|Home | About | Journals | Submit | Contact Us | Français|
The fly brain is formed by approximately hundred paired lineages of neurons, each lineage derived from one neuroblast. Embryonic neuroblasts undergo a small number of divisions and produce the primary neurons that form the functioning larval brain. In the larva, neuroblasts produce the secondary lineages that make up the bulk of the adult brain. Axons of a given secondary lineage fasciculate with each other and form a discrete bundle, the secondary axon tract (SAT). Secondary axon tracts prefigure the long axon connections of the adult brain, and therefore pathway choices of SATs made in the larva determine adult brain circuitry. Drosophila Shotgun/E-cadherin (DE-cad) and its binding partner Armadillo/β-catenin (β-cat) are expressed in newly born secondary neurons and their axons. The fact that the highly diverse, yet invariant pattern of secondary lineages and SATs has been recently mapped in the wild-type brain enabled us to investigate the role of DE-cad and β-cat with the help of MARCM clones. Clones were validated by their absence of DE-cad immuno-reactivity. The most significant phenotype consists in the defasciculation and an increased amount of branching of SATs at the neuropile-cortex boundary, as well as subtle changes in the trajectory of SATs within the neuropile. In general, only a fraction of mutant clones in a given lineage showed structural abnormalities. Furthermore, although they all globally express DE-cad and β-cat, lineages differ in their requirement for DE-cad function. Some lineages never showed morphological abnormalities in MARCM clones, whereas others reacted with abnormal branching and changes in SAT trajectory at a high frequency. We conclude that DE-cad/β-cat form part of the mechanism that control branching and trajectory of axon tracts in the larval brain.
Neurons are morphologically characterized by long, highly branched cytoplasmic extensions called neurites (dendrites and axons). Axonal pathfinding and the pattern of neurite branching determines the wiring of neuronal circuits. Thus, guiding axons along a specific pathway, or setting a branchpoint at one place of the neurite versus another will alter significantly the way in which that neuron is connected to other neurons. The developing brain of Drosophila has proven to be an excellent model system to unravel the genetic mechanisms that control pathway choices and branching behavior of neurons.
The Drosophila brain is formed by a stereotyped set of approximately 100 neuroblasts that appear in the early embryo (Urbach and Technau, 2003; Younossi-Hartenstein et al., 1996). Each neuroblast produces a small lineage of primary neurons during the embryonic period. Neurons that belong to one lineage remain clustered together; likewise, their axons form a coherent bundle, the primary axon tract (PAT). Primary axons then elaborate axonal and dendritic arbors which establish the neuropile of the larval brain. After a period of mitotic quiescence that lasts from mid-embryogenesis to mid larval development, neuroblasts resume their activity and produce much larger lineages of secondary neurons. Similar to primary axons, axons of a given secondary lineage fasciculate with each other and form a discrete bundle, the secondary axon tract (SAT) within the larval brain (Dumstrei et al., 2003b; Pereanu and Hartenstein, 2006). SATs most often remain a single, undivided tract as they enter the neuropile; in certain lineages, the SAT splits into two or even three branches at the cortex-neuropile boundary, and these SAT branches travel along separate pathways in the neuropile to connect to specific compartments. The pathways defined by the SATs in the larval brain define long axon connection of the adult brain (Dumstrei et al., 2003b; Nassif et al., 2003; Pereanu and Hartenstein, 2006; Pereanu et al., 2009). The only features added to secondary neurons during metamorphosis are the proximal and terminal branches that represent sites of postsynaptic input and presynaptic output. It follows that pathway choices of SATs made during the larval stage determine the “macrocircuitry”, that is, the pattern in which brain compartments are connected. To investigate the mechanisms underlying circuitry, it therefore seems appropriate to study the impact of genes on the pathway choices of lineages made at the larval stage. In this paper we have used the MARCM technique to analyze the role of Drosophila E-cadherin (DE-cad) and its binding partner, Armadillo/β-catenin (β-cat) in the formarion of SATs in the larval brain.
