Abstract
Behavioral syndromes are suites of two or more behaviors that correlate across environmental contexts. The aggression–boldness syndrome links aggression, boldness, and exploratory activity in a novel environment. Although aggression–boldness has been described in many animals, the mechanism linking its behavioral components is not known. Here we show that mutation of the gene encoding fibroblast growth factor receptor 1a (fgfr1a) simultaneously increases aggression, boldness, and exploration in adult zebrafish. We demonstrate that altered Fgf signaling also results in reduced brain histamine levels in mutants. Pharmacological increase of histamine signaling is sufficient to rescue the behavioral phenotype of fgfr1a mutants. Together, we show that a single genetic locus can underlie the aggression–boldness behavioral syndrome. We also identify one of the neurotransmitter pathways that may mediate clustering of these behaviors.
Introduction
Many animal species exhibit behavioral syndromes (Sih et al., 2004a,b) [also called personalities (Boissy, 1995; Wolf et al., 2007), coping styles (Benus et al., 1991; Koolhaas et al., 1999), or temperaments (Réale et al., 2007)], suites of two or more behaviors that correlate across environmental contexts. Behavioral syndromes constrain the direction of evolution, affect an animal's future fitness, maintain individual differences in behavior, and may explain maladaptive behaviors (Sih et al., 2004a; Wolf et al., 2007). From an adaptive evolution point of view, the existence of behavioral syndromes is puzzling, because greater flexibility in behavior would give animals a selective advantage (Sih et al., 2004a; Bell and Sih, 2007; Wolf et al., 2007). According to the aggression–boldness syndrome, animals that are more aggressive will also be bolder and more explorative in novel environments (Huntingford, 1976). Aggression–boldness has been described in many animals, including mammals (Benus et al., 1991), birds (Dingemanse et al., 2007), and fish (Huntingford, 1976; Brick and Jakobsson, 2001; Schjolden et al., 2005; Moretz et al., 2007; Bourne and Sammons, 2008). However, the mechanism underlying its formation is not well understood and has been suggested to be caused by either pleiotropic genetic (Deng et al., 1994; Dingemanse et al., 2007) or hormonal (Koolhaas et al., 1999; Veenema et al., 2003) constraint or selection driven by optimal adaptation to the environment (Bell, 2005; Bell and Sih, 2007), with each behavioral component having a separate genetic basis.
Similar to other vertebrates, there is a correlation between aggression and boldness in zebrafish, although the relationship varies depending on the strains analyzed (Moretz et al., 2007). Zebrafish display characteristic agonistic postures, including erection of the fins, biting, tail thrashing, and short bouts of fast swimming (Gerlai et al., 2000). Aggression is exhibited by both sexes (Moretz et al., 2007) and can be influenced by both habitat complexity and rearing conditions (Basquill and Grant, 1997; Marks et al., 2005). Furthermore, different wild-type strains show varying aggression levels suggesting a genetic component to its control (Moretz et al., 2007). Boldness, measured as the number of approaches to a novel object, has also been shown to differ between zebrafish strains, suggesting a genetic basis to this behavior as well (Wright et al., 2003).
In this study, we examine aggression, boldness, and exploration in a zebrafish mutant with reduced fibroblast growth factor receptor 1a (Fgfr1a) function, spiegeldanio (spd) (Rohner et al., 2009). Our results provide evidence that modulation of a single genetic locus, fgfr1a, links the three components of the aggression–boldness behavioral syndrome, a phenotype that may be at least partly mediated through a reduction of histamine (HA) levels in the brain.
Materials and Methods
Animal strains, care, and maintenance.
Adult zebrafish were maintained using standard fish-keeping protocols and in accordance with institute guidelines for animal welfare. Two days before each behavioral assay, the fish were moved to an isolated room and kept under similar conditions to minimize environmental variance and interference from background noise during testing.
Behavioral methods.
Behavioral analyses were performed on 3- to 6-month-old adult Tübingen wild-type, spiegeldaniot3R705H and spiegeldaniohu3264 fish of both sexes. Behavioral experiments were performed using Videotrack and LabWatcher from ViewPoint Life Sciences and a digital camera from Point Gray Research. All behavior experiments were completed between 1:00 P.M. and 4:00 P.M. to minimize circadian differences. One week before analysis, fish were habituated to the stress of being netted. Fish were daily subjected to two rounds of being caught in a fish net, held gently for 3 s, and released into a second tank. To reduce observer bias in the manual analysis of aggressive behavior, films were quantified by two independent observers who were unaware of the genotype being analyzed. Results presented in here show the mean of both observers' findings, and the correlation between the two results was estimated using the Pearson's product-moment correlation coefficient (r = 0.86). In all experiments, the age, sex, and size was carefully matched between fish of different genotypes.
