In many animals, a functioning olfactory sensory system is important for food and mate search, kin recognition, predator avoidance, and complex behaviors such as homing. How these responses are established is poorly understood, but it depends on both intrinsic developmental processes as well as epigenetic influences. The development of the olfactory system is a complex and fascinating process. In vertebrates, this sensory system arises through highly orchestrated interactions between the peripherally derived olfactory sensory neurons and the developing olfactory bulb of the central nervous system (CNS). After differentiating, the sensory neurons establish connections with the developing olfactory bulbs. Making correct connections with the CNS is a complex problem due to the large number of olfactory sensory neurons (OSNs), which number from 106 OSNs in fishes to 10–50 times more in mammals (Hildebrand and Shepherd, 1997). Within the developing olfactory bulbs the sensory axons terminate in characteristic clusters called glomeruli, initially described over a century ago by Santiago Ramon y Cajal (Ramon y Cajal, 1897). These glomeruli are formed in part by the convergence of axons from OSNs that express the same olfactory receptors (Mori et al., 1999). As the sensory system develops, the axons of the OSNs use multiple cues to guide their outgrowth from the sensory epithelium to the bulb. These cues include pioneer neurons (Klose and Bentley, 1989; McConnell et al., 1994; Whitlock and Westerfield, 1998), cell surface molecules (Yoshihara et al., 1997; Alenius and Bohm, 2003), and the very olfactory receptors that on the dendritic side of the OSN transduce the presence of chemical cues in the environment (Singer et al., 1995; Strotmann et al., 2004). The genes encoding the olfactory receptors number around 100 in fishes (Laberge and Hara, 2001) and 1,000 in mammals (Buck and Axel, 1991). Thus, the developing sensory neurons must choose to express a given receptor type, initiate axonogenesis, and find their correct targets in the developing olfactory bulb.
The mature olfactory system allows the animal to respond to a variety of chemical cues in the environment, which convey information about food, conspecifics, and mates (Brennan, 2004). For example in fishes, behavioral and physiological studies have shown that they respond to amino acids that act as feeding cues, (Jones, 1992), as well as to social olfactory cues such as bile acids (Li et al., 1995), pheromones (von Frisch, 1938, 1941), and hormones (Sorensen et al., 1998; for reviews, see Hara, 1971; Sorensen and Caprio, 1997; Whitlock, 2004).
To add to the complexity of its development, the functioning of the olfactory system can also depend on epigenetic interactions. For example, fishes such as salmon form and retain memories of the odor cues associated with their place of birth in a process called olfactory imprinting (Hasler and Scholz, 1983). This olfactory imprinting then contributes to the fishes' successful return to reproduce at their stream of birth after several years at sea. Finally, it is known that olfactory neurons regenerate during life, yet olfactory function remains relatively stable.
The cellular and molecular processes that underlie the development of the olfactory system, the bases for olfactory coding, imprinting, and invariance despite regeneration of olfactory neurons, are poorly understood. Especially in the invertebrates, the nematode C. elegans (Bargmann, 1993) and the fruitfly D. melanogaster (Carlson, 1996), behavioral genetic screens have been used successfully to dissect olfactory function. In the zebrafish, D. rerio, there have been several large-scale genetic screens (Dreiver et al., 1996; Haffter et al., 1996). Yet, no mutants specifically affecting the development or function of the olfactory system have been isolated despite the existence of a rich literature on the behavior of fishes to draw upon for the design of a behavioral genetic screen (Hoar and Randall, 1971).
Here we describe the results of a pilot screen to isolate mutants defective in olfactory behavior. From the progeny of 50 chemically mutagenized fish, we isolated the laure (lre) mutant, which was isolated based on the lack of an aversive response to L-cysteine, and subsequently shown to be defective in response to L-serine and L-alanine but not to taurocholic acid. Preliminary characterization of lre mutant fish revealed a complex etiology: olfactory sensory neurons had normal physiological responses to most odors tested and a near-normal response to cysteine; however, lre mutant fish had a reduced number of sensory neurons and abnormal axonal endings in the olfactory bulb. Our work demonstrates the feasibility of a behavior-based screen for the genetic analysis of the vertebrate olfactory system.
