We have investigated the temporal pattern of appearance, cell lineage, and cytodifferentiation of selected sensory organs (sensilla) of adult Drosophila. This analysis was facilitated by the discovery that the monoclonal antibody 22C10 labels not only the neuron of the developing sensillum organ, but the accessory cells as well. The precursors of the macrochaetes and the recurved (chemosensory) bristles of the wing margin divide around and shortly after puparium formation, while those of the microchaetes and the stout and slender (mechanosensory) bristles of the wing margin divide between 9h and 18 h after puparium formation (apt). The onset of sensillum differentiation follows the terminal precursor division within a few hours. Four of the cells in an individual microchaete organ are clonally related: A single first-order precursor cell divides to produce two second-order precursors; one of these divides into the neuron and thecogen cell, the other into the trichogen cell and tormogen cell. Along the anterior wing margin, two rounds of division generate the cells of the mechanosensory sensilla; here, no strict clonal relationship seems to exist between the cells of an individual sensillum. At the time of sensillum precursor division, many other, non-sensillum-producing cells within the notum and wing proliferate as well. This mitotic activity follows a spatially non-random pattern.

The integument of most insects bears a collection of epidermally derived sensory organs (sensilla), distributed on the body surface in a characteristic pattern. Each sensillum consists of a group of highly specialized cells, including one or more sensory neurons and, depending on the insect species and the type of sensillum, two to five accessory cells (Zacharuk, 1985; Keil and Steinbrecht, 1986). The accessory cells provide sheath-forming processes around the neuron(s) and produce the cuticular apparatus that receives the external stimulus.

The ontogeny of the sensilla can be subdivided into three events: pattern formation, sensillum precursor division, and sensillum cell differentiation. Pattern formation refers to the mechanism that controls the spatial distribution of the different sensilla. This includes determining (a) which cells will give rise to sensilla and which will become epidermal cells, and (b) which type of sensillum will appear at a given site. During a brief proliferatory interval prior to the onset of differentiation, certain cells (which will be referred to as sensillum precursor cells) generate the presumptive sensillum cells in a number of mitotic divisions. In the differentiation step, presumptive sensillum cells, having been assigned their individual fates, undergo a fixed series of morphogenetic and cytodifferentiative changes which lead to their final shape and function.

We are interested in understanding the mechanisms that control these developmental steps. As part of our approach to this problem, it was necessary to obtain a more comprehensive description of normal adult sensillum development in Drosophila. In particular, the spatial and temporal pattern of sensillum precursor cell division in Drosophila had not been investigated. Lees and Waddington (1942) described the development of the trichogen cell and tormogen cell of the macrochaetes. They were unable, however, to determine the earlier history of these cells, or to describe the development of the other cells of the macrochaetes. The differentiation of sensory neurons in the wing has been studied in the light microscope, using specific antibodies (Murray et al. 1984). Sensillum differentiation has been analyzed on the ultrastructural level only for several types of larval sensilla (Hartenstein, 1988).

In this account we describe the pattern of precursor cell divisions and cytodifferentiation for the bristles of the adult notum and wing. Electron microscopy and labeling of sensillum cells by specific antibodies were combined to study the structure of the sensilla at different pupal stages. The pattern of sensillum precursor cell division was studied in a series of pulse-labeling experiments using the base analogue bromo-deoxyuri-dine (BrdU), and by applying colchicine, which blocks mitotic cells during metaphase.

Antibody labeling

Specimens of Drosophila melanogaster (Canton S) were collected as white prepupae and left for defined intervals at 25 °C, after which nota and wings were dissected in phosphate-buffered saline (PBS; 0.1M; pH 7.3) and fixed for 10min in 4% paraformaldehyde in PBS. After several washes in 0.1 M-PBS containing 0.3% Triton X-100, the preparations were incubated for Ih in O.IM-PBS containing 10% goat serum, 0.3 % Triton X-100, and the monoclonal antibody 22C10 or 21A6 and/or 8C5 (all kindly provided by Dr S. Benzer) at a dilution of 1:50. After several washes in PBS, they were then incubated for 1 h in HRP-conjugated rabbit anti-mouse IgG (Boehringer) diluted at 1:50 in 0.1 M-PBS containing 10% goat serum and 0.3% Triton X-100. Preparations were washed several times in PBS and incubated in diaminobenzidine (DAB, Sigma) at 0.1% in 0.1 M-phosphate buffer (pH 7.3) containing 0.006 % hydrogen peroxide. The reaction was interrupted after 5–10 min by diluting the substrate with 0.1 M-phosphate buffer. Preparations were dehydrated in graded ethanol (70%, 90%, 95%, 5 min each; 100%, 15 min) and acetone (5 min) and left overnight in a mixture of Epon and acetone (1:1). They were then mounted in a drop of fresh Epon and coverslipped.

Electron microscopy

For electron microscopy, staged pupae (2h, 8h, 14h, 20h, 24 h, 32 h apf) were dissected in a solution containing 2% glutaraldehyde and 2% paraformaldehyde in PBS. Dissected nota and wings were left for 2h in this solution; they were washed in several changes of 0.15 M-cacodylate buffer (pH 6.9) and postfixed for 30 min in 2% osmium tetroxide in 0.15 M-cacodylate buffer. Specimens were washed several times in PBS and dehydrated in graded ethanol and acetone (see above). Preparations were left overnight in a 1:1 mixture .of Epon and acetone and then for 5–10 h in unpolymerized Epon. They were transferred to molds, oriented, and placed at 60°C for 12 hours to permit polymerization of the Epon. Blocks were sectioned with an LKB Ultrotome III. Sections were mounted on net grids and treated with uranyl acetate and lead citrate (Reynolds, 1963). Sections were inspected and photographed with a Philips EM 300.

Application of bromo-deoxyuridine (BrdU)

The base analogue BrdU, which is incorporated into replicating DNA, was applied either by incubating dissected tissue from staged pupae (2 h intervals from 0–32 h apf) in Ringers containing BrdU (Sigma) at a concentration of 0.01 HIM, or by injecting it into pupae. For the injections, staged pupae (0–14 h apf, every 2 h; 15–24h apf, every 1 h) were mounted, ventral side down, in a row on double-stick Scotch tape. A micropipette (tip diameter 10—20 μm) was attached to a micromanipulator, filled (from the front) with approximately 0.03 id of ImM-BrdU (in PBS), and inserted into the pupa from the caudal end. The pipette was pushed forward so that its tip reached the level of the thorax. Injection was by air pressure, applied manually by a 1 ml syringe connected to the pipette with plastic tubing. Pupae were then placed in a moist chamber at 25°C for an appropriate interval, after which they were dissected in PBS.

For visualization of incorporated BrdU, the preparations were fixed for 2 min in modified Carnoy’s fixative (100% ethanol: glacial acetic acid, 6:2), washed several times in PBS, and incubated for 50 min in 2 N-HCI to denature the DNA. After this step they were washed for 30 min in several changes of PBS containing 0.3 % Triton X-100. The preparations were then incubated for 1h in 0.1 M-PBS containing 10% goat serum, 0.3% Triton X-100, and a monoclonal antibody against BrdU (Beckton-Dickinson) at a dilution of 1:5. In several experiments, the sensillum-specific MAb 22C10 was added to this solution at 1:50. After several washes in PBS, preparations were incubated for 1 h in HRP-conjugated rabbit anti-mouse IgG (Boehringer), diluted at 1:50 in 0.1 M-PBS containing 10% goat serum and 0.3% Triton X-100. The histochemical reaction and further treatment of the preparations were identical to that described above (‘Antibody labeling’).

Following injection into pupae, BrdU seems to become unavailable or be inactivated within less than 4h (see also Schubiger and Palka, 1987). Only nuclei replicating DNA during this interval (and their progeny) are labeled. Thus, for example, BrdU injected into pupae at 20 h apf is never incorporated into the nuclei of microchaete trichogen cells, which undergo endoreplication between 24 h and 26 h apf (see Results and Fig. 6). In other instances, BrdU may become unavailable even more rapidly (i.e., within Ih); see Fig. 7D,E.

We frequently observed that in preparations doubly labeled with anti-BrdU and MAb 22C10, the labeling of sensillum cell bodies by 22C10 was weak; labeling of axons by 22C10, however, remained strong.

Application of colchicine

Colchicine blocks microtubule elongation and arrests dividing cells in metaphase. It was applied either by incubating dissected tissue (at 1h intervals from 14–20h apf) in Ringers containing colchicine (Sigma) at 10 mM, or by injecting it into pupae according to the protocol described above for the BrdU injection.

For visualization of metaphase nuclei following colchicine injections, the preparations were stained with basic fuchsin.

Structure of selected notum and wing sensilla

Eight types of sensilla are found on the notum and wing of Drosophila (Miller, 1950; Bryant, 1978; Hodgkin and Bryant, 1978; Kankel et al. 1980). Five of these are external mechanoreceptors: the microchaetes and macrochaetes of the notum and proximal wing margin (tegula and costa), the stout bristles and the slender bristles of the anterior wing margin, and the campaniform sensilla on the veins and the radius of the wing (for nomenclature of the wing sensilla see Palka et al. 1979; Cole and Palka, 1982). Each of these sensilla bears an aporous shaft (bristles) or dome (campaniform sensilla). There is one class of presumably chemosensory sensilla, the recurved bristles, on the anterior wing margin; these have a uniporous shaft. Subepidermal stretch receptors, or chordotonal organs, are found in the radius of the wing. Finally, a group of sensilla trichodea occur on the tegula; it is uncertain whether these are mechano- or chemosensory.

We have examined the structure of the notum chaetae and wing margin bristles at mid-pupal stages in ultrastructural detail (Fig. 1; see also Palka et al. 1979, for ultrastructure of wing marginal bristles). The structure of all of these sensilla is fairly similar and corresponds to the ‘archetype’ of an insect external sensillum (Zacharuk, 1985; see bottom right diagram in Fig. 9). The mechanosensory sensilla are innervated by one, and the chemosensory sensilla by five, bipolar sensory neurons. These are located subepidermally and project apical dendrites which possess a wide proximal part (inner dendritic segment) and a narrow distal part (outer or ciliary dendritic segment). The dendrite(s) is enclosed in three concentric sheaths. The inner sheath, formed by the thecogen cell, is cylindrical; the middle and outer sheaths, formed by the trichogen cell and the tormogen cell, respectively, are mesaxon-like. Apically, the sheath processes are connected to each other, to the dendrite(s), and to the surrounding epidermal cells by belt desmosomes.

Fig. 1.

