ABSTRACT
The development of the sensory neuron pattern in the antennal disc of Drosophila melanogaster was studied with a neuron-specific monoclonal antibody (22C10). In the wild type, the earliest neurons become visible 3h after pupariation, much later than in other imaginai discs. They lie in the center of the disc and correspond to the neurons of the adult aristal sensillum. Their axons join the larval antennal nerve and seem to establish the first connection towards the brain. Later on, three clusters of neurons appear in the periphery of the disc. Two of them most likely give rise to the Johnston’s organ in the second antennal segment. Neurons of the olfactory third antennal segment are formed only after eversion of the antennal disc (clusters tl-t3). The adult pattern of antennal neurons is established at about 27 % of metamorphosis.
In the mutant lozenge3(Iz3), which lacks basiconic antennal sensilla, cluster t3 fails to develop. This indicates that, in the wild type, a homogeneous group of basiconic sensilla is formed by cluster t3. The possible role of the lozenge gene in sensillar determination is discussed.
The homoeotic mutant spineless-aristapedia (ssa) transforms the arista into a leg-like tarsus. Unlike leg discs, neurons are missing in the larval antennal disc of ssa. However, the first neurons differentiate earlier than in normal antennal discs. Despite these changes, the pattern of afferents in the ectopic tarsus appears leg specific, whereas in the non-transformed antennal segments a normal antennal pattern is formed. This suggests that neither larval leg neurons nor early aristal neurons are essential for the outgrowth of subsequent afferents.
Introduction
Drosophila melanogaster has proven to be a powerful subject for studying neurogenesis at cellular, genetic and molecular levels. Progress has mainly derived from the development of the embryonic central and peripheral nervous system (Campos-Ortega and Knust, 1990; Ghysen and Dambly-Chaudière, 1989; Jan and Jan, 1990), and from cellular differentiation in the eye disc (Baneijee and Zipursky, 1990). These studies have shown that some properties like cell-cell communication are essential in each of these systems, while others (e.g. cell lineage) appear to be crucial only in particular systems. Similarly, the analysis of neurogenesis in insect appendages suggests that the significance of certain mechanisms may often be very different in different systems. For example, in embryonic grasshopper appendages, pioneer neurons appear to be essential for pathfinding of afferents (Klose and Bentley, 1989), whereas in wing discs of Drosophila this is certainly not the case (Blair et al. 1987). Obviously, one has to be very cautious when inferring from one system to another, not only with respect to neurogenesis but also to neurobiology in general.
It is therefore important to analyze many kinds of neural subsystems. We have been focussing on the antennal system of Drosophila, because we believe that it differs in many developmental, functional and structural aspects from the other prominent adult sensory system, the visual system. Various aspects of the adult antennal system of Drosophila have been studied, including (1) the sensillar pattern on the antennae (Stocker et al. 1983; Venkatesh and Singh, 1984; Stocker and Gendre, 1988; Foelix et al. 1989), (2) the sensory projections within the antennal lobes (Stocker et al. 1983), (3) the subdivision of the antennal lobes into distinct units, the glomeruli (Stocker et al. 1983, 1990; Pinto et al. 1988), and (4) the types of interneurons connecting the antennal lobes with higher brain centers (Stocker et al. 1990). In addition, mutants affecting olfactory function have been isolated (Siddiqi, 1987; Helfand and Carlson, 1989; Lilly and Carlson, 1989; McKenna et al. 1989; Woodard et al. 1989; Ayyub et al. 1990). Mapping of odour-induced activity suggested that processing of olfactory information relies on functional compartments within the antennal lobes (Rodrigues and Buchner, 1984; Rodrigues, 1988).
Thus, although anatomical, functional and genetic data have been collected, few attempts have been made to study neurogenesis in the Drosophila antenna (Jan et al. 1985), probably because antennal discs are less accessible during metamorphosis than other imaginai discs. Yet, a number of studies have addressed various aspects of antennal development, including pattern formation, growth and determination (Postlethwait and Schneiderman, 1971a; Haynie and Bryant, 1986), homoeotic transformation (Postlethwait and Schneiderman, 19716; Postlethwait and Girton, 1974; Morata and Lawrence, 1979; Struhl, 1982), and the morphogenesis of the discs in tissue culture (Schneider, 1964, 1966; Milner and Haynie, 1979; Milner et al. 1984).
In the present study, we first describe the development of the neuron pattern in the antennal disc of the wild type using a monoclonal antibody that binds to sensory neurons in general. We observe that the first neurons appear in the disc well after the onset of metamorphosis. They are located at the tip of the prospective antenna suggesting that - as in other insect appendages - they might represent ‘pioneer neurons’ (Bate, 1976). In order to test whether these early neurons are essential for the outgrowth of antennal afferents, we compare the wild-type pattern with the pattern in the homoeotic mutant spineless-aristapedia (ssa). ssa transforms the arista into a tarsus but leaves the other antennal segments unaffected (Morata and Lawrence, 1979). By structural and temporal criteria, the first neurons that appear in the transformed disc are distinct from the first neurons in normal antennal discs. Nevertheless, the pattern of afferents in the entire appendage appears to be normal, i.e., corresponding to the external phenotype. Hence, a crucial ‘pioneering’ role of early antennal neurons can be rejected. Finally, we analyze sensory pattern formation in the lozenge3(lz3) mutant, which lacks basiconic antennal sensilla (Stocker and Gendre, 1988). By comparing the development of wild-type and lz3 antennal discs, we tried to identify the precursors of basiconic sensilla in the wild type and to understand whether this mutant affects the differentiation of basiconic neurons or another step of sensillum formation.
