THE JOURNAL OF COMPARATIVE NEUROLOGY 417:349 –365 (2000)

Proliferation and Programmed Cell Death of Neuronal Precursors in the Mushroom Bodies of the Honeybee ¨ FER, AND DAGMAR MALUN OLGA GANESHINA,* SABINE SCHA Institut fu¨r Neurobiologie, Freie Universita¨t Berlin, 14195 Berlin, Germany

ABSTRACT We have studied proliferation and programmed cell death in the brain of the honeybee during metamorphosis. DNA fragmentation detection using the TUNEL method combined with 5-bromodeoxyuridine incorporation experiments reveal that in the mushroom bodies neurogenesis is terminated by extensive apoptosis. Proliferation of mushroom body neuroblasts is active until the fourth day of pupal development, ceasing abruptly within 1 day after the onset of apoptosis in the mushroom body proliferative clusters. Inside the mushroom bodies, apoptosis spreads from the apical ends of proliferative clusters, beneath the brain’s surface, toward the basal ones. The distributions of apoptotic cells and those in the S phase of the cell cycle overlap significantly. Electron microscopic analysis gives further evidence that mushroom body neuroblasts themselves undergo programmed cell death. We suggest that programmed cell death may be the main factor controlling the final number of Kenyon cells produced during metamorphosis. The overlap in time and space between proliferation and apoptosis raises the question of whether the neuronal precursors switch to programmed cell death during the progression of the cell cycle, or afterwards. J. Comp. Neurol. 417: 349 –365, 2000. © 2000 Wiley-Liss, Inc. Indexing terms: neurogenesis; insect neuroblasts; apoptosis; fluorescence anti-BrdU immunocytochemistry; electron microscopy

The proper establishment of mature structures in the adult animal requires precise coordination of the three main developmental processes—proliferation, differentiation, and programmed cell death (PCD). It is known that proliferation and PCD occur simultaneously in various differentiating tissues (Kallen, 1965; Holman et al., 1996). In the development of both vertebrate and invertebrate nervous systems, PCD of mature and differentiating neurons has been clearly demonstrated and extensively studied (for reviews see Truman, 1984; Truman et al., 1992; recent studies: D’Mello, 1998; Caldero´ et al., 1998), whereas only few studies report PCD in neuronal precursors (Nordlander and Edwards, 1969; Carr and Simpson, 1981; Monsma and Booker, 1996; Champlin and Truman, 1998). On the other hand, very recent studies on mammals and birds indicate that cell death of neuronal precursors may be common in the development of the central nervous system (Blaschke et al., 1998; Diaz et al., 1999). For the development of the insect brain, the degeneration of neuroblasts has been described for the monarch butterfly, Danaus plexippus, and the moth Manduca sexta (Nordlander and Edwards, 1969; Booker and Truman, 1987; Monsma and Booker, 1996). Furthermore, it has © 2000 WILEY-LISS, INC.

been shown that the timing of extensive PCD in the outer optical anlage of Manduca coincides with a gradual decrease in the number of cells replicating DNA, i.e., those passing through the S phase of the cell cycle (Champlin and Truman, 1998). It is not yet known, however, whether the direct elimination of neuronal precursors by PCD is a common mechanism by which neurogenesis is terminated. Here we describe PCD of neuronal precursors in the mushroom bodies (MBs) of the honeybee. The MBs are paired structures in the insect brain involved in the processing of sensory information and in functions related to learning and memory (Davis, 1993; Menzel et al., 1994; Heisenberg, 1998). Drosophila mutants with structural or biochemical abnormalities in the MBs are deficient in olfactory learning (Heisenberg et al., 1985; Nighorn et al.,

Grant sponsor: GRK; Grant number: 120; Grant sponsor: DFG; Grant number: Ma 1449/3-1; Grant number: SFB 515 C7. * Correspondence to: Dr. Olga Ganeshina, Institut fu¨r Neurobiologie, Freie Universita¨t Berlin, Ko¨nigin-Luise-Str. 28-30, 14195 Berlin, Germany. E-mail: [email protected] Received 11 May 1999; Revised 15 October 1999; Accepted 25 October 1999.

350

O. GANESHINA ET AL.

1991; Han et al., 1992). Intrinsic neurons of the MBs, the Kenyon cells, arise from MB neuroblasts. Chemical ablation of neuroblasts during neurogenesis with hydroxyurea (Truman and Booker, 1986) was sucessfully applied to MBs (de Belle and Heisenberg, 1994; Malun, 1998). Flies with chemically ablated MB neuroblasts fail at olfactory conditioning (de Belle and Heisenberg, 1994). The honeybee is a common model for investigating mechanisms of learning and memory (Menzel and Mu¨ller, 1996). The study of neurogenesis in MBs of this insect is especially important, because it may help us develop new experimental approaches for manipulation with the precursors of the MB intrinsic neurons. Neurogenesis in the MBs of different insect species varies both in the arrangement of neuronal precursors and in timing. In Diptera, Lepidoptera, and Heteroptera, large solitary neuroblasts, dividing asymmetrically, produce ganglion mother cells, which, in turn, divide once to produce neurons. In Hymenoptera (including the honeybee), Blattoidea, Orthoptera, Odonata, and some others, MB neuroblasts form compact groups, referred to as proliferative clusters (Panov, 1957). Solitary neuroblasts are easily recognizable by their stereotyped position and large size, whereas ganglion mother cells and neurons arising from individual neuroblasts form distinctive columns. In contrast, neuroblasts in MB proliferative clusters are only slightly bigger than their progeny, which surround the cluster and do not form columns. In the honeybee this makes it difficult to determine exactly the mode of neuroblast divisions. In general, neurogenesis in MBs outlasts that in other parts of the CNS. Starting from embryonic or early postembryonic stages, it can continue until late metamorphosis, as in Drosophila, or even in the adult, as in the cricket Achaeta domesticus (Panov, 1957; Ito and Hotta, 1992; Cayre et al., 1996). With the honeybee, Panov (1957) observed mitotic figures in MB neuroblasts up to the fourth day of pupal development and noticed signs of degeneration among the neuroblasts on the sixth day. Using the BrdU technique, Fahrbach et al. (1995) demonstrated that no proliferation of MB neuroblasts can be detected in the honeybee brain immediately after adult eclosion. The goal of the present study was to visualize proliferation and PCD with specific markers and to determine spatial and temporal relationship between the two processes in neurogenesis. Our data can provide a basis for further investigations of the mechanisms regulating neurogenesis. Incorporation of 5-bromodeoxyuridine (BrdU) combined with fluorescence immunocytochemistry (Gratzner, 1982) was used to identify proliferating cells, and, because most developmentally regulated neuronal death occurs via apoptosis (Kerr et al., 1972), we used the TUNEL method for specifically labelling apoptotic cells (Gavrielli et al., 1992) to identify PCD in the proliferative clusters. Some of the results have already been published in abstract form (Ganeshina et al., 1998).