Electron micrographs showing ultrastructural details of developing microchaetes (A, B, D), non-innervated long hairs of the posterior wing margin (C), and multiply innervated bristles of the anterior wing margin (E-G). A and B are oblique sections of a notum at 24 h apf. A depicts the apical part of an individual microchaete, with a central thecogen cell (th) surrounded by the trichogen cell (tr) and tormogen cell (to). The tip of the dendrite (d) of the sensory neuron (out of the plane of the section) has not reached the apical surface of the thecogen cell which ensheaths it. B shows the cell bodies of the sensory neuron (ne), trichogen cell, and tormogen cell. C depicts an oblique section of a posterior wing margin at 24 h apf. The non-innervated long hairs which form two rows along the posterior wing margin are composed of two ceils, a central trichogen cell (tr) and a tormogen cell (to). D shows an individual microchaete at 32 h apf. In the plane of the section are the tormogen cell and the trichogen cell, which has started to produce a shaft (sh). Cuticle has been deposited on the apical surface of epidermal cells, tormogen cell, and shaft (arrowheads). The curved dendrite is sectioned at both a basal level and an apical level. Basally (lower part of micrograph), it is enclosed within a thin process of the thecogen cell; apically, it has completely penetrated this cell and is surrounded by the dendritic cap, an electron-dense extracellular sheath secreted by the thecogen cell. The inset at bottom right shows the cell bodies of the sensory neuron and the closely apposed thecogen cell. E-G show transverse sections, at three different levels, of the same multiply innervated anterior wing margin bristle (32 h apf). E represents the basalmost section; it shows the five dendrites (d), surrounded by the cylindrical process of the thecogen cell and the mesaxon-like processes of the trichogen cell and tormogen cell. The tormogen cell occupies only half of the perimeter; a process of what is presumably the tormogen cell of a neighboring bristle (to’) covers the other half. F and G represent progressively more apical levels. The dendrites converge apically and terminate at approximately level F; in G, only the apical extension of the thecogen cell and the enclosing processes of the trichogen cell and tormogen cell are present. Other symbol: pc, pupal cuticle. Bar: 3 μm.

Fig. 1.

Electron micrographs showing ultrastructural details of developing microchaetes (A, B, D), non-innervated long hairs of the posterior wing margin (C), and multiply innervated bristles of the anterior wing margin (E-G). A and B are oblique sections of a notum at 24 h apf. A depicts the apical part of an individual microchaete, with a central thecogen cell (th) surrounded by the trichogen cell (tr) and tormogen cell (to). The tip of the dendrite (d) of the sensory neuron (out of the plane of the section) has not reached the apical surface of the thecogen cell which ensheaths it. B shows the cell bodies of the sensory neuron (ne), trichogen cell, and tormogen cell. C depicts an oblique section of a posterior wing margin at 24 h apf. The non-innervated long hairs which form two rows along the posterior wing margin are composed of two ceils, a central trichogen cell (tr) and a tormogen cell (to). D shows an individual microchaete at 32 h apf. In the plane of the section are the tormogen cell and the trichogen cell, which has started to produce a shaft (sh). Cuticle has been deposited on the apical surface of epidermal cells, tormogen cell, and shaft (arrowheads). The curved dendrite is sectioned at both a basal level and an apical level. Basally (lower part of micrograph), it is enclosed within a thin process of the thecogen cell; apically, it has completely penetrated this cell and is surrounded by the dendritic cap, an electron-dense extracellular sheath secreted by the thecogen cell. The inset at bottom right shows the cell bodies of the sensory neuron and the closely apposed thecogen cell. E-G show transverse sections, at three different levels, of the same multiply innervated anterior wing margin bristle (32 h apf). E represents the basalmost section; it shows the five dendrites (d), surrounded by the cylindrical process of the thecogen cell and the mesaxon-like processes of the trichogen cell and tormogen cell. The tormogen cell occupies only half of the perimeter; a process of what is presumably the tormogen cell of a neighboring bristle (to’) covers the other half. F and G represent progressively more apical levels. The dendrites converge apically and terminate at approximately level F; in G, only the apical extension of the thecogen cell and the enclosing processes of the trichogen cell and tormogen cell are present. Other symbol: pc, pupal cuticle. Bar: 3 μm.

The thecogen cell has a small, flattened cell body attached to the apical surface of the neuron. The cellular sheath formed by the thecogen cell around the dendrite(s) stops apically a short distance from the tip of the dendrite; here it is replaced by the dendritic cap, an electron-dense extracellular sheath around the tip of the dendrite which is secreted by the thecogen cell (Schmidt and Gnatzy, 1971). Both trichogen cell and tormogen cell have large cell bodies, located on the opposite side of the neuron from the thecogen cell body. The tormogen cell lies level with the epidermal layer, while the cell body of the trichogen lies beneath the tormogen cell. The sheath process of the trichogen cell continues apically as the shaft-forming process; the sheath process of the tormogen cell continues as the socket-forming process. This latter collar-like process is asymmetric: At the side of the cell body it is thicker and higher than on the opposite side. This difference results in the typical asymmetric shape of the socket, which may determine the polarity of the shaft of the sensillum (or vice versa). The shaft always points in the direction of the higher part of the socket.

The majority, if not all, of the microchaetae of the notum possess a fourth, subepidermally located accessory cell (see Fig. 6D). The small cell body of this cell is attached to the sensory axon at short, variable distances from the soma. It probably forms a glial process around the axon and the basal soma and may be homologous to the cell called the soma sheath cell in Drosophila larval sensilla (Hartenstein, 1988).

Pattern of bristle sensilla on the notum and wing

The notum bears 26 macrochaetes, 13 on each side, which occur at invariant locations and therefore have been individually named (Fig. 2B; see Ferris, 1950). Microchaetes occur regularly spaced within the trapezoidal field which is delimited by the anterior boundary of the notum anteriorly, the suture between scutum and scutellum posteriorly, and a line connecting the presutural and posterior postalar macrochaetes on both sides. Within this field, there are 209±2.2 (males) or 244±13.7 (females) microchaetes (Fig. 2C). The average ratio of microchaetes to epidermal cells is 0.07, or approximately 14 epidermal cells for every microchaete. The ovoid region surrounding the dorsocentral macrochaetes has a lower density of microchaetes (Fig. 2B). The apparent increase in density from posterior to anterior seems to be due mostly to a smaller diameter of the intervening epidermal cells.

Fig. 2.

Pattern of distribution and time of appearance of adult sensilla of the anterior wing margin (A) and notum (B). Drawings derive from camera-lucida reconstructions of the right heminotum and right wing margin (not including the costa) of an individual male fly of the Canton S (wildtype) strain. Different sensillum types are represented by different symbols. Circles and ovals represent singly innervated (mechanosensory) sensilla, triangles symbolize multiply innervated (chemosensory) sensilla. Large circles represent macrochaetes (notum), medium circles represent microchaetes (notum), small circles represent slender bristles (wing margin), and ovals represent stout bristles (wing margin). Differential shading is used to indicate different times in pupal development at which various sensilla appear. The first expression by any cell of the antigen recognized by the MAb 22C10 is taken for the time of appearance. Black symbols represent sensilla which are present before 7h after puparium formation, speckled symbols denote sensilla appearing between 7h and 12 h apf, and white symbols represent sensilla appearing between 16 h and 20 h apf. The table (C) shows the average numbers (±S.D.) of the different types of sensilla, derived from reconstructions of 10 individuals. Results for female and male flies are listed separately. Females, which are larger than males, correspondingly have a somewhat larger number of all types of sensilla (except for macrochaetes). For the wing margin, sensilla arc listed as belonging to either the ‘Dorsal Row’ or the ‘Ventral Row’. The Ventral Row consists of the ventral component of the triple row (vTR) and the ventral component of the double row (vDR), which are identical in the pattern and types of bristles contained within them. The Dorsal Row consists of the medial and dorsal components of the triple row (mTR, dTR) and the dorsal component of the double row (dDR). Note that the intervals between multiply innervated bristles in both the dorsal and ventral row become larger and more irregular going from the proximal towards the distal part of the anterior wing margin. Symbols: A: vTR, mTR, dTR: ventral, medial, and dorsal components of the triple row; vDR, dDR: ventral and dorsal components of the double row; II, III: second and third longitudinal veins. B: Individual macrochaetes are designated as follows, uh, Ih: upper and lower humerais; ps: presutural; anp, pnp: anterior and posterior notopleurals; asa, psa: anterior and posterior supraalars; adc, pdc: anterior and posterior dorsocentrals; apa, ppa: anterior and posterior postalars; asc, psc: anterior and posterior scutellars. C: si, mi: singly and multiply innervated bristles; me: microchaetes; epi: epidermal cells. * Intervals are given as the average number of singly innervated bristles that intervene between two neighboring multiply innervated bristles. The point of intersection of the second longitudinal vein (II) with the wing margin was chosen as the boundary between ‘proximal’ and ‘distal’ wing margin. **The average ratio of microchaetes to epidermal cells was determined from camera-lucida drawings of whole mounts in which both microchaetes and epidermal cells were labeled [by double labeling with the MAbs 22C10 and 8C5 (the latter labels nuclei of all cells; Fujita et al. 1982)].

Fig. 2.

Pattern of distribution and time of appearance of adult sensilla of the anterior wing margin (A) and notum (B). Drawings derive from camera-lucida reconstructions of the right heminotum and right wing margin (not including the costa) of an individual male fly of the Canton S (wildtype) strain. Different sensillum types are represented by different symbols. Circles and ovals represent singly innervated (mechanosensory) sensilla, triangles symbolize multiply innervated (chemosensory) sensilla. Large circles represent macrochaetes (notum), medium circles represent microchaetes (notum), small circles represent slender bristles (wing margin), and ovals represent stout bristles (wing margin). Differential shading is used to indicate different times in pupal development at which various sensilla appear. The first expression by any cell of the antigen recognized by the MAb 22C10 is taken for the time of appearance. Black symbols represent sensilla which are present before 7h after puparium formation, speckled symbols denote sensilla appearing between 7h and 12 h apf, and white symbols represent sensilla appearing between 16 h and 20 h apf. The table (C) shows the average numbers (±S.D.) of the different types of sensilla, derived from reconstructions of 10 individuals. Results for female and male flies are listed separately. Females, which are larger than males, correspondingly have a somewhat larger number of all types of sensilla (except for macrochaetes). For the wing margin, sensilla arc listed as belonging to either the ‘Dorsal Row’ or the ‘Ventral Row’. The Ventral Row consists of the ventral component of the triple row (vTR) and the ventral component of the double row (vDR), which are identical in the pattern and types of bristles contained within them. The Dorsal Row consists of the medial and dorsal components of the triple row (mTR, dTR) and the dorsal component of the double row (dDR). Note that the intervals between multiply innervated bristles in both the dorsal and ventral row become larger and more irregular going from the proximal towards the distal part of the anterior wing margin. Symbols: A: vTR, mTR, dTR: ventral, medial, and dorsal components of the triple row; vDR, dDR: ventral and dorsal components of the double row; II, III: second and third longitudinal veins. B: Individual macrochaetes are designated as follows, uh, Ih: upper and lower humerais; ps: presutural; anp, pnp: anterior and posterior notopleurals; asa, psa: anterior and posterior supraalars; adc, pdc: anterior and posterior dorsocentrals; apa, ppa: anterior and posterior postalars; asc, psc: anterior and posterior scutellars. C: si, mi: singly and multiply innervated bristles; me: microchaetes; epi: epidermal cells. * Intervals are given as the average number of singly innervated bristles that intervene between two neighboring multiply innervated bristles. The point of intersection of the second longitudinal vein (II) with the wing margin was chosen as the boundary between ‘proximal’ and ‘distal’ wing margin. **The average ratio of microchaetes to epidermal cells was determined from camera-lucida drawings of whole mounts in which both microchaetes and epidermal cells were labeled [by double labeling with the MAbs 22C10 and 8C5 (the latter labels nuclei of all cells; Fujita et al. 1982)].