Behavioral Responses of Young Zebrafish to Odors
In order to identify odors that elicited strong behavioral responses suitable for carrying out a behavioral screen, we tested the behavioral response of young (3–4-day-old) juvenile fish to a variety of amino acids. For this, 25 fish were placed in the center of a Petri dish, exposed to test odors, and the number of fish that had left the center of the dish 30 sec after the addition of the odor were counted. Responses were compared to those obtained after exposure to distilled water (diluent) (Fig. 1; and see Experimental Procedures section). No responses above control were observed for L-serine or L-alanine (tested at 3× 10−5M; data not shown). By contrast, L-cysteine (3 × 10−5M) elicited an aversive response from 3-day juveniles (Fig. 1B,D,I). The juveniles moved away from the L-cysteine stimulus with small darting movements toward the outside edge of the Petri dish (Fig. 1B,D). Out of groups of 25 fish, an average of 21.2 ± 3.2 fish (i.e., about 85%; range 11–25; n = 10 groups tested) rapidly moved away from the odor after L-cysteine was added, compared to 5 ± 1.3 (i.e., about 20%; range 2–7; n = 10 groups tested) following addition of distilled water as well as in undisturbed fish (Fig. 1I; p < 0.0001, t-test for Independent Samples). The juvenile fish clearly moved away from the L-cysteine when exposed to the odor at a final concentration of 10−4 M and 3 × 10−5 M (Fig. 1), but did not respond at concentrations of 10−6 M or lower (not shown). This aversive behavioral response to L-cysteine (3 × 10−5 M) could be elicited starting at 3 days, persisted in 2- (Fig. 1E,F) and 3- (Fig. 1G, H) week-old fish, and remained strong in adulthood (see Figs. 3,4).
Isolation of laure Mutant
Using the aversive response to L-cysteine as our behavioral assay, we screened at 3 days the diploid progeny of each of 50 females derived from mutagenized males and wild-type females (Fig. 2; see Experimental Procedures section; Henion et al., 1996). From these 50 clutches, we identified one for which the number of responding juveniles was significantly reduced compared to that of wild-type controls (p < 0.0001; Fig. 1I) and was similar to that obtained in controls exposed to distilled water or in undisturbed fish. The female that produced these mutant fish was mated (Fig. 2B) and the reduced response to L-cysteine phenotype was confirmed in the F2 progeny at 3 days, as well as at 2 and 3 weeks of age indicating that this abnormal response was not due to a slower development or maturation of the olfactory system of the non-respondents. We named the mutant laure (lre), after the character in the novel Perfume (Süskind, 1986). The lre mutation is recessive and fish homozygous for the mutation are viable and fertile. The lre mutation was maintained in heterozygotes, and carriers of the mutation were identified at each generation. The lre mutation segregated with approximately Mendelian frequencies and the olfactory responses of lre homozygotes have remained unchanged over 4 years of outcrossing and testing (K.E.W., unpublished data). This suggests that the behavioral defects of lre are due to a single mutation or to mutations at closely linked loci.
We tested the response of homozygous adult lre mutants to L-cysteine in a Y-maze and recorded their behavior on video (Fig. 3A,B). We then quantified the olfactory response over time by counting the fish in each quadrant of the maze (Fig. 3B) at 15-sec intervals for 9 min. These data are summarized in Figure 3C. It is evident that wild-type fish (Fig. 3A) avoided the L-cysteine baited arm (Fig. 3C, quadrant I; asterisk). In contrast, the lre mutants (Fig. 3B) did not avoid the side with the L-cysteine odorant (Fig. 3C, quadrant I; asterisk). For each quadrant analyzed (I–IV), the lre mutant response was significantly different from the wild-type fish (p < 0.0001).
Behavioral Deficits of the lre Mutant
Once a line of lre carriers was established, we tested lre mutants and their siblings to determine the specificity of the olfactory deficit. Fish homozygous for the lre mutation showed deficits in response to amino acids (Fig. 4) when compared to their control siblings when tested in their fifth week of age. We tested three amino acids, L-cysteine, L-serine, and L-alanine, by adding solution (all were added at 10−5 M) to one side of the testing chamber and recording the response of the fish. A response of 50% (Fig. 4) represents no preference for the odor. The responses of lre fish to L-cysteine, L-serine, and L-alanine were each significantly different from those of the phenotypically wild-type siblings (p < 0.0001; Fig. 4). The lre fish showed no preference for the amino acids L-cysteine and L-serine, and an aversive response to L-alanine, which is opposite to that of the siblings. Strikingly the lre mutants did not differ from their siblings in their response to taurocholic acid (p = 0.085; Fig. 4). In response to the control odorant, distilled water, the lre mutants and their siblings were not significantly different from each other (p = 0.076; Fig. 4). Therefore, the lre mutants appear thus far to be defective in their response to amino acids, but not bile acids.
Morphological Defects of the lre Mutant
Gross morphological examination of the lre mutant embryos showed that the olfactory organ was present. We used the anti-calretinin antibody (Winsky et al., 1989) to visualize the olfactory sensory neurons. This antibody appears to recognize both microvillar and ciliated olfactory receptors in trout (Porteros et al., 1997) and the pattern of expression we observed is similar to that reported in embryos expressing green fluorescent protein (GFP) under the Olfactory Marker Protein (OMP) promoter (Yoshida and Mishina, 2003). The wild type axonal projections in the olfactory bulb are very stereotyped at 3 days of development, and form three main axon bundles, which we termed medial (M), center (C), and lateral (L) (Fig. 5E,G). The first apparent defect observed in lre fish was that the axonal projections from the sensory olfactory epithelium (OE, see Fig. 5G) to the developing olfactory bulb appeared abnormal during the first 3 days of development: although olfactory axons did reach the bulb, they formed fewer branches and less distinct glomeruli, which first appeared at 36 hr (Fig. 5A, wild-type; Fig. 5B, lre), and became more distinct at 48 hr (Fig. 5C, wild-type; Fig. 5D, lre) and 72 hr (Fig. 5E, wild-type; Fig. 5F, lre) as additional olfactory sensory axons entered the developing olfactory bulb. The disruption in axonal projections seen in the lre mutants had variable penetrance, with the more severe phenotype being no or only remnant medial and center projections and no distinct glomerular condensations in the lateral projections into the bulb (Fig. 5G). The severe phenotype was seen in at least 50% of the mutants in any clutch (n>45). The less severe phenotype still lacked distinct glomerular condensations in the lateral projection but had remnant medial and center projections.
