Dscam

DEVELOPMENTAL BIOLOGY

See the embryonic expression pattern of Dscam at the Berkeley Drosophila Genome Project Patterns of Gene Expression Site.

Dscam is expressed on axons in the embryonic CNS. Dscam RNA is expressed in Bolwig's organ as well as more generally within the CNS and PNS. The protein product is exclusively expressed on axon processes (Schmucker, 2000).

Analysis of Dscam diversity in regulating axon guidance in Drosophila mushroom bodies

Dscam is an immunoglobulin (Ig) superfamily member that regulates axon guidance and targeting in Drosophila. Alternative splicing potentially generates 38,016 isoforms differing in their extracellular Ig and transmembrane domains. Dscam mediates the sorting of axons in the developing mushroom body (MB). This correlates with the precise spatiotemporal pattern of Dscam protein expression. MB neurons express different arrays of Dscam isoforms and single MB neurons express multiple isoforms. Two different Dscam isoforms differing in their extracellular domains introduced as transgenes into single mutant cells partially rescued the mutant phenotype. Expression of one isoform of Dscam in a cohort of MB neurons induces dominant phenotypes, while expression of a single isoform in a single cell does not. It is proposed that different extracellular domains of Dscam share a common function and that differences in isoforms expressed on the surface of neighboring axons influence interactions between them (Zhan, 2004).

Dscam is expressed selectively in young MB axons and is required to form a single highly organized nerve bundle and for the segregation of MB axon branches. Using DNA microchip analysis, it was demonstrated that MB neurons express many different Dscam isoforms, that different isoforms combinations are expressed in single cells, and that they differ from the array of isoforms expressed in photoreceptor neurons. Single isoforms containing different extracellular domains substantially rescue the null mutant phenotype in single mutant cells. Rescue activity was restricted to TM2-containing isoforms. This likely reflects differences in the localization of TM1 and TM2 isoforms in MB neurons. Expression of single Dscam isoforms in cohorts of MB neurons induced strong dominant phenotypes. Conversely, expression of a single isoform in isolated neurons did not. It is concluded that different Dscam isoforms have a common function largely accounting for the ability of single isoforms to provide partial rescuing activity. It is proposed that appropriate patterning in the MB requires that each MB axon express Dscam isoforms differently from its neighbors (Zhan, 2004).

Previous studies have demonstrated that Dscam is required for MB formation. Using genetically mosaic animals, it was discovered that the removal of Dscam in large clones results in massive defects in adult MB structure, including collapse and fusion of lobes. In single-cell clones, while Dscam mutant MB neurons invariably bifurcated at the base of the peduncle, branches frequently fail to segregate to different lobes. The developmental analysis described here demonstrates an earlier additional requirement for Dscam function in regulating MB development (Zhan, 2004).

A striking feature of MB organization is the precise topographic relationship between dendrites, axons, and synapses of MB neurons. These neurons fall into morphologically and molecularly (e.g., neurotransmitter expression) distinct classes. The axons in the peduncle are arranged in a concentric manner with a smooth gradient of developmental age in which the oldest fibers are located peripherally. Dendrites are arranged in a zonal fashion that shares a simple topographic relationship to axons not only in the peduncle but also in their layered organization in the lobes. Although MBs in different insects vary considerably in their size and complexity, they share a common topographic organization. This conservation of structure argues that the spatial relationship between different cellular elements in the MB is of functional significance (Zhan, 2004).

Developmental studies have revealed a common mechanism for promoting fiber organization in the MB. In Drosophila, the honey bee, cricket, and cockroach, different MB neuronal classes (e.g., gamma, alpha'ß', and alphaß) are generated in a sequential fashion from a common neuroblast. For instance, gamma neurons are generated prior to alpha'ß' and are followed by alphaß. Cells generated from each neuroblast form a separate cluster from which a single fascicle projects into the peduncle. At the entrance to the peduncle, axons from different fascicles sort such that young axons project into the center of the developing peduncle where they displace progressively older axons peripherally. The relative position of axons established in the peduncle is conserved in the dendritic calyces and lobes. Since MB neurons are generated in a parallel sequential fashion from multiple neuroblasts, mechanisms must exist to promote the selective association of axons of the same age (Zhan, 2004 and references therein).

Based on the Dscam expression pattern and mutant phenotypes as well as biochemical studies, it is proposed that Dscam promotes the association between young axons. Dscam protein is first observed on MB axons as they converge in the peduncle, and it remains expressed on these axons as they branch into the lobes. Expression is dynamic, and Dscam protein is downregulated as fibers mature. In the absence of Dscam, a single core fiber in the peduncle and lobes may be replaced by multiple phalloidin-positive fascicles (typically two or three). These ectopic core fibers are surrounded by a thin layer of axons that stain with neither phalloidin nor Fas II in a manner reminiscent of the organization of the single fiber bundle that forms in the wild-type peduncle. Indeed, it is proposed that in the absence of Dscam young fibers emerging from neurons generated from the same neuroblasts frequently do not converge in the peduncle but rather remain separate. Hence, Dscam ensures that MB neurons generated by different neuroblasts at the same time selectively fasciculate to form a single integrated neuropil (Zhan, 2004).

Defects in the lobe structure may be secondary consequences of earlier defects in the peduncle, or Dscam also may be required in an independent step for different MB branches to select the appropriate pathways in the developing lobes. The branch segregation defects may reflect a role for Dscam in repelling sister branches, thus promoting extension into different lobes. The notion that Dscam could mediate repulsive response is consistent with the phenotypes associated with the overexpression of a single isoform in the MB, as described in this study, as well as in Bolwig's nerve and in olfactory neurons, as well as loss-of-function phenotypes in the olfactory system (Zhan, 2004 and references therein).

The primary motivation for these studies on Dscam function in the MB was to assess whether the diversity of Dscam proteins produced by alternative splicing is functionally important. Microarray data revealed that MB neuronal populations express many different isoforms of Dscam and that single neurons express multiple isoforms. The array of isoforms made in MB neurons is distinct from other neurons, such as photoreceptors isolated from the developing eye disc. While incorporation of alternative exon 4s and 6s into transcripts appeared largely random in these different cell populations, preferential use of alternative exon 9s was observed for other neurons in the developing brain. These findings are consistent with genetic studies showing that the deletion of various alternative exon 4s does not lead to MB defects, arguing for redundancy between them (Zhan, 2004).

Despite the complex pattern of Dscam isoforms expressed in MB neurons, single isoforms of Dscam encoded by cDNA transgenes, including an isoform containing an alternative exon 9 rarely expressed in these neurons, substantially rescues the adult branch segregation phenotype observed in single mutant cells. Whether rescue reflects restoration of the core fiber phenotype in the peduncle or appropriate segregation of axons at the base of the peduncle upon bifurcation is not known. Regardless, these studies argue that all isoforms of Dscam containing TM2 share a common underlying molecular function (Zhan, 2004).

While Dscam proteins containing different extracellular domains provide similar rescuing activity, this is restricted to TM2-containing isoforms. This correlates with differences in protein localization. TM2-containing isoforms are distributed along the entire cell surface, while TM1-containing isoforms were largely excluded from axons. These observations suggest that sequences within TM1, TM2, or both play important roles in regulating the subcellular distribution of Dscam (Zhan, 2004).

In summary, these studies argue that TM1- and TM2-containing isoforms are sorted to different neuronal compartments. It is important to emphasize, however, that the protein localization data are based solely on protein expressed in excess of endogenous levels. Indeed, while transgenes expressing TM1-containing proteins are not sufficient to provide substantial rescue activity, a missense mutation within TM1 leads to defects in MB development similar to Dscam nulls. Hence, it is likely that mechanisms exist to deliver TM1-containing isoforms to axons. Recent biochemical experiments suggest that this may occur through cis interactions with TM2-containing isoforms (Zhan, 2004).