TABLE 1. Honeybee Developmental Stages Stages

Features

P0/1 P2 P3 P4 P4 “late”

White body, white eyes White body, light brown eyes White body, brown eyes White body, dark brown eyes First signs of pigmentation in the wing bases or in the tibiotarsal joints of the first leg pair; rest of the body is white; eyes are very dark brown, nearly black Tibiotarsal joints of two first leg pairs are slightly pigmented; joints of the third leg pair are absolutely white; rest of the body is slightly yellowish; eyes are nearly black; MB proliferative clusters are shrunken by one-third to one-half relative to their maximal length Tibiotarsal joints of two first leg pairs are pigmented; joints of the third pair are either white or slightly pigmented; eyes are nearly black; MB proliferative clusters are shrunken by two-thirds to three-fourths relative to their maximal length Body and legs are yellow; tibiotarsal joints of all three leg pairs are distinctly pigmented, claws are brown; eyes are nearly black Head and thorax are light brown; the abdomen and legs are yellow; claws are dark brown; leg joints, wing bases, and legs have a pointed pattern; eyes are black; wings are light grey and folded Head and thorax are dark grey; abdomen and legs are light brown; leg joints, wing bases, and claws are dark brown; wings grey and folded The body and legs are dark brown; wings are straightened

P5 “early”

P5

P6 P7

P8

P9

Determination of the honeybee worker pupa age, adapted with permission from Eichmu¨ller (1994). In the honeybee worker larval development takes 6 days from larval hatching to the 6th larval day. After the prepupal period (two days), pupal development is completed within 10 days from pupal ecdysis (PO) to adult emergence (P9).

The stages P0 –P1g corresponded to subsequent days of pupal development. On days 4 and 5, pupae were subdivided into four groups: P4, P4 “late”, P5 “early”, and P5. To obtain homogeneous groups of animals we prepared a set of easily distinguishable color samples in the range from yellow to very deep brown to compare the pigmentation of eyes and joints. In addition, for the fine subdivision into P5 “early” and P5 we used relative size of the MB proliferative clusters. The work was done according to regulations and guidelines for animal experiments set out by the current German legislation.

General light and electron microscopy Brains were dissected and subsequently fixed for 1–1.5 hours at 20°C with 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.3) or with a mixture of 2.5% glutaraldehyde and 1.5% formaldehyde in the same buffer (pH 7.3), containing 0.1% CaCl2. After rinsing in buffer and postfixation in 1% OsO4 in the same buffer for 1 hour, the specimens were rinsed again, dehydrated, and embedded in Epon 812. Some of the specimens were stained en bloc with 0.5% uranyl acetate in 70% alcohol during dehydration. For light microscopy, frontal 2– 4 ␮m sections were mounted on slides and stained with toluidine blue (Trump et al., 1961). For electron microscopy, thin sections were cut with a diamond knife, picked up on mesh grids, and stained with 0.4% lead citrate. All grids were viewed on a Philips 208 transmission electron microscope.

BrdU labelling MATERIALS AND METHODS Animals Honeybee (Apis mellifera carnica) worker pupae were taken from the hive and staged according to Eichmu¨ller (1994). Pigmentations of eyes, joints, and legs were used as criteria for identifying developmental stages (Table 1).

To study the pattern of proliferation in the MBs during pupal development, we examined BrdU incorporation into replicating nuclei of proliferating cells during the S phase of the cell cycle, visualizing the incorporation with the anti-BrdU fluorescence immunocytochemistry (Gratzner, 1982). Pupae were injected into the neck with 3 ␮l of 6.25% BrdU (Amersham) dissolved in bee saline (130 mM

PROLIFERATION AND CELL DEATH IN MUSHROOM BODIES NaCl, 6 mM KCl, 2 mM MgCl2, 7 mM CaCl2, 160 mM glucose, 10 mM Hepes at pH 6.7) and subsequently kept in an incubator (33°C) for 40 minutes or 1, 2, or 5 hours. Probably because of a diffusion barrier for BrdU, incorporations were achieved only when the time between injection and chemical fixation of the animal was at least 1 hour. Five hour intervals between BrdU injections and fixation, which we referred here to as 5 hours BrdU incorporation, were used to increase the number of incorporations and to follow the pattern of proliferation in MBs during metamorphosis on the day time scale (see below). One or two hour posttreatment survival periods were used to follow the pattern of proliferation during a very limited apoptotic phase and to match it precisely with the temporal characteristics of apoptosis. These short BrdU pulses allowed us to follow the pattern of proliferation during stages, which differed from each other only by several hours, and minimized the effect of experimental intervention. Immediately after decapitation, the heads were fixed overnight in 4% formaldehyde in phosphate-buffered saline (PBS), pH 7.3, at 4°C. After several rinses in PBS, brains were removed from the head capsule, dehydrated, and embedded in Paraplast. Frontal serial sections (4 or 7 ␮m) were cut and mounted on polylysine-coated slides. After dewaxing and rehydration, sections were treated with 2 N HCl to denature DNA, washed in PBST (PBS containing 0.1% Triton X-100), pH 7.3, and preincubated in blocking solution (PBST containing 5% normal goat serum and 0.5% bovine serum albumin) for 1 hour. Subsequently, sections were incubated with primary monoclonal anti-BrdU antiserum (Amersham, Arlington Heights, IL) overnight at 4°C, rinsed with PBST, and incubated for 1 hour with CY3-coupled secondary antibodies (Dianova, Hamburg, Germany; 1:100, in blocking solution). To evaluate the time course of proliferation in the MBs, we counted the number of profiles labelled with BrdU in the proliferative clusters after 5 hours of BrdU incorporation on 7 ␮m serial sections. Counts were done at the P2, P3, P4, and P5 stages. Because the same nuclei could be counted on adjacent sections, the number of labelled profiles exceeded the number of labelled nuclei. Generally, the number of profiles is proportional to the number of nuclei, with a coefficient of proportionality dependent on thickness of sections, size, shape, and orientation of nuclei. Correction for overestimating can be made if the nuclei have a uniform size and a round shape (Abecrombie, 1946; Konigsmark, 1970). In practice, however, such assumptions are not valid, and the calculations may lead to erroneous results (Coggeshall and Lekan, 1996). Because in the context of the present study the absolute numbers of labelled nuclei were less important than their temporal profile and anatomical distribution, we reported the uncorrected number of profiles rather than number of labelled nuclei. We did not observe changes in the distribution of the size, shape, and orientation of the nuclei at the stages we studied. We believe that the reported changes in the number of profiles reflect changes of the number of BrdU incorporations. Control specimens, which were treated as described above except that injection of BrdU was omitted or BrdU were substituted by Ringer solution, did not show a label in MBs. Also omitting the treatment with primary antibodies during the BrdU immunostaining resulted in the absence of the labelling.

