P., York, T., Sikela, J. M., Williams, R. W., & Miles, M. F. 2005. Ethanol-responsive brain region expression networks: implications for behavioral responses to acute ethanol in DBA/2J versus C57BL/6J mice. J Neurosci, 25(9): 2255-2266. Klein, C., Wendling, K., Huettner, P., Ruder, H., & Peper, M. 2006. Intra-subject variability in attention-deficit hyperactivity disorder. Biol Psychiatry, 60(10): 1088-1097. Kratz, O., Diruf, M. S., Studer, P., Gierow, W., Buchmann, J., Moll, G. H., & Heinrich, H. 2009. Effects of methylphenidate on motor system excitability in a response inhibition task. Behav Brain Funct, 5: 12. 157

References

Kreek, M. J., Nielsen, D. A., Butelman, E. R., & LaForge, K. S. 2005. Genetic influences on impulsivity, risk taking, stress responsivity and vulnerability to drug abuse and addiction. Nat Neurosci, 8(11): 1450-1457. Kuntsi, J., Rogers, H., Swinard, G., Borger, N., van der Meere, J., Rijsdijk, F., & Asherson, P. 2006. Reaction time, inhibition, working memory and 'delay aversion' performance: genetic influences and their interpretation. Psychol Med, 36(11): 1613-1624. Lambourne, S. L., Humby, T., Isles, A. R., Emson, P. C., Spillantini, M. G., & Wilkinson, L. S. 2007. Impairments in impulse control in mice transgenic for the human FTDP-17 tauV337M mutation are exacerbated by age. Hum Mol Genet, 16(14): 1708-1719. Lander, E., & Kruglyak, L. 1995. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet, 11(3): 241-247. Laughlin, R. E., Grant, T. L., Williams, R. W., & Jentsch, J. D. 2011. Genetic dissection of behavioral flexibility: reversal learning in mice. Biol Psychiatry, 69(11): 1109-1116. Laurin, N., Misener, V. L., Crosbie, J., Ickowicz, A., Pathare, T., Roberts, W., Malone, M., Tannock, R., Schachar, R., Kennedy, J. L., & Barr, C. L. 2005. Association of the calcyon gene (DRD1IP) with attention deficit/hyperactivity disorder. Mol Psychiatry, 10(12): 1117-1125. Leth-Steensen, C., Elbaz, Z. K., & Douglas, V. I. 2000. Mean response times, variability, and skew in the responding of ADHD children: a response time distributional approach. Acta Psychol (Amst), 104(2): 167-190. Leussis, M. P., Frayne, M. L., Saito, M., Berry, E. M., Aldinger, K. A., Rockwell, G. N., Hammer, R. P., Jr., Baskin-Hill, A. E., Singer, J. B., Nadeau, J. H., Sklar, P., & Petryshen, T. L. 2009. Genomic survey of prepulse inhibition in mouse chromosome substitution strains. Genes Brain Behav, 8(8): 806-816. Levin, E. D., Conners, C. K., Silva, D., Hinton, S. C., Meck, W. H., March, J., & Rose, J. E. 1998. Transdermal nicotine effects on attention. Psychopharmacology (Berl), 140(2): 135-141. Levin, E. D., Conners, C. K., Sparrow, E., Hinton, S. C., Erhardt, D., Meck, W. H., Rose, J. E., & March, J. 1996. Nicotine effects on adults with attention- deficit/hyperactivity disorder. Psychopharmacology (Berl), 123(1): 55-63. Li, L., Du, Y., Li, N., Wu, X., & Wu, Y. 2009. Top-down modulation of prepulse inhibition of the startle reflex in humans and rats. Neurosci Biobehav Rev, 33(8): 1157-1167. Lidow, M. S., Roberts, A., Zhang, L., Koh, P. O., Lezcano, N., & Bergson, C. 2001. Receptor crosstalk protein, calcyon, regulates affinity state of dopamine D1 receptors. Eur J Pharmacol, 427(3): 187-193. Lien, Y. J., Hsiao, P. C., Liu, C. M., Faraone, S. V., Tsuang, M. T., Hwu, H. G., & Chen, W. J. 2011. A genome-wide linkage scan for distinct subsets of schizophrenia characterized by age at onset and neurocognitive deficits. PLoS One, 6(8): e24103. Lien, Y. J., Liu, C. M., Faraone, S. V., Tsuang, M. T., Hwu, H. G., Hsiao, P. C., & Chen, W. J. 2010. A genome-wide quantitative trait loci scan of neurocognitive performances in families with schizophrenia. Genes Brain Behav, 9(7): 695- 702.

158

References

Lips, E. S., Cornelisse, L. N., Toonen, R. F., Min, J. L., Hultman, C. M., Holmans, P. A., O'Donovan, M. C., Purcell, S. M., Smit, A. B., Verhage, M., Sullivan, P. F., Visscher, P. M., & Posthuma, D. 2011. Functional gene group analysis identifies synaptic gene groups as risk factor for schizophrenia. Mol Psychiatry. Lipszyc, J., & Schachar, R. 2010. Inhibitory control and psychopathology: a meta- analysis of studies using the stop signal task. J Int Neuropsychol Soc, 16(6): 1064-1076. Liu, D., Singh, R. P., Khan, A. H., Bhavsar, K., Lusis, A. J., Davis, R. C., & Smith, D. J. 2003. Identifying loci for behavioral traits using genome-tagged mice. J Neurosci Res, 74(4): 562-569. Liu, J. Z., McRae, A. F., Nyholt, D. R., Medland, S. E., Wray, N. R., Brown, K. M., Hayward, N. K., Montgomery, G. W., Visscher, P. M., Martin, N. G., & Macgregor, S. 2010. A versatile gene-based test for genome-wide association studies. Am J Hum Genet, 87(1): 139-145. Logue, S. F., Swartz, R. J., & Wehner, J. M. 1998. Genetic correlation between performance on an appetitive-signaled nosepoke task and voluntary ethanol consumption. Alcohol Clin Exp Res, 22(9): 1912-1920. Loos, M., Pattij, T., Janssen, M. C., Counotte, D. S., Schoffelmeer, A. N., Smit, A. B., Spijker, S., & van Gaalen, M. M. 2010a. Dopamine receptor D1/D5 gene expression in the medial prefrontal cortex predicts impulsive choice in rats. Cereb Cortex, 20(5): 1064-1070. Loos, M., Staal, J., Schoffelmeer, A. N., Smit, A. B., Spijker, S., & Pattij, T. 2010b. Inhibitory control and response latency differences between C57BL/6J and DBA/2J mice in a Go/No-Go and 5-choice serial reaction time task and strain- specific responsivity to amphetamine. Behav Brain Res, 214(2): 216-224. Loos, M., van der Sluis, S., Bochdanovits, Z., van Zutphen, I. J., Pattij, T., Stiedl, O., Smit, A. B., & Spijker, S. 2009. Activity and impulsive action are controlled by different genetic and environmental factors. Genes Brain Behav, 8(8): 817-828. Lowe, N., Kirley, A., Hawi, Z., Sham, P., Wickham, H., Kratochvil, C. J., Smith, S. D., Lee, S. Y., Levy, F., Kent, L., Middle, F., Rohde, L. A., Roman, T., Tahir, E., Yazgan, Y., Asherson, P., Mill, J., Thapar, A., Payton, A., Todd, R. D., Stephens, T., Ebstein, R. P., Manor, I., Barr, C. L., Wigg, K. G., Sinke, R. J., Buitelaar, J. K., Smalley, S. L., Nelson, S. F., Biederman, J., Faraone, S. V., & Gill, M. 2004. Joint analysis of the DRD5 marker concludes association with attention-deficit/hyperactivity disorder confined to the predominantly inattentive and combined subtypes. Am J Hum Genet, 74(2): 348-356. Lush, I. E. 1989. The genetics of tasting in mice. VI. Saccharin, acesulfame, dulcin and sucrose. Genet Res, 53(2): 95-99. Lynch, M., & Walsh, B. 1998. Genetics and Analysis of Quantitative Traits. Sunderland, MA, USA: Sinauer Associates. Manly, K. F., Cudmore, R. H., Jr., & Meer, J. M. 2001. Map Manager QTX, cross- platform software for genetic mapping. Mamm Genome, 12(12): 930-932. Mansvelder, H. D., van Aerde, K. I., Couey, J. J., & Brussaard, A. B. 2006. Nicotinic modulation of neuronal networks: from receptors to cognition. Psychopharmacology (Berl), 184(3-4): 292-305. McCaughran, J., Jr., Bell, J., & Hitzemann, R. 1999. On the relationships of high- frequency hearing loss and cochlear pathology to the acoustic startle response

