Sensory apparatus of the antennal base
The locust antenna is a true appendage (c.f. leg: Gibson and Gehring 1988; Casares and Mann 1998) comprising three articulations: a basal scape which links the antenna to the head capsule, a short intermediary pedicel, and a distal elongated flagellum which is not actively motile (Fig. 1a; see Gewecke 1972, 1979; Chapman 1982). The antennal flagellum is also subdivided into articulations but these do not represent true segments and so have been termed meristal annuli (Chapman 2002). The major nerve root of the antenna is the antennal nerve which conducts sensory fibers originating from mechanosensory hairs, propioceptive chordotonal organs, campaniform sensilla and olfactory sensilla from all three segments to the deutocerebrum of the brain (for a full description see Chapman and Greenwood 1986; Ochieng et al. 1998; Chapman 2002). In addition, the scape possesses a musculature to move the pedicel (Fig. 1a) and this musculature is innervated by axons from motoneurons in the brain (see Gewecke 1972, 1979).
Johnston´s organ is located in the pedicellar segment and in the adult locust the organ comprises symmetrical ventral and dorsal cell clusters each of which are then subdivided into medial and lateral fields. An adult cluster comprises six to seven scolopales which insert into a membrane at the pedicellar/flagellar junction (Fig. 1a; inset, i). The cell clusters project axons into pedicellar nerves that run basally to fasciculate with respective medial or lateral nerves in the scape and then join the antennal nerve (Fig. 1a; see Gewecke 1972, 1979 for details). In the hemimetabolous locust, Johnston´s organ must be functional on hatching and so develops during embryogenesis. Labeling with neuron-specific α-HRP reveals that the innervation pattern, distribution of sensory organs, and organization of clusters of sensilla in Johnston´s organ existing in the antennal base at mid-embryogenesis (Fig. 1b) already bears a great similarity to that of the adult (c.f. Figure 1a). Our findings suggest that major developmental steps building Johnston´s organ occur prior to mid-embryogenesis. We therefore decided to limit the scope of our present study, which involved screening a total of 235 preparations, to this same time span.
The pedicel, along with the other articulations of the antenna can be identified at the end of embryogenesis via epifluorescent illumination which causes the septal-like cuticular bands to autofluoresce (Fig. 1c). However, this method only functions with cuticularization (after mid-embryogenesis), and does not reveal early epithelial domains. Immunolabeling against the GPI-linked cell surface antigen Lachesin, however, can be shown to identify the epithelial domain of the pedicel at very early embryonic stages, even prior to the formation of Johnston´s organ (Fig. 1d). At mid-embryogenesis (Fig. 1e), Lachesin expression in the scape and pedicel clearly matches the segmentation of the antennal base revealed later by epifluorescent illumination (c.f. Figure 1c) and so allows an unambigous identification of these domains. In combination with neuron-specific labels such as α-HRP (see Figs. 2, 4 below), the α-Lach label allows the cellular development of Johnston´s organ to be followed from its earliest origins within a molecularly identifiable region of the antenna.
Sensory precursors and initial lineages of Johnston´s organ
We identified the sense-organ precursors (SOPs) generating the neurons of Johnston´s organ (JO) in the epithelial domain of the pedicel by triple-immunolabeling against the epithelial cell marker Lachesin (α-Lach), the mitosis marker phosphohistone-3 (α-PH3), and the neuron-specific marker horseradish peroxidase (α-HRP) (Fig. 2). Our analysis here initially focusses on the ventral epithelium but applies equally to the dorsal region (see Fig. 5 below).
Beginning at about 36% of embryogenesis (Fig. 2a), a single large Lach-positive/PH3-positive mitotically active cell is present in the region of the pedicellar domain where the clusters of Johnston´s organ will form. The precursor is itself HRP-negative, and at this age has generated a small lineage of Lach-positive progeny, two of which are also HRP-positive and so represent the initial neurons of this cell cluster of the JO. At 42% (Fig. 2b), three PH3-positive mitotically active cells are generating lineages of the JO in the ventral Lach-positive domain of the pedicel. Towards mid-embryogenesis (48%, Fig. 2c), six PH3-positive progenitors are seen distributed throughout the ventral Lach-positive domain of the pedicel, the dendritic projections of their progeny project apically towards the border with the flagellum. Comparisons across a range of preparations of this age show that the number of active progenitors and lineages in the ventral Lach-positive pedicellar domain already matches the compliment of cell clusters belonging to the mature Johnston´s organ (see Fig. 6, c.f. Figure 1).
Along with the SOP, a smaller mitotically active cell can be seen in lineages at both 42% (Fig. 2b) and 48% (Fig. 2c) of embryogenesis. A time series imaged at higher resolution (Fig. 2d: 38%, 40%, 42%) allows SOPs and their lineages to be analysed more precisely. The data establish that there is only a single large SOP associated with each progressively expanding lineage which therefore constitutes a single clone. At each age there is a smaller cell associated with the clone and in the same location with respect to the SOP. At two of these ages (38%, 42%) the cell in this location is mitotically active, while at 40% it is between cell cycles. We can clearly not be sure that this smaller cell is the same cell each time, and it could equally represent a series of similar second order precursors, each generated asymmetrically by the SOP before dividing.