Swimming behavior of fish in behavioral setups.
Photographs of behavioral setups and swimming traces (see Fig. 1) were captured using the reader function in Videotrack.
Aggression.
The mirror-induced aggression protocol was modified from Gerlai et al. (2000). Adult fish were maintained in groups of 20 animals from birth until the day of testing. On the day before analysis, the fish were moved to a novel 10 L glass fish tank. During testing, fish were placed in a small glass aquarium measuring 15 × 10 × 30 cm. Both long sides and one short side of the aquarium were covered in a white opaque material. The fourth side of the tank was left clear, and a mirror was placed outside of the tank offset at an angle of 22.5° (Gerlai et al., 2000). The behavior of individual fish was recorded using a high-speed digital camera set to 100 frames/s and connected to the Videotrack program. Films were manually analyzed in LabWatcher (ViewPoint Life Sciences. The results were analyzed and processed in Excel (Microsoft).
Boldness.
Boldness was measured using two protocols: the time spent close to a novel object and the time spent on the nonpreferred side of a condition place-preference tank. Novel object boldness gave more reproducible results and so was used when testing the effect of histaminergic pathway drugs on both wild-type and spd fish. “Novel object boldness” was measured as the amount of time spent close to a novel object (Wright et al., 2003). The novel object was made by filling a 15-cm-long clear plastic tube with dark modeling clay. The novel object was suspended at one end of the tank, midway in the water column. Boldness was measured as the total time spent within one body length of the object in a 10 min period. “Place-preference boldness” was measured as the amount of time spent on the nonpreferred side of a place-preference tank. Fish were placed into a 10 L tank, the preferred side of which was colored brown and the nonpreferred side of which was white with two black spots (Ninkovic and Bally-Cuif, 2006). Fish were allowed to habituate to this setup for 2 d. Boldness was calculated as the percentage of time spent on the nonpreferred white side of the tank. Both behavioral protocols gave similar results and so were assumed to measure the same behavior in zebrafish: boldness.
Exploration of a novel environment.
Exploration was quantified as the amount of time needed to explore the arms of a novel environment. For this experiment, a T-shaped tank with one long arm and two short arms (arm lengths, 30 × 30 × 45 cm) was used. The fish were not habituated to this setup. Fish were released into the long arm of tank, and the experiment started when they crossed a designated start line. Exploratory behavior was measured as the time taken to explore and swim within one body length of the end of each arm.
Locomotion.
Locomotion was measured in a 10 L plastic fish tank with clear walls. Locomotion was defined as the distance swum in a 5 min period after a 10 min adjustment to the new tank.
Fish were tracked using a high-speed infrared camera with the Videotrack software set to a detection threshold of 11, inactive/small threshold of 5, and small/large threshold of 10. The mean total distance swam (small distance and large distance) was then calculated for each animal. Locomotion was used as a readout for general behavior of the fish. Because the aggression, boldness, and exploration tests all involved swimming, a reduction in locomotion (for example, after drug treatment) would skew the behavioral results.
Olfaction.
The olfactory test was modified from Vitebsky et al. (2005). Adult fish were tested for their reaction to the aversive amino acid l-serine. A glass fish tank (15 × 10 × 30 cm) was covered with white opaque material. Two flexible plastic tubes were attached to opposing corners of the tank. The tubes were connected to a 15 ml Falcon tube filled with either saline (control) or 104 m l-serine. The tank was divided into two areas, and the time spent in each area was measured using Videotrack. Olfaction was quantified as the time spent away from the source of aversive l-serine.
Drug treatments.
Fluoxetine hydrochloride, tacrine hydrochloride, imetit dihydrobromide, and galantamine hydrochloride were purchased from Sigma-Aldrich. Drug doses were chosen in line with those used in other species (Morisset et al., 1996). In some experiments, fish were treated with drugs before their behavior was measured. Drugs were either administered intraperitoneally (1 mg/g tacrine hydrochloride, imetit dihydrobromide, or galantamine hydrochloride, once per day for 3 d; Sigma-Aldrich) or directly into the tank water (3 μg/ml fluoxetine hydrochloride for 3 h; Sigma-Aldrich) (Lynn et al., 2007). For control experiments, animals were treated with saline instead of drug.
Histochemistry reagents.