To determine whether the disruption in axonal projections was due to a decrease in the number of sensory neurons in the OE, we counted the number of sensory neuron cell bodies in the OE of both wild-type and lre animals. At two days of development, there were significantly fewer calretinin immunopositive cells in the OE of lre animals (left OE: 25 ± 5.1; right OE: 25.8 ± 4.7; n = 14 embryos) compared to the number in wild type animals (left OE: 41.4 ± 6.7; right OE: 41.8 ± 6.1; n = 14 embryos). This difference was maintained in older fish such that at 3 days of age, the OE of the mutants still displayed far fewer cells (left OE: 29.6 ± 8.2; right OE: 29.4 ± 10.7; n = 20 embryos), than did the wild-type (left OE: 59.1 ± 5.3; right OE: 58.1 + 4.7; n = 20 embryos). An additional phenotype observed was that the calretinin labeling of the olfactory nerve in the lre mutants was particulate in appearance (Fig. 6B, arrow) as compared to that seen in the wild-type (Fig. 6A, arrow). This phenotype may result from a disruption in calcium regulation as calretinin is proposed to act in calcium buffering within the neuron (Baimbridge et al., 1992).
We used anti-acetylated tubulin immunolabeling to examine the gross structure of the CNS axonal scaffold of lre mutants. No gross defects were evident in the anterior commissure (Fig. 7A,B, arrow), post-optic commissure (Fig. 7A,B, asterisk), or trigeminal ganglia (Fig. 7C,D, arrow) as judged by anti-acetylated-immunoreactivity (-IR) at 3 days of development. Thus, while lre mutants showed a reduction in the number of olfactory sensory neurons and a disruption in the structure of the olfactory glomeruli in the olfactory bulb, they showed no obvious gross defects in the axonal scaffolding of the developing CNS.
Presence of Pioneer Neurons in lre Mutant Fish
We have previously reported the existence of a population of neurons, distinct from the olfactory sensory neurons, that forge the first pathway from the olfactory sensory epithelium to the developing olfactory bulb. These pioneer neurons are an early developing population of neurons, and can be visualized using the zns-2 monoclonal antibody (Whitlock and Westerfield, 1998, 2000). Using this antibody, we found that lre mutant animals had pioneer neurons, but that they were abnormal in their morphology (Fig. 8). At 24 hr, the patterns of zns-2-IR in the wild-type (Fig. 8A) and the mutant (Fig. 8D) appeared the same. By 36 hr, zns-2-IR showed recognizable glomerular-like formations in the developing olfactory bulb of the wild type animals (Fig. 8B, arrow) like those previously reported in juvenile zebrafish (Dynes and Ngai, 1998), whereas such formations were absent in lre mutant animals (Fig. 8E). At 52 hr, zns-2-IR of wild-type fish showed an increased glomerular complexity (Fig. 8C, arrow). By contrast, the glomeruli of lre mutants visualized by zns-2-IR were still reduced and less distinct (Fig. 8F, arrow). Therefore, although lre mutant fish did have pioneer neurons in the developing olfactory bulb (as judged by zns-2-IR), there were fewer glomeruli than normal and the condensations were less distinct.
Projection of Olfactory Axons in Adult lre Mutants
The defects in axonal projections observed in lre juvenile fish were still present in the adult animals (Fig. 9). In adult fish, there normally is a characteristic pattern of termination of the olfactory sensory axons in the glomerular layer of the olfactory bulb (Baier and Korsching, 1994). In wild-type animals, it is possible to see the entry of the olfactory nerve (Fig. 9A, arrow; transverse section) as well as the clusters of axon terminals around the edge of the olfactory bulb (Fig. 9C, arrowheads, horizontal section). In contrast to the wild-type animals, adult lre mutant fish showed a less distinct olfactory nerve (Fig. 9B, arrow) with fewer axonal projections (Fig. 9D, arrowheads, horizontal section) around the edge of the bulb. In addition, the olfactory bulb appeared much smaller in the lre mutant (Fig. 9B,D) than in the wild-type animals (Fig. 9A,C). This reduction was approximately 15% in the medial lateral axis and 30% in the dorsal-ventral axis. Thus, in the lre mutant the olfactory bulb was smaller, with fewer axons terminations, and these terminations had a disorganized appearance.