Materials and methods
Pupae of the wild-type strain Sevelen and of the mutants lozenge3(lz3/ C(l)) and spineless-aristapedia (ssa / In(3LR)P88,ss−) were used. Larvae were raised at 25 °C on standard cornmeal medium. Wandering third instar larvae were collected from the wall of the stock bottles. For the timing of the metamorphic stages, white prepupae were removed from the bottles and kept at 25 °C in Petri dishes on moist filter paper up to the desired age. The stages given in this paper are indicated as hours after puparium formation (h AP) at 25 °C. Due to the 20 min duration of the white prepupal stage and to developmental variability, the accuracy of staging is about 1 h.
Immunocytochemistry was performed by applying the monoclonal antibody 22C10 (kindly provided by Dr S. Benzer, Caltech, Pasadena) to cryosections of pupae and to whole mounts of antennal discs. For both procedures, phosphate-buffered 4 % paraformaldehyde (PF) was used for fixation, and antibody staining was performed according to Buchner et al. (1986).
For cryosectioning, the posterior tip of the pupae was removed with iridectomy scissors. Pupae were fixed for 4h in PF and immersed overnight in 25 % glucose (in Drosophila Ringer’s) for cryoprotection. After freezing in liquid nitrogen, 10gm thick sections were cut on a cryomicrotome. Whole mounts of antennal discs were prepared as follows: After removal of the puparium, pupae were dissected in Trisbuffered saline with 0.2% Triton X-100. The dissecting procedure was adapted according to the developmental stage. From 0 to 6h AP the mouth hooks were pulled anteriorly out of the prepupae as the eye-antennal discs and the CNS remained attached to them. Between 6 to 10 h AP, the CNS with the discs attached was isolated by gently pushing the whole complex through a dorsal, longitudinal slit made in the anterior half of the pupa. At 10 h, at the onset of head eversion, gentle pressure was applied on the thorax that caused the pupal head to protrude. Since the eye-antennal discs adhere to the pupal cuticle, they could be isolated by dissecting the corresponding cuticular area. The discs were removed from the cuticle after fixation. From 12h AP onward, i.e., after head eversion, the antennal discs were dissected by splitting the head sagittally. CNS and fat droplets were washed out with a pipette. The two halves were fixed in PF and, following immunocytochemistry, the antennal discs could be identified. Whole mounts of the discs were embedded in Faure’s solution.
Results
Morphogenetic movements of the eye and antennal discs during metamorphosis
In wandering third instar larvae, the eye discs cover the anterior surface of the brain hemispheres (Fig. 1A,B). The antennal discs lie in front of the eye discs, closely apposed to a marked dorsal bend of the oesophagus. They consist of several concentric folds (Fig. 1A), the centralmost of which will form the arista, whereas the peripheral folds develop into the more proximal segments of the antenna (Vogt, 1946a). The prominent Bolwig’s nerve, originating from the larval photoreceptor, passes over the surface of the antennal and eye discs (Fig. 1A) (Bolwig, 1946; Zipursky et al. 1984; Steller et al. 1987). Two additional supraoesophageal nerves run close to the antennal disc (Fig. 1B,C), i.e., the larval antennal nerve carrying sensory axons from the antenno-maxillary complex and the labral nerve consisting mainly of pharyngeal afferents (Campos-Ortega and Hartenstein, 1985).
The neuronal elements of the wild-type antennal disc during early metamorphosis as displayed by mAb 22C10. Anterior is to the left. (A) White prepupa (Oh AP; whole mount). Behind the morphogenetic furrow of the eye disc (ed) (arrow), sensillar differentiation is already well advanced, whereas in the concentric antennal disc (ad) no neurons are stained. (B) 3h AP (whole mount). The eye discs are attached to the CNS by the optic stalks (arrowheads). In the center of the antennal disc, the first neuronal cell bodies are visible (arrow). (C) Higher magnification of B showing early neurons (arrowheads). (D) 4h AP. Horizontal cryosection through right antennal disc. Axons (ax) of two early neurons (arrow) appear to grow towards the larval antennal nerve (lan). bl, brain lobe; bn, Bolwig’s nerve; In, labral nerve. Bar, 50 μm.
The neuronal elements of the wild-type antennal disc during early metamorphosis as displayed by mAb 22C10. Anterior is to the left. (A) White prepupa (Oh AP; whole mount). Behind the morphogenetic furrow of the eye disc (ed) (arrow), sensillar differentiation is already well advanced, whereas in the concentric antennal disc (ad) no neurons are stained. (B) 3h AP (whole mount). The eye discs are attached to the CNS by the optic stalks (arrowheads). In the center of the antennal disc, the first neuronal cell bodies are visible (arrow). (C) Higher magnification of B showing early neurons (arrowheads). (D) 4h AP. Horizontal cryosection through right antennal disc. Axons (ax) of two early neurons (arrow) appear to grow towards the larval antennal nerve (lan). bl, brain lobe; bn, Bolwig’s nerve; In, labral nerve. Bar, 50 μm.
