cis-Regulatory Sequences and Functions

A distal 2.2 kb region, part of the the 4.8 kb region contiguous with the start site, directs slit expression to midline glia, cardioblasts, ventral epidermis and ventral pharynx, antennomaxillary complex, midgut and Malpighian tubules. The proximal 2.6 kb of the larger 4.8 kb region directs expression in stage 10; activity is restricted to the brain. A 1.0 kb fragment, including the first intron of the structural gene (but not the proximal promoter region) directs expression to cardioblasts, a subset of brain, and midline glia. Thus two promoter regions can direct transcription to midline glia; a distal upstream fragment and an enhancer located in the first intron (Wharton, 1993).

A single DNA motif has been found in the intronic enhancer, regulated by single-minded. It consists of an E-box element ACGTG, also found in a Toll neural promoter. The same element is responsible for single-minded autoregulation in the midline (Wharton, 1994).

During Drosophila embryogenesis the CNS midline cells have organizing activities that are required for proper elaboration of the axon scaffold and differentiation of neighboring neuroectodermal and mesodermal cells. CNS midline development is dependent on Single-minded, a basic-helix-loop-helix (bHLH)-PAS transcription factor. Fish-hook/Dichaete, a Sox HMG domain protein, and Drifter (Dfr), a POU domain protein, act in concert with Single-minded to control midline gene expression. single-minded, Dichaete, and drifter are all expressed in developing midline cells, and both loss- and gain-of-function assays reveal genetic interactions between these genes. The corresponding proteins bind to DNA sites present in a 1 kb midline enhancer from the slit gene and regulate the activity of this enhancer in cultured Drosophila Schneider line 2 cells. Dichaete directly associates with the PAS domain of Single-minded and the POU domain of Drifter; the three proteins can together form a ternary complex in yeast. In addition, Dichaete can form homodimers and also associates with other bHLH-PAS and POU proteins. These results indicate that midline gene regulation involves the coordinate functions of three distinct types of transcription factors. Functional interactions between members of these protein families may be important for numerous developmental and physiological processes (Ma, 2000).

To address whether the sim, Dichaete, and dfr genes might functionally interact to regulate development of the embryonic CNS midline, whether they exhibit overlapping expression in developing midline cells was examined. This was accomplished using anti-Dichaete and anti-Dfr sera, as well as a P[3.7sim-lacZ] marker that mimics sim midline expression. P[3.7sim-lacZ] embryos were immunostained using anti-ß-gal and either anti-Dichaete or anti-Dfr sera. Prominent overlapping expression was detected between Sim and Dichaete in developing CNS midline cells from stage 8 throughout the remainder of germ band extension. Overlap was also detected in a subset of prospective foregut cells. Similar overlapping expression was also detected between Sim and Dfr. Midline coexpression of Dichaete and Dfr was detected by immunostaining wild-type embryos with anti-Dichaete and anti-Dfr sera. Both genes are expressed together in the CNS midline throughout germ band extension. In germ band-retracted embryos, Dichaete exhibits overlapping expression with Sim and Dfr in the midline glia. Dichaete and Dfr are also detected together in lateral cells of the thoracic ganglia and a subset of ventral epidermal cells. These analyses indicate that sim, Dichaete, and dfr are coexpressed in developing CNS midline cells. The midline expression of these three genes also overlaps that of the slit gene, which is a downstream target of Sim (Ma, 2000).

