Early Development of Mushroom Bodiesin the Brain of the Honeybee Apis melliferaas Revealed by BrdU Incorporationand Ablation ExperimentsDagmar MalunInstitut fur NeurobiologieFreie Universitat Berlin14195 Berlin, Germany
Abstract
In the honeybee the mushroom bodiesare prominent neuropil structures arrangedas pairs in the dorsal protocerebrum of thebrain. Each mushroom body is composed ofa medial and a lateral subunit. Tounderstand their development, theproliferation pattern of mushroom bodyintrinsic cells, the Kenyon cells, wereexamined during larval and pupal stagesusing the bromodeoxyuridine (BrdU)technique and chemical ablation withhydroxyurea.
By larval stage 1, ∼40 neuroblasts arelocated in the periphery of theprotocerebrum. Many of these stem cellsdivide asymmetrically to produce a chain ofganglion mother cells. Kenyon cellprecursors underly a different proliferationpattern. With the beginning of larval stage 3,they are arranged in two large distinct cellclusters in each side of the brain. BrdUincorporation into newly synthesized DNAand its immunohistochemical detectionshow high mitotic activity in these cellclusters that lasts until mid-pupal stages.The uniform diameter of cells, thehomogeneous distribution of BrdU-labelednuclei, and the presence of equally dividingcells in these clusters indicate symmetricalcell divisions of Kenyon cell precursors.
Hydroxyurea applied to stage 1 larvaecaused the selective ablation of mushroombodies. Within these animals a variety ofdefects were observed. In the majority ofbrains exhibiting mushroom body defects,either one mushroom body subunit on one
or on both sides, or three or four subunits(e.g., complete mushroom body ablation)were missing. In contrast, partial ablation ofmushroom body subunits resulting in smallKenyon cell clusters and peduncles wasobserved very rarely. These findingsindicate that hydroxyurea applied duringlarval stage 1 selectively deletes Kenyonstem cells. The results also show that eachmushroom body subunit originates from avery small number of stem cells anddevelops independently of its neighboringsubunit.
Introduction
The mushroom bodies of insects are thoughtto play a major role in processing and storage ofchemosensory information (Menzel et al. 1974,1994; Erber et al. 1980; Heisenberg et al. 1985;Davis 1993; de Belle and Heisenberg 1994; Ham-mer and Menzel 1995). This notion was basedoriginally on the fact that the main input to themushroom bodies comes from the antennal lobes,the first central station in the olfactory pathway.The participation of mushroom bodies in olfactorylearning and memory was subsequently more di-rectly investigated particularly in fruitflies and hon-eybees using various experimental techniques andsimultaneous examination of behavioral responsesto trained odor stimuli. One approach used elec-trophysiological methods to characterize neuronstaking part in neuronal circuits of the mushroombodies in the honeybee (Hammer 1993; Mauelsha-gen 1993). Recordings from two identified neu-rons (VUMmx1 and PE1) during olfactory condi-tioning demonstrated their participation in nonas-sociative and associative olfactory learning. In
another approach, temporary blocking of mush-room body function of honeybees through localcooling led to retrograde amnesia during the periodof a few minutes following olfactory learning (Men-zel et al. 1974; Erber et al. 1980). Furthermore,mushroom body structural mutants obtained by agenetic approach in Drosophila exhibit impairedolfactory learning and memory (Heisenberg et al.1985; Davis and Dauwalder 1991; Skoulakis et al.1993; for review, see Davis 1993). Also in Dro-sophila, an ablation procedure has been estab-lished to selectively delete mushroom bodies (deBelle and Heisenberg 1994). Classical conditioningof these animals demonstrated that mushroombodies mediate associative odor learning in flies.Taken together, these results indicate a central roleof mushroom bodies in learning and memory.
The mushroom bodies are also characterizedby their neuronal plasticity during developmentand even during early adult life. Studies on themetamorphosis of the mushroom bodies in Dro-sophila demonstrated that reorganization in theperiphery (e.g., complete degeneration and re-placement of olfactory sense organs) during pupa-tion parallels extensive neural reorganization ofKenyon cell processes within the mushroom bod-ies (Technau and Heisenberg 1982). In adult flies,vision affects the volume of mushroom bodies aswas shown recently by Barth and Heisenberg(1997). Mushroom bodies in honeybees are alsostructurally highly plastic in adult life (Withers etal. 1993; Fahrbach et al. 1995a). Durst et al. (1994)found that the transition of nursing bees to forag-ing bees 7–10 days after emergence is accompa-nied by a drastic volume change of the calyx re-gions. These structural plasticities are indicative ofsynaptic reorganization at the input side of themushroom bodies and may reflect the connectivityadaptions related to learning processes in foragingbees.
For further clarification of the function ofmushroom bodies of honeybees, its selective elim-ination as described for Drosophila would bebeneficial. However, for obvious reasons, the hon-eybee is not as easily accessible for genetic ex-periments as Drosophila. Thus, a manipulation ofmushroom body size and structure by interferencewith its ontogenetic development might be prom-ising. So far, no successful attempt was made toeliminate mushroom bodies during developmentas was performed in Drosophila (de Belle and Hei-senberg 1994). To achieve precise and selectivemushroom body ablation by this method, the ori-
gin of the intrinsic cells, the Kenyon cells that formthe mushroom bodies (Kenyon 1896), and the tem-poral proliferation pattern of Kenyon precursorcells need to be explored. Thus, in a first approachof this study, the mitotic activity in larval stageswas examined using the BrdU technique. Althoughthe exact number of origin Kenyon cell neuro-blasts could not be identified by this method, theapproximate time period (larval stage 1) of the on-set of Kenyon cell proliferation was determined. Ina second approach, the DNA synthesis inhibitorhydroxyurea was applied to larvae in this timeframe.
Defects induced in the bee differ from thosedescribed for Drosophila in many respects. Thesedifferences are likely to be owing to the number oforigin Kenyon cell neuroblasts, to their prolifera-tion pattern, and to the overall structural organiza-tion of mushroom bodies in bees and flies. Theresults of this study also show that in honeybeesselective mushroom body ablation with hydroxy-urea is a feasible method that provides the oppor-tunity to test the involvement of mushroom bodies innonassociative and associative behavioral paradigms.
