The goal of this review is to acquaint the reader with the newly established vertebrate model organism for genetics, the zebrafish Danio rerio, and the opportunity it provides us to understand the molecular and cellular mechanisms of behavior and behavioral disorders. I will start by presenting the salient features of zebrafish for genetic study, the powerful molecular genetic tools developed in zebrafish, and the impact of sequencing the whole genome. I will go on to discuss the use of zebrafish to study sensory perception and locomotor regulation by outlining the behavioral assays developed for large-scale genetic screening. I will focus on several key mutants that affect either the development or function of specific neuronal circuits involved in the behavior. These parts will be brief, as they have been recently reviewed elsewhere (Baier 2000; Drapeau et al. 2002; Goldsmith & Harris 2003; Kratz et al. 2002; Li 2001; Neuhauss 2003; Whitfield 2002). Finally, I will discuss the use of zebrafish to study more complex behaviors including goal-directed behavior (motivation and reward), emotions (anxiety and fear), learning and memory and the role of brain monoamine systems in modulating these behaviors.
How our brain is wired and subsequently generates functional output, ranging from sensing and locomotion to emotion, decision-making and learning and memory, remains poorly understood. Dys-regulation of these processes can lead to neurodegenerative, as well as neuro-psychiatric, disorders. Molecular genetic together with behavioral analyses in model organisms identify genes involved in the formation of neuronal circuits, the execution of behavior and mechanisms involved in neuro-pathogenesis. In this review I will discuss the current progress and future potential for study in a newly established vertebrate model organism for genetics, the zebrafish Danio rerio. Where available, schemes and results of genetic screens will be reviewed concerning the sensory, motor and neuromodulatory monoamine systems. Genetic analyses in zebrafish have the potential to provide important insights into the relationship between genes, neuronal circuits and behavior in normal as well as diseased states.
Zebrafish: a new vertebrate genetic model organism for understanding neural development, function and disease
Zebrafish, a freshwater tropical teleost fish common to most pet stores, was first chosen for laboratory study by the late George Streisinger about 30 years ago. His vision was to apply molecular genetics to unravel neural development down to single cell and single molecule in a vertebrate (for a review of detailed historical perspective, see Grunwald & Eisen 2002). Why did Streisinger choose zebrafish? The reasons are quite obvious today: zebrafish is a diploid vertebrate with a good balance of complexity and simplicity. It is a small creature of only 3–5 cm in length, and reproduces robustly. These properties make it easy to maintain a large number of animals in a relatively small space, a prerequisite for carrying out large-scale genetic study. Such a prerequisite is found in invertebrate model organisms such as Caenorhabditis elegans and Drosophila, but is lacking in vertebrate model organisms such as mice. The embryo and larva of zebrafish are transparent and develop amazingly rapidly: in merely 5 days, swimming, self-feeding larvae, also known as fry, emerge from fertilized zygotes. This entire process unravels in a Petri dish (Fig. 1).
Since the introduction of zebrafish into the laboratory, many milestones have been achieved that firmly establish zebrafish as a prominent genetic model organism for biology and medicine. Streisinger's pioneering work on establishing homozygous diploid clones and lethal-free strains (Streisinger et al. 1981), and pilot mutagenesis screening (Walker & Streisinger 1983), laid an important foundation for later large-scale genetic screens. Therein ∼4000 embryonic lethal mutations were identified that affect virtually all aspects of visible developmental processes (Driever et al. 1996; Haffter et al. 1996). In addition to genetics, cell biological and anatomical studies uncovered the segmental structure of the brain and identified the role of cell lineage in generating diverse groups of neurons (Kimmel 1993; Kimmel et al. 1990).