Both loss-of-function and gain-of-function assays were used to detect genetic interactions between sim, Dichaete, and dfr. Mutants are known to show genetic interactions in CNS midline differentiation and in Slit protein expression. Potential cooperative interactions between sim, Dichaete, and dfr in regulating slit gene transcription were examined through the use of a P[1.0slit-lacZ] marker. This reporter contains a portion of a slit intron that drives lacZ expression mimicking that of the native slit gene in developing midline glia; P[1.0slit-lacZ] expression is first detected in germ band-extended stage 11 embryos and is maintained throughout the remainder of embryogenesis. Dichaete null mutant embryos exhibit a misplacement and loss of midline glia, as detected via anti-ß-gal immunostaining. P[1.0slit-lacZ] is expressed normally in stage 11 Dichaete mutant embryos, but during germ band retraction the number of midline glia becomes reduced from wild type, and many cells are located at aberrant ventral positions within the nerve cord. Similar, although less severe, defects are observed in dfr mutant embryos, where some midline glia are displaced from their normal positions. Notably, ß-gal-expressing midline glia are still detected in both Dichaete and dfr mutants, indicating that unlike Sim, Dichaete and Dfr are not absolutely required for P[1.0slit-lacZ] expression or midline glial development (Ma, 2000).

A dfr-Dichaete double mutant strain was used to examine whether Dichaete and dfr might act together to regulate midline gene expression. Embryos mutant for both Dichaete and dfr exhibit much more severe defects in P[1.0slit-lacZ] expression than either Dichaete or dfr single mutants. Although P[1.0slit-lacZ] is activated normally in stage 11 dfr-Dichaete double mutant embryos, there is a striking loss of midline P[1.0slit-lacZ] expression during germ band retraction. This synergistic effect strongly suggests that Dichaete and Dfr function together to regulate slit transcription. These functions may be mediated directly through Dichaete and Dfr binding sites present in the slit 1 kb regulatory region. Another, nonexclusive possibility is that Dichaete and Dfr might indirectly control slit transcription by regulating the expression of sim. To address this possibility P[3.7sim-lacZ] expression was examined in wild-type and dfr-Dichaete embryos. Compared with wild-type embryos, dfr-Dichaete double mutants exhibit a severe decrease in P[3.7sim-lacZ] expression, a phenotype that first becomes apparent during germ band retraction. Thus, Dichaete and dfr also influence sim expression and hence may indirectly influence the expression of a wide array of midline genes (Ma, 2000).

Because homozygous sim mutants exhibit severe CNS midline defects, it is not informative to analyze the phenotypes of Dichaete-sim or dfr-sim double mutants. Instead, potential interactions between Dichaete and sim were examined via a gain-of-function approach using the Gal4/UAS targeted gene expression system. A P[GMR-Gal4] strain that drives Gal4 expression in and behind the morphogenetic furrow in the developing eye imaginal disc was crossed to P[UAS-Dichaete] and P[UAS-sim] strains. P[GMR-Gal4]/+;P[UAS-Dichaete]/+ animals exhibit a moderate eye roughening with disruption of ommatidia organization and loss of mechanosensory bristles. In contrast, ectopic sim expression results in essentially normal eye morphology. The effects of Dichaete and sim coexpression reveal a nonadditive phenotype; there is a stronger disorganization of ommatidia and mechanosensory bristles than seen in flies expressing Dichaete or sim alone, and there is also a dramatic loss of eye pigmentation. These results indicated that ectopic expression of Dichaete and sim synergistically alters normal eye development, and supports the hypothesis that these genes can interact functionally (Ma, 2000).

Analysis of a 380 bp slit midline regulatory fragment has indicated the presence of a single CNS midline element (CME), through which Sim::Tgo heterodimers act. The CME is located within 300 bp of the distal end (farther from the promoter in the native slit gene) of this fragment. An inverted TTCAAT repeat (TTCAATTTCATTGAA) is located 20 bp proximal to the CME. This sequence resembles a (A/T)(A/T)CAAT consensus binding site for Sox proteins, although binding of Sox proteins to a TTCAAT sequence has not been reported. Because sequences present in an extended 1 kb slit DNA fragment are required for normal levels of slit expression in vivo, additional DNA sequences have been obtained. This analysis indicated that no other CMEs are present in the 1 kb slit DNA fragment. However, two perfect Dfr consensus binding sites, ATGCAAAT and CATAAAT, located within 500 bp of DNA proximal to the CME were identified. These two Dfr binding sites are separated by ~150 bp and flank a consensus Dichaete binding site, TACAAT. These data suggest that Dichaete, Sim, and Dfr may all bind to sites present in the 1 kb slit regulatory DNA fragment. To test this possibility, DNA gel mobility shift assays were performed using the Dichaete HMG domain and full-length Dfr protein on double-stranded oligonucleotide probes corresponding to sequences from the slit 1 kb fragment. The Dichaete HMG domain binds strongly to a 26 mer probe containing the TACAAT site. In contrast, Dichaete does not bind consistently to a 26 mer probe containing both TTCAAT sites, suggesting that Dichaete can distinguish between closely related DNA sequences. Dfr protein binds very strongly to a 33 mer probe that contains the ATGCAAAT site, and less strongly to a 32 mer probe containing the CATAAAT site. Dfr binds the ATGCAAAT site both as an apparent monomer and a dimer, because two distinct bands with reduced mobilities are detected. The 1 kb slit fragment thus may integrate the actions of at least three different types of regulatory proteins, represented by Sim, Dichaete, and Dfr (Ma, 2000).