159

References

(ASR) and prepulse inhibition of the ASR in the BXD recombinant inbred series. Behav Genet, 29(1): 21-30. McClure, S. M., Laibson, D. I., Loewenstein, G., & Cohen, J. D. 2004. Separate neural systems value immediate and delayed monetary rewards. Science, 306(5695): 503-507. McDonald, M. P., Wong, R., Goldstein, G., Weintraub, B., Cheng, S. Y., & Crawley, J. N. 1998. Hyperactivity and learning deficits in transgenic mice bearing a human mutant thyroid hormone beta1 receptor gene. Learn Mem, 5(4-5): 289-301. McIlwain, K. L., Merriweather, M. Y., Yuva-Paylor, L. A., & Paylor, R. 2001. The use of behavioral test batteries: effects of training history. Physiol Behav, 73(5): 705- 717. Mendel, G. 1866. Versuche über Pflanzen-Hybriden. Verh. Naturforsch. Ver. Brünn, 4: 3-47. Mendrek, A., Blaha, C. D., & Phillips, A. G. 1998. Pre-exposure of rats to amphetamine sensitizes self-administration of this drug under a progressive ratio schedule. Psychopharmacology (Berl), 135(4): 416-422. Milich, R., Balentine, A. C., & Lynam, D. R. 2001. ADHD Combined Type and ADHD Predominantly Inattentive Type Are Distinct and Unrelated Disorders. Clinical Psychology: Science and Practice, 8(4): 463-488. Millan, M. J., Newman-Tancredi, A., Quentric, Y., & Cussac, D. 2001. The "selective" dopamine D1 receptor antagonist, SCH23390, is a potent and high efficacy agonist at cloned human serotonin2C receptors. Psychopharmacology (Berl), 156(1): 58-62. Misener, V. L., Luca, P., Azeke, O., Crosbie, J., Waldman, I., Tannock, R., Roberts, W., Malone, M., Schachar, R., Ickowicz, A., Kennedy, J. L., & Barr, C. L. 2004. Linkage of the dopamine receptor D1 gene to attention-deficit/hyperactivity disorder. Mol Psychiatry, 9(5): 500-509. Morar, B., Dragovic, M., Waters, F. A., Chandler, D., Kalaydjieva, L., & Jablensky, A. 2010. Neuregulin 3 (NRG3) as a susceptibility gene in a schizophrenia subtype with florid delusions and relatively spared cognition. Mol Psychiatry. Muir, J. L., Everitt, B. J., & Robbins, T. W. 1996. The cerebral cortex of the rat and visual attentional function: dissociable effects of mediofrontal, cingulate, anterior dorsolateral, and parietal cortex lesions on a five-choice serial reaction time task. Cereb Cortex, 6(3): 470-481. Munafo, M. R., Thiselton, D. L., Clark, T. G., & Flint, J. 2006. Association of the NRG1 gene and schizophrenia: a meta-analysis. Mol Psychiatry, 11(6): 539-546. Nieuwenstein, M. R., Aleman, A., & de Haan, E. H. 2001. Relationship between symptom dimensions and neurocognitive functioning in schizophrenia: a meta- analysis of WCST and CPT studies. Wisconsin Card Sorting Test. Continuous Performance Test. J Psychiatr Res, 35(2): 119-125. Nigg, J. T., Willcutt, E. G., Doyle, A. E., & Sonuga-Barke, E. J. 2005. Causal heterogeneity in attention-deficit/hyperactivity disorder: do we need neuropsychologically impaired subtypes? Biol Psychiatry, 57(11): 1224-1230. Nolan, K. A., D'Angelo, D., & Hoptman, M. J. 2010. Self-report and laboratory measures of impulsivity in patients with schizophrenia or schizoaffective disorder and healthy controls. Psychiatry Res.

160

References

Oberlin, B. G., & Grahame, N. J. 2009. High-alcohol preferring mice are more impulsive than low-alcohol preferring mice as measured in the delay discounting task. Alcohol Clin Exp Res, 33(7): 1294-1303. Oliver, Y. P., Ripley, T. L., & Stephens, D. N. 2009. Ethanol effects on impulsivity in two mouse strains: similarities to diazepam and ketamine. Psychopharmacology (Berl), 204(4): 679-692. Otsuki, T., Kajigaya, S., Ozawa, K., & Liu, J. M. 1999. SNX5, a new member of the sorting nexin family, binds to the Fanconi anemia complementation group A protein. Biochem Biophys Res Commun, 265(3): 630-635. Patel, S., Stolerman, I. P., Asherson, P., & Sluyter, F. 2006. Attentional performance of C57BL/6 and DBA/2 mice in the 5-choice serial reaction time task. Behav Brain Res, 170(2): 197-203. Pattij, T., Janssen, M. C., Loos, M., Smit, A. B., Schoffelmeer, A. N., & van Gaalen, M. M. 2007. Strain specificity and cholinergic modulation of visuospatial attention in three inbred mouse strains. Genes Brain Behav, 6(6): 579-587. Pattij, T., & Vanderschuren, L. J. 2008. The neuropharmacology of impulsive behaviour. Trends Pharmacol Sci, 29(4): 192-199. Paxinos, G., & Watson, C. 1998. The rat brain in stereotaxic coordinates. Sydney: Academic Press. Peirce, J. L., Lu, L., Gu, J., Silver, L. M., & Williams, R. W. 2004. A new set of BXD recombinant inbred lines from advanced intercross populations in mice. BMC Genet, 5: 7. Perry, J. L., Larson, E. B., German, J. P., Madden, G. J., & Carroll, M. E. 2005. Impulsivity (delay discounting) as a predictor of acquisition of IV cocaine self- administration in female rats. Psychopharmacology (Berl), 178(2-3): 193-201. Perry, R. J., & Hodges, J. R. 1999. Attention and executive deficits in Alzheimer's disease. A critical review. Brain, 122 ( Pt 3): 383-404. Petryshen, T. L., Kirby, A., Hammer, R. P., Jr., Purcell, S., O'Leary, S. B., Singer, J. B., Hill, A. E., Nadeau, J. H., Daly, M. J., & Sklar, P. 2005. Two quantitative trait loci for prepulse inhibition of startle identified on mouse chromosome 16 using chromosome substitution strains. Genetics, 171(4): 1895-1904. Phillips, T. J., Belknap, J. K., & Crabbe, J. C. 1991. Use of recombinant inbred strains to assess vulnerability to drug abuse at the genetic level. J Addict Dis, 10(1-2): 73- 87. Phillips, T. J., Crabbe, J. C., Metten, P., & Belknap, J. K. 1994. Localization of genes affecting alcohol drinking in mice. Alcohol Clin Exp Res, 18(4): 931-941. Piazza, P. V., Deminiere, J. M., Le Moal, M., & Simon, H. 1989. Factors that predict individual vulnerability to amphetamine self-administration. Science, 245(4925): 1511-1513. Pietras, C. J., Cherek, D. R., Lane, S. D., Tcheremissine, O. V., & Steinberg, J. L. 2003. Effects of methylphenidate on impulsive choice in adult humans. Psychopharmacology (Berl), 170(4): 390-398. Pietrzak, R. H., Mollica, C. M., Maruff, P., & Snyder, P. J. 2006. Cognitive effects of immediate-release methylphenidate in children with attention- deficit/hyperactivity disorder. Neurosci Biobehav Rev, 30(8): 1225-1245. Plappert, C. F., Pilz, P. K., & Schnitzler, H. U. 2004. Factors governing prepulse inhibition and prepulse facilitation of the acoustic startle response in mice. Behav Brain Res, 152(2): 403-412. 161