The cadherins form a family of widely expressed adhesion molecules that, according to many studies in vertebrates and invertebrates alike, (Hirano et al., 2003; Tepass et al., 2000) play a central role in patterning of neuronal connectivity. Cadherins consist of an extracellular domain with tandem cadherin repeats, a single membrane-spanning segment, and a cytoplasmic region (Hill et al., 2001; Nollet et al., 2000; Tepass et al., 2000). The so-called classical cadherins interact in a dynamic manner with the actin cytoskeleton via binding to a complex of cytoplasmic proteins, the catenins, The Drosophila genome contains two well studied classical cadherins, DE-cad and DN-cadherin (DN-cad) (Hill et al., 2001; Tepass et al., 2000).. DN-cad is expressed in differentiating neurons and is mainly involved in late events of neuronal development, in particular the formation and maintenance of synaptic connections (Hummel and Zipursky, 2004; Iwai et al., 2002; Prakash et al., 2005). Interfering with DN-cad function disrupts axon-target interaction in the optic lobe and antennal lobe-to-mushroom body projection (Lee et al., 2001; Zhu and Luo, 2004). DE-cad is expressed and required at early developmental stages in epithelial cells; it is downregulated in non-epithelial cells of embryos, including the central nervous system (Oda et al., 1994; Tepass et al., 1996). By contrast, high levels of DE-cad appear postembryonically in the brain where expression can be seen transiently in neuroblasts and secondary neurons, as well as in glia and trachea (Dumstrei et al., 2003b). Knock-down experiments of DE-cad (by expression of a dominant negative construct) revealed a complex larval brain phenotype, consisting in defects of neuroblast proliferation, as well as neuronal and glial morphogenesis (Dumstrei et al., 2003a; Wang et al., 2004). To elucidate the role of DE-cad in a more detailed manner, we decided to undertake an analysis of homozygous loss-of-function clones in the background of a heterozygous (i.e., functionally wild-type) brain. This approach is necessitated by the fact that flies homozygous for loss-of-function alleles in DE-cad or β-cat (as well as most other genes with essential function in neuronal development) are embryonically lethal, and can therefore not be studied at later developmental stages
To generate labeled mutant we used the MARCM technique (Lee and Luo, 1999) in which somatic recombination removes an inhibitory Gal80 construct, which then allows Gal4 driven GFP expression in a clone. The clonal strategy has been widely used in the imaginal discs (eye, wing) and has yielded decisive insight into gene function during pattern formation. It should be pointed out that mutant clones can only be interpreted if the corresponding wild-type clones are well characterized. This has been the case for structures like the eye disc or wing disc epithelium, but not the brain, where every clone (i.e., every lineage) has a different morphology. We have recently mapped and characterized all lineages of the larval brain (Pereanu and Hartenstein, 2006). This map enables us to recognize the identity of individually labeled lineages, be it wild type, or mutant, which, for the first time, makes the analysis of brain clones, induced by somatic recombination, a feasible task. Our data show that DE-cad and β-cat are involved in pathway choices of SATs. Mutant clones had extraneous branches that embarked upon abnormal trajectories in the neuropile. There seemed to be significant differences in the requirement for DE-cad function; some lineages showed no morphological abnormality in MARCM clones, whereas others reacted with abnormal branching at a relatively high frequency. This finding emphasizes the need to assay many lineages when conducting genetic studies of neural development; different lineages/neurons will often differ in their requirement for a given gene.
Oregon R flies were used as wild-type stock. The following lines were requested from the Bloomington stock center: (1) hs-FLP, elav-Gal4, UAS-mCD8-GFP/FM7c; (2)hs-FLP, elav-Gal4, UAS-mCD8-GFP; FRT42D; (3) FRT42B, UAS-GFP/CyO; (4) FRT19A; (5) y, w*, arm4/FM7c; (6) y, arm1/Fm7c; (7) w*, hs-FLP, FRT19A tub-Gal80; (8) elav-gal4; and (9)chaT-gal4, UAS-GFP. FRT42B, shgR69, UAS-GFP/CyO and FRT42B, hsπmyc, tub-gal80/CyO; Tub-Gal4/TM6B were provided by Dr. Carthew (Hayashi and Carthew, 2004).