Electrophysiological Responses to Odors in the lre Mutant
In order to determine whether the olfactory epithelium of lre mutants was physiologically active, we made electro-olfactogram recordings from the adult zebrafish (Fig. 10; n = 5 per genotype and odorant tested). A univariate analysis of response versus strain (wild type or lre) and odor yielded a significant odor main effect (response to artificial water (AFW) is different from responses to other stimuli) but a non-significant genotype main effect and a non-significant genotype-odor interaction. Analyses using individual independent t-tests revealed that only the response of lre mutants to L-cysteine was significantly reduced compared to control (P < 0.05), although the response of the mutant fish to L-cysteine was clearly not zero. Thus, in general, the physiological responses to the odorants tested were not statistically different between the wild type and lre mutants, although a decrement in the response to L-cysteine may reflect the loss of sensory neurons in the periphery.
Behavioral Screens for Genes Affecting Olfactory Behaviors
Here we describe the isolation and initial characterization of an olfactory behavioral mutant in zebrafish. lre mutant fish did not show the strong olfactory avoidance to L-cysteine that is seen in wild-type fish as early as 3 days of age. As young adults, the lre mutants continue to lack a response to L-cysteine and in addition do not respond normally to the amino acids L-serine and L-alanine. Thus far, the behavioral defect appears to be specific to amino acid responses (food cues), as lre fish were able to respond to taurocholic acid (social cue) (Sorensen and Caprio, 1997). The physiological responses of the lre mutants, as measured by electro-olfactogram, were overall quite similar to those of control fish. Nevertheless, lre fish showed a defect in the number of OE neurons, and in the appearance of the olfactory bulb.
In both Drosophila (Carlson, 1996) and C. elegans (Chou et al., 1996), behavioral genetic screens have been used successfully to uncover mutations affecting olfactory behaviors. The subsequent analysis of these mutants has defined genes important for olfactory system function. For example, the Drosophila acj6 gene, which was defined using a behavioral olfactory screen, encodes a POU domain-containing transcription factor that appears to play a role in the expression of olfactory receptors in the fly's antennae (Clyne et al., 1999). By contrast, other genes originally defined on the basis of the olfactory behavioral defects of mutants have subsequently been shown to have a much broader function. For example, the smell blind mutants (Lilly and Carlson, 1990), have since been shown to be allelic to mutations in paralytic, the Drosophila sodium channel gene (Lilly et al., 1994). Similarly, olfactory behavioral screens in C. elegans have resulted in the isolation of mutations affecting both the development and function of the olfactory sensory system (Sengupta et al., 1993). These mutants have led to the identification of genes coding for olfactory receptors such as odr-10 (Sengupta et al., 1996), as well as cyclic nucleotide gated channels involved in chemoreception, and that also cause axon guidance defects in olfactory sensory neurons (Coburn and Bargmann, 1996; Coburn et al., 1998).
The lre mutant was isolated among the progeny of 50 mutagenized fish using a behavioral assay. Because the olfactory sensory neuron phenotype has segregated with the behavior phenotype over five generations and with hundreds of embryos scored, we assume that both are caused by a single mutation, although they could also be due to two closely-linked lesions. A definite resolution of this question will have to wait until the mutation is mapped, cloned, and rescued. We also do not know if the lre mutation affects a function that is only required for olfactory system development and/or function, or whether it represents a hypomorphic allele of a gene that is required more generally for CNS or even for cell function. In Drosophila, for example, adult flies bearing the optomotor blind (omb) mutant allele, ombH31, show an impaired optomotor behavioral response; yet, more severe mutant alleles are lethal (Pflugfelder and Heisenberg, 1995). Likewise, in zebrafish, a behavioral screen based on the fish's escape response was used to isolate the night blind a (nba) mutation (Li and Dowling, 1997; Li, 2001). Interestingly, nba heterozygotes show degeneration that is restricted to the retina, whereas homozygotes die as embryos, show degeneration throughout the CNS, as well as abnormalities in other organs such as the heart (Li, 2001; Maaswinkel et al., 2003), indicating that nba has functions outside of the retina. Future isolation of additional lre alleles, and mapping and molecular identification of the lre gene will aid in resolving these important questions.
In zebrafish, the use of behavioral genetic screens to uncover mutations affecting nervous system function was first proposed by George Streisinger (Grunwald et al., 1988). In subsequent years, behavioral genetic techniques have been used to isolate mutations affecting motor (Granato et al., 1996) and visual (Brockerhoff et al., 1995) responses. Here we report the first use of a behavioral genetic approach to screen for mutations affecting the development and/or function of the olfactory sensory system in the zebrafish. Our pilot screen shows several features that are useful for a high throughput behavioral screen. In particular, animals were tested at a relatively young age (3–4 days) thereby reducing the time that carrier females must be maintained. In addition, the normal response to L-cysteine is robust and fast, with most wild-type fish rapidly moving away from the test odor. This feature makes subtle deviations from a normal response easily detectable even when testing the progeny of carrier females.