References

Poulos, C. X., Le, A. D., & Parker, J. L. 1995. Impulsivity predicts individual susceptibility to high levels of alcohol self-administration. Behav Pharmacol, 6(8): 810-814. Powell, S. B., Zhou, X., & Geyer, M. A. 2009. Prepulse inhibition and genetic mouse models of schizophrenia. Behav Brain Res, 204(2): 282-294. Puglisi-Allegra, S., & Cabib, S. 1997. Psychopharmacology of dopamine: the contribution of comparative studies in inbred strains of mice. Prog Neurobiol, 51(6): 637-661. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M. A., Bender, D., Maller, J., Sklar, P., de Bakker, P. I., Daly, M. J., & Sham, P. C. 2007. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet, 81(3): 559-575. Relkovic, D., Doe, C. M., Humby, T., Johnstone, K. A., Resnick, J. L., Holland, A. J., Hagan, J. J., Wilkinson, L. S., & Isles, A. R. 2010. Behavioural and cognitive abnormalities in an imprinting centre deletion mouse model for Prader-Willi syndrome. Eur J Neurosci, 31(1): 156-164. Riley, B., & Kendler, K. S. 2006. Molecular genetic studies of schizophrenia. Eur J Hum Genet, 14(6): 669-680. Rizzi, T. S., Arias-Vasquez, A., Rommelse, N., Kuntsi, J., Anney, R., Asherson, P., Buitelaar, J., Banaschewski, T., Ebstein, R., Ruano, D., Van der Sluis, S., Markunas, C. A., Garrett, M. E., Ashley-Koch, A. E., Kollins, S. H., Anastopoulos, A. D., Hansell, N. K., Wright, M. J., Montgomery, G. W., Martin, N. G., Harris, S. E., Davies, G., Tenesa, A., Porteous, D. J., Starr, J. M., Deary, I. J., St Pourcain, B., Davey Smith, G., Timpson, N. J., Evans, D. M., Gill, M., Miranda, A., Mulas, F., Oades, R. D., Roeyers, H., Rothenberger, A., Sergeant, J., Sonuga-Barke, E., Steinhausen, H. C., Taylor, E., Faraone, S. V., Franke, B., & Posthuma, D. 2011. The ATXN1 and TRIM31 genes are related to intelligence in an ADHD background: evidence from a large collaborative study totaling 4,963 subjects. Am J Med Genet B Neuropsychiatr Genet, 156(2): 145-157. Robbins, T. W. 2002. The 5-choice serial reaction time task: behavioural pharmacology and functional neurochemistry. Psychopharmacology (Berl), 163(3-4): 362- 380. Robbins, T. W. 2005. Chemistry of the mind: neurochemical modulation of prefrontal cortical function. J Comp Neurol, 493(1): 140-146. Robinson, E. S., Eagle, D. M., Mar, A. C., Bari, A., Banerjee, G., Jiang, X., Dalley, J. W., & Robbins, T. W. 2008. Similar effects of the selective noradrenaline reuptake inhibitor atomoxetine on three distinct forms of impulsivity in the rat. Neuropsychopharmacology, 33(5): 1028-1037. Rodriguez-Oroz, M. C., Jahanshahi, M., Krack, P., Litvan, I., Macias, R., Bezard, E., & Obeso, J. A. 2009. Initial clinical manifestations of Parkinson's disease: features and pathophysiological mechanisms. Lancet Neurol, 8(12): 1128-1139. Rodriguez, L. A., Plomin, R., Blizard, D. A., Jones, B. C., & McClearn, G. E. 1994. Alcohol acceptance, preference, and sensitivity in mice. I. Quantitative genetic analysis using BXD recombinant inbred strains. Alcohol Clin Exp Res, 18(6): 1416-1422.

162

References

Romberg, C., Mattson, M. P., Mughal, M. R., Bussey, T. J., & Saksida, L. M. 2011. Impaired attention in the 3xTgAD mouse model of Alzheimer's disease: rescue by donepezil (Aricept). J Neurosci, 31(9): 3500-3507. Rommelse, N. N., Arias-Vasquez, A., Altink, M. E., Buschgens, C. J., Fliers, E., Asherson, P., Faraone, S. V., Buitelaar, J. K., Sergeant, J. A., Oosterlaan, J., & Franke, B. 2008. Neuropsychological endophenotype approach to genome-wide linkage analysis identifies susceptibility loci for ADHD on 2q21.1 and 13q12.11. Am J Hum Genet, 83(1): 99-105. Rothman, R. B., Baumann, M. H., Dersch, C. M., Romero, D. V., Rice, K. C., Carroll, F. I., & Partilla, J. S. 2001. Amphetamine-type central nervous system stimulants release norepinephrine more potently than they release dopamine and serotonin. Synapse, 39(1): 32-41. Sabol, K. E., Richards, J. B., Broom, S. L., Roach, J. T., & Hausknecht, K. 2003. Effects of stimulus salience and methamphetamine on choice reaction time in the rat: central tendency versus distribution skew. Behav Pharmacol, 14(7): 489-500. Sagvolden, T., Russell, V. A., Aase, H., Johansen, E. B., & Farshbaf, M. 2005. Rodent models of attention-deficit/hyperactivity disorder. Biol Psychiatry, 57(11): 1239-1247. Samocha, K. E., Lim, J. E., Cheng, R., Sokoloff, G., & Palmer, A. A. 2010. Fine mapping of QTL for prepulse inhibition in LG/J and SM/J mice using F and advanced intercross lines. Genes Brain Behav. Sawaguchi, T., & Goldman-Rakic, P. S. 1991. D1 dopamine receptors in prefrontal cortex: involvement in working memory. Science, 251(4996): 947-950. Schachar, R. J., Forget-Dubois, N., Dionne, G., Boivin, M., & Robaey, P. 2010. Heritability of Response Inhibition in Children. J Int Neuropsychol Soc: 1-10. Scholes, K. E., & Martin-Iverson, M. T. 2009. Relationships between prepulse inhibition and cognition are mediated by attentional processes. Behav Brain Res, 205(2): 456-467. Scholes, K. E., & Martin-Iverson, M. T. 2010. Disturbed prepulse inhibition in patients with schizophrenia is consequential to dysfunction of selective attention. Psychophysiology, 47(2): 223-235. Seamans, J. K., Floresco, S. B., & Phillips, A. G. 1998. D1 receptor modulation of hippocampal-prefrontal cortical circuits integrating spatial memory with executive functions in the rat. J Neurosci, 18(4): 1613-1621. Seamans, J. K., & Yang, C. R. 2004. The principal features and mechanisms of dopamine modulation in the prefrontal cortex. Prog Neurobiol, 74(1): 1-58. Sergeant, J., & van der Meere, J. 1990. Additive factor method applied to psychopathology with special reference to childhood hyperactivity. Acta Psychol (Amst), 74(2-3): 277-295. Sergeant, J. A., Geurts, H., Huijbregts, S., Scheres, A., & Oosterlaan, J. 2003. The top and the bottom of ADHD: a neuropsychological perspective. Neurosci Biobehav Rev, 27(7): 583-592. Solanto, M. V., Abikoff, H., Sonuga-Barke, E., Schachar, R., Logan, G. D., Wigal, T., Hechtman, L., Hinshaw, S., & Turkel, E. 2001. The ecological validity of delay aversion and response inhibition as measures of impulsivity in AD/HD: a supplement to the NIMH multimodal treatment study of AD/HD. J Abnorm Child Psychol, 29(3): 215-228.

163

References

Sonuga-Barke, E. J. 2002. Psychological heterogeneity in AD/HD--a dual pathway model of behaviour and cognition. Behav Brain Res, 130(1-2): 29-36. Sonuga-Barke, E. J. 2005. Causal models of attention-deficit/hyperactivity disorder: from common simple deficits to multiple developmental pathways. Biol Psychiatry, 57(11): 1231-1238. Sonuga-Barke, E. J., Taylor, E., Sembi, S., & Smith, J. 1992. Hyperactivity and delay aversion--I. The effect of delay on choice. J Child Psychol Psychiatry, 33(2): 387-398. Spencer, S. V., Hawk, L. W., Jr., Richards, J. B., Shiels, K., Pelham, W. E., Jr., & Waxmonsky, J. G. 2009. Stimulant treatment reduces lapses in attention among children with ADHD: the effects of methylphenidate on intra-individual response time distributions. J Abnorm Child Psychol, 37(6): 805-816. Spijker, S., Houtzager, S. W., De Gunst, M. C., De Boer, W. P., Schoffelmeer, A. N., & Smit, A. B. 2004. Morphine exposure and abstinence define specific stages of gene expression in the rat nucleus accumbens. Faseb J, 18(7): 848-850. Steinmetz, J. E., Blankenship, M. R., Green, J. T., Smith, G. B., & Finn, P. R. 2000. Evaluation of behavioral disinhibition in P/NP and HAD1/LAD1 rats. Prog Neuropsychopharmacol Biol Psychiatry, 24(6): 1025-1039. Stevens, J. P. 2002. Applied Multivariate Statistics for the Social Sciences. Philadelphia: Lawrence Erlbaum. Storey, J. D., & Tibshirani, R. 2003. Statistical significance for genomewide studies. Proc Natl Acad Sci U S A, 100(16): 9440-9445. Sun, M., Thomas, M. J., Herder, R., Bofenkamp, M. L., Selleck, S. B., & O'Connor, M. B. 2007. Presynaptic contributions of chordin to hippocampal plasticity and spatial learning. J Neurosci, 27(29): 7740-7750. Swerdlow, N. R., Filion, D., Geyer, M. A., & Braff, D. L. 1995. "Normal" personality correlates of sensorimotor, cognitive, and visuospatial gating. Biol Psychiatry, 37(5): 286-299. Thorne, G. L., Dawson, M. E., & Schell, A. M. 2005. Attention and prepulse inhibition: the effects of task-relevant, irrelevant, and no-task conditions. Int J Psychophysiol, 56(2): 121-128. Tolliver, G. A., Sadeghi, K. G., & Samson, H. H. 1988. Ethanol preference following the sucrose-fading initiation procedure. Alcohol, 5(1): 9-13. Toplak, M. E., Dockstader, C., & Tannock, R. 2006. Temporal information processing in ADHD: findings to date and new methods. J Neurosci Methods, 151(1): 15-29. Trommer, B. L., Hoeppner, J. A., & Zecker, S. G. 1991. The go-no go test in attention deficit disorder is sensitive to methylphenidate. J Child Neurol, 6 Suppl: S128- 131. Tunbridge, E. M., Harrison, P. J., & Weinberger, D. R. 2006. Catechol-o- methyltransferase, cognition, and psychosis: Val158Met and beyond. Biol Psychiatry, 60(2): 141-151. Tusher, V. G., Tibshirani, R., & Chu, G. 2001. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A, 98(9): 5116-5121. Uylings, H. B., Groenewegen, H. J., & Kolb, B. 2003. Do rats have a prefrontal cortex? Behav Brain Res, 146(1-2): 3-17. Vaidya, C. J., Austin, G., Kirkorian, G., Ridlehuber, H. W., Desmond, J. E., Glover, G. H., & Gabrieli, J. D. 1998. Selective effects of methylphenidate in attention 164