To generate DE-cad mutant clones using the MARCM technique, two different mating schemes were used. (1) males of hs-FLP, elav-Gal4, UAS-mCD8-GFP/FM7c; FRT42D tub-Gal80 were crossed with females of hs-FLP, elav-Gal4, UAS-mCD8-GFP; FRT42D shgR69/CyO flies. (2) males of hs-FLP; FRT42B, shgR69, UAS-GFP/+ were crossed with females of FRT42B, hsπmyc, tub-gal80/CyO; elav-gal4/TM6B (Hayashi and Carthew, 2004). To generate comparable wild-type clones, two different mating schemes were used. (1) males of hs-FLP, elav-Gal4, UAS-mCD8-GFP/FM7c; FRT42D Tub-Gal80 were crossed with female hs-FLP, elav-Gal4, UAS-mCD8-GFP; FRT42D/CyO flies. (2) males of hs-FLP; FRT24B, UAS-GFP/+ were crossed with FRT42B, hsπmyc, tub-gal80/CyO; elav-gal4/TM6B.
To generate arm mutant clones, males of hs-FLP, FRT19A tub-Gal80; elav-Gal4, UAS-GFP/UAS-GFP were crossed with females of FRT19A arm1/FM7c flies. Control clones were generated by crossing males of hs-FLP, FRT19A, Tub-Gal80; elav-Gal4, UAS-GFP/UAS-GFP with females of FRT19A/FRT19A.
Embryos were collected on apple juice plates for 1 hour at room temperature. Embryos were aged until first instar [24 hour After Egg Laying (AEL)] and heat shocked for 30 minutes at 37°C. Larvae were aged at room temperature until late 3rd instar (about 120 AEL) and brains were dissected.
Larval brains were dissected in PBS and fixed in PBT (PBS with 0.1% Triton X-100) containing 4% formaldehyde for 30 minutes at room temperature. Standard staining protocols were used (Ashburner, 1989). Primary antibodies were used at the following concentrations: mouse anti-Neurotactin (Hortsch et al., 1990; 1:10; Hybridoma Bank); mouse anti-Syntaxin (Fujita et al., 1982; 1:10; Hybridoma Bank); rat anti-DEcad (Dumstrei et al., 2003b; 1:500); mouse anti-acetylated tubulin (1:1000; Sigma T6793); mouse anti-β-galactosidase (1:50; Sigma); rat anti-DN-cadherin (Iwai et al., 1997; 1:20; Hybridoma Bank). Secondary antibodies were purchased from Jackson Laboratory and used at the manufacturer’s recommended concentrations. Stained brains were mounted in Vectashield (Vector Laboratory; H-1000).
Confocal images were taken on a Biorad MRC1024ES microscope using Laser sharp version 3.2 software. Complete series of optical sections were taken at 2µm intervals. Images were analyzed using the ImageJ software. Generation of three-dimensional models was done with the Amira 3.0 software (Mercury Computer System, Inc.).
100 larval brains were dissected in ice-cold PBS and homogenized in 0.2 ml of lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 1 mM DTT, 1 mM PMSF). Proteins were immunoprecipitated according to the methods described in Dumstrei et al. (2002), using anti-β-cat antibody (Developmental Studies Hybridoma Bank) at 1:50 dilution. Each immunoprecipitation experiment was done in duplicates. Co-immunoprecipitated proteins were separated on 8% SDS-PAGE and transferred to Western blots. Blots were probed with anti-DE-cad (Dumstrei et al., 2003b) at 1:1000 dilution; anti-β-cat at 1:500 dilution and anti-α-catenin (Oda et al., 1993) at 1:500 dilution.
Larval brains were dissected in cold PBS and fixed in 4% paraformaldehyde, 2.5% glutaraldehyde in 0.1M phosphate buffer, pH 7.3, for 24 hours at 4°C. The brains were post-fixed for 60 minutes with 1% osmium tetroxyde in 0.1M phosphate buffer, pH7.3, at 4°C. Specimens were washed several times with distilled water and dehydrated in graded acetone series at 4°C. Specimens were incubated in 1:3, 2:2, 3:1 Epon:acetone ratio for 2 hours, 3 hours, and overnight, respectively. Brains were transferred to unpolymerized Epon and incubated overnight. They were then transferred to molds, oriented and placed at 60°C for 16 hours to permit polymerization of Epon. Blocks were sectioned at 0.06µm and mounted on slot grids and treated with uranyl acetate and lead citrate.