In situ hybridization was performed according to Norton et al. (2008) and antibody staining according to Topp et al. (2008). The following in situ hybridization probes were used in this study: dusp6; fgfr1a; mao; npvf; pomcb; slc6a4a; and slc6a4b. For gene information, refer to www.zfin.org. Histamine n-methyl transferase (hnmt) was amplified using the primers GGCTCAATGACTAAGATTCAAG and GCTCCATTCAAAACACTTGT. hnmt was cloned into the StrataClone PCR Cloning Vector pSC-A (Stratagene). The following antibodies were used: anti-phosphorylated extracellular signal-regulated kinase (pERK) [mitogen-activated protein kinase (MAPK); 1:1000; Sigma], anti- proliferating cell nuclear antigen (PCNA) (1:1000; Dako), anti-gonadotrophin releasing hormone (GnRH) (1:500; Sigma), and anti-histamine (1:2000; gift from Dr. P. Panula, University of Helsinki, Helsinki, Finland). TUNEL staining was performed using the Neurotacs 2 in situ apoptosis detection kit (Trevigen). Specimens were photographed with an Axioplan2 stereomicroscope for in situ staining (Carl Zeiss) or Axioplan2 ApoTome for antibody labeling (Carl Zeiss). Figures were mounted in Adobe Photoshop version CS2 (Adobe Systems). Nissl staining was performed on cryosections of the brain according to standard protocols.
Radioimmunoassay.
Whole adult brains were snap frozen in liquid nitrogen and stored at −80°C until processing. Brains were homogenized by grinding up in 150 μl of 0.2N HCLO4. After centrifugation at 10,000 × g for 5 min at 4°C, the supernatant was neutralized to pH 6–8 by addition of an equal volume of 1 m potassium borate and then passed through a 10 kDa filter (Nanosep 10 kDa; Pall Life Sciences, VWR). Two aliquots of 100 μl were analyzed for histamine levels by radioimmunoassay (RIA Histamine kit, Immunotech), following the instructions of the manufacturer.
ELISA.
Blood plasma was collected from pools of three male and three female adult fish (each yielding between 1 and 5 μl of blood, to make a total of 25 μl). Fresh blood was mixed 1:10 with a citrate buffer and stored in heparinized hematocrit tubes. Blood samples were centrifuged at 4°C for 10 min, and the plasma was aspirated. ELISA was performed using either a testosterone immunoassay kit (DNOV002) or a 17β-estradiol immunoassay kit (DNOV001) from AbCys.
Real-time PCR.
Quantitative PCR analysis (qPCR) was performed by Eurofins. Primers were designed and optimized by Eurofins (qPCR primer sequences available on request). PCRs were performed on an ABI 7900 SDS machine using a SYBR Green Universal Master Mix (Exiqon). cDNA was synthesized from 1 μg of RNA extracted from whole brains. The PCR conditions were 56°C for 10 min, 94°C for 10 min, followed by 40 cycles of 60°C for 1 min and 94°C for 15 s. Analysis was performed in duplicate using 10 biological replicates for each genotype. All results were separately normalized to the expression level of four housekeeping genes: gapdh, mapta, neurod, and tubb2b. Because all housekeeping genes showed similar results, only data from neurod are presented in this study. qPCR results are shown as an expression ratio calculated relative to neurod. Relative expression levels were calculated using the arithmetic comparative method, (2−ΔΔCt), which includes a correction between E (PCR amplification efficiency) for the target and housekeeping gene (Livak and Schmittgen, 2001). Results were expressed as relative expression ratios (REs) using neurod as a reference gene in the formula RE = (E ref)Ct ref/ (E target)Ct target, where Ref is the housekeeping gene, target is the gene of interest, and Ct is the crossing point (Filby and Tyler, 2005). For statistical analysis, the Student's t test was performed with n = 10 for both genotypes.
qPCR primer sequences.
Primer sequences used for quantitative real-time PCR analysis of neurotransmitter encoding gene expression in spd and wild-type brains included the following: hdc forward, TCCAGCAGTAATCATCCACTCGT; hdc reverse, TTAGAGGAGCGAGGCGTCTC; hnmt forward, GAAAGAGTTCAGTAAAAATGCTCCTGA; hnmt reverse, ATATGATCCTGCCGTCCACC; hrh1 forward, GAAACCTGCGTCCATCACG; hrh1 reverse, AAGTTTTCGTTGCAGAGCGG; hrh2 forward, CACGCCCATTCTTCAAAAGG; hrh2 reverse, CCATTTGTAGCGCTGTTATCGTC; hrh3 forward, TTCTGGCTGCTGTGGATCAA; and hrh3 reverse, TTGCTCGGACACAGTAATTTGC.
Statistics.