The pattern of sensilla on the anterior wing margin has been described in detail by Palka et al. (1979). Here (excluding the tegula and costa), sensilla stand in three rows proximally (triple row) and two rows distally (double row; see Fig. 2A). The boundary between triple row and double row is a short distance distal of the point where the second longitudinal vein merges with the wing margin; the double row ends where the third longitudinal vein merges with the wing margin. The three components of the triple row are composed of different types of sensilla. In the dorsal component, there are only multiply innervated, recurved (chemosensory) bristles. In the middle component, there are only singly innervated, stout (mechanosensory) bristles. The ventral component possesses multiply innervated, recurved bristles and singly innervated, slender (mechanosensory) bristles; the recurved bristles are spaced apart from each other by an average of four slender bristles (Fig. 2C). The double row possesses a dorsal and a ventral component which are almost identical, both consisting of recurved and slender bristles.

The ventral and medial component of the triple row, and both components of the double row, are composed of sensilla only, with no intervening epidermal cells. At the distal tip of the wing, the double row continues into a double row of long, non-innervated hairs. In the density and morphology of the shaft, these hairs are very similar to the slender bristles of the anterior wing margin. Although no true cuticular socket is visible surrounding these hairs, it is clear from electron microscopy that, during early and mid-pupal stages, a tormogen-like cell surrounds each shaft-forming cell (Fig. 1C).

Labeling of sensillum cells with MAb 22C10

In previous studies, the monoclonal antibody 22C10 has been used to label peripheral sensory neurons, including the photoreceptors of the eye, and selected central neurons in the Drosophila embryonic and pupal nervous systems (Zipursky et al. 1984; Canal and Ferrus, 1986; Tomlinson and Ready, 1987; Hartenstein, 1988). Early in the course of the experiments described here, we observed that 22C10 labels a number of cell types other than neurons. In developing sensilla at the pupal stage, particularly in the chaetae of the notum, it labels the neuron and three of the accessory cells (see Figs 3 and 9). The different sensillum cells display a rather dynamic staining pattern. The neuron, thecogen cell, and trichogen cell are labeled immediately following the terminal division of the sensillum precursor cells (Fig. 9). At this stage, the sensillum cells are not yet morphologically distinguishable. In the case of the multiply innervated wing marginal bristles, even premitotic neural precursors are labeled (see Fig. 8). The bristle shaft elaborated by the trichogen cell is strongly labeled from the beginning of its formation; shortly afterward (particularly in the macrochaetes), the trichogen cell body becomes heavily labeled as well. During later pupal development, the labeling of neurons, including dendrites and axons, remains strong. Labeling of the cell bodies of the thecogen cell and the trichogen cell becomes weaker. At these later stages, the socket formed by the tormogen cell also becomes labeled.

Fig. 3.

Temporal pattern of appearance of the sensilla of the notum and wing. A–E show whole-mount preparations of dissected nota (A, B) and right wings (C-E), labeled with the MAb 22C10. In A and B, the dashed line represents the dorsal midline. A represents a stage (8h apf) when the left and right wing discs have just fused at the dorsal midline to form the notum. Certain of the macrochaetes (asc: anterior scutellar; psc: posterior scutellar; pdc: posterior dorsocentral) are already present. The other labeled structures are persisting axons of the larval mesothoracic (nt2) and metathoracic (nt3) nerves. The nt2 axons branch to innervate the persisting dorsal larval musculature (mu), around which the indirect flight muscles will later be assembled. B shows a notum at 36h apf; all macrochaetes [asc, psc, pdc (see above); adc: anterior dorsocentral; psa: posterior supraalar; apa: anterior postalar] and microchaetes (me) are present and have formed axons and bristle shafts. C shows the right wing disc of a pupa at 2 h apf. The wing blade has not yet everted. The distal half of the dorsal and ventral rows (dR, vR) of multiply innervated sensilla have already appeared (the future distal tip of the wing blade is indicated by a star). Each of these sensilla is represented by a single precursor cell which will subsequently divide to give rise to the neurons and, possibly, the thecogen cell of the individual sensilla (see Fig, 8). The neurons of three campaniform sensilla [TSM (1): the first of the twin sensilla of the margin; ACV: anterior crossvein sensillum; L3-2: the second of the three sensilla on the third longitudinal vein; see Murray et al. (1984) for nomenclature] have also appeared. In D (8h apf), the wing blade has everted and all multiply innervated sensilla (mi) are present. Only the dorsal row (dR) is in the plane of focus. By this time each sensillum (arrowheads point to three examples) contains 2–4 cells. In E (24h apf), the singly innervated bristles (si) have also appeared. Arrowheads point to the multiply innervated bristles of the dorsal triple row, which are somewhat out of the plane of focus. The inset shows a small segment of the anterior wing margin at higher magnification to demonstrate the precise layering of the different cell types. In the middle are the cell bodies of the trichogen cells (tr; strongly labeled). Externally is the layer of tormogen cells (to; not labeled); internally, and immediately apposed to the marginal nerve (mn), are the neurons and thecogen cells (ne/th; irregularly labeled). Bar: 50 μm.

Fig. 3.

Temporal pattern of appearance of the sensilla of the notum and wing. A–E show whole-mount preparations of dissected nota (A, B) and right wings (C-E), labeled with the MAb 22C10. In A and B, the dashed line represents the dorsal midline. A represents a stage (8h apf) when the left and right wing discs have just fused at the dorsal midline to form the notum. Certain of the macrochaetes (asc: anterior scutellar; psc: posterior scutellar; pdc: posterior dorsocentral) are already present. The other labeled structures are persisting axons of the larval mesothoracic (nt2) and metathoracic (nt3) nerves. The nt2 axons branch to innervate the persisting dorsal larval musculature (mu), around which the indirect flight muscles will later be assembled. B shows a notum at 36h apf; all macrochaetes [asc, psc, pdc (see above); adc: anterior dorsocentral; psa: posterior supraalar; apa: anterior postalar] and microchaetes (me) are present and have formed axons and bristle shafts. C shows the right wing disc of a pupa at 2 h apf. The wing blade has not yet everted. The distal half of the dorsal and ventral rows (dR, vR) of multiply innervated sensilla have already appeared (the future distal tip of the wing blade is indicated by a star). Each of these sensilla is represented by a single precursor cell which will subsequently divide to give rise to the neurons and, possibly, the thecogen cell of the individual sensilla (see Fig, 8). The neurons of three campaniform sensilla [TSM (1): the first of the twin sensilla of the margin; ACV: anterior crossvein sensillum; L3-2: the second of the three sensilla on the third longitudinal vein; see Murray et al. (1984) for nomenclature] have also appeared. In D (8h apf), the wing blade has everted and all multiply innervated sensilla (mi) are present. Only the dorsal row (dR) is in the plane of focus. By this time each sensillum (arrowheads point to three examples) contains 2–4 cells. In E (24h apf), the singly innervated bristles (si) have also appeared. Arrowheads point to the multiply innervated bristles of the dorsal triple row, which are somewhat out of the plane of focus. The inset shows a small segment of the anterior wing margin at higher magnification to demonstrate the precise layering of the different cell types. In the middle are the cell bodies of the trichogen cells (tr; strongly labeled). Externally is the layer of tormogen cells (to; not labeled); internally, and immediately apposed to the marginal nerve (mn), are the neurons and thecogen cells (ne/th; irregularly labeled). Bar: 50 μm.

In addition to the sensillum accessory cells, other non-neural tissues are labeled as well. The pupal myoblasts and muscles are labeled strongly. A weak but reproducible reaction is observed with the trichogen-like cells that form the long non-innervated hairs along the posterior margin of the wing.

In all labeled cells, we have observed that the antigen is cytoplasmic; it is perhaps a cytoskeletal protein common to certain cells that elaborate long cytoplasmic extensions.

Temporal pattern of appearance of notum and wing sensilla

The sensilla of the notum and wing margin appear at different times during early pupal development (Figs. 2 and 3). Taking the first expression by any of the presumptive sensillum cells of the antigen recognized by the MAb 22C10 for the time when a given sensillum appears, the following picture emerges. There are, on both notum and wing, two phases during which sensilla appear. During the first phase (prior to 8 h apf until 12 h apf), the macrochaetes of the notum appear. Three of the macrochaetes - the anterior and posterior scutellars and the posterior dorsocentral - appear before the others. Microchaetes appear relatively synchronously in a second phase between 18 h and 20 h apf. For the wing margin, the first phase (around puparium formation) includes the chemosensory recurved bristles of the dorsal and ventral rows. In the second phase (16–20 h apf), the mechanosensory stout bristles and slender bristles, as well as the non-innervated hairs of the posterior margin, appear.

Our analysis suggests that the pattern differences between the double row and the triple row (Fig. 2A) may be viewed primarily as differences in the spatial organization of the singly innervated bristles. The multiply innervated bristles appear first, and are spaced at regular intervals in a dorsal and a ventral row along the entire anterior margin (Fig. 2C); no boundary between the later triple row and double row is evident. Then, in the entire ventral row, singly innervated bristles fill in the spaces left between neighboring multiply innervated bristles. But in the dorsal row, the singly innervated bristles behave differently at the proximal wing margin and the distal wing margin. Distally, in the territory of the later double row, they behave exactly as their ventral counterparts, fitting in between the multiply innervated bristles and forming slender bristle shafts. Proximally, in the later triple row, they seem instead to form a separate row (the medial triple row); these bristles develop stout shafts, instead of the slender shafts expressed by their counterparts in the ventral row and the distal dorsal row. In this view, then, the entire wing margin is characterized by a dorsal and a ventral row of bristles or hairs; in the triple row, the dorsal row has two components, one consisting entirely of recurved chemosensory bristles and one consisting entirely of stout mechanosensory bristles (Fig. 2A, C).

Spatiotemporal pattern of cell proliferation in the wing disc

The cells of the wing imaginal disc proliferate during larval and early pupal stages. BrdU pulse-labeling experiments reveal that there are two phases of high mitotic activity during the pupal period, one around puparium formation, the other between 9h and 20 h apf. Division is almost absent in the notum and wing between these phases and after 24 h apf.

During both phases, proliferation occurs in a distinct spatial pattern. On the notum, depending on the time of BrdU injection, labeled cells (i.e. cells that replicated their DNA during the period when BrdU was available, and their progeny) are clustered in regularly spaced groups. There is a central-to-peripheral gradient in the density of proliferating cells. In a continuous belt along the periphery of the notum most, if not all, cells replicate DNA once or twice during both phases described above. In the dorsocentral direction from this belt of high proliferatory activity, there is a zone in which only a minority of cells replicate. In the dorsocentral region itself, almost no cell division whatsoever occurs during the pupal period, except for some small clusters along the dorsal midline and the precursors of the microchaetes, which are discussed in the section below. The significant finding is that, throughout the notum, nuclei that simultaneously replicate DNA at any given time are not randomly distributed, but seem to lie in more or less evenly spaced clusters (Fig. 4).

