Despite the great variability of teleost peripheral olfactory organ shape and number of lamellae (Hansen and Zielinski,2005), in all fish the olfactory epithelium is composed of bipolar olfactory sensory neurons, sustentacular cells and basal stem cells. The olfactory sensory neurons project their dendritic processes, ending with olfactory knobs, to the apical surface, and their axons to the olfactory bulb, where they form topographically organized synapses with mitral cells (Thommesen,1978; Riddle and Oakley,1991; Nikonov et al.,2005). Fish possess three different types of olfactory neurons and they have different apical surface, localization of the soma and length of the dendrite (Morita and Finger,1998; Hamdani and Døving,2007). Besides ciliated and microvillous receptor cells (Hansen and Zeiske,1998; Farbman,2000), common to all other vertebrates, new egg-shaped receptor neurons, the crypt cells, have been discovered in actinopterygean fish (Hansen et al.,1997; Hansen and Finger,2000) and, recently, in some Chondrichthyes (Ferrando et al.,2006,2007). These cells are localized in the upper portion of the olfactory epithelium and bear both microvilli and cilia. In crucian carp, Carassius carassius, it has been demonstrated that crypt neurons project their axons to the ventral part of the olfactory bulb, a region associated with reproductive behavior (Weltzien et al.,2003). This finding suggested that crypt cells express olfactory receptors for sex pheromones (Hamdani and Døving,2006). However, their function is still unclear.
In a previous work (Lazzari et al.,2007), we described the histological and ultrastructural organization of the olfactory epithelium in the guppy, Poecilia reticulata. Such fish could be considered an example of microsmatic fish, due to the morphology of the olfactory mucosa and both sexual and antipredator visual-based behaviors (Evans et al.,2004; Hibler and Houde,2006). The aim of the present study was to investigate, in Poecilia reticulata, the molecular aspects of olfactory sensory neurons and their projections into the olfactory bulb, still unknown in microsmatic fish. In particular, we reported the absence of immunostaining for PGP 9.5 and the localization of calretinin, two neuronal markers of the olfactory epithelium of mammals (Bastianelli et al.,1995; Weiler and Benali,2005) and fish (Porteros et al.,1997; Saito et al.,2004; Yamamoto et al.,2004; Castro et al.,2008), and the immunolabelling pattern of a novel marker for crypt cells, S100 protein (Germanà et al.,2004,2007). In addition, we used lectin binding to evaluate the molecular characteristics of cell surface glycoconjugates, both in the epithelium and in the bulbs, and to reveal potential segregated central projections, as seen in other species (Riddle et al.,1993; Franceschini et al.,1999).
MATERIALS AND METHODS
Twenty-two adult (6–8-months old) P. reticulata of both sexes (female body length approximately 5 cm and male body length approximately 3 cm), were purchased locally from the Acquario Fossolo, Bologna, Italy, maintained in aquaria at 25°C in a natural light-dark cycle and fed once daily 6 days per week. All procedures were in conformity with the guidelines of the European Communities Council Directive (86/609/CEE), the current Italian legislation for the use and care of animals, and the guidelines of the U.S. National Institute of Health. The Ethic-Scientific Committee of the University of Bologna also approved this study.
Specimens were sacrificed in two distinct periods: 11 (six females and five males) were sacrificed in April–June, and 11 (six females and five males) in October–December. The animals were anesthetized with 0.1% 3-aminobenzoic acid ethyl ester (MS-222, Sigma Chemical, St. Louis, MO), killed by decapitation and their heads, after removal of the dorsal cranium, rapidly immersed in a modified Bouin solution consisting of saturated aqueous solution of picric acid and formalin (ratio 3:1) for 24 hr. The picric acid was removed by long washing in 0.1 M disodium phosphate buffer, pH 7.4. The heads were then decalcified in 0.25 M EDTA (ethylenediaminetetraacetic acid) in 0.1 M disodium phosphate buffer, pH 7.4 for 3–7 days, depending on the degree of bone tissue mineralization. Specimens were then embedded in Paraplast plus (Sherwood Medical, St. Louis, MO; mp 55–57°C), horizontally (twenty animals) and transversely (two animals) sectioned (5 μm) with Leitz 1516 microtome and mounted on silanated slides. Immunohistochemical and lectin histochemical investigations were performed on each animal using adjacent sections.