Secondary lineages of the larval brain are formed during the late second to late third instar. SATs grow into the neuropile compartments that had been formed by primary neurons during late embryonic and early larval stages. Each secondary lineage is defined by its position within the brain cortex and the invariant trajectory followed by its axons; based on these characteristics, lineages were mapped (Pereanu and Hartenstein, 2006). Secondary lineages share a number of generic properties. In most, if not all, lineages the sequence in which secondary axons form is reflected in their position within the SAT, and is also correlated with the expression of DE-cad and the organization of the cytoskeleton. As axons first grow out, they are thin, actin-rich, and express DE-cad (“early secondary axons”; Fig.1A–D). When reaching the glial layer that surrounds the neuropile surface (Fig.1J), these axons stall and grow out a tuft of filopodia. Subsequently, axons further elongate as a coherent bundle into the neuropile, often after following the neuropile surface for various distances. These older axons (“mature secondary axons”) become thicker and assemble more microtubules (Fig.1D–G). At the same time, the expression of DE-cad ceases (Fig.1F, G, I). Within the neuropile, SATs form bundles that consist exclusively of mature secondary axons that do not express DE-cad (Fig.1H, I). The axons forming later grow through the center of the pre-existing SAT tract (Fig.1D, middle). In other words, when looking at the cross section of a SAT (Fig.1E), older axons (thick, many MTs, DE-cad-negative) are always at the periphery of the tract, younger ones (thin, few MTs, DE-cad-positive) in the center. This peculiar manner of axon growth may enable older axons better to serve as “guides” for their younger siblings.
In the Drosophila embryo, DE-cad is expressed in epithelial cells where it interacts with β-cat and α-catenin (α-cat) (Tepass et al., 2000). The cadherin-catenin complex is strongly enriched in the zonula adherens, a narrow belt of adherens junctions encircling the apical pole of epithelial cells. By contrast, in neuroblasts and neurons of the larval brain, DE-cad is localized diffusely in the cell membrane. Electron microscopic analysis of larval brain cortex indicates that there are few, if any, adherens junction-type membrane specializations in between neuronal somata, or axons (Fig.1A, E, H; data not shown). However, cell membranes of SAT fibers are closely packed; the inter-cellular cleft in between fibers is less than 8nm (compared to the 20–25 nm cleft in between Drosophila epithelial cells).
The DE-cad expression pattern in the secondary lineages of the larval brain coincides with the expression of its presumed binding partner, β-cat (Fig. 2A–C) and α-cat (data not shown). In addition to secondary neuroblasts and neurons, β-cat is expressed strongly in the neuropile; this most likely reflects the fact that β-cat also forms a complex with DN-cad, which is expressed on primary neurons forming in the neuropile. Given that few or no adherens junctions exist among neuronal somata and axons, we wondered whether DE-cad and the catenins form a regular protein complex as described for epithelia. To this end, co-immunoprecipitation (coIP) experiments were performed on protein extracts from late larval brains. In wild type, the epithelial optic anlagen constitute a major part of the brain (Fig.2E), which would affect the interpretation of our results. We therefore generated brains lacking optic anlagen by expressing a dominant negative construct of EGFR, in the embryonic optic lobe primordium. In both transgenic lines, larvae and adult flies were viable, but completely lacked optic lobe and eye (Fig.2D). CoIP of extracts of these brains confirmed that DE-cad, β-cat and α-cat form a protein complex, just as in epithelial cells (Fig.2F).