The goal of forward genetic screens is to reveal the molecular identity of genes affected by mutations (Fig. 2). Genomic technologies such as RAPD (Rapid Amplification of Polymorphic DNA) and AFLP (Amplified Fragment Length Polymorphism) have been used to positionally clone genes defined by mutations (Donovan et al. 2000; Guo et al. 2000; Zhang et al. 1998). A genetic linkage map composed of more than 4000 polymorphic microsatellite markers (Knapik et al. 1998) and two radiation hybrid maps composed of both microsatellite markers and ESTs (Expressed Sequence Tags) (Geisler et al. 1999; Hukriede et al. 1999) have been constructed. A syntenic relationship between zebrafish and mammalian species has been determined (Barbazuk et al. 2000; Woods et al. 2000). Recently, rapid mapping of zebrafish mutations with SNPs (Single Nucleotide Polymorphisms) and oligonucleotide microarrays has also been developed (Stickney et al. 2002). These technological advancements greatly facilitate gene identification from mutations. In addition, tools for insertional mutagenesis have been established to isolate insertional mutants that could be rapidly cloned (Amsterdam et al. 1999). Complementing forward genetic analysis, reverse genetic methods are also in place. These include the morpholino antisense knockdown, which is a simple method to inactivate known genes in early embryos (Nasevicius & Ekker 2000). Recently, TILLING (Targeted-Induced Local Lesions IN Genomes) (McCallum et al. 2000), a method originally developed in Arabidopsis, has been successfully adapted to zebrafish (Wienholds et al. 2002). Mutations ranging from null (complete loss of function) to hypomorphic (partial loss-of-function) alleles can be isolated in virtually any gene of interest (Fig. 3). Finally, both forward and reverse genetic studies in zebrafish will be greatly facilitated by the ongoing Sanger Center zebrafish genome sequencing project, which is due to be completed in the year 2003. Sequencing data are freely available in GenBank as well as at the Sanger Center web site and a centralized web-based database (ZFIN) (Sprague et al. 2001). Taken together, zebrafish hold great potential for studying developmental processes, organ function and human diseases (Dooley & Zon 2000; Shin & Fishman 2002).
Zebrafish are highly suitable for studying behavior to elucidate the role of genes in the formation and function of neuronal circuitry. Zebrafish embryos (0–5 days old) already exhibit simple sensory and locomotor abilities. Zebrafish larvae (5 days to 2-weeks old) are free-living and need to hunt for food and escape from predators, thus they possess many patterns of behavior. Yet both embryos and larvae are optically transparent, making them highly accessible to cellular study of neuronal circuitry. In fact, single neuronal connection or activity can be directly visualized with the utilization of fluorescent proteins, and single neuron ablation can also be achieved with a laser (Fetcho & Liu 1998).
Genes themselves do not generate behavior. Rather, genes control the development and function of neuronal circuits, which produce behavior. In general, genetic analysis of behavior is expected to identify two classes of mutations: one affects the formation (development) of neuronal circuits, and the other alters their function. Both classes of mutations, when specific enough, will provide important insights into the neuronal circuitry that underlie behavior. Therefore, in the following paragraphs, both classes of mutations will be discussed in the context of specific behavior.
It is also important to keep in mind that the expressivity or function of genes, as well as the connectivity patterns of neuronal circuits, can be modified by the external environment, leading to behavioral variability and plasticity (Crabbe et al. 1999). In addition, genetic background can also have a strong influence on behavior (Gerlai 1996). Therefore, when designing behavioral assays, the factors of environment and genetic background should be taken into consideration. Although an ideal behavioral assay that is suitable for large-scale genetic screens should be robust enough to show minimal sensitivity to these factors, such an assay is not always obtainable. Therefore, genetic background and environmental factors need to be tightly monitored and controlled during the study of behavior. In zebrafish, several strains of different genetic background are available. AB is a laboratory strain that has been bred for many generations in the Oregon lab and may be a good choice for behavioral study. Other strains include Wik, Tu, TL and India (Johnson & Zon 1999). These strains show high interstrain genetic polymorphisms, and therefore are likely to exert different behavior.
In summary, the zebrafish offers a unique advantage for large-scale genetic screens to be carried out in a vertebrate. Compared to Drosophila and C. elegans, the nervous system of zebrafish is more comparable to that of humans. Compared to mice, zebrafish is more amenable to forward genetic study, which has proven difficult and costly to carry out in mice. In the following paragraphs I will discuss the behavioral genetic study to understand multiple sensory systems, the motor systems and the neuromodulatory monoamine systems. Behaviors that are largely reflexive in nature are considered relatively simple behaviors, whereas non-reflexive behaviors, which are generally termed here as ‘goal-directed behaviors’, are considered more complex behaviors.
Vision, olfaction and mechanosensory transduction
Contact with the external environment is through the sensory systems, including eyes, nose, ears and peripheral nerves. Sensing the surroundings is the first step toward the generation of complex behavior. Studies in zebrafish are providing considerable insights into the formation of sensory circuits, and are holding great potential for elucidating the function of sensory systems at an organismal level.
The visual system
Among all the sensory systems in zebrafish, the visual system is the best studied, due to its great accessibility. Mutants that affect the development as well as function of the visual system have been isolated through both morphological and behavioral screens (Baier et al. 1996; Brockerhoff et al. 1995; Li & Dowling 1997; Malicki et al. 1996a; Neuhauss et al. 1999).