Zebrafish Behavioral Responses to Odorants
It has previously been reported that adult zebrafish show a behavioral response to pheromones (Chen and Martinich, 1975; Bloom and Perlmutter, 1977; Van den Hurk et al., 1987; Van der Kraak, 1989; Hall and Suboski, 1995) and to amino acids (Steele et al., 1990, 1991). Zebrafish clearly respond to alarm pheromone as well as to the amino acids, L-serine, alanine, and cysteine with characteristic attractive or aversive responses, similar to those described in other fishes (von Frisch, 1938, 1941; Hara, 1971, 1993). Amino acids, in general, play a role in feeding responses in fishes. Whether they move toward or away from a given amino acid depends on the concentration, the species of fish, and perhaps on whether it is a learned response (Valentincic et al., 1994). The developmental onset of behavioral responses to amino acids in fishes is not well characterized, although it has recently been shown that juvenile zebrafish respond to amino acid mixtures starting on the fourth day of development (Lindsay and Vogt, 2004). The later time of onset of behavioral response reported in that study compared to what we report here may be due to the lower temperature at which the embryos were raised or to the fact that they used amino acid mixtures whereas we used single amino acids. L-cysteine, which elicits the earliest behavioral response in juvenile zebrafish in our study, is also an effective amino acid stimulus in adult zebrafish (Michel and Lubomudrov, 1995). Similarly, it has been shown that juvenile trout respond to excitatory amino acids known to be involved in complex feeding behaviors in the adult as early as the alevin stage (the time before they have resorbed their yolks) (Valentincic et al., 1999). This stage is developmentally comparable to 3–4 days of age in zebrafish (Kimmel et al., 1995).
Segregation of Physiological Responses Accompanies Distinct Behavioral Responses to Odorants
There is a large literature demonstrating that in fishes amino acids are potent olfactory stimulants of feeding behaviors (Jones, 1992), and that bile acids are potent olfactory stimuli that appear to function in the recognition of conspecifics (Li et al., 1995). Strikingly, the lre mutants show a loss of or abnormal response to the amino acids L-cysteine, L-serine, and L-alanine when compared to their siblings, yet they show a normal behavioral response to taurocholic acid, a bile acid. The sensory processing of these odorants is such that there is a chemotopy in the olfactory bulb. Studies using calcium imaging (Friedrich and Korsching, 1997; Friedrich and Korsching, 1998) and physiological recording (Nikonov and Caprio, 2001), to characterize the activity patterns elicited in the olfactory bulb by amino acids and bile salts, have demonstrated that the medial regions of the bulb process bile salt input and the lateral regions amino acids and nucleotides. This chemotopy, which appears to separate the processing of feeding cues (amino acids) from social cues (bile salts), is present in both channel catfish (Nikonov and Caprio, 2001; Nikonov et al., 2005) and zebrafish (Friedrich and Korsching, 1998). Thus the loss of behavioral responses to the amino acids but not taurocholic acid in the lre mutant suggests that the defect may result in part from the disruption and or loss of the sensory fibers in the lateral chain of glomerular modules that respond to amino acids.
In the olfactory system, axon guidance is a multi-step process with pioneer neurons initiating the connection between the developing olfactory placode and the telencephalon (Whitlock and Westerfield, 1998) and olfactory receptors contributing to the targeting of the axons to the glomeruli in the developing olfactory bulb (Lin and Ngai, 1999; Mori et al., 1999). Loss of pioneer neurons results in misrouting of the axons of the olfactory sensory neurons into neighboring commissures in the zebrafish (Whitlock and Westerfield, 1998). lre mutant animals had pioneer neurons and the axons of the olfactory sensory neurons did not misroute into neighboring commissures. Nevertheless, these mutants did show a disruption of the pioneer neurons' axonal processes within the developing olfactory bulb. Therefore, the olfactory sensory axon defect observed in lre fish is not a defect in pathfinding from the OE to the olfactory bulb, but most likely related to the segregation of axons within the bulb.
It has been proposed that the olfactory receptors play a role in axon guidance, raising the possibility that the anatomical defects of lre mutant fish could be due to abnormal olfactory receptor expression. Olfactory receptors have been isolated in the zebrafish and their genomic organization analyzed (Barth et al., 1996; Kratz et al., 2002). In zebrafish there are approximately 100 receptor genes, which fall into two genomic clusters, although only one cluster has been well characterized (Kratz et al., 2002). In PCR-based studies using primers designed to amplify olfactory receptor families 2, 4, 7, 5, 9, 13 (Rivard et al., 2003), we tested whether there was a difference in receptor expression between the lre mutant and the wild type animals. All receptors tested were detectable; however, a quantitative analysis was not carried out and more subtle differences in the levels of expression cannot be ruled out (M. Rivard and K.E.W., unpublished data). The fact that these receptors were expressed in lre mutants is consistent with our finding that the olfactory sensory neurons are physiologically responsive to odorants. Thus, preliminary analysis of receptor expression suggests that the axon guidance defects observed in lre are not correlated with gross changes in olfactory receptor expression.