Utilizing the MARCM technique (Lee and Luo, 1999), ectopic GFP expression was induced in individual neuroblasts in the larval brain. Heat-pulses inducing somatic recombination were applied during the early larval stage, so that only secondary lineages (i.e., neurons born in the larval period), consisting of neuroblasts, GMCs, and neurons were visualized by the presence of the GFP protein. The morphological characteristics of a GFP-labeled wild-type control clone are shown in Fig.3A. The neuroblast is located at the surface of the cortex. Clusters of 3–4 GMCs, distinguishable from neurons by their larger size and reduced Elav expression (not shown), cling to the lateral surface of the neuroblast. Secondary neurons typically form a wedge-shaped or cylindrical cluster that reaches from the surface to close to the cortex-neuropile boundary (Fig.3A’). The axons formed by all of the neurons of one lineage remain as a tight bundle, the SAT, that projects towards the neuropile, penetrates the glial layer at the cortex-neuropile boundary, and travels for various distances within the neuropile. Approximately 30% of the lineages split into two branches with different trajectories at the cortex-neuropile interface; about 10% split into three branches.
DE-cad mutant clones showed absence or significant reduction of DE-cad expression in cell bodies and axons (Fig.3B, B’). The structural abnormalities of these clones included changes in cell body position and axon defasciculation, but most prominently, an increased number of branches at the cortex-neuropile boundary, as well as abnormal SAT trajectories in the neuropile. Figure 4 summarizes the characteristics of brain lineages that can be scored when identifying mutant phenotypes. All lineages are characterized by a highly stereotyped SAT pattern. This pattern includes (1) the point of entry into the neuropile; (2) direction of SAT at cortex-neuropile boundary (SAT follows straight radial path into neuropile or turns and follows neuropile surface); (3) presence or absence of SAT branching at the cortex neuropile boundary; (4) fasciculation of SAT with other tracts in the neuropile; (5) branching of SAT in the neuropile. Figure 4A and C give examples SATs branching at cortex-neuropile boundary (e.g., BLVp1/2) or within neuropile (e.g., BAmv2). Using these criteria, we can reliably recognize almost all lineages individually (Fig.4D). In several cases (e.g., DALcm1/2; CP2/3), lineages form pairs in which clusters of cell bodies lie next to each other and produce separate SATs, but then merge into one single tract with a common trajectory and target. Five lineages (DPMm1, DPMpm1 and DPM2, CM3, CM4) stand out by their much larger size (Fig.4D, bottom panel). These so-called “mega-lineages” follow a different proliferatory pattern (Bello et al., 2008), in that the neuroblast first produces several symmetrically dividing “amplifying progenitors”, whose daugher cells then behave like regular neuroblasts. The mega-lineages have 4–5 times more neurons than regular lineages, and produce complex arrays of SATs. Since the pattern of the SATs of mega-lineages is not yet fully worked out, clones in these lineages were not considered in the present study.
The frequency of clones with and without morphological phenotypes are listed in Fig.4E. Overall 43 out of 134 clones (32%) showed abnormalities. If we compared the frequency with which individual lineages were affected (Fisher exact analysis) to the overall score, none of the scores were significantly lower or higher than the overall 32%, due to the relatively small n-value. If we grouped lineages by their topology (e.g., all BA, DPL, DPM etc added together), some significant differences materialized (Tab.1).The grouping of lineages in families is based on a common primary target. Thus, for example, all lineages whose SAT enter the neuropile in or directly adjacent to the BA compartment (the larval antennal lobe) are designated as “BA lineages” (Pereanu and Hartenstein, 2006).
Comparing the ratio of abnormal clones that belong to a given lineage family to the overall score, one family, DPM, stands out by having a particularly high number of clones with abnormal phenotypes. Comparing lineage families to each other showed that BA lineages (antennal projection neurons and other lineages whose SATs project into the ventro-anterior brain) showed abnormal phenotypes less often than DPL and DPM lineages (located in the dorsal brain). Similarly, mushroom body lineages showed significantly fewer clones with abnormalities than DPM lineages, and marginallly fewer than DPL. The data imply that the requirement for DE-cad varies: in terms of SAT branching and trajectory, some lineages are more resistant to the loss of DE-cad than others. This conclusion was further supported by the result of an experiment where we used the atonal-Gal4 driver (Hassan et al., 2000), expressed in BLD5, to generate and visualize more clones specifically in this lineage. Out of more than 15 clones overall, not a single one showed ectopic branches or fasciculation defects.
