SEARCH

SEARCH BY CITATION

Keywords:

  • memory;
  • learning;
  • insect;
  • mushroom bodies;
  • sensory integration;
  • evolution

Abstract

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED

In most insects with olfactory glomeruli, each side of the brain possesses a mushroom body equipped with calyces supplied by olfactory projection neurons. Kenyon cells providing dendrites to the calyces supply a pedunculus and lobes divided into subdivisions supplying outputs to other brain areas. It is with reference to these components that most functional studies are interpreted. However, mushroom body structures are diverse, adapted to different ecologies, and likely to serve various functions. In insects whose derived life styles preclude the detection of airborne odorants, there is a loss of the antennal lobes and attenuation or loss of the calyces. Such taxa retain mushroom body lobes that are as elaborate as those of mushroom bodies equipped with calyces. Antennal lobe loss and calycal regression also typify taxa with short nonfeeding adults, in which olfaction is redundant. Examples are cicadas and mayflies, the latter representing the most basal lineage of winged insects. Mushroom bodies of another basal taxon, the Odonata, possess a remnant calyx that may reflect the visual ecology of this group. That mushroom bodies persist in brains of secondarily anosmic insects suggests that they play roles in higher functions other than olfaction. Mushroom bodies are not ubiquitous: the most basal living insects, the wingless Archaeognatha, possess glomerular antennal lobes but lack mushroom bodies, suggesting that the ability to process airborne odorants preceded the acquisition of mushroom bodies. Archaeognathan brains are like those of higher malacostracans, which lack mushroom bodies but have elaborate olfactory centers laterally in the brain. J. Comp. Neurol. 513:265–291, 2009. © 2009 Wiley-Liss, Inc.

The present account addresses the following question: what constitutes an insect mushroom body such that its organizational ground plan can be reconciled with current models of its supposed functional properties? The reason why this brain center must be subject to special scrutiny is in large part a result of its acquired reputation as an odorant processor and as a higher brain center that plays a cardinal role in cognition. Common opinion is that mushroom bodies play an indispensable role in olfactory discrimination as well as in olfactory learning and memory (Heisenberg,1998,2003; Perez-Orive et al.,2002; Akalal et al.,2006). The latter is suggested from memory deficits caused by targeted disruption of mushroom body development. by suppression of the cyclic AMP cascade within its neuropils, or by selective inhibition of protein signaling or synaptic transmission in its intrinsic neurons, the Kenyon cells (deBelle and Heisenberg,1994; Zars et al.,2000; McGuire et al.,2001; Ferris et al.,2006). Studies of one neopteran insect in particular, the locust Schistocerca americana, have led to the idea that the mushroom bodies play crucial roles in the encoding and representation of specific odors (Laurent et al.,2001; Cassenaer and Laurent,2007).

However, one problem with these and related studies is that they focus on mushroom bodies that possess a calyx supplied by neurons from the olfactory lobes. The calyx is a distal neuropil of the mushroom body composed of the dendrites of this center's intrinsic neurons (see Fig. 1). In many species of insects, the calyx is supplied with olfactory input from the antennal lobes, but, in some, it also receives relays from the visual system, and, in one species of gyrinid beetle, visual afferents are the exclusive inputs (Strausfeld, unpublished data). The calyx leads to a stalk and a system of lobes that are formed by the axon-like extensions of intrinsic neurons (see Fig. 1). Because for most investigators this organization defines the mushroom body (Flögel,1878; Kenyon,1896; Sánchez y Sánchez,1933), practically all studies over the last 3 decades have been interpreted under the assumption that the calyces alone receive the mushroom body's inputs and the lobes provide its outputs (Menzel,2001). The general view then is that because Kenyon cell dendrites in the calyces are clearly postsynaptic and are the recipients of information (Steiger,1967) their axon-like extensions in the pedunculus and lobes must be overwhelmingly presynaptic, imparting information to efferent neurons (Menzel and Manz,2005; Cassenaer and Laurent,2007). For example, a major role of the mushroom bodies in mediating learned behaviors in the honey bee has been attributed to one type of efferent neuron, called “Pe1,” recordings from which demonstrate a decrease in responsiveness as a result of an acquired association of an olfactory and gustatory stimulus, presumably at the level of the calyx (Mauelshagen,1993; Menzel and Manz,2005; Okada et al.,2007). Similarly, it has been proposed that the response properties of a population of efferents in the locust mushroom bodies, called “β-L neurons,” specifically maintain the synchrony of oscillations that are characteristic of odor representations relayed to the calyx (Cassenaer and Laurent,2007). Models of how mushroom bodies might function as memory centers also stress the afferent supply to the calyces as being of crucial importance in associating olfactory stimuli with a conditioned stimulus, either a reward or punishment, that is delivered by modulatory neurons that globally innervate the mushroom body lobes (Schwaerzel et al.,2002,2007).

thumbnail image

Figure 1. Summary diagram of the principal elements of mushroom bodies in dicondylic insects. Top: Mushroom bodies with paired calyces (Ca) equipped with Kenyon cells from cell bodies known as globuli cells (gl). Kenyon cells (magenta) have dendritic branches in the calyces, which also receive inputs from sensory neuropils (green): diagrammed here are afferents from the antennal lobes that terminate laterally outside the calyx in the lateral horn neuropil of the protocerebrum (L ho). Afferents and efferents (aff, eff; brown) also supply and extend from various locations in the lobes (see Strausfeld,2002). Bottom: Calyxless mushroom body with globuli cells (gl) providing Kenyon cells (Kc) with axon-like branches to the vertical and median lobes (V, M). Inputs (afferent; aff) and outputs (efferents; eff) extend to and from various locations along the lobes.

Download figure to PowerPoint

The widely held notion that the calyces receive the mushroom body's inputs and its lobes provide its outputs is, however, conceptually flawed. Frederick Kenyon in 1896 already proposed that other neurons supply terminals to the lobes (Fig. 1). Four recent studies (Ito et al.,1998; Li and Strausfeld,1999; Strausfeld,2002; Tanaka et al.,2008) unequivocally demonstrate that, in addition to providing outputs (efferents), the mushroom body's lobes receive abundant inputs (afferents) as well as the aforementioned modulatory neurons. Indeed, if there were no such inputs, the observer would be hard pressed to explain why, in a mushroom body whose calyces receive almost exclusively olfactory relays, the efferent neurons from its lobes can respond to modalities other than olfaction, such as acoustic, visual motion, and tactile (Li and Strausfeld,1997,1999). Such recordings have been done extensively in cockroaches, where some 14 unique types of efferent neurons responding to a variety of sensory modalities have been identified morphologically and functionally.

If it is accepted that mushroom bodies in general give rise to efferent neurons that respond to multimodal stimuli, this raises pivotal questions about the basic structure of these centers and thus their functional organization and evolution. If mushroom body calyces generally receive most of their inputs from the antennal lobes, what is the role of this specialized stream of olfactory input with regard to the mushroom body's ability to integrate other types of multimodal information received by its lobes? Furthermore, does the massive innervation of the mushroom body calyx by olfactory afferents in phylogenetically divergent species, such as the locust, cockroach, fruit fly, and honey bee, support the contention that this brain region evolved and continues to function primarily as an olfactory processing center? If so, it would be expected that, first, mushroom bodies initially arose in conjunction with the origin of antennal lobes and their projection neurons. Second, species lacking antennal lobes should also lack mushroom bodies. In support of the first scenario, recent studies of the basal apterygotes Thermobia domestica and Lepisma saccharina reveal mushroom bodies with calyces that receive input from small glomerular antennal lobes (Farris,2005; Schactner et al.,2005; Farris, unpublished). However, classical studies by Hanström et al. (1940) on representatives of the most basal group of insects, the Archaeognatha, show no evidence of mushroom bodies, although descriptions of this group identify antennal lobes. As will be described here, the absence of mushroom bodies but presence of antennal glomeruli has been confirmed in the Archaeognatha by using modern immunofluorescence techniques that can better reveal the detailed structure of the brain. Regarding the notion that insects without antennal lobes should lack mushroom bodies, it was demonstrated over 130 years ago that insects that lack antennal lobes nevertheless possess mushroom bodies (Flögel,1878). Curiously, although the antennal lobes and calyces are reduced or absent in these mostly aquatic species, the mushroom body lobes remain (Fig. 1) and presumably serve processing functions other than the perception of odors (Strausfeld et al.,1998). That such insects are wholly “anosmic” is not, however, proven; they may detect water soluble molecules via gustatory pathways, and a recent account of the antennae of the dragonfly identifies pore sensilla that may represent hygro- and chemoreceptors (Rebora et al.,2008). Finally, in insects that do possess calyces, these structures are remarkably diverse in morphology and across species. Particularly, Hymenoptera and Coleoptera can receive substantial gustatory and/or visual input to their calyces in addition to, or instead of, an olfactory input (Schröter and Menzel,2003; Strausfeld et al.,1998; Farris,2005; Strausfeld et al., unpublished data). Clearly, then, calyces are not dedicated solely to olfactory processing.

This account compares four types of insect brains, representing key phylogenetic positions and behavioral ecologies, to demonstrate that irrespective of whether or not mushroom bodies are equipped with calyces their lobes reveal a common ground plan of organization. The species selected include “typical” terrestrial neopterans with glomerular antennal lobes, aquatic neopteran insects that have secondarily lost their antennal lobes, basal palaeopteran insects that also lack antennal lobes, and the most basal of insects: the primitively wingless apterygotes. The present observations are also compared with published accounts of mushroom bodies in additional species of terrestrial neopterans that span the insect phylogenetic tree.

The present results demonstrate that what unites the mushroom bodies of all insects from the apterygote Zygentoma forward is that the mushroom body lobes comprise systems of parallel axon-like processes that are supplied by the terminals of afferent neurons from protocerebral centers. These synapse onto Kenyon cells, and the Kenyon cells synapse onto the dendritic trees of numerous efferent neurons (Strausfeld et al., unpublished data). We argue that this organization, which represents the Zygentoma + Pterygote ground plan of the mushroom bodies, is independent of the presence of a calyx. This is discussed with respect to orthodox views of mushroom body function and with respect to a plausible scenario of mushroom body evolution.