References

deficit hyperactivity disorder: a functional magnetic resonance study. Proc Natl Acad Sci U S A, 95(24): 14494-14499. van der Meere, J., Gunning, B., & Stemerdink, N. 1999. The effect of methylphenidate and clonidine on response inhibition and state regulation in children with ADHD. J Child Psychol Psychiatry, 40(2): 291-298. van Gaalen, M. M., Brueggeman, R. J., Bronius, P. F., Schoffelmeer, A. N., & Vanderschuren, L. J. 2006a. Behavioral disinhibition requires dopamine receptor activation. Psychopharmacology (Berl), 187(1): 73-85. van Gaalen, M. M., van Koten, R., Schoffelmeer, A. N., & Vanderschuren, L. J. 2006b. Critical involvement of dopaminergic neurotransmission in impulsive decision making. Biol Psychiatry, 60(1): 66-73. Van Tol, H. H., Wu, C. M., Guan, H. C., Ohara, K., Bunzow, J. R., Civelli, O., Kennedy, J., Seeman, P., Niznik, H. B., & Jovanovic, V. 1992. Multiple dopamine D4 receptor variants in the human population. Nature, 358(6382): 149-152. Vannoni, E., Voikar, V., & Wolfer, D. P. 2010. Higher motor impulsivity in DBA/2 than in C57BL/6 mice revealed by a novel simple reaction time task in intellicage., Presented at: Society for Neuroscience. San Diego, USA. Varty, G. B., Walters, N., Cohen-Williams, M., & Carey, G. J. 2001. Comparison of apomorphine, amphetamine and dizocilpine disruptions of prepulse inhibition in inbred and outbred mice strains. Eur J Pharmacol, 424(1): 27-36. Ventura, R., Alcaro, A., Cabib, S., Conversi, D., Mandolesi, L., & Puglisi-Allegra, S. 2004. Dopamine in the medial prefrontal cortex controls genotype-dependent effects of amphetamine on mesoaccumbens dopamine release and locomotion. Neuropsychopharmacology, 29(1): 72-80. Verdejo-Garcia, A., Lawrence, A. J., & Clark, L. 2008. Impulsivity as a vulnerability marker for substance-use disorders: review of findings from high-risk research, problem gamblers and genetic association studies. Neurosci Biobehav Rev, 32(4): 777-810. Wade, T. R., de Wit, H., & Richards, J. B. 2000. Effects of dopaminergic drugs on delayed reward as a measure of impulsive behavior in rats. Psychopharmacology (Berl), 150(1): 90-101. Waldman, I. D., & Gizer, I. R. 2006. The genetics of attention deficit hyperactivity disorder. Clin Psychol Rev, 26(4): 396-432. Watanabe, A., Toyota, T., Owada, Y., Hayashi, T., Iwayama, Y., Matsumata, M., Ishitsuka, Y., Nakaya, A., Maekawa, M., Ohnishi, T., Arai, R., Sakurai, K., Yamada, K., Kondo, H., Hashimoto, K., Osumi, N., & Yoshikawa, T. 2007. Fabp7 maps to a quantitative trait locus for a schizophrenia endophenotype. PLoS Biol, 5(11): e297. Waterston, R. H., Lindblad-Toh, K., Birney, E., Rogers, J., Abril, J. F., Agarwal, P., Agarwala, R., Ainscough, R., Alexandersson, M., An, P., Antonarakis, S. E., Attwood, J., Baertsch, R., Bailey, J., Barlow, K., Beck, S., Berry, E., Birren, B., Bloom, T., Bork, P., Botcherby, M., Bray, N., Brent, M. R., Brown, D. G., Brown, S. D., Bult, C., Burton, J., Butler, J., Campbell, R. D., Carninci, P., Cawley, S., Chiaromonte, F., Chinwalla, A. T., Church, D. M., Clamp, M., Clee, C., Collins, F. S., Cook, L. L., Copley, R. R., Coulson, A., Couronne, O., Cuff, J., Curwen, V., Cutts, T., Daly, M., David, R., Davies, J., Delehaunty, K. D., Deri, J., Dermitzakis, E. T., Dewey, C., Dickens, N. J., Diekhans, M., Dodge, S., Dubchak, I., Dunn, D. M., Eddy, S. R., Elnitski, L., Emes, R. D., Eswara, P., 165

References

Eyras, E., Felsenfeld, A., Fewell, G. A., Flicek, P., Foley, K., Frankel, W. N., Fulton, L. A., Fulton, R. S., Furey, T. S., Gage, D., Gibbs, R. A., Glusman, G., Gnerre, S., Goldman, N., Goodstadt, L., Grafham, D., Graves, T. A., Green, E. D., Gregory, S., Guigo, R., Guyer, M., Hardison, R. C., Haussler, D., Hayashizaki, Y., Hillier, L. W., Hinrichs, A., Hlavina, W., Holzer, T., Hsu, F., Hua, A., Hubbard, T., Hunt, A., Jackson, I., Jaffe, D. B., Johnson, L. S., Jones, M., Jones, T. A., Joy, A., Kamal, M., Karlsson, E. K., Karolchik, D., Kasprzyk, A., Kawai, J., Keibler, E., Kells, C., Kent, W. J., Kirby, A., Kolbe, D. L., Korf, I., Kucherlapati, R. S., Kulbokas, E. J., Kulp, D., Landers, T., Leger, J. P., Leonard, S., Letunic, I., Levine, R., Li, J., Li, M., Lloyd, C., Lucas, S., Ma, B., Maglott, D. R., Mardis, E. R., Matthews, L., Mauceli, E., Mayer, J. H., McCarthy, M., McCombie, W. R., McLaren, S., McLay, K., McPherson, J. D., Meldrim, J., Meredith, B., Mesirov, J. P., Miller, W., Miner, T. L., Mongin, E., Montgomery, K. T., Morgan, M., Mott, R., Mullikin, J. C., Muzny, D. M., Nash, W. E., Nelson, J. O., Nhan, M. N., Nicol, R., Ning, Z., Nusbaum, C., O'Connor, M. J., Okazaki, Y., Oliver, K., Overton-Larty, E., Pachter, L., Parra, G., Pepin, K. H., Peterson, J., Pevzner, P., Plumb, R., Pohl, C. S., Poliakov, A., Ponce, T. C., Ponting, C. P., Potter, S., Quail, M., Reymond, A., Roe, B. A., Roskin, K. M., Rubin, E. M., Rust, A. G., Santos, R., Sapojnikov, V., Schultz, B., Schultz, J., Schwartz, M. S., Schwartz, S., Scott, C., Seaman, S., Searle, S., Sharpe, T., Sheridan, A., Shownkeen, R., Sims, S., Singer, J. B., Slater, G., Smit, A., Smith, D. R., Spencer, B., Stabenau, A., Stange-Thomann, N., Sugnet, C., Suyama, M., Tesler, G., Thompson, J., Torrents, D., Trevaskis, E., Tromp, J., Ucla, C., Ureta- Vidal, A., Vinson, J. P., Von Niederhausern, A. C., Wade, C. M., Wall, M., Weber, R. J., Weiss, R. B., Wendl, M. C., West, A. P., Wetterstrand, K., Wheeler, R., Whelan, S., Wierzbowski, J., Willey, D., Williams, S., Wilson, R. K., Winter, E., Worley, K. C., Wyman, D., Yang, S., Yang, S. P., Zdobnov, E. M., Zody, M. C., & Lander, E. S. 2002. Initial sequencing and comparative analysis of the mouse genome. Nature, 420(6915): 520-562. Watson, J. D., & Crick, F. H. 1953. Molecular structure of nucleic acids; a structure for deoxyribose nucleic acid. Nature, 171(4356): 737-738. Wilhelm, C. J., & Mitchell, S. H. 2008. Rats bred for high alcohol drinking are more sensitive to delayed and probabilistic outcomes. Genes Brain Behav, 7(7): 705- 713. Wilhelm, C. J., & Mitchell, S. H. 2009. Strain Differences in Delay Discounting Using Inbred Rats. Genes Brain Behav. Williams, G. V., & Goldman-Rakic, P. S. 1995. Modulation of memory fields by dopamine D1 receptors in prefrontal cortex. Nature, 376(6541): 572-575. Willott, J. F., & Erway, L. C. 1998. Genetics of age-related hearing loss in mice. IV. Cochlear pathology and hearing loss in 25 BXD recombinant inbred mouse strains. Hear Res, 119(1-2): 27-36. Willott, J. F., Tanner, L., O'Steen, J., Johnson, K. R., Bogue, M. A., & Gagnon, L. 2003. Acoustic startle and prepulse inhibition in 40 inbred strains of mice. Behav Neurosci, 117(4): 716-727. Wills, T. A., Vaccaro, D., & McNamara, G. 1994. Novelty seeking, risk taking, and related constructs as predictors of adolescent substance use: an application of Cloninger's theory. J Subst Abuse, 6(1): 1-20.

