roundabout

DEVELOPMENTAL BIOLOGY

Embryonic

The robo expression pattern was examined in the embryonic CNS. The in situ hybridization pattern of ROBO mRNA in Drosophila shows it to have elevated and widespread expression in the CNS. Multiple monoclonal and serum antibodies were raised against portions of Robo protein and the same staining pattern was observed with all of them. Robo is first seen in the embryo weakly expressed in lateral stripes during germband extension [Images]. At the onset of germband retraction, Robo expression is observed in the neuroectoderm. By the end of stage 12, as the growth cones first extend, Robo is seen on growth cones which project ipsilaterally, including pCC, aCC, MP1, dMP2, and vMP2. Strikingly, little or no Robo expression is observed on commissural growth cones as they extend toward and across the midline. However, as these growth cones turn to project longitudinally, their level of Robo expression dramatically increases. Robo is expressed at high levels on all longitudinally projecting growth cones and axons. In contrast, Robo is expressed at nearly undetectable levels on commissural axons. This is striking since 90% of axons in the longitudinal tracts also have axon segments crossing in one of the commissures. Thus, Robo expression is regionally restricted. Robo expression is also seen at a low level throughout the epidermis and at a higher level at muscle attachment sites. In stage 16-17 embryos, faint Robo staining can be seen in the commissures but at levels much lower than observed in the longitudinal tracts. For those axons that never cross the midline, Robo is expressed on their growth cones from the outset; for the majority of axons that do cross the midline, Robo is expressed at high levels on their growth cones only after they cross the midline (Kidd, 1998a).

Most of the neurons of the ventral nerve cord send out long projecting axons that cross the midline. In the Drosophila CNS, cells of the midline give rise to neuronal and glial lineages with different functions during the establishment of the commissural pattern. The development of midline cells is fairly well understood. In the developing ventral neural cord, 7-8 midline progenitor cells per abdominal segment generate about 26 glial and neuronal cells, i.e. 3-4 midline glial cells, 2 MP1 neurons, 6 VUM neurons, 2 UMI neurons, as well as the median neuroblast and its support cells. The VUM neurons comprise motoneurons as well as interneurons, which project through the anterior and posterior commissures. Genetic studies indicate that the VUM neurons are involved in the initial attraction of commissural growth cones. The MP1 neurons are ipsilateral projecting interneurons, which participate in the formation of specific longitudinal axon pathways. The median neuroblast divides during larval and pupal stages. Contrary to what occurs in the grasshopper CNS, the Drosophila median neuroblast does not generate midline glial cells. In Drosophila, the midline glial cells develop from a set of 2-3 progenitors located in the anterior part of each segment. A function of the midline glial cells during the maturation of the segmental commissures has been found, such that two midline glial cells migrate along cell processes of the VUM-midline neurons to separate anterior and posterior axon commissures. If this migration is blocked, a typical fused commissure phenotype develops. Toward the end of embryogenesis, midline glial cells are required for the formation of individual fascicles within the commissures (Hummel, 1999 and references).

The glial cells present repulsive signals to the Roundabout receptor in addition to a permissive contact-dependent signal helping commissural growth cones across the midline. A novel repulsive component is encoded by the karussell gene. In stage 12 karussell mutant embryos, several short FasII-positive cell processes project toward the CNS midline. In older embryos, FasII expression is found on axons crossing the midline in more than 50% of the neuromeres. The distribution of the 22C10 antigen (see Futsch) in karussell embryos uncovers only a few abnormalities. Thus a small subset of normally ipsilateral axons project contralateral in karussell mutants. The majority of axons found in the circles around the RP1 neurons are likely to be commissural axons that cross the midline more than once. This suggests that karussell encodes a novel component of the repulsive signaling pathway. karussell is shown to act in parallel to commissureless and roundabout gene functions, or it may act downstream in a common regulatory hierarchy (Hummel, 1999).

To determine which CNS midline cells present the repulsive signal recognized by Robo, double mutant embryos were analyzed. In commissureless mutant embryos no commissures are formed. The gene pointed is specifically expressed in the midline glial cells and controls differentiation of this cell type. In commissureless pointed mutant embryos few commissures are formed. Similarly in commissureless/slit mutant embryos commissures do form. Since pointed as well as slit affect differentiation of the midline glial cells it is suggested that these cells present the repulsive ligand to Robo. In the absence of differentiated midline glial cells no repulsive ligand can be present and growth cones can cross the midline. This finding implies that disruption in glial differentiation at the midline should lead to a robo-like phenotype as well (Hummel, 1999).

What is the function of midline glia in commissure formation? Genetic data suggest that, in addition to being a permissive substrate for commissural growth, the midline glial cells present the repulsive signal to axons that should not cross the midline. Indeed, many examples of FasciclinII-positive axons crossing the midline are found in mutants in which the development of the midline glial cells is affected. The same defect can be observed when differentiation of the midline glia is impaired by directed overexpression of argos, which is a negative regulator of the EGF-receptor pathway. This finding was not unexpected since the Commissureless protein, which regulates Robo expression, is found in high levels in these cells. It is proposed that one important function of the midline glial cells is to act as a control post dictating who can and cannot cross. They prevent commissural axons from crossing the midline more than once and ensure that ipsilateral projecting axons never cross the midline. These two processes might be regulated by different processes (karussell/roundabout) (Hummel, 1999).

The following model is proposed for commissure formation. The initial growth of commissural growth cones towards the midline in stage 12 embryos is guided by an attractive signal expressed by the midline neurons. Presumably, this attraction is mediated by early Netrin expression in the midline neurons or alternatively by the action of a Schizo/Weniger attractive system. At this early developmental stage the midline glial cells are elongated in shape, contacting the epidermis with their basal side and are assumed to send out cellular processes contacting the VUM-midline neurons at the dorsal side of the nervous system. The midline glial cells express a repulsive signal that is conveyed to lateral axons via the Robo receptor and/or the karussell gene product. This repulsive function restricts the first axons to cross the midline just anterior of the VUM neurons. The midline glial cells also express a contact dependent permissive guidance cue helping the axons to cross the midline. Subsequently, neuron-glia interaction at the midline results in the migration of the midline glial cells along processes of the VUM neurons (Hummel, 1999).

In situ hybridization and immunocytochemistry studies show that all three robos are expressed in the embryonic CNS during the period of axon outgrowth. robo expression begins first at embryonic stage 10. robo2 expression is first visible at stage 11 and becomes restricted to a smaller subset of neurons later in development (by stage 15). robo3 expression does not begin until late stage 13 and is limited to fewer neurons. Comparing the cells that express robo, robo2, and robo3 gives clues about the potential roles the three different Robo receptors might play during axon guidance in terms of two different events. They function both during the early establishment of midline crossing decisions and later during the establishment of lateral position (i.e., the location and choice of specific longitudinal axon pathways in the medial-lateral axis). robo2 RNA can be detected in the aCC and pCC neurons at early stage 13. The expression level of robo2 in these cells increases throughout stage 13. robo2 is transiently expressed in a variety of other pioneer neurons in the CNS, including MP1, dMP2, and vMP2. All of these growth cones normally project ipsilaterally without crossing the midline. The four axons from pCC, vMP2, MP1, and dMP2 initially selectively fasciculate as they extend in a pairwise fashion and transiently display a high affinity for one another; they all express high levels of Fas II. However, they subsequently selectively defasciculate during the time that pCC and vMP2 pioneer the medial Fas II pathway, while MP1 ultimately pioneers the intermediate Fas II pathway. The defasciculation of these axons and their separation to form these two distinct longitudinal pathways occurs when robo2 expression in all of these neurons declines; this is the same period when robo3 appears in a subset of these neurons (Simpson, 2000a).

robo3 is expressed later than robo2 and in a highly restricted subset of CNS neurons. robo3 is not expressed at early or midstage 13 but, by late stage 13, begins to be expressed in MP1 (which pioneers the intermediate Fas II pathway) and aCC (which is a motoneuron that exits the CNS and extends into the periphery). robo3 expression increases throughout stage 14 in both MP1 and aCC. robo3 mRNA is not detected in pCC, vMP2, or dMP2 (Simpson, 2000a).

The pCC, vMP2, MP1, and dMP2 growth cones pioneer the first two longitudinal axon pathways. All four growth cones initially extend right next to the midline but normally do not cross it. In a robo mutant, all four growth cones cross and recross the midline. In a slit mutant, all four growth cones enter the midline and do not leave it. From the beginning of axon outgrowth, robo is expressed in all four neurons. Similarly, robo2 is transiently expressed in all four neurons by early stage 13. However, it is not until late stage 13 that robo3 is expressed at low levels in two of these four neurons. Thus, robo and robo2 are expressed early enough in these ipsilaterally projecting pioneer neurons to prevent them from entering or crossing the midline, whereas robo3 is not. As robo3 expression begins, robo2 expression becomes more restricted. As development proceeds, both robo2 and robo3 expression becomes restricted to a pattern that specifies the lateral position of axons (Simpson, 2000a).

Antibody staining using monoclonal and polyclonal antisera raised (in mouse) against the three different Robos supports the mRNA expression data. Robo and Robo2 proteins appear earlier than Robo3 and, in general, appear to be expressed on many if not all of the early ipsilaterally projecting axons. Later in development, as Robo3 protein appears, the patterns of expression resolve into a restricted pattern for Robo2 and Robo3. Robo, Robo2, and Robo3 are found on the longitudinal tracts of the CNS scaffold but not in the commissural segments of contralaterally projecting axons. All three Robos are expressed on growth cones as revealed by immunoelectron microscopic analysis. Robo is present across the entire medial-lateral span of the longitudinal pathways, while Robo3 is expressed on axons in the lateral two thirds, and Robo2 is further restricted to the lateral third only of the longitudinal axon pathways. Immunocytochemistry also shows that the Robo2 protein is found in the heart, the early trachea, and the lateral body wall muscles, where it subsequently resolves to the muscle attachment sites (Simpson, 2000a).

Slit and Robo control cardiac cell polarity and morphogenesis

Basic aspects of heart morphogenesis involving migration, cell polarization, tissue alignment, and lumen formation may be conserved between Drosophila and humans, but little is known about the mechanisms that orchestrate the assembly of the heart tube in either organism. The extracellular-matrix molecule Slit and its Robo-family receptors are conserved regulators of axonal guidance. This study reports a novel role for the Drosophila slit, robo, and robo2 genes in heart morphogenesis. Slit and Robo proteins specifically accumulate at the dorsal midline between the bilateral myocardial progenitors forming a linear tube. Manipulation of Slit localization or its overexpression causes disruption in heart tube alignment and assembly, and slit-deficient hearts show disruptions in cell-polarity marker localization within the myocardium. Similar phenotypes are observed when Robo and Robo2 are manipulated. Rescue experiments suggest that Slit is secreted from the myocardial progenitors and that Robo and Robo2 act in myocardial and pericardial cells, respectively. Genetic interactions suggest a cardiac morphogenesis network involving Slit/Robo, cell-polarity proteins, and other membrane-associated proteins. It is concluded that Slit and Robo proteins contribute significantly to Drosophila heart morphogenesis by guiding heart cell alignment and adhesion and/or by inhibiting cell mixing between the bilateral compartments of heart cell progenitors and ensuring proper polarity of the myocardial epithelium (Qian, 2005).

Early embryonic events of heart formation are remarkably similar between Drosophila and vertebrates, in that two bilaterally symmetrical strips of precardiac mesoderm fuse as a linear tube at the ventral or dorsal midline in both systems. Although there is much interest in understanding the basis of heart-tube assembly, little is known about the underlying molecular-genetic mechanisms that orchestrate this and other morphogenetic processes. Drosophila Slit, an EGF- and LRR-containing secreted protein, is expressed in the heart, and thus may participate in heart morphogenesis. Slit functions as a repulsive ligand for the Roundabout (Robo) family of receptors in the CNS and acts both attractively and repulsively in trachea and somatic muscles. In vertebrates, there are three slit and three robo genes. Among them, Slit3 is expressed prominently in the developing atrial walls of the murine heart. A Slit3 gene-trap mouse exhibits abnormal heart formation, including an apparent enlargement of the right ventricle. Whether or not this heart defect is secondary to other embryonic defects is not known, nor is the genetic or cellular mechanism underlying this phenotype. It is also not known which of the Robo receptors and other Slit proteins play a role in heart development (Qian, 2005).

To assess the role of Slit in Drosophila heart, slit null-mutant embryos (slit²) were analyzed by labeling the heart with antibodies against Dmef2, a muscle-specific transcription factor expressed in all myocardial and other muscle cells. When the bilateral rows of myocardial progenitors have reached the dorsal midline, they fail to align properly in slit mutants compared to wild-type. A similar phenotype is observed in robo,robo2 double-mutant embryos. In contrast, only subtle alignment defects are found in robo or robo2 single mutants. Unlike robo or robo2,robo3 mutants in combination with robo or robo2 do not cause additional heart defects, and thus robo3 is unlikely involved in cardiac development. Similar alignment phenotypes were observed with nmr^H15lacZ reporter, a marker for myocardial nuclei, in slit mutants. Although the dorsal migration of the myocardial progenitors does not seem to be affected, their highly regular arrangement is already perturbed before they reach the midline, as manifested in gaps and double rows. Visualization of the pericardial cells with Zfh-1 shows that their alignment with the myocardial cells is also perturbed in slit-robo mutants. At stage 16, two types of phenotypes can be distinguished: Type I consists of irregular cell arrangements, and type II, in addition, has large gaps. These two types of phenotypes are found in roughly equal proportion in slit and robo,robo2 double mutants. These defects are unlikely caused by abnormalities in cardiac lineage specification or in ectodermal epithelium formation during dorsal closure (Qian, 2005).