During metamorphosis this organization changes dramatically, since the eye and antennal discs undergo complex morphogenetic movements (Fig. 2). In the eye disc, the cell bodies of the retinular neurons are initially located on the ‘outer’ surface of the disc and their axons point towards the brain hemispheres (Figs 1D, 2). Between 4 and 10 h AP the posterior edge of the eye disc detaches from the underlying brain and migrates forward over the antennal disc, whereas the anterior edge of the eye disc remains in place (Fig. 2). Consequently, the antennal disc becomes covered laterally by the eye disc. The retinular cell bodies are now situated on the ‘inner’ disc surface opposite the antennal disc. The three larval nerves have essentially disappeared and the adult antennal nerve, which is formed by the fasciculation of axons from the newborn antennal neurons (see below), becomes visible.
Morphogenesis of eye and antennal discs. At 10 h AP, the prospective eye surface becomes transiently apposed towards the antennal discs. As indicated schematically, this leads to an inverted orientation of visual receptor neurons. Later on the eye discs fold back to their adult-specific location. During head eversion, the antennal discs move far ventrally to the surface of the head (15 h AP) and then upward and forward into a preformed pair of cuticular pockets (arrows).
Morphogenesis of eye and antennal discs. At 10 h AP, the prospective eye surface becomes transiently apposed towards the antennal discs. As indicated schematically, this leads to an inverted orientation of visual receptor neurons. Later on the eye discs fold back to their adult-specific location. During head eversion, the antennal discs move far ventrally to the surface of the head (15 h AP) and then upward and forward into a preformed pair of cuticular pockets (arrows).
At 10h AP, the eversion of the head begins, i.e., the discs evaginate (see below), enlarge and migrate towards their adult-specific location (Fig. 2). By squeezing the thorax with forceps, head eversion can be mimicked, which allows the complex movements involved to be followed. The eye discs fold back over the brain hemispheres, restoring the initial spatial orientation (Fig. 2: 15 h AP). The antennal discs detach from the eye discs and move from their position in the interior of the pupa towards the ventro-lateral surface of the head (Fig. 2: 15 h AP). Then they migrate in an antero-dorsal direction and arrive between 18 and 24 h AP in a preformed pair of cuticular pockets that occupy the site of the adult antennae (Fig. 2). During this movement, the tip of the prospective arista points laterally, due to its attachment to the ventrolateral puparial membrane. The extensive forward migration of the antennal anlage is accompanied by a dramatic elongation of the antennal nerve.
The sensory neuron pattern in the developing antennal disc of the wild type
The mAb 22C10 recognizes epitopes on peripheral neurons in general (Zipursky et al. 1984). Using this antibody, very early stages of neuronal differentiation can be visualized, i.e., cell bodies with a dendrite and a short axonal outgrowth. The overall staining patterns of antennal discs of the same ages are highly invariant. Differences in the dendritic staining properties allowed us to follow the fate of distinct groups of neurons up to the mature stage. An estimate of the age of individual neurons is given by the length of their axonal processes. As a control for successful staining, we used the well-known differentiation pattern of sensory neurons in the eye disc (Fig. 1A) (Zipursky et al. 1984). Table 1 summarizes the developmental periods of the various neuron clusters to be described below.
Neuron pattern before disc eversion
In the antennal disc of white prepupae, no neurons are demonstrated by mAb 22C10, although in the eye disc neuronal differentiation is already well advanced (Fig. 1A). It is only at 3h AP when the first 2 to 4 neurons appear near the center of the antennal disc (Fig. 1B,C). Judging from their location in the prospective distalmost part of the antenna, it is likely that these neurons will become part of the aristal sense organ (Foelix et al. 1989). Their axons seem to grow towards the larval antennal nerve (Fig. ID) and thus to establish the first connection towards the brain. Until 6h AP, a heavily stained and a lightly stained cluster of about 10 neurons each appear at a more peripheral annular location of the disc (Fig. 3).
Whole mount of wild-type antennal disc at 6h AP (stippled) showing the prospective aristal neurons (arrow), as well as a heavily (h) and a lightly (1) stained cluster of neurons, ed, eye disc. Anterior is to the left. Bar, 50 μm.
At 8 h AP, the early formed neurons become located within a conspicuous, slightly acentric protrusion of the disc, which can be followed up to the arista (Fig. 4B). Most of the other neurons are arranged within three distinct clusters (j1, j2, je) (Fig. 4A,B). Each cluster forms its individual nerve trunk towards the center of the disc. There they join the pupal antennal nerve that has apparently been established by the future aristal fibers. The dendrites within clusters j1 and j2 are opaque, which is typical of dendrites in the mature Johnston’s organ in the second antennal segment. One of these clusters is likely to stem from the heavily stained cluster at 6h AP, while the other probably arises later. Cluster je, possibly derived from the lightly stained cluster at 6h AP, consists of both lightly and heavily stained neurons. In between the major clusters, scattered, lightly stained neurons appear, which may become neurons of external sensilla on the second (or possibly third) segment. In the periphery of the antennal disc, three large neurons become visible (Fig. 4A,B: 1-3). Their locations remain highly invariant during subsequent development. Judged by axon formation, neuron 2 develops in advance to neurons 1 and 3.