The ability of Dichaete, Dfr, Sim, and Tgo to directly control slit transcription was examined using transient transcription assays in cultured Drosophila S2 cells. The P[1.0slit-lacZ] construct was used as a reporter with various combinations of plasmids that express Dichaete, Dfr, Sim, or Tgo. Dichaete modestly activates P[1.0slit-lacZ] transcription, indicating that in both yeast and fly cells, Dichaete can function as a direct transcriptional activator. Dfr results in little if any activation of P[1.0slit-lacZ], and Dfr and Dichaete together do not exhibit any increased activation over the levels observed for Dichaete alone. Neither Sim nor Tgo alone is able to activate the P[1.0slit-lacZ] reporter, because only background levels of expression are detected. Furthermore, Sim and Tgo together yield only minimal activation. These results imply that although Sim::Tgo heterodimers strongly activate expression of a P[6XCME-lacZ] reporter (>150 units) that contains six multimerized CMEs, additional factors are required to achieve high levels of reporter expression. Significantly, the combination of either Dichaete and Sim::Tgo or Dfr and Sim::Tgo both result in relatively high levels of activation. Thus, both Dichaete and Dfr strongly enhanced the ability of Sim::Tgo heterodimers to activate slit transcription. Comparable levels of activation are observed when all four proteins are expressed together. Taken together, the DNA binding and transcriptional activation assays provide additional evidence that regulation of slit expression in the midline glia requires functional interactions between Dichaete, Dfr, Sim, and Tgo (Ma, 2000).

Functional interactions between Sim, Dichaete, and Dfr may also regulate the midline expression of other genes, including sim and breathless (btl). Thus, sim has autoregulatory functions, and the combined functions of dfr and Dichaete are also required for sustained midline sim expression. In addition, a 2.8 kb interval in the P[3.7sim-lacZ] transgene used in this study contains six evolutionarily conserved CMEs as well as several consensus Dichaete and Dfr binding sites. btl encodes an FGF receptor homolog whose expression in the CNS midline and tracheal cells has been shown to depend, respectively, on Dfr as well as Sim and Tgo, or Trh and Tgo. A 200 bp btl midline/tracheal regulatory region contains three evolutionarily conserved CMEs. Inspection of this region also revealed the presence of a conserved consensus ATCAAT Dichaete binding site located in a 40 bp interval between CME2 and CME3, as well as a conserved consensus GATAAAT Dfr binding site located 40 bp downstream of CME3. Thus, functional interactions between Sim, Dichaete, and Dfr could be a general mechanism to regulate gene transcription during CNS midline development (Ma, 2000).