All error bars denote SEM. For aggression, boldness, exploration, locomotion, cell counts, and qPCR analysis, we used Welch's two-sample t test allowing for unequal variances. For olfactory behavior, fluoxetine hydrochloride, tacrine hydrochloride, imetit dihydrobromide, and galantamine hydrochloride treatment data were analyzed by two-way ANOVA with genotype and drug. First, a model with interaction was fitted. If there was a significant interaction effect, the genotype difference was tested in the subgroups of drug treatment. If the interaction term was not significant, a two-way ANOVA without interaction was fitted. If appropriate, post hoc comparisons were performed by t tests with p value adjustment for multiple testing according to Holm (1979). When necessary, data were log-transformed to normalize the distribution. Statistical significance was depicted as follows: NS (nonsignificant), p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001. In all cases the number of animals tested is denoted by n.
Results
Reduction of Fgfr1a increases aggression in zebrafish
During routine stock keeping, we noticed that spiegeldanio [fgfr1at3R705H/t3R705H, also called spd (Rohner et al., 2009)] mutants showed an increased tendency to kill other fish and to bite conspecifics during pairing. To quantify these observations, we measured agonistic behavior by using mirror-image stimulation (Rowland, 1999; Gerlai et al., 2000; Clotfelter et al., 2007). Fish are unable to recognize their own mirror image and thus attack as if an intruder is present (Rowland, 1999). Homozygous spd mutants were clearly more aggressive than siblings during a 5 min period (Figs.1a–c,j, 2a). Whereas wild-type fish serially reduced aggression levels after initiation, spd mutants maintained high levels of agonistic behavior throughout the experiment, failing to downregulate aggression over time (Fig. 1k). Agonistic behavior was not sex specific, in agreement with other studies of zebrafish aggression (Moretz et al., 2007) (Fig. 2b). We therefore used fish of both sexes in our subsequent behavioral analyses.
To confirm that alteration of fgfr1a activity underlies the observed behavioral defects, we measured aggression levels in fish mutant for a second fgfr1a allele, fgfr1ahu3264. Both fgfr1a alleles harbor strong hypomorphic mutations in the kinase domain of fgfr1a (Rohner et al., 2009) (Fig. 3a). However, hu3264 shows variable penetrance and juvenile lethality as homozygotes, suggesting that it causes a more severe reduction of fgfr1a activity than the t3R705H allele (Rohner et al., 2009). The lethality of juvenile fgfr1ahu3264/hu3264 fish precluded their use in adult behavioral experiments. We thus quantified aggression in transheterozygous fish (fgfr1at3R705H/hu3264). Like homozygous spdt3R705H/t3R705H, transheterozygous fgfr1at3R705H/hu3264 fish showed a significant increase in aggression compared with siblings (Fig. 2c). Fish heterozygous for either allele showed a nonsignificant dominant trend increase in aggression compared with wild type, suggesting that a gene dosage effect may exist for this phenotype. We decided to use fgfr1at3R705H/t3R705H in our subsequent experiments (hereafter referred to as homozygous spd mutants).
spiegeldanio mutants exhibit hallmarks of the aggression–boldness syndrome
Because an aggression–boldness syndrome has been observed in some zebrafish strains (Moretz et al., 2007), we surmised that the increased aggression levels in homozygous spd might be paralleled by increases in boldness and exploratory activity. Boldness was quantified by using two behavioral assays. We first measured the amount of time spent close to a novel object [novel-object boldness(Wright et al., 2003)]. Neither wild-type nor homozygous spd fish showed a preference for either side of an empty boldness tank (Fig. 2f) or a difference in the distance swum in the presence of the novel object (Fig. 2g). However, homozygous spd mutants spent significantly more time within one body length of the novel object than wild-type fish (Fig. 1d–f,l), suggesting an increase in boldness. The second boldness paradigm recorded the amount of time spent on the nonpreferred side of a biased place-preference tank [place-preference boldness(Ninkovic and Bally-Cuif, 2006)]. Wild-type fish spent on average 14.7% of the time on the nonpreferred (frightening) side compared with 35.9% of time spent by homozygous spd mutants (Fig. 1m). The similar results of both assays (place preference and reaction to a novel object) support the idea that they measure the same behavior in zebrafish. Together, the results of both these tests suggest that spd exhibits enhanced boldness compared with siblings. Similar to aggression levels, fish heterozygous for either allele show a nonsignificant trend increase in boldness, and both spd and transheterozygous fish show significantly increased boldness behavior (Fig. 2d).
We next analyzed exploratory activity in a novel environment, a T-shaped aquarium. We recorded the amount of time needed to swim within one body length of each end of the tank. Homozygous spd mutant fish explored this environment in significantly less time than wild-type siblings (Fig. 1g–i,n), thereby showing increased exploration. However, homozygous spd mutants did not exhibit changes in locomotion compared with wild type (Fig. 2e). Together, these results suggest that reduction of Fgf signaling in spd results in concomitant increases in aggression–boldness.