Fig. 4.

Pattern of DNA-synthesizing cells in the pupal abdomen (A), notum (B), and head (C). Bromodeoxyuridine (BrdU) was injected into pupae at 16 h apf; pupae were dissected and fixed at 24 h apf. All photographs show whole-mount preparations of pupal tissues doubly labeled with a MAb against BrdU (to visualize cells that had replicated their DNA within a few hours after 16 h apf, as well as their progeny) and with the MAb 22C10 (to identify sensillum cells). A shows one abdominal hemisegment; anterior is to the left, dorsal at the top. In the middle are axons of the persisting larval segmental nerve (sn) around which myoblasts (mb) of the dorsal muscles have accumulated. Replicating nuclei (and their progeny) arc confined to a narrow transverse strip posterior to the nerve (arrowheads). B shows the posterior part of a heminotum; anterior is toward the bottom, lateral to the right. Some macrochaetes (psc: posterior scuteliar; pdc: posterior dorsocentral) and microchaetes (me) are visible. Labeled epidermal nuclei lie in patches which are non-randomly spaced (arrowheads). The same is true for patches of labeled epidermal cells in the head capsule (C). Bar: 5 μm.

Fig. 4.

Pattern of DNA-synthesizing cells in the pupal abdomen (A), notum (B), and head (C). Bromodeoxyuridine (BrdU) was injected into pupae at 16 h apf; pupae were dissected and fixed at 24 h apf. All photographs show whole-mount preparations of pupal tissues doubly labeled with a MAb against BrdU (to visualize cells that had replicated their DNA within a few hours after 16 h apf, as well as their progeny) and with the MAb 22C10 (to identify sensillum cells). A shows one abdominal hemisegment; anterior is to the left, dorsal at the top. In the middle are axons of the persisting larval segmental nerve (sn) around which myoblasts (mb) of the dorsal muscles have accumulated. Replicating nuclei (and their progeny) arc confined to a narrow transverse strip posterior to the nerve (arrowheads). B shows the posterior part of a heminotum; anterior is toward the bottom, lateral to the right. Some macrochaetes (psc: posterior scuteliar; pdc: posterior dorsocentral) and microchaetes (me) are visible. Labeled epidermal nuclei lie in patches which are non-randomly spaced (arrowheads). The same is true for patches of labeled epidermal cells in the head capsule (C). Bar: 5 μm.

To confirm the observation that the clusters of simultaneously replicating cells in the notum are distributed non-randomly, we carried out a statistical analysis of the pattern. Clusters of labeled nuclei were charted with a camera lucida for 10 specimens. A grid of equally sized squares, drawn on transparent paper, was randomly superimposed on each chart. The number of clusters falling into each individual square was counted. Assuming a random spatial distribution, and knowing the average density of the clusters, one can calculate the expected numbers of squares with n clusters in them using the Poisson distribution. The observed numbers were compared with these expected numbers using the Chi-square test. The difference between the observed distribution and random expectation was found to be highly significant (p < 0.001).

In the wing, cell division also follows a regular pattern (Fig. 5). During both phases of proliferation, cells of the presumptive wing blade that replicate DNA at the same time (and their progeny) form elongated clusters. The spatial arrangement of these clusters corresponds to that of the wing margins and wing veins (see also Schubiger and Palka, 1987). During the second proliferatory phase, there appears in each cluster a distinctive inward-out gradient in the time of DNA replication (Fig. 5C-E). Thus, cells along a narrow strip destined to become the veins replicate DNA first, followed by cells that lie adjacent to the presumptive veins and will become part of the intervein tissue. A similar progression is evident at the anterior and posterior wing margin: The externalmost cells (which, as will be shown later on, will become the tormogen cells and trichogen cells of the marginal bristles) replicate DNA first, along with the vein cells; they are followed by more interiorly located cells (which, for the anterior margin, are destined to become neurons and thecogen cells), which replicate along with the cells of the intervein tissue.

Fig. 5.

Pattern of DNA-synthesizing cells in the pupal wing. In A, a wing disc at 3 h apf was incubated for 2 h in BrdU-containing Ringers solution; in B-E, BrdU was injected into pupae at 11-13 h apf; these were then dissected and fixed at 24 h apf. All photographs show whole-mount preparations of pupal tissues labeled with a MAb against BrdU to visualize cells that had replicated their DNA during the time when BrdU was available, and their progeny. The dorsal surface is in the plane of focus; anterior is to the top, distal to the right. In A, labeled cells on the wing blade form radial stripes (large arrowheads) which in location roughly correspond to the longitudinal wing veins Ill, IV, and V. Small arrowhead points to one example of the pairs of labeled cells that appear at regular distances along the anterior wing margin; these probably correspond to replicating precursors of the multiply innervated sensilla (see Fig. 8). In B, labeled cells are also confined to the anterior and posterior wing margin and to the tissue around veins III, IV, and V. C-E show segments of the anterior wing margin (upper row), third longitudinal vein (middle row) and posterior wing margin (bottom row); these preparations were also labeled with MAb 22C10. In C, BrdU had been injected at 11 h apf. On the wing blade, labeled cells form a single thin strip within the third vein; at the anterior and posterior margins, labeling is confined to the externalmost rows of cells (i.e. the presumptive tormogen cells and trichogen cells of marginal sensilla and long hairs). Injecting slightly later (D), BrdU-containing cells occupy a different position. At the margin, they lie more interiorly, nearer the marginal nerve (mn), although there are still some scattered labeled cells in the external rows; on the blade, fewer cells lie on the vein itself, and more flank the vein tissue on either side. Injecting BrdU at 13 h apf (E) accentuates this picture; labeled cells are found among the innermost rows of the anterior wing margin (neurons and thecogen cells of marginal sensilla). No labeled cells are found along the posterior margin, which has no neurons or thecogen cells. On the blade, the vein itself is almost free of labeled cells [note 22C10-labeled nerve at center of third vein (nIII)]; labeled cells occupy narrow strips on either side of the vein. Bar: 20μm.

Fig. 5.

Pattern of DNA-synthesizing cells in the pupal wing. In A, a wing disc at 3 h apf was incubated for 2 h in BrdU-containing Ringers solution; in B-E, BrdU was injected into pupae at 11-13 h apf; these were then dissected and fixed at 24 h apf. All photographs show whole-mount preparations of pupal tissues labeled with a MAb against BrdU to visualize cells that had replicated their DNA during the time when BrdU was available, and their progeny. The dorsal surface is in the plane of focus; anterior is to the top, distal to the right. In A, labeled cells on the wing blade form radial stripes (large arrowheads) which in location roughly correspond to the longitudinal wing veins Ill, IV, and V. Small arrowhead points to one example of the pairs of labeled cells that appear at regular distances along the anterior wing margin; these probably correspond to replicating precursors of the multiply innervated sensilla (see Fig. 8). In B, labeled cells are also confined to the anterior and posterior wing margin and to the tissue around veins III, IV, and V. C-E show segments of the anterior wing margin (upper row), third longitudinal vein (middle row) and posterior wing margin (bottom row); these preparations were also labeled with MAb 22C10. In C, BrdU had been injected at 11 h apf. On the wing blade, labeled cells form a single thin strip within the third vein; at the anterior and posterior margins, labeling is confined to the externalmost rows of cells (i.e. the presumptive tormogen cells and trichogen cells of marginal sensilla and long hairs). Injecting slightly later (D), BrdU-containing cells occupy a different position. At the margin, they lie more interiorly, nearer the marginal nerve (mn), although there are still some scattered labeled cells in the external rows; on the blade, fewer cells lie on the vein itself, and more flank the vein tissue on either side. Injecting BrdU at 13 h apf (E) accentuates this picture; labeled cells are found among the innermost rows of the anterior wing margin (neurons and thecogen cells of marginal sensilla). No labeled cells are found along the posterior margin, which has no neurons or thecogen cells. On the blade, the vein itself is almost free of labeled cells [note 22C10-labeled nerve at center of third vein (nIII)]; labeled cells occupy narrow strips on either side of the vein. Bar: 20μm.

Pattern of division of sensillum precursor cells

Sensillum precursor cells divide during the two phases of general proliferatory activity described in the previous section (Figs 6, 7 and 8). The precursors of the macrochaetes and the chemosensory wing marginal bristles divide during the first proliferatory phase (around puparium formation; Fig. 8), those of the microchaetes and wing marginal mechanosensory bristles during the second phase [9–14 h apf for the wing marginal bristles (Fig. 7), 14–18 h apf for the microchaetes (Fig. 6)].

Fig. 6.

Pattern of division of microchaetc precursor cells. A shows schematic drawings of a small part of the notum at three consecutive pupal stages (approximately 10 h, 14h, and 18 h apf; see time scale). Dotted circles represent epidermal cells; cells with heavy outlines symbolize microchaete precursors (pl: first-order precursor; pila, pllb: second-order precursors) and their progeny. One particular first-order precursor, whose further division pattern is indicated by arrows, is shaded. Points of bifurcation of the arrows, projected on the time scale below, give the times when the divisions indicated by the bifurcations take place. In this and the following three Figures (Figs 7-9), all developmental times should be considered to have an error of ±lh, due to imprécisions in staging and variations in the rate of progress between successive pupal stages. The lightly shaded area represents a single microchaete into which all four progeny of one first-order precursor, whose names are given to the right, are incorporated. The fifth cell (glia cell) is of unknown origin (indicated by question mark). The tormogen cell undergoes one round of cndoreplication (ER), the trichogen cell two rounds. B and C show fuchsin-stained whole-mount preparations of nota fixed at 24 h apf. In these experiments, the division of sensillum precursors had been blocked by injecting colchicine. In B, colchicine was injected at approximately 14h apf (see ‘B’ on time scale). First-order precursors (pl), undergoing no further division, have increased in volume; they are labeled with the MAb 22C10. In C, colchicine was injected at 16 h apf (‘C’ on time scale). In some cases, both secondary precursors were blocked before their division, yielding pairs of metaphase cells (plla/b); in other cases, plia had already divided shortly prior to the injection, leaving only pllb to be blocked in metaphase. D and E show parts of whole-mount preparations of nota at 24 h apf, labeled with anti-BrdU and MAb 22C10. In D, BrdU was injected at 16 h apf (‘D’ on time scale); the progeny of both plia and pllb (to: tormogen cell; tr: trichogen cell; th: thecogen cell; ne: neuron) have incorporated BrdU. In addition, the glia cell (gl) and small clusters of epidermal cells along the dorsal midline (arrowhead) have incorporated BrdU. In E, BrdU was injected at 17 h apf (‘E’ on time scale); only the progeny of pllb (th, ne) have incorporated the analogue. In F and G, nota dissected at 26 h and 28 h apf (‘F’,‘G’ on time scale), respectively, were incubated for 2h in BrdU-containing Ringers and then labeled with anti-BrdU and MAb 22C10. In F, trichogen cells, which endoreplicate at approximately 26 h apf, have labeled nuclei; in G, tormogen cell nuclei are labeled. Bar: 20 μm

Fig. 6.