A NOTE ON TERMINOLOGY

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED

The uniramous deutocerebral appendages of insects are known as the “antennae,” which are homologous to the “first antennae” of crustaceans, which are called “antennules” (Boxshall,2004), which are also homologous to the chelicerae of the chelicerates (Boxshall,2004). The paired appendages that supply the crustacean tritocerebrum are biramous, and the accepted term for these is “antennae”, or sometimes “second antennae” (Boxshall,2004). These appendages have been lost in the insects, although rudiments can appear transiently during embryogenesis (Tamarelle,1984). Thus, the single pair of deutocerebral appendages in insects ought to be termed the “antennules.” However, the unfortunate use of the term “antennae” by insect morphologists has a long tradition and because of this has been retained for this account, bearing in mind that insect antennae and crustacean “antennae” are not homologous.

Intrinsic neurons of the mushroom bodies are usually referred to as “Kenyon cells,” following Howse's (1974) first use of the eponymous term. Here, both terms are used interchangeably. “Extrinsic neuron” or “element” is a generic term referring to afferent and efferent profiles that extend to and from the lobes at different levels.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED

Insects

Adult Machilis germanica (Machilidae, Archaeognatha) were collected from colonies located in damp limestone walls in the gardens of the Botanical Gardens, University of Würzburg. Damselflies and dragonflies (Calopteryx splendens, Libellula depressa, Odonata) and mayflies (Potamanthus luteus, Ephemeroptera) were captured near streams and rivers in the vicinity of Würzburg. Dragonflies (Perithemis tenera, Libellulidae, Odonata) and Japanese beetles (Popillia japonica, Scarabaeidae, Coleoptera) were collected in the Morgantown, West Virginia, region. Individuals of Lepisma saccharina (Zygentoma) were caught locally in Arizona. Small numbers of adult tiger beetles (Cicindela ocellata ocellata, Cicindelidae), soldier beetles (Chauliognathus lecontei), fig beetles (Cotinus mutabilis), and diving beetles (Thermonectus marmoratus [also referred to as Thermonectes marmoratus]) were collected locally in the vicinity of Tucson or at the Willcox Playa and from cattle tanks near Nogales. Terrestrial and aquatic hemipterans (the cicada Diceroprocta semicincta and back-swimmer Notonecta undulata) were obtained in riparian regions near Arivaca, Arizona. Specimens of Notonecta glauca were obtained from pools of the University of Würzburg botanical gardens. Blowflies (Phaenicia sericata) were obtained from breeding colonies in the ARL Division of Neurobiology. Adult Thermobia domestica (Zygentoma) were collected from breeding colonies maintained at 35°C on a 12:12-hour light:dark cycle.

Histology

Golgi impregnations.

We employed the combined Golgi Colonnier and Golgi Rapid procedures described by Li and Strausfeld (1997) and Farris and Strausfeld (2001). This method provides crisp impregnation of neurons in all the brains so treated and is a “ubiquitous” Golgi procedure that requires no further modification. The numbers of neurons impregnated can, however, be increased by a second cycle of osmium-dichromate treatment, followed by a second silver impregnation. The method used is as follows. After opening of the heads of cooled animals submerged in 2.5% potassium dichromate containing 1.3 g sucrose/100 ml solution, fat bodies and trachea are cleaned off the brain's surfaces, and then the brain was removed and placed in a 1:5 admixture of 25% glutaraldehyde and 2.5% potassium dichromate containing 1.3 g sucrose/100 ml (4 days at 4°C), followed by a second chromation (3 days at 4°C) in 2.5% potassium dichromate and 1% osmium tetroxide (99:1). Tissue was washed for 5 seconds in distilled water and then immersed in 0.75% silver nitrate in distilled water or in borax-borate buffer adjusted to pH. 7.2. After 24 hours of impregnation, tissue was dehydrated, embedded in plastic, and serial sectioned. Some brains of T. marmoratus were processed with double impregnation to increase the number of elements impregnated in the mushroom bodies. After the first silver nitrate treatment, tissue was washed in distilled water and placed again in dichromate-osmium for 3 days and then placed in 0.75% silver nitrate.

Reduced silver staining.

Bodian's (1936) original method was used on brains fixed in AAF (85 ml 98% ethanol, 5 ml glacial acetic acid, 10 ml 35% formaldehyde), dehydrated, cleared in terpineol, and embedded in Paraplast Plus (Sherwood Medical, St. Louis, MO). No species-specific modifications were necessary, and all brains on which the method was used stained with equal clarity.

Immunolabeling

It is not the aim of this account to ascribe epitopes to neuropils. Rather, immunolabeling is used here exclusively as a histological label to resolve brain regions, such as the mushroom bodies, by DCO. Brains were labeled with antisera that in Drosophila and the cockroach Periplaneta americana have been demonstrated to reveal specific subdivisions of the mushroom body lobes (Sinakevitch et al.,2001; Strausfeld et al.,2003; Farris et al.,2004; Sjöholm et al.,2005; Farris,2005). After labeling with secondary antibody conjugated to a fluorescent dye, immunolabeled material was observed with a Zeiss Pascal three-line confocal microscope.

Anti-DCO and phalloidin.

Polyclonal antibody (generously provided by Dr. Daniel Kalderon; see Lane and Kalderon,1993) raised against DCO (the catalytic subunit of Drosophila cAMP-dependent protein kinase) reliably labels all Kenyon cell subpopulations in the mushroom bodies of neopteran insects (Farris and Sinakevitch,2003; Farris and Strausfeld,2003; Farris et al.,2004). The anti-DCO polyclonal antibody was generated by cloning the DCO cDNA sequence into a pAR3040 T7 expression vector (Studier and Moffatt,1986); then, after purification of the protein from sedimentation of inclusion bodies, this was used to immunize rabbits (Lane and Kalderon,1993) and the final antibody was purified using DCO protein immobilized on nitrocellulose (Harlow and Lane,1988). For the present study, brains were fixed in situ in 4% paraformaldehyde in phosphate-buffered saline (PBS; from tablets; pH 7.4; Sigma, St. Louis, MO). Dissected out brains were stored in fixative until processing for immunostaining. They were washed in PBS and then embedded in 7% agarose. Agarose blocks were sectioned on a vibratome at 50–70 μm and the sections washed in PBS containing 0.1% Triton X-100 (PBST). Sections were blocked in PBST containing 10% normal goat serum (NGS). The primary antiserum was used at a concentration of 1:1,000 in PBS. After overnight incubation at 20°C, sections were washed in PBST and incubated in Texas red-conjugated goat antirabbit secondary antibody (Molecular Probes, Eugene, OR) at a concentration of 1:500. In the confocal images, anti-DCO labeled neurons are shown red, magenta, or pink-brown. For DCO/phalloidin double staining, Oregon green-conjugated phalloidin (Molecular Probes) was also added at 1:500 concentration. After 12 hours, sections were washed in PBST and cleared in 60% glycerol in PBS for 1 hour, followed by 80% glycerol in PBS for 1 hour. Sections were mounted on slides in 80% glycerol and coverslipped, using clear nail polish to seal the edges of the slide (Rodriguez and Deinhardt,1960). In confocal images, phalloidin labeling is shown green.

Taurine and aspartate.

Antisera against taurine and aspartate (GEMAC, Talence, France) were separately generated in rabbits immunized with L-aspartate or taurine conjugated to poly-L-lysine (MW 60,000) by means of glutaraldehyde and purified by adsorption on the glutaraldehyde-treated protein carriers (Campistron et al.,1986a,b). Affinity and specificity tests were performed on thyroglobulin-linked immunoabsorbent assay (ELISA) and on rat cerebellum (Campistron et al.,1986a,b) to demonstrate negligible cross-reactivity to glutamate, GABA, taurine (for tests of antiaspartate), aspartate (for tests of antitaurine), glycine, and β-alanine. For the present study, brains were fixed overnight in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.3) and then embedded in 7–8% agarose or gelatin/albumin and vibratome sectioned at 60–100 μm. After being washed in PBST, sections were blocked in PBST containing 10% normal swine serum (Dako Corp., Carpinteria, CA). Sections were incubated overnight with aspartate antiserum (1:500) or taurine antiserum (1:500) at 20°C. After washing in PBST, secondary goat anti-rabbit immunoglobulins conjugated to Texas red (1: 250; Jackson Immunoresearch Laboratories, West Grove, PA) or anti-rabbit immunoglobulins conjugated to Alexa 568 (Molecular Probes) were applied overnight. After washing in PBST, sections were mounted on slides and coverslipped under 80% glycerol. Preparations incubated in the absence of primary antiserum showed no staining. Specificity of immunolabeling was controlled as described by Campistron et al. (1986a,b): taurine (Sigma; T9931), L-aspartic acid (Sigma; A-8949), and L-glutamic acid (Sigma; G-6904) were each conjugated to bovine serum albumin (BSA) by glutaraldehyde (Tau-G-BSA, Asp-G-BSA, Glu-G-BSA). After adsorption of taurine antiserum with Tau-G-BSA (10−4 M, concentration with respect to the amino acid), there was no labeling after secondary antibody treatment. Adsorption of taurine antiserum with Asp-G-BSA or Glu-G-BSA did not block staining with the secondary antibody. Aspartate immunostaining was undiminished after incubation of the antiserum with Tau-G-BSA (10−4 M) and Glu-G-BSA (10−4 M), whereas adsorption of aspartate antiserum with 10−4 M Asp-G-BSA abolished immunostaining.