Some of the results have been published inabstract form (Malun 1997).
Materials and Methods
TOLUIDINE BLUE STAINING
Histological staining was performed on braintissue of worker honeybees (Apis mellifera car-nica) obtained from regular hives. The develop-mental stage of animals was determined accordingto Bertholf (1925). Brains were dissected underbee saline (130 mM NaCl, 6 mM KCl, 2 mM MgCl2,7 mM CaCl2, 160 mM glucose, 10 mM HEPES at pH6.7 and 500 mOsmoles; all chemicals are fromSigma, Deisenhofen, Germany) from the head cap-sule, immersed in fixative solution [2.5% (wt/vol)glutaraldehyde in phosphate buffer at pH 7.1] over-night at 4°C, rinsed in buffer, and postfixed in 1%OsO4 for 1 hr at room temperature. The tissue wassubsequently dehydrated and embedded in Dur-cupan (Fluka, Buchs, Switzerland). Serial sections(5 µm thick) were cut with a microtome (Om U3ultramicrotome, Reichert) and stained with tolu-idine blue. The sections were embedded in Entel-lan (Merck, Darmstadt, Germany) and examinedwith a microscope (Polyvar, Reichert-Jung). Serialreconstructions of the sectioned tissue were car-ried out to perform cell counts.
MUSHROOM BODY DEVELOPMENT IN THE HONEYBEE
&L E A R N I N G M E M O R Y
91
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
To visualize DNA synthesis in dividing cells,5-bromo-28-deoxyuridine (BrdU, Sigma, Deisen-hofen, Germany) was injected (∼2 µl of a 6.25-mg/ml BrdU/saline solution) into larval stage 3 animalsand older larvae and into pupae. Because the smallsize of animals of early larval stages (stage 1 and 2)precluded injections, these animals were fed witha BrdU solution [6.25 mg of BrdU per ml of royaljelly/dH2O solution (1:1)]. BrdU was allowed toincorporate 3–7 hr in larvae and 10–22 hr in pu-pae. Brains were dissected, fixed in 4% paraform-aldehyde, embedded in paraplast, and cut into 5- to7-µm-thick sections with a frontal orientation. Sub-sequently, using an immunohistochemical proce-dure, the sections were stained with a monoclonalantibody to BrdU (Amersham Buchler, Braunsch-weig, Germany) and a secondary antibody conju-gated to Cy3 (Dianova, Hamburg, Germany). Thesections were dehydrated, mounted with Entellan,and examined with a laser confocal microscope(see below).
CONFOCAL MICROSCOPY
Samples were examined with a confocal laserscanning microspope (Leica TCS-4D) equippedwith a Leitz microscope (DM RBE) and a Krypton/Argon laser light source. The Cy3 signal of theBrdU immunostaining was excited with the 568-nm line of the Krypton/Argon laser and detectedwith the 590-nm long-pass filter. Single optical sec-tions were scanned from 7-µm-thick sections of thebrain. Subsequently, some preparations were ex-amined for their autofluorescence using an appro-priate filter set (excitation wavelength, 515 nm;barrier filters, 476 and 488 nm). The autofluores-cent signal of the aldehyde-fixed tissue emphasizedthe shape of cell clusters or regions of neuropil(Fig. 3D, below). Images showing the BrdU signaland the signal taken with autofluorescence helpedto examine the position of mitotically active cellswithin the tissue. Figure panels were created inAdobe Photoshop, and photographs were takenwith a slide maker (Lasergraphics Personal LFRPlus).
ABLATION EXPERIMENTS
Hydroxyurea selectively deletes dividing cells(Truman and Booker 1986; Prokop and Technau
1994). To ablate proliferating Kenyon cell neuro-blasts, hydroxyurea at a concentration of 0.5–3.5mg/ml of royal jelly/distilled H2O solution (1:1)was fed to stage 1 larvae for 5 hr in multiwells in ahumid chamber at 35°C. Control larvae were fedwith royal jelly/distilled H2O solution only. Afterthe treatment, the larvae were put back into thehive for further development. At different develop-mental stages, the brains were dissected and pre-pared for subsequent histological examination(PKAII immunohistochemistry; see below). Thirty-six animals showed mushroom body defectsand were investigated in detail. Five animalswere taken at larval stage 5, and 31 animals weretaken either just after or up to 6 days after adulthatching.
PKAII IMMUNOHISTOCHEMISTRY
To emphasize the shape of the mushroom bod-ies but also that of other neuropil areas, immuno-histochemistry was performed on 7-µm-thick para-plast sections using a monoclonal antibody to theregulatory subunit of the cAMP-dependent proteinkinase type II (PKAII; Muller 1997). As described indetail by Muller (1997), the PKAII was visualizedusing a primary monoclonal antibody (final dilu-tion, 1:200), a biotinylated goat anti-mouse second-ary antibody (final dilution, 1:2000; Boehringer,Ingelheim, Germany), and streptavidin-alkalinephosphatase (Sigma, Deisenhofen, Germany),5-bromo-4-chloro-3-indolyl phosphate (AppliChem,Darmstadt, Germany), and blue tetrazolium(Sigma) as staining reagents.
Results
BRDU STUDY AND TOLUIDINE BLUE STAINING
In the adult honeybee brain, the mushroombodies appear as prominent neuropil regions ar-ranged as pairs in the dorsal protocerebrum (Fig.1). Each mushroom body consists of 170,000densely packed Kenyon cells (Witthoft 1967). Theneuropil of each mushroom body is composed oftwo subunits: a pair of cup-shaped structures, thecalyces, and two peduncles. Kenyon cell neuriteswithin the bipartite peduncle divide to enter boththe a- and the b-lobes, respectively (Mobbs 1982;Rybak and Menzel 1993). Thus, the lobes are fusedproducts of Kenyon cell processes from bothmushroom body subunits.