Wild-type antennal disc before (A,B) and after (C,D) évagination. Whole-mount preparation (A) and corresponding camera-lucida drawing (B) of 8 h disc. Anterior is to the left. Heavily staining clusters jl and j2 consist of putative neurons of the Johnston’s organ in the 2nd antennal segment. Ouster je bears a mixed population of heavily and lightly stained neurons. Three large neurons (1-3) are located at the extreme periphery of the disc. Aristal neurons (ar) lie in a protrusion (the prospective arista) outside the focal plane of A. (C) 14 h AP (whole mount). After disc eversion, antennal segments are easily distinguishable. In addition to the precursors of the Johnston’s organ (jo) on the 2nd segment (s), neuron clusters tl-t3 have emerged on the 3rd segment (t). The arista with the neurons of the aristal sense organ (ar) protrudes from the 3rd segment. For a camera-lucida drawing of the same preparation, see Fig. 7A. (D) At 26h AP (whole mount), the anatomy resembles that of an adult antenna and the neuron pattern is comparable to the adult distribution of sensilla. Three nerves extend from the 3rd (t) to the 2nd (s) segment, an, antennal nerve, ed, eye disc. Bar, 50 μ m.
Wild-type antennal disc before (A,B) and after (C,D) évagination. Whole-mount preparation (A) and corresponding camera-lucida drawing (B) of 8 h disc. Anterior is to the left. Heavily staining clusters jl and j2 consist of putative neurons of the Johnston’s organ in the 2nd antennal segment. Ouster je bears a mixed population of heavily and lightly stained neurons. Three large neurons (1-3) are located at the extreme periphery of the disc. Aristal neurons (ar) lie in a protrusion (the prospective arista) outside the focal plane of A. (C) 14 h AP (whole mount). After disc eversion, antennal segments are easily distinguishable. In addition to the precursors of the Johnston’s organ (jo) on the 2nd segment (s), neuron clusters tl-t3 have emerged on the 3rd segment (t). The arista with the neurons of the aristal sense organ (ar) protrudes from the 3rd segment. For a camera-lucida drawing of the same preparation, see Fig. 7A. (D) At 26h AP (whole mount), the anatomy resembles that of an adult antenna and the neuron pattern is comparable to the adult distribution of sensilla. Three nerves extend from the 3rd (t) to the 2nd (s) segment, an, antennal nerve, ed, eye disc. Bar, 50 μ m.
Neuron pattern after disc eversion
Eversion of the antennal disc occurs between 10 and 16 h AP. At 14 h AP, the third segment (funiculus) becomes distinct from the second (Figs 4C, 7A). Within the elongated arista, four to six neurons of the aristal sense organ are present. Three clusters of five to eight neurons each t1-t3) emerge on the proximal part of the funiculus. Thus, neuronal differentiation in this segment is delayed by about 6h with respect to the second segment. Four nerves extend from the third segment to the second: one of them comes from the aristal sensillum, the others from the clusters t1-t3. The organization of the heavily labelled Johnston’s organ in the second segment is already well visible. A fourth large neuron (Fig. 7A: 4) becomes manifest at the extreme periphery of the disc. Since the delineation between first and second segments is not distinct, thé adult identity of the neurons 1 – 4 remains unclear. Judging by their large size, we think that they might become associated with large bristles on the first segment.
At 19 h AP, the number of neurons in each cluster has increased. In addition, newly bom neurons appear on the funiculus in between clusters t1-t3. 22 h AP the growth of antennal afferents towards the CNS is reflected by a considerable increase in the diameter of the antennal nerve. At about 26 h AP, which approximates 27 % of adult development, the adult distribution of neurons of the Johnston’s organ is established (Fig. 4D). Similarly, the pattern of neurons on the third segment corresponds to the adult-specific sensillum pattern. In the adult, a large subtype of basiconic sensilla occurs as a pure cluster on the medial edge of the funiculus, whereas a small subtype is interspersed with coeloconic and trichoid sensilla in the other regions of the third segment (Stocker et al. 1983). When using the arista anlage as a landmark, it is very likely that cluster t3 (Figs 4C, 7A) will form the pure cluster of large basiconic sensilla. The nerve that originates from cluster t3 apparently joins the aristal nerve.
Antennal neurogenesis in lozenge3
The mutant lozenge is known for a variety of pattern abnormalities that affect eyes, antennae, tarsal claws and gonads. In the allele lz3, basiconic sensilla of the funiculus including all of their cellular elements are lacking (Stocker and Gendre, 1988). Moreover, the numbers of the other two types of sensilla on the funiculus are modified. There are fewer trichoid and more coeloconic sensilla despite a reduced size of the antennae. Studying the development of such a mutant may help to understand the neuronal pattern formation in the wild-type antenna.
The neuronal pattern on the lz3 antennal disc is initially similar to the wild type. This refers, e.g., to the appearance of the presumptive aristal neurons at 3h AP, or of the three clusters j1, j2 and je at 10h AP (Fig. 5A,B). The onset of disc eversion (though not its termination) is delayed by approx. 2h. Whether this time-lag is an effect of the mutation or due to interstrain variability remains unknown. A significant deviation from the wild-type pattern occurs only after eversion. Whereas clusters t1 and t2 show a normal pattern, no neurons can be seen at the t3 site (Fig. 5C; cf. Fig. 7A with B). Since the (3 disc region of the wild type will form the medial edge of the funiculus (bearing exclusively large basiconic sensilla; see above), this observation provides additional evidence that cluster t3 consists of neurons of prospective large basiconic sensilla. Furthermore, the nerve originating in the lz3 arista does not fuse with one of the two remaining funicular nerves, in agreement with the fact that aristal axons and t3 axons form a common nerve in the wild type. In summary, our results demonstrate that, in the antennae of lz3, large basiconic neurons are never bom and that the pure basiconic region of the wild type originates from a homogeneous neuron cluster.