Several lines of evidence suggest that Dscam diversity may be required for normal MB development: (1) expression of a single isoform in a single mutant neuron provides only partial rescue; (2) expression of single isoforms in multiple MB neurons induces a strong dominant phenotype; (3) single MB neurons express multiple isoforms, and different neurons express different combinations of them. While different isoforms share a common feature of homophilic binding, single isoforms show a striking specificity for identical isoforms. They bind strongly to the same isoform and weakly, if at all, to different isoforms. It remains possible that the dominant phenotypes induced when many neurons express a single isoform may, in part, reflect differences in expression levels intrinsic to the method of producing large and small clones using the MARCM system. However, given the biochemical data, it is highly likely that interactions between neuronal cell surfaces expressing the same isoform of Dscam will be different from neurons expressing different isoforms (Zhan, 2004).

It is proposed that Dscam plays two distinct functions in MB neurons: one based on interactions between isoforms expressed on the surface of different neurons and the other based on interactions between isoforms on branches of the same neuron. It is envisioned that these interactions promote different levels of signal, with low levels promoting contact-dependent attraction, leading to fasciculation, and high levels promoting contact-dependent repulsion. In both the peduncle and lobes, Dscam on different MB growth cones promotes core fiber formation. As different MB neurons express different combinations of Dscam isoforms, this fasciculation function may be a consequence of low signaling levels. This may result from weak interactions between different isoforms or interactions between the small fraction of identical isoforms that may be shared by different MB neurons. At the base of the peduncle, MB growth cones bifurcate and then segregate to different lobes. In the absence of Dscam, axons bifurcate, as in wild-type, but frequently fail to segregate. Here, interactions between identical isoforms of Dscam expressed on sister branches of the same neurons may produce a strong signal activating a repellent response, thereby preventing axons from extending into the same lobe. Indeed, the diversity of Dscam expression on MB axons ensures that only sister branches express an identical set of Dscam isoforms. Upon segregating to the two lobes, weaker interactions between Dscam on axons of different neurons promote fasciculation. Since the biochemical properties of Dscam proteins and the gain- and loss-of-function phenotypes of Dscam throughout the nervous system are complex, Dscam diversity may contribute to wiring specificity in different ways in different developmental contexts (Zhan, 2004).

Effects of Mutation or Overexpression

To critically assess whether Dscam is required for axon guidance, loss-of-function mutations were identified. Dscam is localized to region 43B1-B3. Four lethal complementation groups have been mapped to this region. Sequences abutting a P element lethal insertion into one of these complementation groups, l(2)43Bc, were identical to sequences in a P1 genomic clone containing Dscam sequences. This insertion maps 1200 nucleotides downstream from exon 4.1; exon 4 encodes part of the second Ig domain. Two inversion alleles of l(2)43Bc also break within the Dscam locus. In(LR)43b71kIA breaks in the 8.8 kb intron separating exons 2 and 3. A second inversion, In(2R)DX8, also breaks within the Dscam locus, although the precise breakpoint was not identified. A single EMS allele, l(2)43Bc¹, fails to complement the inversions and the P allele. All alleles were early larval lethal. They have been renamed as follows: l(2)43Bc¹ = Dscam^E1; In(LR)43b71kIA = Dscam^X1; In(2R)Dx8 = Dscam^X2; and the P allele = Dscam^p (Schmucker, 2000).

Axon bifurcation results in the formation of sister branches, and divergent segregation of the sister branches is essential for efficient innervation of multiple targets. Axon bifurcation occurs when one growth cone is split into two and the two growth cones diverge. The directional segregation of the growth cones and thus the sister branches derived from a single axon are essential for propagating neural signals in divergent directions. If axon bifurcation leads to formation of 'twin' growth cones, how can these growth cones faithfully project away from each other and toward different targets? This simultaneous extension and divergence of twin growth cones might be achieved simply by mutual repulsion mechanisms. Alternatively, during or soon after bifurcation, individual growth cones might further differentiate and acquire distinct sets of guidance receptors through interactions among one another or with different glial cells. From a genetic mosaic screen, it has been found that a lethal mutation in Dscam specifically perturbs segregation of axonal branches in the mushroom bodies. Single axon analysis further reveals that Dscam mutant axons generate additional branches, which randomly segregate among the available targets. Moreover, when only one target remains, branching is suppressed in wild-type axons while Dscam mutant axons still form multiple branches at the original bifurcation point. It is concluded that Dscam controls axon branching and guidance such that a neuron can innervate multiple targets with minimal branching (Wang, 2002).

To facilitate mosaic analysis in the complex CNS, a novel genetic mosaic system, called MARCM has been developed, in which only the homozygous cells lacking GAL80 are uniquely labeled in mosaic tissues. Using the MARCM genetic mosaic system, the wild-type development of the Drosophila MB has been described with unprecedented single-cell resolution. Moreover, by creating clones of MB neurons homozygous for various mutations in mosaic organisms, it is possible to elucidate the molecular mechanisms controlling different aspects of MB development. Interestingly, from a genetic mosaic screen designed for identifying novel mutations that cause specific defects in the development of MB neurons, a lethal mutation was found that specifically hinders the formation of bifurcated axon bundles. If the mutant clone is created before the birth of neurons that normally project two perpendicular fascicles, these neurons will instead extend their axons in only one of the two original directions. One mutant neuroblast clone in the whole MB can alter the projections of the other three wild-type clonal units. Phenotypic analysis of single neurons further reveals that mutant axons give rise to additional branches at the bifurcation point and that the branches are distributed randomly among the accessible pathways. When only one path exists, wild-type neurons project a single process along the path while mutant neurons generate multiple branches. In addition, defects in axon extension are observed in certain mutant MB neurons. Mapping by recombination and complementation reveals this interesting lethal mutation is an allele of Dscam, suggesting a novel molecular mechanism for mediating formation and guidance of axonal branches (Wang, 2002).

The MBs are paired structures, one in each brain lobe; and each MB is derived from four neuroblasts (Nbs), each of which generates a similar set of three distinct types of neurons. Gamma neurons, which are generated prior to the mid-3^rd instar stage, project axons into the gamma lobe at the adult stage; alpha'/ß' neurons, which are generated in late 3^rd instar, have bifurcated axons that form the alpha' and ß' lobes; and alpha/ß neurons, which are generated after puparium formation (PF), project their bifurcated axonal branches into the alpha and ß lobes. In a genetic mosaic screen, a novel lethal mutation, l(2R)MB99, was identified that causes various defects in the guidance of bifurcated alpha/ß axons (Wang, 2002). When subsets of cells within wild-type MB Nb clones are selectively marked using GAL4-201Y, all labeled axons extend either into the gamma lobe or the alpha and ß lobes; the alpha lobe extends dorsally while the gamma and ß lobes extend medially toward the midline. Only a small number of the late pupal-born alpha/ß neurons are marked, so the alpha and ß lobes appear very thin and faint as compared with the gamma lobe. In addition, these labeled axons occupy the center of the alpha and ß lobes. These GAL4-201Y-marked alpha and ß lobes are hereafter referred to as the core alpha and ß lobes. Notably, the core alpha and ß lobes are morphologically indistinguishable, consistent with the fact that individual alpha/ß axons bifurcate into two branches that project away from each other into the alpha and ß lobes, respectively. A total of 50 l(2R)MB99 mutant Nb clones were collected for detailed phenotypic analysis. In contrast with normal looking gamma lobes, abnormal alpha/ß lobes are observed in every Nb clone homozygous for the l(2R)MB99 mutation. Being much thicker and denser than in wild-type clones, the core alpha/ß lobes seem to be composed of many more axons in the l(2R)MB99 mutant clones. But, unlike wild-type alpha/ß axons, most mutant core alpha/ß processes appear to fail to reach the ends of the alpha/ß lobes. Because no change in the number of cell bodies can be detected (about 30 GAL4-201Y-positive alpha/ß neurons in each Nb clone), the morphological changes observed in mutant core alpha/ß bundles imply that individual alpha/ß axons acquire supernumerary but short-ending branches in mutant clones. In addition, dramatic changes in the configuration of the alpha/ß lobes are observed in 38% of mutant Nb clones. Instead of bifurcating axons at a right angle, these mutant clones project all of their alpha/ß axons in only one direction, either dorsally or medially. Interestingly, these uni-directionally extending mutant processes can exist as two distinct bundles running side by side, or can be fasciculated into a single bundle. Another phenotype that suggests defects in the divergent segregation of alpha/ß axonal branches is detected in about 20% of mutant Nb clones, where differences in the thickness of the axon bundles exist between the dorsal and medial fascicles. These complicated, wide-ranging abnormalities underscore the importance of the phenotypic analysis of individual axons (Wang, 2002).