Malun
&L E A R N I N G M E M O R Y
92
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
This study was devoted to examining the post-embryonic proliferation pattern of neuroblasts,particularly of Kenyon cell neuroblasts, during de-velopment of the mushroom bodies. Neuroblastswere identified in toluidine blue-stained sections oflarval brains as large chromatin-poor cells thatwere located predominantly in the periphery ofthe brain (Fig. 2B, larval stage 3). In stages 1, 2, and3, the only stages examined for cell counts (n = 2–4 for each stage), ∼40 large neuroblasts with a di-ameter of ∼20 µm were found per brain hemi-sphere within the protocerebrum. Sections ofBrdU-treated larvae showed labeled nuclei of somestem cells as early as stage 1 indicating their mitoticactivity (Fig. 3A,B). Daughter cells were smaller indiameter (∼8 µm) than the stem cells from whichthey derived. These findings together with theabove-mentioned constant number of large neuro-blasts in different larval stages indicate that thesestem cells reproduce themselves and that the smalldaughter cells are products of an asymmetrical celldivision (Fig. 2B). According to Edwards (1969),
Doe and Goodman (1985), and Truman and Bate(1988), in the insect nervous system, the progenyof neuroblasts that are generated by this type ofcell division are called ganglion mother cells. Acharacteristic feature of this proliferation pattern isthat ganglion mother cells are arranged in columnsjust underneath the stem cells (Zacharias et al.1993). In the protocerebrum of bees, this patternof large neuroblasts associated with columns ofganglion mother cells appeared with the beginningof larval stage 2 (Fig. 2B).
By larval stage 2, a distinct cell cluster ap-peared within the dorsal protocerebrum of eachbrain hemisphere. This cell cluster was easily dis-
Figure 1: Schematic diagram of an adult honeybeebrain in a frontal view. The paired mushroom bodies(shaded areas) consist of a medial (mc) and a lateralcalyx (lc), a peduncle (p), and an a- (a) and a b-lobe (b).The somata of the Kenyon cells (kc) lie mainly in thecenter of the cup-shaped neuropils of the median andlateral calyces. The main input connections of the mush-room bodies are shown: Olfactory projection neuronsfrom the antennal lobes (al) project via the medial an-tenno-glomerular tract (magt) exclusively to the lip re-gion (lip) of the calycal neuropil and terminate withinthe lateral protocerebrum (lpl). Projection neurons leav-ing the antennal lobe within the lateral antennoglomeru-lar tract (lagt) innervate the lateral protocerebrum andthe calyces. Visual fiber tracts originating in the opticlobes [medulla (me); lobula (lo)] project into both brainhemispheres via the anterior commissure (ac) and inner-vate the collar of the calyces. (an) Antennal nerve; (sog)suboesophagylganglion; (oc) ocelli; (d) dorsal; (l) lateral.All following figures refer to the same orientation.
Figure 2: Toluidine blue-stained semithin plastic sec-tion of a brain at larval stage 3. The dorsal protocere-brum of the right side of the brain is shown. (A) Two cellclusters (arrows) that constitute the proliferation centersof Kenyon cells contain densely stained small neuro-blasts of uniform diameter. (Double arrow) Left cell clus-ter; (arrow) right cell cluster. (B) Ten micrometersdeeper, the lateral cell cluster (arrow) reaches its greatestextension. Three large neuroblasts (large arrowheads)are located in the periphery of the protocerebrum. Onelarge neuroblast gives rise to a column of smaller gan-glion mother cells (small arrowheads). (d) Dorsal; (l) lat-eral. Scale bar, 50 µm.
MUSHROOM BODY DEVELOPMENT IN THE HONEYBEE
&L E A R N I N G M E M O R Y
93
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
tinguished from the surrounding tissue by its tightconglomeration and by the great number of cellswith BrdU-labeled nuclei (Fig. 3C,D). One larvalstage later, a second cell cluster became visiblelaterally to, but in close apposition with, the firstcluster (Figs. 2A,B and 3E,F). Because of their rela-tive positions within the protocerebrum, they willbe referred to as medial and lateral cell groups(Figs. 2A,B and 3E,F). These two cell groups con-stitute the proliferation centers of the Kenyon cells(this paper; Panov 1957). In toluidine blue-stainedtissue, cells within these clusters had a uniformappearance with a homogeneous dark cytoplasmand a diameter of ∼10 µm (Fig. 2A,B). Thus, theyare smaller than the large neuroblasts in the pe-riphery of the brain. Data gathered from both to-luidine blue- and Bodian-stained sections and fromBrdU immunolabeling revealed that the two cellclusters of each hemisphere persisted and showed
mitotic activity until mid-pupal stages (pupal stage4; Fig. 4A,B).
Throughout larval and pupal development,these proliferation centers were never observed tobe associated with large neuroblasts. This fact, to-gether with the above-mentioned uniform size ofthe cells, and homogeneous distribution of BrdU-stained nuclei within the cell groups strongly sug-gest that the Kenyon cell precursors undergo sym-metrical cell divisions. This assumption is strength-ened by findings gathered by classical cytology; inBodian-stained brains, Kenyon cell precursorswere observed regularly that divide equally (Fig.4B). Because of their high mitotic activity and theirproliferation pattern, I consider them as small neu-roblasts that clearly differ from the large commonneuroblasts in the periphery of the brain that per-form asymmetric cell division and that are associ-ated with columns or rows of their progeny.
Figure 3: Confocal micrographs of paraffinsections through larval brains after BrdU in-corporation and anti-BrdU immunohisto-chemistry to visualize mitotic activity. (A,B)Two consecutive sections of a larval stage 1brain. BrdU immunofluorescence is visible innumerous cell nuclei including nuclei oflarge neuroblasts in the periphery of the pro-tocerebrum (arrows in A). A large neuroblastperforms asymmetric cell division that is notyet completed (B, arrow). The large nucleusof the neuroblast is unlabeled, whereas thesmall nucleus of the ganglion mother has al-ready entered S phase of a further mitoticcycle. (Inset) The dividing neuroblast (arrow)at higher magnification. (C,D) Right side ofthe brain at larval stage 2. Several BrdU-im-munolabeled nuclei are concentrated withinone cell group (arrow in C) that is also visibleas a distinct cell group in confocal micro-graphs scanned for autofluorescence (arrowin D). The cluster constitutes the medial pro-liferation center for Kenyon cells. Large neu-roblasts in the periphery of the protocere-brum are indicated by arrowheads. (E,F) Bylarval stage 4, two distinct cell clusters con-taining numerous BrdU-labeled nuclei (ar-rows) have arisen in each side of the proto-cerebrum. (F) The two proliferation centers ofthe left brain hemisphere at higher magnifi-cation. Scale bars in A, C, and F, 50 µm; in Binset, 25 µm; in E, 100 µm.