351

Apoptosis detection To visualize cells undergoing PCD, we applied the TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labelling) method. The tissue was fixed, washed, embedded in Paraplast, serially sectioned, and mounted as described above. After dewaxing and rehydration, sections were postfixed in 4% formaldehyde (in PBS) for 15 minutes. Following thorough rinsing in PBS, proteinase K (10 ␮g/ml) was applied to the sections for 8 minutes at 20°C. Subsequent treatment of the tissue was performed according to the protocol provided with the DNA Fragmentation Detection Kit (Amersham-Promega). Briefly, the sections were rinsed in PBS and treated with equilibrium buffer for 20 minutes in a humidified chamber. After removal of excess buffer, the sections were exposed to the incubation buffer containing TdT enzyme for 1 hour in a humidified chamber at 37°C. Subsequently, the sections were treated with the stop/wash solution, rinsed in PBS, and finally mounted with Slow Fade (Molecular Probes, Eugene, OR) under glass coverslips. For negative control, TdT enzyme was substituted by distilled water.

Labelling of proliferating and apoptotic cells on adjacent sections To match the distribution of apoptotic and proliferating cells precisely, we selected P4, P4 “later”, and P5 “early” animals (two individuals each). After BrdU injection and incubation of the animals, the brains were dissected, fixed, and treated as above. Serial Paraplast sections (4 ␮m) were cut, and each ribbon was carefully divided into short (three or four sections) fragments using a stereomicroscope. The fragments were alternatively mounted on two slides, and one slide was treated for BrdU, the other for TUNEL labelling. Thus we obtained adjacent sections that revealed cell proliferation and apoptosis, respectively. The sections were viewed using a confocal microscope Leica TCS-4D with a krypton/argon laser. The final figures of the micrographs were assembled with Photoshop (version 2.5) on a Macintosh computer.

RESULTS General structure of the proliferative clusters and their pattern of proliferation The paired honeybee MBs consist of two cup-like calyces (input regions for sensory information), connected via peduncles to an ␣-lobe and a ␤-lobe (output neuropiles) (Mobbs, 1982). In the adult, most of the somata of the MB intrinsic neurons, the Kenyon cells, lie within the calycal cups, but a smaller population also surrounds the calycal neuropiles. The four MB proliferative clusters, one for each calyx, can already be identified in the second larval stage (Malun, 1998). In the pupal brain they lie on top of the peduncle within the growing calycal cups, which, as in the adult, are elongated in an anterior–posterior direction. Figure 1 shows a frontal section through the bee brain on the fourth day of pupal development (P4 “late” stage). The proliferative clusters of the medial calyces are easily recognizable inside the calycal cups. The proliferative clusters of the lateral calyces are not visible, because they are located posterior to the section. The neuroblasts have nu-

352

O. GANESHINA ET AL.

Fig. 1. Frontal section through the bee brain on day 4 of pupal development (P4 “late”). Proliferative clusters of medial calyces are marked with asterisks. KC, Kenyon cell somata; MC, medial calyx; LC, lateral calyx; arrows mark degenerating cells. Scale bar ⫽ 200 ␮m.

clei about 7–9 ␮m in diameter and can be readily distinguished from much smaller Kenyon cells, which from early larval stages are continuously shed by the proliferative clusters (Panov, 1957; Malun, 1998) (Figs. 1, 2c, 3a,c). Small larval Kenyon cells with 4 ␮m nuclei are located outside the calycal cups, whereas the larger ones with 7 ␮m nuclei corresponding to a younger generation are located above and within the calycal cups (Panov, 1957). The youngest Kenyon cells with 4 ␮m nuclei, which start appearing during day 2 or 3 of pupal development (Panov, 1957), flank the proliferative cluster (Fig. 3a,c). They form the median group of Kenyon cells in the adult (Panov, 1957). Within the proliferative clusters a few smaller (3– 6 ␮m) nucleus profiles containing nucleoli are scattered among 7–9 ␮m profiles. BrdU incorporation experiments revealed that from day 2 of pupal development the MBs were the only structures in the bee brain undergoing extensive proliferation (Fig. 2a,b). Until P4, most cells within the cluster were labelled after 5 hours of BrdU incorporation (Fig. 2b), suggesting that at these stages the clusters consisted mainly of proliferating cells. Also, there was no obvious gradient in the distribution of BrdU-labelled nuclei along the proliferative clusters, either in the anterior–posterior or in the basal–apical axis (Fig. 2b). One hour of BrdU incorpora-

tion at P4 also showed the even distribution of BrdUlabelled nuclei (see Fig. 5a), indicating approximately the same rate of proliferation throughout the cluster. In the P4 “late” pupa, a gradient in the distribution of labelled nuclei became evident after 2 hours of BrdU incorporation, with a higher density of incorporations in the basal part of the cluster, close to the peduncle, and fewer profiles in the apical part closer to the brain surface (Fig. 3b). At the same stage degenerating pycnotic cells appeared, which occupied predominantly the apical part of the cluster (Figs. 1, 3a). On P5 “early” the proliferative clusters became smaller, the number of pycnotic profiles increased (Fig. 3c), and only few BrdU-labelled nuclei were localized mainly on the bottoms of the clusters (Fig. 3d). By day 6 of pupal development, the proliferative clusters disappeared.

Identification of apoptosis in the MB proliferative clusters Apoptosis is characterized by a common intracellular molecular machinery (Kerr et al., 1972), including the fragmentation of nuclear DNA into short olygonucleotides. We used the TUNEL method (Gavrielli et al., 1992) to detect fragmented DNA by enzymatic attachment of labelled nucleotides to the free ends of the fragments.

PROLIFERATION AND CELL DEATH IN MUSHROOM BODIES

353

During metamorphosis, extensive apoptosis in the MBs could be detected only within a narrow time window from day 4 to day 6 of pupal development and was restricted to the proliferative clusters (Fig. 4). On P4, the first apoptotic nuclei appeared at the apical ends of the clusters (Fig. 5b). During the following hours (P4 “late” and P5 “early”) apoptosis spread over the cluster from its apical tip to the basal region (Fig. 4b,c) with a tendency for apoptotic nuclei to concentrate at the boundary with the surrounding Kenyon cell bodies. At the same time the proliferative clusters shrunk from the apical end and lost contact with the brain surface. Serial sections revealed that they also shrunk along their anterior–posterior extension. During day 5 of pupal development, the remnants of the apoptotic cells accumulated on apical end of the vanishing cluster to form a triangular structure, which could also be seen with Nomarski optics in unlabelled preparations (data not shown). On P6 a few apoptotic profiles marked the former position of proliferative clusters at the bottom of the calycal cups. Thus extensive apoptosis in the MB proliferative clusters started at about the time when proliferation began to cease.