Neurogenesis on the lz3 antenna. Whole-mount preparation (A) and corresponding camera-lucida drawing (B) of 10h disc. Clusters j1, j2 and je are present (cf. Fig. 4A,B). From the large peripheral neurons 1-3 only 2 is in focus. (C) At 16h AP, after eversion, clusters t1 and t2 have emerged, whereas cluster t3 (cf. Fig. 4C) is lacking (arrow). For a camera-lucida drawing of the same preparation, see Fig. 7B. an, antennal nerve; ar, arista with aristal sense organ; ed, eye disc; jo, Johnston’s organ. Anterior is to the left. Bar, 50 μ m.
Neurogenesis on the lz3 antenna. Whole-mount preparation (A) and corresponding camera-lucida drawing (B) of 10h disc. Clusters j1, j2 and je are present (cf. Fig. 4A,B). From the large peripheral neurons 1-3 only 2 is in focus. (C) At 16h AP, after eversion, clusters t1 and t2 have emerged, whereas cluster t3 (cf. Fig. 4C) is lacking (arrow). For a camera-lucida drawing of the same preparation, see Fig. 7B. an, antennal nerve; ar, arista with aristal sense organ; ed, eye disc; jo, Johnston’s organ. Anterior is to the left. Bar, 50 μ m.
Antennal neurogenesis in spineless-aristapedia
In the homoeotic mutant ssa, the distal region of the antennal disc, the arista, becomes transformed into the corresponding leg region, the tarsus (Morata and Lawrence, 1979). The ssa gene appears to be required only in the distal portion of the appendage (Struhl, 1982). Its lack of activity leads to an aberrant disc phenotype at about 1/3 of the third larval instar (Vogt, 1946b). Since the first neurons that appear in the wildtype disc are aristal neurons, a distally transformed antenna might help to elucidate the putative guiding capability of these cells.
Whereas normal leg discs carry a whole set of larval neurons (Jan et al. 1985; Tix et al. 1989), this is not the case for homoeotically transformed ss° antennal discs. However, the first 2-4 neurons appear already at the white prepupal stage (Fig. 6A), i.e., earlier than in the normal antennal disc (Table 1). They are located in the center of the disc, the region that will develop into a tarsus. At 5h AP the protrusion of the future tarsus is already very obvious, and - unlike normal antennal discs - two parallel axonal processes extend from the distal tip towards the antennal part of the disc (Fig. 6B).
Neurogenesis on the ssa antennal disc (whole mounts). (A) White prepupa (Oh AP). In contrast to the wild type, the first neurons (arrow) are already present in the antennal disc (ad). (B) At 5h AP, two parallel axons (arrow) are stained within the tarsal outgrowth of the disc. (C) At 15 h AP, during the slightly retarded eversion of the disc, axons originating from the tarsal neurons are labeled (arrow) in addition to the three normal clusters j1, j2 and je. (D) 17h AP. The two parallel nerves in the ectopic tarsus (ta) are characteristic of a normal tarsal pattern. In the 3rd segment of the everted antennal disc, the clusters t1-t3 are visible as in a normal antenna. For a camera-lucida drawing of the same preparation, see Fig. 7C. ed, eye disc; jo, Johnston’s organ; 1 – 3, large peripheral neurons. Anterior is to the left. Bar, 50 μ m.
Neurogenesis on the ssa antennal disc (whole mounts). (A) White prepupa (Oh AP). In contrast to the wild type, the first neurons (arrow) are already present in the antennal disc (ad). (B) At 5h AP, two parallel axons (arrow) are stained within the tarsal outgrowth of the disc. (C) At 15 h AP, during the slightly retarded eversion of the disc, axons originating from the tarsal neurons are labeled (arrow) in addition to the three normal clusters j1, j2 and je. (D) 17h AP. The two parallel nerves in the ectopic tarsus (ta) are characteristic of a normal tarsal pattern. In the 3rd segment of the everted antennal disc, the clusters t1-t3 are visible as in a normal antenna. For a camera-lucida drawing of the same preparation, see Fig. 7C. ed, eye disc; jo, Johnston’s organ; 1 – 3, large peripheral neurons. Anterior is to the left. Bar, 50 μ m.
Camera-lucida drawings of everted antennal discs corresponding to micrographs in Figs 4C, 5C and 6D. The location of the 2nd antennal segment (s) with the Johnston’s organ is shaded. (A) Wild type, 14 h AP. (B) lz3, 16 h AP. Note the absence of cluster t3 (arrow). (C) ssa, 17 h AP. The ectopic tarsus (ta) shows two parallel nerves like in normal tarsi (arrows), ar, aristal neurons; t, 3rd antennal segment; t1 – 13, clusters of neurons on the 3rd segment; 1 – 4, large neurons at highly invariant peripheral locations.
Camera-lucida drawings of everted antennal discs corresponding to micrographs in Figs 4C, 5C and 6D. The location of the 2nd antennal segment (s) with the Johnston’s organ is shaded. (A) Wild type, 14 h AP. (B) lz3, 16 h AP. Note the absence of cluster t3 (arrow). (C) ssa, 17 h AP. The ectopic tarsus (ta) shows two parallel nerves like in normal tarsi (arrows), ar, aristal neurons; t, 3rd antennal segment; t1 – 13, clusters of neurons on the 3rd segment; 1 – 4, large neurons at highly invariant peripheral locations.