An interesting observation was made that blind fish tend to have a darker appearance compared to normal fish when placed in a light background (Neuhauss et al. 1999). This appearance is particularly easy to see in larval zebrafish. As for many lower vertebrates, zebrafish possess the ability to change their appearance as a means of environmental adaptation. That is, the melanin pigments (melanosomes) in their pigment cells (melanophores) disperse or aggregate in response to the intensity of light. This cellular behavior is mediated by retinal-hypothalamic projections, which in turn regulate the secretion of two pituitary hormones that control melanosome movements (Balm & Gronevald 1998). Because of the inability to sense light, blind fish adapt to have a dark appearance. This simple assay can be used as a primary screening method for visual impairment in larval zebrafish.
In addition to screening for dark fish, two other robust and perhaps more specific and quantifiable assays for vision are the optokinetic response (OKR) and optomotor response (OMR), which are largely applied to larval zebrafish. In the OKR assay, the fish are partially immobilized with methyl cellulose in a Petri-dish, which is placed inside a rotating drum. A smooth pursuing eye movement in the direction of the moving objects followed by a rapid saccadic eye movement is consistently observed in normal fish (Clark 1981). In the OMR assay, fish swim in the direction of the moving objects when placed in a circular container surrounded with rotating black and white stripes (Clark 1981). Recently, computer-generated images (black and white stripes) are presented below a chamber containing freely moving larval zebrafish. It is observed that larval fish also swim in the direction of the moving stripes. As a consequence, fish congregate towards the other end of the chamber (Neuhauss et al. 1999). Both OKR and OMR are robust, i.e. ∼90% fish will respond to the stimuli; therefore, these assays are highly suitable for genetic screens.
A behavioral test based on the visually mediated escape response has been developed to quantitatively measure visual sensitivity in adult zebrafish (Li & Dowling 1997). The adult fish is confined to a circular container with a pole in the center. Upon visually encountering a black segment rotating outside the container (serving as a threatening object), the fish immediately turn and rapidly swim away from the black segment. ∼80–90% of time the fish will respond to the stimuli in a positive way (i.e. an escape response is evoked).
These behavioral assays share simple and robust features. Genetic screens have been conducted using these assays. A number of interesting mutants have been isolated. Among them, a mutation named lakritz (lak) was originally identified to have a dark appearance (Kelsh et al. 1996), and later found to be defective in both OMR and OKR assays (Neuhauss et al. 1999). The behavioral defect of lak is caused by a specific developmental deficit of retinal ganglion neurons due to a mutation in the ath5 gene, a vertebrate homologue of the Drosophila bHLH gene atonal (Kay et al. 2001). In Drosophila, the atonal gene is required for the development of photoreceptors in the compound eye and peripheral chordotoal organs (Jarman et al. 1993; Jarman et al. 1994). In vertebrates including zebrafish, xenopus and mice, ath5 is required to determine retinal ganglion neuron differentiation. Although carrying out seemingly different functions in Drosophila and vertebrates, the biochemical mechanism of action of the atonal/ath5 gene appears to be highly conserved. Another mutant (belladonna) displays an OKR behavior that is directionally reversed. Cellular study showed that optic chiasm failed to form in the mutant, suggesting the possible underlying circuitry defect (Rick et al. 2000). In addition to larval stage-based screens, a genetic screen in adult zebrafish has identified dominant mutations that lead to night blindness (nb) (Li & Dowling 1997). Heterozygous nba and nbb mutations have specific visual defects due to adult onset retinal neurodegeneration, whereas homozygous mutations for both loci are embryonic lethal due to widespread neurodegeneration.
Can zebrafish mutations be useful to understand human visual disorders? The phenotypes of mutant zebrafish have shown resemblances to certain conditions of human retinal disorders. For example, the nagie oko mutant, which has an eye and brain patterning defect, showed a similarity to the oculo-cerebro-renal human disorder and encoded a membrane-associated guanylate kinase (Wei & Malicki 2002). This suggests that the studies of zebrafish mutations might provide insights into human retinal neurodegeneration.
The olfactory system
Besides vision, another way to sense the environment is through smell. Olfactory perception involves odorant interaction with receptors that are expressed on the surface of odorant receptor (OR) neurons in the olfactory epithelium of the nose. About 100 genes that encode ORs have been identified in zebrafish (Kratz et al. 2002; references therein). Four main classes of water-soluble odorants are detectable by the fish olfactory system: amino acids, gonadal steroids, bile acids and prostaglandins (Laberge & Hara 2001).
Electrophysiological and activity-dependent labeling techniques have been employed to investigate the capability of zebrafish to detect different odorants. Recently, polyamines as well as the monoamine histamine have been shown to elicit significant olfactory responses, and it was found that a novel transduction pathway that is distinct from the ones used by amino acids or bile acids may mediate the response to polyamines (Michel et al. 2003). In addition, optical imaging of neuronal activity in the olfactory bulb has been employed to determine the structural requirement and minimal number of different zebrafish olfactory receptors that respond to a series of naturally occurring amino acids and structurally related compounds. This analysis suggests that odorant detection requires several different receptors even for relatively simple odorants, and individual receptors require the presence of certain molecular features represented in the odorant molecules (Fuss & Korsching 2001).