Spatial and temporal relationship between proliferation and apoptosis To see both processes in the same preparations, we performed labelling of apoptotic and proliferating cells on adjacent Paraplast sections and were thus able to match the distributions of proliferating and apoptotic cells in the same animal at the same time. At the onset of apoptosis (P4), when only a few profiles of apoptotic cells could be detected per individual (Fig. 5b), profiles of BrdU-labelled nuclei were evenly distributed throughout the cluster (Fig. 5a). As the number of apoptotic nuclei increased, and as they spread throughout the proliferative cluster (Fig. 5d,f), the distribution of BrdU incorporations became uneven, with a tendency for labelled nuclei to concentrate at the bottom of the cluster (Fig. 5c,e). Significant overlap was evident, however, in both distributions. To describe the temporal aspects of cell proliferation, we counted the number of labelled profiles per cluster after 5 hours of BrdU incorporation at P2, P3, P4, and P5. Generally the number of profiles is proportional to the number of BrdU-labelled nuclei (see Materials and Methods), which, in turn, depends on the total number of proliferating cells, on the rate of proliferation, and on the duration of the S phase of cell cycle. When incorporation time is longer than the duration of cell cycle, the number of labelled cells is independent of the duration

Fig. 2. Distribution of BrdU-labelled nuclei (5 hours incorporation) before the onset of apoptosis and the structure of the MB proliferative cluster at these stages. a: BrdU labelling on day 2 of pupal development (P2). Four proliferative clusters are the only brain structures incorporating BrdU. b: BrdU labelling on day 3 of pupal development (P3). Proliferative clusters of the medial calyces are visible on the section. Most of the nuclei in the clusters are labelled. c: A proliferative cluster on day 3 of pupal development (P3); toluidine blue staining. Neuroblasts are easily distinguishable from the smaller postmitotic neurons (Kenyon cells) surrounding the cluster. No degenerating cells are in the cluster. Neuroblasts are marked with large arrows; small profiles scattered among neuroblasts are marked with small arrows. KC, Kenyon cell somata; CN, calycal neuropile. Scales bars ⫽ 200 ␮m in a; 100 ␮m in b; 50 ␮m in c.

354

O. GANESHINA ET AL.

Fig. 3. Structure of MB proliferative clusters after the onset of apoptosis, toluidine blue staining (left panels) and the distribution of BrdU-labelled nuclei at these stages after 2 hours of incorporation of BrdU (right panels). a,b: P4 “late” stage; a few degenerating pycnotic cells in the apical part of the cluster are marked with arrows. Note bundles of Kenyon cell axons (arrowheads), which run down towards the peduncle along the border of the proliferative cluster and among the youngest neurons (median group of Kenyon cells) that flank it; a

gradient in the distribution of BrdU-labelled nuclei (b) along their basoapical axis is evident. c,d: P5 “early” stage. Proliferative clusters are shrunken along their basoapical axis and contain more degenerating cells; a few BrdU-labelled nuclei are located at the bottom of the proliferative clusters. PC, proliferative cluster; KC, Kenyon cells; CN, calycal neuropile; P, peduncle. Scales bars ⫽ 50 ␮m in a,c; 100 ␮m in b,d.

of S phase and reflects both the total number of proliferating cells and the rate of their proliferation. The cell cycle duration of the neuroblasts in Drosophila brain has been estimated to be about 1–2 hours during the active phase of neurogenesis (Truman and Bate, 1988; Ito and Hotta, 1992), and in MBs it did not exceed 5 hours at the end of neurogenesis (Ito and Hotta, 1992). Therefore, we used the number of labelled profiles after 5 hours BrdU incorporation to characterize the proliferation activity in MBs. Until P4 proliferation was active, nearly 1,000 labelled profiles per cluster were counted, with only a small ten-

dency to decrease (Fig. 6a). After the onset of apoptosis (P4) proliferation ceased abruptly within 1 day. Given that the fragmentation of DNA is considered to be a late event in apoptosis, the onset of apoptosis (or commitment to it) could be several hours prior to DNA fragmentation (for review see D’Mello, 1998). Therefore, in reality, proliferation and apoptosis overlapped earlier than we could detect. Thus apoptosis in MB proliferative clusters started 1 day before the proliferation of neuronal precursors stopped. During this period the distributions of cells still undergoing DNA replication, and those commenc-

PROLIFERATION AND CELL DEATH IN MUSHROOM BODIES

355

ing apoptosis significantly overlapped within the clusters.

Do only neuroblasts proliferate and undergo PCD in proliferative clusters? Starting at P4, there was an increasing number of cells in the MB proliferative clusters that did not incorporate BrdU and underwent PCD. At P3–P4 within the proliferative clusters the smaller nucleus profiles were scattered among the larger ones on 2– 4 ␮m toluidine blue-stained sections (Figs. 2c, 3a,b). They could correspond either to superficial aspects of neuroblasts or to smaller cells. Insofar as the mode of the MB neuroblast divisions is not yet established for the honeybee, one cannot exclude the possibility that these smaller profiles represent neuroblasts just after symmetric division or ganglion mother cells and/or youngest Kenyon cells at the start of their differentiation (in the case of an asymmetric division mode). Also, in MBs of Periplaneta americana clusters of glial cells that incorporated BrdU were found at the bottom of calycal cups (Cayre et al., 1996). To determine whether the cells within the clusters are exclusively neuronal precursors and whether differentiating neurons or glial cells are also eliminated by PCD there, we studied the ultrastructure of the area before and during the apoptotic phase of MB development. We compared the cell populations within and outside the cluster and on the boundary between the proliferative cluster and the surrounding Kenyon cells. The proliferative cluster. One day before the onset of apoptosis (P3) the cell population within the proliferative cluster appeared to be homogeneous (Fig. 7). The cells were of irregular shape; large nuclei contained dispersed chromatin and had profiles of one to three large nucleoli. The cytoplasm of these cells was of moderate electron density and contained numerous ribosomes and clusters of mitochondria; membranous organelles were represented by the Golgi complex and few tubules of the smooth endoplasmic reticulum. The plasma membranes of adjacent cells were in tight contact, so that there was extremely little intercellular space. Some of the cell bodies showed small finger-like cytoplasmic extrusions that lacked cytoplasmic organelles or cytoskeleton. At P5, “early,” various apoptotic profiles appeared in the proliferative clusters. Figure 8 shows a probable temporal consequence of ultrastructural manifestation of PCD. The presumed early stage was characterized by small membrane-bound spherical dense bodies (diameter 1–2 ␮m), which appeared in the cytoplasm of the neuroblasts (Fig. 8a– c) and contained densely packed ribosomes and several mitochondria. At the second stage, the entire cell shrank considerably, became spherical, and transformed into a single large dense body (Fig. 8b). Subsequently,

Fig. 4. Apoptosis in proliferative clusters revealed with the TUNEL method. a: Apoptosis in the brain of a pupa on P5 “early” stage. MB proliferative clusters of the medial calyces are visible on the section. Apoptotic nuclei are located exclusively within the MB proliferative clusters. Above the MBs nonspecific fluorescence of ocelli is visible. b: The distribution of apoptotic nuclei on P4 “late”; c: on P5 “early” stages. Note the predominant distribution of apoptotic nuclei at the border of proliferative cluster. Scale bars ⫽ 100 ␮m in a; 50 ␮m in b,c.