Figure 5 illustrates typical structural phenotypes encountered. A–F show an MB lineage where the SAT (forming part of the peduncle) gives off a side branch within the neuropile, near the beginning of the peduncle. More frequently than ectopic branches in the neuropile we encountered abnormalities starting at the cortex neuropile boundary. Shown in Fig.5G–I is a DPLl1 lineage whose normally single SAT splits into two at the point where it enters the neuropile; both branches then follow a similar trajectory. A similar phenotype can be seen for the clone affecting the CP2/3 lineage (Fig.5J–L). CP2/3 form a pair of lineages whose SAT normally splits into a dorsal branch, crossing over the peduncle, and a ventral branch, traveling first parallel to the peduncle and then converging onto the spur region (Fig.5L). In the mutant clone, the dorsal branch is divided into two components traveling close to each other, but separately, a behavior never encountered in a wild type CP2/3 clone. The examples shown in Fig.5M–O and P–R illustrate cases where supernumerary branches follow different trajectories. The BLVp1/2 clone, normally branched into two long SATs (Fig.5M, O) forms two extra branches that follow trajectories not seen in any wild-type lineages (arrows in Fig.5N). Similarly, the mutant BLD3/4 clone shown in Fig.5P–R generates two supernumerary branches at the cortex-neuropile boundary with novel trajectories (arrows): projecting ventro-laterally this branch reaches the optic lobe neuropile, where it forms a tuft of filopodia, before turning sharply medially to join a commissural tract. A second ectopic branch projects straight ventrally towards the ventral nerve cord.
An additional aspect of SAT pathway abnormalities within the neuropile was revealed in brains that contained several mutant clones together. In such cases, it could be observed that SATs that in wild type remain separate throughout their pathway, fused in the mutant clones. Shown in Fig,6A–C is DALcm1/2 and BAlp2. One branch of DALcm1/2, in wild type, turns ventrally and forms a tract towards the ventral nerve cord. BAlp2 has an upward directed, unbranched SAT. Both come within less than 5µm from each other, but do not fasciculate (Fig.6B). In brains where both lineages lacked DE-cad, the SATs merged (Fig.6C, E). A second example is the pair DPLal1 and DALl1. In wild type, these form characteristic crescent shaped SATs that arch around each other without touching (Fig.6E). In the mutant setting, the SATs fuse at the point of closest proximity (Fig.6D, F). The phenotype shown by the mutant clone in DPMl3 (Fig.6G–I) shows similarities to the fusions above, although only a single lineage is involved here. Thus, the wild-type DPMl3 splits into two branches at the point where it enters the neuropile: an antteriorly directed branch (“a” in Fig.6G, I) and a posteriorly directed one (“p”) The latter curves ventrally and medially and splits into a commissural component (arrowhead in Fig.6G, I) and a short ipsilateral branch that ends close to the commissure. The posterior branch of the mutant DPMl3 (Fig.6H, J) initially follows the same pathway as its wild-type counterpart; however, it does not enter the commissure, but turns backward and terminates in contact with the tip of the anterior branch of the same lineage, thereby forming a closed circular SAT.
β-cat is the cytoplasmic binding partner of DE-cad and shows an overlapping expression in secondary lineages and their SATs (see above). Correspondingly, loss of function MARCM clones in β-cat have similar phenotypes as DE-cad clones (Fig.7). The abnormalities included mostly excessive branching at the cortex-neuropile boundary, as shown for the BLD3/4 lineage depicted in Fig.7A. Within 43 total clones, we observed several clones whose SAT did not enter the neuropile at all, but skirted along the neuropile surface (Fig.7B–D). Cell bodies of the clone shown here are located in the DPMl territory. The course taken by the SAT also have similarity to the DPMl3/4 pattern (see Fig.6G–J above). One branch grows postero-ventrally, the other anteriorly. However, both branches remain superficially at the neuropile surface and deviate from their normal pattern. The posterior branch does not enter the commissure but continues stright downward towards the ventral nerve cord. The anterior branch terminates at a superficial position near the dorsal lobe of the mushroom body.
An interesting phenotype shown by several of the β-cat mutant clones concerned a downregulation of the global neuronal marker, neurotactin. In wild type, neurotactin is expressed on all secondary neurons and their axons from the time they grow out. Mutant clones often showed branches that had lost neurotactin expression (Fig.4A’, A”). In these clones, cell bodies and the proximal segment of the SAT (within the cortex) were positive for neurotactin, but one or more of the supernumerary branches penetrating into the neuropile were neurotactin-negative. This finding points at regulatory relationships between the different adhesion complexes formed by DE-cad and neurotactin.