Until 10 h AP, 2 – 4 additional neurons develop at the distal end of the tarsus. Judging from their numbers and their pattern of afferents, these neurons do not correspond to the aristal neurons of the normal antennal disc. In contrast, the non-transformed portion of the disc shows a wild-type pattern of labeled neuron clusters and afferent nerves.
Unlike the wild type, a cuticular, bag-like protrusion is formed around the homoeotic tarsus as soon as it has reached the surface of the ventro-lateral head region. This structure does not correspond to the normal antennal pockets which are situated more anteriorly and dorsally, and which are also present in ssa. Since the ectopic tarsal neurons stain relatively weakly at this stage, it is possible that the bag structure blocks to some extent the diffusion of the antibody. Thus, our view of the neuronal pattern in the tarsus between 5 and 15 h AP may be slightly incomplete.
At 15 h AP, the mutant antennal disc is in the process of eversion, i.e., with a delay of about 3 h (Fig. 6C). As in normal development, the periphery of the disc is occupied by the clusters j1,j2 and je, while in its center, at a somewhat lower plane, the first neurons of the third segment are labeled. At about 17 h AP eversion has completed (Fig. 6D). As the disc is migrating towards its adult location along the surface of the head, the homoeotic tarsus becomes pulled out of the cuticular bag. This process apparently facilitates the labeling of tarsal neurons. Boundaries between tarsomeres are represented by sets of neuronal cell bodies whose axons join the two parallel tarsal nerves. In the nontransformed portion of the disc, clusters t1-t3 appear on the funiculus (Figs 6D, 7C), although they are less distinct than in the wild type, because of spatial changes caused by the ectopic tarsus. At least four nerves are seen to extend from the third to the second antennal segment. Three of them arise from the three clusters t1-t3, whereas one comes from the ectopic tarsus. The second nerve of the ectopic tarsus could not be followed upon entering the funiculus.
Around 23 h AP the ectopic tarsus reaches its adult length. Additional neurons differentiate whose axons join the two major tarsal nerves. One of the two ectopic tarsal nerves runs straight through the funiculus and enters the second segment at a similar location to the normal aristal nerve. The neuronal pattern of the nontransformed segments is similar to that of the normal antenna at this stage.
Discussion
The specific binding properties of mAb 22C10 to peripheral neurons (Zipursky et al. 1984) allowed us to follow the differentiation of antennal neurons during the crucial early phase of metamorphosis. The location of three neuron clusters each on the second and third antennal segments, and of four characteristic neurons at the periphery of the disc was found to be highly invariant. Other neurons that are scattered in between these clusters, or individual neurons within the clusters, show considerable positional variability. Hence, although the fate of neuron clusters can be determined, neurons may not be followed individually during metamorphosis (except for the four cell bodies at the periphery of the disc). This is consistent with the observation that individual sensilla on the adult funiculus cannot be identified by position (data not shown).
A first wave of neuronal differentiation occurs in the antennal disc immediately after the onset of metamorphosis (Table 1). During disc eversion, in particular between 10 and 15 h AP, only few additional neurons become labeled. Thereafter, differentiation is resumed, and a neuron pattern corresponding to the adult sensillum pattern becomes established by 26 h AP (i.e. about 27 % of metamorphosis). These findings parallel cell proliferation dynamics in the antennal disc (Post-lethwait and Schneiderman, 1971a), which is characterized by a decrease of cell division from 2 h to 8 h AP and a second wave of proliferation that reaches its maximum 14 h AP and terminates around 18 h AP (cf. Table 1). Thus, all cellular elements of the sensilla must have been born before 26 h AP, our latest stage studied.
Morphogenesis of the eye and antennal discs
During metamorphosis, the anterior part of the pupa undergoes complex morphogenetic movements, termed head eversion, which include migration, growth and évagination of the imaginai discs involved (Fig. 2). Since it is possible to imitate eversion to a certain degree by applying pressure on the thorax, we believe that normal head eversion may be triggered by intrinsic pressure generated in the pupal thorax or abdomen, i.e., by ‘inflating’ the head part of the pupa.
After pupariation, the eye discs transiently withdraw from the brain hemispheres and undergo an intriguing inversion of their anatomy in which the prospective surface of the eye becomes apposed to the antennal discs (Fig. 2: 10 h AP). It is only during head eversion that the eye discs fold back and the adult-like anatomy becomes established (Fig. 2: 15 h AP). A forward movement of the lateral disc margins has been reported from cultured eye-antennal discs attached to the cephalic ganglia (Schneider, 1964), although the transiently inverted stage of the disc remained unnoticed. The significance and the mechanism of these processes are not yet understood.
During head eversion, the antennal discs, too, are subject to complex movements. Antennal discs initially reach the surface of the head far ventro-laterally (Fig. 2: 15 h AP), at a position which may reflect the ‘ventral’ quality of antennae (Postlethwait and Schneiderman, 19716). Their adult location is attained only after a second phase of forward and upward migration. The antenno-cerebral connection is established before the onset of these movements.
The cuticular pockets formed by the pupal epidermis, which will contain the antennae after their forward migration, develop long before the antennal discs arrive. Hence, their formation is not caused by the discs. In contrast, the cuticular bags that ensheath the ectopic tarsus in the ssa antenna appear to be induced by the transformed appendage, because no such structures occur at this site in the wild type. Since similar sheaths have been found to envelop various types of surgically generated ectopic appendages (Schmid et al. 1986), we suggest that any part of the pupal epidermis may be capable of producing such structures.