Based on the effects on animal behavior, odorants can also be classified as chemo-attractants, repellents or neutral stimulants. Zebrafish are attracted to amino acids and repelled by copper (Steele et al. 1990). How different odorants elicit attractive or aversive behavior in vertebrates is likely to be complex and involve not only the sensory system but also neuromodulatory as well as motor systems. Saturating genetic screens of chemoattractive/chemorepellent behavior has been carried out in C. elegans (Bargmann et al. 1993), which identified highly conserved odorant receptor genes and elucidated sensory neuron axon guidance. However, genes and pathways that are involved in mediating the attraction/avoidance behaviors have not been identified in C. elegans. In addition, the simple C. elegans nervous system is quite different from the vertebrate olfactory system, which contains the olfactory epithelium, the bulb and higher processing centers. Therefore, future genetic study of chemoattractive or chemorepellent behavior in zebrafish may provide important insights into the understanding of molecules and neural network in mediating such attraction/avoidance behavior.
Hearing and mechanosensation
The vertebrate inner ear is the organ of hearing and balance. Defects within the neuroepithelium of the inner ear can lead to congenital deafness in humans. Although the zebrafish ear does not contain the specialized hearing organ cochlea, other features are preserved in zebrafish. Specialized sensory hair cells are found within the zebrafish inner ear. In addition, hair cells are also present in the lateral line neuromasts that are involved in balancing and detecting water flow. A genetic analysis of inner ear development and function will provide insights into understanding human deafness as well as mechanisms underlying mechanosensory transduction.
Since the zebrafish ear is readily visible in developing embryos, simple screens for defects in ear morphology have been carried out (Malicki et al. 1996b; Whitfield et al. 1996). In addition, behavioral assays were also employed: since the inner ear functions in maintaining balance, mutations that show defects in balance were isolated (Granato et al. 1996). The ‘circler’ type of zebrafish mutants have trouble keeping their postural balance, often swim on their sides or back, and many fail to exhibit the acoustic/vibrational startle reflex. The acoustic/vibrational startle reflex is present in 72-h-old larvae and involves a relatively simple neural circuit: the sensory hair cells activated by auditory or vibrational stimuli cause the firing of the eigth cranial sensory ganglia, which activate the Mauthner neurons. The Mauthner neurons subsequently activate motor neurons that output the startle response. Monitoring the activity of these neurons through calcium imaging analysis revealed that in many of the circler mutants these neurons do not display calcium signals in response to vibration but do so in response to touch. This analysis suggests that these circler mutants have defects in the periphery involved in sensing vibration, whereas the hindbrain neurons are likely to be normal. Gross and fine structure analysis of inner ear as well as recordings from the lateral line hair cells allowed further classification of circler mutants into five different classes (Nicolson et al. 1998). Among the circler mutants, the molecular identity of mariner has been revealed to encode myosin VIIA (Ernest et al. 2000). Mutations in this gene are also linked to human deafness, the Usher 1B syndrome, thus validating zebrafish as a model for human hearing disorders. Recently, a morpholino-based reverse genetic analysis of a zebrafish homologue of the Drosophila NompC TRP channel revealed that it plays an essential role in vertebrate sensory hair cell mechanotransduction (Sidi et al. 2003).
In addition to screening for balancing mutants, a dominant screen for hearing defects has been conducted (Bang et al. 2002). The assay monitors a rapid escape reflex in response to a loud sound. Over 6500 fish were tested, and 1% of them were found to be non-responsive. Subsequent X-ray imaging analysis revealed defects in conductive elements of the auditory system. However, none of the putative mutations were transmitted to the next generation, suggesting that these mutations were somatic rather than germ-line in nature. It is possible that a loss of two copies of a particular gene is necessary to produce a heritable hearing mutant phenotype; therefore, a recessive screen using the assay may turn out to be fruitful.
The motor system
The first motor behavior of zebrafish is a spontaneous contraction at 17 h after fertilization, and it depends upon a simple spinal network consisting of fewer than 20 primary motor-, sensory- and interneurons (Saint-Amant & Drapeau 1998; Saint-Amant & Drapeau 2001). By 27 h old, embryos respond to a gentle touch by swimming away for a short distance. More frequent spontaneous beat-and-glide swimming begins at 5 days. In addition, 5-day-old larvae acquire sensory cue-induced reflective movement such as the optomotor response (visual cue) and escape behavior (mechanosensory cue), as well as goal-directed motion (e.g. feeding, and escape from predators).