Malun
&L E A R N I N G M E M O R Y
94
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
As shown above, larval stage 2 was the firststage in which Kenyon cell precursors weregrouped in the medial cell clusters and were there-fore clearly distinguishable from surrounding cells(Fig. 3C,D). At that time, only a small number ofKenyon precursor cells had been formed. Theclusters consisted of ∼20 cells. Thus, the preced-ing stage was most likely the stage in which theKenyon cell neuroblasts began to proliferate.Therefore, I expected hydroxyurea application tobe most effective in interfering with cell prolifera-tion by stage 1. Blockade of cell division at thisearly stage should cause strong ablation effects ofmushroom bodies.
To obtain a precise picture of structural de-fects within the brain, a PKA immunostaining pro-cedure was applied. The anti-PKA antiserum labelsneuropil structures and particularly intensively, themushroom body neuropil (Muller 1997). Althoughthe overall intensity of the PKA immunostaining inthe present study varied among animals, strong la-beling was always observed within the neuropil ofthe mushroom bodies (Fig. 5). Thus, PKA immu-nolabeling provided a useful indicator for mush-room body defects. PKA immunostaining in un-treated animals served as a control and reference toanimals with hydroxyurea-induced ablation (Fig.5A).
Ablation of mushroom bodies was observed in31 brains out of 37 hydroxyurea-treated animalsthat passed adult hatching. The adult eclosion ratedepended on the hydroxyurea concentration ap-plied. Highest viability (30%) was observed in ani-mals treated with low concentrations of hydroxy-
urea, that is, 0.5–3.5 mg/ml, which is below theconcentrations used for chemical ablation of mush-room bodies in Drosophila (de Belle and Heisen-berg 1994) and is the range of concentrations usedfor partial ablation of Drosophila mushroom bod-ies (Ito et al. 1997). Higher hydroxyurea concen-tration applied to stage 1 honeybee larvae resultedin considerable larval and pupal lethality. Threeout of eight animals studied at larval stage 5 ex-hibited a normal mushroom body morphology,whereas five larvae showed mushroom body de-fects. All 45 animals (larvae and pupae) were stud-ied in detail for hydroxyurea-induced defects. Sizeand external anatomy of animals exhibiting mush-room body defects were not distinguishable fromcontrol animals. No obvious adverse effects of thehydroxurea treatment outside of the brain wereobserved. The gross morphology, shape, and vol-ume of brain structures other than the mushroombodies (e.g., central complex, optic lobes, and an-tennal lobes) were not affected in the brains ofhydroxyurea-treated animals (Fig. 5). This, how-ever, does not exclude small defects (e.g., ablationof single cells or small neuropilar compartments)that would not have been detected with the histo-logical examination used in this study. The pres-ence or absence of the calyces is the most reliableindicator for mushroom body defects because thecalyces are formed by the Kenyon cell dendritesand constitute morphologically prominent neuro-pil regions. Consequently, the ablation of Kenyoncell neuroblasts would be most likely reflected inmorphological changes within the calyces and thepeduncle of each calyx. The a- and b-lobes, how-ever, are the fused structures of Kenyon cells fromboth the medial and the lateral calyx (Rybak and
Figure 4: Mitotic activity of Kenyon precur-sor cells in the mushroom body proliferationcenters during pupal development. (A) Confo-cal micrograph of a paraffin section through abrain at pupal stage 4 after BrdU incorpora-tion and anti-BrdU immunohistochemistry tovisualize mitotic activity. Numerous labelednuclei are homogeneously distributed withinthe mushroom body proliferation center. (c)Developing calyx neuropil. (B) Bodian-stainedsections of a brain at pupal stage 2. Cell divi-sion (the anaphase of the mitocic cycle isshown) of Kenyon cell precursors occurswithin the mushroom body proliferation cen-
ter. The two daughter cells (arrows) are of equal diameter indicating symmetrical cell division. (Inset) Symmetricallydividing Kenyon cell precursors (arrows) during telophase of the mitotic cycle. Scale bar in A, 100 µm; in B, 25 µm.
MUSHROOM BODY DEVELOPMENT IN THE HONEYBEE
&L E A R N I N G M E M O R Y
95
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
Menzel 1993) and could not contribute as un-equivocal indicators for defects in individual mush-room body substructures. Mushroom body defectsin larvae were characterized by the presence orabsence of the mushroom body proliferation cen-ters and the ‘‘calyx developmental zone’’ that com-prises the precursor structure of the future calycalneuropil (Menzel et al. 1994). Thus, larvae showeda similar distribution of the various mushroom de-fects as adult animals, and therefore, findings onlarvae were included into data gathered on adultanimals.
The various mushroom body defects observedare grouped into five main classes and are summa-rized in Figure 6: (1) In one group of animals(n = 10), the medial calyx in one hemisphere wascompletely absent (Fig. 5B). This became obviousby close examination of the level of the brain inwhich the lateral calyces showed their greatest ex-tension and their peduncles projected into the pro-
Figure 6: Schematic diagram of hydroxyurea-inducedmushroom body defects indicated by the presence andabsence (crossed out) of calyces of the mushroom bod-ies. Number of animals with each type of defects (n) andtheir relative representation are shown.