Thus Dscam is involved in regulating the bifurcation of MB axons. Instead of having two branches that project away from each other, Dscam mutant axons give rise to supernumerary branches, through repeated bifurcation, that fail to extend in divergent directions. It is likely that homophilic interactions between identical Dscam molecules mediate novel mechanisms that coordinate induction of axon bifurcation with segregation of sister branches. In addition, Dscam mutant axons can alter the projections of wild-type axons within the same MB. Further investigations into the cellular basis of such non-cell-autonomous effects will shed new light on how the insect olfactory learning and memory center acquires its normal projection patterns during development (Wang, 2002). Several different scenarios can account for the observed correlation of ectopic bifurcation with abnormal segregation in Dscam mutant axons. One possibility is that Dscam activity might play a direct role in both divergent segregation of sister branches and suppression of ectopic bifurcation. These two activities could occur via a common mechanism or through two independent signaling events. Alternatively, the role of Dscam in divergent segregation of sister branches could be a secondary consequence of its suppression of axon bifurcation, or vice versa. The idea is favored that Dscam directly controls both formation and segregation of axonal branches because in single mutant cells, abnormal segregation is not always coupled with the generation of additional branches and vice versa. Most likely, Dscam normally prevents sister branches from extending along the same path and consequently suppresses additional bifurcation after the sister branches have occupied all available paths (Wang, 2002).

One mechanism that can explain coordination of axon bifurcation with divergent segregation is Dscam-dependent growth cone collapse. When sister growth cones contact each other, homophilic interactions between Dscam molecules may lead to growth cone collapse or fusion. Thus, when axons bifurcate, only divergently split growth cones could survive, initiating separation of sister branches. Moreover, new growth cones continue to bifurcate until all target sites receive branches; but only one growth cone derived from any individual neuron could traverse a given path because the Dscam-mediated homophilic interactions would immediately collapse adjacent sister growth cones into one. This mechanism would couple formation of axonal branches to divergent guidance of sister branches. The final result would be that individual axons would innervate multiple targets through minimal bifurcation. In contrast, sister branches may extend along the same path after loss of Dscam activity. Consequently, when bifurcation is induced in mutant axons, they randomly distribute their branches and often generate additional branches through repeated bifurcation (Wang, 2002).

This model also suggests that induction of axon bifurcation may not always lead to formation of sister branches. Growth cones split in response to bifurcation-inducing signals. However, if twin growth cones cannot migrate divergently, then no axon bifurcation would be expected after futile cycles of splitting followed by collapse. This mechanism can explain why bifurcation is suppressed in wild-type axons when only one trajectory is left. Despite the lack of bifurcated bundles, induction of axon bifurcation persists in mosaic MBs since Dscam mutant axons still generate multiple branches at the original bifurcation point. Taken together, these observations further support the proposal that divergent guidance is essential for the survival of twin growth cones in the presence of normal Dscam activity (Wang, 2002).

Remarkably, Drosophila Dscam may exist in numerous isoforms through alternative splicing, and Dscam is widely required for axon arborization in distinct CNS neurons. Given that various Dscam isoforms exhibit distinct features in their extracellular domains while sharing common intracellular structures, it is likely that activation of a common Dscam-dependent signaling pathway is dynamically and differentially regulated in distinct growth cones expressing distinct sets of Dscam's. Although immunohistochemistry using an anti-Dscam Ab has revealed general expression of Dscam in most pupal brain structures, it remains unclear whether distinct Dscam isoforms are expressed in different neurons. However, delicate cell type-specific controls of Dscam signaling could be achieved if homophilic interactions were restricted to identical or certain pairs of Dscam isoforms. In addition, activation of Dscam might lead to different developmental changes, depending on the types of cellular structures that are involved. For instance, Dscam-Dscam interactions between split growth cones may result in reunification of growth cones. In contrast, homophilic Dscam binding could mediate contact-dependent attraction of growth cones when the interactions occur between growth cones and their guiding substrates. Such Dscam-mediated growth cone guidance might be crucial to normal extension of GAL4-201Y-positive alpha/ß neurons as well as correct pathfinding of the Bolwig's nerve. If the Dscam pathway is widely used to mediate various growth cone behaviors, it is understandable that numerous Dscam isoforms would be needed to confer diverse activation patterns in different types of growth cones (Wang, 2002).

Analysis of the non-cell-autonomous effects of Dscam mutant clones provides convincing evidence that the first-born alpha'/ß' neurons play a crucial role in shaping the projection patterns of all later-born MB neurons. If Nb clones are induced after the initiation of alpha'/ß' neuron production, both dorsal and medial MB lobes are always observe. In contrast, if early alpha'/ß' neurons are made homozygous for a Dscam mutation, absence of either the dorsal or medial lobe is observed in about one-third of mosaic MBs. Interestingly, consistent patterns exist between alpha'/ß' lobes and alpha/ß lobes. Given that alpha'/ß' neurons, unlike gamma neurons, maintain their projection patterns through metamorphosis, the effects of early alpha'/ß' neurons on the final organization of MB lobes support roles of alpha'/ß' axons as pioneer axons that guide the projections of adult-specific MB neurons. Because alpha'/ß' axons and alpha/ß axons form distinct bundles, the effects of alpha'/ß' axons on the projections of alpha/ß axons may be due to pioneer axon-mediated changes in the arrangement of local glial cells. Since one MB is composed of four indistinguishable clonal units, how several cells within one unit can dominate the entire MB projection patterns remains to be elucidated (Wang, 2002).

The demonstration of Drosophila Dscam's essential roles in the bifurcation of MB axons immediately suggests many lines of research for future investigation of the molecular mechanisms underlying growth cone splitting. For instance, it remains to be determined whether the Dock/Pak signaling pathway is involved in bifurcation of various axons. In addition, it is uncertain whether the vertebrate Dscam plays similar roles during development of the CNS. Identification of additional molecules that control bifurcation of axons in Drosophila may generate new insights into how similar neuronal morphogenetic processes are regulated in vertebrates (Wang, 2002).

Different classes of olfactory receptor neurons (ORNs) in Drosophila innervate distinct targets, or glomeruli, in the antennal lobe of the brain. Specific ORN classes require the cell surface protein Dscam (Down Syndrome Cell Adhesion Molecule) to synapse in the correct glomeruli. Dscam mutant ORNs frequently terminate in ectopic sites both within and outside the antennal lobe. The morphology of Dscam mutant axon terminals in either ectopic or cognate targets is abnormal. Target specificity for other ORNs is not altered in Dscam mutants, suggesting that different ORNs use different strategies to regulate wiring. Multiple forms of Dscam RNA are detected in the developing antenna, and Dscam protein is localized to developing ORN axons. A role is proposed for Dscam protein diversity in regulating ORN target specificity (Hummel, 2003).

Dscam expression was assessed in the developing olfactory system using both in situ hybridization and immunohistology. Between 30% to 40% of pupal development, Dscam RNA is seen in the third antennal segment that contains differentiating ORNs. At this stage of development, many ORNs extend axons into the antennal lobe. Within the target region, Dscam RNA is observed in most, and perhaps all, neuronal cell bodies surrounding the antennal lobe; this region includes the cell bodies of both projection neurons and local interneurons. Dscam protein is highly enriched in ORN axons in early to mid pupal development, with little immunoreactivity apparent on cell bodies, as assessed using an antibody directed to a domain shared by all Dscam isoforms. Similarly, little Dscam immunoreactivity was observed on cell bodies of interneurons and projection neurons. At 40% pupal development, Dscam protein is seen on ORN axons in the peripheral nerve fiber layer surrounding the developing antennal lobe. Dscam is also detected on the centrally located dendritic processes of antennal lobe neurons. At later stages, Dscam immunoreactivity is seen uniformly distributed within developing glomeruli and is markedly downregulated in the antennal lobe of adult flies. In contrast to the vertebrate olfactory system, ORNs in flies are not generated throughout the life of the animal, and hence, it would not be surprising that genes that regulate targeting in the fly olfactory system would be downregulated in the adult (Hummel, 2003).