166

References

Winstanley, C. A., Eagle, D. M., & Robbins, T. W. 2006a. Behavioral models of impulsivity in relation to ADHD: translation between clinical and preclinical studies. Clin Psychol Rev, 26(4): 379-395. Winstanley, C. A., Olausson, P., Taylor, J. R., & Jentsch, J. D. 2010. Insight into the relationship between impulsivity and substance abuse from studies using animal models. Alcohol Clin Exp Res, 34(8): 1306-1318. Winstanley, C. A., Theobald, D. E., Cardinal, R. N., & Robbins, T. W. 2004. Contrasting roles of basolateral amygdala and orbitofrontal cortex in impulsive choice. J Neurosci, 24(20): 4718-4722. Winstanley, C. A., Theobald, D. E., Dalley, J. W., Cardinal, R. N., & Robbins, T. W. 2006b. Double dissociation between serotonergic and dopaminergic modulation of medial prefrontal and orbitofrontal cortex during a test of impulsive choice. Cereb Cortex, 16(1): 106-114. Winstanley, C. A., Theobald, D. E., Dalley, J. W., & Robbins, T. W. 2005. Interactions between serotonin and dopamine in the control of impulsive choice in rats: therapeutic implications for impulse control disorders. Neuropsychopharmacology, 30(4): 669-682. Wong, A. A., & Brown, R. E. 2006. Visual detection, pattern discrimination and visual acuity in 14 strains of mice. Genes Brain Behav, 5(5): 389-403. Yang, R. J., Mozhui, K., Karlsson, R. M., Cameron, H. A., Williams, R. W., & Holmes, A. 2008. Variation in mouse basolateral amygdala volume is associated with differences in stress reactivity and fear learning. Neuropsychopharmacology, 33(11): 2595-2604. Young, J. W., Crawford, N., Kelly, J. S., Kerr, L. E., Marston, H. M., Spratt, C., Finlayson, K., & Sharkey, J. 2007. Impaired attention is central to the cognitive deficits observed in alpha 7 deficient mice. Eur Neuropsychopharmacol, 17(2): 145-155. Young, J. W., Finlayson, K., Spratt, C., Marston, H. M., Crawford, N., Kelly, J. S., & Sharkey, J. 2004. Nicotine improves sustained attention in mice: evidence for involvement of the alpha7 nicotinic acetylcholine receptor. Neuropsychopharmacology, 29(5): 891-900. Young, J. W., Light, G. A., Marston, H. M., Sharp, R., & Geyer, M. A. 2009a. The 5- choice continuous performance test: evidence for a translational test of vigilance for mice. PLoS One, 4(1): e4227. Young, J. W., Powell, S. B., Scott, C. N., Zhou, X., & Geyer, M. A. 2011. The effect of reduced dopamine D4 receptor expression in the 5-choice continuous performance task: Separating response inhibition from premature responding. Behav Brain Res, 222(1): 183-192. Young, S. E., Friedman, N. P., Miyake, A., Willcutt, E. G., Corley, R. P., Haberstick, B. C., & Hewitt, J. K. 2009b. Behavioral disinhibition: liability for externalizing spectrum disorders and its genetic and environmental relation to response inhibition across adolescence. J Abnorm Psychol, 118(1): 117-130. Zahrt, J., Taylor, J. R., Mathew, R. G., & Arnsten, A. F. 1997. Supranormal stimulation of D1 dopamine receptors in the rodent prefrontal cortex impairs spatial working memory performance. J Neurosci, 17(21): 8528-8535. Zelenin, S., Aperia, A., & Diaz Heijtz, R. 2002. Calcyon in the rat brain: cloning of cDNA and expression of mRNA. J Comp Neurol, 446(1): 37-45.

167

168

Nederlandse samenvatting

Achtergrond Het risico op het ontwikkelen van een psychiatrische ziekte, zoals ‘attention deficit hyperactivity disorder’ (ADHD), schizofrenie of verslaving, wordt voor een aanzienlijk deel bepaald door genetische aanleg. Het identificeren van genen die bijdragen aan de ontwikkeling van een psychiatrische ziekte kan inzicht verschaffen in de moleculaire processen ten grondslag aan de ziekte, en daarnaast mogelijk handvatten bieden om de ziekte te behandelen. Ondanks grote genetische associatie studies is het aantal genen dat tot dus ver in verband is gebracht met het risico om deze ziektes te ontwikkelen beperkt gebleven. Bovendien verklaren deze genen tezamen slechts een kleine fractie van het totale genetische risico om de betreffende ziekte te ontwikkelen. De verwachting is dan ook dat er veel genen niet geïdentificeerd zijn gebleven. Psychiatrische aandoeningen worden gediagnostiseerd aan de hand van het optreden van een scala aan symptomen. Er is daarbij een grote heterogeniteit onder patiënten met dezelfde aandoening waarbij niet alle symptomen in dezelfde mate voorkomen in iedere patiënt. Een wetenschappelijk breed gedragen idee is dat verschillende genen ten grondslag liggen aan verschillende ziekte symptomen. Om nieuwe relaties te leggen tussen genen en psychiatrische ziekten wordt daarom voorgesteld om de aandoening op te splitsen in goed gedefinieerde symptomen en te zoeken naar genen die specifiek bijdragen aan deze ziekteverschijnselen. In deze context zijn symptomen van overmatige impulsiviteit en aandachtsproblemen zeer relevant, omdat ze worden waargenomen bij meerdere ziektes, zoals ADHD, schizofrenie en verslaving. Er bestaan uitstekende mogelijkheden om aspecten van aandacht en impulsiviteit in modelorganismen te meten. Het grote voordeel van het uitvoeren van studies in modelorganismen, zoals knaagdieren, is de mogelijkheid om een statistisch verband tussen een gen en een bepaald gedrag experimenteel vast te stellen. Door het ingrijpen in de functie van een gen kan het functionele verband tussen gen en gedrag worden onderzocht, iets dat in humane studies zeer ingrijpend, zo niet onmogelijk is. Het doel van het onderzoek in dit proefschrift was het identificeren van genen die een rol spelen in impulsiviteit en aandacht in knaagdieren en, waar mogelijk binnen de looptijd van het onderzoek, de functionele rol van het betreffende gen nader te onderzoeken.