Figure 5: PKA immunostaining on paraffinsections of adult brains of a control animal (A)and of hydroxyurea-treated animals (B–F). Im-munolabeling is present in all neuropilar areasbut is especially strong in the mushroom bod-ies. (A–C,E,F) Same levels of the brains fromfrontal serial sections. (D) A more anterior sec-tion. (A) Brain of a control animal. Strong PKAimmunolabeling is found within the calyces ofthe mushroom bodies. (B) The medial calyx ofthe left hemisphere is deleted. (C) The medialcalyces in both brain hemispheres are ablated.The calyces extend to the midline of the brain.The volume of the protocerebrum is reducedowing to the mushroom body ablation. (D) Thelateral and the medial calyces of the left hemi-sphere are ablated, whereas both calyces of theopposite hemisphere are present. The sectionshows the a-lobe in the intact side and thecomplete absence of the a-lobe in the hemi-sphere with mushroom body ablation. The pro-tocerebrum without mushroom body is greatlyreduced in its size. Although strong PKA im-munostaining is detectable in all neuropil re-gions, the strongest labeling is found within themushroom bodies. (E) Only the lateral calyx ofthe right hemisphere is present. (F) Brain withcomplete mushroom ablation. The immunos-taining does not detect any parts of the mush-room bodies. The volume of the protocere-brum has greatly shrunk. (al) Antennal lobe;(lc) lateral calyx; (ml) medial calyx; a, a-lobe.Scale bar, 500 µm.
Malun
&L E A R N I N G M E M O R Y
96
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
tocerebral neuropil. Although only one calyx de-veloped on one side, it had the same dimension asthe corresponding lateral calyx on the oppositebrain side. Furthermore, the position and thecourse of the peduncles were the same for bothlateral calyces (Fig. 5B). In more anterior sections,the region in which the medial calyces reachedtheir greatest extension, the medial calyx of theintact hemisphere lacked its counterpart on theother side of the brain. In the brain hemisphereswith the medial calyx missing, the a-lobe wasgreatly reduced (half the size of a-lobes in controlanimals). Moreover, even the whole protocere-brum appeared decreased in size on one side, be-cause the surface of protocerebral outline becameflatter. (2) The majority of animals (n = 17) lackedthe medial calyx in both hemispheres, whereas thelateral calyces appeared normal with respect totheir size (Fig. 5C). However, in these brains thecup-shaped neuropil of the lateral calyces ap-peared flatter and extended into the region nor-mally occupied by the medial calyx and evenreached the midline of the brain. (3) The fact thatthe protocerebrum decreased in size by ablation ofmushroom body subunits became even more obvi-ous in brains with complete loss of a mushroombody in one brain side (Fig. 5D). The contralateralside developed the full set of calyces, peduncles,and lobes. These defects were observed very rarely(n = 1). (4) Four animals had developed only onelateral calyx. The medial and lateral calyces of oneand the medial calyx of the other hemisphere werecompletely missing (Fig. 5E). (5) Finally, two ani-mals showed a complete mushroom body ablationon both sides. Because the mushroom body in hon-eybees represents 10% of the total bee brain vol-ume (Mobbs 1982), the protocerebrum of theseanimals had decreased considerably in size (Fig.5F). Immunolabeling of PKAII for these animals didnot provide any evidence for the existence ofKenyon cells.
In two animals (not listed in the diagram of Fig.6), a partial ablation within mushroom body sub-units was observed: One brain had developed justa small clump of presumptive Kenyon cells that didnot give rise to a prominent calyx neuropil butformed just a very small, rudimentary peduncle(not shown). Another animal that was sacrificedand processed at the last larval stage (larval stage 5)had developed two cell clusters of Kenyoncell precursors in one side and a single cell clusterof smaller size within the opposite side (notshown).
Discussion
KENYON PRECURSOR CELLS DIVIDESYMMETRICALLY
The present study did not succeed in deter-mining the number of origin Kenyon cell neuro-blasts that form the two proliferation centers ofeach side of the brain. Neither toluidine blue stain-ing nor the BrdU technique revealed large neuro-blasts on top of the Kenyon cell cluster that wouldgive rise to ganglion mother cells via asymmetricalcell divisions and, subsequently, via a final sym-metrical division, to a pair of Kenyon cells as de-scribed for developing mushroom bodies in Dro-sophila (Ito and Hotta 1992; Ito et al. 1997; Tetta-manti et al. 1997). Possibly, Kenyon cells inhoneybees do not derive from large conspicuousneuroblasts as in Drosophila but from smaller stemcells. A more reasonable explanation, however,would be that large origin Kenyon cell neuroblastslying in the periphery of the brain begin to dividesymmetrically very early during larval development(larval stage 1) and give birth to two identical smallneuroblasts as it was shown for the monarch but-terfly Danaus by Nordlander and Edwards (1970).By performing this proliferation pattern, the originneuroblasts would exist just for a short period oftime and, therefore, would be difficult to detect.Likewise, the uniform size of small mitotically ac-tive neuroblasts observed in the proliferation clus-ters as early as larval stage 2 strongly suggests thatthe small daughter neuroblasts that derive from ori-gin neuroblasts also perform symmetrical cell divi-sions. The proliferation centers formed by theseneuroblasts enlarge until late larval and early pupalstages and exhibit mitotic activity until mid-pupalstages, whereas neurogenesis is absent in thebrains of adult honeybees (Fahrbach et al. 1995b;this paper). The proposed proliferation pattern canexplain the development of the enormous numberof 170,000 Kenyon cells in honeybees (Witthoft1967). The findings of the present study are inagreement with data published by Panov (1957,1960) who described two main types of prolif-eration patterns in the developing honeybeebrain. Panov distinguished between (1) large neu-roblasts that divide asymmetrically, thereby repro-ducing themselves and generating smaller ganglionmother cells that subsequently produce a pair ofneurons in a symmetrical division, and (2) prolif-eration centers in which small neuroblasts dividesymmetrically to produce small neuroblasts that in
MUSHROOM BODY DEVELOPMENT IN THE HONEYBEE
&L E A R N I N G M E M O R Y
97
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
turn divide symmetrically. Panov further proposedthat in honeybees, both proliferation patterns takeplace during the development of the optic lobes,whereas the Kenyon cells are generated exclu-sively by the second type of proliferation pattern.The latter observation was recently confirmed bydata published on the development of the honey-bee brain by S.M. Farris, G.E. Robinson, and S.E.Fahrbach (unpubl.).