To assess the diversity of Dscam isoforms expressed in the developing ORNs, third antennal segments at 30% pupal development were isolated and the RNA isoforms expressed were analyzed using RT-PCR. Seventy-two different clones from six different antennae were sequenced, and sixty-eight of these comprised different combinations of alternative exons 4, 6, and 9 (encoding amino acids in Ig2, Ig3, and Ig7, respectively). No striking preferential patterns of alternative exon utilization or combinations of exons were apparent. While Dscam protein is largely expressed in ORNs at this stage, the possibility cannot be ruled out that Dscam RNA but not protein is expressed in other cells. Since markers for different subclasses of ORNs are expressed after glomeruli are formed, it has not yet been possible to assess the forms of Dscam expressed in different subclasses of ORNs as they target to specific glomeruli. Nevertheless, these data are consistent with many different forms of Dscam protein being produced by ORNs, since they project into the target region and that individual ORNs express multiple Dscam isoforms (Hummel, 2003).

Dscam plays a complex role in regulating the formation of specific connections in the fly olfactory system. In the absence of Dscam, two classes of antennal ORNs, Or47a and GH298, project into the antennal lobe, as they do in wild-type, but terminate at ectopic sites they encounter as they project to their cognate glomeruli. That Dscam prevents inappropriate targeting is underscored by the robust mistargeting of three classes of maxillary palp ORNs to neuropil outside the antennal lobe. In addition, ORN axons that reach the correct glomerulus do not elaborate a normal terminal region. They fail to extend thin processes throughout the cognate glomerulus as in wild-type and remain tightly segregated in local domains within the glomerulus. All ORNs studied that branch in wild-type to the contralateral glomerulus frequently fail to do so in Dscam mutants. It seems likely that defects in branching result from abnormalities in the interactions between ORNs and their cognate ipsilateral glomeruli. For some ORNs, branches target accurately to their cognate glomeruli on the contralateral antennal lobe, whereas in others they terminate in inappropriate locations. While there are differences in Dscam phenotypes in different ORNs, as well as their penetrance and expressivity, these observations support a critical role for Dscam in mediating interactions between growth cones and neuropil within potential target regions (Hummel, 2003).

Are alternative forms of Dscam required for targeting? In recent studies, attempts have been made to assess whether expression of a single form of Dscam would rescue the targeting defects in specific olfactory receptor neurons. Since expression in ORNs leads to a severe dominant phenotype with loss of glomerular structure, it has not been possible to critically address the importance of multiple isoforms for targeting using this experimental approach. In future studies, the importance of alternative splicing in target selection will be assessed by analyzing loss-of-function mutants in which the number of alternative exons is reduced at the endogenous locus and by expressing additional isoforms of Dscam in ORNs (Hummel, 2003).

It is thought that ORNs and projection neurons express molecular labels that allow them to match-up in distinct glomeruli Could the alternative forms of Dscam provide such labels? This is indeed a possibility, since Dscam is expressed on both ORNs and projection neurons during glomerular formation. That Dscam could act as a short-range attractant is consistent with the observation that mammalian Dscam proteins can bind to each other when expressed in transfected cells. In Drosophila, forms expressed on specific ORNs and the appropriate target projection neurons would selectively adhere to each other. Alternatively, Dscam could act as a short-range repellent. Projection neurons that extend dendrites selectively into only a single glomerulus may express a diverse set of Dscam isoforms, and different ORN subclasses express forms of Dscam that interact with isoforms expressed in dendrites of all projection neurons, except those in their cognate glomeruli. As a consequence, ORNs connect to their appropriate targets by being excluded from all other potential targets. Repulsion may be mediated by interactions between the same Dscam isoforms or between different isoforms; different isoforms may modulate the extent of repulsion between ORNs and target glomeruli, and hence, competitive interactions between ORN classes may contribute to specificity (Hummel, 2003).

While Dscam is necessary for ORN targeting, it is clear that it is not the only ORN determinant regulating this process. Indeed, some ORN targeting is independent of Dscam (i.e., Or22a and Or23a), and in many cases, subclasses affected by Dscam mutations exhibit only partially penetrant phenotypes. This may reflect the overlapping function of other Dscams (i.e., Dscams 2-4; these Dscams do not come in multiple forms) that are expressed within the olfactory system or the activity of redundant pathways utilizing other cell surface recognition molecules. Indeed, other targeting proteins, such as N-cadherin and Flamingo, contribute to targeting of some but not all classes of ORNs. These studies allude to a complex combinatorial mechanism regulating olfactory receptor neuron targeting in Drosophila. The availability of markers for different neurons exhibiting different targeting specificities within a common ganglion provides a unique opportunity to explore the mechanisms regulating the formation of diverse patterns of synaptic connectivity (Hummel, 2003).

Dendritic patterning by Dscam and synaptic partner matching in the Drosophila antennal lobe

In the olfactory system of Drosophila melanogaster, axons of olfactory receptor neurons (ORNs) and dendrites of second-order projection neurons typically target 1 of ~50 glomeruli. Dscam, an immunoglobulin superfamily protein, acts in ORNs to regulate axon targeting. Dscam acts in projection neurons and local interneurons to control the elaboration of dendritic fields. The removal of Dscam selectively from projection neurons or local interneurons leads to clumped dendrites and marked reduction in their dendritic field size. Overexpression of Dscam in projection neurons causes dendrites to be more diffuse during development and shifts their relative position in adulthood. Notably, the positional shift of projection neuron dendrites causes a corresponding shift of its partner ORN axons, thus maintaining the connection specificity. This observation provides evidence for a pre- and post-synaptic matching mechanism independent of precise glomerular positioning (Zhu, 2006).

The shift of position of the projection neurons of the VA1d glomerulus, caused by the overexpression of Dscam in VA1d neurons, offers a unique opportunity to investigate the logic of olfactory circuit assembly. Because projection neuron dendrites form a coarse spatial map before ORN axons invade, ORN axons could use the projection neuron dendritic map as a cue to determine their spatial position in the antennal lobe. To date, however, there is no experimental support for this hypothesis. Alternative proposals are that ORN axons recognize non–projection neuron cues within or surrounding the antennal lobe and/or that they self-organize through axon-axon interaction to reach their spatially invariant targets (Zhu, 2006).

The overexpression of Dscam in projection neurons alters the initial spatial map of the projection neuron dendrites before ORN invasion; therefore, it was asked whether the change in the spatial map of projection neuron dendrites affects the targeting of the corresponding ORN axons. If projection neuron dendritic maps are used as cues for ORN axon targeting, one would predict a corresponding shift of ORN axon position. Alternatively, if the spatial positions of ORN axons are solely determined by interactions among ORN axons or with non-projection neuron cues, changing the spatial map of projection neuron dendrites would lead to a mismatch of projection neuron dendrites and ORN axons and would result in a change in connection specificity (Zhu, 2006).

To distinguish between these two possibilities, ORN axon innervation patterns in the antennal lobe were examined in response to the dendritic map shift induced by Dscam overexpression in projection neuron dendrites. The ORNs innervating the VA1d glomerulus express Or88a. To label these ORN axons, the transgene Or88a-CD2 was used in which the Or88a promotor drives the expression of the membrane marker CD2. In the same brain, Mz19⁺ projection neurons were labeled by the membrane marker mCD8-GFP using the Gal4/UAS system. In the wild type, Or88a axons synapse with Mz19⁺ projection neuron dendrites innervating VA1d, which is adjacent to the DA1 glomerulus. In 13 brain hemispheres where Mz19⁺ projection neurons also overexpressed Dscam, the spatial position of the VA1d projection neuron dendrites shifted ventrally, away from the DA1 glomerulus. Notably, Or88a axons shifted to a ventral position accordingly in each case, maintaining the correct connections with the VA1d projection neuron dendrites. In the seven hemispheres in which Dscam overexpression did not induce a dendritic shift, Or88a axon targeting was normal. These results demonstrated that Or88a axons recognized VA1d projection neuron dendrites, thereby maintaining their connection specificity even when these dendrites were shifted to a different position (Zhu, 2006).