Here we have isolated a mutant with a clear defect in the behavioral response to amino acids. However, the lre mutant's etiology is complex: lre mutant animals showed relatively normal electrophysiological responses to various odors including L-cysteine, a reduced number of olfactory neurons, and altered glomerular structure of the olfactory bulb. The relative simplicity of the behavioral screen used should allow for the isolation of additional lre alleles, which, together with mapping and subsequent molecular identification of the lre gene, will lead to a more complete understanding of the function of lre in olfactory function and/or development. In addition, the use of our screen for the isolation of other mutants defective in olfactory behavior coupled with the application of the simple tests used here for characterizing the defects of lre mutants, will allow for the genetic dissection of this complex sensory system in a vertebrate.
In order to develop a genetic screen for defects in olfactory behavior, we first characterized behavioral responses of juvenile wild-type zebrafish to amino acids. We were particularly interested in behavioral responses that could be elicited in young fish, as they would simplify the subsequent screening for mutations affecting olfactory behaviors. Juvenile zebrafish hatch by the third day after fertilization and previous work has shown that juvenile zebrafish respond to amino acid mixtures at 4 days after fertilization (Lindsay and Vogt, 2004). Juvenile zebrafish were tested in 35-mm Petri dishes on day 3–4 and scored as to whether they moved when the odorant was added, which determined whether the odorant should be pursued further. Ten dishes, each dish containing 20–25 juvenile fish in 5 ml of embryo medium (Westerfield, 1993), were dechorionated at two days post fertilization and tested in the afternoon of the third day (78 hr after fertilization), which is before the juveniles become vibration sensitive (day 4; K.E.W., unpublished data). Each Petri dish was placed on a white background and centered on a 0.5-cm circle that was drawn on the white background (see Fig. 1A–D). The fish were gently swirled into the center of the dish, and 10 μl of odorant solution was added to the center of the dish, taking care to not disturb the water surface. The number of fish that had moved out of the inner circle 30 sec after odorant delivery was scored (Fig. 1C,D). Fish lying on the border of the inner circle were counted as being in the inner circle. Statistical analyses of juvenile fish behavior were done using a t-test for Independent samples using the VassarStats Program (http://faculty.vassar.edu/lowry/VassarStats.html). Odorants L-serine, L-alanine, and L-cysteine (Sigma Aldrich) were each tested with concentrations of 10−4, 10−5, and 10−6 M. Distilled water was the diluent and was also tested.
We used fish from an ongoing genetic screen (Henion et al., 1996) in which male fish were mutagenized with N-ethyl-N-nitrosourea, (ENU), which is known to make point mutations and small deletions. Mutagenized males were crossed to wild-type females (AB strain, Oregon) and their progeny raised to adulthood. Eggs from individual female progeny (Fig. 2A) were then collected, activated with UV-inactivated sperm, and pressure treated in order to create progeny diploid for the female genome (Corley-Smith et al., 1999). Diploid progeny were then tested at 3 days for their response to L-cysteine. Females that carried a mutation that impaired the behavioral response to this amino acid would be expected to produce approximately 50% homozygous mutant fish, whereas the remaining animals would be normal. In the absence of a mutation that interfered with the behavioral response to L-cysteine, approximately 15% of the progeny would be expected to not respond to this odorant (the average fraction of wild-type fish that do not appear to respond to this odor; see Results and Fig. 1I). Thus, clutches in which >20% of fish failed to respond were considered for further testing. For this they were rinsed and re-tested 2 hr later. Poor responders were then tested for touch sensitivity. Those not responding to touch (2 clutches) were eliminated as not being olfactory-specific mutants.
The progeny of two females (out of 50 tested) included >20% non-respondents in the screen described above and these females were out-crossed separately to wild-type fish (AB strain, Oregon) and the two corresponding families generated (Fig. 2B). The F1 progeny from these crosses would have consisted of fish heterozygous for the original mutagenized chromosomes (ca. 50%) and the rest would be homozygous wild-type. Single pair matings of F1 progeny were set up for each family and their F2 progeny screened at 3 days for lack of L-cysteine (3 × 10−5 M) aversive response (Fig. 2B). For one of the 2 families, 100% of the F2 progeny of each of 20 different pair matings showed a wild-type response and this line was not pursued further. For the second family, 4 out of 7 single pair crosses produced embryos that showed the original behavioral phenotype (see Fig. 1I). These clutches were raised and re-tested as individuals at 2 and 3 weeks (Fig. 2B). The individuals not responding to L-cysteine at 2 and 3 weeks were isolated and grown to adulthood. The mutation present in this family was named laure (lre).