In Drosophila, mushroom body neuroblastsundergo asymmetric cell divisions throughout lar-val stages and during pupal stages until almost theend of metamorphosis (Ito and Hotta 1992). How-ever, in the optic lobes of Drosophila, like in theoptic lobes of Apis, neuroblasts first divide sym-metrically to produce a large number of stem cells,whereas the later neuroblasts divide in a typicalasymmetrical fashion (Truman et al. 1993). Thisseems to indicate that symmetrical cell divisionpredominantly takes place when a large number ofneuroblasts needs to be generated over a short pe-riod of time.
According to the proposed pattern of prolif-eration of Kenyon precursor cells in honeybees,large origin Kenyon cell neuroblasts would lieamong, and be indistinguishable from, the otherlarge neuroblasts in the periphery of the protoce-rebrum for a very short period, that is, before theystart to divide and then disappear. The total num-ber of these protocerebral neuroblasts appears tobe rather constant (∼40 per hemisphere) in earlypostembryonic stages, throughout larval stages 1and 2, and even after the two proliferation centersof each mushroom body have been formed (stage3). If the mushroom proliferation center would de-rive from a large subpopulation of these stem cells,it should be expected that the latter disappear assoon as they start their symmetrical cell division(see above). This would diminish the total numberof these large conspicious neuroblasts. However,such a decrease was not observed. The loss of avery small number of stem cells (e.g., their conver-sion into small neuroblasts) would not be discov-ered by cell counts. The present study thereforesuggests that Kenyon cells are generated by a verysmall number of large origin neuroblasts like inanother holometabolous insect, the butterfly Dan-aus, where cell divisions of Kenyon cell neuro-blasts were found to take place within two prolif-eration clusters (in each side of the brain), each ofwhich were shown to derive from three large neu-roblasts, respectively (Nordlander and Edwards1970). The data gathered with the ablation tech-
nique further support the notion of a similar situ-ation in the honeybee.
This study demonstrates that hydroxyurea ap-plied to honeybees at larval stage 1 leads to specificablations within the mushroom bodies. Althoughthe defects varied in detail (summarized in Fig. 6),the prevalent feature was the complete deletion ofindividual mushroom body subunits, whereas thepersisting subunits retained their overall shape andsize. These findings can be explained in conjunc-tion with the results of the BrdU study and histo-logical data: During early larval development, themedial proliferation centers of Kenyon cell precur-sors appeared about one larval stage earlier thanthe lateral cell clusters. Therefore, cell prolifera-tion of the stem cells forming the medial prolifera-tion centers is likely to start before proliferation ofthe cells producing the lateral cluster. Immediatelyafter their appearance, both clusters are character-ized by their structural integrity producing twoclearly separate units although they are not delim-ited by a continuous glial sheath (Hahnlein andBicker 1997).
In this context, two striking features of thehydroxyurea experiments have to be mentioned:(1) Ninety-four percent of the animals displayinghydroxyurea-induced mushroom body defectsshowed complete ablations of mushroom bodysubunits even when the hydroxyurea concentra-tion was reduced to threshold concentrations (0.5mg/ml). Smaller or rudimentary mushroom bodieswere observed in just very rare occasions (2 out of36 animals). In these animals, some small Kenyoncell neuroblasts probably escaped hydroxyurea ab-lation and formed a small number of Kenyon cells.(2) Mushroom body ablation or reduction predomi-nantly affected the medial calyces. One hundredpercent of the animals exhibit mushroom body ab-lation of one or both medial calyces, whereas in20% of the animals, one or both lateral clayceswere missing.
These findings together with the above-men-tioned histological data suggest that each cell clus-ter arises from separate stem cells. Consequently,in animals exhibiting only one mushroom bodysubunit in each hemisphere, this remaining sub-unit is not a merged structure but is rather gener-ated by Kenyon cell precursors of a single prolif-
Malun
&L E A R N I N G M E M O R Y
98
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
eration cluster. This further strengthens the ideathat the two cell clusters develop functionally sepa-rately and that intercellular interactions betweenthem, that is, exchange of cells, does not occur.After ablation of one mushroom body subunit, thecalyx neuropil of the remaining mushroom bodysubunit just seems to take over freed space withinthe protocerebrum. Thereby, the calyx gives up itsprominent cup-shaped structure and gets a flatterappearance. Yet, as a result of the ablation of onemushroom body subunit, the protocerebrum isclearly diminished in volume and so are the a- andb-lobes. The high incidence of animals with ab-lated medial mushroom bodies also shows that hy-droxyurea applied to newly hatched larvae main-ly affected stem cells of the medial cell cluster,whereas the stem cells forming the lateral cell clus-ter escaped ablation. This finding is compatiblewith an earlier onset of proliferative activity thatleads to the medial cell cluster and, ultimately, themedial calyx. Although they were observed in veryrare occasions (n = 2) only, animals with completemushroom body ablation suggest that the periodsduring which origin neuroblasts of the medial andthe lateral calyces proliferate can overlap. Abouthalf of the population of animals with mushroombody ablation exhibit asymmetrical defects wherethe two sides of the brain differ with respect to thenumber of mushroom body subunits present (Figs.5 and 6). Animals with such defects (Fig. 5B,D,E)underline the ‘‘all-or-none’’ character of the abla-tions induced by hydroxyurea. They strongly sug-gest that mushroom body subunits originate from asmall number of Kenyon stem cells, possibly asingle neuroblast, respectively. They also showthat the hydroxyurea concentration applied wasclose to threshold.
Overall, it was surprising that the portion ofanimals with complete mushroom body ablation inthis study was very small. However, the reason forthis observation might be that only very low con-centrations of hydroxyurea (0.5–3.5 mg/ml) leadto survival and hatching of imagines. As revealedby the BrdU study presented here, some stem cellsincluding presumable Kenyon cell neuroblasts inthe protocerebrum are mitotically active during lar-val stage 1, the stage hydroxyurea had been ap-plied. Therefore, higher concentrations of hy-droxyurea might have caused a higher incidence ofcomplete mushroom body ablation but might havealso affected other proliferating cells. The resultingdamage throughout the brain might have led tolower survival rates.