Overexpressing Dscam in Mz19⁺ projection neurons shifted the VA1d dendrites to a ventral position equivalent to VA1lm, which is normally innervated by ORN axons expressing Or47b. Would Or47b axons still innervate the same position in the antennal lobe and therefore connect incorrectly with the ventrally shifted VA1d projection neuron dendrites? Or would they avoid the ventrally shifted VA1d dendrites and target a new position? To address this question, the Or47b axons were labelled using the transgene Or47b-CD2. It was found that, in each hemisphere, when Dscam overexpression caused a ventral shift of VA1d dendrites, Or47b axons shifted to a dorsal position and innervated the glomerulus situated ectopically between DA1 and VA1d, avoiding the VA1d dendrites completely. In hemispheres where no projection neuron dendritic position shift occurred, Or47b axons targeted normal positions. Note that for this VA1lm glomerulus, neither the pre- nor the postsynaptic partners had been subjected to any genetic modifications. Their positional shifts were the sole result of the expression of Dscam in projection neurons in glomeruli neighboring VA1lm. Together, these data supported the notion that glomerular targeting, at least in the two ORN classes examined, requires specific recognition between ORN afferents and projection neuron dendrites (Zhu, 2006).

Individual neurons express different arrays of isoforms in a largely stochastic fashion. It is proposed that such expression provides each neuron with a unique identity. Biochemical studies demonstrate that Dscam promotes isoform-specific binding: isoforms sharing the same extracellular domain bind to each other, whereas isoforms that are different do not. Furthermore, interactions between identical Dscam proteins seem to promote contact-dependent repulsion. One phenotype in the Dscam mutant axon is the failure to segregate sister branches, which is also the case for projection neuron axons (Zhu, 2006).

How might Dscam contribute to dendrite branching? To efficiently innervate the target areas, some neurons possess an intraneuronal tiling mechanism such that dendritic branches of the same neuron do not overlap. On the basis of expression studies in other neurons, it is proposed that the array of Dscam isoforms expressed on a dendrite's surface provides each projection neuron and local interneuron with a mechanism by which to distinguish its own dendrites from those of neighboring cells. Thus, Dscam isoforms expressed on sister branches will be the same; however, they will share little, if any, overlap with the isoforms expressed on dendrites of other projection neurons or local interneurons innervating the same glomurulus. Thus, Dscam diversity provides a mechanism by which dendrites from the same neuron can avoid each other as they elaborate their receptive fields, while overlapping with dendritic processes of other cells within the neuropil (Zhu, 2006).

Developmental studies have led to a model in which ORNs and projection neurons initially develop spatial maps that are independent of each other. When the two spatial maps converge, the connection specificity between a given ORN-projection neuron pair can be determined solely by matching the positional coordinates of the respective spatial maps for ORN axons and projection neuron dendrites. Such an extreme model would predict that if a change is made to the positions of projection neuron dendrites before and independent of the arrival of ORN axons, the positional maps of ORN axons and projection neuron dendrites would be misaligned. Under these circumstances, ORN axons should innervate projection neuron dendrites positioned as they would be in the wild type, rather than as they are in the reordered state; thus the ORN axons would select the incorrect target. Alternatively, these two maps may be rather coarse and specific recognition between appropriate ORN axons and their dendritic targets may be required to refine the map (Zhu, 2006).

The findings that changing the spatial map of the projection neuron dendrites also leads to corresponding changes of the ORN axon map strongly argues against a strict spatially regulated matching of the afferent axonal and dendritic maps. The data indicate that ORN axons are influenced by their corresponding postsynaptic projection neuron dendrites in determining their spatial position in the antennal lobe, at least at the local level for the two classes of ORNs analyzed. This finding by no means contradicts the contribution of projection neuron-independent mechanisms in ORN axon targeting. Such mechanisms may be used to set up a coarse map for the ORN axons, limiting the approximate spatial position that a given ORN axon can target. ORN axons would then be faced with only a few projection neuron dendritic targets in the neighborhood, which they can sample through ORN-projection neuron recognition in order to solidify the final synaptic partners. Sensory input and synaptic activity are unlikely to play a role in matching ORN axons and projection neuron dendrites because this process is completed before olfactory receptor expression and synaptic maturation. It is more likely that these processes recognize each other through yet to be identified 'chemoaffinity tags' (Zhu, 2006).

Homophilic Dscam interactions control complex dendrite morphogenesis

Alternative splicing of Dscam results in up to 38,016 different receptor isoforms proposed to interact by isoform-specific homophilic binding. Dscam controls cell-intrinsic aspects of dendrite guidance in all four classes of dendrite arborization (da) neurons. Loss of Dscam in single neurons causes a strong increase in self-crossing. Restriction of dendritic fields of neighboring class III neurons appeared intact in mutant neurons, suggesting that dendritic self-avoidance, but not heteroneuronal tiling, may depend on Dscam. Overexpression of the same Dscam isoforms in two da neurons with overlapping dendritic fields forced a spatial segregation of the two fields, supporting the model that dendritic branches of da neurons use isoform-specific homophilic interactions to ensure minimal overlap. Homophilic binding of the highly diverse extracellular domains of Dscam may therefore limit the use of the same 'core' repulsion mechanism to cell-intrinsic interactions without interfering with heteroneuronal interactions (Hughes, 2007).

This study shows that Dscam has an important cell-intrinsic function in dendrite development of da neurons. Dscam is required for steering the growth of sister branches to ensure correct dendrite morphogenesis but is not required for other mechanisms of dendrite patterning. Dscam loss-of-function mutations result in strong disruption of dendrite morphogenesis in different classes of da sensory neurons. The phenotypic defects included uneven spacing of dendritic branches, a strong increase in dendritic self-crossing, and highly abnormal dendritic fascicles or tangles. All the observed phenotypes are consistent with the possibility that loss of Dscam results in a lack of self-avoidance of sister dendrites. Consistent with a role of Dscam in dendrite-dendrite repulsion, it was found that Dscam overexpression in da neurons, which normally have overlapping dendritic fields, forced the respective dendrites to segregate from each other. In addition, gain-of-function phenotypes resulting from overexpression of single Dscam isoforms or a Dscam isoform lacking the cytoplasmic domain are also consistent with the possibility that repulsion between sister branches is controlled by Dscam signaling. It is therefore suggested that in da neurons, direct isoform-specific homophilic Dscam-Dscam interactions result in signal transduction events that lead to repulsion of dendrites expressing identical Dscam isoforms. This model is consistent with previous biochemical studies, Dscam's role in bifurcating MB axons, and Dscam's function in projection neurons of the olfactory system (Hughes, 2007).

Previous expression studies have shown that single photoreceptor neurons of the same type express different Dscam isoforms. Similarly, expression of a large diversity of Dscam isoforms has also been found in olfactory neurons and MB neurons. Based on these findings and the observation that experimentally forced expression of identical isoforms in da neurons causes dominant phenotypes, it is highly likely that different da neurons also express diverse Dscam isoforms. Considering the large diversity of Dscam isoforms, the possibility that dendrites from different neurons present identical Dscam isoforms seems minimal. In contrast, dendritic sister branches of the same cell, even though they are likely expressing multiple isoforms, will at significant frequency encounter homophilic Dscam-Dscam interactions. This model is consistent with the finding that the expression of the same Dscam isoform causes the segregation of normally overlapping dendritic fields. Based on this one might expect that a functionally critical threshold of Dscam diversity must exist. However, no obvious morphogenesis defects of class I neurons, such as changes in self-crossing, number of dendritic termini, or dendritic area, were detected in homozygous animals bearing the reduced diversity allele Dscam^C22-1. Also no obvious defects were detected in dendrite morphogenesis of class IV neurons in Dscam^C22-1 animals. Although it is possible that some aspects of dendrite morphogenesis are altered in Dscam^C22-1 and were not identified in these experiments, the possibility is favored that a few thousands or even significantly fewer Dscam isoforms are sufficient to still ensure nonoverlapping expression of identical isoforms in neighboring da neurons. However, it is proposed that reducing the diversity of Dscam isoforms below a certain threshold would lead to scenarios where different neighboring da neurons express the same isoforms and it would not be possible to limit Dscam-mediated repulsion of dendrites to cell-intrinsic sister dendrite interactions. This would likely lead to strong morphogenesis and functional defects throughout the Drosophila PNS (Hughes, 2007).