Given the cardiac abnormalities of slit and robo mutants, the expression pattern of slit and its receptors in the developing heart were examined. Slit protein is first detected in the heart at stage 14, uniformly distributed within the myocardial cytoplasm. As the bilateral rows of cardiac progenitors align at the dorsal midline, Slit accumulates at the contact sites between them. Like Slit, Robo initially displays a similarly uniform cortical localization within myocardial cells. Once they reach the midline, Robo enriches strongly at the dorsal (apical) surface of the cell. In contrast, Robo2 is present in pericardial cells located ventrally to the myocardial cells. Unlike Slit and Robo, Robo2 does not accumulate at the midline but remains in pericardial cells. In robo mutants, however, robo2 is ectopically expressed in myocardial cells and enriches at the dorsal midline, similar to Robo in wild-type embryos. Thus, robo2 apparently compensates for a myocardial loss-of-robo function, and this compensation is consistent with their redundant requirement in cardiac morphogenesis (Qian, 2005).

Although Slit and Robo are indeed expressed in the heart, indirect effects cannot be ruled out because they function in multiple tissues. To address whether slit/robo acts autonomously within the heart, tissue- and cell-type-specific rescue experiments were performed. slit and robo expression within myocardial cells is sufficient to rescue the slit and robo,robo2 phenotype, respectively, in promoting normal heart morphogenesis (Qian, 2005).

Because slit and robo are expressed at the cardiac midline and are required for heart cell alignment, it was asked if local mislocalization of these proteins also causes cardiac morphogenesis defects. Myocardial-specific (tinCΔ4-driver) or pan-mesodermal (twi24B-driver) overexpression of slit does not produce significant cardiac alignment defects or only with low penetrance, suggesting that augmenting Slit levels in myocardial cells hardly perturbs cardiac cell alignment. Mesodermal robo overexpression, however, results in frequent alignment defects, as does ectopic expression of slit in pericardial cells. Interestingly, in those embryos that exhibit virtually normal cardiac alignment, Slit accumulates continuously at the cardiac midline. In contrast, the embryos with significant abnormalities also mispattern Slit. Precise midline accumulation of Slit thus seems to be critical to correctly align and assemble the heart tube (Qian, 2005).

Because similar cardiac misalignment defects occur in robo,robo2 as in slit mutants, it was asked whether Slit accumulation is affected in robo,robo2 embryos. Indeed, without robo and robo2, Slit no longer concentrates evenly at the contact point between the myocardial cells. Thus, loss of Robo receptors compromises Slit accumulation at the dorsal midline. When robo2 is misexpressed in myocardial cells by using tinCΔ4-Gal4, a premature midline accumulation of Slit is observed, and upon contact of the bilateral cardiac rows, Slit no longer concentrates at the cardiac midline. It may be also that misexpressed Robo2 receptors trap Slit in the cytoplasm and prevent its proper secretion. When Robo or Robo2 is expressed throughout the mesoderm, the Slit pattern is also severely disrupted, and the heart tube is frequently misaligned. Because pan-mesodermal expression of slit is of little consequence, it may be that the localization of Robo is crucial for Slit accumulation at the midline. However, slit mutants do not exhibit correct Robo patterning either, thus implying that slit is necessary but not sufficient (or instructive) for Robo localization (Qian, 2005).

Previous reports suggest a role of cardiac cell-polarity acquisition in heart morphogenesis. Failure to correctly polarize the cardiac epithelium may result in misalignments that are independent of the earlier specification and differentiation events. To study the polarity of the cardiac epithelium in slit mutants, Dlg was examined. Dlg localizes to the baso-lateral sides of myocardial epithelium before contact of the bilateral rows is established, and to the apical-lateral sides after contact. Unlike in the dorsal ectoderm, cardiac Dlg localization is severely compromised in slit mutants as the bilateral heart primordia come in contact. Because a polarity phenotype is manifest only upon heart-tube assembly, slit does not appear to be required for guiding the cardiac epithelium to the dorsal midline or for initiating its polarity before contact, but rather for correctly switching its polarity from basal-lateral to apical-lateral. Examination of myocardial polarity of slit mutants with two other basal-lateral to apical-lateral makers, α-Spectrin and Armadillo, shows defects similar to those observed with Dlg. In addition, the transmembrane protein Toll, which is present on the apical-lateral surface of myocardial cells during, but not before, the cardiac alignment process, was examined. As with Dlg, α-Spectrin, and Armadillo, Toll protein is no longer restricted to the apical-lateral sides of the myocardial cells in slit mutants. Toll mislocalization can be rescued by expressing a slit transgene in the hearts of slit mutants. The disruption in apical-lateral patterning of all cell-polarity makers examined suggests an important function of slit in polarity acquisition and maintenance. Consistent with this conclusion is the accumulation of Slit and Robo at the dorsal myocardial midline, which potentially mediates the switch in myocardial cell polarity as a prerequisite for heart-tube formation (Qian, 2005).

In contrast to the apical-lateral localization of the previous markers, Dystroglycan (Dg) is heavily enriched at both apical and basal sides of myocardial membrane, but is excluded laterally. Interestingly, in slit mutant hearts, polarized Dg localization does not seem to be significantly altered despite the severe cardiac morphogenetic defects. This is in contrast to Tbx20 neuromancer (nmr) mutants, in which myocardial polarity is also disrupted, including Dg localization (Qian, 2005).

It was anticipated that there are numerous molecules involved in generating or maintaining cardiac cell polarity in conjunction with slit/robo during heart morphogenesis, but mutants of some key factors may be early lethal or have pleiotropic effects. Thus, genetic interactions between cell-polarity genes and slit were examined in relation to cardiac morphogenesis. For this purpose, various transheterozygous combinations between were made between slit and polarity genes that are expressed in the heart, including dg, dlg, and shortgun (shg), encoding E-cadherin, and mutants previously shown to have cardiac defects. As single heterozygotes, they do not have detectable heart abnormalities, but removal of one copy of slit and dg, shg, or dlg results in defective cardiac morphogenesis. In contrast, crumbs(crb) does not interact with slit in the heart, which is consistent with the lack of (polarized) Crb localization in the cardiac epithelium. Taken together, these observations suggest that slit and cell-polarity genes cooperate in aligning the myocardium. Slit/Robo localization is also perturbed in nmr mutants, suggesting that Tbx20-mediated transcriptional activities also influence Slit/Robo localization in the heart (Qian, 2005).

Slit is well known as a repellent signal that emanates from the CNS midline and patterns axon tracks, muscles, and tracheal branches. Slit can also act as an attractant, but in all cases seems to be secreted from another cell type from its receptors. In contrast, during Drosophila heart morphogenesis, both Slit and Robo originate from the same cells, i.e., from the cardiomyocytes as they align at the dorsal midline. During this apparently autocrine process, Slit ligands and Robo receptors relocalize from the myocardial circumference to accumulate between the bilateral cell rows, mediating aligned adhesion between these rows. It is presently unknown how Slit and Robo relocalize to the apical side of the heart, but this process is likely to require the function of cell-polarity genes, such as dlg and others, that genetically interact with slit and are repolarized themselves. It may also be that a Slit molecule can bind Robo receptors on both sides of the midline, perhaps in a cooperative manner, which would then lead to a progressive accumulation of both receptors and ligands at the midline and thus to a precise alignment of the bilateral rows. This is reminiscent of the attractive Robo-Slit interaction during muscle patterning: Robo is made in the muscles of adjacent segments and accumulates at the Slit-secreting muscle-attachment sites between the segments. Regardless of the difference in cellular origin, Slit may bind Robo receptors across the segment boundary, just Slit may interact with Robo proteins across the midline between the myocardial rows. Such a Robo-Slit-mediated adhesion process is also consistent with the observed myocardial-epithelium repolarization, which would bring the bilateral rows of cells in close proximity. In slit mutants, morphogenetic defects not only include failed alignments but also double alignments and intercalation. Thus, mutant cardiomyocytes often reach the midline and get in close proximity with the contralateral side but then seem to intermix. Therefore, it is proposed that Robo-Slit act as heterophilic cell-adhesion molecules mediating coordinated stereotyped alignment as well as inhibiting cell mixing. In conclusion, it is proposed that Slit/Robo proteins act in concert with cell-polarity genes in guiding and maintaining myocardial (and pericardial) cell alignment, which is likely a prerequisite for later morphogenetic events, such as formation of a continuous cardiac lumen precisely at the position of Slit localization (Qian, 2005).

Pupal

Positioning sensory terminals in the olfactory lobe of Drosophila by Robo signaling

Olfactory receptor neurons and the interneurons of the olfactory lobe are organized in distinct units called glomeruli. Expression patterns and genetic analysis has been used to demonstrate that a combinatorial code of Roundabout (Robo) receptors act to position sensory terminals within the olfactory lobe. Groups of sensory neurons possess distinct blends of Robo and Robo3 and disruption of levels by loss-of-function or ectopic expression results in aberrant targeting. In wild type, most of the neurons send collateral branches to the contralateral lobe. The data suggest that guidance of axons across brain hemispheres is mediated by Slit-dependent Robo2 signaling. The location of sensory arbors at distinct positions within the lobe allows short-range interactions with projection neurons leading to formation of the glomeruli (Jhaveri, 2004).

The Drosophila olfactory lobe is composed of about 50 glomeruli located at fixed positions within the mediolateral, anteroposterior and dorsoventral axis. Sensory neurons expressing a given candidate odorant receptor target to the same glomeruli and also send projections to the contralateral lobe. Adult olfactory neurons differentiate within the first one-third of pupal life, radiate over the lobe anlage and transit across the midline. Sensory neurons invade the lobe during the next one-third of pupation and form distinct glomeruli (Jhaveri, 2004).

Antibodies against the three Robo receptors were used to examine their localization in olfactory neurons during pupal life. The patterns of Robo, Robo2 (Leak -- FlyBase) and Robo3 are rather dynamic and appear markedly different when examined early during lobe development, when compared with later after glomeruli are formed. During the first ~20 hours after puparium formation (APF), when the olfactory neurons are on the surface but have not yet invaded the lobe, Robo is expressed uniformly on all afferent axons. Robo2 is present at low levels in all neurons but is enriched in regions lateral to the commissure. A careful examination of confocal sections through a number of pupal lobes stained with anti-Robo2 suggests that immunoreactivity is lower as axons transit the midline than just prior to/after crossover. Expression of Robo2 declines in older pupae and is no longer detectable by ~40 hours APF. Axons that express high Robo3, lie at more medial positions in the outer nerve layer. The analysis of patterns of expression indicates that populations of neurons possess unique combinations of Robo, Robo2 and Robo3 that change during development (Jhaveri, 2004).

robo3 expression in the embryonic peripheral nervous system has been shown to be regulated by the proneural gene atonal (ato). In the adult olfactory system, ato specifies a subset of neurons that are the first to develop and appear to guide the rest of the axons into the lobe. In ato1/Df(3R)p13 animals, these 'pioneers' fail to form and the rest of the neurons stall at the entry to the olfactory lobe. A subset of the Ato-independent neurons express Robo3. Furthermore, only a subset of the Ato-dependent neurons visualized by Ato::GFP express Robo3. As expected, these occupy medial positions in the outer nerve layer. These data together suggest that Robo3 is not expressed in genetically defined subset of neurons in the pupal olfactory system (Jhaveri, 2004).

Sensory neurons begin to invade the lobe from about 25 hours APF and the first signs of glomerular organization become apparent by around 36 hours APF. Glomerular formation occurs sequentially and by 60 hours APF most of the glomeruli have formed. The entry of glial cell processes and concomitant increase in lobe volume, results in some re-organization of glomerular position and the adult pattern can only be discerned by about 80 hours APF. Robo and Robo3 are enriched in subsets of sensory neurons as they terminate within the lobe. Robo is detected in most axons, although at differing levels, while Robo3 is strongly enriched in terminals within a smaller number of glomeruli. A comparison of stained 60 hour APF lobes with the adult glomerular map suggests that Robo3-expressing neurons tend to preferentially target more dorsomedial locations. An estimation of Robo and Robo3 immunoreactivity in identified glomeruli supports the idea of a combinatorial code in determining sensory neuron position (Jhaveri, 2004).