In humans, autosomal dominant loss-of-function mutations in FGFR1/KAL2 cause Kallmann syndrome (KS), which comprises both anosmia/hyposmia and hypogonadotropic hypogonadism caused by a reduction of GnRH levels and hypoplasia of the olfactory bulb and tract (Dodé et al., 2003). KS is a developmental disease with highly variable penetrance; patients with the same mutation within FGFR1 show a range of symptoms, from normosmia and fertility to complete KS. Because spd codes for one of the two zebrafish orthologs of FGFR1/KAL2 (Fig. 3a), we asked whether homozygous spd mutants share phenotypic similarities with KS. We found that homozygous spd are able to detect the aversive amino acid l-serine (Vitebsky et al., 2005) (Fig. 3b) and so have at least some ability to detect odors. Homozygous spd also have no changes in the expression of GnRH in the preoptic area (Fig. 3c) and similar blood levels of testosterone and 17β-estradiol as wild-type fish (Fig. 3e). Therefore, homozygous spd mutants do not share obvious phenotypes with KS patients.
spd mutants have a major reduction of Fgf signaling in the inferior lobe of the hypothalamus
Compared with mice and medaka deficient in fgfr1 activity (Deng et al., 1994; Yokoi et al., 2007), homozygous spd mutants have a surprising lack of early developmental phenotype, perhaps because of the presence of two paralogous fgfr1 genes with similar expression patterns during development (Rohner et al., 2009). However, only fgfr1a is expressed in the adult zebrafish brain, and the expression of fgfr1a is globally similar in both wild-type and mutant fish. To examine the Fgf signaling defect in spd, we analyzed expression of the dual specificity phosphatase enzyme dusp6 (Fig. 4a,b) and pERK (Fig. 4c,d), readouts of Fgf signaling in the adult brain (Topp et al., 2008). We observed a reduction of both dusp6 and pERK in the periventricular nucleus of the inferior lobe of the hypothalamus (PVN; Fig. 4a–d,k). However, other brain areas, such as the central nucleus of the hypothalamus (Hc), were essentially unaffected (Fig. 4a–d,k). During closer inspection, the ventricle surrounded by the PVN appears larger in homozygous spd (Fig. 4j) than wild type (Fig. 4i), supporting our observation that Fgf function is altered in this region. However, analysis of anti-PCNA antibody (marking dividing cells; Fig. 4e,f) and TUNEL staining for apoptosis (Fig. 4g,h) did not reveal addition defects between wild-type and mutant hypothalami. We examined more genes expressed in the Hc and PVN (Fig. 4l–o). All of the genes that we analyzed were similarly expressed in the inferior hypothalamus of both genotypes, indicating that the PVN is not globally reorganized. Thus, homozygous spd mutants have a major reduction of Fgf signaling in the PVN.
We next expanded our characterization of the mutant brain to include other neural markers. The preoptic area of the anterior hypothalamus (POA) is another area of the zebrafish brain implicated in the control of aggression and dominance (Larson et al., 2006). In contrast to the PVN, the POA looked morphologically normal, and we found similar expression of POA marker genes in both wild-type and homozygous spd (Fig. 4p–s).
Behavioral alterations in spd are caused by reduced histaminergic neurotransmitter signaling
We next aimed to identify the neurotransmitter pathways linking Fgf signaling and behavior. 5-HT acting via Htr1a (Gross et al., 2002; Clotfelter et al., 2007) and Htr1b receptors (Ramboz et al., 1996) has been identified as having a key role in the control of aggression and boldness/anxiety in fish and mouse. Treatment of fish with the selective 5-HT reuptake inhibitor fluoxetine hydrochloride increases 5-HT levels and has been shown to reduce aggression levels in fish (Perreault et al., 2003; Lynn et al., 2007). We first observed an increase in the number of neurons expressing the 5-HT transporter gene slc6a4a in the raphe nucleus of spd (Fig. 5a,b,d).
To assess the contribution of 5-HT signaling to the behavior of spd, we gave fish an acute nonsedative (see Fig. 8a) dose of fluoxetine and measured subsequent changes to aggression, boldness, and exploration. Surprisingly, although fluoxetine reduced aggression levels in both genotypes, it failed to blunt the increased aggression in spd (Fig. 5e). We next measured place-preference boldness and exploration. Modifying 5-HT signaling with fluoxetine caused a significant increase in the place-preference boldness levels of both wild-type and homozygous spd without rescuing the mutant behavior (Fig. 5f); thus, similar to the aggressive behavior, 5-HT neurotransmission does not underlie the increased boldness in spd. Conversely, fluoxetine treatment did not significantly alter exploratory activity in either homozygous spd or wild type (Fig. 5g), ruling out a role for 5-HT signaling in this behavior. We thus conclude that 5-HT signaling does not underlie the Fgfr1a-dependent aggression–boldness syndrome. However, it is possible that the increase in slc6a4a expression seen in the dorsal raphe nucleus of mutants may be connected to the behavioral changes that we observe in a 5-HT-independent manner.