Pathfinding of antennal afferents
In confirmation of recent reports, our data demonstrate that larval antennal discs lack neurons (Jan et al. 1985; Tix et al. 1989). However, in contrast to Jan et al. (1985), who described the emergence of the first neuron clusters at 6h AP at the periphery of the disc, we observe an early group of two to four neurons already 3 h AP. These neurons lie in the center of the disc and can be followed through successive stages to become part of the adult aristal sensillum. They are obviously the first neurons to extend their axons. These fibers seem to grow towards the larval antennal nerve, which they follow to the brain. They are apparently able to navigate through the disc tissue and may therefore provide a pioneer function for the later developing afferents.
The idea of specialized neurons that pioneer the establishment of connections in insect appendages has come from many observations in both hemi- and holometabolous insects (Bate, 1976; Ho and Goodman, 1982; Jan et al. 1985; Caudy and Bentley, 1986). The physiological conditions under which these neurons differentiate have been analyzed in detail (Bentley and Caudy, 1983a; Lefcort and Bentley, 1987). Recent experimental data suggest that in embryonic limb buds of grasshoppers, pioneer neurons are essential in guiding subsequent afferents, i.e., in their absence, the afferents fail to reach the CNS and to establish the nerve pattern characteristic of that appendage (Bentley and Caudy, 19836; Klose and Bentley, 1989). In order to understand whether the early aristal neurons in the antennal disc serve a similar crucial role, we studied neurogenesis in the homoeotically transformed antenna of ss°. Unlike the normal antennal disc, the first neurons emerge already at pupariation. Moreover, from 5h AP onward two pairs of parallel nerves develop in the ectopic tarsus, and at 17 h AP boundaries between tarsomeres become labeled by sets of neuronal cell bodies. Both structural patterns are peculiar to leg discs and do not occur in normal antennal discs or in any other type of imaginai disc (Jan et al. 1985; Tix et al. 1989; Lakes and Pollack, 1990). In spite of this likely distal transformation of the sensory neurons, the pattern of peripheral nerves in the non-transformed proximal segments is similar to that in a normal antenna (Fig. 7C). Also, afferent connections towards the brain are apparently normal. Hence, a crucial pioneering function of early aristal neurons can be ruled out.
The pattern of neurogenesis in the ssa antennal disc allows also study of possible pioneer functions in the normal leg. Leg discs of third larval instars possess a small number of neurons that have established central projections already during embryogenesis (Jan et al. 1985; Tix et al. 1989; Lakes and Pollack, 1990). These connections will apparently be used as afferent pathways during metamorphosis. Unlike normal leg discs, the first neurons in the ss° antennal disc emerge only at pupariation, as mentioned above. Thus, despite the lack of larval neurons, adult axons successfully navigate to the base of the transformed segments and establish a nerve pattern characteristic of the normal tarsal pattern. Analogous observations have been made in homoeotically transformed antennal discs of Antennapedia (Tix et al. 1989), and in supernumerary legs induced by the mutant l(l)ts726 (Lienhard and Stocker, 1987). Since the formation of duplicated legs in this mutant starts in the third larval instar, long after early neurons have been formed, it is likely that some of the supernumerary leg anlagen lack neurons. Yet, the afferents from such duplications are always able to find the CNS and even to mediate normal reflex responses.
These data suggest that neither aristal neurons nor larval leg neurons are essential in guiding sensory axons towards the disc base and to establish the pattern of afferents characteristic of that appendage. How normal patterns of afferents are formed in these appendages, may be explained by several mechanisms. (1) Early neurons could be replaced by any other neurons that develop at similar locations. Hence, in the ectopic tarsus of ss°, larval neurons may be substituted by the first neurons that emerge distally. (2) The generation of an antenna-like pattern in the non-transformed part of the ss” antennal disc may be based on the homology between legs and antennae (Postlethwait and Schneiderman, 1971b). According to this idea, leg axons would be able to organize an antennal nerve pattern in the non-transformed part because of their capacity to read the appropriate cues in the antennal disc. (3) Essentially every neuron of a disc might be capable of reading signals intrinsic to its own disc. Hence, in this model, none of the neurons would possess a crucial pioneering function.
Although our data do not allow us to distinguish between these explanations, the third model is supported by experimental data in the wing disc, which clearly show that axons of newly bom neurons establish the future pathways of wing nerves without the aid of pioneer neurons (Palka, 1986a,b). Transplantation of neurons from a variety of imaginai discs on genetically aneural wing discs suggested that sensory axons are guided towards the wing base in a relatively nonspecific way by polarity cues and by particular pathways localized in the wing epithelium (Blair et al. 1987). In summary, none of the data collected so far have succeeded in demonstrating the presence of pioneer neurons that play an essential role of pathfinding within Drosophila imaginai discs.
Our data suggest that the larval antennal nerve serves as a pathway towards the brain for the adult antennal fibers. The distal part of this nerve, anterior to the antennal disc, disappears during early metamorphosis, probably because the afferents from the larval antennomaxillary complex degenerate. The persistence of its proximal portion may be due to the arrival of the first antennal afferents before degeneration of the larval afferents. The fact that afferents from the transformed antenna of ssa join the antennal nerve underlines the significance of fasciculation and of its non-specificity as major principles of peripheral nerve formation, a fact that has also been demonstrated in earlier experimental work (Ghysen and Deak, 1978; Schmid et al. 1986). It is apparently more economical to collect axons into a small number of fiber bundles, regardless of their identity than to let each fiber explore its own pathway towards the base of the appendage.