Brains at different pupal ages were stained with antibodies against the secreted ligand Slit. A sheet of cells in the midline of the sub-esophageal ganglion expresses high levels of Slit. Immunoreactivity declines in later pupae (after 60 hours APF) and is absent in the adult. The midline cells do not express the glial marker Reverse Polarity (Repo). Other regions in the midbrain closely associated with groups of Repo-positive glial cells were also labeled by anti-Slit. The diffuse nature of the staining makes it difficult to ascertain whether the glia are the source of secreted Slit in the midbrain. At 20 hours APF, the boundaries of the olfactory lobes are clearly demarcated by the presence of surrounding glial cells. Slit levels within the lobe neuropil is significantly higher than that of the background. Expression can be detected from 14 hour APF and begins to decline by 60 hours APF (Jhaveri, 2004).

The MARCM method combined with ey-FLP generates large patches of homozygous tissue in the eye-antennal disc. Since flip-out occurs early, phenotypes generated in mature neurons result from a lack of gene function from the beginning of differentiation. Clones of robo2¹ and robo3¹ were generated and targeting of a small number of sensory neurons marked by the Or22a-Gal4 transgene was examined. Sensory neurons expressing Or22a normally project to glomerulus designated DM2 and cross-over to the contralateral lobe in the inter-antennal commissure (Jhaveri, 2004).

Neurons lacking Robo2 function (robo2¹ clones) fail to cross over to the contralateral lobe and terminate at the midline forming small 'glomerular-like' structures. The terminals show immunoreactivity against the synaptic marker nc82. Targeting to DM2 occurs normally although in many (13 out of 16) cases the glomeruli appear less intensely innervated by GFP-expressing neurons. Loss of Robo3 function (robo3¹ clones), however, affected targeting of axons rather dramatically. In all cases, some mutant neurons did project correctly to DM2 although a subset of axons strayed to ectopic sites. Commissure formation was unaffected. The erroneously placed terminals formed 'glomerular-like' organizations as revealed by staining with mAbnc82, but these did not correspond in shape or position to those previously identified. A large irregular shaped 'glomerulus' located ventrally in the posterior region of the lobe was most frequently observed. In about half the preparations, an additional site was observed in a dorsolateral location. Such ectopic targets were never found in control animals carrying the or22a-Gal4 (14.6) transgene (Jhaveri, 2004).

Because Robo is expressed rather generally in olfactory neurons, loss-of-function was studied by targeted misexpression of antagonists of signaling, rather than in clones. SG18.1-Gal4 expresses in a large fraction of olfactory neurons thus revealing most of the glomeruli as well as the antennal commissure. Ectopic expression of commissureless (comm) using SG18.1-Gal4 resulted in disorganization of glomerular patterning with a weak effect on the commissure. Comm has been shown to downregulate Robo, although its effect on Robo2 and Robo3 is less well understood. The phenotype of Comm ectopic expression suggests that Robo is necessary for determining sensory neuron position within the lobe. Abelson kinase (Abl) phosphorylates the CC0 and CC1 domains of Robo, thus antagonizing signaling. Ectopic expression of either Abl or a constitutively active Dcdc42^v12 completely abolishes glomerular formation. Sensory neurons expressing Dcdc42^v12 (SG18.1::Dcdc42^v12) show an attraction for the midline and terminate there forming 'glomerular-like' structures at the midline. Results from loss-of-function clones predict such a phenotype for robo2 nulls. Constitutive activation of Dcdc42 is known to affect cytosketal dynamics generally, and could phenocopy a loss-of-function of all Robo receptors (Jhaveri, 2004).

Ectopic expression demonstrates that levels and location of Robo receptor expression are important for three-dimensional patterning of sensory terminals. Robo was ectopically in sensory neurons to test whether the domains and levels of receptors are instructive in positioning of sensory terminals within the lobe. SG18.1::GFP was used to drive Robo in olfactory neurons; the positions and morphology of glomeruli could be visualized by GFP. Robo is expressed endogenously in all olfactory neurons and the small increase in level caused by driving a single copy of the UAS-robo transgene did not significantly alter lobe morphology. Higher levels achieved by driving three copies of the transgene abrogated glomerular formation. Changing the nature of the Robo code by misexpressing Robo3, however, resulted in a dorsomedial shift of projections. The commissure forms normally when either Robo or Robo3 are misexpressed. Ectopic expression of Robo2, however, completely abolishes commissure formation with a less severe effect on glomerular morphology (Jhaveri, 2004).

Whether the genetic elements participating with Robo signaling in other well-studied systems also operate in the Drosophila adult olfactory system was also tested. A deficiency for the Slit region was crossed into an SG18.1 UAS-GFP UAS-robo2 recombinant. In this situation, where endogenous levels of the ligand were reduced by 50%, commissure formation, which is disrupted by the ectopic expression of Robo2, was restored and glomerular morphology also returned to normal. Targeted down-regulation of Robo signaling by misexpression of Comm or activated Dcdc42^v12, respectively, also suppress the phenotype caused by elevated Robo2 (Jhaveri, 2004).

These data argue that sensory neuron positioning within the lobe is determined by signaling through the Robo receptors. Reduction of Slit levels suppress the effect of receptor overexpression, demonstrating that the phenotypes are mediated through endogenous ligand. In this case, alteration of the geometry of the Slit gradient by misexpression would be expected to alter terminal positioning of sensory neurons. High Slit expression was driven in glial cells around and within the lobe using loco-Gal4. Staining of the adult lobes in these animals with an antibody against the synaptic marker mAbnc82 revealed the presence of ectopic glomeruli outside the normal lobe circumference. Increasing Slit levels further using multiple copies of the transgene led to more severe effects (Jhaveri, 2004).

The model proposes that olfactory neurons traveling in the outer nerve layer possess a different combination of Robo receptors that respond to Slit by branching into the lobe and arborizing at specific positions. In order to understand this positional code, a Gal4 line was selected that would allow expression in a set of neurons projecting to identified glomeruli to be driven from early during development. lz-Gal4;UAS-GFP labels two glomeruli -- DM6 and DL3 -- during development and in the adult brain, thus providing a means to examine the location of selective sensory neurons when the combinations of Robo are altered. A change in the levels of any of the three Robo receptors, caused by misexpression using the lz-Gal4 driver, altered the positions of these identified terminals. The phenotypes showed variable expressivity; however, it was possible to categorize preferred positions for the terminals in each treatment (Jhaveri, 2004).

Elevated Robo levels shift DL3/DM6 neurons to more central locations. Robo3 misexpression shifted the positions of the arbors most frequently to a mediodorsal axis. Large irregular-shaped glomeruli were frequently observed. The changes in neuronal positions observed by Robo2 misexpression were somewhat surprising given the hypothesis that Robo2 is involved largely in commissure formation. It is suggested that high levels of Robo2 induced by lz-Gal4 could interfere with the function of endogenous receptors. Robo2 misexpression most frequently produced cases where projections were seen terminating within a single lobe (Jhaveri, 2004).

The ectopic 'glomeruli' produced by alterations in the Robo code showed a normal organization of cellular elements. In the wild type, terminal branches of sensory neurons remain at the periphery of each glomerulus. Dendritic arbors of the lobe interneurons, filled the entire glomerulus as seen by GFP driven by GH146-Gal4 or the synapse specific marker mAbnc82. Glomeruli produced by misexpression of any of the Robo receptors also showed a similar organization as evidenced by mAbnc82 staining (Jhaveri, 2004).

This expression and genetic data suggests a model for axon guidance in the olfactory lobe. Neurons arriving at the olfactory lobe in the antennal nerve express Robo, and those expressing high levels of Robo3 additionally decussate onto the medial side of the outer nerve layer. The position of an axon in the nerve layer is influenced by Slit levels, although the identity of the cells that contribute Slit still needs to be elucidated. Several regions of Slit expression have been detected in the brain, although the cells at the midline express highest levels. Robo2, which is expressed at very low levels in all sensory neurons, is elevated after the axons cross the midline thereby preventing re-crossing. Later during pupation, sensory axons branch into the lobe and terminate at distinctive positions regulated by their unique Robo code in response to Slit levels. This allows short-range interactions with the dendritic arbors of projection neurons leading to formation of glomeruli (Jhaveri, 2004).

Compartmentalization of visual centers in the Drosophila brain requires Slit and Robo proteins

Brain morphogenesis depends on the maintenance of boundaries between populations of non-intermingling cells. Molecular markers have been used to characterize a boundary within the optic lobe of the Drosophila brain; Slit and the Robo family of receptors, well-known regulators of axon guidance and neuronal migration, were found to inhibit the mixing of adjacent cell populations in the developing optic lobe. The data suggest that Slit is needed in the lamina to prevent inappropriate invasion of Robo-expressing neurons from the lobula cortex. Slit protein surrounds lamina glia, while the distal cell neurons in the lobula cortex express all three Drosophila Robos. The function of these proteins in the visual system was examined by isolating a novel allele of slit that preferentially disrupts visual system expression of Slit and by creating transgenic RNA interference flies to inhibit the function of each Drosophila Robo in a tissue-specific fashion. Loss of Slit or simultaneous knockdown of Robo, Robo2 and Robo3 causes distal cell neurons to invade the lamina, resulting in cell mixing across the lamina/lobula cortex boundary. This boundary disruption appears to lead to alterations in patterns of axon navigation in the visual system. It is proposed that Slit and Robo-family proteins act to maintain the distinct cellular composition of the lamina and the lobula cortex (Tayler, 2004).

The optic lobes are comprised of four processing centers derived from two distinct populations of precursor cells. In several regions of the optic lobe, cells derived from these different sets of progenitors lie immediately adjacent to one another but do not intermingle. This type of organization is found at the interface of the lamina and the lobula cortex, which are derived from the outer and inner optic anlagen, respectively. Distal cell neurons form the anterior edge of the lobula cortex and are located immediately adjacent to the posterior face of the lamina. Distal cell neurons are closely appositioned to glia at the posterior edge of the developing lamina. This study examines the mechanisms that prevent the distal cell neurons of the lobula cortex from intermingling with the lamina glia (Tayler, 2004).

A novel role has been identified for Slit and the Robo receptors as key factors that prevent mixing between adjacent groups of cells in the fly brain. The secreted protein Slit surrounds the lamina glia on one side of the boundary while Robo family proteins (receptors for Slit) are expressed by the distal cell neurons on the other side of the boundary. Loss of Slit expression or tissue-specific inhibition of Robo family expression in distal cell neurons causes the intermingling of lamina glia and distal cell neurons. It is proposed that Slit protein in the lamina keeps Robo-expressing neurons within the normal confines of the lobula cortex, establishing the sharp boundary between these two regions. Given the conservation of Slit and Robo signaling in axon guidance throughout evolution, Slit and Robo family members may also regulate boundary formation in the brains of other animals. Interestingly, humans with mutations in ROBO3 exhibit defects in hindbrain morphology, although the underlying developmental defect in humans is not known (Tayler, 2004).

RNAi knockdown of Robo family protein expression in the optic lobe using the Sca-Gal4 driver causes robust defects in distal cell neuron positioning. In addition to driving gene expression in the inner proliferation center neuroblasts and distal cell neurons, Sca-Gal4 also drives expression in R8 photoreceptor axons and neuroblasts of the outer proliferation center and neurons of the medulla cortex. Inhibition of Robo family expression only in the photoreceptors caused no detectable defects. In addition, knockdown of all three Robo family proteins in the medulla cortex using apterous-Gal4 had no effect on distal cell neuron behavior, and no defects in medulla neuron movement or axon targeting were identified in either slit mutants or Robo family knockdowns. Taken together with Robo family protein expression data, the Robo family knockdown analysis strongly supports a requirement for Robo family receptors in distal cell neurons in preventing them from invading the lamina neuropil (Tayler, 2004).

In the Drosophila visual system, Slit protein is present in a continuous zone from the base of the lamina into the underlying medulla neuropil. Although Slit mRNA is detected within the optic lobe, and Slit:lacZ expression is detected in medulla glia at the base of the lamina and in medulla cortex neurons, the optic lobe does not appear highly sensitive to the precise source or concentration of Slit. Attempts to use mosaic analysis to further define the cells in which slit function was required were unsuccessful, since no phenotypes were observed, despite the generation of large marked patches of slit² mutant tissue in the visual system and the use of the Minute technique to maximize mutant clone size. It is suspected that the diffusibility of Slit protein combined with the large number of Slit-expressing cells in the optic lobe permitted the remaining heterozygous and wild-type cells in the mosaic animals to provide sufficient Slit to support proper optic lobe development. In addition, expression of Slit in photoreceptors under the control of GMR-Gal4 rescued the photoreceptor projection phenotype of slit mutants as effectively as more general expression of Slit in the optic lobe using Omb-Gal4. Thus, delivery of Slit to these neuropil regions may be sufficient to restore the boundary between the lobula cortex and the lamina (Tayler, 2004).