The sections were deparaffinized, rehydrated, immersed in citrate buffer at pH 6.0 and treated with microwaves for 10 min at 750 W for antigen retrieval. After blocking with 10% normal goat serum (NGS; Vector Laboratories, Burlingame, CA), the sections were incubated separately overnight at 4°C with three primary antibodies: (i) rabbit polyclonal antibodies against PGP 9.5 (1:300; Z 5116; DAKO Cytomation, Glostrup, DK), a neuron-specific ubiquitin C-terminal hydrolase, (ii) rabbit polyclonal antibodies against rat calretinin (1:2000; AB5054; Chemicon International, Temecula, CA) and (iii) rabbit polyclonal antibodies against bovine S100 (1:1000; Z 0311; DAKO Cytomation, Glostrup, DK), two calcium-binding proteins expressed in chemosensory neurons. After incubation the sections were washed and incubated for 1 hr and 30 min in peroxidase-labeled goat anti-rabbit IgG (1:100; Vector Laboratories, Burlingame, CA). The immunoreaction was visualized using the intensified diaminobenzidine method, as reported by Adams (1981). Negative controls were obtained by omission of the primary antibody, replaced by 10% NGS. Sections of olfactory epithelium from mouse and rat were chosen as positive controls.
Six horseradish peroxidase–conjugated lectins (Sigma-Aldrich, Steinheim, DE) were used: Wheat germ agglutinin (WGA), Glycine max agglutinin (SBA), Bandeiraea simplicifolia agglutinin isomer B4 (BSA-I-B4), Ulex europaeus agglutinin I (UEA-I), Ricinus communis agglutinin I (RCA120), Helix pomatia agglutinin (HPA). Furthermore, two biotinylated lectins (Vector Laboratories, Burlingame, CA): Lycopersicon esculentum agglutinin (LEA) and Sophora Japonica agglutinin (SJA), were also used for this investigation. For all lectins the concentration was 10 μg/mL. The time of incubation was 3 hr for the HRP-conjugated lectins and 30 min for the biotinylated lectins. In the case of biotinylated lectins, after incubation, the slides were washed in TBS (Tris Buffer Saline) 0.1 M and then incubated in the avidin–biotin peroxidase complex (Vectastain ABC kit, Vector Laboratoires, Burlingame, CA) for 1 hr in a moist chamber at room temperature. Lectin detection was performed with 3.3 Diaminobenzidine (DAB, Sigma Chemical, St. Louis, MO) following the method modified by Adams (1981). Lectin controls included: competitive inhibition with the appropriate sugar (100–200 mM) for 1 hr at room temperature, buffered saline in place of the lectin and use of mouse and rat olfactory and vomeronasal organs as positive control tissues.
Image Acquisition and Processing
Images of immuno- and lectin histochemistry were visualized using an Olympus BH-2 microscope and photographed using an Olympus DP11 digital camera. Figures were assembled by using Adobe Photoshop CS3 for the Mac and adjusted for brightness, contrast, and color balance. Images were resized and rotated for purposes of presentation, but in no case was their content altered.
Estimation of Crypt Cell Density and Statistical Analysis
We applied the physical disector method (Sterio,1984) on parallel horizontal sections obtained from the olfactory organs of 10 males and 10 females. The mean crypt cell diameter was approximately 13 μm, suggesting an optimal height of the disector around 5 μm (Gundersen, 1986), the section thickness we chose. Beginning randomly from the dorsal end of the olfactory epithelium, every 10th section and its adjacent section (the reference and the look up sections) were sampled and incubated with polyclonal anti-S100. In female olfactory organs we collected an average of 12–13 dissectors, and 8–9 in male olfactory organs. Sections excluded from the counting were used for calretinin, PGP 9.5 and lectin detection. The image analysis software ImageJ (version 1.38r) was used to measure the area of the olfactory epithelium in each chosen section and the Cell Counter plug-in (v.2) was used to count the S100-positive crypt cells. The density of crypt cells was calculated as follows:
The area calculated by the image analyser was multiplied by the mean section thickness to give Vdis.
The data collected were expressed as density of crypt cells in 100,000 μm3 and reported as mean and standard error of the mean in male and female. Statistical analysis of results was performed using Student's t-test. A P value of less than 0.05 (two-sided) was used as the level of significance for rejecting the null hypothesis that there was no difference between sexes.
The immunohistochemical and lectin patterns showed no differences between sexes and between periods of tissue sampling.