Hoofdstuk 2 Zowel mensen als dieren devalueren de waarde van een beloning indien die beloning lang op zich zal laten wachten. Bij patiënten met ADHD of een

169

Nederlanse samenvatting verslaving is vastgesteld dat zij gemiddeld genomen minder lang op een grote beloning willen wachten. De neiging om een snelle, kleinere beloning te kiezen wordt omschreven als een impulsieve keuze, een gedrag dat dankzij een operante taak ook in ratten kan worden gemeten. In hoofdstuk 2 heb ik gevonden dat impulsief keuzegedrag van een individuele rat kan worden voorspeld aan de hand van de expressie van dopamine receptor D1 en D5 in de mediale prefrontale cortex (mPFC) van de desbetreffende rat. Deze correlatie, een statistisch verband, heb ik functioneel onderzocht door stoffen, die signalering via dopamine receptor D1 en D5 beïnvloeden, in de mPFC te injecteren voorafgaand aan de taak. De daarop volgende verandering in impulsief keuze gedrag toonde het causale verband aan tussen deze dopamine receptoren en het impulsieve gedrag. Dit was de eerste studie die een verband aantoonde tussen dopamine receptor D1 en D5 in de mPFC en impulsief keuze gedrag. Dit resultaat verschafte het inzicht in de signaalsystemen die impulsieve keuze sturen en daarmee mogelijk een rol spelen in de etiologie van ADHD.

Hoofdstuk 3 Voor kwantificeren van aandacht en impulsiviteit in de mens zijn verschillende gecomputeriseerde responstaken ontwikkeld. In hoofdstuk 3 heb ik deze humane taken vertaald naar bestaande en nieuwe gedragstesten en parameters om aandacht en impulsiviteit in muizen te meten. Ter validatie van deze muizentaken heb ik het effect van amfetamine onderzocht, een stof die aandacht en impulsiviteit in de mens beïnvloed. In de zogenaamde 5-choice serial reaction time task (5-CSRTT) gaat gedurende 1 s een licht stimulus aan één van de vijf response gaten in de wand van een zogenaamde operante kooi. Een correcte respons in het verlichtte gat werd beloond door het verstrekken van een smakelijke voedselbeloning in een bakje in de tegenoverliggende wand van de kooi. Impulsief gedrag in deze taak was gedefinieerd als het responderen vóórdat het stimuluslicht aan gaat. De klassieke maat voor aandacht in deze taak is het percentage correcte responsen. Daarnaast heb ik een nieuwe parameter geïntroduceerd, de variabiliteit in correcte reactie snelheiden, een parameter die zeer gevoelig bleek in verschillende studies met ADHD patiënten versus controles. Daarnaast heb ik een Go/No-Go taak ontwikkeld, waarin muizen werden beloond als ze reageerden op een lichtstimulus (Go-stimulus) in een respons gaatje. Om responsinhibitie te meten werd één op de vijf stimuluspresentaties gekoppeld aan een luide toon (No-Go stimulus). Onder die conditie werd een muis beloond als er géén respons werd gemaakt (inhibitie van responsen). Het aantal niet geïnhibeerde responsen tijdens No-Go stimuli was een maat voor impulsief gedrag. Daarnaast werd wederom in deze taak de variabiliteit in correcte reactie snelheiden gemeten als maat voor aandacht. Twee muizenlijnen, DBA/2J and C57BL/6J, vertoonden verschillen in deze en andere ADHD-gerelateerde gedragsparameters, in het bijzonder in de 5-CSRTT. Daarnaast bleek amfetamine voornamelijk op gedrag in de 5-CSRTT effect te

170

Nederlandse samenvatting hebben. Daaruit concludeerde ik de 5-CSRTT van grote waarde was voor vervolgonderzoek naar de genetische basis aandacht en impulsiviteit.

Hoofdstuk 4 'Novelty seeking' en impulsiviteit worden beiden in verband gebracht met het risico op het ontwikkelen van een verslaving en zouden daarom een overeenkomstige genetische basis kunnen hebben. Er bestaan veel ingeteelde muizenlijnen, waarvan broertjes en zusjes genetisch identiek zijn, vergelijkbaar met eeneïge tweelingen. Deze lijnen zijn uitermate geschikt om de genetische aanleg van een bepaald gedrag te bepalen. In dit hoofdstuk heb ik de mate van novelty seeking en impulsiviteit gemeten in meerdere individuen van 12 genetisch diverse muizenlijnen. Het bleek dat 34% van de fenotypische variatie tussen muizen in de mate van impulsief gedrag werd verklaard door genetische variatie. Ik heb aangetoond dat de mate van novelty seeking geen voorspellende waarde had voor de mate van impulsiviteit in de 5-CSRTT, waaruit ik kon concluderen dat beide vormen van gedrag een verschillende genetische basis hebben. Het is daarom uitermate relevant om voor beide typen van gedrag de onderliggende genen te identificeren, in de verwachting dat die genen bijdragen aan verschillende aspecten van verslavingsgedrag.

Hoofdstuk 5 In mens en muis wordt een luide geluidspuls gevolgd door een schrikreactie waarbij verschillende spieren worden samengetrokken. De intensiteit van de schrikreactie is kleiner dan normaal wanneer ongeveer 30 tot 500 ms vóór de luide puls een zachte puls, de zogenaamde prepulse, wordt gegeven. Dit fenomeen heet prepulse inhibitie (PPI). Schizofrenie wordt in verband gebracht met verminderde PPI en aandachtscapaciteit. In hoofdstuk 5 heb ik de vraag gesteld in hoeverre deze twee symptomen genetisch gerelateerd zijn. Ik heb aangetoond dat de mate van PPI in muizenlijnen (zogenaamde BXD lijnen) geen voorspellende waarde hebben voor de mate van aandacht in de 5-CSRTT. Door gebruik te maken van de bekende genetische architectuur van het cohort BXD lijnen kon ik segmenten op het muizengenoom identificeren, zogenaamde ‘quantitative trait loci’ (QTL), waarin genen liggen die verantwoordelijk zijn voor de gedetecteerde verschillen in gedrag. Ik heb een QTL voor PPI gedetecteerd op chromosoom 17 en een QTL voor aandacht op chromosoom 16. Beide QTL bevatten vele genen. Door gebruik te maken van verschillende databases heb ik een lijst met meest waarschijnlijke kandidaat genen geselecteerd. Vervolgstudies aan zowel PPI, en aan aandacht-gerelateerde genen zullen meer inzicht geven in de verschillende moleculaire mechanismen, die ten grondslag liggen aan psychiatrische ziektes zoals schizofrenie.

171

Nederlanse samenvatting

Hoofdstuk 6 In hoofdstuk 6 heb ik een QTL voor impulsief gedrag in de 5-CSRTT geïdentificeerd op chromosoom 14. In het QTL voor impulsiviteit bevonden zich 4 genen met daarin verschillen in DNA sequentie die leiden tot een verandering in het eiwit. Naast deze 4 mogelijke kandidaten heb ik een gen geïdentificeerd (Nrg3) dat in de mens geassocieerd is met schizofrenie. Bovendien heeft Nrg3 van alle 36 genen in het impulsiviteit QTL de meest significante associatie met de ADHD diagnose, waardoor dit gen een zeer bruikbare kandidaat is voor vervolgstudies. Uit het feit dat het QTL voor aandacht op een ander chromosoom ligt (hoofdstuk 5), in combinatie met de afwezigheid van een significante correlatie tussen impulsiviteit en aandacht, kan worden geconcludeerd dat aandacht en impulsiviteit genetisch niet zijn gerelateerd. De afzonderlijke genetische basis voor aandacht en impulsiviteit in de muis bevestigen het idee dat subgroepen van ADHD patiënten kunnen bestaan met voornamelijk symptomen van aandachtsstoornissen dan wel impulsiviteit, hetgeen van invloed kan zijn op de diagnose van ADHD.

Hoofdstuk 7 Impulsiviteit is vaak in verband gebracht met alcoholverslaving in de mens. In dit hoofdstuk heb ik in een proefdiermodel onderzocht of verhoogde impulsiviteit een oorzaak of een gevolg is van alcohol zoekgedrag. Ik heb de meest impulsieve BXD lijn in de 5-CSRTT (BXD16, zie Hoofdstuk 6) en een controle lijn (C57BL/6J) onderzocht in een nieuwe taak van aandacht en impulsiviteit. Evenals in de 5-CSRTT was BXD16 ook in de nieuwe simpele reactie tijd taak impulsiever dan C57BL/6J en had een verhoogde variabiliteit in reactietijden. Ik heb een tweede groep muizen getraind in een zelfadministratie taak, waarin muizen zichzelf eerst sucrose konden toedienen, gevolgd door een transitie fase, waarna muizen uiteindelijk respondeerden voor een 15% alcohol oplossing. Terwijl BXD16 en C57BL/6J even gemotiveerd waren om zichzelf sucrose toe te dienen, was de motivatie en zoekgedrag voor alcohol hoger in BXD16 dan in C57BL/6J. Ondanks dat aanvullende studies nodig zijn om de genetische relatie tussen impulsiviteit en alcohol zoekgedrag aan te tonen, kon ik voor de eerste keer in de muis vaststellen dat verhoogde impulsiviteit een risicofactor is voor alcohol zoekgedrag.