To broaden our analysis to include other neurotransmitter pathways, we used qPCR to examine the expression of 40 components of the dopaminergic, 5-HTergic, noradrenergic, and histaminergic neurotransmitter pathways, including synthesis enzymes, receptors, and transporters. The majority of these genes showed no change in expression, whereas the gene encoding hnmt was significantly upregulated in homozygous spd mutants compared with wild type (Fig. 6a), representing a 1.06-fold change in mRNA levels. hnmt curbs histaminergic signaling in the brain by converting HA to its metabolite N-methylhistamine (Haas et al., 2008). hnmt is widely expressed throughout the brain, with similar distribution in both wild type and spd (Fig. 6i–n). This is in keeping with the widespread projection pattern of histaminergic neurons, whose cell-bodies are clustered in the Hc (Eriksson et al., 1998). Although the fold change of hnmt RNA was very small, it prompted us to investigate whether spd mutants have an imbalance in histaminergic neurotransmitter signaling. Analysis of neurotransmitter levels by radioimmunoassay (Fig. 6b) and with an HA-specific antibody (Fig. 6c–h) confirmed the reduction of HA in the mutant brain. The decrease in antibody labeling was seen in all brain areas, although a few HA-positive fibers were still seen in the raphe nucleus of spd (Fig. 6d,g). We reasoned that, if decreased HA was a relevant downstream component of impaired Fgf signaling, increasing HA levels should rescue some aspects of the spd phenotype. We treated homozygous spd mutants with a nonsedative dose (see Fig. 8b) of the Hnmt inhibitor tacrine hydrochloride (Horton et al., 2005) and observed an increase in HA levels in the brain (Fig. 6b). We then measured aggression, boldness, and exploration of tacrine-treated fish. Treatment with tacrine did not affect aggression or novel-object boldness levels in wild type, but reduced aggression and novel-object boldness in homozygous spd (Fig. 7a,b). Tacrine treatment had a weaker effect on place-preference boldness, only causing a nonsignificant trend reduction of boldness in spd (Fig. 8e). Thus, the novel-object boldness test appeared to be more sensitive to changes in HA level, and so we used it in our subsequent pharmacological experiments. Tacrine treatment increased the amount of time needed to explore the T-shaped tank in both wild-type and homozygous spd mutants, bringing both genotypes to the same exploration level (Fig. 7c). At low concentrations, tacrine hydrochloride is specific for hnmt, whereas at high concentrations it can also inhibit the acetylcholine breakdown enzyme acetylcholinesterase (Horton et al., 2005). To rule out a contribution of acetylcholine signaling to the spd behavioral phenotype, we treated homozygous spd mutants with a nonsedative dose (Fig. 8c) of the acetylcholinesterase inhibitor galantamine hydrochloride (Houghton and Howes, 2005) and measured aggression. Treatment with galantamine did not affect the aggression levels of either wild-type or mutant fish (Fig. 8f), ruling out a contribution of acetylcholine signaling to the aggressive behavior we observed.
In zebrafish, histaminergic neurotransmitter signaling is transduced by three receptors encoded by the genes hrh1, hrh2, and hrh3 (Peitsaro et al., 2007). All three genes are expressed at similar levels throughout the brain of wild-type and homozygous spd mutant fish. In the PVN, only hrh3 expression was easily detectable, whereas hrh1 was absent and hrh2 was only weakly expressed. To test whether the HA-linked behavior changes could be mediated by the histamine H3 receptor (Hrh3), we treated fish of both genotypes with a nonsedative dose (Fig. 8d) of the selective Hrh3 agonist imetit dihydrobromide (Garbarg et al., 1992). We measured aggression, novel-object boldness, and exploratory activity of imetit-treated fish. Treatment with imetit rescued the behavioral phenotype of homozygous spd mutants. Both aggression and novel-object boldness were reduced compared with wild-type siblings (Fig. 7d,e), whereas the amount of time needed to explore a T-shaped aquarium was increased to the level of wild-type fish (Fig. 7f). Therefore, the combined results from these experiments suggest that modulation of HAergic signaling can rescue the Fgfr1a-dependent aggression–boldness phenotype observed in homozygous spd mutants.
Discussion
In this study, we have analyzed spiegeldanio mutant fish that have a reduction of fgfr1a activity in the brain. Homozygous spd mutants show increases in aggression, boldness, and exploration, three components of the aggression–boldness behavioral syndrome. These behaviors are related to a reduction in histamine neurotransmitter levels in spd, and pharmacological manipulation of HA signaling is sufficient to rescue the observed behavioral defects.