COMPARISON TO THE MUSHROOM BODYDEVELOPMENT IN DROSOPHILA
Organization and development of mushroombodies in honeybees and Drosophila differ in manyrespects. Drosophila develop just one unpairedmushroom body with just one calyx and one pe-duncle per side of the brain. Each mushroom bodyconsists of 2500 Kenyon cells only (Hinke 1961) incontrast to 170,000 in honeybees (Witthoft 1967).In Drosophila, each mushroom body is generatedby four Kenyon cell neuroblasts (Ito and Hotta1992), and each neuroblast is capable of formingautonomously all of the mushroom body structures(Ito et al. 1997). It was proposed that each of theseneuroblasts produces a series of ganglion mothercells by asymmetrical divisions. The latter werethought to perform a single symmetrical cell divi-sion generating a pair of Kenyon cells. The largenumber of honeybee Kenyon cells cannot be ac-counted for by such a proliferation scheme: On theassumption of one neuroblast of 85,000 Kenyoncells per mushroom body subunit in the bee, thisproliferation pattern would require 42,500 asym-metrical cell divisions by each neuroblast in ∼310hr (13 days, larval stage 1 to pupal stage 4, e.g., 137cell divisions per hour). This clearly cannot reflectthe real situation even when assuming a greaternumber of origin neuroblasts per mushroom bodysubunit.
Hydroxyurea-induced defects are also differentfrom ablations induced by hydroxyurea describedfor honeybees in this study. The effects of hydroxy-urea in Drosophila depended on the concentrationapplied and ranged from partial to complete mush-room body ablation (de Belle and Heisenberg 1994;Ito et al. 1997). Low hydroxyurea concentrationsled to survival of one or two of the four stem cellsand consequently to smaller mushroom bodies (deBelle and Heisenberg 1994; Ito et al. 1997). In hon-eybees the prevalent effect of hydroxyurea appli-cation was the complete ablation of individualmushroom body subunits, whereas other subunits,in the same brain remained completely intact.Thus, mushroom body subunits in honeybees mustoriginate from a very small number of neuroblastsas in Danaus or, more likely, just from one neuro-blast (see above). Conversely, a greater number ofneuroblasts per proliferation center should havecaused small mushroom bodies like those de-scribed for Drosophila. Such defects, however,could not be found in the honeybee.
In summary, the results presented in this study
MUSHROOM BODY DEVELOPMENT IN THE HONEYBEE
&L E A R N I N G M E M O R Y
99
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
demonstrate that in honeybees each of the fourmushroom body subunits develops postembryoni-cally from a proliferation cluster containing Ken-yon precursor cells that divide in a symmetricalfashion until mid-pupal stages. Cells within thesemitotically active cell clusters derive possibly froma single origin Kenyon cell neuroblast, respec-tively.
The ablation experiments further showed thathydroxyurea treatment of young larvae inducesspecific and precise mushroom body defects.Therefore, this technique provides a promisingtool to study the role of mushroom bodies forlearning and memory in honeybees.
AcknowledgmentsI gratefully acknowledge the financial support of Dr.
Randolf Menzel. Dr. Uli Muller kindly provided theantiserum to PKAII. Dr. Sabine Schafer kindly provided thespecimen shown in Figure 4B. I also acknowledge thecontribution of Dr. Olga Ganeshina in preparing thespecimen shown in Figure 4A. I thank Drs. Randolf Menzel,Sabine Schafer, and Bernhard Zimmermann for constructivediscussion and comments on the manuscript. This work wassupported by a scholarship from the Kommission zurForderung von Nachwuchswissenschaftlerinnen der FreienUniversitat Berlin.
The publication costs of this article were defrayed inpart by payment of page charges. This article must thereforebe hereby marked ‘‘advertisement’’ in accordance with 18USC section 1734 solely to indicate this fact.
ReferencesBarth, M. and M. Heisenberg. 1997. Vision affects mushroombodies and central complex in Drosophila melanogaster.Learn. & Mem. 4: 219–229.
Bertholf, L.M. 1925. The moults of the honeybee. J. Econom.Entomol. 242: 593–605.
Davis, R.L. 1993. Mushroom bodies and Drosophila learning.Neuron 11: 1–14.
Davis, R.L. and B. Dauwalder. 1991. The Drosophila duncelocus: Learning and memory in flies. Trends Genet.7: 224–229.
de Belle, J.S. and M. Heisenberg. 1994. Associative odorlearning in Drosophila abolished by chemical ablation ofmushroom bodies. Science 263: 692–695.
Doe, C.Q. and C.S. Goodman. 1985. Early events in insectneurogenesis: I. Development and segmental differences inthe pattern of neuronal precursor cell. Dev. Biol.111: 193–205.
Durst, C., S. Eichmuller, and R. Menzel. 1994. Developmentand experience lead to increased volume of
subcompartments of the honeybee mushroom body. Behav.Neural Biol. 62: 259–263.
Edwards, J.S. 1969. Postembryonic development andregeneration of the insect nervous system. Adv. InsectPhysiol. 6: 97–137.
Erber, J., T. Masuhr, and R. Menzel. 1980. Localization ofshort-term memory in the brain of the bee, Apis mellifera.Physiol. Entomol. 5: 343–358.
Fahrbach, S.E., T. Giray, and G.E. Robinson. 1995a. Volumechanges in the mushroom bodies of adult honey bee queens.Neurobiol. Learn. & Mem. 63: 181–191.
Fahrbach, S.E., J.L. Strande, and G.E. Robinson. 1995b.Neurogenesis is absent in the brains of adult honey bees anddoes not explain behavioral neuroplasticity. Neurosci. Lett.197: 145–148.
Hahnlein, I. and G. Bicker. 1997. Glial patterning duringpostembryonic development of central neuropiles in the brainof the honeybee. Dev. Genes Evol. 207: 29–41.
