Kappe neurons, a novel population of olfactory sensory neurons

Abstract

Perception of olfactory stimuli is mediated by distinct populations of olfactory sensory neurons, each with a characteristic set of morphological as well as functional parameters. Beyond two large populations of ciliated and microvillous neurons, a third population, crypt neurons, has been identified in teleost and cartilaginous fishes. We report here a novel, fourth olfactory sensory neuron population in zebrafish, which we named kappe neurons for their characteristic shape. Kappe neurons are identified by their Go-like immunoreactivity, and show a distinct spatial distribution within the olfactory epithelium, similar to, but significantly different from that of crypt neurons. Furthermore, kappe neurons project to a single identified target glomerulus within the olfactory bulb, mdg5 of the mediodorsal cluster, whereas crypt neurons are known to project exclusively to the mdg2 glomerulus. Kappe neurons are negative for established markers of ciliated, microvillous and crypt neurons, but appear to have microvilli. Kappe neurons constitute the fourth type of olfactory sensory neurons reported in teleost fishes and their existence suggests that encoding of olfactory stimuli may require a higher complexity than hitherto assumed already in the peripheral olfactory system.

Figure 1. G_o-like immunoreactivity reveals a distinct population of sparse, pear-shaped sensory neurons in zebrafish olfactory epithelium.

(a) G_o-ir (green) is seen in a sparse population of pear-shaped cells in horizontal sections of the olfactory epithelium (short-fixed), using DAPI as counter-stain (blue); r, radial distance. Top right inset at higher magnification shows a G_o-ir-positive cell (green), co-labeled with zns2 (red), and visible nucleus (DAPI, blue). Bottom left inset at higher magnification shows a G_o-ir-positive cell with initial axon segment (ax) and cap (cp). (b) At higher magnification the apical position of G_o-ir-positive cells (green) is clearly visible. ø_v, vertical cell diameter; ø_h, horizontal cell diameter; h, laminar height; dotted half-circle, the apical dendritic part of G_o-ir-positive olfactory sensory neurons resembles a cap. (c) Nine G_o-ir-positive cells show the typical range of morphologies for these neurons. (d) Whole mount of adult zebrafish olfactory bulb double-labeled with anti- G_o and anti-zns2 antibodies, dorsal view. Zns2 labels all glomeruli, whereas G_o-ir labels a single medial glomerulus (yellow).The olfactory nerves were cut at the entrance to the olfactory bulb before staining. (e) Horizontal vibrotome cross-section (100 μm) reveals the extremely dorsal position of the Go-immunoreactive glomerulus in each olfactory bulb. A single, thick axon bundle is seen entering the glomerulus. (f,g) One shape parameter and three spatial parameters were quantified for the G_o-ir-positive cell population, and shown as histogram (f) and empirical cumulative distribution function, ECDF (g). From left to right: ratio of horizontal to vertical diameter, laminar height (normalized to maximal height), radial distance (normalized to maximal radius), and number of cells per 10 μm horizontal cross section of the olfactory epithelium; x axis units and labels are valid for both (f) and (g). Scale bars correspond to 50 μm (a), 5 μm (b), and 100 μm (d, e).

Figure 2. G_o-ir-positive neuron population is different from crypt neurons.

(a) Double labeling with anti-G_o antibody (green) and anti-S100 antibody (red, marker for crypt neurons) in cryostat sections of short-fixed olfactory epithelium reveals two mutually exclusive sensory neuron populations. Insets, single neurons at higher magnifications. Note the differences in morphology of these two cell populations; arrow in top right insert, the cap-like structure typical for G_o-ir-positive neurons. (b–c) Higher magnifications show the typical shapes of G_o-ir-positive neurons (pear-shaped) and crypt neurons (globose), indicated by filled arrow heads and open arrow heads, respectively. (d–e, g–h) One shape parameter and three spatial parameters (see Fig. 1b for graphical explanation) were quantified for the TrkA-ir-positive cell population and the corresponding empirical cumulative distribution function, ECDF, was compared with that of G_o-ir-positive neurons; ***, distributions of TrkA-ir and G_o-ir cells are significantly different (p < 10⁻⁶), as assessed by Kolmogorov-Smirnov test of the unbinned distributions. (d), ratio of horizontal to vertical diameter [diameter ratio (ø_h/ø_v)], (e), laminar height normalized to maximal height is shown. (f), Absence of co-label for G_o-ir and TrkA-ir cell populations. (g) Relative radial distance of labeled cells is shown. (h) Number of cells per 10 μm horizontal cross section of the olfactory epithelium was analysed for G_o-ir and TrkA-ir-positive neurons. (i) Maximal vertical distance (Δ max) of distributions as indicated; ø ratio, diameter ratio; height, normalized laminar height; radius, normalized radial distance; z axis, section number (ordinal). Vertical distance can range between 0 (identical curves) and 1 (no overlap of x value range). Scale bars correspond to 100 μm (a) and 10 μm (b, c).