Dendrite development of da neurons requires at least four distinguishable patterning mechanisms: (1) growth of dendrites and dendritic branches emanating from the same cell has to be controlled such that relatively even spacing between dendrites with minimal overlap is achieved (self-avoidance); (2) for any given class or type of neuron, the dendritic growth has to obey a characteristic polarity and likely limits the extension of the primary dendritic branches (dendrite architecture); (3) the degree of branching has to be adapted to the type of sensory neuron (stereotyped branching); (4) inhibitory interactions with nearby neurons are needed to control the size of dendritic fields such that a complete but nonredundant innervation of a receptive area by functionally uniform groups of neurons is achieved (heteroneuronal tiling) (Hughes, 2007).

It has been speculated that self-avoidance and tiling might depend on the same molecular mechanism and may not require distinct signals. In such a scenario, isoneuronal dendrites could be developmentally identical to 'like' heteroneuronal dendrites. In this study it was suggested that this may not be the case. Dscam function is required for correct spacing of dendrites due to self-avoidance of sister branches but is unlikely required for other mechanisms of dendrite patterning. These results suggest that the repulsive mechanism or mechanisms underlying hetero-neuronal tiling are molecularly different from the mechanism controlling repulsive interactions underlying self-avoidance. It seems likely that homophilic Dscam-Dscam interactions represent the major molecular system controlling isoneuronal dendrite-dendrite repulsion in Drosophila. In this specific context of dendrite morphogenesis, the diversity of Dscam ensures that this repulsive function is restricted to cell-intrinsic interactions, as only dendrites of the same cell are likely to express identical isoforms. As such, the molecular diversity of Dscam is less likely to provide each neuron with a unique 'identity' but rather provides a molecular buffer for enabling 'tolerance' between neurons (Hughes, 2007).

Several studies have revealed examples consistent with the notion that Dscam signaling can lead to neurite repulsion. It has been proposed that this repulsive function can be mediated by direct homophilic Dscam-Dscam interactions. For example, it has been shown that the trajectory of interneurons overexpressing a single isoform of Dscam is disrupted upon encountering midline cells that overexpress the identical isoform. The strongest support for a direct Dscam-Dscam interaction has been provided by a series of impressive biochemical experiments, in which it was shown that from a randomly chosen set of 11 Dscam isoforms, each one binds to itself but not to others. All three variable Ig domains of Dscam are required for homophilic binding specificity. In addition, recent studies described that overexpression of Dscam in a subset of projection neurons connecting with specific glomeruli (termed DA1 and DC3) resulted in a strong gain-of-function phenotype, again consistent with a repulsive interaction due to homophilic Dscam interactions. This gain-of-function phenotype was found to be dependent on Dscam signaling, as a deletion of the cytoplasmic domain in a Dscam isoform (Dscam1.30.30.1ΔC) blocked this dominant phenotype. Similarly, it was found that overexpression of Dscam1.30.30.1ΔC in ddaE neurons blocked the repulsion of sister dendrites and instead lead to abnormal fasciculation. Although the endogenous physiological function revealed by these experiments is unclear, they nevertheless are consistent with the hypothesis that Dscam can function as a cell-surface receptor mediating neurite repulsion (Hughes, 2007).

How is the homophilic Dscam-Dscam interaction transformed into a repulsive action rather than a stable adhesion? Dscam has been initially identified as a tyrosine phosphorylated receptor functioning upstream of the adaptor molecule Dock. Dock binds to Dscam via SH2 as well as SH3 domains and serves to recruit the effector kinase Pak to the plasma membrane where it can be activated by Rac or Cdc42. Pak has been implicated in several signaling pathways that control cytoskeletal rearrangement, including pathways underlying neurite repulsion. By examining the effect of a constitutively membrane-bound form of Pak in da neurons, it was found that Pak signaling can influence dendrite morphogenesis. However, loss-of-function analysis provided no evidence for a direct role of Pak in dendrite morphogenesis or self-avoidance of class I neurons. Therefore, at least in class I neurons, Dscam signaling likely bypasses Pak and utilizes alternative downstream components that have yet to be identified. Although a signaling pathway controlling heteroneuronal tiling and branching in da neurons has been described, signaling pathways that control self-avoidance are currently unknown (Hughes, 2007).

It is important to note that Dscam function is not only required for controlling neurite repulsion. For example, it has been proposed that Dscam controls axon guidance of Bolwig's nerve by signaling through Dock and Pak in response to an as of yet unknown guidance cue present at an intermediate target. In early developing MB fibers, Dscam is required for axon bundling and fasciculation, thereby mediating adhesive interactions. In addition, one distinct function of Dscam in mechanoreceptor neurons appears to mediate a growth-promoting role rather than repulsion. Importantly, the role of Dscam diversity in the development of mechanosensory neuron projections suggests the possibility that Dscam is not only involved in homophilic interactions of neurites emanating from the same cell. In fact, it has been proposed that in the somatosensory system Dscam isoforms have instructive roles controlling targeting decisions of axonal branches. Future studies will have to address the molecular differences that allow for such a versatile use of Dscam receptors (Hughes, 2007).

Drosophila sensory neurons require Dscam for dendritic self-avoidance and proper dendritic field organization

A neuron's dendrites typically do not cross one another. This intrinsic self-avoidance mechanism ensures unambiguous processing of sensory or synaptic inputs. Moreover, some neurons respect the territory of others of the same type, a phenomenon known as tiling. Different types of neurons, however, often have overlapping dendritic fields. Dscam is required for dendritic self-avoidance of all four classes of Drosophila dendritic arborization (da) neurons. However, neighboring mutant class IV da neurons still exhibit tiling, suggesting that self-avoidance and tiling differ in their recognition and repulsion mechanisms. Introducing 1 of the 38,016 Dscam isoforms to da neurons in Dscam mutants is sufficient to significantly restore self-avoidance. Remarkably, expression of a common Dscam isoform in da neurons of different classes prevented their dendrites from sharing the same territory, suggesting that coexistence of dendritic fields of different neuronal classes requires divergent expression of Dscam isoforms (Soba, 2007).

Given the necessity for different classes of da neurons to express at least one isoform of Dscam to ensure self-avoidance, it is important to explore the possible consequences of having multiple da neurons in the same region express the same Dscam isoform. Different classes of da neurons presumably respond to different sensory inputs, and hence their coverage of overlapping regions of the body wall will allow the animal to detect different types of sensory stimuli at the same physical locations. Expression of a common Dscam isoform made it almost impossible for different classes of da neurons to occupy the same territory, which is likely associated with deprivation of all but one type of sensory input at any one location. These considerations provide a plausible explanation for the existence of a vast number of Dscam splice variants. For Dscam to mediate self-recognition manifested as dendritic self-avoidance in da neurons, it may be crucial to limit the expression of each of the Dscam isoforms to one or a small subset of neurons in any region of the nervous system. This allows neighboring neurons to interact in ways distinct from the interactions that occur between isoneuronal processes of an individual neuron. Indeed, microarray analyses of individual Drosophila photoreceptor cells and mushroom body neurons have found distinct Dscam isoforms expressed in neighboring cells. This supports the idea that an intricate splicing mechanism ensures that individual neurons can distinguish between self and nonself. Given that divergent Dscam isoform expression is required for coexistence of da neurons in the same territory, it is likely that other mechanisms are involved in recognition of neighboring class IV da neurons for the purpose of tiling (Soba, 2007).