Imaging and reconstructions

Histological preparations were viewed with a Leitz Orthoplan microscope equipped with a Sony DKC 5000 CCD digital camera linked to an Apple G4 computer equipped with graphic software. Images of Golgi-impregnated neurons were reconstructed from stacked optical sections captured at an initial magnification of ×600, employing a planapochromat oil-immersion objective. About 25–50 optical images were captured and layered in register, then made transparent by using the Photoshop darkening function, and shadows were removed section by section before flattening the image. For larger areas, several such reconstructions were seamlessly montaged. This results in images showing each process exactly in focus throughout a depth of 25–50 μm. Immunolabeled images were acquired with a three-line Zeiss Pascal or an Olympus Fluoview 1000 confocal microscope. Line reconstructions of brain regions were made by tracing outlines from serial sections captured with the Sony DKC 5000 CCD digital camera. Three or four images were superimposed as stacks by using fiducial markers such as tracheoles or prominent fiber tracts to maintain orientation. Each digitized stack was printed onto paper, and outlines were drawn in Smartsketch (Futureware Software, Inc., San Diego, CA). Images were imported into Photoshop as TIF files. Shading and hidden line elimination provided plastic images of brains viewed frontally, with the neuraxial ventral surface facing to the front. Optic lobes were omitted from these reconstructions.

RESULTS

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED

“Olfactory” mushroom bodies of terrestrial insects equipped with calyces, as exemplified by coleopterans

Olfaction is a key sensory modality for insects that possess antennae equipped with olfactory sensilla for detecting odors associated with food and conspecifics. Specific odorant receptor neurons that recognize a characteristic set of ligands target a specific antennal lobe glomerulus having a unique identity and position (Chambille and Rospars,1985). This organization, best known from the fruit fly Drosophila melanogaster (Fishilevitch and Vosshall,2005; Couto et al.,2005), is probably ubiquitous across the Neoptera, although genes encoding odorant receptor channels in any one taxon will be distinct from those of another (Robertson and Wanner,2006; Wanner et al.,2007). Typically, each glomerulus provides projection neurons to the mushroom body calyces. This organization typifies terrestrial dicondylic insects.

“Terrestrial” is used here to denote an animal that spends most if not all of its adult life above water and that forages or hunts on land. Here, paradigmatic examples of terrestrial species are coleopterans: the predatory tiger beetle Cicindela ocellata (see Fig. 3A), the herbivorous scarab beetle Cotinus mutabilis (Fig. 2I,J), and the pollen-feeding soldier beetle Chauliognathus leconteis (Fig. 2A–H,K), all of which have mushroom bodies with paired calyces that are supplied by one antennocerebral tract comprising the axons of relay neurons from glomerular antennal lobes (see schematic in Fig. 2A and micrographs in Fig. 2B–D,F–H). The order Coleoptera originated in the Early Permian as the first holometabolous insects and today comprises the largest number of species of any insect order or, for that matter, any animal order (Grimaldi and Engel,2005). Not all Coleoptera are terrestrial, however; some hunt and forage either on or beneath the water surface as adults. Whirligig beetles and diving beetles are two examples; the second here represented by the dytiscid Thermonectus marmoratus.

thumbnail image

Figure 2. Mushroom bodies of terrestrial Coleoptera. AH: Reconstruction (A) from successive reduced-silver-stained sections (B–D,F–H) of the mushroom body of the soldier beetle Chauliognathus lecontei (E). I: Mushroom body of the fig beetle Cotinus mutabilis (J) labeled with anti-DCO to resolve Kenyon cells (magenta) and phalloidin (green), which reveals microglomeruli of the calyces (Ca). In beetles possessing antennal lobes (Ant Lo), projection neurons supply ascending tracts to the calyces (Ca) through the antennocerebral tract (ACT; green in A–D,F–H). In Coleoptera, each calyx is supplied by two identical populations of globuli cells (gl). The two calyces supply axons to the pedunculus (Ped) and the medial (M) and vertical (V) lobes. In some species, such as Chauliognathus lecontei, the concentric organization of each calyx is represented through the length of the pedunculus and its lobes. K: One of the paired glomerular antennal lobes (Ant Lo) of Cotinis mutabilis, with the striate mechanosensory neuropil dorsocaudal to it (Mech S). FB, fan-shaped body of the central complex. Scale bars = 50 μm in B (applies to B–D,F–H); 50 μm in I,K.

Download figure to PowerPoint

thumbnail image

Figure 3. AG: Vertical and medial lobes of comparable size in a terrestrial tiger beetle (A: Cicindela ocellata) and aquatic sunburst and whirligig beetles (D: Thermonectus marmoratus; F: Dineutus sublineatus); beetles show no clear differences in terms of size relative to other brain areas. Large argyrophilic processes indicate “extrinsic” (afferent and efferent) neurons to and from the lobes. B,C,F,G: Golgi impregnations of afferent supply to the lobes of T. marmoratus and D. sublineatus. B: In T. marmoratus both the vertical lobe (cross-section in square brackets) and the medial lobe (C; brackets) receive terminals (highlighted solid arrows) from the protocerebrum. F,G. In D. sublineatus, the vertical lobe (F) and medial lobe (G) receive inputs (highlighted solid arrow) and provide outputs (open arrow). Arrowheads in F indicate Kenyon cell processes. Scale bars = 50 μm.

Download figure to PowerPoint

Regardless of adult feeding preference, terrestrial beetles are endowed with antennae equipped with odorant receptors, the axons of which supply prominent antennal lobes divided into glomerular domains (Hansson et al.,1999; Larsson et al.,2001,2003). Projection neurons from each glomerulus extend axons into the antennocerebral tract and provide systems of collaterals that supply the mushroom body calyces (green outlines, Fig. 2B–D,F–H; see also Larsson et al.,2004; Farris,2008a,b). These projection neurons terminate in the lateral protocerebrum in a region known as the “lateral horn” (Strausfeld,1976). In insect orders such as the Hymenoptera, Lepidoptera, and Diptera, the mushroom bodies are also supplied by a second (outer) antennocerebral tract, which essentially reverses the course of the inner tract by providing collaterals to the lateral horn first and terminating in the mushroom bodies (Tanaka et al.,2004). There is no evidence for an outer antennocerebral tract in the beetles surveyed so far (Larsson et al.,2004; Farris,2008b).

As described from observations of scarabid beetles (Larsson et al.,2004; Farris and Roberts,2005), the coleopteran mushroom body, like that of other neopterans, has two calyces ranging across various species-specific degrees of fusion; from a single neuropil to two completely separate neuropils (Fig. 2I). Regardless of calyx morphology, distinct populations of Kenyon cells provide an equal contribution to the pedunculus and its division into the medial and vertical lobes (Larsson et al.,2004). Reduced silver staining and Golgi impregnation reveal that, in insects such as cockroaches, flies, honey bees, and beetles, afferents enter the lobes from regions of the protocerebrum, whereas the dendrites of efferent (output) neurons project from the lobes to defined protocerebral regions (Li and Strausfeld.,1997,1999; Ito et al.,1998; Strausfeld,2002). The same principal organization of afferents to and efferents from the lobes is observed in the mushroom bodies of terrestrial beetles (Larsson et al.,2004).

In summary, mushroom bodies of terrestrial beetles reflect the ability of these species to detect odors. These beetles have pronounced calycal neuropils that are supplied by collaterals of olfactory relay (projection) neurons. Variations in this basic ground plan occur, however, and are associated with species-specific behavioral ecologies. In feeding generalist scarabaeids, for example (Fig. 2J), separation of the two calyces is pronounced and their neuropils are further elaborated into two discrete zones, each supplied by a distinct type of projection neuron terminal. One zone is supplied from the antennal lobes; the other is supplied by projection neurons that originate in the optic lobes (Farris and Roberts,2005; Farris,2008b).

Mushroom bodies of aquatic coleopterans

It is commonly held that the olfactory supply from the antennal lobes determines the primary functional role of the mushroom body in olfactory processing. However, if the function of the mushroom body were solely that of an olfactory center, it might be expected that an evolutionary absence of olfactory inputs, or the transformation of these to another modality, should result in the diminution or even the absence of the mushroom bodies.

In terrestrial species, the antennae supply two major streams of sensory axons to the brain (Boyan et al.,2002). One carries information about olfactory stimuli to the glomerular antennal lobes of the deutocerebrum (Fishilevich and Vosshall,2005); the other carries mechanosensory afferents from the scapus and pedicellus to striate neuropil of the dorsal lobes (Fig. 2K), regions that have been ascribed by various authors to the tritocerebrum because of their segmental affinities and the passage of their heterolateral axons via suboesophageal commissures (Viallanes,1887; Nishino et al.,2005). Only relays from the olfactory pathway, that is, via the antennal lobes, normally supply the calyces. However, mutants of Drosophila melanogaster, such as Antennopedia, transform the antennae into legs, resulting in a mechanosensory supply to the antennal lobes (Stocker and Lawrence,1981). In these mutant flies, transformation of the antennae into legs does not result in a loss of the mushroom bodies, even though the cellular organization of the antennal lobes more resembles neuropil serving a thoracic appendage rather than one serving an olfactory one (Stocker and Lawrence,1981; Strausfeld, unpublished). This suggests that the determination of mushroom body neuroblasts and their production of the proper Kenyon cell progeny assume the correct morphological fates independent of the afferent supply to the putative calyces, and although no laboratory experiments have succeeded in selectively deleting the antennal lobes to determine whether this affects mushroom body development natural selection had accomplished such experiments by the late Jurassic with the appearance of coleopteran species adapted for life underwater. Species of aquatic beetles thus far sampled appear to have lost their antennal lobes. Slender annulated antennae, which the Gyrinidae (whirligig beetles) use to detect surface vibrations (Bendele,1986), supply mechanoreceptor axons to a striate dorsal lobe neuropil corresponding to the striate dorsal lobe of terrestrial coleopterans, which is likewise supplied by mechanosensory afferents (Fig. 2K). This region is homologous to mechanosensory neuropil of the Drosophila tritocerebrum supplied by mechanosensory neurons from the scapus and pedicellus (Kamikouchi et al.,2006).

If the mushroom bodies are olfactory processing centers, their organization should reflect the absence of an antennal lobe. Indeed, this is so, but only with regard to the appearance of the calyces. Whereas the calyces of aquatic species are very much reduced or in some species even absent (Berger,1878; Flögel,1878), the pedunculus and lobes of aquatic beetles appear to differ little from those of their terrestrial counterparts. They can be as large as in terrestrial ones of comparable brain size (Fig. 3A,D,E) and the lobes receive inputs from the protocerebrum (Fig. 3B,C,F,G), as they do in terrestrial species (Larsson et al.,2004).