The flat mucosa of Poecilia reticulata presented an intense calretinin immunoreactivity restricted to the sensory folds, while the nonsensory areas were negative. Positive cells were distributed in the upper half of the epithelium, while cells in the lower layers appeared negative (Fig. 1A). Calretinin-positive neurons showed intensely stained dendrites and somata. Among them, some crypt cells, identifiable by their ovoid shape, were visible in the proximity of the apical surface (Fig. 1B). Calretinin immunoreactivity was both nuclear and cytoplasmic. In the lamina propria some olfactory fibres, randomly distributed in the axon bundles, were stained (Fig. 1A). Poecilia reticulata possesses sessile olfactory bulbs and, consequently, long olfactory nerves. During their route to the bulbs, the immunopositive axons observed in the lamina propria appeared more confined in the lateral portion of the olfactory nerve (Fig. 1C) and penetrated the glomerular layer in the lateral region, as could be seen in horizontal and transverse sections of the olfactory bulbs (Fig. 1D,E). The medial axon bundles and the glomeruli they form were only weakly labeled.
Also S100 staining was exclusively localized in the sensory epithelium (Fig. 1F). The crypt cells showed a very intense reactivity and their ovoid shape was well defined (Fig. 1F,G). Other cellular elements in various depths of the olfactory epithelium appeared moderately colored compared to crypt cells (Fig. 1F). They seemed to belong to different subpopulations of olfactory neurons. They exhibited reaction product in both cell body and dendrite. The germinative basal layer was generally devoid of S100 protein labelling but some scattered elements, perhaps immature crypt cells, were positive (Fig. 1H). The axon bundles, emerging from the mucosa (Fig. 1G), and the olfactory nerve were strongly colored (Fig. 1I). In contrast with the intensity of immunolabelling in the terminal conical expansion of the olfactory nerve, the S100-positive axonal endings in the bulb were only moderately stained. The glomeruli they innervated spread over different areas, even if the major amount appeared to converge in the medial region (Fig. 1I,J).
For their quantification, crypt cells were identified by their S100 immunopositivity, their typical shape and their localization in the upper layer of the mucosa. The tissues were sampled in different periods of the year but the density of crypt cells remained steady (Fig. 2A). We further observed that their number was low and that they were randomly distributed: in some regions we counted more than one crypt cell per section (see Fig. 1F,G) while in others there were no S100-positive crypt cells. Moreover, the calculated density of crypt cells in female olfactory organ was significantly higher (P = 0.013) than the corresponding male value (Fig. 2B).
We could not observe any PGP 9.5-immunoreactivity either in the epithelium or in the olfactory bulbs, while control tissues gave a well-defined positive signal (data not shown).
Among the eight lectins used in this study, only two labeled both the mucosa and the bulbs: WGA and HPA. Their binding patterns in the olfactory epithelium were very similar. The superficial mucous layer and numerous receptor cells ubiquitously distributed in the mucosa were intensely labeled and it was not possible to distinguish subpopulations (Fig. 3A,B). The reaction products were localized in the soma and in the dendritic processes, especially in their apical region, where the olfactory knobs emerge in the free surface. Crypt cells showed a very faint staining, and they were identified through their superficial localization and shape. The axon bundles in the lamina propria were strongly labeled, as the olfactory nerve and the bulbar fibre layer. Horizontal sections of the olfactory bulb showed numerous HPA- and WGA-labeled glomeruli, mostly localized in the antero-medial and lateral regions, while they were less abundant in the medial and absent in the postero-medial layers (Fig. 3D,E). In transverse sections, WGA and HPA-stained glomeruli were observed in the ventral and lateral regions, showing identical patterns (Fig. 3F,G).
Incubation with LEA exhibited binding localized in the olfactory mucosa (Fig. 3C). The surface of nonsensory epithelium, where goblet mucous cells were visible, was negative (data not shown). Some scattered crypt cells appeared positive to this lectin binding (Fig. 3C).
The reactions with the other lectins gave no evident results both in the olfactory organ and in the bulbs (data not shown).