Conclusie In dit proefschrift heb ik proefondervindelijk kunnen vast stellen dat het opsplitsen van complexe ziektes in duidelijk gedefinieerde symptomen de translatie naar proefdiermodellen mogelijk maakt, en dat dit een efficiënte wijze is om genen en ziekte met elkaar in verband te brengen. Dankzij deze dissectie van complexe ziektebeelden zijn kandidaat genen gevonden voor schizofrenie die ófwel bijdragen aan symptomen van verlaagde PPI ófwel aan symptomen van gebrek aan aandacht. Ook de concepten ‘novelty seeking’ en impulsiviteit bleken

172

Nederlandse samenvatting op genetisch niveau goed te onderscheiden te zijn. Hetzelfde geldt voor de belangrijkste ADHD symptomen, waarvoor ik kandidaat genen heb gevonden, die ófwel bijdragen aan impulsiviteit, ófwel aan aandachtsgebrek. Door het gebruik van modelorganismen zoals muizen en ratten in dit proefschrift, is de vervolgstap, het aantonen van een causaal verband tussen gen en gedrag, binnen handbereik. De QTL en genen geïdentificeerd in dit proefschrift maken het mogelijk een beter inzicht te krijgen in moleculaire processen die ten grondslag liggen aan ziektes zoals ADHD, schizofrenie en verslaving, en brengen daarmee behandelmethoden dichterbij.

173

174

Dankwoord

Ik heb het lang kunnen rekken, maar het lijkt nu echt af te zijn. Van eind 2003 tot en met eind 2009 heb ik mij meer dan 100% kunnen uitleven in de speeltuin die MCN heet. In talloze wetenschappelijke projecten heb ik kennis kunnen maken met een grote verscheidenheid aan onderwerpen en onderzoekstechnieken. Gedurende die tijd heb ik dankzij collega's, vrienden en familie een fantastische tijd gehad. Op zoek naar een goede stage plek begon dit traject in 2001 tijdens een gesprek met Guus. Guus, ik kon toen nog niet vermoeden dat ik een decennium later onveranderd enthousiast op de VU rond zou lopen. Ik ben je dankbaar voor het feit dat je gedurende al die jaren dankzij jouw brede interesse en strategische manoeuvres nieuwe uitdagingen wist te bedenken en realiseren. Daardoor heb ik me in afgelopen jaren in van alles kunnen storten, van hypermoderne proteomics dataset tot en met panels BXD muizen. Sabine, ook jou ben ik zeer dankbaar voor je onuitputtelijke wetenschappelijke enthousiasme en energie. Proefschrift overleg sessies met jullie duurden steevast tot ver na reguliere werktijd en leidden altijd tot nieuwe inzichten en experimenten. Lique, jij hebt mij tijdens mijn stage 2002–2003 essentiële dierexperimentele technieken bijgebracht die ik tot op de dag van vandaag toepas. Marcel, jou enthousiasme voor operante taken is op mij overgeslagen, evenals de XL macro's die daarmee gepaard gingen. Mieke, bedankt voor jouw hulp bij de locale toediening ratten studies en de eerste farmacologische experimenten in de muisjes (wat beten ze ons graag!). Rob en Halfdan, bedankt voor hulp bij experimenten en goede zorg voor dieren. Ton en Taco, bedankt voor jullie kritische blik op farmacologische interventies en zelfadministratie procedures en de manuscripten die daaruit volgden. Leontien bedankt voor jou feedback op dit proefschrift. En tot slot wat betreft de afdeling Anatomy en Neurosciences, Tommy, ik heb genoten van het samen nadenken en opzetten van nieuwe operante taken en onze fietstochten. Ik heb veel geleerd van jouw genuanceerde commentaar en scherpe manier van definiëren tijdens het schrijven van gedragspapers. Oliver, heel erg bedankt voor jouw hulp, en die van Anton en Rene, om Bk ruimtes in orde te krijgen zodat ik daar in 2005 muizen experimenten kon voorzetten. Jouw gestructureerde en grondige manier werken zijn een belangrijk voorbeeld voor mij geweest. Het screenen van de eerste BXD muizen was niets geworden zonder de hulp van Iris. Logistiek en dierverzorging en waren in goede handen bij Carool en Sara. In de periode dat ik geen dierproeven kon doen (eind 2005 – eind 2007) door verbouwingen heb ik leuke tijden beleefd achter de PC samen met Roel, Joost en Pim in een poging proteomics data te doorgronden. Pim, helaas is het bedrijf PiMaLogis nooit van de grond gekomen, maar de pijplijn gelukkig wel! Het

175

Dankwoord opzetten van de productie van viruspartikels in die periode was nooit gelukt zonder hulp William en mensen uit de groep van Joost Verhage, Erich en Ruben in het bijzonder. Yvonne, jij begon ook in 2001 bij MCN, bedankt voor je hulp in het moleculaire lab, tot op de dag van vandaag, met kloneringen, kleuringen, virus productie en niet te vergeten het organiseren van borrels en feesten. Bedankt dat je mijn paranimf wilt zijn! Rolinka, dankzij jouw nuchtere houding en rechttoe rechtaan aanpak waren we ruim vóór de officiële oplevering al met proeven in het MouseHouse begonnen en als vliegende keep heb je veel experimenten ondersteund. Jorn, bedankt dat jij, eerst als supergemotiveerde stagestudent en vervolgens als vaste kracht, de basis hebt gelegd voor grote 5- CSRTT datasets. Christiaan, Bert en andere mensen van het UPC bedankt voor de fok van BXD lijnen. Ruud, jij raakte op een later stadium betrokken bij het werk beschreven in dit proefschrift, bedankt voor de goede verzorging van de muizen en jouw bijdrage aan de experimenten die uit mijn proefschrift volgden. Zonder de hulp van stage studenten Thomas, Antonia, Joost, Youssef, Gert Jan, Jorn, Chavely, Brenda, Martha, Sanne en Stefan zou het onderzoek nog een paar jaar langer geduurd hebben, dank jullie wel! In mijn begin AIO jaren heb ik waardevolle tips gehad van Floor, Eisuke, Edwin, Volker en vooral kamergenoot Michiel die met nietsontziende ironie het leven als AIO kon relativeren. Michel en Désirée, bedankt voor de super gezellige avonden die meestal begonnen in stelling en eindigden in de binnenstad. AIO kamer maatjes Harold en Danielle, ook jullie bedankt voor discussies en ontspanning tijdens koffiepauzes, fietsen en hardlopen. Brigitte, bedankt voor hulp met de papierwinkel die gepaard gaat met promoveren. Dankzij Sophie heb ik de wereld van multivariate statistiek beter leren. Dit proefschrift is mede tot stand gekomen dankzij de kritische inbreng van, samenwerking met, en goede werksfeer gecreëerd door: Andrea, Annemieke, Arjen, Bart, Bastijn, Eva, Evelyn, Jan, Jeroen, Jochem, Joost, Ka Wan, Loek, Marion, Mark, Marlène, Matthijs, Nelleke, Ning, Nutabi, Patricia, Peter, Pieter, Priyanka, Remco, René, Robbie, Ronald, Tariq en andere CNCR en Sylics collega's. Vrienden uit Wageningen en Roosendaal zorgden voor ontspanning tijdens feesten, kampeerweekenden en sport evenementen. Bart en Erik, met jullie kon ik onderzoekssuccessen en strubbelingen delen en dankzij mijn paranimf rol tijdens jullie promotie kon ik wennen aan het idee zelf ook ooit klaar te zijn. Erik, fijn dat je op 10 mei mijn paranimf wilt zijn! David en Pim, als ik jullie zie, altijd een goed gesprek. Paul, altijd in voor een nieuwe uitdaging, waar gaan we skiën? Elles, Bjorn, Claudia, Joyce en Sascha, bedankt voor jullie interesse, steun en gezelligheid gedurende al deze jaren. Sander, vriend, fietsmaatje en ingenieur met goede tips voor programmeren en data analyse, wat ben ik blij dat jij mijn broer bent! Dan mijn ouders, lieve pa en ma, jullie staan aan de basis van alles, hebben mij alle kans gegeven me te ontwikkelen. Bedankt dat jullie tot op de dag van vandaag als één blok achter me staan, wat ik ook doe. Femke en Casper, fijn

176

Dankwoord dat jullie de uitknop van mijn computer feilloos weten te vinden, wetenschap heeft extra kleur gekregen sinds jullie er zijn. Lieve Ilse, mijn reismaatje onderweg naar later, in het begin was jij mijn telefonische hulplijn, daarna mijn thuis, mijn dagelijkse steun en toeverlaat. Zonder jou was dit boekje nog lang niet af gekomen. Schatje, bedankt voor je begrip tijdens weekenden en avonden waarop mijn computer belangrijker leek, er zijn samen nog zoveel avonturen te beleven!