Instead of possessing Kenyon cells that are endowed with densely branched dendritic trees that form a discrete calyx, in most aquatic coleopterans Kenyon cells are either devoid of dendrites or give rise to occasional wispy dendrites that extend far out into neuropils of the superior protocerebrum (Fig. 4A). Phalloidin labeling fails to reveal microglomeruli at a distal level that is equivalent to the calyces of a terrestrial species (Fig. 4a1). In terrestrial species, microglomeruli in the calyx (see, e.g., Fig. 2I) represent the sites of synaptic input onto Kenyon cells (Steiger,1967; Yasuyama et al.,2002; Farris et al.,2004; Frambach and Schurmann,2004; Groh et al.,2006; Lent et al.,2007). However, the absence of microglomeruli in T. marmoratus does not mean that Kenyon cells are devoid of all manner of afferent input distally, because, as shown by Golgi impregnations, dendrites extending out into the surrounding protocerebrum are equipped with spiny or claw-like specializations indicative of postsynaptic sites (Fig. 4A).

thumbnail image

Figure 4. Mushroom bodies of insects with secondary anosmia or reduced olfaction. A: Golgi impregnation showing that Kenyon cells of the diving beetle Thermonectus marmoratus (C) provide a loose arrangement of dendrites (arrowed) that extend into the surrounding protocerebral neuropil. DCO labeling (magenta, inset a1) shows two bundles of Kenyon cell neurites (arrowheads) from globuli cells projecting into the protocerebrum without providing a definable calyx or even a remnant that, as in Odonata, would be represented by microglomeruli (green, inset b1). B: Kenyon cells of the mushroom body of the banded demoiselle damselfly Calopteryx splendens (D) provide long (solid arrows) and short (s, open arrows) dendrites; the long ones extend into the surrounding protocerebrum, whereas the short dendrites provide their claw-like specializations at the pedunculus. In the dragonfly Perithemis tenera, four bundles of neurites (arrowheads in inset b1) originate from four sets of globuli cells (labeled pale magenta with DCO in b1). These provide microglomeruli (labeled green with phalloidin, inset b2) within and surrounding the head of the pedunculus, shown in cross-section. Scale bars = 50 μm in A,a1,B,b1; 20 μm in b2.

Download figure to PowerPoint

An exception to the reduced or absent calyx of aquatic beetles is found in the Gyrinidae, such as the whirligig beetle, Dineutus sublineatus. Although its antennae supply only striate mechanosensory neuropils, its mushroom body nevertheless has robust calyces. However, instead of being supplied by olfactory projection neurons, the calyces are supplied by relays from the optic lobe medullas subtending the upper (aerial) compound eyes. In this taxon, the mushroom body calyces have become dominated by a sensory modality that is neither olfactory nor gustatory (Strausfeld, unpublished observations).

In aquatic coleopterans, the long, parallel axon-like processes of Kenyon cells do not differ obviously from those of terrestrial species. Kenyon cells of aquatic beetles, like those of their terrestrial cousins, appear to derive from two adjacent clusters of perikarya even when the calyx is entirely absent (Fig. 4A, inset a1). The two bundles of neurites converge to supply a stout pedunculus, which then divides into a medial and vertical lobe. Along their lengths, the long processes of Kenyon cells are supplied by afferent endings just as they are in the lobes of terrestrial species possessing a defined calyx. Thus, as evidenced by sections cut across equivalent levels and stained by reduced silver (Fig. 3A,D,E), the fibroarchitecture of the lobes of aquatic coleopterans seems as elaborate as that of a terrestrial species. Golgi impregnations reveal inputs to the lobes (Fig. 3B,C,F,G) as they do in terrestrial species. In summary, the lack of an antennal lobe is usually matched by a pronounced reduction or even loss of the calyx but not by decreased elaboration of the pedunculus and lobes.

Mushroom bodies of the Heteroptera: aquatic, terrestrial, and nonfeeding terrestrial species

The Heteroptera are a diverse group of carnivorous and phytophagous species considered to be the sister group of the holometabolous insects (Grimaldi and Engel,2005). As previously described by Pflugfelder (1936), the mushroom body of the terrestrial milkweed bug Oncopeltus fasciatus (Fig. 5A) possesses a large fused calyx formed by two groups of Kenyon cells (and a small accessory calyx supplied by a third). The larger primary calyx is supplied by afferents from the antennal lobe. The parallel processes of the Kenyon cells continue into a typical pedunculus and lobes, the latter decorated with an array of orbicular swellings (Fig. 5A). As in the Coleoptera, several taxa of the heteropteran suborder Hemiptera have independently adopted an aquatic life style. These species are here exemplified by the water strider Aquarius remigis and the water boatmen Notonecta glauca and N. undulata. These insects have also retained the mushroom body lobes, but their calyces have undergone a profound diminution or even loss, as indicated by the complete absence of microglomeruli (Fig. 5B), although in some species the lobes are prominent (Fig. 5C,D,H). Despite the absence of even a vestigial calyx, there is still a trace of an antennocerebral tract that reaches a position at the pedunculus where a calyx would be in a species equipped with antennal lobes (Fig. 5B,E).

thumbnail image

Figure 5. Mushroom bodies of the milkweed bug Oncopeltus fasciatus (A) compared with aquatic and water-adapted hemipterans (BH) lacking antennal lobes and calyces. A–C,F,G show DCO/phalloidin labeling; D,E,H are Bodian stained. A: O. fasciatus mushroom body showing globuli cells (gl), calyx (Ca), and pedunculus (Ped) extending through the lateral protocerebrum (L Pr) and giving rise to a medial lobe (M) equipped with numerous ovoid swellings. B: Notonecta undulata. Distal part of the pedunculus of this back-swimmer supplied by globuli cells. The absence of microglomeruli suggests complete loss of calyx. The diffuse fascicle of axons (brackets) comprises the vestigial antennocerebral tract. C: N. undulata. Medial lobe of mushroom body showing many swollen outgrowths. This contrasts with the medial lobe of the water strider Aquarius remigis, shown in H, which terminates as a single bulb. D: Aquarius remigis. The vertical lobes of the wider strider have two divisions (1,2) flanked by a smaller lobe (asterisk). E: N. glauca. The vestigial antennocerebral tract (brackets) stained by reduced silver. In this species, the mushroom body is simple, arising from one population of globuli cells with a pedunculus ending in a system of ovoid swellings, as shown in N. undulata in C. The mushroom body of N. glauca is reminiscent of the ephemeropteran mushroom body shown in Fig. 6A. F: Diceroprocta semicincta. The pronounced vertical (V) and medial (M) lobes of the cicada mushroom body that lacks a calyx. G: Diceroprocta semicincta. Mushroom body globuli cells (gl), pedunculus (ped), and residual antennocerebral tract (brackets). H: A. remigis. The large bulbous medial lobe of the water strider. Scale bars = 50 μm.

Download figure to PowerPoint

thumbnail image

Figure 6. Comparisons of ephemeropteran (A, a2, a3) and zygentoman (BD) mushroom bodies. A: In the mayfly Potamanthus luteus (a1), mushroom bodies are simple. A single cluster of globuli cells (gl) provides a pedunculus (Ped), the head of which contains an assembly of microglomeruli (green in inset a2, which is a DCO/phalloidin preparation corresponding to the white-outlined box in A), indicative of an attenuated calyx. The pedunculus extends medially to terminate as a few discrete swellings (arrows, also inset a3, a DCO/phalloidin preparation that corresponds to the black-outlined box in A). There is no evidence of a discrete vertical lobe in this taxon. B: Thermobia domestica. Mushroom bodies of the Zygentoma (inset b1, Lepisma saccharina) have a small calyx (Ca) composed of two components: a dense central aggregate of microglomeruli (here revealed by DCO) flanked by larger ones (curved brackets). These are supplied by two streams of axons from the antennocerebral tract. The smaller of the two (straight brackets, left) carries axons from the lobus glomerulatus. The larger tract carries axons of antennal lobe projection neurons. Globuli cells (gl) are organized in two clusters. C,D: Ctenolepisma longicaudata (silverfish) possesses mushroom bodies equipped with a medial lobe from which protrude grape-like swellings called “Trauben” (T) and a prominent vertical lobe (V). Bodian stains also show the calyx (Ca) supplied by a substantial antennocerebral tract (bracket in D). EB, ellipsoid body. Scale bars = 50 μm in A–D; 25 μm in a2.

Download figure to PowerPoint

Within the heteropteran suborder Homoptera, adult cicadas are a famously short-lived terrestrial species, despite spending up to 17 years underground as feeding nymphs (Marshall,2001). As seen in the aquatic Hemiptera, the brain of the adult cicada Tibicen spp. reveals a robust pair of mushroom body lobes (Fig. 5F) provided by a small cloud of Kenyon cell somata. Despite the complete lack of microglomeruli and thus the absence of a calyx (Fig. 5G), the most distal portion of the pedunculus is intersected by a vestigial but recognizable antennocerebral tract, as described above for the aquatic Notonecta (Fig. 5B,E). The tract comprises a small bundle of neurites originating from perikarya that would, in a species equipped with an antennal lobe, normally provide systems of dendrites to the antennal lobe's glomeruli.