Intriguingly, even duplicated class I neurons most likely express distinct Dscam isoforms, since they occupy the same territory, but repel each other when expressing a common Dscam isoform. While the specific isoforms expressed by individual da neurons are currently unknown, the general principle of enabling dendrites of a single neuron to avoid one another without imposing recognition and avoidance of neighboring neurons underscores the importance of Dscam diversity. Since Dscam has been shown to primarily interact in an isoform-specific manner, Dscam-based self-recognition and self-avoidance depends on expression of the same isoforms. This is illustrated by the repulsive function of Dscam in vivo, where it induces branch retraction and avoidance of dendrites expressing identical isoforms. The self-avoidance phenomenon seems to be conserved in axonal development as well: in mushroom body neurons Dscam ensures the proper segregation and targeting of the two axonal processes to different lobes (Soba, 2007).

This study reveals significant differences between self-avoidance, which encompasses isoneuronal crossing and bundling of dendrites, and tiling. Dscam is essential for dendritic self-avoidance (dendritic crossing and bundling) of all classes of da neurons, but not for tiling of class IV da neurons. Thus, different recognition mechanisms are used for self-avoidance and tiling. The phenomenon of tiling, in which the dendrites of different neurons of the same class avoid one another, relies on components of the Hpo/Trc/Fry pathway. Whereas this signaling pathway also contributes to the avoidance of isoneuronal dendritic crossing of class IV da neurons, it is not required to prevent bundling of class IV da neuron dendrites, nor is it important for self-avoidance of the other three classes of da neurons. It is an interesting open question as to what intracellular machineries are employed to prevent the dendrites of each da neuron in class I-III from crossing one another. The involvement of the Hpo/Trc/Fry pathway in preventing crossing, but not bundling, of dendrites from the same class IV da neuron further suggests the possibility of multiple mechanisms for self-avoidance. While fasciculation of multiple processes, e.g., those in axon guidance, is a widely used mechanism to ensure accurate projection to common target areas, bundling of isoneuronal dendrites defeats the purpose of dendritic field coverage and nonredundant signal processing. Given the ample opportunities for neighboring dendritic branches of the same neuron to bundle, there is likely a specialized mechanism that repels branching dendrites from each other as they respond to cues for their extension (Soba, 2007).

Dendrite self-avoidance is controlled by Dscam

Dendrites distinguish between sister branches and those of other cells. Self-recognition can often lead to repulsion, a process termed 'self-avoidance.' Dendrite self-avoidance in Drosophila da sensory neurons requires cell-recognition molecules encoded by the Dscam locus. By alternative splicing, Dscam encodes a vast number of cell-surface proteins of the immunoglobulin superfamily. Interactions between identical Dscam isoforms on the cell surface underlie self-recognition, while the cytoplasmic tail converts this recognition to dendrite repulsion. Sister dendrites expressing the same isoforms engage in homophilic repulsion. By contrast, Dscam diversity ensures that inappropriate repulsive interactions between dendrites sharing the same receptive field do not occur. The selectivity of Dscam-mediated cell interactions is likely to be widely important in the developing fly nervous system, where processes of cells must distinguish between self and nonself during the construction of neural circuits (Matthews, 2007).

These data demonstrate a cell-autonomous role for Dscam function in self-avoidance in all classes of Drosophila da neurons. da neurons associate closely with the epidermis as they extend across the body wall; thus, their dendrites create a two-dimensional meshwork in which developing branches frequently encounter other dendrites. This is in contrast to the layout of the CNS, in which axons and dendrites usually elaborate in three dimensions. By examining da neurons, it was possible to analyze the behavior of individual branches within a single dendritic arbor at high resolution. This allowed critical quantitative examination of the mechanisms underlying selective recognition between dendrites. The data show that when deficient in Dscam function, individual dendrites do not recognize sister branches and fail to initiate repulsion, leading to a breakdown in self-avoidance. Individual branches of Dscam mutant cells often failed to evenly disperse across their territory. Additionally, processes from specific da neurons gathered at nonrandom, discrete target sites within their territory (see below) (Matthews, 2007).

Dscam is likely to play a similar role in the CNS based on axonal- and dendritic-arborization phenotypes. In the olfactory system, for example, the terminal processes of single mutant olfactory receptor neurons, projection neurons, and interneurons form clumps. While this might reflect a phenotype of self-avoidance, the resolution of these studies was not sufficient to distinguish between self-avoidance and other mechanisms such as branch extension and synapse formation. By contrast, single-branch resolution has been achieved for Dscam defects in the axonal projections of MB neurons. Dscam is required for proper segregation of sister axon branches, and analogous to the self-avoidance control in da neurons, specific isoforms do not appear to provide instructive cues for this segregation event. It has been argued that this reflects a role for Dscam in mediating self-recognition and repulsion between these axons. In da neurons, dendrite self-avoidance defects were separable from growth, branching, and targeting errors and were fully penetrant. Thus, these data directly implicate Dscam in self-avoidance and demonstrate this role at the level of interactions between individual branches (Matthews, 2007).

The simplest model for a direct role for Dscam in self-recognition is one in which identical Dscam ectodomains on the surfaces of isoneuronal dendrites recognize each other and induce a subsequent repulsive signal that is mediated by domains in the cytoplasmic tail. This model is supported by both in vitro and in vivo data presented in this paper. First, identical Dscam isoforms expressed in two cell populations in vitro induced their aggregation in an isoform-specific manner, showing that Dscam provides cells with the ability to distinguish between different cell surfaces. Second, ectopic expression of identical Dscam isoforms on the dendrites of different cells, which normally overlap, promoted growth away from each other (Matthews, 2007).

How can in vitro adhesion be reconciled with in vivo repulsion? The data suggest that the dendrites of da neurons convert an initial Dscam-dependent cell-surface interaction into a repulsive response, which leads to dendrite separation and receptive field elaboration. da dendrites expressing a form of Dscam in which the cytoplasmic domain was replaced with GFP formed stable bridges. These data are reminiscent of studies demonstrating that complexes of ephrin-A2 and EphA3 are intermediates in heterophilic repulsive interactions in cell culture. Ephrin-A2 is normally cleaved by a metalloprotease, and cleavage-resistant mutations lead to more stable interactions between growth cones and target cells (Matthews, 2007).

The signal transduction mechanism promoting repulsion is poorly understood. At least some self-avoidance activity derives from sequences encoded by exon 18 in the Dscam cytoplasmic tail, which includes a polyproline motif. In previous studies, the Dock adaptor protein was shown to bind to this region as well as to other sites on the cytoplasmic domain and to act downstream of Dscam in axon guidance. While Dock has been implicated in the repulsive signaling downstream from the slit receptor, Robo, loss-of-function Dock mutations caused no obvious self-avoidance defects in da neurons. Dock may not function in self-avoidance or, alternatively, it may be redundant with other signaling pathways. The Tricornered (Trc) signaling pathway was previously shown to regulate tiling and self-avoidance in class IV neurons; however, examination of animals carrying transheterozygous mutant combinations did not uncover a genetic interaction between trc and Dscam (Matthews, 2007).

Alternative splicing of Dscam pre-mRNA can generate an enormous number of distinct cell-surface receptors. Is Dscam diversity, or any specific Dscam isoform, necessary for self-avoidance in individual da neurons? The results argue that while diversity is not strictly required for self-recognition and repulsion, it is crucial to prevent inappropriate repulsive interactions from occurring between the dendrites of different cells. This may be a central function for Dscam diversity both in different functional groups of sensory neurons in the PNS, which must sample input from overlapping regions of the body wall, and in regions of the CNS with much more highly intermingled dendritic and axonal processes (Matthews, 2007).

Previous data suggest an analogous function for Dscam diversity in mediating the sorting of axons in the developing MB. MB axon phenotypes were partially rescued by expression of single isoforms, whereas ectopic expression across multiple cells gave dominant effects, in which axons were guided to improper targets. These data together with studies described in this paper indicate that axons that project along a common fascicle or dendrites with overlapping fields must express sufficiently different isoform repertoires. Supporting this scenario, expression of Dscam isoforms in MB neurons, as well as photoreceptor subtypes, appears to be specified through a stochastic mechanism whereby each neuron expresses a biased, yet largely nonspecific set of isoforms. Given the complexity of the Dscam locus, it is reasonable to expect that different roles for diversity will be observed in different cell populations or even different processes of a cell. For example, these results are not incompatible with Dscam diversity also contributing to wiring in a more deterministic fashion wherein specific isoforms are required for elaborating different aspects of neural circuits (Matthews, 2007).