Cadherins are structural membrane molecules expressed throughout the developmental history of most, if not all, animal cell types. Most cells express multiple cadherins in a dynamic pattern, such that one cadherin is turned on at an early stage and is later replaced by another cadherin. Given their widespread expression it is no wonder that cadherins participate in many different functions, both developmentally and in the mature organism. In the larval Drosophila brain, the reduction of DE-cad by expression of a dominant-negative construct results in abnormalities of neuroblast proliferation, glial morphology, layering of neurons, and axon patterning (Dumstrei et al., 2003a). Abnormalities in a wide range of morphogenetic processes have also been observed in vertebrate. In zebrafish, hypomorphic mutation of E-cadherin showed flattened anterior neural tissue, scattered trigeminal ganglia, and caused abberant axon bundles from the trigeminal ganglia (Shimizu et al., 2005). In mouse, loss of E-cadherin (Cdh1) during neurogenesis results in loss of polarity and adhesion among radial glial cells, disrupting the integrity of the ventricular zone (Rasin et al., 2007). How can one envision, in more general terms, the cellular mechanism for which cadherin function is essential? Studies carried out so far point at a number of different actions of cadherins.
(1) Differential expression of cadherins is involved in specifying brain compartments and neural circuits: it has been known for a long time that boundaries between tissues are set up by cell sorting (Townes and Holtfreter, 1955). If cells within one population start expressing two different adhesion molecules (or different levels of the same molecule) they will sort, such that cells with the same adhesion molecule will group together. It is widely speculated that the fact that cadherins are expressed in highly diverse and regionally specific patterns in the developing vertebrate nervous system points at a role of these molecules in “neural sorting”, in the sense that neurons or axons sort out into specific brain compartments or tracts depending on the type of cadherin they express (Redies, 2000; Tepass et al., 2000; Yagi and Takeichi, 2000). Experimental evidence for the role of cadherins in sorting neuronal connectivity by matching discrete populations of neurons with their targets comes from studies of the visual system. In the chicken optic tectum, N-cadherin is concentrated at laminae b–e. If N-cadherin is blocked by antibodies, retinal axons fail to stop at these target laminae (Sanes and Yamagata, 1999). In a similar manner, DN-cad may regulate target specifity in visual and olfactory sensory afferents (Hummel and Zipursky, 2004; Lee et al., 2001; Zhu and Luo, 2004).
(2) Cadherin-mediated adhesion modulates signaling pathways, or plays a permissive role during cell signaling interactions: Numerous examples of cases where the cadherin-mediated contact between cells activates specific signaling cascades are known (e.g., (Williams et al., 2001; Yap and Kovacs, 2003). In addition, a permissive effect of cadherins on signaling can be envisaged whereby cell contacts at a certain time or space need to be established/stabilized in order for specific signal-receptor interactions to occur.
We speculate that it is this second role of cadherins that most likely explains the axonal abnormalities that we see in DE-cad loss of function clones. As the first secondary axons start to extend as a cohesive bundle (SAT), they have to grow only for a short distance before reaching the glial layer surrounding the neuropile. After traversing this layer the SAT continues to extend, either entering the neuropile immediately, or by growing for various distances, and in different directions, along the glial layer (Dumstrei et al., 2003b; Pereanu and Hartenstein, 2006). Since lineages differ with respect to the trajectory of their SAT, one has to assume that specific signals expressed locally within the lineage and/or their environment guide the choice of the SAT which direction to follow. The fact that SATs of DE-cad mutant lineages make the wrong choice in a significant fraction of cases indicates that DE-cad-mediated adhesion, possibly between axons and neuropile glia, or between axons themselves, is required at the “choice point” to receive the proper “instructions”. We prefer this permissive model of DE-cad function to an instructive model because DE-cad expression on SATs and neuropile glia appears to be uniform for all lineages, despite the fact that lineages differ widely in their pathway choices.