Mushroom bodies in Ephemeroptera and Odonata

Until very recently, it was thought that extant bristletails (Archaeognatha), which are reminiscent of certain late Devonian fossils, represent the earliest terrestrial insects (Labandeira et al.,1988). However, the age of the first insect might have to be placed significantly earlier as a result of a contemporary reinterpretation by Engel and Grimaldi (2004) of a fossil insect called Rhyniognatha hirsti, originally discovered in the 1920s in Rhynie Chert, which is dated to the early Devonian. The mouth parts and associated apodemes of this specimen are more similar to those of primitive winged insects than to those of any archaeognathan. Apart from the exciting implication that winged insects had already evolved by 410 million years ago, the structure of this animal's jaws suggests affinities with the Odonatoptera, a group of insects that have aquatic naiads (larvae) and winged imagoes. This group includes today's damselflies and dragonflies (Odonata). These Palaeoptera are basal taxa that are additionally characterized by possessing reduced “setaceous” antennae that supply axons to mechanosensory neuropils. Both damselflies and dragonflies lack multiglomerular antennal lobes, as do their naiads, yet their brains have prominent mushroom bodies (Svidersky and Plotnikova2004; Farris,2005).

Golgi impregnations, reduced silver stains, and immunolabeling show no evidence for the presence of a neopteran-like calyx in either the Odonata (Fig. 4B, insets b1,b2; see also Fig. 7A) or the Ephemeroptera (Fig. 6A), the most basal group of winged insects. Nevertheless, there are some interesting differences between these two groups. In the Ephemeroptera, each mushroom body consists of a slender pedunculus that provides several bulbous structures at its base (Fig. 6A, inset a3). In apterygote zygentomans (see, e.g., Fig. 6, inset b1), the mushroom bodies lobes give rise to similar bulbs or “Trauben” (Farris,2005), exemplified in Figure 6C,D from Ctenolepisma longicaudata. In the ephemeropteran P. luteus (Fig. 6A), an extension of the pedunculus toward the midline gives rise to tuber-like components of the medial lobe that are situated beneath a simple two-layered central complex (Fig. 6A, see also inset a3). These swollen elements of the medial lobe (Fig. 6A) are penetrated by axons that branch within them, suggesting both inputs and outputs at these locations. Anti-DCO and phalloidin staining demonstrates that the mayfly mushroom body possesses a cluster of microglomeruli (Fig. 6A, inset a2) localized within the most distal level of the pedunculus at a level where, in terrestrial neopterans, the calyx is normally located. The presence of microglomeruli in P. luteus indicates that afferent input of some kind is targeting Kenyon cells at this level.

thumbnail image

Figure 7. Mushroom body of the dragonfly Libellula depressa. A: Reduced-silver section showing two of the four bundles of neurites (arrows) arising from several thousand globuli cells and supplying the two necks (ne) of the pedunculus (ped). B: Top-down view showing the two populations of globuli cells, each supplying half of the mushroom body. The two necks of the pedunculus (ne) are surrounded by a swathe of protocerebral neuropil (Pro). There is no evidence in silver stains of a discrete calyx, but antisera against phalloidin reveal microglomeruli (Fig. 3B, inset b2). C: Golgi impregnation of a few dozen Kenyon cells supplying one pedunculus neck and then extending through the pedunculus and medial lobe. DF: Enlargements of boxed areas in C show details of Kenyon cell dendrites that extend into the protocerebrum from the distal pedunculus (D) or that arise at several levels along the length of the pedunculus and medial lobe (E,F). Scale bars = 50 μm.

Download figure to PowerPoint

In contrast to those of the Ephemeroptera, mushroom bodies in the Odonata are massive structures (Figs. 4B, 7A–F). Their intrinsic neurons derive from twinned clusters of globuli cells that provide two necks that converge to form a single pedunculus (Fig. 7A,B). As observed in neopteran insects, but apparently lacking both in the Ephemeroptera and possibly in the primitive apterygote Thermobia domestica (Farris,2005), the odonate pedunculus and its vertical and medial lobes are divided into parallel divisions. These can be separately resolved using appropriate immunocytological markers that demonstrate their similarity to divisions identified in neopteran insects, such as the cyclorrhaphan flies Drosophila melanogaster (Strausfeld et al.,2003) and Phaenicia sericata (Fig. 8). This degree of similarity suggests that parallel divisions of the mushroom bodies, composed of distinct Kenyon cell subpopulations that are generated sequentially by mushroom body neuroblasts (Lee et al.,1999), originated in the Palaeoptera and have been conserved in the Neoptera as part of the basal ground plan of this neuropil. Because the divisions in Aeschna correspond to similar parallel divisions first identified in the lobes of flies, it is convenient to use the same terms (α, α′, β, β′, γ) to indicate comparable elements (Fig. 8). The parallel organization of divisions in the odonate lobe is further confirmed by reduced silver preparations, which reveal three domains throughout the length of the pedunculus and lobes. Each domain is characterized by its distinctive arrangement of fibers and extrinsic elements (Figs. 9A–C, 10A), the latter again revealing the substantial supply to the lobes by afferent neurons (Fig. 10B).

thumbnail image

Figure 8. Parallel divisions of the mushroom body lobes. In neopteran insects, here exemplified by Phaenicia serricata (B,C,E), mushroom bodies consist of several discrete parallel divisions. Antisera against aspartate, taurine, and DCO distinguish divisions of the vertical and medial lobes (B,C,E). In odonate insects, here exemplified by the dragonfly Aeschna, the same terms (α, α′, β, β′, γ) can be used to indicate comparable subdivisions in the lobes (A,D,F inset, H). In the fly, intense DCO immunoreactivity identifies the lobelet (E), whereas, in Aeschna, an intense DCO-positive structure at the bulb of the vertical lobe (D) indicates a vertical branch of the γ-lobe. As in the fly (B), aspartate reveals two parallel divisions in the odonate mushroom body, best seen in cross-sections of the pedunculus (inset to F). Aspartate labels the head of the odonate pedunculus (brackets, F) and the fly calyx (G). Taurine identifies the α′ division in the fly (C) and a comparable swelling at the tip of the odonate vertical lobe (asterisk in H). Scale bars = 50 μm.

Download figure to PowerPoint

thumbnail image

Figure 9. Parallel divisions of the mushroom body of the odonate Libellula depressa. A: Cross-sections of the pedunculus distinguish three parallel divisions, each of which will provides the corresponding components of the vertical (α, α′) and medial (β, β′) lobes, with the γ division contributing to both. Note that each division has its own characteristic fibroarchitecture and extrinsic (input and output) elements. B: Afferent neurons invading the base of the pedunculus. C: The junction of the vertical and medial lobes. Afferent neurons extending to the α′/β′ (double arrows) shown with the initial branches of two large efferent neurons arising from the α/β and γ divisions (single arrows). Scale bars = 50 μm.

Download figure to PowerPoint

thumbnail image

Figure 10. Parallel and local organization of the lobes of Libellula depressa. A: Golgi impregnations demonstrate that, at specific locations along the lobes, Kenyon cells provide collaterals (brackets) indicative of regions in which they participate in local circuits. B,C: Reduced silver impregnations reveal the participation of efferent (single black arrow in B, white arrows in C) and afferent (double (white and black arrows in B) neurons at the bulbous ending of the vertical (α) lobe. The spur (sp) indicates the peduncular divide at which Kenyon cells divide to send collaterals into the medial and vertical lobes. Scale bars = 50 μm.

Download figure to PowerPoint

Studies on preemergent odonate brains, with the vital stain methylene blue, have proposed that the most distal part of the aeschnid mushroom body receives afferents from the optic lobe (Svidersky and Plotnikova,2004). Evidence supporting a distal area that is morphologically distinct as a zone receiving an afferent supply is suggested by DCO/phalloidin staining. DCO-positive bundles of processes originate from clusters of globuli cells and extend through the pedunculus (Fig. 4B, inset b1). At a position just beneath where the bundles of Kenyon cell neurites merge to form the pedunculus proper, phalloidin labeling reveals punctate microglomeruli (Fig. 4B, inset b2) that correspond to those typifying the calyces of terrestrial neopterans (Steiger,1967; Yasuyama et al.,2002; Farris et al.,2004; Frambach et al., 2004; Groh et al.,2006; Lent et al.,2007; Mashaly et al.,2008). As demonstrated via Golgi impregnations (Figs. 4B, 7A), Kenyon cells in odonates also provide a very loose arrangement of dendrites that extend into the surrounding protocerebral neuropil, where they may receive afferent relays from the visual and other sensory neuropils. Dendritic branches also originate along the length of Kenyon cell processes in the pedunculus and extend sometimes for scores of micrometers out from the pedunculus into other protocerebral regions (Fig. 7D–F). This arrangement is similar to that observed in aquatic coleopterans, where dendritic extensions from Kenyon cells likewise reach into the surrounding protocerebrum (Fig. 4A).

Antennal lobes and mushroom bodies in apterygotes

There are two extant lineages of primitively wingless, or apterygote, insects. These are the Zygentoma, which comprise the firebrats and silverfish, and the Archaeognatha or bristletails. Recent phylogenies have placed the Zygentoma as a sister group of the pterygote insects (Paleoptera and Neoptera; Mendes,2002; Giribet et al.,2004; Grimaldi and Engel,2005), with the Archaeognatha basal to this grouping (Hovmöller et al.,2002; Carapelli et al.,2007). Thermobia domestica, a zygentoman of the family Lepismatidae, has recently been demonstrated to possess mushroom bodies with all of the elements of those of the terrestrial Neoptera (Farris,2005; see also Fig. 6B–D). The mushroom body of T. domestica consists of two groups of Kenyon cells, the dendrites of which supply a single calyx. Their axon-like processes extend through the pedunculus into medial and vertical lobes, where they provide numerous branches that constitute bulbous Trauben as occurs in other species in this order, such as the Australian Ctenolepisma longicaudata (Fig. 6C,D). However, it is not possible to discern parallel divisions in the medial lobes of Zygentoma that in Neoptera reflect the sequential neurogenesis of type I–III Kenyon cell populations (Fig. 8; see also Farris and Sinakevitch,2003). Nevertheless, the vertical lobe of Ctenolepisma longicaudata terminates as two swellings reminiscent of the α and α′ lobes of the cyclorrhaphan mushroom body (see Fig. 8E, or their odonate equivalent, Fig. 8D).