The effects of overexpression and ectopic expression of Slit and Robo proteins were examined in the optic lobe. Overexpression of Slit in the optic lobe using GMR-Gal4, Sca-Gal4, Omb-Gal4 or the more ubiquitously expressed Tubulin-Gal4 did not generate detectable phenotypes in the optic lobe. The failure to generate strong overexpression phenotypes could reflect the increased Slit expression within the lamina that accompanied overexpression in other regions using these Gal4 drivers. However, overexpression of Robo2 under the control of Sca-Gal4 dramatically distorted the shape of the lobula cortex, causing the distal cell neurons to move around the ventral and dorsal edges of the lamina. Since distal cell neurons normally encounter Slit protein at the posterior face of the lamina, this redistribution could reflect repulsion from regions of Slit expression. Overexpression of Robo or Robo3 caused no detectable defects (Tayler, 2004).

Effects of mutation and ectopic expression

On each side of the midline of the Drosophila CNS, axons are organized into a series of parallel pathways. The midline repellent Slit, previously identified as a short-range signal that regulates midline crossing, also functions at long range to pattern these longitudinal pathways. In this long-range function, Slit signals through the receptors Robo2 and Robo3. Axons expressing neither, one, or both of these receptors project in one of three discrete lateral zones, each successively further from the midline. Loss of robo2 or robo3 function repositions axons closer to the midline, while gain of robo2 or robo3 function shifts axons further from the midline. Local cues further refine the lateral position. Together, these long- and short-range guidance cues allow growth cones to select with precision a specific longitudinal pathway (Rajagopalan, 2000).

Forced expression of Robo2 or Robo3 repositions axons further from the midline. Increased expression of Robo does not. Clearly, the repulsive signal provided by Robo2 and Robo3 is qualitatively different from the Robo signal. What is the basis for this difference? One interesting possibility is that only Robo2 and Robo3 detect the long-range graded Slit signal, while Robo responds only to the short-range signal that regulates midline crossing. In vivo, Slit exists in a least three isoforms: a full-length 190 kDa glycoprotein, and 140 kDa N-terminal and 55 kDa C-terminal fragments produced by proteolytic cleavage of the full-length protein. At present, it is not known in which of these isoforms the various activities of Slit reside. Once this issue has been resolved, it will be interesting to test this idea by comparing the affinities of each of the Robo receptors for the different Slit isoforms (Rajagopalan, 2000).

Another possibility is that Robo2 and Robo3 transduce a qualitatively different signal from Robo by activating a different set of signal transduction pathways inside the growth cone. This is an appealing idea, since it is in their cytoplasmic domains that Robo2 and Robo3 differ most from Robo. Both Robo2 and Robo3 lack cytoplasmic motifs that are found in all other known Robo family receptors in various species, and are required in Robo for it to regulate midline crossing. In Robo signaling, these motifs are thought to mediate interactions with Ena and Abl, though it is evident that Robo must signal through other pathways as well. Receptor tyrosine phosphatases and the calmodulin and Sos-Ras pathways have also been implicated in Robo signaling, though their roles are even less clear. Too little is known about Robo signal transduction at this point to predict how the pathways activated by Robo2 and Robo3 might differ (Rajagopalan, 2000).

What forces counter the Slit gradient in the longitudinal pathways to prevent axons from simply continuing down the gradient and out to the periphery? One possibility would be a second gradient. It could be a repulsive countergradient or a parallel attractive gradient. In the vertebrate spinal cord, Slit and Semaphorin chemorepellents are expressed on both sides of the longitudinal pathways, 'squeezing' axons into a narrow corridor between the two repulsive centers. In Drosophila, there is little to suggest that such squeezing occurs. No known chemorepellent is expressed at the lateral edges of the CNS. If not by a repulsive countergradient, then might Slit instead be balanced by the parallel gradient of an attractant secreted from the midline? Netrins would be an obvious candidate for this attractant. However, the current model for guidance at the midline proposes that commissural axons lose sensitivity to Netrins and any other midline attractants as they cross. This remains to be tested in Drosophila, but if it is true, as seems likely, then the fact that most longitudinal axons have first crossed the midline would argue against the idea that Slit is balanced by a graded midline attractant (Rajagopalan, 2000).

A second graded signal to balance the Slit gradient therefore seems unlikely. In contrast, there is strong evidence that repulsion by Slit is balanced by local interactions within the longitudinal tract. This is revealed by the behavior of the Ap axons when they are forced to misexpress Robo2 or Robo3. As a result, they move down the Slit gradient, but not uniformly, and not out of the CNS. Instead, they appear to latch on to one of two alternative lateral pathways. This strongly suggests that local cues within the longitudinal tract provide a short-range attractive force that can overcome the long-range repulsive influence of Slit (Rajagopalan, 2000).

There is also other evidence to support the notion that local cues counter the Slit signal: this comes from the initial experiments that led to the formulation of the labeled pathways hypothesis itself. The idea of specific pathway labels was inspired largely by the behavior of a single neuron, called the G neuron, in the grasshopper embryo. This neuron extends an axon that grows across the midline and contralateral longitudinal tract until its growth cone meets a lateral fascicle known as the A/P fascicle. It then turns anteriorly along this pathway, fasciculating tightly with the P axons. What does the G growth cone do when the P axons are ablated? It continues further laterally! This behavior, a mystery when it was first observed in the 1980s, can now be readily understood as the continued extension of the G growth cone down the Slit gradient. At the same time, it provides further evidence that long-range repulsion from the midline is balanced by short-range cues provided by single fascicles within the longitudinal tract (Rajagopalan, 2000).

It is proposed that lateral pathway choices are specified by two interdependent mechanisms: a Robo code and a fasciculation code. The Robo code specifies the broad zone within which a growth cone should select a pathway, while the final choice of a pathway within that zone is specified by its fasciculation code. The two systems therefore act as the coarse and fine tuning for lateral pathway selection. With such a Robo code in place, it is necessary only to differentially label the pathways within a given zone. For this a relatively small number of surface molecules should suffice (Rajagopalan, 2000).

Two groups of axons, the Sema2b and the Ap axons provide an instructive example to illustrate how this system might work. Sema2b axons occupy a lateral position in the nerve cord and extend axons across the midline in the anterior commissure. The cell bodies of AP axons are located laterally, and these axons grow initially toward the midline before turning, without crossing, to continue anteriorly near the medial edge of the ipsilateral longitudinal tract. The Sema2b neurons have the Robo code of Robo+Robo3 and an unknown fasciculation code, and project their axons along a fascicle near the middle of the longitudinal tract. The Sema2b growth cones approach their target fascicle from the medial side, having crossed the midline and so, most likely, having lost their senstivity to the long-range attractive cues it provides. Within the medial (Robo-only) zone, they encounter a fascicle that expresses the appropriate fasciculation code. They do not select this pathway, however, because the long-range repulsive influence of Slit at this point is stronger than the short-range attractive forces provided by these fasciculation cues. Instead, they continue to migrate down the Slit gradient into the next zone, the intermediate Robo+Robo3 zone. Here they encounter another fascicle with the same fasciculation code and now, since the Slit signal has become weaker, short-range attraction exceeds long-range repulsion and they turn to follow this pathway. When robo3 function is removed, the Sema2b growth cones no longer detect the long-range repulsive Slit signal, and so they select instead the first attractive pathway they encounter (Rajagopalan, 2000).

The Ap neurons have a Robo code of Robo-only, and, as for the Sema2b neurons, their fasciculation code too is unknown. Their growth cones make a lateral approach toward their medial target fascicle. As ipsilateral axons that project toward but not across the midline, they respond to both its long-range attractive signals (most likely the Netrins) and its short-range repulsive cue (Slit). They also respond to short-range attractive cues (pathway labels), and, when forced to express Robo2 or Robo3 will also respond to long-range repulsion from the midline (Slit again). Initially, long-range attraction is the predominant force, and the Ap growth cones migrate toward the midline. En route to their medial target fascicle they encounter two alternative pathways that express the appropriate fasciculation cues. However, the short-range attraction these pathways offer is insufficient to overcome the pull of the midline. It is not until the Ap growth cones are closer to the midline, and begin to sense Slit as a short-range repellent (acting through Robo), that the midline loses its appeal and the Ap growth cones turn to follow instead the short-range attractive cues of their target fascicle. If the Ap axons are forced to express either Robo2 or Robo3, they can also sense Slit as a long-range repellent. The midline no longer beckons, and so the Ap growth cones are far more likely to take one of the alternative pathways they encounter out in the lateral or intermediate zones. Most often they choose the one in the intermediate Robo+Robo3 zone (Rajagopalan, 2000).

A delicate interplay between long-range graded cues and short-range pathway labels thus underlies the exquisite precision of lateral pathway selection in the Drosophila CNS. It would not be surprising to find similar mechanisms at work in the many other regions of invertebrate and vertebrate nervous systems in which axons are patterned into a series of parallel pathways (Rajagopalan, 2000).

Roundabout (Robo) in Drosophila is a repulsive axon guidance receptor that binds to Slit, a repellent secreted by midline glia. In robo mutants, growth cones cross and recross the midline, while, in slit mutants, growth cones enter the midline but fail to leave it. This difference suggests that Slit must have more than one receptor controlling midline guidance. In the absence of Robo, some other Slit receptor ensures that growth cones do not stay at the midline, even though they cross and recross it. The Drosophila genome is shown to encode three Robo receptors and Robo and Robo2 have distinct functions, which together control repulsive axon guidance at the midline. The robo,robo2 double mutant is largely identical to slit (Simpson, 2000a).

Mutations in robo2 were generated to determine if Robo2 has an essential function -- whether it plays a role in midline guidance, and, in particular, whether its presence drives axons to leave the midline in robo mutants. When examined with mAb BP102 against all CNS axons, the robo2 mutant looks slightly abnormal but much closer to wild-type than does the robo mutant. In the robo2 mutant, some axons ectopically cross the midline. This ectopic crossing phenotype is much weaker and less penetrant than in the robo mutant. In the robo2 mutant there is disorganization of the longitudnal tracts. At stage 16, Fas II is normally expressed on four major longitudinal axon pathways, of which three are clearly visible in a single optical focal plane and are diagnostic for lateral positioning. One of the Fas II pathways (the pCC pathway) is medial, another is intermediate (the MP1 pathway), and a third is lateral (this one is the last to form). A fourth Fas II pathway is more ventral directly below the medial Fas II pathway (Simpson, 2000a).

The disorganization of the Fas II pathways appears as 'braiding,' since, instead of maintaining their parallel alignment (i.e., medial, intermediate, and lateral), the three diagnostic Fas II bundles on each side of the CNS now cross over and intermittently join with each other on their own side. Segments that show misrouting of axons between bundles on the same side of the midline are more common than those that show axons crossing the midline. The frequency of aberrations is higher in the excision/deficiency embryos as compared to the excision/excision embryos, but this may be due to the fact that the deficiency removes a number of genes in addition to robo2 -- notably robo3. Heterozygosity for one robo can enhance the null phenotype of another; robo2 dominantly enhances a robo mutation. Thus, it is plausible that the increase in robo2 defects in the excision/deficiency combination is due to heterozygosity for robo3 rather than to any additional reduction in Robo2. The robo2 phenotype can also be visualized using anti-Connectin mAb. Connectin is a cell adhesion molecule that is expressed in the CNS by a subset of axons that fasciculate in two longitudinal axon pathways, one medial and the other intermediate to lateral. Some of these axons cross in the anterior commissure, where they also express Connectin. In robo2 mutants, the two Connectin pathways are often fused together into a single group of axons. The Fas II and Connectin staining patterns suggest that the loss of function of robo2 affects the ability of these axons to locate their correct lateral position and to form their correct pattern of longitudinal axon pathways. robo mutants, however, still show two Connectin pathways, but axons in the medial of the two Connectin pathways appear to ectopically cross the midline (just as the medial Fas II axons abnormally cross the midline) (Simpson, 2000a).

The ectopic crossing of axons in robo2 mutants indicates that Robo2 does indeed contribute to midline guidance as well as to lateral position. To determine if Robo2 supplies the repulsive force that drives axons to leave the midline in robo mutants, robo,robo2 double mutants were generated by recombination. The robo, robo2 double mutants were examined with mAbs 1D4 and BP102 and found to be phenotypically identical to slit. All axons are initially attracted to the midline (presumably guided in part by Netrins). But once these axons enter the midline, they are unable to leave. In a robo mutant alone, the axons leave the midline but recross it. In the double mutant, they never leave the midline, just as in a slit mutant. Thus, Robo and Robo2 together can account for all of the function of Slit in midline guidance. In the absence of Robo, it is the small amount of Robo2 on the growth cones that drives them to leave the midline, even though they can cross and recross the midline (Simpson, 2000a).

The relative contribution of Robo and Robo2 to prevention of crossing can be clarified by examining their ability to dominantly enhance each other (i.e., the phenotype generated by removing 100% of one protein is enhanced by removing 50% of the other protein). Heterozygosity for robo in a robo2 null background (robo^+/- robo2^-/-) increases the midline disruption. These embryos show a dramatic increase in ectopic midline crossing as compared to robo2 mutants alone, and the crossing involves all three of the Fas II longitudinal pathways (not just the medial Fas II pathway, as seen in robo mutants alone). Thus, one copy of robo (presumably producing 50% of protein) is not sufficient to prevent crossing, but it is sufficient to prevent axons from lingering at the midline in the absence of robo2 (Simpson, 2000a).