Calcium imaging in combination with laser lesioning has been employed to study neuronal circuits regulating larval escape in response to tactile stimuli (Liu & Fetcho 1999; O'Malley et al. 1996). Calcium indicators loaded into hindbrain neurons through back filling allowed the monitoring of the activity of labeled neurons in living zebrafish. Furthermore, specific laser killing of selected reticulospinal neurons leads to delayed behavioral response, functionally validating the role of these neurons in the behavior.
A mutant named space cadet was isolated for its locomotion defect (Lorent et al. 2001): normal larvae exhibit a C-bend pattern of escape upon tactile or vibrational stimuli, followed by accelerated swimming through a series of fast, bilateral tail flextures. However, the space cadet mutant often performs multiple C-bend towards the same side, resulting in circumferential rotation. In addition, the mutant displays repeated tail flextures to the same side. Detailed cellular study revealed that the motor defect is correlated with axonal defects in a small population of commissural hindbrain spiral fiber neurons. Severing their axons in the wildtype can phenocopy the mutant phenotype, strongly suggesting that the space cadet gene is involved in the development of the circuitry modulating fast turning. An additional commissural axonal defect in the retina suggests that space cadet may have a general role in commissural neuron axonal pathfinding.
One has to keep in mind that in addition to the neural system, body structures or muscle strength can affect movement. The effect of fin size was recently found to affect the swimming performance, swimming behavior and routine activity of zebrafish (Plaut 2000). Therefore, in a genetic screen that uses locomotion as a read-out, both neural and non-neural mechanisms need to be taken into consideration.
Neuromodulation by brain monoamine systems
Behaviors such as OKR, OMR and escape in response to touch are mostly reflexive in nature, thus relatively simple and usually involving simpler neuronal circuits. In contrast, behaviors that are goal-directed (driven by the ‘intention’ to survive or reproduce, e.g. food hunting, escape from predators) or emotion-related (e.g. aggression, anxiety and fear) are considerably more complex. In addition to the requirement of sensory or motor systems, neuromodulatory systems (e.g. dopamine, noradrenaline and serotonin systems) play important roles.
Dopamine, noradrenaline and serotonin belong to the family of monoamines (Bjorklund & Hokfelt 1984). They are synthesized from the dietary amino acids tyrosine (for dopamine and noradrenaline) and tryptophan (for serotonin) through a series of enzymatic reactions (Fig. 4). The key enzymes involved in the synthesis of dopamine and noradrenaline are TH (tyrosine hydroxylase) and d.b.h. (dopamine beta-hydroxylase). TPH (tryptophan hydroxylase) is involved in the synthesis of serotonin. L-AADC (L-aromatic amino acid decarboxylase) is involved in synthesizing all three monoamine neurotransmitters.
In the vertebrate central nervous system, the dopaminergic (DA) neurons are located in the basal forebrain (mainly the hypothalamus), midbrain, olfactory bulb and retina. Noradrenergic (NA) neurons are mainly present in the locus coeruleus, the brain noradrenergic center. Serotonergic (5HT) neurons are found in the basal forebrain (mainly the hypothalamus) and hindbrain (raphe nuclei). These neurons innervate many different regions of the brain and spinal cord, and play important modulatory roles in regulating locomotor coordination, neuro-endocrine and vital organ function, motivation and reward, emotional balance, mood, attention, planning and social behavior (Goldstein et al. 1998). Dysfunction of monoamine systems in humans is associated with movement disorders, drug and alcohol addiction, anxiety and depression and schizophrenia. In the following paragraphs I will discuss our work on studying the effects of drugs of abuse on zebrafish and anxiety-like behavior in zebrafish, as well work from other laboratories on several complex behaviors.
Behavior induced by drugs of abuse in zebrafish
Addictive substances (e.g. drugs of abuse, alcohol) possess powerful reinforcing properties that lead to compulsive seeking behavior. In mammals, the brain dopamine system, mainly the ventral tegmental area (VTA) and their projection territory, namely the nucleus accumbens (NAC), are heavily implicated in the process. Although a theoretical framework exists, the molecular and cellular basis of drug-seeking is not well understood. Can zebrafish genetics be used to dissect the molecular components as well as cellular circuitry involved in drug-seeking behavior?
Effects of drugs of abuse on locomotor activity
Previous studies have shown that many drugs of abuse can lead to hyperlocomotor activity when administered acutely, and such hyperlocomotor activity may involve the same brain dopamine system that mediates the rewarding effect of drugs (Koob et al. 1998; Phillips & Shen 1996). Therefore, one way to study the effects of drugs of abuse is to examine the locomotor behavior upon their acute or chronic administration. Genetic screens for locomotor sensitivity to alcohol have been carried out in Drosophila, and a mutation named cheapdate (allelic to amnesia, a previously isolated mutant defective in olfactory learning) was identified (Moore et al. 1998). The mutation disrupts a neuropeptide that is involved in the activation of the cAMP pathway.