It is important to point out that the requirement of different lineages for DE-cad is apparently not uniform. Thus, for some lineages (example: most BA lineages; BLD5), abnormalities were never observed in mutant clones; in other cases, the opposite was true, and mutant clones often caused abnormal pathway choices. This finding suggests that the specific cues determining the pathway choice of a given SAT are more “robust” for some lineages, promoting the right choice even in the absence of permissive factors such as DE-cad. The finding also underscores the importance of assaying many lineages, rather than one, when evaluating the function of a gene in neural development. Negative findings in a genetic analysis that focuses on one (or a small group of lineages, such as the mushroom body, which has been so far used almost exclusively as a “testing ground” for genetic studies on brain development) do not imply that the gene in question plays no role in other neurons. Looking at different lineages in the Drosophila larval brain has only recently become possible, since the morphology, different for each one of the 100 or so lineages, has been described (Pereanu and Hartenstein, 2006).
The DE-cad loss of function phenotype that we observe in clones is relatively mild, compared to the findings in a previous study where we utilized a dominant negative construct to inhibit DE-cad in larval brain development (Dumstrei et al., 2003a). First, the mutant clones reveal no gross defects in size or neuronal position. We had seen such defects in experiments where a dominant negative DE-cad construct was driven, either in the entire brain (neurons plus glia), or in glia alone. In the latter case we could verify directly that the glial scaffold (“trophospongium”) around neuroblasts and neurons, formed by the widespread processes of cortex glia, is absent. Taken together, the findings suggested that DE-cad mediates interactions among cortex glia which are required for the formation of the trophospongium. Once in place, the trophospongium is able to support proper neuroblast division and neuronal layering, even if DE-cad is removed from the neurons by somatic recombination. Another possibility to keep in mind is perdurance, where the DE-cad inherited from the neuroblast prior to clone induction, is sufficient to function in early processes like neuroblast division and layering of neurons.
The DE-cad LOF phenotype in mutant clones is also quite mild with respect to axon fasciculation and pathfinding. Pathway abnormalities were observed in a substantial fraction of, but not all, cases; only rarely did individual axons within the SAT of a clone defasciculate before entering the neuropile. Here it is important to keep in mind that aside from DE-cad, many additional adhesion molecules are expressed in developing neurons. For example DN-cad and neurotactin: DN-cad is expressed at a low level in differentiating neurons, and starts to be upregulated in the late larva. Neurotactin is expressed in all secondary neurons from the time onward when they are born. It is plausible that all three proteins, in addition to other adhesion molecules (Fung et al., 2008), would share in the function of mediating contact between axons/axon-glia at the point where the SAT enters the neuropile and “searches for” specific cues guiding its further progress into the neuropile.
Our findings indicate that β-cat, similar to its role in epithelial morphogenesis, interacts with DE-cad in the developing nervous system. Loss of β-cat in individual clones results in phenotypes similar to those, if not often stronger, than those observed in DE-cad mutant clones. Many studies in the vertebrate field support a role of β-catenin and other proteins that bind to the cadherin-catenin complex in neural development, and particular in synaptic target selection, synaptogenesis and synaptic function (Elul et al., 2003; Murase et al., 2002; Togashi et al., 2002; Yu and Malenka, 2004). It has been speculated that the role of β-catenin in axons is similar to that in epithelial cells, namely to mediate the dynamic interaction between membrane adhesion complexes and the actin cytoskeleton (Thoumine et al., 2006). It will be important to learn more about the molecular basis of this link in growing axons. Axons and their growth cones navigate in a three-dimensional space; their interaction with the microenvironment is extremely dynamic, judging from the live recordings of filopodia that extend from progressing growth cones. This is different from epithelial cells which form two dimensional sheets, and generally move in concert. One might expect differences in the mechanism that integrate cell-contact (adhesion complex) and cell movement (cytoskeleton). Genetic screens and in vivo studies in the Drosophila brain provide a useful tool to make progress in understanding this mechanism.
We are grateful to Dr. RW Carthew for providing us with fly stocks. This work was supported by NIH Grant RO1 NS29357-15 to V.H. and the Ruth L. Kirschstein National Research Service Award GM007185 to S.S.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.