Heterozygosity for robo2 in a robo null background (robo^-/-robo2^+/-) leads to a different enhancement in the midline phenotype. Just as in a robo mutant, so too in a robo^-/-robo2^+/- mutant; it is only the axons in the medial Fas II pathway that ectopically enter and cross the midline. However, this subset of axons usually does not leave the midline, and, instead, the two medial Fas II pathways fuse and run along the midline. (In a slit mutant -- or robo,robo2 double homozygous mutant -- all three Fas II pathways are fused along the midline.) Thus, whereas one copy of robo (in the absence of robo2) is sufficient to prevent axons from staying at the midline, one copy of robo2 (in the absence of robo) is not (Simpson, 2000a).

Robo and Robo2 also cooperate in other developmental processes. Slit, Robo, and Robo2 function during mesoderm migration. After gastrulation in Drosophila, many myoblasts migrate laterally away from the ventral midline. In slit mutant embryos, some mesoderm cells do not migrate away from the midline and, instead, form muscles abnormally near the midline that often stretch across the midline. A weak version of this phenotype is observed in the robo mutant, suggesting that it alone cannot control mesoderm migration away from Slit. A similarly weak phenotype is observed in the robo2 mutant. However, a strong phenotype is observed in the robo,robo2 double mutant. This phenotype is very similar to the slit phenotype; many mesodermal cells do not migrate away from the midline, and, instead, some developing muscles are found ectopically crossing the midline. Thus, Robo and Robo2 appear to cooperate in controlling mesoderm migrations away from the midline. Robo and Robo2 also appear to cooperate in governing proper cell migrations and alignment of cardioblasts in the embryonic heart and in the further development of muscle, including the identification of proper insertion sites (Simpson, 2000a).

Overexpression of robo2 demonstrates that Robo2 can act as a repulsive axon guidance receptor. Moreover, it reveals an important difference between Robo and Robo2. The UAS-GAL4 system was used to drive robo2 expression in all neurons in the embryonic CNS. An expression series of increasing levels of Robo2 was generated. A characteristic phenotypic series was observed based on increasing levels of Robo2 that is different from what is seen with Robo. At the high end of expression levels, both genes generate a commissureless-like phenotype in which no axons cross the midline. However, increasing Robo expression leads to a simple phenotypic series of increasing severity of the commissureless phenotype. Interestingly, something quite different is observed with Robo2. A low level of Robo2 overexpression results in inappropriate midline crossing reminiscent of a partial robo loss-of-function phenotype and, with increasing levels of Robo2, of a complete loss of function of robo. As levels of Robo2 continue to increase, the response becomes biphasic. The proclivity to cross the midline (and thus mimic the robo loss of function) is replaced at higher levels of Robo2 by an increasing tendency to avoid the midline (and thus mimic the robo gain of function). This biphasic phenotypic series with increasing levels of Robo2 is different from what is observed with Robo and suggests two opposing functions with different thresholds. In one case, moderate levels of Robo2 appear to be able to interfere with midline repulsion. One interpretation is that Robo2 disrupts Robo signaling, either by competing for Slit binding or by decreasing Robo's output strength. Robo2 is found to be capable of heterodimerizing with Robo (as well as both receptors being capable of homodimerizing). If the heterodimer has a weaker repulsive output than a Robo homodimer, then this could explain the decrease in midline repulsion at low increased levels of Robo2 (Simpson, 2000a).

However, Robo2 does not just interfere with midline repulsion; it can also mediate it. Higher levels of ectopic Robo2 lead to the opposite phenotype in which axons fail to cross the midline. Evidently, Robo2 does have a repulsive output, just not as strong as that of Robo. Sufficient levels of Robo2 are capable of generating a complete commissureless phenotype. Thus, at low levels, Robo2 decreases the strength of Robo signaling and permits inappropriate midline crossing, while, at higher levels, Robo2 is capable of mediating sufficient repulsive signaling to prevent midline crossing entirely (Simpson, 2000a).

The commissureless phenotype observed at the higher levels of Robo2 overexpression can be partially genetically suppressed by heterozygosity (i.e., removing one copy) of robo, slit, or enabled. Although the number of commissures that form in these backgrounds is increased, the phenotype is more complex than simple suppression because in many cases the crossovers that now occur are inappropriate. Adding a robo dominant-negative transgene (truncated just after the transmembrane domain) changes the phenotype at all levels of Robo2. The Robo dominant negative (roboDN) increases the ectopic crossing seen at low levels of Robo2 overexpression, and it causes ectopic crossing at higher levels of Robo2 overexpression as well. It is unclear whether this is suppression by interference with Robo2 repulsion directly or, alternatively, whether it results from cumulative loss of repulsion by reducing the efficacy of the Robo pathway. However, increasing levels of RoboDN in a wild-type background only look like a robo loss of function, no matter how much RoboDN is added, and not like a robo,robo2 double mutant or slit mutant. This suggests that the RoboDN affects Robo output and not Robo2 output, making the second alternative above seem more likely (Simpson, 2000a).

Ectopic expression of low levels of Robo2 by all neurons causes ectopic crossing of axons reminiscent of a robo mutant. A possible explanation is that small amounts of Robo2 can interfere with repulsion by Robo. Perhaps Robo2, which lacks some of the conserved motifs found in the Robo cytoplasmic domain, has a less robust repulsive output than Robo. Extra Robo2 could interfere with Robo by dimerizing with it and creating a weaker receptor. Alternatively, Robo2 might interfere by competing for Slit binding or by sequestering downstream signaling components needed by Robo. In vitro analysis shows that the cytoplasmic domains of Robo2 and Robo can bind to one another (and homodimerize), suggesting that the interference might be direct (Simpson, 2000a).

The in vitro translated cytoplasmic domains of Robo and Robo2 can bind to GST-fusion proteins containing the cytoplasmic domain of Robo or Robo2. The homodimeric interactions are favored over the heterodimer by ~4-fold. The binding of Robo to Robo2 and of Robo to itself is not altered in GST-Robo fusion proteins individually lacking conserved motif CC1, 2, or 3, nor in one lacking the 67 amino acids closest to the transmembrane domain. Further experiments to determine which cytoplasmic domains are sufficient and necessary for in vitro Robo and Robo2 dimerization are in progress.

Although Robo and Robo2 can interact in vitro, it is not known if they heterodimerize in vivo. They are coexpressed in certain cells and thus have the opportunity to function cooperatively, but they can clearly function independently, presumably as homodimers. Robo can maintain a relatively normal CNS scaffold in the absence of Robo2. Robo2 can prevent the medial and lateral pathways from crossing the midline and all axons from lingering at the midline, in the absence of Robo. Although heterodimers have not yet been detected in vivo due to problems with coimmunoprecipitation sensitivity in whole-embryo preparations, the genetic results described above (i.e., the biphasic phenotypic series with increasing levels of Robo2) are consistent with this possibility (Simpson, 2000a).

Commissureless protein can downregulate Robo2 as well as Robo. comm overexpression in midline glia and early neurons using Scabrous-GAL4 can reduce the level of Robo2 protein in CNS axons just as it reduces the levels of Robo. In comm gain-of-function embryos, the phenotype is robo like, but there is more disorganization of the outer (i.e., intermediate and lateral) pathways, presumably because Comm is downregulating Robo2 as well as Robo. In comm null mutants, Robo2 is still localized to the lateral pathways of the CNS scaffold (and Robo3 to the intermediate and lateral pathways), indicating that Comm is not required for the lateral restriction of Robo2 and Robo3. This restriction of Robo2 and Robo3 to specific subsets of neurons appears to be largely transcriptional as revealed by in situ hybridization (Simpson, 2000a).

In contrast, the dramatic increase of Robo protein levels as growth cones cross the midline is, at least in part, regulated by Comm. The distinction is as follows: which neurons express any particular Robo family member (or combination of Robos) appears to be largely transcriptionally controlled, whereas when a given neuron displays on its axons any particular Robo family member (after the onset of transcription) appears to be controlled by other mechanisms, including Comm. Moreover, where a neuron expresses any particular Robo family member (i.e., the commissural versus longitudinal axon segment) also appears to be controlled by other mechanisms (Simpson, 2000a).

The comm gain of function shows that Comm can downregulate both Robo and Robo2. But does it normally regulate more than just Robo? In the original midline mutant screen paper, the robo;comm double mutant was described as looking just like robo when stained with mAb BP102 (which labels all CNS axons). If the double mutant was indeed indistinguishable from robo alone, then this would suggest that Comm normally only regulates Robo. But this is not the case; distinct differences are observed when the double (robo;comm) mutant is compared with robo alone, using mAb 1D4 to stain the three major Fas II pathways. In a robo mutant, the axons in the medial Fas II pathway cross and recross the midline, while the axons in the intermediate and lateral Fas II pathways do not cross the midline. In contrast, in a robo;comm double mutant, the intermediate Fas II pathway is also perturbed and can be seen crossing the midline. At the very least, this result shows that, in the absence of Robo, Comm still has some additional function that is revealed by removing them both together. Since this additional function affects midline guidance, it is speculated that this additional function involves its regulation of Robo2 and/or Robo3. There are several alternative ways in which one might interpret the additional phenotypes seen in the robo;comm double mutant. Distinguishing between these models requires having probes for the different subsets of Fas II axons (medial versus intermediate versus lateral); such probes are not yet available, although work is underway to generate these tools (Simpson, 2000a).

Can Comm also downregulate Robo3? It is very difficult to do the same experiment as with Robo and Robo2. Both Robo and Robo2 proteins are expressed early in both the CNS and surrounding tissues. Comm can be overexpressed early only in the CNS, and differential reduction of Robo or Robo2 protein in the CNS compared to the surrounding tissue can be assessed. However, Robo3 is neither expressed early enough nor in tissues outside the nervous system for a similar comparison. The fact that the robo,robo2;comm triple mutant looks like the robo,robo2 double mutant (in which no axons leave the midline) suggests that if loss of Comm increases the level of Robo3, it does not do so sufficiently to allow any axon to escape the midline. But Robo3 may simply be too weak on its own, even when released from putative Comm downregulation, to repel axons away from the midline. All of these results and interpretations are further complicated by the existence in the Drosophila genome of a gene encoding a second Comm-like protein. Both Comms are all capable when overexpressed of downregulating Robo and Robo2. How they function to regulate the different Robos is under investigation (Simpson, 2000a).

Comm is expressed on midline cells; Robo is expressed in a dynamic fashion on growth cones and appears to function as an axon guidance receptor. robo function is dosage-sensitive. commissureless and roundabout lead to complementary mutant phenotypes in which either too few or too many axons cross the midline. The robo;comm double-mutant phenotype is identical to robo alone, suggesting that in the absence of robo, comm is no longer required. Overexpression of comm is also dosage-sensitive and leads to a phenotype identical to robo loss-of-function. Comm controls Robo expression; increasing Comm leads to a reduction of Robo protein. The levels of Comm and Robo appear to be tightly regulated to assure that only certain growth cones cross the midline and that those growth cones that do cross never do so again (Kidd, 1998b).

A large-scale screen has been carried out for mutations that affect the development of CNS axon pathways in the Drosophila embryo. Embryos were screened from over 13,500 balanced lines and over 250 mutant lines were saved whose phenotypes suggest possible defects in growth cone guidance. Two new genes, commissureless (comm) and roundabout (robo) are described in this paper. Mutations in comm lead to an absence of nearly all CNS axon commissures, such that growth cones that normally project across the midline extend instead only on their own side. Mutations in robo lead to the opposite misrouting, such that some growth cones that normally extend only on their own side, now project across the midline. The phenotypes of these two genes suggest that they may encode components of attractive and repulsive signaling systems at the midline that either guide growth cones across the midline or keep them on their own side (Seeger, 1993).

In robo mutants, axons that normally project ipsilaterally can cross and recross the midline. Growth cones expressing Robo are believed to be repelled from the midline by the interaction of Robo and its ligand Slit, an extracellular protein expressed by the midline glia. To help understand the cellular basis for the midline repulsion mediated by Robo, time-lapse observations were taken to compare the growth cone behavior of the ipsilaterally projecting motorneuron RP2 in robo and wild-type embyros. In wild-type embryos, filopodia can project across the midline but are quickly retracted. In robo mutants, medial filopodia can remain extended for longer periods and can develop into contralateral branches. In many cases RP2 produces both ipsilateral and contralateral branches, both of which can extend into the periphery. The growth cone also exhibits longer filopodia and more extensive branching both at the midline and throughout the neuropile. Cell injections in fixed stage 13 embryos has confirmed and quantified these results for both RP2 and the interneuron pCC. The results suggest that Robo both repels growth cones at the midline and inhibits branching throughout the neuropile by promoting filopodial retraction (Murray, 1999).