Figure 5

PROLIFERATION AND CELL DEATH IN MUSHROOM BODIES

Fig. 6. Temporal aspects of proliferation. Abscissa shows the days of pupal development corresponding to stages P2, P3, P4, and P5. a: Number of BrdU-labelled nuclear profiles in a MB proliferative cluster after 5 hours of BrdU incorporation; mean and SD (error bars). N and n denote the number of animals and the number of clusters, respectively. Apoptosis is detected starting from the fourth day (P4); the corresponding columns in the figure are indicated in black. Note the abrupt cessation of proliferation after the onset of apoptosis. b: Ecdysteroid titer as recorded in hemolymph. Redrawn from Feldlaufer and coauthors (1985). In the original drawing, age was given as day postoviposition. Day 9 postoviposition corresponds to the day of pupation; day 11 corresponds to P0. The two curves correspond to two experiments. Note that the peak of the ecdysteroid titer corresponds to the onset of apoptosis.

nuclei with condensed chromatin lost their nuclear envelopes (Fig. 8c), whereas some cytoplasmic organelles— mitochondria, endoplasmic reticulum, and ribosomes— remained unchanged. At a late stage, cell debris formed conglomerates of extremely electron-dense material

Fig. 5. Labelling of apoptotic and BrdU-labelled cells on adjacent sections at the different stages of the apoptotic phase of MB development. The overlap in distributions of apoptotic and BrdU-labelled nuclei is visible. a,c,e: BrdU labelling after 1 hour (a) or 2 hours (c,e) of BrdU incorporation. b,d,f: Labelling of apoptotic cells with the TUNEL method. a,b: Onset of apoptosis (P4); c and d correspond to P4 “late” and e and f correspond to P5 “early” stage. In b nonspecific fluorescence of ocelli is visible above the MBs. Scale bar ⫽ 50 ␮m.

357

(Fig. 8d), observed predominantly at the apical end of the cluster. Cells outside the proliferative cluster. Besides being smaller, the Kenyon cells that surrounded the proliferative clusters differed from neuroblasts in several respects (Fig. 9). They had a more regular, usually oval, shape and a spherical or slightly oval nucleus surrounded by comparatively little cytoplasm with evenly distributed mitochondria. Thin cytoplasmic processes of glial cells, apparently containing glycogen granules, isolated adjacent Kenyon cell bodies from each other (Figs. 9, 10). We encountered only a few glial cell somata (Fig. 10a), so each glial cell must have a huge tree of cytoplasmic processes. A complex mesh of glycogen containing glial cell processes surrounding Kenyon cells has already been described for adult ants (Landolt, 1965). Primary neurites (0.5– 0.7 ␮m), growing from somata (Fig. 10b) ran singly or in bundles among the Kenyon cells (Fig. 9). They contained microtubules and mitochondria and resembled neurites in the calycal neuropile (Fig. 10c). The border of the cluster. Prominent bundles of Kenyon cell axons ran toward the peduncle along the border of the proliferative cluster (Figs. 3a,c, 11a). Here a thin “transition zone” contained very small cell profiles (3–5 ␮m), which could represent newborn neurons, because processes of neurons and the glial cells grew among them (Fig. 11b). At the P5 “early” stage many apoptotic profiles representing early, middle, and late stages of cell death were also observed in this region (Fig. 11b– d). Because the edges of the transitional zone were not clearly defined and newborn Kenyon cells were hardly recognizable in thin sections, it was difficult to determine which cell type underwent PCD there. However, some apoptotic cells were presumed to be newborn neurons because of their shape, pattern of mitochondria distribution, and location close to Kenyon cell somata (Fig. 11c). We occasionally observed late apoptotic profiles being engulfed by glial cell processes (Fig. 11b), indicating that glia may participate in the ultimate removal of apoptotic remnants. Kenyon cell axons running to the peduncle through the transitional zone directly contacted many apoptotic profiles on their way (Fig. 11d). Comparison of the ultrastructure of proliferative clusters, the area of Kenyon cell somata, and the border between them indicated that the first distinctive signs of Kenyon cell differentiation at an EM level were redistribution of mitochondria and growth of neurites. The neurite profiles were characterized by microtubules and mitochondria and resembled neurite profiles in calycal neuropile. We never observed the neurite profiles within proliferative clusters, so that morphological differentiation of the Kenyon cells at an EM level occurred outside the proliferative clusters, most probably starting at their boundary. The glial cells were characterized by long cytoplasmic processes, which were easily recognizable because of the glycogen granules. The glial cells or their processes were never found among neuroblasts within proliferative clusters. Because we could not find any signs characteristic of differentiating neurons or glial cells, we consider the proliferative clusters to consist mostly (or even exclusively) of neuronal precursors. Thus our ultrastructural data provide evidence that the MB proliferative clusters consist mostly of neuronal precursors, which in the middle of pupal development exhibit features typical for apoptosis: chromatin condensation,

358

O. GANESHINA ET AL.

Fig. 7. Ultrastructure of MB proliferative cluster 1 day before the onset of apoptosis (P3). Arrows mark small cytoplasmic processes of neuroblasts. N, nuclei; M, mitochondria; G, Golgi complex; SER, smooth endoplasmic reticulum. Scale bar ⫽ 1 ␮m.

Fig. 8. Apoptosis in the MB proliferative clusters at P5 “early.” a: An early stage of apoptosis in a neuroblast. A spherical membranebound dense body is seen in the cytoplasm. b: Neuroblasts at early and middle stages of apoptosis. Membrane-bound dense bodies in two neuroblasts at the early stage of apoptosis are marked with small

arrows. A neuroblast in the middle stage of apoptosis is marked with a large arrow. c: An apoptotic cell (large arrow) after “dissolution” of the nuclear envelope. d: The late stage of apoptosis. NB, neuroblast; N, nucleus of neuroblast; DB, dense body; Ch, condensed chromatin. Scale bars ⫽ 1 ␮m.

360

O. GANESHINA ET AL.

Fig. 9. Kenyon cell somata inside the calycal cup; P5 “early.” Large arrows mark the processes of glial cells, isolating the neighboring Kenyon cell somata (KC). They are easily recognizable because of abundant glycogen granules (small arrows). The bundles of neurites are marked with arrowheads. Scale bar ⫽ 1 ␮m.

Fig. 10. Ultrastructure of cells outside proliferative clusters on day 5 of pupal development (P5 “early”). a: A glial cell is located among the Kenyon cell somata. Note the cytoplasmic processes, containing glycogen granules (arrows). b: A Kenyon cell soma with a primary process (arrowhead). c: The same primary process; microtu-

bules are easily recognizable. d: A Kenyon cell soma near calycal neuropile. Note the resemblance between neural processes running among Kenyon cell somata (see also Fig. 9) and in the neuropile (arrowheads). KC, Kenyon cell soma; GC, glial cell; CN, calycal neuropile. Scale bars ⫽ 1 ␮m in a,b,d; 0.5 ␮m in c.