In the present study, we investigated the possible presence and localization of calretinin, S100 and PGP 9.5, and examined the carbohydrates moieties of glycoconjugates in the olfactory system of a microsmatic fish, Poecilia reticulata. Our results demonstrate that calretinin is present in olfactory receptor cells localized in the upper half of the olfactory epithelium, and in crypt cells, while it is absent in neurons belonging to the basal layer, as observed also in other fish species (Saito et al.,2004; Yamamoto et al.,2004; Germanà et al.,2007) and in mammals (Kimura and Furukawa,1998). It is known that mouse olfactory and rat vomeronasal epithelia are organized in distinct laminar layers, each colonized by cell bodies of sensory neurons expressing the same odorant receptor gene (Strotmann et al.,1996; Inamura et al.,1999). A similar spatial segregation exists also in fish, where morphologically different olfactory receptor cells are arranged in different layers. In particular, the microvillous subtype is localized in the middle layer, while ciliated neurons are located in deep layers of the epithelium (Morita and Finger,1998; Hansen et al.,2004; Hamdani and Døving,2007), an aspect observed also in the guppy (Lazzari et al.,2007). Moreover, we observed that calretinin immunoreactivity in the olfactory nerve is restricted to the external area, in accordance with what Bastianelli and Pochet (1994) reported in the rat nerve bundles. At the bulbar level, these positive axons arborize in glomeruli of the lateral region. Previous works in Carassius carassius showed that this synaptic region, connected with the lateral olfactory tract, mediates feeding behavior (Hamdani et al.,2001b). Retrograde axonal transport studies (Hamdani et al.,2001a; Valentinčič et al.,2005) discovered that the lateral glomerular layer is innervated by microvillous olfactory neurons, known to be stimulated by amino acids (Lipschitz and Michel,2002). We could hypothesize that in Poecilia reticulata calretinin prevalently identifies the subtypes of crypt and microvillous cells in the mucosa. A relationship between calretinin expression and sensitivity to aminoacids, or other chemicals associated with feeding, could exist in the guppy. Even in salmonids a spatial segregation of calretinin-positive glomeruli exists. However, they are present also in the ventral glomerular layer, probably because calretinin-expressing olfactory neurons belong to all morphological subtypes, spreading over more bulbar regions (Porteros et al.,1997; Castro et al.,2008).
The S100 immunostaining pattern in the olfactory epithelium comprises both crypt olfactory sensory neurons and other cell types in deeper layers. We also identified some undifferentiated basal cells, even if we could not establish if they belonged to crypt cells or to other S100-positive olfactory neurons. In zebrafish, the only fish in which the S100 antiserum has been tested, exclusively crypt neurons were labeled, while their central processes and all the remaining sensory cells in the epithelium appeared negative (Germanà et al.,2004,2007). Instead, Kerschbaum and Hermann (1992) revealed, in Xenopus laevis, an immunohistochemical pattern similar to the one we observed in Poecilia. The strong immunoreactivity of the olfactory nerve depends on the localization of S100 proteins in ensheating cells. However, while in mammals S100 is restricted to ensheating glia (Au et al.,2002), in Poecilia we find it also in olfactory axons and glomeruli. In particular, the S100 positive fibres arborize and innervate different regions, preferably in the medial glomerular layer, and are not restricted to the ventral area where crypt cells were reported to project (Hansen et al.,2003; Hamdani and Døving,2006). This seems to confirm that, in Poecilia, S100 marks different subsets of sensory cells in the olfactory neuroepithelium, with topographically distinct projections. Moreover, even if in crypt cells calretinin and S100 are co-expressed, the two calcium binding protein sera show only partially overlapping patterns, suggesting that not all olfactory cells expressing calretinin or S100 belong to the same subpopulations. Calcium-binding proteins modulate neuron excitability serving different functions, not only calcium buffering (Kimura and Furukawa,1998; Donato,2003), and they could have a role in specific sensory transduction, as already suggested by Germanà et al. (2004) even if their function in olfactory sensory neurons is still unknown.
Crypt cells are supposed to respond to sex pheromones (Hamdani and Døving,2006). The counting revealed that Poeciliareticulata possesses a low number of crypt neurons, compared with the density reported by Hamdani and Døving (2006) in Carassius carassius, confirming the electron microscopical results of Lazzari et al. (2007). This difference could be related to a limited role of olfaction in the mating behaviors of guppy. In fact, it is known that female guppies choose males mainly on color pattern (Evans et al.,2004). Moreover, Hamdani and Døving (2007) reported that the number of mature crypt cells of the crucian carp increases during spawning season. In the Poecilia, instead, we observed no seasonal variations, in accordance with guppy year-round reproduction (Houde,1997). However, it is not clear why crypt cells density in females is higher than in males. Perhaps the reason resides in the roles, well distinct between sexes, of odorant cues in Poecilia mate recognition: male guppies use pheromone detection to distinguish virgin females, even if they copulate also with mated females (Guevara-Fiore et al.,2009). Male pheromones are not only useful in choosing partners in absence of visual stimuli, but seem to induce also female guppies to avoid males and reduce the risk of predation (Shohet and Watt,2004). To date, which substances crypt cells perceive has not been deeply investigated. Schmachtenberg (2006) discovered that crypt cells respond to amino acids, but without testing sex pheromones. We also reported that calretinin-positive neurons, including crypt cells, exclusively project in the lateral olfactory bulb and not in the ventral olfactory bulb; in the crucian carp the former is associated with amino acid detection and feeding behavior, while the latter with sex pheromone detection and reproductive behavior (Lastein et al.,2006). It could be possible that two subtypes of crypt neurons exist: S100-positive crypt cells and calretinin-positive crypt cells which are able to recognize odorants not necessarily involved in reproduction. To investigate this hypothesis we attempted to count the number of calretinin-positive crypt cells but the uniform staining intensity makes it hard to distinguish among all subtypes of positive neurons and, as a consequence, the estimated number could be uncorrect.