The Zygentoma possess antennal lobes. These glomerular neuropils are small relative to the layered mechanosensory neuropils that are also supplied by the antennae (Schactner et al.,2005). Nonetheless, afferents from the antennal lobes of Thermobia project to the mushroom body calyx and from there to the lateral protocerebrum via a tract that is equivalent to the inner antennocerebral tract of neopterans (Fig. 6B,D; see also Farris,2005,2008a). A parallel tract originating in the lobus glomerulatus (see Fig. 13) also carries afferents to an accessory calyx (curved brackets, Fig. 6B), where it supplies just two prominent microglomeruli of the ventral calyx. Inputs from the lobus glomerulatus to the calyces via an equivalent tract are also observed in the Orthoptera (Weiss,1981; Frambach and Schurmann,2004). Studies of maxillary palp inputs to this neuropil in the Orthoptera (Blaney and Chapman,1969; Ignell et al.,2000) suggest that its ascending tract carries gustatory information to the calyx. It is likely that in zygentomans the calyces similarly receive a combination of gustatory and olfactory information. In conclusion, the mushroom bodies and their inputs in the Zygentoma are reminiscent of those of orthopterans.

Kenyon cells have a peculiarly high affinity for anti-DCO antibody (Farris,2005). However, monocondylic insects, here represented by the archaeognathan Machilis germanica (Fig. 11A), are unlike any other insect in that they uniquely lack any evidence of mushroom bodies, including the lobes. Anti-DCO immunostaining on three individuals failed to reveal any neurons in the brain that were labeled by this procedure, contrasting with the anti-DCO affinity of Kenyon cells in all other insects surveyed. Phalloidin immunolabeling reveals that the deutocerebrum of M. germanica looks very much like that of Thermobia. It consists of a glomerular neuropil, supplied by the antennal nerve, lying immediately rostral to a larger striate neuropil that is situated at the level of the tritocerebrum and supplied by the mechanosensory axon bundle of the antenna (Fig. 11C). A medial glomerular neuropil is innervated by a nerve arising from more posterior segments and thus likely represents the tritocerebral lobus glomerulatus. These results suggest that the acquisition of glomerular olfactory lobes, and thus the ability to process airborne odorants, occurred prior to the acquisition of mushroom bodies. It is also unlikely that the mushroom bodies were secondarily lost in Machilis. As we demonstrate here, in other species that have lost their calyces, the mushroom body lobes always persist even in species that lack antennal lobes and that have very short adult lives. Neither condition applies to archaeognathans. However, although there is no evidence for mushroom bodies in Machilis, the lateral horn of the protocerebrum is pronounced (S L Pr in Fig. 11B). This suggests that in Machilis, as in other insects (Homberg et al.,1988), this part of the protocerebrum is supplied by relays from the antennal lobes and is likely to be homologous to regions of the lateral protocerebrum that in malacostracan crustaceans receive the terminals of projection neurons from the antennular lobes (Sullivan and Beltz,2001).

thumbnail image

Figure 11. An insect without a mushroom body, the archaeognathan Machilis germanicus (A). B: DCO/phalloidin shows no evidence of a mushroom body in the archaeognathan brain. However, there is a clearly delineated superior lateral protocerebrum (S L Pr), the organization of which is reminiscent of the medulla terminalis of a malacostracan crustacean or the lateral horn olfactory neuropil of other insects. C: The machilid antenna provides discrete olfactory glomeruli (olf glom) and a more caudal antennal mechanosensory neuropil (ant mech). Gustatory neuropil (gust) is also glomerular and is situated in the tritocerebrum. Note the simple layered arrangement of the central complex (CC), again reminiscent of that of malacostracan crustaceans. Me, medulla; Lo, lobula; S M Pr, superior medial protocerebrum. Scale bars = 50 μm.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED

Resolving the role of the insect mushroom bodies requires more than a study of three “model” species, in which the dominant position of the olfactory pathway detracts from considerations about other aspects of organization that may have nothing at all to do with odor perception. A case in point is the gyrinid mushroom body, where the calyces have been entirely commandeered by the visual system (Strausfeld, unpublished). Clues to the overall function of the mushroom bodies are likely to emerge from comparative studies that consider Kenyon cell circuits in the most basal insect lineages (barring those species that have since acquired highly derived behavioral ecologies). To what degree are circuits in the lobes ubiquitous? Although this question will be considered in a forthcoming study (Brown et al., unpublished), the present discussion will consider the ground plan of the mushroom bodies, their evolution, and their likely function.

What is the functional identity of the mushroom bodies?

That the mushroom bodies are olfactory processing centers has been advocated primarily in studies on the orthopteran, Schistocerca americana, in which subsets of Kenyon cells are thought to serve as coincidence detectors that encode odor identity (Perez-Orive et al.,2002; Cassenaer and Laurent,2007). An objection to this view is that such studies cannot legitimately generalize across insects, because it is unclear whether this and closely related taxa are representative even of the Orthoptera, because the locust's antennal lobes are unusual in that they are aglomerular. Each olfactory receptor axon diverges to many small glomeruli and thus several projection neurons (Ignell et al.,2001; Anton et al.,2002). In most insects, all olfactory receptor neurons carrying a given receptor protein converge on a single specific glomerulus, from which information is then transmitted to the mushroom bodies via a small number of mostly uniglomerular projection neurons (Bhalerao et al.,2003; Tanaka et al.,2004). However, there is evidence for coincidence detection and signal “sparsening” in the mushroom bodies of the honey bee Apis mellifera and the fruit fly Drosophila melanogaster, both of which have conventional olfactory glomeruli of the latter type. These studies therefore support the view that mushroom bodies indeed encode precise information about the identity of odors (Wang et al.,2004; Szyszka et al.,2005).

Are, then, the mushroom bodies primarily olfactory centers, or might they play a broader role in multisensory integration? This question is of central concern because it is unlikely that mushroom bodies evolved concurrently with the acquisition of airborne odor perception. This is implied by the absence of mushroom bodies in the Machilidae despite the presence of olfactory glomeruli. In the Machilidae, relays from the antennal lobes supply an extensive neuropil that makes up a large volume of the lateral protocerebrum. This area is relatively much larger than that of the lateral horn, a delineated dorsolateral region of the lateral protocerebrum that receives endings from the antennal lobes of dicondylic insects (Tanaka et al.,2004) and where the organization of terminals of antennal lobe projection neurons best manifest odortypic representation (Tanaka et al.,2004; Lin et al.,2006; Jefferis et al.,2007). Taken together, the protocerebral termination area for antennal lobe relays in monocondylic insects and the smaller lateral horn of the Dicondylia are most reminiscent of neuropils in the malacostracan lateral protocerebrum (the medulla terminalis and hemiellipsoid body), in which olfactory relay neurons from the antennular lobes terminate (Sullivan and Beltz,2001,2005).

The mushroom body ground plan and its elaboration in the Neoptera

The present results demonstrate that many species of neopteran insects possess mushroom bodies that either lack calyces or have drastically reduced calyces, a deficit that correlates with the absence of antennal lobe glomeruli (Fig. 12). Such calyxless mushroom bodies occur mostly in aquatic species secondarily lacking antennal lobes. These species provide the results of an experiment that cannot as yet be performed in the laboratory: the selective ablation of primary olfactory neuropil. Calyxless species include diving coleopterans and aquatic, as well as surface, hemipterans: two examples are the water boatmen (e.g., Notonecta glauca) and water striders (e.g., Gerris remigis), the latter relying on mechanosensory signals for prey and conspecific detection, orientation, and courtship (Murphey,1971; Lang,1980), all behaviors that in many terrestrial species are driven by odor perception. Notwithstanding that sensory axons from the antennae of these hemipterans supply exclusively striate mechanosensory neuropils, these taxa possess large and elaborate mushroom bodies that lack calyces but have elaborate lobes.

thumbnail image

Figure 12. Comparisons of mushroom bodies in insects lacking a calyx and antennal lobes with an insect (Cicindela ocellata) equipped with both. Mushroom body neuropils are shown in green, and globuli cell clusters supplying them are magenta. Among the six midbrains shown here, only that of the tiger beetle Cicindela ocellata possesses antennal lobes (orange) and an ascending tract supplying calyces. The mushroom body of basal palaeopterans (exemplified by the damsel fly Agrion virgo and mayfly Potamanthus luteus) possesses lobes; those of the odonate are substantial. Secondarily aquatic neopterans, here the sunburst beetle Thermonectus marmoratus and the hemipteran Notonecta undulata (back-swimmer), have calyxless mushroom bodies, the latter showing tuberous outgrowths from the base of the pedunculus that are reminiscent of those in Potamanthus (see Fig. 6A). The terrestrial homopteran, the cicada Diceroprocta semicincta, lacks antennal lobes but possesses calyxless mushroom bodies.

Download figure to PowerPoint

At least one group of exclusively terrestrial neopterans lacks antennal lobes and calyces. These are the cicadas, the antennae of which are minute and predominantly mechanosensory, and in whose brains antennal lobes are conspicuously absent. However, cicadas possess mushroom bodies with substantial medial lobes, albeit foreshortened vertical lobes (Diceroprocta; Fig. 12). The cicadas are short lived, do not feed as adults, and are notable for their use of auditory rather than olfactory cues for mate location. This truncated and specialized adult life history may explain the loss of the antennal lobe and concomitant loss of the calyx.

The second group of insects with calyxless mushroom bodies are the two ancient orders of Palaeoptera: the Ephemeroptera, comprising the mayflies, and the Odonata, comprising the damselflies and dragonflies. The latter possess substantial protocerebra and are equipped with enormous optic lobes but lack glomerular antennal lobes and calyces. Phalloidin labeling reveals copious microglomerular-like structures in the distal pedunculus, indicating substantial afferent input to this region that is likely to originate from the optic lobes (Svidersky and Plotnikova,2004). These microglomeruli are clustered within and around the passage of Kenyon cell processes at a level in the mushroom body that, in other species, would provide the calyces. As demonstrated here, odonate mushroom bodies can be huge (Agrion; Fig. 12), comprising many thousands of intrinsic neurons and subdivided into divisions that are reminiscent of neopteran mushroom bodies.