Fig. 11. Ultrastructure of the transitional zone at the boundary of proliferative cluster. a: One day before the onset of apoptosis (P3); note the Kenyon cell axons, running toward the peduncle (arrowheads). b: After the onset of apoptosis (P5 “early”). Presumable newborn Kenyon cells are marked with asterisks. Glial sheath, typical for KC, is not yet complete (large arrow). Note the neural processes among the newborn KC (arrowheads). A conglomerate of cell debris

(late stage of apoptosis) is located inside the glial cell, which is identifiable because of distinctive glycogen granules (small arrows). c: Dense body (early signs of apoptosis) in a newborn KC (asterisk). Neural processes are marked with arrowheads. d: Middle stage of apoptosis; note the direct contact between apoptotic cell and neural process. NB, neuroblast; KC, Kenyon cell; DB, dense body; CD, cell debris; GC, glial cell. Scale bars ⫽ 1 ␮m.

PROLIFERATION AND CELL DEATH IN MUSHROOM BODIES dissolution of the nuclear envelope, increased electron density of the cytoplasm where organelles retain a normal morphology, and shrinkage of cells (Kerr et al., 1972). Morphological differentiation of Kenyon cells, recognizable at the EM level, definitely occurs outside the clusters and most probably starts at their boundary. At P4 –P5 not only neuronal precursors but also newborn Kenyon cells possibly undergo PCD in this region.

DISCUSSION We have shown that in the honeybee MBs neurogenesis is very active until day 4 of pupal development, when other brain neuroblasts have already ceased proliferation. The intense and prolonged proliferation is not surprising, because the final number of Kenyon cells in the adult honeybee is extremely high compared to that in most other insect species: about 340,000 or roughly one-third of the entire neuronal population of the brain (Wittho¨ft, 1967). During days 4 and 5 of pupal development the proliferation abruptly ceases from nearly 1,000 proliferating cell profiles within the cluster to zero. The first signs of apoptosis are detected when proliferation is still massive. The clear coincidence in time and space of both PCD and the cessation of proliferation supports the hypothesis that MB neuroblasts are eliminated by switching from proliferation to apoptosis. We suggest that the timing and regulation of this process may be the critical parameter determining the final number of Kenyon cells in the adult. This mechanism may act together with finer regulation of the rate of Kenyon cell production during neurogenesis. It is well known that both proliferation and PCD that occur in a variety of tissues during insect metamorphosis are under strong ecdysteroid control (for review see Truman, 1996; Cayre et al., 1997). In honeybee metamorphosis the ecdysteroid titer rises about five- to sixfold between day 2 and day 4 of pupal development and returns to its previous level by day 6 (Feldlaufer et al., 1985). Our results clearly demonstrate a strong coincidence between the onset of apoptosis and the sharp peak of the ecdysteroid titer (compare Fig. 6a and b). Thus, it is reasonable to propose that ecdysteroid pulse triggers the switch from proliferation to apoptosis during MB development. A similar coincidence has been demonstrated recently for the optical lobe of Manduca sexta (Champlin and Truman, 1998). Experiments in vitro reveal that low and moderate levels of ecdysteroids reversibly sustain proliferation in metamorphic optical lobes, whereas a high level of the hormone triggers extensive apoptosis (Champlin and Truman, 1998). A stereotyped subset of postmitotic mature motoneurons innervating Manduca prolegs has been demonstrated to die in response to prepupal peak of 20hydroecdysone (Weeks and Truman, 1985; Weeks et al., 1992). In contrast, the withdrawal of ecdysteroids is required to induce apoptosis in “doomed” neurons of the ventral cord of Drosophila and Manduca (Truman and Schwartz, 1984; Robinow et al., 1993). If the final elimination of different neuronal precursors in the insect CNS is dependent on ecdysteroids, then possibly each individual case requires a particular ecdysteroid level in the hemolymph. This supposition arises from the lack of temporal correlation between the end of neurogenesis in different parts of the CNS and developmental ecdysteroid pulses in hemolymph (Truman, and Bate, 1988; Bainbridge and Bownes, 1988; Ito and Hotta, 1992; Truman et

363

al., 1994). Other distant and/or local external signals may also act together with ecdysteroids or change the requirements for ecdysteroid concentration. For example, juvenile hormone has been shown to be involved in regulating proliferation and apoptosis both in Manduca optic lobes (Monsma and Booker, 1996) and in neurogenesis of adult MBs of crickets (Cayre et al., 1994; 1996). In addition, locally acting factors can either rescue neurons from PCD or, as with the recently discovered factor neurocidin (Choi and Fahrbach, 1995), induce PCD in specific subsets of neurons. In Drosophila, nitric oxide (NO) has been shown to have an antiproliferative effect (Kuzin et al., 1996) and to be a regulator of cell survival (Nicotera et al., 1997). It is not yet known how NO synthase is expressed in the bee MBs during metamorphosis; in the adult it is restricted to the neuropile (Mu¨ller, 1997). According to our EM observations, apoptotic profiles at the border of proliferative clusters are often situated in the vicinity of or even in direct contact with neurites. Thus, the outgrowing processes of the differentiating Kenyon cells may be a source for a possible local factor affecting cell fate there. Glia may also affect the cell fate (Xiong and Montell, 1995) and have been shown to participate in the removal of apoptotic cells during development of the insect nervous system (Sonnenfeld and Jacobs, 1995). Mushroom body glia have been studied recently in adult honeybees (Ha¨hnlein and Bicker, 1996) and also during pupal development (Ha¨hnlein and Bicker, 1997). Our EM data indicate that the glia may eliminate the remnants of apoptotic cells at the boundary of the proliferative clusters (Fig. 11b). The ultrastructural changes within MB neuroblasts show features typical for apoptosis: shrinkage of cells, chromatin condensation, dilution of nuclear envelope, and fragmentation of cells. We also describe the presence in the cytoplasm of membrane-bound spherical dense bodies, which we interpret as an early manifestation of apoptosis. To our knowledge, such structures have so far not been described in other apoptotic cells. However, tissue-specific differences in cytological characteristics of apoptosis in different tissues are well known (Stocker et al., 1978; Macagno, 1979; for review see Truman, 1984). Taken together, our data provide evidence that neuronal precursors themselves undergo PCD at a time when cell division in the clusters continues. The question of how the switch from proliferation to apoptosis is related to the cell cycle is a general one. It has been recently discovered that both proliferation and apoptosis are tightly coupled by cell cycle regulators such as retinoblastoma protein (pRb) and transcription factor p53, affecting both cell division and cell death (for review see Herwig and Strauss, 1997; King and Cidlowski, 1998). This coupling is believed to be the mechanism that eliminates genetically abnormal cells or cells with aberrations in the machinery of their cell cycle (King and Cidlowski, 1998). The switch to apoptosis may occur during cell cycle progression, or, alternatively, postmitotic cells may undergo PCD. Various agents are known to induce apoptosis in proliferating cells at different phases of the cell cycle (Asano et al., 1996; Hensey and Gautier, 1997; Ninomiya et al., 1997; Ye et al., 1998). In vitro studies on different neuronal cell lines have shown that in some cases apoptosis follows arrest of the cell cycle in G1 (Howard et al., 1993); in other cases, cell death occurs after reentry into S phase (Ninomiya et al., 1997). The actual relationships between both processes during development can be demonstrated only in the intact ani-

364

O. GANESHINA ET AL.

mal, in which the entire set of distant and local signals is preserved. In vivo experiments on intact chick dorsal root ganglia revealed that degenerating pycnotic cells labelled by 3H-thymidine appear as early as 2 hours after application of the thymidine (Carr and Simpson, 1981). Because the switch to PCD may precede the morphological signs of degeneration by several hours (for review see D’Mello, 1998), one may suggest that in chick dorsal root ganglia apoptosis is triggered in cells that are still in their cell cycle. In the case of the honeybee MBs the temporal and spatial overlap between proliferation and apoptosis can be considered as an indirect indication in support of the hypothesis that commitment to PCD in neuroblasts occurs during cell cycle progression. The identification of early events of apoptosis and of the phase of the cell cycle at the level of single cells in situ could be used to check this hypothesis.