The adult fate of neuron clusters
By using numbers, locations and staining properties of labeled neurons as criteria, we were able to identify the adult fate of many sensory neurons or neuron clusters on the antennal disc. As mentioned before, the early neurons in the center of the disc are likely to become the neurons of the aristal sensillum (Foelix et al. 1989). Clusters jl and j2 emerge in the periphery of the disc before its eversion and exhibit heavy dendritic staining. In the adult, similarly labeled neurons occur only in the proprioceptive Johnston’s organ of the second antennal segment. Thus, both the location in the disc and the staining pattern suggest that j1 and j2 neurons are presumptive mechanoreceptors of the Johnston’s organ. In contrast, lightly stained neurons in the je cluster may become integrated in external sensilla of the second segment. From the three additional clusters t1-t3 that differentiate on the funiculus after disc eversion, t3 is missing in the lz3 mutant from the very beginning (Fig. 7B). The adult funiculus of this mutant lacks a group of large basiconic sensilla (Stocker and Gendre, 1988) at a position corresponding to the t3 position on the disc. Thus, we conclude that cluster t3 consists of neurons of large basiconic sensilla. Consequently, coeloconic, trichoid and small basiconic sensilla of the large mixed funiculus region (Venkatesh and Singh, 1984) may be derived from clusters t1 and t2 and from later neurons emerging between these clusters. An analysis of sensillar development at higher resolution would require the isolation of monoclonal antibodies or enhancer trap lines that recognize specific types of sensory neurons.
An intriguing observation is that neurogenesis in the antennal disc begins in the prospective distal segments, jumps to the most proximal segments and ends in between. In the homologous leg disc, the first neurons develop at the tip as well, and then the temporal emergence of neurons is governed by the type of sensillum rather than by any directional sequence (Held, 1990). Since the sensillar types on second and third antennal segments are distinct from each other, the pattern of development observed might be controlled by similar rules as in leg discs. In non-homologous discs neurogenesis apparently follows different programs. For example, in the wing disc, neurons form synchronously at many sites of the wing blade (Blair and Palka, 1985), whereas in the eye disc, neuronal differentiation obeys a very strict spatiotemporal order (Zipursky et al. 1984).
Determination of antennal sensilla: the lozenge gene
The detection of genes that interact with sensillar development and the establishment of powerful techniques of cellular labeling have considerably expanded our understanding of sensory development in insects. According to a current hypothesis (Ghysen and Dambly-Chaudière, 1989; Jan and Jan, 1990), differentiation in the embryonic sensory system is initiated by ‘proneural’ genes, which make groups of cells competent to become neuronal precursors. By cell-cell communication, controlled by ‘neurogenic’ genes, the first committed precursor in a group inhibits its neighbours from becoming additional precursors. Afterwards, ‘selector’ genes determine the type of sensillum that will develop. Finally, ‘cell lineage’ genes interact with the lineage relationships within a sensory organ, i.e., with the descent of all cells of an external sensillum from a single sensory mother cell (Bodmer et al. 1989).
Our developmental data show that on the lz3 antennal disc, neurons of the basiconic cluster t3 are never formed. This demonstrates that the lozenge gene is involved in the determination of antennal sensilla rather than in neuronal differentiation, lz3 and a number of other mutations and deletions at the lz locus are characterized by the lack of basiconic antennal sensilla (Stocker and Gendre, 1988). However, this defect results in two distinct phenotypes on the funiculus. The site occupied in the wild type exclusively by basiconic sensilla is devoid of any type of sensilla in the mutant. This suggests that lozenge blocks the commitment of sensillum precursor cells in general, i.e., that it acts as a proneural gene. However, in the mixed sensilla region of lz3, the number of coeloconic sensilla is significantly enlarged (whereas the density of the third major type of sensilla, the trichoid sensilla, remains unchanged). This suggests that basiconic sensilla have been replaced by coeloconic sensilla, which favors the idea that lozenge acts as a selector gene. The paradoxical situation of two distinct phenotypes (lack of sensilla vs. transformation) may be explained by two different mechanisms. Either the lozenge gene functions as a selector gene on the whole funiculus, or it behaves as a proneural gene in the pure region, and as a selector gene in the mixed region. In both cases, its activity would be controlled by positional cues. For example, in the first case, information laid down early in development would allow exclusively the generation of basiconic sensilla in the pure basiconic region, and of both coeloconic and basiconic sensilla in the mixed region. In a somewhat similar way, ‘prepat-tem’ genes acting upstream of proneural genes are believed to divide the embryo into different domains (Jan and Jan, 1990).
Various tools such as cell-specific mAbs or enhancer trap lines should allow further elucidation of antennal neurogenesis and the role the lz gene plays therein. Moreover, the pattern of BrdU incorporation could improve our understanding of cellular proliferation dynamics (Schubiger and Palka, 1987; Bodmer et al. 1989). Finally, studying how lz is affected by prepattem, proneural and selector genes may clarify its position within the hierarchy of these genes.
ACKNOWLEDGEMENTS
We thank Mrs N. Gendre for technical assistance and Mr H. Gachoud for photographie work. We are grateful to Dr R. F. Foelix, K. Stôrtkuhl and Dr H. Tobler for critically reading this manuscript. This study was supported by the Swiss National Science Foundation (Grant No. 31-25626.88).