The original model for Robo proposed the existence of a midline repellent ligand for which Robo was the receptor. The results presented here show that Robo's effects are not confined to the midline. In addition to preventing axons from crossing the midline, Robo also appears to inhibit filopodial extension and subsequent branch formation throughout the neuropile. Recently, in vitro binding studies and genetic analysis have identified Robo's ligand as the extracellular matrix protein Slit. As expected, in the CNS slit is only transcribed by the midline glia. The location of the Slit protein, however, is less clear. Antibodies raised against a C-terminal fragment of the protein show that Slit is found on the surface of axons throughout the neuropile. Recently it has emerged that Slit is proteolytically cleaved into two fragments, one of which, the C-terminal fragment, appears to be more readily diffusible (Brose, 1999). Thus, different antibodies may recognize different protein fragments with different expression patterns. Furthermore, evidence exists that Slit can function as a diffusible chemorepellent both in vitro (Brose, 1999) and in the Drosophila embryo (Kidd, 1999). These studies, together with the current results, suggest that Slit, or a fragment thereof, may interact with Robo in regions of the neuropile away from the midline. These results reinforce the concept that accurate growth cone guidance depends on a delicate balance of multiple attractive and repulsive cues. When a factor such as Robo is removed, the highly stereotypic trajectories of identified neurons are replaced with more variable results. Thus in the case of both RP2 and pCC, ipsilateral axons do not simply cross the midline but exhibit various trajectories and may even bifurcate into multiple axons (Murray, 1999).

The establishment of axon trajectories is ultimately determined by the integration of intracellular signaling pathways. Here, a genetic approach in Drosophila has demonstrated that both Calmodulin and Son of sevenless signaling pathways are used to regulate which axons cross the midline. A loss in either signaling pathway leads to abnormal projection of axons across the midline and these increase with roundabout or slit mutations. When both Calmodulin and Son of sevenless are disrupted, the midline crossing of axons mimics that seen in roundabout mutants, although Roundabout remains expressed on crossing axons. Calmodulin and Son of sevenless also regulate axon crossing in a commissureless mutant. These data suggest that Calmodulin and Son of sevenless signaling pathways function to interpret midline repulsive cues that prevent axons crossing the midline (Fritz, 2000).

A novel CaM inhibitor, called kinesin-antagonist (KA), has been expressed using the neurogenic enhancer element of the fushi tarazu gene (ftzng) in a subset of CNS neurons that normally do not cross the midline. KA expression decreases endogenous CaM activation of target proteins in the growth cone and this leads to specific axon guidance defects including stalls at selected choice points, failure to fasciculate properly and abnormal crossing of the midline. robo and slit mutations and KA interact synergistically to increase the number of axon bundles abnormally crossing the midline. KA also induces axon bundles to cross the midline in the absence of Comm protein. Sos-dependent crossovers are enhanced by KA or by slit mutation. KA and Sos also interact to increase the number of axon bundles crossing in a comm mutant. Thus, the data demonstrate that both CaM and Sos signaling pathways are required to prevent certain axons crossing the midline (Fritz, 2000).

Whether CaM and Sos-mediated signaling is working directly downstream of Robo or in closely associated, but parallel signaling pathways to prevent axons from crossing is difficult to ascertain from this genetic data alone. If these signaling pathways lie downstream of Robo, the data suggest that both CaM and Sos are activated upon Slit binding to Robo, and result in growth cone repulsion. Interestingly, increased levels of calcium have been implicated in growth cone retraction and growth cone collapse, two ways in which a growth cone may respond to a repulsive agent. In addition, retrograde actin flow, which leads to filopodial retraction, is stimulated by CaM activation of myosin light chain kinase. Two other CaM target proteins, cAMP adenylyl cyclase and phosphodiesterase, regulate cAMP cellular concentrations thus altering neuronal response to Netrin 1 and other guidance cues. Activation of a Sos signaling pathway can affect cytoskeletal dynamics by activating various GTPases known to regulate growth cone behavior and axon guidance. Moreover, the cytoplasmic tail of Robo, known to be essential for signaling function, has a tyrosine residue that could recruit Sos via Drk or dreadlocks (dock), another SH2-SH3 adapter protein that affects axon guidance. Alternatively, Robo may bind Enabled, a known substrate for Abelson tyrosine kinase (Abl), which has been implicated in commissure formation. If Sos binds to phosphotyrosine residues on Ena (also via an adapter protein) it could be indirectly recruited to Robo (Fritz, 2000 and references therein).

Another possibility is that a disruption in both the CaM and Sos signaling pathways indirectly causes abnormal crossovers. CaM has been identified as a player downstream of several guidance molecules. Indeed, the gaps in the longitudinal connectives observed with increasing copies of KA in a comm mutant or in KA robo mutants, which are not seen in robo mutants alone, suggest CaM may function downstream of other guidance cue receptors to allow extension through the connective. Once these signals are attenuated by expression of KA, axons may inadvertently cross the midline. However, if CaM only functions in cell adhesive mechanisms within the connectives, it is difficult to explain why axons cross the midline in comm mutants when no other axons cross and the presence of Slit is still being read by Robo (Fritz, 2000 and references therein).

Since CaM and Sos appear to interpret a midline repulsive cue, the existence of an additional midline repulsion system working in parallel to Robo represents an interesting possibility. In robo mutants, axons cross the midline but then move to the longitudinal connective, instead of collapsing at the midline as observed in slit mutants. It has been suggested that this occurs because the continued presence of Slit at the midline is detected by a second receptor system, and candidate genes include a second robo gene or karussel. As the data shows, heterozygous slit mutations interact very strongly with single copies of KA, Sos or KA Sos together, to force axons across the midline. The interaction between Sos and slit mutations, especially when compared to the lack of Sos and robo interaction, is particularly striking. It seems that if the activity of both repulsion systems is decreased due to the reduction of a common ligand (Slit), a disruption in CaM and/or Sos signaling dramatically increases midline crossing errors. Most of these results, including the synergistic effects of KA and Sos, robo and slit mutations, the robo-like phenotype of KA Sos mutants, and the enhancement of crossovers in comm mutants can be explained by a parallel decrease in both midline repulsive systems upon disruption of the CaM and Sos signaling pathways. Thus while the mechanisms by which CaM and Sos contribute to an axon guidance decision at the midline remain unclear, the data clearly indicate that CaM and Sos signaling pathways are critical to the transduction of repulsive information at the midline (Fritz, 2000 and references therein).

Axons in the bilateral CNS of Drosophila decide whether or not to cross the midline before following their specific subsequent pathways. In commissureless mutants, the RP3 and V motoneuron axons often fail to cross the midline but subsequently follow the mirror-image pathways and innervate corresponding muscle targets on the ipsilateral side. Conversely, in roundabout mutants, the RP2 and aCC motoneuron axons sometimes cross the midline abnormally but their subsequent pathways and synaptic targeting are the perfect mirror images of those seen in wild type. Furthermore, within a single segment of these mutants, bilateral pairs of motoneuron axons can make their midline decisions independent of one another. Thus, neither the particular molecular experience of the growth cones nor the decision at the midline caused by these mutations affects growth cone ability to respond normally to subsequently presented cues (Wolf, 2000).

The logic motivating these experiments is that growth cones will retain their ability to respond normally to all subsequent cues if they rely on an experience-independent preprogramming (Preprogram model). But, if their normal mode of operation is one of continual reprogramming, and the final axon pathways are a result of a specific sequence of interactive reprogramming, missing the normal cues at the midline should lead to a subsequent deviation from the normal pathways at one point or another (Reprogram model). It was reasoned that by following the axon pathways of individual neurons affected by midline mutations, these scenarios could be examined experimentally. The first set of experiments with comm mutants provided cases in which growth cones were prevented from crossing the midline. However, due to the bilateral symmetry of the molecular and cellular organizations across the midline, the RP3 and V motoneuron growth cones, when they failed to cross the midline, still found themselves surrounded by the same microenvironment that they would normally have experienced after crossing the midline. The net effect is essentially equivalent to a cellular transplant across the midline. Without experiencing the Comm protein on the midline glia, and despite the abnormal decision to not cross the midline, these growth cones are nevertheless perfectly capable of responding normally to all subsequent cues, allowing them to follow the mirror-image peripheral pathways all the way to their respective target muscles and to initiate synaptogenesis there (Wolf, 2000).

The second set of experiments with robo mutants provided complementary cases in which growth cones were made to abnormally cross the midline, presumably due to either full or partial loss of the ability of the growth cones to respond to midline repulsion signals. The Robo protein is a widely expressed neuronal growth cone receptor, and its deletion offers a means to disrupt the midline decisions of growth cones independent of comm mutations. In all cases, when they cross the midline abnormally, the RP2 and aCC motoneuron growth cones retain normal responsiveness to all subsequent cues, selecting the mirror-image pathways and muscles on the other side of the midline (Wolf, 2000).

These results clearly demonstrate that neither disrupted midline decisions nor a lack of midline signaling molecules affect the ability of the motoneuron growth cones to respond normally to cues encountered beyond the midline. It is concluded that, under the situations examined, the growth cones rely on an experience-independent preprogramming for their navigation through complex in vivo environments (Wolf, 2000).

The bipotential ganglion mother cells, or GMCs, in the Drosophila CNS asymmetrically divide to generate two distinct post-mitotic neurons. The midline repellent Slit (Sli), via its receptor Roundabout (Robo), promotes the terminal asymmetric division of GMCs. In GMC-1 of the RP2/sib lineage, Slit promotes asymmetric division by down regulating two POU proteins, Nubbin and Mitimere. The down regulation of these proteins allows the asymmetric localization of Inscuteable, leading to the asymmetric division of GMC-1. Consistent with this, over-expression of these POU genes in a late GMC-1 causes mis-localization of Insc and symmetric division of GMC-1 to generate two RP2s. Similarly, increasing the dosage of the two POU genes in sli mutant background enhances the penetrance of the RP2 lineage defects whereas reducing the dosage of the two genes reduces the penetrance of the phenotype. These results tie a cell-non-autonomous signaling pathway to the asymmetric division of precursor cells during neurogenesis (Mehta, 2001).

How is the Sli signal transmitted from outside to inside? Previous results show that one of the receptors for Sli is the transmembrane protein encoded by the robo locus. To determine if the effect of Sli signaling on GMC-1 is mediated via Robo, the expression of Robo in the GMC-1->RP2/sib and GMC-1-1a->aCC/pCC lineages was examined. Double staining of wild-type embryos with anti-Eve and anti-Robo shows that both these GMCs express Robo. Consistent with this, in robo null mutants, GMC-1 and GMC1-1a were found to divide symmetrically to generate two RP2s and two aCCs at the expense of sib and pCC. Although the penetrance of the RP2 lineage phenotype was low in robo mutants, the facts that Robo is expressed in GMC-1 and that the phenotype was observed only in robo null mutants argue that Robo at least partially transmits the Sli signal and promotes the asymmetric division of GMC-1 into RP2 and sib. Since three additional robo genes, robo2, robo3 and robo4 exist in Drosophila, the weak penetrance is likely to be due to genetic redundancy between these robo genes (Mehta, 2001).

The following picture emerges from this study. The Sli-Robo signaling down regulates the levels of Nub and Miti in late GMC-1, allowing the asymmetric localization of Insc and the asymmetric division of GMC-1. The possibility is entertained that loss of sibling cells in sli mutants would mean that some projections will be duplicated, while others are eliminated. Depending upon the extent, this might have an overall bearing on the pathfinding defects in sli mutants. Since Sli signaling is conserved in vertebrates, it is possible that this signaling may regulate generation of asymmetry during vertebrate neurogenesis as well (Mehta, 2001).

Given that Slit can bind directly to Netrin, and can also act via Robo receptors to silence Netrin attraction, might midline repulsion by unc-5 depend in any way on repulsion mediated by Slit and its Robo receptors? To test this, embryos were generated carrying both the elav-GAL4 and UAS-Unc5 transgenes, and that also were homozygous for one or more of the null alleles slit², robo¹, and robo2⁴ (Keleman, 2001).

The commissureless phenotype of pan-neural Unc5 embryos is essentially unaltered in the robo and robo2 single mutant backgrounds. The phenotype is more difficult to interpret when either slit or both robo and robo2 function is eliminated. The CNS phenotype observed in these embryos is intermediate between that of pan-neural Unc5 embryos and either slit or robo robo2 embryos. In some segments, axons are entirely collapsed at the midline as in slit or robo robo2 mutants, but in other segments axons are separated into two bundles, one on each side of the midline. This argues against a direct role for Slit in Unc5-mediated repulsion, since clearly Unc5 misexpression does have an effect in the absence of Slit. It does, however, suggest an indirect role. For example, repulsion by Netrin and Unc5 may only be effective in keeping axons away from the midline when it is added on top of the repulsive signal transduced via Slit and its Robo receptors. Another, not exclusive, possibility is that this intermediate phenotype is due to the ventral displacement of midline cells that occurs in both slit and robo robo2 mutants (Keleman, 2001).