ACKNOWLEDGMENTS We thank Randolf Menzel for fruitful discussions and general support; Ian Meinertzhagen for valuable critical comments and careful correction of the manuscript; and Mark Feldlaufer, Hans-Joachim Pflu¨ger, and Uli Mu¨ller for critical reading. We especially thank Klaus Hausmann for his hospitality at the EM, Rosi Hahmann for technical assistance in electron microscopy, and Mary Wurm for linguistic corrections.

LITERATURE CITED Abecrombie M. 1946. Estimation of nuclear population from microtome sections. Anat Rec 94:239 –247. Asano M, Nevins JR, Wharton RP. 1996. Ectopic E2F expression induces S phase and apoptosis in Drosophila imaginal discs. Genes Dev 10:1422– 1432. Bainbridge SP, Bownes M. 1988. Ecdysteroid titers during Drosophila metamorphosis. Insect Biochem 18:185–197. Blaschke AJ, Weiner JA, Chun J. 1998. Programmed cell death is a universal feature of embryonic and postnatal neuroproliferative regions throughout the central nervous system. J Comp Neurol 396:39 – 50. Booker R, Truman JW. 1987. Postembryonic neurogenesis in the CNS of the tobacco hornworm, Manduca sexta. II. Hormonal control of imaginal nest cell degeneration and differentiation during metamorphosis. J Neurosci 7:4107– 4114. Caldero´ J, Prevette D, Mei X, Oakley RA, Li L, Milligan C, Houenou L, Burek M, Oppenheim RW. 1998. Peripheral target regulation of the development and survival of spinal sensory and motor neurones in the chick embryo. J Neurosci 18:356 –370. Carr VM, Simpson SB. 1981. Rapid appearance of labelled degenerating cells in the dorsal root ganglia after exposure of chick embryos to tritiated thymidine. Dev Brain Res 2:15–162. Cayre M, Strambi C, Strambi A. 1994. Neurogenesis in an adult insect brain and its hormonal control. Nature 368:5–58. Cayre M, Strambi C, Charpin P, Augier R, Meyer MR, Edwards JS, Strambi A. 1996. Neurogenesis in adult insect mushroom bodies. J Comp Neurol 371:300 –310. Cayre M, Strambi C, Charpin P, Augier R, Strambi A. 1997. Inhibitory role of ecdysone on neurogenesis and polyamine metabolism in the adult cricket brain. Arch Insect Biochem Physiol 35:85–97. Champlin DT, Truman JW. 1998. Ecdysteroid control of cell proliferation during optic lobe neurogenesis in the moth Manduca sexta. Development 125:269 –277. Choi MK, Fahrbach SE. 1995. Evidence for an endogenous neurocidin in the Manduca sexta ventral nerve cord. Arch Insect Biochem Physiol 28:273–289. Coggeshall RA, Lekan HA. 1996. Methods for determining number of cells and synapses: a case for more uniform standards of review. J Comp Neurol 364:6 –15.

Davis RL. 1993. Mushroom bodies and Drosophila learning. Neuron 11:1– 14. de Belle JS, Heisenberg M. 1994. Associative odor learning in Drosophila abolished by chemical ablation of mushroom bodies. Science 263:692– 695. Diaz B, Pimentel B, De Pablo F, De La Rosa EJ. 1999. Apoptotic cell death of proliferating neuroepithelial cells in the embryonic retina is prevented by insulin. Eur J Neurosci 11:1624 –1632. D’Mello SR. 1998. Molecular regulation of neuronal apoptosis. Curr Top Dev Biol 39:187–213. Eichmu¨ller S. 1994. Vom Sensillum zum Pilzko¨rper: Immunhistologische und ontogenetische Aspekte zur Anatomie des olfaktorischen Systems der Honigbiene. PhD Thesis, Freie Universita¨t Berlin. Fahrbach SE, Strande JL, Robinson GE. 1995. Neurogenesis is absent in the brains of adult honey bees and does not explain behavioral neuroplasticity. Neurosci Lett 19:145–148. Feldlaufer MF, Herbert EW Jr, Svoboda JA, Thompson MJ, Lusby WR. 1985. The major ecdysteroid from the pupa of the honey bee, Apis mellifera. Insect Biochem 15:597– 600. Ganeshina OT, Malun D, Scha¨fer S. 1998. Development of mushroom bodies in honeybee—proliferation and programmed cell death. Eur J Neurosci 10(Suppl. 10):9. Gavrielli Y, Shermann Y, Ben-Sasson SA. 1992. Identification of programmed cell death in situ via specific labelling of nuclear DNA fragmentation. J Cell Biol 19:493–501. Gratzner HG. 1982. Monoclonal antibody to 5-bromo- and 5-iodo-deoxyuridyne: a new reagent for detection of DNA replication. Science 218:474 – 475. Ha¨hnlein I, Bicker G. 1996. Morphology of neuroglia in the antennal lobes and mushroom bodies of the brain of the honeybee. J Comp Neurol 367:235–245. Ha¨hnlein I, Bicker G. 1997. Glial patterning during postembryonic development of central neuropiles in the brain of the honeybee. Dev Genes Evol 207:29 – 41. Han K, Levin LR, Reed RR, Davis RL. 1992. Preferential expression of the Drosophila rutabaga gene in mushroom bodies, neural centers for learning in insects. Neuron 9:619 – 627. Heisenberg M. 1998. What do the mushroom bodies do for the insect brain? An introduction. Learning Mem 5:1–10. Heisenberg M, Borst A, Wagner S, Byers D. 1985. Drosophila mushroom body mutants are deficient in olfactory learning. J Neurogenet 2:1–30. Hensey C, Gautier J. 1997. A developmental timer that regulates apoptosis at the onset of gastrulation. Mech Dev 69:183–195. Herwig S, Strauss M. 1997. The retinoblastoma protein: a master regulator of cell cycle, differentiation and apoptosis. Eur J Biochem 246:581– 601. Holman SD, Collado P, Skepper JN, Rice A. 1996. Postnatal development of a sexually dimorphic, hypothalamic nucleus in gerbil: a stereologic study of neuronal number and apoptosis. J Comp Neurol 376:315–325. Howard MK, Burke LC, Mailhos C, Pizzey A, Gilbert CS, Lawson WD, Collins MKL, Thomas NSB, Latchman DS. 1993. Cell cycle arrest of proliferating neuronal cells by serum deprivation can result in either apoptosis or differentiation. J Neurochem 60:1783–1791. Ito K, Hotta Y. 1992. Proliferation pattern of postembryonic neuroblasts in the brain of Drosophila melanogaster. Dev Biol 149:134 –148. Kallen B. 1965. Degeneration and regeneration in the vertebrate central nervous system during embryogenesis. Progr Brain Res 14:77–96. Kerr JFR, Willie AH, Currie AR. 1972. Apoptosis: a basic biological phenomenon with wide ranging implications in tissue kinetics. Br J Cancer 26:239 –257. King KL, Cidlowski JA. 1998. Cell cycle regulation and apoptosis. Annu Rev Physiol 60:601– 617. Konigsmark BW. 1970. Methods for the counting of neurones. In: Nauta WJH, Ebbesson SOE, editors. Contemporary research methods in neuroanatomy. Heidelberg: Springer. p 315–338. Kuzin B, Roberts I, Peunova N, Enikolopov G. 1996. Nitric oxide regulates cell proliferation during Drosophila development. Cell 87:639 – 649. Landolt AM. 1965. Elektronenmikroskopische Untersuchungen an der Perikaryenschicht der Corpora pedunculata der Waldameise (Formica lugubris Zett.) mit besonderer Beru¨cksichtigung der Neuron-GliaBeziehung. Z Zellforsch 66:701–736. Macagno ER. 1979. Cellular interactions and pattern formation in the development of the visual system of Daphnia magna (Crustacea, Branchiopoda). Dev Biol 73:206 –238.