Figure 3. G_o-ir-positive neurons do not co-localize with established markers for ciliated and microvillous neurons.

(a–d), Double labeling of G_o-ir-positive cells with different markers is analysed in horizontal cryostat sections of olfactory epithelia; dashed line, basal border; dotted line, apical border of the sensory layer; scale bar, 20 μm. (a) Double fluorescent labeling of anti-Go antibody (green) with RFP (red) expressed in ciliated neurons in Tg(OMP:lyn-mRFP) shows absence of co-localization; filled grey arrowhead, G_o-ir-positive cell; open arrowhead, RFP-positive cell. (b) Double fluorescent labeling of anti-G_o antibody with Venus expressed in microvillous neurons in Tg(TRPC2:Venus) line shows absence of co-localization. G_o-ir signal is set to green, Venus signal is set to red; filled arrowhead, G_o -ir-positive cell; open arrowhead, Venus-positive cell. (c) Double fluorescent labeling of anti- G_o antibody (green) with in situ hybridisation signal from TRPC2 probe shows absence of co-localization; filled grey arrowhead, G_o-ir-positive cell; open arrowhead, TRPC2-positive cell. (d) Double fluorescent labeling of anti-G_o antibody (green) with anti-calretinin antibody (red) shows absence of co-localization; filled grey arrowhead, G_o-ir-positive cell; open arrowhead, calretinin-positive cell. (e) The empirical cumulative distribution function (ECDF) for a cell shape parameter (diameter ratio) shows distributions for TRPC2 and OMP-positive cells to be different from each other as well as from G_o-ir and TrkA-ir, shown for comparison here. (f) Quantification of co-label for G_o-ir and markers for microvillous, ciliated and crypt neurons (as indicated) shows 0 to 2% co-label (yellow) in G_o-ir-positive neurons. Such small percentages amount to a handful of cells in an entire olfactory epithelium, and are likely to accrue from the dense packing of cells, dendrites, cilia and microvilli, at the limit of light-microscopic resolution. (g) The empirical cumulative distribution function (ECDF) for a cell localisation parameter (laminar height) shows distributions for TRPC2 and OMP-positive cells to be different from each other as well as from G_o-ir and TrkA-ir-positive cells, shown for comparison here. (h) Maximal vertical distance (Δ max) of distributions as indicated; ø ratio, diameter ratio; height, normalized laminar height. Significance of distribution differences is assessed by Kolmogorov-Smirnov test of the unbinned distributions; a, p < 10⁻⁶; b, p > 0.6.

Figure 4. Kappe neurons are tubulin-negative and actin-positive.

Double labeling of G_o-ir-positive cells with anti-tubulin or anti-actin antibody is analysed in horizontal cryostat sections of olfactory epithelia. (a) Double fluorescent labeling of G_o-ir (green) and tubulin (red) shows absence of co-localization; scale bar 10 μm. The insets represent magnified images of single neurons taken at 100× magnification, 0.1 μm optical sections. (b) Double fluorescent labeling of G_o-ir (green) and actin (red) shows co-localization: G_o-ir positive neurons exhibit highly localized actin staining at the apical surface of their cell bodies, the expected position for microvilli. Scale bar 10 μm. The insets show single neurons, images taken at 100× magnification, 0.1 μm optical sections. (c) Quantification of co-label for G_o-ir and actin or tubulin, respectively, shows over 90% co-label (yellow) for actin, but less than 10% co-label for tubulin. The small number of G_o-ir/tubulin co-labeled cells is likely to result from the dense packing of cells, dendrites, cilia and microvilli, at the limit of light-microscopic resolution. (d) Schematic representation of four types of olfactory sensory neurons with their laminar position. Ciliated neurons (orange) have round somata and slender dendrites that terminate in bundles of cilia on the epithelial surface. They constitue the most basal layer of olfactory sensory neuron. Microvillous neurons (blue) have bundles of microvilli on their apical surface. Crypt neurons (red) are globular-shaped and carry both microvilli and cilia on their apical surface. They are located more apical than microvillous neurons. G_o-ir-positive kappe neurons (green) are pear-shaped with an apical appendage resembling a cap (German: Kappe), have no cilia, and are located even more apical than crypt neurons. Kappe neurons (green) constitute a novel olfactory sensory neuron population.