177

Curriculum Vitae

Maarten Loos werd geboren op 25 december 1977 in Roosendaal. Na het behalen van het VWO diploma aan het Norbertus Lyceum te Roosendaal in 1996 begon hij in hetzelfde jaar met een studie Bioprocestechnologie aan de toenmalige Landbouw Universiteit Wageningen. Na het behalen van de propedeuse (cum laude) was hij gedurende een jaar bestuurslid van de sportstichting van de universiteit. Tijdens zijn eerste afstudeervak onderzocht hij de effecten van chronische stress op de hypofyse van varkens aan de leerstoelgroep Fysiologie van mens en Dier in Wageningen onder begeleiding van Dr. E. van der Beek. In zijn twee afstudeervak onderzocht hij het effect van voedsel deprivatie op gen expressie in twee muizenlijnen aan de afdeling Moleculaire en Cellulaire Neurobiologie van de Vrije Universiteit Amsterdam onder begeleiding van Dr. A.B. Smit. Tijdens zijn stage in het lab van Dr. L.M. Coolen aan de University of Cincinnati, Ohio in de Verenigde Staten bestudeerde hij de rol van de mediale prefrontale cortex in seksueel gedrag van mannelijke ratten. In 2003 studeerde hij als ingenieur af aan Wageningen Universiteit. Van 2003 tot en met 2009 deed Maarten zijn promotie onderzoek onder begeleiding van Prof. A.B. Smit, Dr. S. Spijker en Dr. T. Pattij aan de genetische basis van impulsiviteit en aandacht in ratten en muizen. De resultaten van dat onderzoek zijn beschreven in dit proefschrift. Van 2010 tot en met 2011 combineerde hij academisch onderzoek aan de afdeling Moleculaire en Cellulaire Neurobiologie van de Vrije Universiteit Amsterdam met onderzoek binnen het bedrijf Sylics (handelsnaam van Synaptologics B.V). Sinds 2012 werkt Maarten full time als scientific officer van Sylics.

178

Peer-reviewed publications

Islinger, M., Luers, G. H., Li, K. W., Loos, M., & Volkl, A. 2007. Rat liver peroxisomes after fibrate treatment. A survey using quantitative mass spectrometry. J Biol Chem, 282(32): 23055-23069. Li, K. W., Miller, S., Klychnikov, O., Loos, M., Stahl-Zeng, J., Spijker, S., Mayford, M., & Smit, A. B. 2007. Quantitative proteomics and protein network analysis of hippocampal synapses of CaMKIIalpha mutant mice. J Proteome Res, 6(8): 3127-3133. Pattij, T., Janssen, M. C., Loos, M., Smit, A. B., Schoffelmeer, A. N., & van Gaalen, M. M. 2007. Strain specificity and cholinergic modulation of visuospatial attention in three inbred mouse strains. Genes Brain Behav, 6(6): 579-587. Loos, M., van der Sluis, S., Bochdanovits, Z., van Zutphen, I. J., Pattij, T., Stiedl, O., Smit, A. B., & Spijker, S. 2009. Activity and impulsive action are controlled by different genetic and environmental factors. Genes Brain Behav, 8(8): 817-828. Rajcevic, U., Petersen, K., Knol, J. C., Loos, M., Bougnaud, S., Klychnikov, O., Li, K. W., Pham, T. V., Wang, J., Miletic, H., Peng, Z., Bjerkvig, R., Jimenez, C. R., & Niclou, S. P. 2009. iTRAQ-based proteomics profiling reveals increased metabolic activity and cellular cross-talk in angiogenic compared with invasive glioblastoma phenotype. Mol Cell Proteomics, 8(11): 2595-2612. Davis, J. F., Loos, M., Di Sebastiano, A. R., Brown, J. L., Lehman, M. N., & Coolen, L. M. 2010. Lesions of the medial prefrontal cortex cause maladaptive sexual behavior in male rats. Biol Psychiatry, 67(12): 1199- 1204. Islinger, M., Li, K. W., Loos, M., Liebler, S., Angermuller, S., Eckerskorn, C., Weber, G., Abdolzade, A., & Volkl, A. 2010. Peroxisomes from the heavy mitochondrial fraction: isolation by zonal free flow electrophoresis and quantitative mass spectrometrical characterization. J Proteome Res, 9(1): 113-124. Klychnikov, O. I., Li, K. W., Sidorov, I. A., Loos, M., Spijker, S., Broos, L. A., Frants, R. R., Ferrari, M. D., Mayboroda, O. A., Deelder, A. M., Smit, A. B., & van den Maagdenberg, A. M. 2010. Quantitative cortical synapse proteomics of a transgenic migraine mouse model with mutated Ca(V)2.1 calcium channels. Proteomics, 10(13): 2531-2535. Loos, M., Pattij, T., Janssen, M. C., Counotte, D. S., Schoffelmeer, A. N., Smit, A. B., Spijker, S., & van Gaalen, M. M. 2010a. Dopamine Receptor

179

Peer-reviewed publications

D1/D5 Gene Expression in the Medial Prefrontal Cortex Predicts Impulsive Choice in Rats. Cereb Cortex, 20(5): 1064-1070. Loos, M., Staal, J., Schoffelmeer, A. N., Smit, A. B., Spijker, S., & Pattij, T. 2010b. Inhibitory control and response latency differences between C57BL/6J and DBA/2J mice in a Go/No-Go and 5-choice serial reaction time task and strain-specific responsivity to amphetamine. Behav Brain Res, 214(2): 216-224. Van den Oever, M. C., Lubbers, B. R., Goriounova, N. A., Li, K. W., Van der Schors, R. C., Loos, M., Riga, D., Wiskerke, J., Binnekade, R., Stegeman, M., Schoffelmeer, A. N., Mansvelder, H. D., Smit, A. B., De Vries, T. J., & Spijker, S. 2010. Extracellular matrix plasticity and GABAergic inhibition of prefrontal cortex pyramidal cells facilitates relapse to heroin seeking. Neuropsychopharmacology, 35(10): 2120- 2133. Counotte, D. S., Goriounova, N. A., Li, K. W., Loos, M., van der Schors, R. C., Schetters, D., Schoffelmeer, A. N., Smit, A. B., Mansvelder, H. D., Pattij, T., & Spijker, S. 2011. Lasting synaptic changes underlie attention deficits caused by nicotine exposure during adolescence. Nat Neurosci, 14(4): 417-419. Dahlhaus, M., Wan Li, K., van der Schors, R. C., Saiepour, M. H., van Nierop, P., Heimel, J. A., Hermans, J. M., Loos, M., Smit, A. B., & Levelt, C. N. 2011. The synaptic proteome during development and plasticity of the mouse visual cortex. Mol Cell Proteomics, 10(5): M110 005413. Guillem, K., Bloem, B., Poorthuis, R. B., Loos, M., Smit, A. B., Maskos, U., Spijker, S., & Mansvelder, H. D. 2011. Nicotinic acetylcholine receptor beta2 subunits in the medial prefrontal cortex control attention. Science, 333(6044): 888-891. Jansen, R., Timmerman, J., Loos, M., Spijker, S., van Ooyen, A., Brussaard, A. B., Mansvelder, H. D., Smit, A. B., de Gunst, M., Linkenkaer-Hansen, K., & The Neuro-Bsik Mouse Phenomics, C. 2011. Novel Candidate Genes Associated with Hippocampal Oscillations. PLoS One, 6(10): e26586. Klemmer, P., Meredith, R. M., Holmgren, C. D., Klychnikov, O. I., Stahl-Zeng, J., Loos, M., van der Schors, R. C., Wortel, J., de Wit, H., Spijker, S., Rotaru, D. C., Mansvelder, H. D., Smit, A. B., & Li, K. W. 2011. Proteomics, ultrastructure, and physiology of hippocampal synapses in a fragile x syndrome mouse model reveal presynaptic phenotype. J Biol Chem, 286(29): 25495-25504. Loos, M., Staal, J., Pattij, T., Smit, A. B., & Spijker, S. 2011. Independent genetic loci for sensorimotor gating and attentional performance in BXD recombinant inbred strains. Genes Brain Behav, In Press.

180