We designed a Y-maze scaled to adult zebrafish based on prior experience using juvenile salmon (K.E.W., unpublished data). Our Y-maze was built from opaque, white 5/16 inch Plexiglas®, with a volume of 20 liters and dimensions of 24” length × 10” width × 6” height. This custom-designed behavior testing system was integrated into our recirculating zebrafish facility racks (Aquatic Habitats, Apopka, FL). The behavior rack housed both the experimental fish (held in 9-liter tanks) and the Y-maze in which they were tested. The common recirculating water eliminated both the possibility of confounding effects due to olfactory cues present in different waters, as well as the stress of being transferred to new waters. Each arm of the Y-maze had a fitting where odorants were added via an IV drip system. The flow rate was adjustable and our experiments were done using a flow rate of approximately 1.2 liters per minute. During odor exposure, the outflow on the Y-maze was run directly into a bucket eliminating the introduction of the odorants into the recirculating system. The whole testing rack was covered with a white semi-opaque cover, which created a fairly uniform visual world around the Y-maze and dispersed the glare from the overhead lights.
Fish were transferred to the maze and allowed to acclimate to the new environment for at least 6 hr before testing. The behavior was then recorded for 10 min on a digital video camera (SONY DCR TRV11) using the remote setting, thereby eliminating human movement around the tank during the recording session. For each trial, the temperature, pH, conductivity, and outflow volume of the Y-maze were recorded. The behavior videos were loaded onto the computer using Adobe Movie Maker. An overlay was used on the screen to divide the Y-maze into quadrants (see Fig. 3B) and the number of fish in each quadrant was recorded every 15 sec for 10 min. The data were plotted as the average number (± SEM) of fish per quadrant over the 10-min period (see Fig. 3C).
Behavioral Responses to Other Odorants
We tested lre fish and their control siblings to determine whether they were defective in response to other odorants. For this, we tested 4-week-old fish from five separate clutches using amino acids L-cysteine, L-serine, L-alanine, and the bile acid taurocholic acid. The control odorant was distilled water that was used to make up the amino acid and bile acid solutions. All fish were tested five times over the course of a week starting at the end of the fourth week after fertilization. For each test, 10 fish were placed in one liter of system water in a container (Rubbermaid 2.4 liter food service container, l = 8.5”, h = 4”, w = 4.5”). The container was divided in half and the odorant (10−5 M) was added to one side. The fish were observed and their position, whether they were in the half with the odor or without the odor, was recorded every 30 sec for 2 min. Statistical analyses of juvenile fish behavior were done using a t-test for Independent samples using the VassarStats Program (http://faculty.vassar.edu/lowry/VassarStats.html).
Trials with wild-type zebrafish were initially done using 20 fish per trial (Fig. 1E–H). Trials were later modified to include only 10 fish (Fig. 4) to streamline counting of fish.
All electrophysiological recording procedures used have been approved by the University of Utah Animal Care and Use Committee. Prior to a recording session, zebrafish were immobilized with an intramuscular injection of Flaxedil (60 mg/g body weight), placed in a silastic-polymer (Sylgard) recording chamber, and immediately provided with flowing artificial fresh water [AFW; (in mM) NaCl, 3; KCl, 0.2; CaCl2, 0.2; HEPES, 1; pH 7.2] to the olfactory epithelium (OE). They were also provided with a separate gill irrigation flow of approximately 3 ml/min of AFW containing a general anesthetic (MS-222, 20 mg/L in AFW). In order to prevent the loss of afferent sensory activity (Spath and Schweickert, 1977), the fish were anesthetized only after immobilization, thus minimizing contact between the anesthetic and the OE. Each zebrafish was given 10 min for the anesthetic to fully act before the small flap of the epithelium covering the left olfactory organ was surgically removed to expose the olfactory lamellae. Throughout the experiment, reflex movements of the gills or eyes were monitored and, if noted, additional anesthetic was provided by increasing the gill irrigation flow. Following each experiment, and while still anesthetized, each fish was measured (total body length) and weighed, then killed by decapitation and sexed.
The olfactory responses of wild-type and lre mutant zebrafish were measured using EOG recording methods as described previously (Michel and Lubomudrov, 1995; Michel and Derbidge, 1997). EOG recordings detect the extracellular ionic flux associated with the summed receptor potentials of the odor-activated olfactory sensory neurons as a negative DC voltage potential shift. A reference electrode was placed on the head and a recording electrode was placed in the left olfactory organ, between adjacent olfactory lamellae and near the midline raphe. Both electrodes were fabricated from silver/silver chloride wires bridged to the fish by way of 3 M NaCl/agar-filled (1–3%) glass electrodes with approximately 10-μm tip diameters. A silver/silver chloride wire, placed in the AFW bath directly beneath the body of the fish, served as the ground electrode. The responses to the olfactory stimuli were amplified (2,000–5,000x) and filtered at 1–2 kHz by a low-noise, differential, DC amplifier and stored digitally (100 Hz; Digidata 1200 A/D board and Axoscope software, Axon Instruments, Union City, CA). Before beginning an experiment, a stable baseline and a response of at least 0.5 mV to 100 μM L-glutamine were required.