Pattern of division of microchaetc precursor cells. A shows schematic drawings of a small part of the notum at three consecutive pupal stages (approximately 10 h, 14h, and 18 h apf; see time scale). Dotted circles represent epidermal cells; cells with heavy outlines symbolize microchaete precursors (pl: first-order precursor; pila, pllb: second-order precursors) and their progeny. One particular first-order precursor, whose further division pattern is indicated by arrows, is shaded. Points of bifurcation of the arrows, projected on the time scale below, give the times when the divisions indicated by the bifurcations take place. In this and the following three Figures (Figs 7-9), all developmental times should be considered to have an error of ±lh, due to imprécisions in staging and variations in the rate of progress between successive pupal stages. The lightly shaded area represents a single microchaete into which all four progeny of one first-order precursor, whose names are given to the right, are incorporated. The fifth cell (glia cell) is of unknown origin (indicated by question mark). The tormogen cell undergoes one round of cndoreplication (ER), the trichogen cell two rounds. B and C show fuchsin-stained whole-mount preparations of nota fixed at 24 h apf. In these experiments, the division of sensillum precursors had been blocked by injecting colchicine. In B, colchicine was injected at approximately 14h apf (see ‘B’ on time scale). First-order precursors (pl), undergoing no further division, have increased in volume; they are labeled with the MAb 22C10. In C, colchicine was injected at 16 h apf (‘C’ on time scale). In some cases, both secondary precursors were blocked before their division, yielding pairs of metaphase cells (plla/b); in other cases, plia had already divided shortly prior to the injection, leaving only pllb to be blocked in metaphase. D and E show parts of whole-mount preparations of nota at 24 h apf, labeled with anti-BrdU and MAb 22C10. In D, BrdU was injected at 16 h apf (‘D’ on time scale); the progeny of both plia and pllb (to: tormogen cell; tr: trichogen cell; th: thecogen cell; ne: neuron) have incorporated BrdU. In addition, the glia cell (gl) and small clusters of epidermal cells along the dorsal midline (arrowhead) have incorporated BrdU. In E, BrdU was injected at 17 h apf (‘E’ on time scale); only the progeny of pllb (th, ne) have incorporated the analogue. In F and G, nota dissected at 26 h and 28 h apf (‘F’,‘G’ on time scale), respectively, were incubated for 2h in BrdU-containing Ringers and then labeled with anti-BrdU and MAb 22C10. In F, trichogen cells, which endoreplicate at approximately 26 h apf, have labeled nuclei; in G, tormogen cell nuclei are labeled. Bar: 20 μm

Fig. 7.

Pattern of division of precursor cells of the singly innervated bristles of the anterior wing margin. A shows schematic drawings of a small segment of the anterior wing margin at three consecutive pupal stages (approximately 6h, 10h, and 14h apf; see time scale). Grey lines denote the rim of the wing margin. Dotted circles represent epidermal cells; circles with heavy outlines symbolize sensillum precursor cells (pl: first-order precursor; pH: second-order precursor) and their progeny. One hypothetical precursor, whose further division pattern is indicated by arrows, is shaded. Points of bifurcation of the arrows, projected on the time scale below, give the times when the divisions indicated by the bifurcations take place. The lightly shaded area represents a single bristle; note that not all of the daughter cells of a single precursor are incorporated into this bristle, and that it is therefore composed of progeny of more than one precursor. The bristle cells, whose names are given to the right, form regular strata parallel to the margin. The trichogen cell undergoes at least one round of endoreplication (ER). B shows a portion of a whole-mount preparation of a wing disc which was dissected at 8.5 h apf, incubated for 2h in BrdU-containing Ringers (see ‘B’ on time scale), and labeled with anti-BrdU. Along the anterior wing margin, there appears a strip of cells 2–3 cells wide (large arrowhead) which contain BrdU. They correspond to the second-order bristle precursors (pH in A). Small arrowheads point to regularly spaced precursors of multiply innervated wing margin bristles (see Fig. 8). In C, the division of the second-order precursors was blocked by injecting colchicine at 12 h apf (‘C’ on time scale), leading to the appearance of 4–5 rows of metaphase cells along the margin (arrowhead); fuchsin stained. D and E show portions of whole-mount preparations of wing discs at 24 h apf, labeled with anti-BrdU and MAb 22C10. In D, BrdU was injected at 12 h apf (‘D’ on time scale); the progeny of second-order precursors giving rise primarily to tormogen cells (to) and trichogen cells (tr) have incorporated the analogue. In E, BrdU injection was at 13 h apf (’E’ on time scale); primarily neurons (ne) and thecogen cells (th) have incorporated the analogue. Small arrowheads in D point to pairs of labeled cells, consisting of one tormogen cell (more external) and one trichogen cell (more internal); each pair belongs to an individual bristle. Large arrowheads in D point to bristles in which only the tormogen cell is labeled. Medium size arrowheads in D and E mark bristles with only the trichogen cell labeled. The wing marginal nerve (mn) is labeled with MAb 22C10. Bar: 20 μm.

Fig. 7.

Pattern of division of precursor cells of the singly innervated bristles of the anterior wing margin. A shows schematic drawings of a small segment of the anterior wing margin at three consecutive pupal stages (approximately 6h, 10h, and 14h apf; see time scale). Grey lines denote the rim of the wing margin. Dotted circles represent epidermal cells; circles with heavy outlines symbolize sensillum precursor cells (pl: first-order precursor; pH: second-order precursor) and their progeny. One hypothetical precursor, whose further division pattern is indicated by arrows, is shaded. Points of bifurcation of the arrows, projected on the time scale below, give the times when the divisions indicated by the bifurcations take place. The lightly shaded area represents a single bristle; note that not all of the daughter cells of a single precursor are incorporated into this bristle, and that it is therefore composed of progeny of more than one precursor. The bristle cells, whose names are given to the right, form regular strata parallel to the margin. The trichogen cell undergoes at least one round of endoreplication (ER). B shows a portion of a whole-mount preparation of a wing disc which was dissected at 8.5 h apf, incubated for 2h in BrdU-containing Ringers (see ‘B’ on time scale), and labeled with anti-BrdU. Along the anterior wing margin, there appears a strip of cells 2–3 cells wide (large arrowhead) which contain BrdU. They correspond to the second-order bristle precursors (pH in A). Small arrowheads point to regularly spaced precursors of multiply innervated wing margin bristles (see Fig. 8). In C, the division of the second-order precursors was blocked by injecting colchicine at 12 h apf (‘C’ on time scale), leading to the appearance of 4–5 rows of metaphase cells along the margin (arrowhead); fuchsin stained. D and E show portions of whole-mount preparations of wing discs at 24 h apf, labeled with anti-BrdU and MAb 22C10. In D, BrdU was injected at 12 h apf (‘D’ on time scale); the progeny of second-order precursors giving rise primarily to tormogen cells (to) and trichogen cells (tr) have incorporated the analogue. In E, BrdU injection was at 13 h apf (’E’ on time scale); primarily neurons (ne) and thecogen cells (th) have incorporated the analogue. Small arrowheads in D point to pairs of labeled cells, consisting of one tormogen cell (more external) and one trichogen cell (more internal); each pair belongs to an individual bristle. Large arrowheads in D point to bristles in which only the tormogen cell is labeled. Medium size arrowheads in D and E mark bristles with only the trichogen cell labeled. The wing marginal nerve (mn) is labeled with MAb 22C10. Bar: 20 μm.

Fig. 8.

Pattern of division of precursor cells of the multiply innervated bristles of the anterior wing margin. A shows schematic drawings of a small segment of the anterior wing margin at four consecutive pupal stages (approximately 2h, 6h, 8h, and 12 h apf; see time scale). Grey lines denote the rim of the wing margin. Dotted circles represent epidermal cells; cells with medium thick outlines denote precursors of singly innervated bristles; heavy outlines symbolize the precursors of one multiply innervated bristle and their progeny. Heavily shaded are the precursors of the neurons and, presumably, the thecogen cell (pn); medium shading marks the precursor of the tormogen cell and the trichogen cell. The lightly shaded area represents a single multiply innervated bristle. Points of bifurcation of the arrows, projected on the time scale below, give the times when the divisions indicated by the bifurcations take place. Dotted arrows indicate divisions for which no direct evidence has been obtained. B shows part of a whole-mount preparation of a wing disc at 2h apf, labeled with the MAb 22C10 (sec ‘B’ on time scale). First-order neural precursors (pn in A) of some of the multiply innervated bristles of both the dorsal row (dR) and the ventral row (vR) are labeled (the star indicates the future distal wing tip). C shows part of a wholemount preparation of a wing disc dissected at 4h apf (‘C’ on time scale) and incubated for 2h in BrdU-containing Ringers. The preparation was labeled with anti-BrdU and MAb 22C10. All along the anterior margin there appear clusters of 22C10positive neural precursors; in several clusters there are precursors which have incorporated BrdU (arrowheads point to examples). D shows part of a whole-mount preparation of a wing disc at 24 h apf, labeled with anti-BrdU and MAb 22C10. BrdU was injected at 6.5 h apf (‘D’ on time scale); the nuclei of tormogen cells (to) and trichogen cells (tr) and some of the neurons (ne) have incorporated the analogue, mn indicates the marginal nerve, labeled with MAb 22C10. Bar: 20 μm.

Fig. 8.

Pattern of division of precursor cells of the multiply innervated bristles of the anterior wing margin. A shows schematic drawings of a small segment of the anterior wing margin at four consecutive pupal stages (approximately 2h, 6h, 8h, and 12 h apf; see time scale). Grey lines denote the rim of the wing margin. Dotted circles represent epidermal cells; cells with medium thick outlines denote precursors of singly innervated bristles; heavy outlines symbolize the precursors of one multiply innervated bristle and their progeny. Heavily shaded are the precursors of the neurons and, presumably, the thecogen cell (pn); medium shading marks the precursor of the tormogen cell and the trichogen cell. The lightly shaded area represents a single multiply innervated bristle. Points of bifurcation of the arrows, projected on the time scale below, give the times when the divisions indicated by the bifurcations take place. Dotted arrows indicate divisions for which no direct evidence has been obtained. B shows part of a whole-mount preparation of a wing disc at 2h apf, labeled with the MAb 22C10 (sec ‘B’ on time scale). First-order neural precursors (pn in A) of some of the multiply innervated bristles of both the dorsal row (dR) and the ventral row (vR) are labeled (the star indicates the future distal wing tip). C shows part of a wholemount preparation of a wing disc dissected at 4h apf (‘C’ on time scale) and incubated for 2h in BrdU-containing Ringers. The preparation was labeled with anti-BrdU and MAb 22C10. All along the anterior margin there appear clusters of 22C10positive neural precursors; in several clusters there are precursors which have incorporated BrdU (arrowheads point to examples). D shows part of a whole-mount preparation of a wing disc at 24 h apf, labeled with anti-BrdU and MAb 22C10. BrdU was injected at 6.5 h apf (‘D’ on time scale); the nuclei of tormogen cells (to) and trichogen cells (tr) and some of the neurons (ne) have incorporated the analogue, mn indicates the marginal nerve, labeled with MAb 22C10. Bar: 20 μm.