Dendritic arbors respond to numerous intrinsic and extrinsic cues during morphogenesis. How might self-avoidance mechanisms operate in the context of these other patterning events during the assembly of neural circuits? Dscam mutant phenotypes in da neurons provide insight into this problem. Mutant isoneuronal dendrites freely overlapped along their length, and the dendrites of some cells collected into tight bundles at stereotyped locations along the body wall. Mutant dendrites rarely grew beyond these specific sites of termination. Interestingly, wild-type dendrites normally projected to these same foci but provided a more diffuse coverage of the surrounding area, very likely because self-avoidance prohibited their overlap. These observations together suggest that Dscam mutant phenotypes reveal coordinates on the body wall that are attractive to dendrites and that there is an important interplay in da neurons between self-avoidance signaling and dendrite guidance mechanisms (Matthews, 2007).

One implication of these observations for circuit assembly is that self-avoidance is likely crucial for the spreading of highly branched dendritic processes that might otherwise tend to fasciculate or respond in unison to localized extrinsic guidance signals. In this way, self-avoidance might act throughout the nervous system to establish properly targeted and fully sampled territories. The analogies between Dscam mutant phenotypes in the brain and those of da neurons described in this study support this notion. Based on these findings, it is proposed that Dscam-mediated self-avoidance plays a widespread role in patterning the fly nervous system. Since Dscam diversity is not seen in vertebrate neurons, it is speculated that analogous mechanisms might exist in which stochastic expression of other families of cell-surface recognition molecules provide the capacity for self-avoidance in the vertebrate brain (Matthews, 2007).

REFERENCES

Agarwala, K. L., et al. (2000). Down syndrome cell adhesion molecule DSCAM mediates homophilic intercellular adhesion. Brain Res. Mol. Brain Res. 79: 118-126. 10925149

Agarwala, K. L., et al. (2001a). DSCAM, a highly conserved gene in mammals, expressed in differentiating mouse brain. Biochem. Biophys. Res. Commun. 281(3): 697-705. 11237714

Agarwala, K. L., et al. (2001b). Cloning and functional characterization of DSCAML1, a novel DSCAM-like cell adhesion molecule that mediates homophilic intercellular adhesion. Biochem. Biophys. Res. Commun. 285(3): 760-72. 11453658

Agarwala, K. L., et al. (2001c). Dscam is associated with axonal and dendritic features of neuronal cells. J. Neurosci. Res. 66(3): 337-46. 11746351

Barlow, G. M., et al. (2002). Mammalian DSCAMs: roles in the development of the spinal cord, cortex, and cerebellum? Biochem. Biophys. Res. Commun. 293(3): 881-91. 12051741

Barlow, G. M., et al. (2002). DSCAM: an endogenous promoter drives expression in the developing CNS and neural crest. Biochem. Biophys. Res. Commun. 299(1): 1-6. 12435380

Celotto, A. M. and Graveley, B. R. (2001). Alternative splicing of the Drosophila Dscam pre-mrna is both temporally and spatially regulated. Genetics 159: 599-608. 11606537

Crayton, M. E., Powell, B. C., Vision, T. J. and Giddings, M. C. (2006). Tracking the evolution of alternatively spliced exons within the Dscam family. BMC Evol. Biol. 6: 16. 16483367

Fuerst, P. G., Koizumi, A., Masland, R. H. and Burgess, R. W. (2008). Neurite arborization and mosaic spacing in the mouse retina require DSCAM. Nature 451(7177): 470-4. PubMed citation: 18216855

Graveley, B. R., et al. (2004). The organization and evolution of the dipteran and hymenopteran Down syndrome cell adhesion molecule (Dscam) genes. RNA 10(10): 1499-506. 15383675

Graveley, B. R. (2005). Mutually exclusive splicing of the insect Dscam pre-mRNA directed by competing intronic RNA secondary structures. Cell 123(1): 65-73. 16213213

Hughes, M. E., et al. (2007). Homophilic Dscam interactions control complex dendrite morphogenesis. Neuron 54(3): 417-27. Medline abstract: 17481395

Hummel, T., et al. (2003). Axonal targeting of olfactory receptor neurons in Drosophila is controlled by Dscam. Neuron 37: 221-231. 12546818

Kohmura, N., Senzaki, K., Hamada, S., Kai, N., Yasuda, R., Watanabe, M., Ishii, H., Yasuda, M., Mishina, M. and Yagi, T. (1998). Diversity revealed by a novel family of cadherins expressed in neurons at a synaptic complex. Neuron 20: 1137-1151.

Kreahling, J. M. and Graveley, B. R. (2005). The iStem, a long-range RNA secondary structure element required for efficient exon inclusion in the Drosophila Dscam pre-mRNA. Mol. Cell. Biol. (23): 10251-60. 16287842

Li, W. and Guan, K. L. (2004). The Down syndrome cell adhesion molecule (DSCAM) interacts with and activates Pak. J. Biol. Chem. 279(31): 32824-31. 15169762

Matthews, B. J., Kim, M. E., Flanagan, J. J., Hattori, D., Clemens, J. C., Zipursky, S. L. and Grueber, W. B. (2007). Dendrite self-avoidance is controlled by Dscam. Cell 129(3): 593-604. Medline abstract: 17482551

Meijers, R., et al. (2007). Structural basis of Dscam isoform specificity. Nature 449: 487-491. Medline abstract: 17721508

Park, J. W., et al (2004). Identification of alternative splicing regulators by RNA interference in Drosophila. Proc. Natl. Acad. Sci. USA 101: 15974-15979. 15492211

Schmucker, D., et al. (2000). Drosophila Dscam is an axon guidance receptor exhibiting extraordinary molecular diversity. Cell 101: 671-684.

Soba, P., et al. (2007). Drosophila sensory neurons require Dscam for dendritic self-avoidance and proper dendritic field organization. Neuron 54(3): 403-16. Medline abstract: 17481394

Wang, J., et al. (2002). Drosophila Dscam is required for divergent segregation of sister branches and suppresses ectopic bifurcation of axons. Neuron 33: 559-571. 11856530

Wang, J., et al. (2004). Transmembrane/juxtamembrane domain-dependent Dscam distribution and function during mushroom body neuronal morphogenesis. Neuron 43: 663-672. 15339648

Watson, F. L., et al. (2005). Extensive diversity of Ig-superfamily proteins in the immune system of insects. Science 309(5742): 1874-8. 16109846

Wojtowicz, W. M., Flanagan, J. J., Millard, S. S., Zipursky, S. L. and Clemens, J. C. (2004). Alternative splicing of Drosophila Dscam generates axon guidance receptors that exhibit isoform-specific homophilic binding. Cell 118(5): 619-33. 15339666

Worby, C. A., et al. (2001). The sorting nexin, DSH3PX1, connects the axonal guidance receptor, Dscam, to the actin cytoskeleton. J. Biol. Chem. 276(45): 41782-9. 11546816

Wu, Q. and Maniatis, T. (1999). A striking organization of a large family of human neural cadherin-like cell adhesion genes. Cell 97: 779-790.

Yimlamai, D., Konnikova, L., Moss, L. G. and Jay, D. G. (2005). The zebrafish down syndrome cell adhesion molecule is involved in cell movement during embryogenesis. Dev. Biol. 279(1): 44-57. 15708557

Yamakawa, K., et al. (1998). DSCAM: a novel member of the immunoglobulin superfamily maps in a Down syndrome region and is involved in the development of the nervous system. Hum. Mol. Genet. 7: 227-237.

Zhan, X. L., Clemens, J. C., Neves, G., Hattori, D., Flanagan, J. J., Hummel, T., Vasconcelos, M. L., Chess, A. and Zipursky, S. L. (2004). Analysis of Dscam diversity in regulating axon guidance in Drosophila mushroom bodies. Neuron 43(5): 673-86. 15339649

Zhu, H., et al. (2006). Dendritic patterning by Dscam and synaptic partner matching in the Drosophila antennal lobe. Nature Neuroscience 9: 349-355. 16474389

date revised: 15 July 2008

Home page: The Interactive Fly © 1997 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.