Compared to Drosophila, the nervous system of zebrafish is more comparable to that of humans. If simple and robust assays that assess the effects of drugs of abuse can be developed in zebrafish, subsequent genetic study could provide important insights into the molecular and cellular mechanisms underlying addiction. Administrations of alcohol to adult zebrafish lead to dose-dependent modification of locomotor activity (Gerlai et al. 2000): at low or intermediate doses, zebrafish exhibit hyperactivity, whereas at high doses, zebrafish become hypoactive and intoxicated. In addition to affecting locomotor activity, alcohol also modifies a number of interesting behaviors including aggression, social behavior, and antipredatory behavior. The effects of alcohol are found to be strain-dependent, suggesting that genotypic difference is accountable for alcohol's effects (Dlugos & Rabin 2003). Besides acute effects, chronic ethanol exposure was examined in Wildtype (in this case, fish with steel-blue body stripes that were purchased from commercial sources) and LFS fish (a strain with long dorsal and pectoral fins). It was found that chronic exposure to 0.5% ethanol for two weeks caused the wild type to swim in a less clustered pattern, whereas LFS fish appeared more clustered. It will be interesting to determine whether chronic ethanol exposure can lead to tolerance, as has been observed in mammals as well as Drosophila (Scholz et al. 2000).
Since larval zebrafish are much easier to handle in behavioral studies and genetic screens, and larval zebrafish are particularly amenable to the study of cellular circuitry, the response of larval zebrafish to alcohol was recently explored (S. Bjerke, B. Lockwood, K. Kobayashi & S. Guo submitted). Larval zebrafish are much less sensitive to handling. They are small-sized and exhibit little social interaction; therefore, a group of larval zebrafish can be tested in the same tank. Like adult zebrafish, larval zebrafish exhibit acute sensitivity to alcohol in a dose-dependent manner. Furthermore, the response is sensitive to genetic background, suggesting that it is genetically modifiable. Ethanol-induced hyperlocomotor activity can be blocked by the addition of a dopamine antagonist, suggesting the involvement of the brain dopamine system. Similar dopaminergic regulation of ethanol-induced hyperlocomotor activity has been observed in rodents and Drosophila (Bainton et al. 2000; Phillips & Shen 1996). In addition, it is also observed in larval zebrafish that alcohol can modify melanosome dispersion in pigment cells. Melanosome dispersion can be regulated through background adaptation. It is not clear whether such a response is mediated through the central nervous system, or whether it is mediated through the direct effect of alcohol on pigment cells. Nevertheless, it provides a direct and robust measure of the biological effects of ethanol in vivo. The finding that larval zebrafish exhibit similar alcohol-induced behavior as adult fish provides an easy means to explore the genetic basis of the biological effects of alcohol.
Effects of drugs of abuse on preference behavior
Although locomotor response to drugs of abuse is a simple way of measuring their biological effects, a more direct way to measure motivation and reward is to examine their preference behavior. Conditioned place preference (CPP) is a relatively simple assay for reward-related behavior (Tzschentke 1998). In CPP, an animal learns to prefer an initially neutral set of stimuli (conditioned stimuli, CS) that is paired with drug exposure (unconditioned stimuli, UCS), and the preference is assessed by the time difference the animal spends in the drug-paired side before and after the conditioning sessions (Fig. 5). In addition to CPP, the self-administration paradigm has been developed in rodents, in which the rewarding effect of a drug is reflected through the animal's action of direct drug intake (Tzschentke 2001).
Using the CPP assay, cocaine directly administered in the tank water was shown to elicit a preference response in adult zebrafish (Darland & Dowling 2001). Only a single trial of visual-cue paired drug exposure is required to elicit a preference. Maximal preference (∼15%) was observed with 10 mg/l cocaine concentration, whereas both lower and higher concentrations of cocaine elicited a smaller preference response. Similarly, zebrafish were found to exhibit preference toward morphine (S. B. Bretaud and S. G. Guo, unpublished data).
Anxiety and fear
Anxiety and fear are emotional states generated in response to perceived or real threats, respectively. Within normal ranges, these emotional states serve to protect individuals from dangerous situations. However, disorders arise when anxiety and fear are in excess and out of control. Anxiety disorders are quite heterogeneous and many forms have been diagnosed in humans. More than 19 million people in the US alone are affected by one form or another. Compounds that are used to reduce anxiety in humans act on GABA, serotonin and dopamine systems (Lister 1990).