Our rationale for attempting to reconstruct lineage relationships between individual sensillum cells using BrdU and colchicine is as follows. First, if after injecting BrdU at a given time and fixing at a later stage (e.g., 24 h apf, when cells are postmitotic and can readily be identified by their differentiated phenotype), one cell of an individual sensillum is found to contain BrdU, and another one is not, these two cells cannot be sister cells. Secondly, if only two cells of’an individual sensillum contain BrdU, and all neighboring cells (including the remaining sensillum cells and the surrounding epidermal cells) do not, these two cells must be sisters. Finally, preparations of pupae injected with colchicine at specific stages can be examined for the presence of mitotic figures to define the timing of mitotic events during precursor cell proliferation. Most importantly, this makes it possible to distinguish between proliferatory and endoreplicative DNA synthesis.

Microchaetes of the notum

The precursor cells of the more peripherally located microchaetes are included within the regularly spaced clusters of dividing, non-sensillum-producing cells described in the previous section. On the dorsocentral notum, the microchaete precursor cells are the only cells to divide; this enabled us to reconstruct the exact sequence of divisions by which these microchaete cells are generated (Fig. 6). Four cells of each dorsocentral microchaete, namely the neuron, thecogen cell, trichogen cell, and tormogen cell, derive from a single precursor cell. This cell, which in the following will be referred to as the first-order precursor (pI), divides at approximately 15 h apf into two second-order precursor cells (pII). One of these (pIIa) divides at approximately 17 h apf to give rise to the presumptive trichogen cell and tormogen cell; the other one (pIIb) divides 1–2 h later and produces the neuron and the thecogen cell. First-order and second-order precursor cells, as well as the sensillum cells prior to the onset of differentiation, are indistinguishable, in terms of size, shape, and level within the cell layer, from the surrounding epidermal cells. Their mitotic spindles are oriented parallel to the epidermal surface.

The majority of soma sheath cells derive from precursors which also carry out their terminal DNA replication at approximately 16 h apf. From our study we were unable to determine the origin of these precursors. As, by inference, they cannot form part of the pairs of second-order precursors giving rise to the other four microchaete cells, they must be located at some distance from these. It is possible that they derive from the adepithelial cells which are located beneath the epidermis and which primarily produce the adult muscles.

Between 18 h and 24 h apf, the microchaete cells rearrange in shape and position (see ‘Differentiation of sensillum cells’ below). Later, accompanying the formation of bristle shafts and sockets, trichogen cells and tormogen cells become polyploid by endoreplication. Similar to the precursor cell divisions, this event is highly coordinated among all microchaetes (Fig. 6). Trichogen cells endoreplicate twice, between 24 h and 26 h apf and between 32 h and 34 h apf. Tormogen cells perform one round of endoreplication, out of phase with the trichogen cells, at approximately 28 h apf. With each endoreplication the nucleus and cytoplasm of a cell become significantly larger; at 34 h, after two rounds of endoreplication in the case of the trichogen cells, and one round of endoreplication in the case of the tormogen cells, the nuclei of these ceils have reached their final size. This finding suggests that little, if any, endoreplication occurs after 34 h apf; however, we did not follow endoreplication by means of BrdU pulse-labeling experiments any further than this stage.

Mechanosensory bristles of the anterior wing margin

The precursor cells of these sensilla divide as part of the second proliferatory phase in the wing (9-14h apf). As these bristles lie immediately adjacent to each other in two continuous rows, it was not possible to discern individual precursor cells. It is clear, however, that the cells of these bristles, like the microchaete cells of the notum, are generated by two rounds of mitosis (Fig. 7). At approximately 9 h apf, the externalmost 1–2 rows of cells on both the dorsal and ventral wing margin divide as a group. The second round of mitosis occurs between 12 h and 14 h apf. These two divisions may be considered analogous to the division of the first-order precursors and the second-order precursors of the microchaetes, respectively. As described above (see ‘Spatio-temporal pattern of cell proliferation in the wing disc’), the external row of ‘secondary precursors’, which ultimately give rise to tormogen cells and trichogen cells, divide earlier than the internal row, which produces neurons and thecogen cells. During the entire proliferatory phase (i.e. between 9h and 14h apf), not only the sensillum precursors, but also numerous adjacent epidermal precursors divide.

The BrdU pulse-labeling experiments provide evidence that there may not exist a strict clonal relationship between the cells of an individual wing marginal bristle. In wings of pupae that had been injected with BrdU at 12 h apf, and fixed at 24 h apf, label is found in the majority of trichogen cells and tormogen cells, as well as some neurons and thecogen cells. In a significant fraction of sensilla [25 % of the analyzed cases (53/212)], the only labeled cell was the trichogen cell or the tormogen cell (Fig. 7D, E). This finding implies that in some sensilla of the wing margin, the trichogen cell and tormogen cell are not sisters, but instead derive from different precursors. In approximately half of these cases (25/212), label was present in several neighboring nuclei, all of which belonged to either the trichogen or the tormogen cell layer, but not both. This implies that a given precursor cell can produce two trichogen cells or two tormogen cells of adjacent bristles, instead of a trichogen and a tormogen of the same bristle. In the remaining cases (28/212), labeled nuclei were present in both layers, so that a bristle with only a labeled trichogen cell was adjacent to another bristle with only a labeled tormogen cell. This suggests that a given precursor may produce a trichogen and a tormogen as usual, but these may become incorporated into separate bristles. Our preparations do not allow us to distinguish, by position or morphology, neurons and thecogen cells in the wing margin; we therefore have no evidence as to whether these types of cells can also be generated in a lineage-independent manner.

The trichogen cells of the wing margin mechanosensory bristles become polyploid as a result of a single round of endoreplication, which takes place at approximately 20h apf (Fig. 7). Unlike the case of the notum microchaetes, the tormogen cells of these wing margin sensilla do not appear to participate in endoreplication.

Chemosensory bristles of the anterior wing margin

The cells of the multiply innervated chemoreceptors are generated in a distinctive way (see also Peters, 1965). At puparium formation, there appear on both the dorsal and ventral wing margin a row of individual cells which are recognized by the MAb 22C10 (Fig. 8B). Before 8h apf most, if not all, of these cells divide twice; possibly, there follows a third division of one or two of the progeny. The resulting groups of cells (all of which are 22C10-positive) can be observed to differentiate as the neurons of the chemoreceptors. Thus, the antigen recognized by the MAb 22C10 is already expressed by the precursors of the neurons (pn) at puparium formation. Previous studies (and the remainder of the present work) would suggest that the appearance of this antigen is associated with the onset of neural differentiation; perhaps this is incorrect. Alternatively, this may be a bona fide case in which a particular aspect of neural differentiation is expressed in premitotic neural precursors.

The precursors of the trichogen and tormogen cells commence their final DNA replication at approximately 6 h apf. These precursors are most likely distinct from the 22C10-positive neural precursors, most of which can be directly observed not to give rise to trichogen or tormogen cells. On the other hand, the thecogen cell could be produced by one of the terminal divisions of the neural precursors.

By analogy to the other sensilla studied here, it seems appropriate to consider both the neural precursor and the precursor of the trichogen and tormogen cells as second-order precursors. We have not yet been able to determine whether there exists a common (first-order) precursor for these two secondary precursors. If it exists, it divides prior to puparium formation.

Other sensilla

We have not directly studied the division of precursor cells for other groups of adult sensilla. Nevertheless, it is possible from our data to estimate the time of these events, based on the time of first appearance of differentiated sensillum cells for each group. The precursors of the macrochaetes of the head and notum, and those of the campaniform sensilla of the wing, divide approximately at the time of puparium formation or shortly after; i.e. at roughly the same time as the precursors of the chemosensory bristles of the anterior wing margin. Precursor cell division for the microchaetes of the head and for the interommatidial bristles occurs at approximately the same time as for the notum microchaetes (14–18h apf). Finally, the precursors of the abdominal bristles divide between approximately 24 h and 30 h apf.

Differentiation of sensillum cells

The cytodifferentiation of all sensilla analyzed in this study is fairly similar. In the following, the development of one class of sensilla, the microchaetes, will be described (Fig. 9).

Fig. 9.

Microchaete differentiation. A-E show parts of whole-mount preparations of nota fixed and labeled with MAb 22C10 at the times indicated by arrows extending from the time scale (middle column; hours after puparium formation); to the right are schematic diagrams of longitudinal sections of a microchaete at different stages (arrows from the time scale). Vertical bars immediately to the right of the time scale show the times of the major phases of sensillum development. A and B: 20h apf. In A, regularly spaced triplets of 22C10-positive cells, which correspond to the presumptive neuron, thecogen cell, and trichogen cell, are illustrated. In B, nuclei of all cells are labeled with the MAb 8C5. In this double-labeled preparation, the cytoplasmic MAb 22C10 label (large arrowheads) of presumptive microchaete cells is very faint. Presumptive microchaete cell bodies are caught at different stages during their subepidermal shift. In the cluster marked 1, all cells lie in the same plane. In the clusters marked 2, the neuron (dark nucleus, arrow) has shifted interiorly, as shown by the fact that it is overlain by another nucleus (small arrowhead; slightly out of focus). In the cluster marked 3, the cell bodies of three cells (neuron, thecogen cell, trichogen cell; all slightly out of focus) have left the epidermal layer; they are covered by epidermal nuclei. The inset is focused on the plane of the three subepidermal sensillum cells. In C (26h apf), the neuron (ne), thecogen cell (th), and trichogen cell (tr) are subepidermal and have a spindle-like shape. Neurons, which have sent out axonal growth cones (gc), are always more darkly labeled by MAb 22C10. D and E: 36h apf. In E, trichogen cells have produced bristle shafts (sh); thecogen cells are very thin and closely apposed to the neuron and are not readily recognized in whole-mount preparations at this stage. Inset at bottom right shows two microchaetes at 40 h apf labeled only with MAb21A6, which reacts specifically with the dendritic cap (de) secreted by the thecogen cell. D shows a macrochaete to demonstrate that, in addition to the neuron and trichogen cell, the tormogen cell (to) has also become 22C10-positive by this stage. The glia cell (gl on diagram at right) does not react with the antibody at any stage. Other symbols: cu: cuticle; so: socket. Bar: 20 μm.

Fig. 9.