Receptor tyrosine phosphatases regulate axon guidance across the midline of the Drosophila embryo, interacting with the Robo system

Neural receptor-linked protein tyrosine phosphatases (RPTPs) are required for guidance of motoneuron and photoreceptor growth cones in Drosophila. These phosphatases have not been implicated in growth cone responses to specific guidance cues, however, so it is unknown which aspects of axonal pathfinding are controlled by their activities. Three RPTPs, known as DLAR, DPTP69D, and DPTP99A, have been genetically characterized thus far. The isolation of mutations in the fourth neural RPTP, DPTP10D, is reported. The analysis of double mutant phenotypes shows that DPTP10D and DPTP69D are necessary for repulsion of growth cones from the midline of the embryonic central nervous system. Repulsion is thought to be triggered by binding of the secreted protein Slit, which is expressed by midline glia, to Roundabout (Robo) receptors on growth cones. Robo repulsion is downregulated by the Commissureless (Comm) protein, allowing axons to cross the midline. The Rptp mutations genetically interact with robo, slit and comm. The nature of these interactions suggests that DPTP10D and DPTP69D are positive regulators of Slit/Roundabout repulsive signaling. Elimination of all four neural RPTPs converts most noncrossing longitudinal pathways into commissures that cross the midline, indicating that tyrosine phosphorylation controls the manner in which growth cones respond to midline signals (Sun, 2000).

The fact that longitudinal axons can be changed into commissural axons by elimination of RPTP activity suggests that tyrosine phosphorylation controls the manner in which growth cones respond to midline repulsive signals. This is consistent with the observation that pharmacological inhibition of tyrosine kinase activity in grasshopper embryos causes a robo-like phenotype in which the longitudinal axon of the pCC neuron crosses the midline and circles back to the ipsilateral side. Further evidence that the effects of the inhibitor may actually be due to blockage of Robo signaling is provided by the recent observation that the Drosophila pCC axon in robo embryos has a unique branched morphology that is identical in appearance to that of the grasshopper pCC in inhibitor-treated embryos (Sun, 2000 and references therein).

The repulsive response to midline signals is encoded within the Robo cytoplasmic domain. The cytoplasmic domains of fly, nematode and mammalian Robo family proteins (Robos) contain conserved tyrosine-containing PYATT sequence motifs, suggesting that these domains could be direct targets for tyrosine kinases. Phosphorylated tyrosine motifs usually function by binding to SH2 and PTB-domain adapter proteins that mediate downstream signaling events. Robo also contains two proline-rich sequences that could interact with SH3-domain adapters. Robo2 has the tyrosine-containing motif, but lacks the proline-rich sequences (Sun, 2000).

How are Robo signaling pathways regulated by RPTPs? There is no evidence at present that the RPTPs directly alter signaling by the Robo protein. It is possible that the RPTPs and Robo feed into separate pathways that only intersect after several signaling steps. There is, however, a known mechanism for RPTP-mediated positive regulation of tyrosine kinase pathways that suggests how DPTP10D and DPTP69D could facilitate Robo signaling. During T cell receptor (TCR) signal transduction, the RPTP CD45 removes an inhibitory C-terminal phosphate group from the Src-family tyrosine kinase Lck, thereby activating it and allowing it to phosphorylate the z chain of the TCR. The phosphorylated z chain in turn binds to an SH2-domain containing tyrosine kinase (ZAP-70), which mediates downstream signaling events. CD45 is required for TCR signaling because in its absence Lck is not activated and thus cannot efficiently phosphorylate the z chain. (Interestingly, CD45 may also be involved in the termination of the TCR signaling response, since it can dephosphorylate the z chain and prevent it from binding to ZAP-70) Another mammalian receptor phosphatase, RPTPa, also dephosphorylates and activates Src-family kinases. Fibroblasts derived from RPTPa knockout mice have reduced Src and Fyn activities, suggesting that RPTPa is an in vivo regulator of Src family kinase function (Sun, 2000 and references therein).

By analogy to these pathways, DPTP10D and DPTP69D might regulate growth cone repulsion by activating Src-family tyrosine kinase(s) that phosphorylate Robos. This could explain the genetic data, since the loss of RPTP function would be expected to cause a decrease in the extent of Robo phosphorylation. One might also propose that positive regulation of repulsion by the RPTPs occurs through direct dephosphorylation of Robos, and that dephosphorylated Robos are more active in signaling. This would be unusual, however, since normally it is the phosphorylated form of a signaling motif that binds to downstream adapters. A variant of the direct interaction model proposes that Robos become phosphorylated on tyrosines after engagement of Slit, and that DPTP10D or DPTP69D are recruited into a Robo/Slit signaling complex by their interactions with the phosphotyrosine motifs. RPTPs might remain bound to these sites for a significant time period, because they often hydrolyze phosphate-tyrosine bonds quite slowly. The RPTPs could then function as adapters themselves, binding to downstream signaling proteins and recruiting them into Robo/Slit receptor complexes. Determining which, if any, of these models is correct will require biochemical or genetic identification of in vivo substrates for RPTPs (Sun, 2000).

Roundabout signalling, cell contact and trophic support confine longitudinal glia and axons in the Drosophila CNS

The mechanisms that generate and maintain the longitudinal axon pathways of the Drosophila CNS are largely unknown. The longitudinal pathways are formed by ipsilateral pioneer axons and the longitudinal glia. The longitudinal glia dictate these axonal trajectories and provide trophic support to later-projecting follower neurons. Follower interneuron axons cross the midline once and join these pathways to form the longitudinal connectives. Once on the contralateral side, longitudinal axons are repelled from recrossing the midline by the midline repulsive signal Slit and its axonal receptor Roundabout. Longitudinal glia also transiently express roundabout, which halts their ventral migration short of the midline. Once in contact with axons, glia cease to express roundabout and become dependent on neurons for their survival. Trophic support and cell-cell contact restrict glial movement and axonal trajectories. The significance of this relationship is revealed when neuron-glia interactions are disrupted by neuronal ablation or mutation in the glial cells missing gene, which eliminates glia, when axons and glia cross the midline despite continued midline repellent signaling (Kinrade, 2001).

The longitudinal pathways are pioneered by four neurons per hemisegment: pCC, MP1, dMP2 and vMP2 -- their axons never cross the midline. These ipsilateral pioneer axons form a scaffold for the later selective fasciculation of follower axons. During the formation of the first longitudinal fascicle, pioneer growth cones also express the Robo receptor, which prevents them from crossing the midline. During their pathfinding, the pioneer axons interact with a class of glial cells, the interface glia, which at the end of embryogenesis overlie the longitudinal axons. The longitudinal glia are the interface glia derived from the segmentally repeated lateral glioblasts, located at the edge of the neuroectoderm. Longitudinal glia, like the midline glia, are reminiscent of vertebrate oligodendrocytes since they originate from highly migratory and proliferative precursors and enwrap CNS axons. The longitudinal glioblasts divide and migrate ventrally until they contact the cell bodies of the pioneer neurons, where they halt at a certain distance from the midline. The first longitudinal fascicle is formed as the descending axons of dMP2 and MP1 meet the ascending axons of pCC and vMP2. Longitudinal glia migrate anteroposteriorly slightly ahead, but in close contact with, the extending pioneer growth cones and stall at choice points relevant for axon guidance and fasciculation. Following the formation of the first longitudinal fascicle, glia continue migrating, occupying choice points to instruct axonal defasciculation and refasciculation. The final pattern of pioneer axon trajectories is dictated by glia (Kinrade, 2001 and references therein).

Interface glia normally overlie the longitudinal connectives of the CNS and are not found over the midline. Mutations in robo cause interface glia to migrate over the midline. However, not all glia migrate over the midline: some remain along the longitudinal tracts, whose position is nevertheless closer to the midline than in normal embryos. This differential effect of robo mutations on glia correlates with similar differences in the axonal phenotype. For instance, only the pCC/MP2 (central) fasII fascicle, but not the outer two fascicles, is affected in robo mutants. The longitudinal glia are responsible for the formation of the three fasII fascicles. Hence, these data suggest that either there is a differential requirement for robo among the longitudinal glia, or that other members of the robo family may also play a role in these glia or that further factors, other than robo function, may determine glial positions along the longitudinal fascicles (Kinrade, 2001).

Since in robo mutants longitudinal glia migrate over the midline, it was of interest to see if robo may be expressed in glia as well as in axons. If so, robo would keep glia away from the midline in normal embryos and would render them insensitive to midline repulsion in robo mutants. Prior to axonogenesis, robo is expressed in two broad longitudinal bands at either side of the sli-expressing midline. Sli protein is also found within these bands at stages 12.2 and 12.1. By stage 13 robo expression resolves into segmental clusters, preceding the overt expression in axons (Kinrade, 2001).

robo is also expressed in one transverse stripe per hemisegment at stage 11. These lateral stripes include the tracheal pits. The longitudinal glioblasts originate just ventrally of the tracheal pits, and they divide as they migrate medially within these lateral bands of robo expression. Glia stop migrating medially as they enter the longitudinal bands of robo expression. Within these robo expression domains glia express robo themselves. Glia also express the longitudinal glia cell membrane marker Heartless. Thus longitudinal glia express robo and respond to midline-derived repulsion within a narrow time window. Expression of robo in normal embryos disappears from glia from stage 13, as glia occupy more dorsal positions over the longitudinal tracts. By stage 14, Robo is clearly present only in axons. Since glia maintain lateral positions in the longitudinal pathways throughout embryogenesis despite ceasing to express robo, further mechanisms must restrict glial movement (Kinrade, 2001).

Trophic support between neurons and glia is a means of restricting glial movement and axonal trajectories. There are reciprocal (although asymmetric) trophic interactions between neurons and glia during pathfinding. Pioneer neurons maintain the survival of longitudinal glia, and glia maintain the survival of follower neurons. When axon-glia interactions are intact, axonal CNS patterning is virtually normal in the absence of cell death, although subtle axonal defects are found, such as longitudinal growth cones projecting toward the midline. However, when neuron-glia interactions are perturbed in rpr mutant embryos, which are unable to undergo programmed cell death, the misrouting of axons across the midline is dramatically enhanced, compared either with rpr mutants alone or with perturbing neuron-glia interactions in embryos in which apoptosis can occur normally. In embryos double mutant for gcm and rpr or upon neuronal ablation in a rpr mutant background, axonal extension along the longitudinal pathways is severely affected, as visualized with fasII. Longitudinal connectives are virtually missing and fasII-positive axons cross the midline in every segment. In the case of ablated rpr mutant embryos the phenotype often resembles (albeit more severely) the robo mutant phenotype. These midline crossing axons express robo. In embryos double mutant for gcm and rpr or upon neuronal ablation in a rpr mutant background there is no robo expression along the longitudinal connectives, but instead all axons expressing robo cross the midline. These observations imply that attraction towards the midline is the default pathway taken by axons and glia despite midline repulsion in the absence of normal axon-glia interactions in the longitudinal pathways. They also mean that the pressure for survival normally keeps cells in contact with their normal neighbors, in this case in lateral positions. Taken together, these data show that in normal embryos control of cell survival and cell-cell contact are means of confining glia and axons laterally (Kinrade, 2001).

Cells in animals are programmed to die unless they receive input from their neighbors. In the nervous system, target cells provide trophic factors to extending axons, thus ensuring correct innervation. In the Drosophila CNS, trophic support between neurons and glia plays an instructive role during the formation of longitudinal pathways. Glia numbers are depleted upon neuronal ablation, and they can be rescued by blocking programmed cell death. Longitudinal glia normally undergo apoptosis at the time they first come into axonal contact. Pioneer neurons do not require longitudinal glia for survival, but they require glia for pathfinding. Thus by regulating glia survival, the pioneer neurons anchor longitudinal glia to their axons to enable their pathfinding. Subsequently, longitudinal glia maintain the survival of follower neurons, thus aiding the maintenance of the axonal fascicles in lateral positions. Altogether these data show that survival pressure is instructive in determining the positions of glia and axons during pathfinding. Evidence is provided in support of this notion. (1) In embryos lacking programmed cell death, some axons project across the midline. (2) When neuron-glia interactions are disturbed in embryos lacking programmed cell death, axons and glia dramatically cross over the midline. These are more severe misroutings than if only neuron-glia interactions are disturbed. This reveals the roles of axon-glia interactions in keeping both axons and glia laterally, and it also shows that combining lack of programmed cell death with other genotypes does not lead to additive but synergistic phenotypes. This means that cells respond differently if their survival needs are removed. Interestingly, blocking programmed cell death has been used as a means of unravelling functions of neurotrophins other than in survival. These observations, however, imply that preventing cell death does not recreate a normal although death-free environment, but instead generates a novel one in which cells are subject to different kinds of pressures. In the normal Drosophila embryo, the pressure for continued maintenance of cell contact functions to keeps axons and glia away from the default midline pathway, and along lateral positions (Kinrade, 2001).