Adequate animal models are needed to study the molecular genetic basis of anxiety and fear. In developing such animal models, it is important to keep in mind that since anxiety is a very complex emotion and involves many different neural systems, a particular behavioral paradigm in an animal model is unlikely to truly mimic the complex human disorders. Therefore, the goal of animal study shall be to develop paradigms that assess anxiety-like behavior in animals and study the molecular and cellular mechanisms in the context of the specific animal behavior. Genes and circuitry identified in animal study will provide candidates for subsequent investigation in humans.
Over the years, several behavioral paradigms have been established in rodents that are likely to measure the state of anxiety or fear in the tested animal, as drugs that are known to reduce human anxiety can also modify such behavior in animals (Finn et al. 2003; Lister 1990). Some of these behavioral paradigms test exploratory behavior (open field, the plus-maze test and light-dark transition), as animals tend to explore less when in a state of anxiety or fear. Other paradigms also measure social behavior, as anxiety or fear decreases social interaction.
Given the strength of zebrafish for forward genetic studies, can it be used to dissect the molecular and cellular basis of anxiety-like behavior? Recent studies have started to address this question. It has been previously noted that adult zebrafish show a natural preference for the dark environment when tested in a tank that is half-white and half-black (Serra et al. 1999). A quantitative test showed that both adult and larval zebrafish exhibit ∼80% preference for the dark side in a light/dark preference chamber. Furthermore, anxiolytic compounds can modify such behavior (B. L. Lau and S. G. Guo, unpublished data).
Other complex behaviors of zebrafish
Cereberal lateralization, once thought to be a human-specific feature, has been found to be widespread in vertebrates (Vallortigara et al. 1999). Behavioral lateralization is revealed in zebrafish by preferential eye use (Miklosi & Andrew 1999; Miklosi et al. 1998; Miklosi et al. 2001). The way zebrafish view a range of different objects and react to these objects was behaviorally recorded. It was found that the right eye was used at the first encounter of an object (e.g. a colored bead), during which a decision has to be made regarding what action to take (e.g. to bite or not to bite). Subsequently, when the object becomes familiar due to habituation, the left eye was used. These findings have several implications. First, in zebrafish, the left and right hemifield carry out different tasks: the left hemifield may have a role in inhibiting a reaction before a decision can be reached. Second, this behavioral paradigm also serves as a way to measure decision-making and goal-directedness. It will thus be interesting to see if monoamine systems play a role in mediating such behavior.
A simple spatial alternation paradigm with a food reward was recently employed to assess learning and memory in zebrafish (Williams et al. 2002). Fish were fed on alternating sides of a divided fish tank, with a red card displayed on one side to serve as a visual cue for orientation. Observations were made at three time points (at the cue, a light tap near the center of the tank, during food delivery, and 5 seconds after the food delivery), and correct responses were compiled for averages and statistical purposes. It was found that adult zebrafish quickly learned to alternate for food, and could recall the learned task after a short period of 10 days. The alternating behavior was extinguished by withholding the food reward. Furthermore, 6–8-week-old juveniles learned the task as well as or better than adult fish, whereas 3–4-week-old zebrafish did not learn the task to any significant level above random chance. It is possible that 3–4-week-old fish are too young to learn or they are too weak to properly complete the task. Taken together, this analysis suggests that higher mental processes such as learning and memory are present in zebrafish, and may be subsequently studied at molecular and cellular levels, harnessing the power of zebrafish forward genetics. It will be interesting to test pharmacologically what neurotransmitter systems (e.g. the monoamine system) may be involved in modulating this behavior.
A genetic analysis of the development of the monoamine system
The developmental ontogeny of DA, NA and 5HT neurons in zebrafish has been characterized (Bellipanni et al. 2002; Guo et al. 1999b; Ma 2003; Rink & Wullimann 2001). Similar to all teleosts examined to date, the majority of DA neurons are found in the basal forebrain while absent from the midbrain. Anatomical as well as tracing experiments have shown that the teleost basal forebrain DA neurons send ascending projections to telencephalon, particularly to the proposed area of striatum (Parent et al. 1984; Rink & Wullimann 2001). Thus, the telesot basal forebrain DA neurons may carry out functions similar to the mammalian midbrain DA neurons. In addition, DA neurons are found in the olfactory bulb and retina. NA neurons are present in very small numbers (∼3–10 in adult zebrafish) in the locus coeruleus. 5HT neurons are found in the epiphysis, basal diencephalon and hindbrain. Because DA neurons selectively degenerate in Parkinson's disease patients, elucidating the mechanisms of DA neuron development will provide important insights into the regeneration of these neurons. A genetic screen for mutations affecting the development of DA and NA neurons has identified a handful of mutations (Guo et al. 1999b). The foggy mutant exhibited a deficit of DA neurons but a surplus of 5HT neurons in the basal diencephalon, and was found to disrupt a regulator of transcription elongation (Guo et al. 2000). This is the first example to show that regulation of transcription elongation plays an important role in vertebrate neuronal development. However, the foggy mutant also has defects in other neuronal types, e.g. neurons in the retina. This precludes its use in addressing the function of basal forebrain DA and 5HT neurons. Nevertheless, the foggy mutant opens a window to look into the role of transcription elongation in regulating neuronal fate determination.