3D confocal reconstructions following α-HRP labeling (Fig. 2e) document the age-dependent increase in neuronal numbers contributing to a lineage (36%:2 progeny, 38%:3 progeny, 41%:5 progeny), and reveal that these neuronal progeny are consolidated into tightly bound cartridges.
Topographic organization of the pedicellar epithelium
Our evidence up to this point is that a subset of Lach-positive/PH3-positive proliferative cells we have identified in the pedicellar domain generate the neuronal progeny of Johnston´s organ. We reasoned that if the locations of these SOPs were fixed, then this should be reflected in a conserved distribution of their initial lineages in the epithelium, which in turn would argue for a topographic organization of the Lach-positive pedicellar domain.
Support for this hypothesis takes the form of a series of confocal images showing the initial cell clusters in the ventral epithelium of the pedicel in five repeat preparations of the same age (39%) after α-HRP labeling (Fig. 3). The data clearly show neuronal clusters of Johnston´s organ as well as campaniform sensilla at remarkably conserved locations in the epithelium across these preparations. This is also the case for a range of other ages investigated (Suppl. Figure 1) from which we infer that the distribution of PH3-positive proliferative precursors generating these lineages is equally conserved and leads us to propose that the future cluster organization of Johnston´s organ is based on a topographic organization of the epithelium.
Developing neuronal clusters of Johnston´s organ
We investigated the developing pattern of cell clusters (clones) making up Johnston´s organ in the pedicellar domain by labeling with epithelial cell-specific α-Lachesin and neuron-specific α-HRP. At 40% of embryogenesis (Fig. 4a), two major clusters of Lach-positive cells are present in the ventral epithelial domain. Each cluster contains a lineage of differentiating neurons co-labeled by α-HRP, some of which are sprouting initial dendritic and axonal processes. By 42% of embryogenesis (Fig. 4b), five clusters of Lach-positive cells are present in the ventral epithelium, only three of which contain HRP-positive neurons at this stage. At 48% of embryogenesis (Fig. 4c), six Lach-positive cell clusters are now present all of which contain HRP-positive neurons with significant axonal and dendritic processes. At 55% of embryogenesis (Fig. 4d), six clusters of cells are still present in the ventral Lach domain, all contain HRP-positive differentiated neurons. These neurons have generated extensive dendritic processes projecting apically towards the flagellum, and axons projecting basally to the scape where they fasciculate with the antennal nerve to the brain (see Fig. 1b).
The cluster number we see at each developmental stage reflects the number of active progenitor cells present, and by mid-embryogenesis matches that for the ventral subregion of the adult JO (c.f. Figure 1a). Subsequent development does not, therefore, appear to involve the generation of additional lineages.
Our data above refer to events in the ventral epithelium, but there is a parallel, symmetrical, development dorsally. In order to map the complete developmental pattern, we labeled cell clusters of the JO with α-HRP and then optically reconstructed these in 3D (Fig. 5). A transverse optical slice through the pedicel at 41% of embryogenesis shows three cell clusters with dendrites extending towards the ventral epithelial surface, and a mirror symmetrical group of three cell clusters with dendrites extending towards the dorsal epithelial surface (Fig. 5a). At 55% of embryogenesis (Fig. 5b), the number of ventral cell clusters has increased to six (c.f. Figure 4d), and a transverse view shows an equal number of dorsal clusters so that the complete JO is now organized circumferentially in the pedicellar epithelium. A 3D confocal reconstruction from a further preparation at 55% of embryogenesis (Fig. 5c) shows that scolopales from all six ventral and six dorsal cell clusters in the pedicel extend in parallel for over 50µm towards the flagellum (c.f. Wolfrum 1990). We were able to visualize the scolopale insertion points in cap cells via Nomarski optics following α-HRP/PO labeling and DAB counterstaining (Fig. 5d).
The developing pattern of cell clusters, and the number of neurons contributing to each cluster of Johnston´s organ up to mid-embryogenesis, are summarized graphically in Fig. 6. Our data reveal that for both ventral and dorsal pedicellar domains: (a) cluster numbers increase in a stepwise manner each 5% of development; (b) cluster numbers are remarkably constant across all preparations, and between antennae, at each given age; (c) the spatial distribution of clusters within a domain changes in a binary fashion with age (Fig. 6, i);
The number of HRP-positive cells/cluster (Fig. 6) also increases with age up to 41% of embryogenesis at which it saturates. The very small variance in neuronal number per cluster between preparations at a given age may be a function of our method. We staged preparations to the nearest 1% (5 hours) of developmental time so that some embryos could have been slightly advanced or retarded temporally compared to others. There was also no significant difference in neuronal numbers comprising the ventral and dorsal cell clusters for a given age (data not shown). It should be emphasized that our data for cell numbers here are based exclusively on HRP-positive, differentiated, neurons. Double-immunolabeling against neuron-specific HRP and the sensory cell marker Lazarillo, however, reveals the presence of one or more extrinsic cells associated with each cluster (Suppl. Figure 2). These extrinsic cells are exclusively HRP-negative/Laz-positive over the age spectrum we tested and their identity must be established in a later study.