PROLIFERATION AND CELL DEATH IN MUSHROOM BODIES Malun D. 1998. Early development of mushroom bodies in the brain of the honeybee Apis mellifera as revealed by BrdU incorporation and ablation experiments. Learning Mem 5:90 –101. Menzel R, Mu¨ller U. 1998. Learning and memory in honeybees: from behavior to neural substrates. Annu Rev Neurosci 19:379 – 404. Menzel R, Durst C, Erber J, Eichmu¨ller S, Hammer M, Hildebrandt H, Mauelshagenj Mu¨ller U, Rosenboom H, Rybak J, Scha¨fer S, Scheidler A. 1994. The mushroom bodies in the honeybee: from molecules to behaviour. In: Schidberger K, Elsner N, editors. Fortschritte der Zoologie 39, neural basis of behavioural adaptations. Stuttgart: Gustav Fischer Verlag. p 81–102. Mobbs PG. 1982. The brain of the honeybee Apis mellifera. I. The connections and spatial organization of the mushroom bodies. Phil Trans R Soc London [Biol] 298:309 –354. Monsma SA, Booker R. 1996. Genesis of the adult retina and outer optic lobes of the moth, Manduca sexta. I. Patterns of proliferation and cell death. J Comp Neurol 367:10 –20. Mu¨ller U. 1997. The nitric oxide system in insects. Progr Neurobiol 51: 363–381. Nicotera P, Brune B, Bagetta G. 1997. Nitric oxide: inducer or supressor of apoptosis? Trends Pharmacol Sci 18:189 –190. Nighorn A, Healy MJ, Davis RL. 1991. The cyclic AMP phosphodiesterase encoded by the Drosophila dunce gene is concentrated in the mushroom body neuropil. Neuron 6:455– 467. Ninomiya Y, Adams R, Morriss-Kay GM, Eto K. 1997. Apoptotic cell death in neuronal differentiation of P19 EC cells: cell death follows reentry into S phase. J Cell Physiol 172:25–35. Nordlander RH, Edwards JS. 1969. Postembryonic brain development in the monarch butterfly, Danaus plexippus plexippus, L. I. Cellular events during brain morphogenesis. Wilhelm Roux Arch 162:19 –217. Panov AA. 1957. Bau des Insektengehirns wa¨hrend der postembryonalen Entwicklung. Rev Entomol URSS 39:269 –284. Robinow S, Talbot WS, Hogness DS, Truman JW. 1993. Programmed cell death in the Drosophila CNS is ecdysone-regulated and coupled with a specific ecdysone receptor isoform. Development 119:1251– 1259. Sonnenfeld MJ, Jacobs JR. 1995. Macrophages and glia participate in the removal of apoptotic neurons from the Drosophila embryonic nervous system. J Comp Neurol 359:644 – 652.

365

Stocker RF, Edwards JS, Truman JW. 1978. Fine structure of degenerating abdominal motor neurons after eclosion in the sphingid moth, Manduca sexta. Cell Tissue Res 191:317–331. Truman JW. 1984. Cell death in invertebrate nervous systems. Annu Rev Neurosci 7:171–188. Truman JW. 1996. Steroid receptors and nervous system metamorphosis in insects. Dev Neurosci 18:87–101. Truman JW, Bate M. 1988. Spatial and temporal patterns of neurogenesis in the central nervous system of Drosophila melanogaster. Dev Biol 125:145–157. Truman JW, Booker R. 1986. Adult specific neurons in the nervous system of the moth, Manduca sexta: selective chemical ablation using hydroxyurea. J Neurobiol 17:613– 625. Truman JW, Schwartz LM. 1984. Steroid regulation of neuronal death in the moth nervous system. J Neurosci 4:274 –280. Truman JW, Thorn RS, Robinow S. 1992. Programmed neuronal death in insect development. J Neurobiol 23:1295–1311. Truman JW, Talbot WS, Fahrbach SE, Hogness DS. 1994. Ecdysone receptor expression in the CNS correlates with stage-specific responses to ecdysteroids during Drosophila and Manduca development. Development 120:219 –234. Trump BF, Smuckler EA, Benditt EP. 1961. A method for staining epoxy sections for light microscopy. J Ultrastruct Res 5:43–348. Weeks J, Truman JW. 1985. Independent steroid control of the fates of motoneurons and their muscles during insect metamorphosis. J Neurosci 5:2290 –2300. Weeks JC, Roberts WM, Trimble DL. 1992. Hormonal control and segmental specificity of motoneuron phenotype during metamorphosis of the tobacco hornworm, Manduca sexta. Dev Biol 149:185–196. Wittho¨ft W. 1967. Absolute Anzahl und Verteilung der Zellen im Hirn der Honigbiene. Z Morphol Tiere 61:160 –184. Xiong W-C, Montell C. 1995. Defective glia induce neuronal apoptosis in the repo visual system of Drosophila. Neuron 14:581–590. Ye K, Ke Y, Keshava N, Shanks J, Kapp JA, Tekmal RR, Petros J, Joshi HC. 1998. Opium alkaloid noscapine is an antitumor agent that arrests metaphase and induces apoptosis in dividing cells. Proc Natl Acad Sci USA 95:1601–1606.