The soulless mutant lacks the locus coeruleus NA neurons. These neurons send their projections widely throughout the entire brain and regulate a variety of functions, such as mood, alertness, sleep-wake and memory acquisition (Barnes & Pompeiano 1991; Cirelli & Tononi 2000). Like DA neurons, locus coerulus NA neurons are also vulnerable in neurodegenerative disorders such as Parkinson's disease and Alzheimer's disease. The soulless gene encodes the paired homeodomain transcription regulator phox2a that is required to specify the locus coeruleus noradrenergic neurons together with subsets of cranial sensory and motor neurons (Guo et al. 1999a). The soulless mutant is larval lethal, due to apparent lack of food intake. The exact neuronal cause of this phenotype is not yet clear.
The too few mutant, in which the basal forebrain DA and 5HT neurons are selectively reduced, disrupted a zinc-finger containing protein that probably acts as a transcription regulator (Levkowitz et al. 2003). Among the mutants isolated, the too few mutant showed the most specific defects in that only DA and 5HT neurons located in the basal forebrain were affected. The too few mutant is adult viable, and has no observable phenotype in terms of sensory modality or movement. Therefore, the too few mutant provides a great opportunity to explore the function of basal forebrain DA and/or 5HT neurons.
A genetic analysis of the function of the monoamine system
The monoamine systems carry out a wide variety of functions, including motor coordination, motivation and reward, anxiety/fear, social interaction and learning and memory, and their dysfunctions are implicated in a range of complex human disorders. A behavioral approach is desirable to reveal the function of neural systems in living organisms. However, behaviors that involve the monoamine system are generally quite complex. To apply a forward behavioral genetic screening strategy to identify genes and pathways in the formation and function of monoamine systems and possibly other related neural systems, the challenge rests on the establishment of appropriate behavioral assays. In order for the assays to work for a large-scale genetic screen, they must have the features of simplicity and robustness. Since it is not possible for a single behavioral assay in any animal model to truly resemble complex human disorders, the approach shall be to understand the behavior at molecular and cellular levels in the context of a particular model organism. Genes and circuitry identified in animal study will provide much-needed candidates for subsequent investigation in humans.
A pilot screen has been carried out to identify mutations that showed altered CPP response to cocaine (Darland & Dowling 2001). Three mutations that displayed abnormally low response to cocaine were identified. Greater than 45% of individuals in the F2 family from all three mutations showed insensitivity to cocaine, suggesting that all three mutations are dominant. Although the frequency of recovering mutations appears to be overly high (3 out of 18 families screened), the behavior was inherited by the F3 generation in a manner that suggests they are dominant single gene mutations. Given this frequency, one would expect that there are many genes that are involved in cocaine CPP and can be mutated to give an identifiable phenotype.
Other behavioral assays described above (e.g. the ethanol-induced locomotor activity, or the light/dark preference) are relatively simple for high throughput genetic screens. The assays that measure behavioral lateralization and learning and memory are very interesting. However, more work is needed to amend them to a higher throughput. It will be interesting to test the too few mutant in these assays and subsequently use these assays for forward genetic screens. Sensory or motor defective mutants will affect these behaviors, but they can be easily distinguished by behavioral assays that assess sensory or motor integrity. Mutants identified could be involved in the development, neural connectivity and function of monoamine systems as well as other neural systems involved in these complex behaviors such as motivation and reward, anxiety/fear, decision-making, functional lateralization and learning and memory.
The past three decades have seen zebrafish grow from a pet store resident to a laboratory model organism. With the build-up of genetic and genomic infrastructure, the development of behavioral assays and the strength for large-scale forward genetic screens, the zebrafish is expected to make valuable contributions to the understanding of genes, brain, behavior and human neurological disorders.
The author wishes to thank Brent Lockwood, Keerthi Krishnan and an anonymous reviewer for helpful comments on the manuscript. Work in the author's laboratory is supported by grants from Searle Scholar's program, Burroughs Wellcome Fund, David and Lucile Packard Foundation, Sandler family Foundation and NIH.