The neuroepithelium, the olfactory nerve and the bulbs of Poecilia are devoid of labelling in PGP 9.5-immunoreactions. In the olfactory system of mammals this pan-neuronal marker is expressed in differentiating neuronal precursors, as well as in differentiated neurons (Weiler and Benali,2005). Its presence in the olfactory system of fish has been investigated only in Verasper moseri, where PGP 9.5 is intensely expressed by sensory neurons in the upper three quarters of the epithelium (Saito et al.,2004; Yamamoto et al.,2004). The positive results we obtained in control tissue seem to discard any hypothesis of false negative, so we can presume that in the olfactory system of guppy PGP 9.5 is not expressed. However, we cannot totally exclude that the Dako anti-bovine PGP 9.5 antiserum did not specifically recognize this protein in the guppy.
Glycoconjugates in the mucous composition and in olfactory neuronal membranes are supposed to be involved in the reception of odorants. Furthermore, their role in cell–cell recognition could be associated with the establishment of correct synaptic connections. For these reasons terminal sugars detected by lectin binding are considered important discriminants among olfactory neuron subpopulations (Plendl and Sinowatz,1998). We reported that only two lectins, WGA and HPA, intensely react with olfactory receptor cells and their projections in the bulbs. Instead, LEA binding preferentially labels neurons in the middle layer of the olfactory epithelium. These findings suggest the presence of N-acetylglucosamine residues on the saccharidic chains of membrane glycoproteins in the olfactory receptor neurons of Poecilia, as revealed by the binding with LEA and WGA. In particular, WGA has higher affinity for dimers and trimers of this sugar, while trimers and tetramers are preferentially bound by LEA (Goldstein and Poretz,1986). A higher density of dimers of N-acetylglucosamine could explain the different patterns displayed by these two lectins. Tetramers appear restricted to a subpopulation of olfactory receptor cells in the middle layer. Moreover, LEA-bound tetramers are absent in the bulbs, while WGA-bound dimers characterize the glomerular layer. HPA lectin preferentially binds to dimers of α-N-acetylgalactosamine (Goldstein and Poretz,1986), which appear ubiquitously distributed in the membranes of guppy olfactory neurons. This sugar residue is diffused in the cell bodies, in the dendritic processes and in the axonal projections in the glomerular layer. The absence of evident SBA binding further confirms that α-N-acetylgalactosamine residues are in dimeric form, because it is known that SBA is more specific for monomers (Goldstein and Poretz,1986). Crypt olfactory sensory neurons are intensely labeled only by LEA and very weakly by WGA in contrast with the intense reaction obtained in the other receptor cells and in the mucous layer. Instead, lectin binding in Verasper moseri and in the elasmobranches Scyliorhinus canicula and Raja clavata demonstrated a high density of glycoconjugates in the crypt-like zone, reactive for PHA-L (Verasper moseri), specific for β-N-acetylgalactosamine, PNA (Scyliorhinus canicula), specific for β-galactose, and WGA (Scyliorhinus canicula and Raja clavata) (Saito et al.,2004; Yamamoto et al.,2004; Ferrando et al.,2006,2007), suggesting a wide range of odorant-receptor interactions. The almost complete absence of lectin binding we observed in the crypt cells of guppies could imply a limited glycosilation of transmembrane proteins or a reduced amount of glycoconjugates in the mucus and, as a consequence, a restricted odorant detection capability.
In summary, we reported that molecular differences exist between the olfactory systems of guppy and macrosmatic fish. In particular, the absence of PGP 9.5 immunoreactivity and a low range of carbohydrates residues identified by lectin binding, along with the morphological description of the olfactory organ (Lazzari et al.,2007), simple and reduced in size, suggest that the guppy can recognize only a restricted panel of substances compared with other teleosts (Riddle et al.,1993; Franceschini et al.,1999) or vertebrates with a developed olfactory system (Franceschini et al.,2000,2001,2003). Otherwise, calcium-binding proteins show similar immunohistochemical patterns both in macrosmatic and microsmatic fish. Crypt cells in the guppy seem to project also to the lateral olfactory bulb, as shown by calretinin immunostaining. Such cells could have other functions than sex odorant captation, even if this hypothesis remains to be investigated.