Among the Neoptera, the most substantial mushroom bodies with respect to the number of constituent neurons are seen in both social and solitary Hymenoptera. Nevertheless, the mushroom bodies of the honey bee Apis mellifera are thus far the best studied (Strausfeld,2002), and their elaboration represents one extreme of mushroom body evolution, in which the calyces are polymodal, supplied by afferents from several sensory modalities. At the other extreme are the mushroom bodies of Ephemeroptera. These are exemplified by the mushroom body of the mayfly Potamanthus luteus, which lacks a calyx and originates from a small cluster of globuli cells. In contrast, the honey bee calyx is divided into 13 discrete concentric zones, each of which is supplied by a characteristic arrangement of terminals that carry olfactory, gustatory, or visual information and invested with a characteristic ensemble of Kenyon cell dendrites (Strausfeld,2002; Ehmer and Gronenberg,2002; Schröter and Menzel,2003). Kenyon cells associated with a particular zone send their parallel extensions into a corresponding stratum that extends for the length of the pedunculus and lobes. As shown by Golgi studies, each stratum receives afferents (inputs) that are restricted to it, and each stratum contains layered dendrites of efferent (output) neurons that extend from it to terminate in the protocerebrum (Strausfeld,2002). In essence, the honey bee mushroom body is not a single isomorphic ensemble of Kenyon cells but is a massively parallel structure that is divided into numerous subunits, each served by its own arrangement of inputs and providing its own outputs. Accordingly, the collar zone of the mushroom body calyx, which receives visual inputs from the optic lobes (Ehmer and Gronenberg,2002), supplies Kenyon cell processes to five parallel strata of the mushroom body's lobes, whereas the calyx lip, which receives inputs from three clusters of antennal lobe glomeruli (Müller et al.,2002; López-Riquelme and Gronenberg,2004), supplies three other parallel strata of the lobes (Strausfeld,2002).

This organization at first sight seems to be very different from the simple mushroom bodies of the Ephemeroptera. In Potamanthus, there is no discrete calyx, and the pedunculus is not divided into parallel divisions. Its lobe merely provides several bulbous outgrowths, each of which additionally receives afferents from the protocerebrum, and each presumably supplies outputs back to the protocerebrum. This morphology is reminiscent of the “Trauben” formed by Kenyon cell branches in the lobes of the apterygote Thermobia domestica, although the latter does possess both a main and an accessory calyx (Farris,2005), the latter supplied by gustatory afferents from the lobus glomerulatus. The Ephemeroptera are acknowledged as the most “primitive” of extant winged insects, the fossil record for which extends back to the early Permian (Grimaldi and Engle,2005). However, although it is tempting to equate the simplicity of the ephemeropteran mushroom body with the basal position of this taxon, it is once again necessary to consider the highly derived life style of adult mayflies. As with the above-mentioned cicadas, mayflies perish just a few hours after adult emergence, which occurs en masse at streams and rivers. There are no specialized mate- or oviposition-locating mechanisms; adults also do not feed, so like cicadas they have little use for olfactory cues or higher brain centers for more complex sensory processing. The extremely abbreviated adult life of mayflies (Carey,2002) has likely favored the reduction of the mushroom bodies in this taxon, so the simplicity of these structures may not be entirely attributed to a primitive character state.

The question remains, however, of how a simple mushroom body, such as exemplified by Potamanthus, might have given rise to the kind of elaborate organization seen in many hymenopterans. One obvious hint is provided by the developmental history of these centers and the unusual properties of mushroom body neuroblasts. Usually, a neuroblast provides a small number of neurons after producing several ganglion mother cell progenitors. In Drosophila, each member of a quartet of neuroblasts (two to each presumptive calyx) produces ganglion mother cells that generate about 500 daughters, the Kenyon cells (Ito et al.,1997). The two fused calyces are each composed of the dendrites of about 1,200 Kenyon cells. This is numerically trivial compared with the honey bee or species of ants, in which neuroblasts associated with mushroom bodies behave differently. In the latter, each of the original quartet of neuroblasts (Ishii et al.,2005) undergoes many divisions to provide many smaller neuroblasts that subsequently provide hundreds of ganglion mother cells and hence an enormous number of Kenyon cells (see also Farris et al.,1999). In honey bees, about 175,000 Kenyon cells are generated for each mushroom body in which the two calyces are clearly separate (Witthöft,1967). Drosophila and Apis thus represent two extremes, but Kenyon cell numbers appear to have been modified many times during insect evolution (Farris and Sinakevitch,2003; Urbach and Technau,2003). The ease with which neuroblast number, and thus Kenyon cell number, appears to respond to adaptive pressure provides a solid mechanism for the observed variation in mushroom body size across the dicondylic insects.

It is not currently known how differences in gene expression produce evolutionary differences in neuroblast number, but it is likely that proteins regulating symmetric vs. asymmetric cell divisions, and thus neuroblast vs. neuron-producing cell divisions, are important. The product of one such gene in Drosophila, called mushroom body defect, functions in the alignment of the mitotic spindle. Mutations of this gene result in a tenfold increase in mushroom body neuroblast number and an enormous increase in mushroom body size (Guan et al.,2000; Bowman et al.,2006; Siller et al.,2006). The mechanisms by which morphologically and functionally distinct Kenyon cell subpopulations have arisen from an ancestrally uniform population remain elusive in large part because the molecular determinants of each subpopulation during the ontogeny of an individual fly are poorly understood at present. There is new evidence, however, that progressively decreasing concentrations of one transcription factor, chinmo, may regulate neuroblast production of larval-born subpopulations (Zhu et al.,2003; Yu and Lee,2007).

Evolution of the mushroom body

The concomitance of calyces and antennal lobes at first sight suggests that these two neuropils might have evolved in tandem with the acquisition of odorant-detecting sensory neurons. One counterargument is that the antennal lobe was already present when the first insects colonized land, but the mushroom body was not. Alternatively, the first insects might, from the outset, have been endowed with the full panoply of odorant receptors, with the antennal lobes and calyces being elements of the basal ground plan of the insect brain. However, the brains of Machilidae, the most basal living lineage of insects, speak against this. Immunohistochemical labeling shows that the Machilis brain possesses glomerular antennal lobes (that is, strictly the antennular lobes; see under Terminology, above), but no mushroom body. Hence olfactory lobes were present in insects before the acquisition of mushroom bodies.

True insects (Archaeognatha + Dicondylia) are likely to have arisen from a crustaceomorph ancestor. Crucially, no extant crustacean (Entomostraca + Malacostraca) possesses paired mushroom bodies, and the crustacean brain thus shares greater affinity with that of the monocondylic Machilidae than with the brain of dicondylic insects. It is the dicondylic apterygote Zygentoma, species such as firebrats and silverfish, whose simple detritivorous life styles are likely to be similar to those of the earliest insects, that may serve as the best representatives for what the earliest insect mushroom bodies looked like.

Recent phylogenetic studies resolve the Zygentoma as a sister group of the Pterygota (Giribet et al.,2004; Grimaldi and Engel,2005), and both groups have dicondylic mouth parts. They are thus distant from the monocondylic Archaeognatha, whose lack of a mushroom body suggests a very ancient status close to the common ancestor of the Malacostraca. The calyces of the zygentoman apterygotes Lepisma and Thermobia receive inputs from an ascending tract that originates from a small cluster of glomeruli supplied by the ipsilateral antenna and that lies just rostral to striate mechanosensory neuropil (Farris,2005). The mushroom body of Thermobia also receives gustatory input from the ipsilateral lobus glomerulatus, a neuropil that receives mechanosensory and gustatory input from the maxillary palp (Farris,2008a). Lobus glomerulatus projection neurons ascend to the protocerebrum, where they invade two enlarged microglomeruli in the ventral calyx. In this respect, the inputs to the zygentoman mushroom body are in essence similar to those of the orthopteran mushroom body, exemplified by crickets and locusts (Frambach and Schurmann,2004), in which the smaller of two calyces (the accessory calyx) receives afferents from the lobus glomerulatus, whereas the larger calyx (the primary calyx) is supplied from the antennal lobe (Fig. 13).

thumbnail image

Figure 13. Similarities between flightless and winged dicondylian insects, here the zygentoman apterygote Lepisma saccharina (“silverfish”, left) compared with the orthopteran Schistocerca gregaria (“American desert locust”, right). In both taxa, mechanosensory terminals from the antennae (AN) supply the tritocerebrum, and olfactory receptors from the antennae supply axons to the antennal lobe in the deutocerebrum. In both, the lobus glomerulatus receives gustatory inputs and, like the antennal lobe, provides a separate stream of projection neuron axons to a discrete region of the calyx (Ca I, accessory calyx). Gustatory and olfactory projection neuron axons appear to extend beyond their specific calyx regions (Ca I, Ca II) into the lateral horn (LH) of the superior lateral protocerebrum. gc, Globuli cells; M, medial lobe; V, vertical lobe.

Download figure to PowerPoint

Do the Zygentoma represent the earliest appearance of mushroom bodies? Here we have to be cautious. In addition to the absence of a deep fossil record for the Zygentoma (the earliest known fossils are relatively recent, from the Lower Cretaceous; Sturm,1998), several neurological features speak against this group being primitive. Its central complex is equipped with a fan-shaped body, superior arch, and ellipsoid body, along with a protocerebral bridge. These features typify the central complex of crown palaeopterans and neopterans and differ from the simple bilayered central body of machilids and ephemeropterans. Another curious aspect of zygentomans is exemplified by the silverfish Ctenolepisma longicaudata, in which the developing meso- and metathoracic ganglia have precisely the same numbers and locations of neuroblasts as in a neopteran insect such as the locust. Moreover, the sequence in which they segregate from the neuroectoderm is the same in both taxa (Truman and Ball,1998). Either this suggests that the entire complement of thoracic neuroblasts was present in the insect ventral nerve cord prior to the evolution of flight, as argued by Truman and Ball, or it may suggest the retention of a developmental program associated with flight in a secondarily flightless insect allied to the Neoptera. Although the suggestion that zygentoman insects are secondarily flightless conflicts with molecular sequence data (Regier et al.,2004), the neuroanatomical observations summarized above could support the divergence of the Zygentoma from within the neopteran stem lineage. Despite the drastic reduction of the visual system, the zygentoman brain shares more similarities with the brain of Neoptera than with the other group of apterygotes, the Archaeognatha, or with the winged Ephemeroptera.