Altered levels of Gq activity modulate axonal pathfinding in Drosophila

A majority of neurons that form the ventral nerve cord send out long axons that cross the midline through anterior or posterior commissures. A smaller fraction extend longitudinally and never cross the midline. The decision to cross the midline is governed by a balance of attractive and repulsive signals. This study has explored the role of a G-protein, Galphaq (G protein alpha49B), in altering this balance in Drosophila. Dgq was originally identified from a head cDNA library as a homolog of mammalian Galphaq. Initial functional characterization suggested that it was a visual-specific G-protein essential for Drosophila visual transduction. A splice variant of Galphaq, dgqalpha3, is expressed in early axonal growth cones, which go to form the commissures in the Drosophila embryonic CNS. Misexpression of a gain-of-function transgene of dgqalpha3 (AcGq3) leads to ectopic midline crossing. Analysis of the AcGq3 phenotype in roundabout and frazzled mutants shows that AcGq3 function is antagonistic to Robo signaling and requires Frazzled to promote ectopic midline crossing. These results show that a heterotrimeric G-protein can affect the balance of attractive versus repulsive cues in the growth cone and that it can function as a component of signaling pathways that regulate axonal pathfinding (Ratnaparkhi, 2002).

cDNA clones corresponding to the dgq gene were isolated in library screens using a fragment from the eye-specific splice variant dgqalpha1. Libraries derived from either embryo or appendage RNAs were screened and dgq-positive cDNA clones were analyzed by restriction digests and PCR. Three classes of cDNA clones were obtained. In the region of the open-reading frame, one of these classes corresponds to a splice variant transcript of the dgq gene, dgqalpha3, known to be expressed in several adult tissues. This class was isolated repeatedly from the embryo cDNA library, as judged by extensive PCR analysis. dgqalpha3-specific transcripts are present in poly(A⁺) RNA extracted from heads, appendages, male and female bodies, and embryos. Another class of cDNA clones was found only in the appendage library and appeared identical to the adult visual Galphaq splice form (dgqalpha1) (Ratnaparkhi, 2002).

The presence of the Dgqalpha3 protein in Drosophila embryos was examined by Western blot analysis of embryo extracts. The antiserum used recognizes the C-terminal end of the mammalian Gq protein. In Drosophila Gq this C-terminal sequence is conserved only in the Dgqalpha3 form. The results obtained indicate that a 39 kDa band, corresponding to the predicted size of the Dgqalpha3 protein, is present in embryos throughout development from as early as 0-8 hr (Ratnaparkhi, 2002).

Presence of dgqalpha3 RNA and protein in embryos suggests an involvement of the dgq gene in Drosophila development. The expression pattern of dgqalpha3 during embryonic development was examined by in situ hybridization with a dgqalpha3 splice variant-specific probe. Although dgqalpha3 RNA is present in earlier stages, tissue-specific expression of dgqalpha3 is first seen in the brain and ventral nerve cord at stage 13. This expression persists until late in development, where in addition, strong expression is seen in an anterior sense organ. This organ corresponds in position to the Bolwig's organ or the larval eye (Ratnaparkhi, 2002).

Expression of Dgqalpha3 during development of the embryonic nervous system was further confirmed by immunohistochemical staining of wild-type embryos with the Gq antiserum. The first indication of Dgqalpha3 expression in the CNS is at early stage 12. This is also the stage at which the pioneer neurons begin formation of axon pathways that give rise to the typical ladder-like appearance of the embryonic CNS, consisting of longitudinal tracts and anterior and posterior commissures that can be visualized with the axonal marker mAb BP102. A similar pattern of expression of anti-Gq and the axonal marker mAb BP102 at early stage 12 suggests that Dgqalpha3 is expressed in the pioneer growth cones that give rise to the commissures. At later stages of development Dgqalpha3 protein expression increases in the axonal tracts of the CNS. In addition, Dgqalpha3 expression was visible in the midgut epithelium at stages 12 (Ratnaparkhi, 2002).

Axonal guidance in the Drosophila CNS requires the interpretation of both attractive and repulsive cues, generated by cells that lie in the midline. The expression pattern of Dgqalpha3 protein suggested that it might be required in early growth cones for the interpretation of these cues. To address this possibility, it was essential to alter Galphaq signaling in a tissue and cell-specific manner. Therefore, transgenic strains were created with a dominant active form of Dgqalpha3, in which a glutamine residue at position 203 was mutated to a leucine. The mutation was made based on previous studies on dominant active forms of Galphaq from mammalian cells and Drosophila. As controls, transgenic lines carrying the wild-type form of Dgqalpha3 were created. Both activated dgqalpha3 (UAS-AcGq3) and dgqalpha3 (UAS-Gq3) cDNAs were placed under the control of the GAL4-inducible UAS promoter that would allow tissue and cell-specific expression. Initially, the C155-GAL4 line, which expresses in all postmitotic neurons, was used in order to study the effect of UAS-AcGq3 expression on axonal development. When stained with mAb BP102, the CNS of C155-GAL4;UAS-Gq3 embryos looked normal. In embryos expressing AcGq3, the pattern of the CNS appeared mildly deranged in that the commissures were thicker, and the neuropil region was broader than usual. More significant differences between the two genotypes were obvious when a monoclonal antibody against Fasciclin II (mAb 1D4) was used. At stage 13, anti-Fasciclin II (anti-Fas II) marks the pioneer axons that go to form the first longitudinal axon pathway, which by stage 16, defasciculates to form three distinct fascicles. These axons project ipsilaterally and do not cross the midline. In embryos of the genotype C155-GAL4;UAS-Gq3, this projection pattern was identical to wild-type embryos, indicating that overexpression of Dgqalpha3 has no effect on Fas II-expressing axons. However, in embryos expressing AcGq3, Fas II-positive axons appeared abnormal in all the embryos examined with variations in the extent of abnormality. One obvious phenotype observed was that of 'stalling' of Fas II-positive axons, which could be seen clearly at late stage 13. At this stage, minute outgrowths from the cell bodies and axonal tracts were also visible. From stage 15 onward, Fasciclin II-expressing axons could be seen crossing the midline. Occasionally a whirling phenotype similar to that observed in robo mutant alleles was seen (Ratnaparkhi, 2002).

From these experiments the fate of the axons that cross the midline was unclear. For this purpose a strain with the Apterous tau-ßgalactosidase (Ap-taußgal) construct was created in which single axons could be observed. Ap-taußgal marks specific Apterous-expressing neurons in each hemisegment of the embryo. Normally these axons project anteriorly on the ipsilateral side to form a distinct Apterous fascicle. In embryos of the genotype C155; UAS-AcGq3, axons from Apterous-expressing neurons no longer remain on the ipsilateral side but are now able to cross the midline. However, unlike axons that crossover in robo mutant embryos, these appear to stall after reaching and crossing the midline (Ratnaparkhi, 2002).

The phenotypes observed in embryos expressing AcGq3 suggest that Gq signaling can drive formation of the commissures and longitudinal tracts. This idea is supported by the phenotype observed in embryos homozygous for Df(2R)vg-C (which uncovers dgq). In these embryos the commissures appear thinner, and there are extensive breaks in the longitudinal tracts. These phenotypes are considerably stronger than those observed for frazzled mutants, which is also uncovered by the same deficiency, indicating that the effect of removing both Dgq and Frazzled is additive. However, these defects could be either caused by erroneous signaling within neurons so that they misinterpret existing cues, or by a non-autonomous mechanism that affects midline guidance cues. The latter would result in misplaced neurons or glia or neurons with changed identity. In Df(2R) vg-C embryos, the pattern of neurons expressing the Even-skipped (Eve) protein appear normal, indicating that the defects seen occur after neuronal patterning is complete (Ratnaparkhi, 2002).

To confirm that the phenotype seen by expression of AcGq3 in the CNS is caused by altered signaling within neurons expressing AcGq3, more restrictive GAL4 drivers were used to express UAS-AcGq3 in specific subsets of neurons of the embryonic CNS. ftz_ng-GAL4 expresses in a small subset of neurons that include mostly motor neurons and some interneurons like vMP2, pCC, dMP2, and MP1. These interneurons pioneer the longitudinal axon tracts, which stain positive for Fasciclin II. In addition, these axons never cross the midline. On expressing UAS-AcGq3 with ftz_ng-GAL4, midline crossing by Fasciclin II-positive axons could be observed. At stage 13, the pCC axon, which normally projects anteriorly on the ipsilateral side, could be seen turning toward the midline. At stage 16, aberrant midline crossing by the medial fascicle could be observed. The number of midline crossovers at this stage is less as compared with C155-GAL4, presumably because of the restricted and comparatively weak expression of the ftz_ng-GAL4 line. Similar results were obtained with eve_ng-GAL4, which expresses in aCC, pCC, and RP2 neurons. The pCC axon can be seen crossing the midline, whereas the aCC and RP2 projections look normal on expression of AcGq3. Axons from Apterous-expressing dorsal cells (dc) can also change their trajectory on expression of AcGq3. Instead of projecting toward the anterior and in an ipsilateral direction as is normal, a fraction of the axons can be seen drifting across the midline. The autonomy of AcGq3 function is further supported by the observation that neurons and glia are patterned normally in C155-GAL4/UAS-AcGq3 embryos, as judged by staining with anti-Eve and anti-Repo antibodies. Taken together these data demonstrate that specific activation of Dgqalpha3 in ipsilaterally projecting neurons causes changes in their axonal trajectories so that they are now able to project across the midline (Ratnaparkhi, 2002).

To understand how Dgqalpha3 acts to change axonal paths, possible interactions with genes known to affect midline guidance were sought. Axons that cross the midline and project along the contralateral longitudinal tract normally need to downregulate expression of Robo, which acts as a receptor for the midline repellant Slit. It is known that Robo downregulation requires Commissureless, but the precise mechanism is not understood. A possible mechanism by which AcGq3 could promote midline crossing was by downregulating Robo. To test this hypothesis, Robo expression was examined in ftz_ng-GAL4;UAS-AcGq3 embryos. Interestingly, Robo is not downregulated visibly in axons that ectopically cross the midline under the influence of AcGq3. The extent of Robo staining seen on these axons that aberrantly cross the midline is comparable with that seen on the longitudinal tracts. Thus, constitutive activation of Dgqalpha3 results in aberrant midline crossing of axons by a mechanism that is independent of Robo downregulation (Ratnaparkhi, 2002).

Another mechanism by which AcGq3 could induce midline crossing is through inhibition of the repulsive signal mediated by Robo. If this were so, then reducing levels of Robo by genetic means should enhance the phenotype of AcGq3. To test this, AcGq3 was expressed using ftz_ng-GAL4 in embryos carrying a single copy of the robo¹ mutant allele. robo¹ is a recessive mutation. However, embryos with one copy of this mutation show midline crossing at a frequency of ~10%. When UAS-AcGq3;robo¹/+;ftz_ng-GAL4 embryos were stained with mAb 1D4, a significant increase in the number of midline crossovers was observed as compared with embryos of the genotype UAS-AcGq3;+/+;ftz_ng-GAL4. This suggests that activation of Dgqalpha3 antagonizes the repulsive output through Robo resulting in excessive midline crossing. The antagonism could be mediated either through phosphorylation of Robo or signaling components that function downstream and/or in parallel with Robo (Ratnaparkhi, 2002).

Phosphorylation of a single tyrosine residue on Robo by Abelson (Abl) tyrosine kinase inhibits Robo repulsive signaling and is needed for normal midline crossing to take place. Expression of a mutant form of Robo in which this tyrosine residue (Y1040) has been replaced with a phenylalanine (in a transgenic strain referred to as UAS-roboY-F), leads to constitutive Robo signaling such that no axons cross the midline, resulting in a complete absence of commissure formation. If AcGq3 acts upstream of Robo, it was predicted that ectopic midline-crossovers, induced by expression of AcGq3, would be reduced in the presence of Robo Y-F. In fact, in embryos expressing both AcGq3 and Robo Y-F, no ectopic crossovers are seen, indicating that AcGq3 could inhibit Robo signaling by promoting Robo phosphorylation. This finding is also supportive of the fact that AcGq3 exerts its effect independent of Commissureless-mediated Robo downregulation. It is possible however, that AcGq3 acts through a parallel pathway that is no longer effective in the presence of Robo Y-F (Ratnaparkhi, 2002).

Both the spatiotemporal pattern of expression and functional analysis of dgq indicate that Gq activation in vivo promotes midline crossing. Axons that cross the midline need to down-modulate their repulsive signaling pathway(s) as well as respond positively to attractive cues. Therefore, whether changes in the levels of 'attractive' signaling such as the Netrin-Frazzled pathway affect the phenotype of AcGq3 was examined. Interestingly, AcGq3 phenotype shows a dosage-dependent interaction with Fra. Removal of a single copy of the Fra gene leads to a threefold reduction in the number of midline crossovers induced by AcGq3. A further reduction was observed on removal of both copies of the Fra gene as seen in embryos of the genotype C155-GAL4/UAS-AcGq3;fra³/fra⁴. Signaling through AcGq3 is thus sensitive to levels of Frazzled in the CNS (Ratnaparkhi, 2002).

To examine the effect, if any, of AcGq3 on the frazzled mutant phenotype, embryos of the genotype C155-GAL4/UAS-AcGq3;fra³/fra⁴ were examined with anti-connectin antibody and BP102. Anti-connectin labels a distinct axon fascicle in the longitudinal connectives, axon projections of SP1 and RP1 neurons that project through the anterior commissure, and a subset of axons that project through the posterior commissure to their contralateral targets. In embryos of the geno