Although behavioral syndromes have been extensively studied, the mechanism underlying their formation is not well understood. Behavioral syndromes could be caused by either genetic or hormonal constraint (Koolhaas et al., 1999; Drent et al., 2003; Veenema et al., 2003; Dingemanse et al., 2007) or by the selection of behaviors in response to environmental pressure (Bell, 2005; Bell and Sih, 2007). In this study, we have shown that polymorphism of a single gene, fgfr1a, is sufficient to induce parallel changes in three behaviors providing evidence that genetic constraint can influence the aggression–boldness behavioral syndrome. However, in another fish, the stickleback, the correlation between aggression and boldness varies depending on the testing environment (Bell, 2005; Bell and Sih, 2007). It thus seems likely that a combination of genetic or hormonal constraint and environmental conditions controls the expression of behavioral syndromes. Aggression is a malleable behavior, whereas boldness tends to be a more stable trait (Bell and Sih, 2007). Although our behavioral measurements were highly reproducible, we saw differences in aggression levels between zebrafish measured in different locations. Fish raised in Munich, Germany, were significantly more aggressive than those raised in Gif-sur-Yvette, France. Nevertheless, spd were always more aggressive than wild-type regardless of testing location (Fig. 9a). Zebrafish aggression levels can be affected by the environmental conditions during development (Marks et al., 2005), and behavioral assays are sensitive to changes in feeding regimen and diet, lighting conditions, pH, time of day, and experimental observer (Wahlsten et al., 2006; Burgess and Granato, 2008). Although we made an effort to maintain similar conditions between laboratories, we cannot rule out the contribution of one (or all) of these variables to the alterations in aggression levels. Conversely, the boldness levels of both wild-type and spd were similar in both locations (Fig. 9b). Therefore, if the correlation between aggression and boldness varies in response to environmental factors, it may be the aggression component of the syndrome that is more likely to be modified.
Although spd mutant fish have a life-long reduction of Fgf signaling, it does not appear to adversely affect their lifespan, at least in the laboratory. Reduction of fgfr1a function therefore appears to occur within an acceptable limit of developmental plasticity (West-Eberhard, 2005): adult behavioral changes occur in the absence of deleterious developmental phenotypes. In this study, we have shown that mutation of fgfr1a increases the number of zebrafish expressing aggression at the higher end of the natural spectrum (Fig. 2a). Although increased aggression is sometimes viewed as a negative trait, this result suggests that the spd behavioral phenotype may rather represent an extreme example of normal behavior (Garrod, 1902).
spiegeldanio mutants harbor a mutation in fgfr1a that is a receptor for Fgfs, a large family of secreted signaling molecules with important roles during development (Thisse and Thisse, 2005). Zebrafish have multiple Fgf receptors that are coexpressed in the adult brain. For example, fgfr1a, fgfr2, and fgfr3 are all expressed in the PVN, central nucleus, and diffuse nucleus of the hypothalamus (Topp et al., 2008). Both fgfr2 and fgfr3 are expressed normally throughout the brain of homozygous spd mutants. The most obvious reduction of Fgf signaling in spd is found in the PVN. The PVN has already been associated with the control of aggression in teleosts. Electrical stimulation of the nucleus of the lateral recess of the third ventricle (equivalent to the zebrafish PVN) causes aggressive interaction with a mirror in the Bluegill (Demski and Knigge, 1971). However, additional brain areas may also play an important role in linking aggression, boldness, and exploration. For example, a recent gene expression analysis of aggressive zebrafish identified gene expression changes in the telencephalon, optic tectum, hypothalamus, and hindbrain (Filby et al., 2010). It is clear that an in-depth analysis of the spd mutant brain is needed to resolve this issue.
Our analysis of spd mutants demonstrates that a reduction of Fgfr1a signaling results in increased expression of the HA breakdown enzyme-encoding gene hnmt, correlating with decreased levels of HA in the brain. Although hnmt expression was only marginally increased in spd, modulation of HA levels was sufficient to reverse the behavioral phenotype. However, such a slight change in the mRNA expression level of hnmt is unlikely to fully explain the changes observed. Thus, an additional (unidentified) mechanism may also disrupt HA levels in spd. Alternatively, the upregulation of hnmt may only occur in some discrete areas of the mutant brain. Our analysis of hnmt levels in whole zebrafish brain extracts precludes us from ruling out this possibility.