The AFW solution was delivered to the OE at a rate of 3 ml/min. A rotary loop injector (Rheodyne, Inc., Rohnert Park, CA) was used to introduce the test odorant (50 ml) into the olfactometer flow. Dye calibration determined that the odor solution arrived at the OE in approximately 8 sec, achieved peak odor concentration (approximately 84% of stock) at around 10 sec, and then returned to baseline levels in another 12–15 sec. The concentrations reported have not been corrected for this dilution. To minimize adaptation, at least 2 min were allowed between odorant tests. The odorants tested in these experiments were: the amino acids: L-glutamine (Gln), L-arginine (Arg), L-glutamate (Glu), and L-cysteine (Cys); the polyamines: agmatine (AGB) and spermidine (Spd); and the bile salt, taurocholic acid (TCA). The amino acid L-glutamine was used as a positive control to establish the viability of the preparation and to allow comparisons with earlier studies. With the exception of TCA (10 μM) and Spd (10 μM), all of the odorants were tested at 100 μM. All chemicals were obtained from Sigma Aldrich.
In order to visualize the olfactory sensory neurons, an anti-calretinin antibody, recognizing calretinin, a calcium binding protein, was used (SWANT, Switzerland). To recognize the pioneer neurons (Whitlock and Westerfield, 1998), we used the zns-2 mouse monoclonal antibody (Trevarrow et al., 1990). Embryos were collected and maintained at 28.5°C in embryo medium (Westerfield, 1993). At each desired stage (24, 36, 48, and 72 hr post-fertilization), embryos were fixed overnight at 4°C in 4% paraformaldehyde in fix buffer (Westerfield, 1993). After fixation, embryos 48 hr and younger were rinsed 3 times for 10 min in 0.1 M phosphate buffer (PB; Westerfield, 1993), once for 10 min in dH2O, and permeabilized in acetone for 7 min at −20°C. For anti-calretinin labeling, embryos 72 hr and older were also permeabilized for 10 sec in trypsin (Sigma; 1 mg/ml in dH2O), rinsed 2 times in dH2O, and 3 times in 0.1 M PB with 0.1% Triton X-100, once in 0.1 M sodium citrate and incubated overnight in 0.1 M sodium citrate in a water bath at 37°C. After blocking in PBDT [0.05 M PB, 2% bovine serum albumin, 1% dimethylsulfoxide (DMSO) and 0.5% Triton X-100 with 4% normal goat serum (ngs)] for a minimum of 2 hr, embryos were incubated in primary antibody overnight at 4°C (anti-calretinin, 1:1,000; zns-2 mouse monoclonal antibody, 1:1,000), and washed 3 times in PBDT.
For fluorescent labeling, embryos were incubated in Alexa Red conjugated goat anti-mouse secondary antibody (1:1,000; Molecular Probes, Eugene, OR) for 4 hr at room temperature or overnight at 4°C. The grainy nature of the signal observed in some photos was due to fluorescent debris that sometimes remained when the fluorescent secondary (Alexa Red) was not centrifuged prior to use. They were then rinsed in PBS, mounted in 90% glycerol, and viewed under a conventional fluorescence microscope as well as a Leica DMR confocal microscope system. For the diaminobenzidine (DAB) coloration reaction, embryos were incubated overnight in peroxidase conjugated goat-anti-mouse antibody (Sternberger Monoclonals; 1:200 in PBDT) at 4°C, rinsed, and incubated overnight in mouse-PAP (Sternberger Monoclonals, 1:500 in PBDT) at 4°C. Embryos were then washed 2 times for 15 min in PBDT, 2 times for 15 min in 0.1M PB, and incubated for 20 min in a DAB mixture (0.05 M PB, 1% DMSO, 0.05% DAB; Sigma Chemical Co., St. Louis, MO). The DAB mixture was removed and replaced with DAB solution containing 0.1% hydrogen peroxide. The reaction was monitored under a dissecting microscope. When the labeling was deemed sufficient, the DAB solution was removed and the embryos were rinsed with 0.1 M PB and mounted in 90% glycerol for viewing.
Embryos were fixed overnight at 4°C in 4% paraformaldehyde, washed in 0.1M PB, and embedded in agar (1.5% agar, 5% sucrose). Blocks with embryos were trimmed and placed in 30% sucrose at 4°C. The preparations were cryosectioned at 20-μ intervals on a Zeiss microtome. Sections were collected on Superfrost slides, dehydrated, and mounted using DPX medium (Electron Microscopy Sciences). Because males and females are different sizes, only males were used.
To visualize olfactory sensory neurons, embryos were stained with the anti-calretinin antibody using the DAB coloration reaction. After staining, the heads were removed and mounted anterior up for the best view of the olfactory organs. Using a Camera Lucida, individual olfactory sensory neurons cell bodies were traced and counted.
K.E.W. thanks M. Rivard for analysis of olfactory receptors expression and Dr. J. Ewer for critical reading of the manuscript. Additional sectioning was also performed by the Department of Biomedical Sciences Facility, Cornell University. This work was supported by a Howard Hughes Summer Scholar Research Award (A.V.), an APA Minority Fellowship (R.R.), and NIH R01 DC04218 (K.E.W.) and DC01418 (W.C.M.).