Hammer, M. 1993. An identified neuron mediates theunconditioned stimulus in associative learning in honeybees.Nature 366: 59–63.
Hammer, M. and R. Menzel. 1995. Learning and memory inthe honeybee. J. Neurosci. 15: 1617–1630.
Heisenberg, M., A. Borst, S. Wagner, and D. Byers. 1985.Drosophila mushroom body mutants are deficient in olfactorylearning. J. Neurogenet. 2: 1–30.
Hinke, W. 1961. Das relative Wachstum der Hirnteile vonCulex pipiens, Drosophila melanogaster, undDrosophila-Mutanten. Z. Morphol. Okol. Tiere 50: 81–118.
Ito, K. and Y. Hotta. 1992. Proliferation pattern ofpostembryonic neuroblasts in the brain of Drosophilamelanogaster. Dev. Biol. 149: 134–148.
Ito, K., W. Awano, K. Suzuki, Y. Hiromi, and D. Yamamoto.1997. The Drosophila mushroom body is a quadruplestructure of clonal units each of which contains a virtuallyidentical set of neurones and glial cells. Development124: 761–771.
Kenyon, F.C. 1896. The brain of the bee. J. Comp. Neurol.6: 133–210.
Malun, D. 1997. Postembryonic neurogenesis anddevelopment of the mushroom bodies in the honeybee. InProceedings of the 25th Gottingen Neurobiology Conference(ed. N. Elsner and H. Wassle). Thieme, Stuttgart, Germany.
Mauelshagen, J. 1993. Neural correlates of olfactory learningin an identified neuron in the honeybee brain. J.Neurophysiol. 69: 609–625.
Menzel, R., J. Erber, and T. Masuhr. 1974. Learning andmemory in the honeybee. In Experimental analysis of insect
Malun
&L E A R N I N G M E M O R Y
100
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
behaviour (ed. L. Barton-Browne), pp. 195–217. Springer,Berlin, Germany.
Menzel, R., C.D. Durst, J. Erber, S. Eichmuller, M. Hammer,H. Hildebrand, J. Mauelshagen, U. Muller, H. Rosenboom, J.Rybak, S. Schafer, and A. Scheidler. 1994. The mushroombodies in the honeybee: From molecules to behaviour.Fortschritte der Zoologie 39. In Neural basis of behaviouraladaptions (ed. K. Schildberger and N. Elsner), pp. 81–102.Fischer, Stuttgart, Germany.
Mobbs, P.G. 1982. The brain of the honeybee Apis mellifera.I. The connections and spatial organization of the mushroombodies. Phil. Trans. R. Soc. Lond. B. Biol. Sci. 298: 309–354.
Muller, U. 1997. Neuronal cAMP-dependent protein kinasetype II is concentrated in mushroom bodies of Drosophilamelanogaster and the honeybee Apis mellifera. J. Neurobiol.33: 33–44.
Nordlander, R.H. and J.S. Edwards. 1970. Postembryonicbrain development in the monarch butterfly, Danausplexippus plexippus L. III. Morphogenesis of centers otherthan the optic lobes. Wilhelm Roux’s Arch. 164: 247–260.
Panov, A.A. 1957. Bau des Insektengehirns wahrend derpostembryonalen Entwicklung. Rev. Entomol. URSS36: 269–284.
———. 1960. Bau des Insektengehirns wahrend derpostembryonalen Entwicklung. III: Sehlappen. Rev. Entomol.URSS 39: 86–105.
Prokop, A. and G.M. Technau. 1994. Normal function of themushroom body defect gene of Drosophila is required for theregulation of the number and proliferation of neuroblasts.Dev. Biol. 161: 321–337.
Rybak, J. and R. Menzel. 1993. Anatomy of the mushroombodies in the honey bee brain: The neuronal connections ofthe alpha-lobe. J. Comp. Neurol. 334: 444–465.
Skoulakis, E.M.C., D. Kalderon, and R.L. Davis. 1993.Preferential expression in mushroom bodies of the catalyticsubunit of protein kinase A and its role in learning andmemory. Neuron 11: 197–208.
Technau, G. and M. Heisenberg. 1982. Neuralreorganization of the corpora pedunculata in Drosophilamelanogaster. Nature 295: 405–407.
Tettamanti, M., J.D. Armstrong, K. Endo, M.Y. Yang, K.Furukubo-Tokunaga, K. Kaiser, and H. Reichert. 1997. Earlydevelopment of the Drosophila mushroom bodies, braincentres for associative learning and memory. Dev. GenesEvol. 207: 242–252.
Truman, J.W. and R. Booker. 1986. Adult-specific neurons inthe nervous system of the moth, Manduca sexta: Selectivechemical ablation using hydroxyurea. J. Neurobiol.17: 613–626.
Truman, J.W. and M. Bate. 1988. Spatial and temporal
patterns of neurogenesis in the central nervous system ofDrosophila melanogaster. Dev. Biol. 125: 145–157.
Truman, J.W., B.J. Taylor, and T.A. Awad. 1993. Formationof the adult nervous system. In The development ofDrosophila melanogaster (ed. M. Bate and A. Martinez-Arias),pp. 1245–1275. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY.
Withers, G.S., S.E. Fahrbach, and G.E. Robinson. 1993.Selective neuroanatomical plasticity and division of labour inthe honey bee. Nature 364: 238–240.
Witthoft, W. 1967. Absolute Anzahl und Verteilung derZellen im Hirn der Honigbiene. Z. Morphol. Tiere61: 160–184.
Zacharias, D., J.L.D. Williams, T. Meier, and H. Reichert.1993. Neurogenesis in the insect brain: Cellular identificationand molecular characterization of brain neuroblasts in thegrasshopper embryo. Development 118: 941–955.
Received January 28, 1998; accepted in revised form April14, 1998.
MUSHROOM BODY DEVELOPMENT IN THE HONEYBEE
&L E A R N I N G M E M O R Y
101
Cold Spring Harbor Laboratory Press on April 17, 2020 - Published by learnmem.cshlp.orgDownloaded from