Functional implications of the mushroom body ground plan

Neuroanatomical observations support a mushroom body ground plan in the terrestrial Dicondylia comprising parallel arrangement of Kenyon cells that possess distal postsynaptic sites in a specialized component called the “calyx” and whose axon-like prolongations provide pre- and postsynaptic sites along the length of the lobes. However, if we take into consideration taxa that have secondarily lost all vestiges of their calyces, including the complete loss of microglomeruli at the level where a calyx is present in other taxa, then the ground plan that would apply to all dicondylic insects is one that comprises the pre- and postsynaptic axon-like processes of Kenyon cell divested of their distal dendrites (Fig. 14). What unites all neopteran taxa is that these axon-like processes contribute to circumscribed domains down the length of the lobes, each of which comprises a system of local circuits that integrate afferent information projected into the lobes from protocerebral neuropils and providing inputs onto dendrites of efferent neurons that extend back into the protocerebrum (Strausfeld,2001). In neopterans, the lobes alone can integrate sensory information, as demonstrated by intracellular recordings showing that multimodal information is relayed by input neurons to the lobes and that context-specific information is relayed from the lobes by efferent neurons extending back out to the protocerebrum. Crucially, these efferent neurons are driven by modalities, such as acoustic, tactile, and visual parameters, that are not represented in the calyces (Li and Strausfeld,1997,1999).

thumbnail image

Figure 14. Elaboration of the mushroom body ground plan as evidenced by the apterygote Zygentoma [upper left panel: calyces receive olfactory inputs (red)]. In all mushroom bodies, with or without a calyx [the latter as in the Ephemeroptera (upper middle) and certain Hemiptera], parallel fibers contribute to local networks that receive inputs (purple) and provide outputs (black). In odonates (upper right), the distal part of the pedunculus reveals a remnant calyx equipped with microglomeruli receiving inputs (blue) from visual interneurons. Calyces are most elaborated when they serve several modalities, as in many species of Hymenoptera (olfactory input red, visual input blue). Modality-specific zones of the calyces supply longitudinally partitioned lobes (lower left). In some Neoptera, the entire calyx is dominated by a modality other than olfaction, as in the “visual” calyces of the Gyrinidae (lower middle). Irrespective of how many modalities supply the calyces, the lobes of most mushroom bodies are divided into parallel or concentric divisions (lower right). These are usually represented by a concentric organization of Kenyon cell populations in the calyces.

Download figure to PowerPoint

Relatively few changes are required for a transition from such a simple ground plan (Fig. 14, upper left) to that of a mushroom body equipped with a calyx serving two or more modalities (Fig. 14, lower left). Successive duplication of an original complement of one neuroblast provides a quadripartite mushroom body and hence four identical sets of circuits operating on the same afferent supply (Ito et al.,1997). Included in the neopteran ground plan is the potential for Kenyon cell numbers to increase and for their processes in the lobes to be subdivided lengthwise into discrete parallel components each of which represents a defined region in the calyx. The latter is nothing more than an enfolding of one or more planar subdivisions, which in the mushroom bodies of some species, such as coleopterans, form a structure of concentric cylinders, whereas, in others, such as the honey bee, they maintain their planar arrangements. Elaboration of the calyx into discrete zones, each receiving a different set of sensory interneurons, as in many species of Hymenoptera, imposes nothing new on the original ground plan of parallel intrinsic neurons and local regions of inputs and outputs in the lobes. However, the acquisition of new sensory inputs to the calyx has been accompanied by an overall increase in the number of Kenyon cells resulting from sequential proliferation of many descendent neuroblasts from the original quartet. The consequent expansion of the calyx, and thus the number of divisions through the lobes, is comparable to an evolutionary trend described from studies of the cortex of mammals in which additional modalities, combinations of modalities, or accentuation of a sensory modality can impose novel cortical territories or expansions of existing ones (Kaas,1995; Catania,2005; Krubitzer,2007).

The presence of a calyx imposes a distinctive property on a system that can also operate without one. A calyx does not alter the basic circuitry of the lobes, but it does impose physiological constraints. Although the basic functional organization of the mushroom body lobes is the same in neopterans as in palaeopteran insects, sensory input to the calyx would mean that, in any local circuit in which a Kenyon cell participates, its synaptic activity will be subject to modulation by the ambient sensory signal relayed to its distal dendrites. A specific odor relayed by antennal lobe projection neurons to the calyces that results in a change of the excitability of a cohort of Kenyon cells (Jortner et al.2007; Finelli et al.,2008) must accordingly modify their participation in local computations in the lobes. This conclusion means that odor encoding by Kenyon cells should not be regarded as the primary function of a mushroom body. Rather, information about an odor provides the context in which multimodal integration is performed at discrete loci within the mushroom body lobes. In species such as the honey bee, those divisions of the mushroom body calyces that receive visual inputs (Paulk and Gronenberg,2008) would likewise modulate local circuits in the relevant divisions of the lobes in the context of the ambient visual signal.

Elaboration through time

Sequential elaboration of the mushroom body, originating from a zygentoman-like ground plan and culminating in a neopteran mushroom body typified by Apis or Drosophila, can be mapped onto a “standard” insect phylogeny (Fig. 15). In this tree, the zygentoman mushroom body is close to representing the basal ground plan.

thumbnail image

Figure 15. Evolution of mushroom bodies. Glomerular olfactory lobes supplied by the first antennae (character 1, see Note on Terminology) are resolved in malacostracan Crustacea and in the Archaeognatha. Projection neurons from the glomerular lobes extend to the lateral protocerebrum of both malacostracan crustaceans and archaeognathans (character 2). Mushroom bodies with calyces and lobes (3), supplied by olfactory glomeruli and gustatory inputs (4) are resolved in the Zygentoma and are thus basal to Palaeoptera and Neoptera. Double calyces (5) and parallel subdivisions of the lobes (6) are resolved in pterygote insects. The representation of modalities in addition to olfaction and taste (visual afferents (7) is resolved in the Odonata and higher Neoptera. Modality substitution (8; vision substituting olfaction) in the calyces or their rudiments has occurred independently in some Coleoptera and Odonata. Secondary loss of antennal lobe glomeruli and the reduction or complete loss of calyces (9, 10) has occurred several times independently within the Palaeoptera and Neoptera. Aquatic species are shown in boldface. The phylogeny on which characters 1–10 are mapped is based on Wheeler et al. (2001).

Download figure to PowerPoint

A pronounced calyx, present in most terrestrial neopteran insects, might suggest an adaptation to the evolution of high-growth flora that lead to ecosystems in which foraging and intraspecific recognition required the detection of airborne plant odors. However, rudimentary calyces probably also equipped the earliest mushroom body. Assuming that the first dicondylic insects were detritovores, then Thermobia and Lepisma may be good modern representatives of a life style similar to that of ancient insects. Both species have small antennal lobes as well as a large columnar mechanosensory neuropil. The latter does not provide an input into the calyces, but the former does. Gustatory inputs also target the calyx. It is thus likely that the mushroom bodies of ancient dicondylic insects played a role in chemosensory processing. This leaves open the question: what do the lobes do that causes them to be retained when olfaction is lost? They are obviously capable of participating in local circuits supplied by multimodal inputs that are important for the behavior of the animal.

A final transition from a calyx receiving one or two sensory modalities to one receiving many is today exemplified by some extant species of Coleoptera and Hymenoptera. For example, scarabid beetles that are generalist herbivores possess calyces that receive both olfactory and visual afferents, whereas scarabs that are dietary specialists have calyces that receive solely olfactory inputs (Farris,2008b). In a manner analogous to the acquisition of new sensory areas in the mammalian cortex, the accommodation of several modalities by the mushroom body calyces of many species of Hymenoptera has resulted in the evolution of novel and discrete calycal subdivisions, each of which defines a discrete and independent functional subdivision of the underlying lobes (Strausfeld,2002).

These observations return the discussion once again to considerations about mushroom body function. Intracellular recordings of output neurons from the lobes of the cockroach mushroom body demonstrate that efferent neurons respond in a context-specific manner to multimodal stimuli (Li and Strausfeld,1997,1999). Then, as now, the interpretation of such results is that mushroom bodies of crown neopterans process sensory cues in the context of odor representation. However, the natural experiment of antennal lobe ablation in aquatic insects provides a mushroom body without a calyx. Insofar as the function of this type of mushroom body is unlikely to be related to olfaction, might the general functional significance of the lobes might be better resolved from the study of such anosmic insects than, say, from study of an olfactory species such as Drosophila, where certain subdivisions of the lobes are required for different aspects of olfactory memory (Pascual and Préat,2001; Yu et al.,2007). However, such roles need not be universal, nor the sole functional attributes. As demonstrated in a preliminary study (Strausfeld et al.,2006), circuitry within the lobes is comparable across widely divergent species, including those that lack antennal lobes. In short, the basic organization of the mushroom body lobes is the same with or without a calyx. It is the functional relevance of the calyxless ground plan that, once understood, may reveal the real significance of the mushroom body.

Acknowledgements

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED

Our insights into the mushroom bodies have profited from various discussions with colleagues. We thank Daniel Kalderon (Department of Biological Sciences, Columbia University) for the gift of anti-DCO antibodies. We are grateful to Kei Ito (University of Tokyo), Martin Heisenberg (Theodor Boveri Institute, University of Würzburg), and Gilles Laurent (California Institute of Technology) for their views. We especially thank Camilla Strausfeld (University of Arizona) for detailed comments on the paper and for extensive editing of the text, and Pavel Masek for the photograph of the machilid shown in Fig. 11A. Partial support for the evolutionary analyses derives from a fellowship to N.J.S. from the John D. and Catherine T. MacArthur Foundation.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. A NOTE ON TERMINOLOGY
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. LITERATURE CITED