Microchaete differentiation. A-E show parts of whole-mount preparations of nota fixed and labeled with MAb 22C10 at the times indicated by arrows extending from the time scale (middle column; hours after puparium formation); to the right are schematic diagrams of longitudinal sections of a microchaete at different stages (arrows from the time scale). Vertical bars immediately to the right of the time scale show the times of the major phases of sensillum development. A and B: 20h apf. In A, regularly spaced triplets of 22C10-positive cells, which correspond to the presumptive neuron, thecogen cell, and trichogen cell, are illustrated. In B, nuclei of all cells are labeled with the MAb 8C5. In this double-labeled preparation, the cytoplasmic MAb 22C10 label (large arrowheads) of presumptive microchaete cells is very faint. Presumptive microchaete cell bodies are caught at different stages during their subepidermal shift. In the cluster marked 1, all cells lie in the same plane. In the clusters marked 2, the neuron (dark nucleus, arrow) has shifted interiorly, as shown by the fact that it is overlain by another nucleus (small arrowhead; slightly out of focus). In the cluster marked 3, the cell bodies of three cells (neuron, thecogen cell, trichogen cell; all slightly out of focus) have left the epidermal layer; they are covered by epidermal nuclei. The inset is focused on the plane of the three subepidermal sensillum cells. In C (26h apf), the neuron (ne), thecogen cell (th), and trichogen cell (tr) are subepidermal and have a spindle-like shape. Neurons, which have sent out axonal growth cones (gc), are always more darkly labeled by MAb 22C10. D and E: 36h apf. In E, trichogen cells have produced bristle shafts (sh); thecogen cells are very thin and closely apposed to the neuron and are not readily recognized in whole-mount preparations at this stage. Inset at bottom right shows two microchaetes at 40 h apf labeled only with MAb21A6, which reacts specifically with the dendritic cap (de) secreted by the thecogen cell. D shows a macrochaete to demonstrate that, in addition to the neuron and trichogen cell, the tormogen cell (to) has also become 22C10-positive by this stage. The glia cell (gl on diagram at right) does not react with the antibody at any stage. Other symbols: cu: cuticle; so: socket. Bar: 20 μm.

Immediately following the terminal division of the sensillum precursors, the presumptive neuron, thecogen cell, trichogen cell, and tormogen cell all lie in the epidermal plane. They are indistinguishable in size and shape, but their arrangement is somewhat variable. Sometimes they form a row; in other cases, they are grouped in a cloverleaf pattern.

The subsequent differentiation of the sensillum can be divided into two phases (Fig. 9). In the first or morphogenetic phase (18–24 h apf), sensillum cells rearrange in shape and position to adopt the basic shape and position typical for a mature sensillum (Figs 1 and 9):

  1. The neuron shifts interiorly. It grows out a dendrite apically, and an axon basally. The dendrite grows through the overlying thecogen cell. The axonal growth cone extends in a caudolateral direction along the basement membrane of the epidermis. Not infrequently, neurons with more than one outgrowing axon can be observed. A small cell with a presumably glial function becomes associated with the axon near its junction with the soma. As noted above, the origin of this cell could not be determined. The elongating axon contacts and fasciculates with the axon of the nearest neighboring microchaete. In this manner, all axons are funneled in a caudolateral direction; at the lateral boundary of the notum they meet the preexisting macrochaete axons. All axons then fasciculate with axons of the larval mesothoracic segmental nerve, which have persisted through metamorphosis (Fig. 3A). They use this guiding track to reach their destination, the thoraco-abdominal ganglion.

  2. The accessory cells converge apically. The trichogen cell and tormogen cell begin wrapping mesaxon-|ike sheath processes around the thecogen cell. The thecogen cell, which lies in the center of the accessory cell cluster, is invaded from beneath by the outgrowing dendrite.

  3. The cell bodies of the accessory cells are displaced subepidermally, while the concentrically arranged sheath processes all stay in contact with the surface. The thecogen cell body is displaced the most; it comes to lie next to the cell body of the neuron. The trichogen cell and tormogen cell bodies, the latter on top of the former (see also Lees and Waddington, 1942), come to lie beside the dendrite; they are located on the opposite side of the dendrite from the thecogen cell body.

The second or differentiative phase (24–44 h apf) is characterized by two main events:

  1. The trichogen cell and the tormogen cell grow out a shaft and a socket, respectively. Both structures may be considered as the apical extension of the concentric sheath processes of these cells. With the onset of shaft formation and socket formation, the trichogen cell and tormogen cell increase in volume, possibly as a consequence of the endoreplication described in a previous section (see ‘Pattern of division of sensillum precursor cells’). The beginning of shaft formation always precedes socket formation.

  2. Cuticle is deposited on the apical surface of all epidermal cells, including the trichogen cell and tormogen cell. The thecogen cell, during this phase, secretes the dendritic cap. This structure, which is specifically recognized by the MAb 21A6 (Zipursky et al. 1984), is first detectable with the antibody at approximately 30 h apf.

For other classes of sensilla (including the macrochaetes of the head and notum, the microchaetes and interommatidial bristles of the head, the bristles of the anterior wing margin, and the abdominal bristles), the events of the first or morphogenetic phase, as well as the onset of endoreplication, follow the division of the respective precursor cells by approximately the same interval as for the microchaetes of the notum. For example, the precursors of the singly innervated bristles of the wing margin complete their divisions by 12-13 h apf (versus 18h for the notum microchaetes), and have reached the end of the morphogenetic phase as described above by approximately 18 h apf (versus 24 h for the notum microchaetes). By contrast, with the exception of endoreplication, the events of the second or differentiative phase are coordinated with the deposition of the imaginal cuticle, which begins simultaneously over the entire head and thorax at approximately 24 h apf, and in the abdomen at 36–40 h apf (Madhavan and Madhavan, 1980). Thus, for the different classes of head and thorax sensilla, these events follow precursor cell division and the morphogenetic phase of cytodifferentiation by a variable interval. Similar observations have been reported for the cercus of the grasshopper (Shankland and Bentley, 1983) and for Drosophila larval sensilla (Hartenstein, 1988), leading to the suggestion that different mechanisms might be responsible for controlling earlier biochemical and morphogenetic events (including endoreplication) on the one hand, and the events associated with cuticle deposition on the other (Shankland and Bentley, 1983).

The role of cell lineage and cell division in sensillum differentiation

Morphological studies of the development of sensilla in a variety of insect species (see Lawrence, 1966, for review) indicated that the cells constituting an individual sensillum are the progeny of a single precursor cell (the sensillum ‘mother cell’). This cell, in a fixed sequence of ‘differentiative’ divisions, was observed to produce non-equivalent daughter cells which differ reproducibly in size, staining properties, or (via nonhorizontal planes of division) position relative to the epidermal layer. It was therefore speculated that these differentiative divisions play a crucial role in determining the fate of the presumptive sensillum cells; for example, determinative factors expressed by the mother cell might be distributed unequally to the different daughter cells (Henke, 1953; Lawrence, 1966). Our observations in Drosophila are relevant to these proposals.

A strict clonal relationship may not always exist among the cells of individual sensilla

Our data suggest that a fixed clonal relationship may not exist among the cells of individual wing margin mechanoreceptors. We have found, for example, that during its terminal division, a given second-order precursor may give rise to two trichogen cells, or two tormogen cells, of adjacent bristles, instead of one trichogen cell and one tormogen cell of the same bristle. Thus, in these exceptional sensilla, the trichogen cell and the tormogen cell are not sister cells.

An alternative explanation of the patterns of BrdU-labeled cells in the anterior wing margin, upon which we base the above conclusion, is an artifact of the labeling technique. For example, due to inhomogeneous diffusion into the wing tissue, the anti-BrdU antibody might bind to only some of the BrdU-containing nuclei, and might then label the tormogen cells, but not the trichogen cells, deriving from two adjacent precursors. We believe that inhomogeneity on this fine a scale is unlikely, but cannot rule it out at present.

If the observation reflects a real phenomenon, it indicates that the sensillum precursors at the wing margin are not determined as to which types of progeny cells they will produce. The determination of sensillum cell fate within a single bristle organ would consequently depend not on cell lineage, but on some non-autonomous, postmitotic mechanism such as cell-cell interaction.

There have been few experimental studies of the causative role of cell lineage in the determination of sensillum cell fate. Clever (1960), studying developing bristles in Galleria, observed that after selective removal of the presumptive sensory neurons by treatment with methylene blue, trichogen and tormogen cells failed to differentiate. This result suggests that cell-cell interaction plays an essential role in establishing sensillum cell fate. A recent study in our laboratory supports this conclusion. We have found that shifting pupae bearing a temperature-sensitive allele of Notch (Ntsl) to the restrictive temperature at 16 h apf causes all progeny of some first-order microchaete precursors to develop as neurons (unpublished observations). This implies that presumptive sensillum cells are not autonomously committed to become either neurons or accessory cells, consistent with a role for a Notch dependent cell-cell interaction step in sensillum cell fate determination.

Displacement of cells during sensillum morphogenesis is independent of precursor cell division

We have found that, prior to the onset of differentiation, all presumptive sensillum cells lie in the plane of the epidermis. This observation contrasts with previous findings in other species, in which the plane of mitosis for some of the sensillum precursor divisions is perpendicular or oblique to the epidermal plane, thereby displacing one or more of the presumptive sensillum cells subepidermally (see Lawrence, 1966; Keil and Steinbrecht, 1986). In Drosophila, the subepidermal segregation of the presumptive neuron, as well as the subepidermal shift of the cell bodies of all accessory cells, occurs following the completion of precursor cell division and is therefore independent of it. We have obtained additional evidence which further supports this point. First, following colchicine injection into pupae, undivided first-order microchaete precursors eventually segregate subepidermally (our unpublished observations), demonstrating that this event does not require cell division. Secondly, in embryos homozygous for the lethal mutation string (stg), no cell division occurs after blastoderm formation (Edgar and O’Farrell, 1989; our unpublished observations). Nevertheless, undivided sensillum precursor cells differentiate as sensory neurons, and this includes their segregation into the subepidermal space (Hartenstein and Posakony, submitted).

The mechanism that causes the presumptive sensory neurons to segregate is unknown. It is possible that it is identical with the mechanism controlling neuroblast segregation in the early embryo. Here, during early gastrulation, undivided cells leave the ectodermal layer and move interiorly. It has been observed that, during this process, the cells take on the shape of a flask (Poulson, 1950). This might reflect an active narrowing of the apical parts of the cells (the neck of the flask) by means of contraction of certain cytoskeletal elements. Following this change in shape, the cells could be ‘squeezed’ out of the ectodermal layer by pressure exerted by the surrounding, actively dividing tissue.

Spatially ordered cell proliferation patterns: Relationship to cellular commitment

During the period when the notum microchaete precursors are dividing (between 14 h and 18 h apf), a substantial fraction of the cells in the notum, wing, and head proliferate. This proliferation seems to be spatially regulated: Single cells, or small clusters of cells, which divide simultaneously are distributed in a non-random, regular pattern. The microchaete precursors are embedded in some of these clusters of dividing cells; that is, variable numbers of epidermal cells surrounding the microchaete precursors divide at the same time (in the dorsocentral region of the notum, the microchaete precursors are the only cells to divide at this time). Thus, one cell in an individual cluster of notum cells may give rise to the cells of a microchaete, whereas the other cells in this cluster produce epidermal cells. In other regions (e.g., the scutellum, pleura, and parts of the head capsule), none of the cells in the dividing clusters will produce a microchaete.