In zebrafish, similar to other animals, HA neurons are found exclusively in the caudal hypothalamus and send projections to many brain areas (Eriksson et al., 1998). We observed a strong reduction of HA-positive fibers throughout the brain of spd (Fig. 6c–h). Although the PVN (the area of the brain with the most striking reduction of Fgf signaling) does not produce HA, it is positive for both hnmt expression and the histamine receptor-encoding genes hrh2 and hrh3 at weak and strong levels, respectively. Rescue of the behavioral phenotype of spd mutants with the Hrh3 agonist imetit (Garbarg et al., 1992) provides evidence that Hrh3 may underlie these behaviors, although we cannot explicitly rule out a contribution of Hrh1 and Hrh2 activity as well. Hrh3 acts as both an autoreceptor and a heteroreceptor (Peitsaro et al., 2000) and has a complex role in the control of HA neurotransmission (Leurs et al., 2005; Jones and Kearns, 2011). Hrh3 knock-out mice have a net decrease of HA levels in the cortex (Toyota et al., 2002) and show reduced anxiety in an elevated plus maze (Rizk et al., 2004) reminiscent of the spd phenotype. Histamine has been linked previously to several behaviors, including aggression, exploration, and boldness (Lecklin et al., 1998; Yanai et al., 1998; Peitsaro et al., 2003; Passani et al., 2004; Masaki and Yoshimatsu, 2006). Thus, HA may have a role as a general regulator of brain function rather than influencing specific behaviors (Wada et al., 1991).
To our knowledge, there are currently no studies linking Fgf and HA, making it difficult to understand how these signaling pathways might interact. However, our data suggest two possible mechanisms. Fgf signaling could act on hnmt expression independently of HA, reducing the activity of the enzyme. A reduction of fgfr1a in spd mutants would lead to a concomitant increase in Hnmt activity and HA turnover. Alternatively, Fgfr1a could interact with Hrh3 signaling at the level of MAPK signaling. Stimulation of Hrh3 receptors results in MAPK activation (Drutel et al., 2001), providing a possible substrate for interaction with Fgf that also uses MAPK as a target (Thisse and Thisse, 2005). Reduction of fgfr1a may decrease the sensitivity of Hrh3-expressing cells to Hrh3 activation. Because we only observe a slight increase in hnmt expression levels in mutants, we tend to favor this second hypothesis. Together, our results suggest that Fgf signaling may sensitize the brain to changes in HAegic signaling levels, thus altering the expression of the aggression–boldness behavior syndrome. Nevertheless, it is important to remember that spd mutants have a life-long reduction of Fgf signaling, meaning that the behavioral phenotype we observe may stem from changes in brain development. It is therefore likely that other neurotransmitter pathways may also mediate aspects of the spd behavioral phenotype. Additional studies will be required to address this issue.
Footnotes
The authors declare no conflict of interest.
The Hubrecht laboratory, Sanger Institute, and MPI-CGB/CRT Dresden are supported by the ZF-Models Integrated Project (Contract LSHG-CT-2003-503496 of the European Commission, respectively, to the laboratories of E. Cuppen, D. Stemple, and M. Brand). The Sanger Institute Zebrafish Mutation Resource is also sponsored by Wellcome Trust Grant WT 077047/Z/05/Z. Work in the L.B.-C. laboratory at the Helmholtz Zentrum München was funded by the Helmholtz Association, the European Union (EU) ZF-Models integrated project under Contract LSHG-CT-2003-503466 and the Center for Protein Science Munich. For work at the CNRS, funds from the EU Seventh Framework Integrated Project NeuroXsys and ZF-Health, the Agence Nationale de la Recherche, Ecole des Neurosciences de Paris, Fondation pour la Recherche Médicale, and Programme Incitatif à la Mobilité d'Équipe programs, and the Schlumberger Association are gratefully acknowledged. We thank Sébastien Bedu and Michel Chaminade for assistance with fish care and behavior work and members of the Bally-Cuif laboratory for scientific comments. We thank Prisca Chapouton and Marion Coolen for discussion of the data and Monika Krecsmarik and Ina Rothenaigner for help with confocal microscopy. We are grateful for help with Nissl staining from Frank Bourrat. The spd aggression phenotype was identified by Stefanie Topp. We are grateful to Pertti Panula, Kristin Tessmar-Raible, and Michael Tsang for reagents. We thank the collaborative efforts of the Hubrecht laboratory, the Sanger Institute Zebrafish Mutation Resource, and the MPI-CGB/CRT Dresden for providing the zebrafish knock-out allele hu3264.
- Correspondence should be addressed to Drs. William Norton and Laure Bally-Cuif, Zebrafish Neurogenetics, Laboratory of Neurobiology and Development, Institut de Neurobiologie Alfred Fessard, Unité Propre de Recherche 3294, CNRS, Avenue de la Terrasse, Gif-sur-Yvette, 91190, France. norton{at}inaf.cnrs-gif.fr; bally-cuif{at}inaf.cnrs-gif.fr