It is evident that this spatially ordered division pattern does not reflect any particular state of commitment of the cells within the clusters versus those without. Both groups of cells will give rise to epidermal cells which are identical in every detectable respect. Thus, it may be inferred that there exists a ‘cell proliferation patterning mechanism’ in the imaginal disc which by itself does not confer a particular state of commitment but instead controls the spatial organization of the cells which divide at any given time. It is possible that this mechanism plays a role in establishing the pattern of sensilla. Thus, for example, the spatially ordered proliferation pattern might constitute an intermediate system of positional information to which other mechanisms, more specifically involved in determining the position of sensilla, refer.

Timing of the sensillum precursor divisions

Our results concerning the temporal pattern of cell division in the wing imaginal disc generally agree with observations made in previous studies. Stumpf (1956) reported that mitotic figures are no longer apparent in the disc at 24 h apf. Similarly, Garcia-Bellido and Merriam (1971a) concluded that cell division in the wing disc ceased by approximately 24 h apf, based upon the failure to observe marked cell clones generated by somatic crossing-over after this time.

We have shown that the precursors of the sensilla of the wing and notum divide during two proliferatory phases which affect the wing disc as a whole. The precursors of the multiply innervated bristles of the wing margin divide before and shortly after puparium formation; from the time at which they commence differentiation, it must be inferred that the precursors of the notum macrochaetes and of the wing campaniform sensilla also divide at approximately this time. The precursors of the notum microchaetes and of the singly innervated wing marginal bristles divide approximately 10 h later. A similar pattern of sensillum precursor division and subsequent sensillum differentiation seems to occur in the Drosophila embryo. Here, the multiply innervated sensilla (basiconical sensilla and certain trichoid and campaniform sensilla) appear first, followed by the singly innervated sensilla (Hartenstein, 1988). Correspondingly, the precursors of the former divide earlier than those of the latter (R. Bodmer, personal communication).

Poodry (1975), on the basis of a change in radiosensitivity, suggested a pattern of terminal divisions of macro- and microchaete precursors in substantial agreement with our results, when corrected for temperature. Our direct observations also confirm the conclusion of Garcia-Bellido and Merriam (19716), using mosaic analysis, that the division separating the trichogen and tormogen of the microchaetes can occur as late as 18 h apf.

Further results of the mosaic studies of Garcia-Bellido and Merriam (1971b) suggest that the microchaete precursors become committed by 40 h before puparium formation. After this time, no somatic clones including both sensilla and epidermal cells could be obtained, indicating that the last division of a cell producing both structures must take place before this time (barring the death of progeny cells). Combined with our result that the two final microchaete precursor cell divisions commence at approximately 14 h apf, this would imply that the microchaete precursors remain undivided for 54 h during the third larval instar and early pupal stage, while other disc cells carry out 4-5 rounds of mitosis. We are currently testing this prediction by applying BrdU to third instar larvae. If indeed microchaete precursors are mitotically quiescent over the entire predicted period, they should not incorporate BrdU and, consequently, their progeny should not be labeled.

We are especially grateful to Dr Seymour Benzer and the members of his laboratory for their continuing generosity in providing monoclonal antibody reagents.

We thank Dr Hilary Ellis and Michael Leviten for critical reading of the manuscript and for helpful discussions. This work was supported by a postdoctoral fellowship (Ha 1445-1) from the Deutsche Forschungsgemeinschaft to V.H. and a Pew Scholars award to J.W.P.

Bryant
,
P. J.
(
1978
).
Pattern formation in imaginal discs
. In
The Genetics and Biology of Drosophila, vol. 2c
(eds.
M.
Ashburner
&
T.R.F.
Wright
), pp.
229
335
.
New York
:
Academic Press
.
Canal
,
I.
&
Ferros
,
A.
(
1986
).
The pattern of early neuronal differentiation in Drosophila melanogaster
.
J. Neurogen
.
3
,
293
319
.
Clever
,
U.
(
1960
).
Der Einfluss der Sinneszellen auf die Borstenentwicklung bei Galleria tnellonella L
.
Wilhelm Roux Arch. EntwMech. Org
.
152
,
137
159
.
Cole
,
E. S.
&
Palka
,
J.
(
1982
).
The pattern of campaniform sensilla on the wing and haltere of Drosophila melanogaster and several of its homeotic mutants
.
J. Embryol. exp. Morph
.
71
,
41
61
.
Edgar
,
B. A.
&
O’farrell
,
P. H.
(
1989
).
Genetic control of cell division patterns in the Drosophila embryo
.
Cell
57
,
177
187
.
Ferris
,
G. F.
(
1950
).
External morphology of the adult
. In
Biology of Drosophila
(ed.
M.
Demerec
), pp.
368
419
.
New York
:
Hafner Publishing Co
.
Fujita
,
S. C.
,
Zipursky
,
S. L.
,
Benzer
,
S.
,
Ferrus
,
A.
&
Shotwell
,
S. L.
(
1982
).
Monoclonal antibodies against the Drosophila nervous system
.
Proc. natn. Acad. Sci. U.S.A
.
79
,
7929
7933
.
Garcia-Bellido
,
A.
&
Merriam
,
J. R.
(
1971a
).
Parameters of the wing disc development of Drosophila melanogaster
.
Devi Biol
.
24
,
61
87
.
Garcia-Bellido
,
A.
&
Merriam
,
J. R.
(
1971b
).
Genetic analysis of cell heredity in imaginal discs of Drosophila melanogaster
.
Proc, natn. Acad. Sci. U.S.A
.
68
,
2222
2226
.
Hartenstein
,
V.
(
1988
).
Development of Drosophila larval sensory organs: spatiotemporal pattern of sensory neurones, peripheral axonal pathways and sensilla differentiation
.
Development
102
,
869
886
.
Henke
,
K.
(
1953
).
Ueber Zelldifferenzierung im Integument der Insekten und ihre Bedingungen
.
J. Embryol. exp. Morph
.
1
,
217
226
.
Hodgkin
,
N. M.
&
Bryant
,
P. J.
(
1978
).
Scanning electron microscopy of the adult of Drosophila melanogaster
. In
The Genetics and Biology of Drosophila, vol. 2c
(ed.
M.
Ashburner
&
T.R.F.
Wright
), pp.
337
358
.
New York
:
Academic Press
.
Kankel
,
D. R.
,
Ferrus
,
A.
,
Garen
,
S. H.
,
Harte
,
P. J.
&
Lewis
,
P. E.
(
1980
).
The structure and development of the nervous system
. In
The Genetics and Biology of Drosophila
, vol.
2d
(ed.
M.
Ashburner
&
T.R.F.
Wright
), pp.
295
368
.
Keil
,
T. A.
&
Steinbrecht
,
R. A.
(
1986
).
Mechanosensitive and olfactory sensilla of insects
. In
Insect Ultastructure
, vol.
2
(ed.
R.C.
King
&
H.
Akai
), pp.
477
516
.
New York
:
Plenum Press
.
Lawrence
,
P. A.
(
1966
).
Development and determination of hairs and bristles in the milkweed bug, Oncopeltus fasciatus (Lygaeidae, Hemiptera)
.
J. Cell Sci
.
1
,
475
498
.
Lees
,
A. D.
&
Waddington
,
C. H.
(
1942
).
The development of the bristles in normal and some mutant types of Drosophila melanogaster
.
Proc. Roy. Soc. Ser. B
131
,
87
110
.
Madhavan
,
M. M.
&
Madhavan
,
K.
(
1980
).
Morphogenesis of the epidermis of adult abdomen of Drosophila
.
J. Embryol. exp. Morph
.
60
,
1
31
.
Miller
,
A.
(
1950
).
The internal anatomy and histology of the imago of Drosophila melanogaster
. In
Biology of Drosophila
(ed.
M.
Demerec
), pp.
420
534
.
New York
:
Hafner Publishing Co
.
Murray
,
M. A.
,
Schubiger
,
M.
&
Palka
,
J.
(
1984
).
Neuron differentiation and axon growth in the developing wing of Drosophila melanogaster
.
Devi Biol
.
104
,
259
273
.
Palka
,
J.
,
Lawrence
,
P. A.
&
Hart
,
H. S.
(
1979
).
Neural projection patterns from homeotic tissue of Drosophila studied in bithorax mutants and mosaics
.
Devi Biol
.
69
,
549
575
.
Peters
,
W.
(
1965
).
Die Sinnesorgane an den Labellen von Calliphora erythrocephala Mg. (DIptera)
.
Z. Morph. Ökol. Tiere
55
,
259
320
.
Poodry
,
C. A.
(
1975
).
A temporal pattern in the development of sensory bristles in Drosophila. Wilhelm Roux Arch
.
EntwMech. Org
.
178
,
203
213
.
Poulson
,
D. F.
(
1950
).
Histogenesis, organogenesis, and differentiation in the embryo of Drosophila melanogaster Meigen
. In
Biology of Drosophila
(ed.
M.
Demerec
), pp.
168
274
.
New York
:
Hafner Publishing Co
.
Reynolds
,
E. S.
(
1963
).
The use of lead citrate at high pH as an electron-opaque stain in electronmicroscopy
.
J. Cell Biol
.
17
,
208
212
.
Schmidt
,
K.
&
Gnatzy
,
W.
(
1971
).
Die Feinstruktur der Sinneshaare auf den Cerci von Gryllus bimaculatus Deg. (Saltatoria, Gryllidae). II. Die Hautung der Faden- und Keulenhaare
.
Z. Zellforsch. mikrosk. Anat
.
122
,
210
226
.
Schubiger
,
M.
&
Palka
,
J.
(
1987
).
Changing spatial patterns of DNA replication in the developing wing of Drosophila
.
Devi Biol
.
123
,
145
153
.
Shankland
,
M.
&
Bentley
,
D.
(
1983
).
Sensory receptor differentiation and axonal pathflnding in the cercus of the grasshopper embryo
.
Devi Biol
.
97
,
468
482
.
Stumpf
,
H.
(
1956
).
Die richtungen der Teilungsspindeln auf dem Puppenfliigel von Drosophila im verlanfe der mitosenperiode
.
Biol. Zbl
.
75
,
17
27
.
Tomlinson
,
A.
&
Ready
,
D. F.
(
1987
).
Neuronal differentiation in the Drosophila ommatidium
.
Devi Biol
.
120
,
366
376
.
Zacharuk
,
R. Y.
(
1985
).
Antennae and sensilla
. In
Comprehensive Insect Physiology, Biochemistry and Pharmacology, vol. 6
(ed.
G.A.
Kerkut
&
L.I.
Gilbert
), pp.
1
69
.
Oxford
:
Pergamon Press
.
Zipursky
,
S. L.
,
Venkatesh
,
T. R.
,
Teplow
,
D. B.
&
Benzer
,
S.
(
1984
).
Neuronal development in the Drosophila retina: Monoclonal antibodies as molecular probes
.
Cell
36
.
15
26
.