Enhancer loops appear stable during development and are associated with paused polymerase

Developmental enhancers initiate transcription and are fundamental to our understanding of developmental networks, evolution and disease. Despite their importance, the properties governing enhancer-promoter interactions and their dynamics during embryogenesis remain unclear. At the β-globin locus, enhancer-promoter interactions appear dynamic and cell-type specific, whereas at the HoxD locus they are stable and ubiquitous, being present in tissues where the target genes are not expressed. The extent to which preformed enhancer-promoter conformations exist at other, more typical, loci and how transcription is eventually triggered is unclear. This study generated a high-resolution map of enhancer three-dimensional contacts during Drosophila embryogenesis, covering two developmental stages and tissue contexts, at unprecedented resolution. Although local regulatory interactions are common, long-range interactions are highly prevalent within the compact Drosophila genome. Each enhancer contacts multiple enhancers, and promoters with similar expression, suggesting a role in their co-regulation. Notably, most interactions appear unchanged between tissue context and across development, arising before gene activation, and are frequently associated with paused RNA polymerase. These results indicate that the general topology governing enhancer contacts is conserved from flies to humans and suggest that transcription initiates from preformed enhancer-promoter loops through release of paused polymerase (Ghavi-Helm, 2014).

Drosophila embryogenesis proceeds very rapidly, taking 18 h from egg lay to completion. Underlying this dynamic developmental program are marked changes in transcription, which are in turn regulated by characterized changes in enhancer activity. However, the role and extent of dynamic enhancer looping during this process remains unknown. To address this, 4C-seq (chromosome conformation capture sequencing) experiments were performed, anchored on 103 distal or promoter-proximal developmental enhancers (referred to as 'viewpoints'), and absolute and differential interaction maps were constructed for each, varying two important parameters: (1) developmental time, using embryos at two different stages, early in development when cells are multipotent (3-4 h after egg lay; stages 6-7), and mid-embryogenesis during cell-fate specification (6-8 h; stages 10-11); and (2) tissue context, comparing enhancer interactions in mesodermal cells versus whole embryo. To perform cell-type-specific 4C-seq in embryos, a modified version of BiTS-ChIP (batch isolation of tissue-specific chromatin for immunoprecipitation) was established. Nuclei from covalently crosslinked transgenic embryos, expressing a nuclear-tagged protein only in mesodermal cells, were isolated by fluorescence-activated cell sorting (FACS; (>98% purity) and used for 4C-seq on 92 enhancers at 6-8 h and a subset of 14 enhancers at 3-4 h. The same 92 enhancers, and 11 additional regions, were also used as viewpoints in whole embryos at both time points. The enhancers were selected based on dynamic changes in mesodermal transcription factor occupancy between these developmental stages and the expression of the closest gene. This study was thereby primed to detect dynamic three-dimensional (3D) interactions, focusing on developmental stages during which the embryo undergoes marked morphological and transcriptional changes (Ghavi-Helm, 2014).

All 4C-seq experiments had the expected signal distribution, with high concordance between replicates. To assess data quality further, ten known enhancer-promoter pairs (of the ap, Abd-b, E2f, pdm2, Con, eya, stumps, Mef2, sli and slp1 genes) were compared, and in all cases the expected interactions were recovered. For example, using an enhancer of the apterous (ap) gene, the expected interaction was detected with the ap promoter, 17 kilobases (kb) away, illustrating the high quality and resolution of the data (Ghavi-Helm, 2014).

In chromosome conformation capture assays, interaction frequencies decrease with genomic distance between regions. To adjust for this, the 4C signal decay was modelled as a function of distance using a monotonously decreasing smooth function. Subtracting this trend, the residual interaction signal was converted to z-scores and interacting regions defined by merging neighbouring high-scoring fragments within 1 kb. Using this stringent approach, 4,247 high-confidence interactions were identified across all viewpoints and conditions, representing 1,036 unique interacting regions (Ghavi-Helm, 2014).

Each enhancer (viewpoint) interacted with, on average, ten distinct genomic regions, less than half (41%) of which were annotated enhancers or promoters. Distal enhancers had a higher than expected interaction frequency with other enhancers. Similarly, promoter-proximal elements had extensive interactions with distal active promoters, 98% of which are >10 kb away. Enhancer-promoter interactions, although not significantly enriched, involve active promoters, with high enrichment for H3K27ac and H3K4me3, and active enhancers, defined by H3K27ac, RNA Pol II and H3K79me3. These results are similar to recent findings in human cells and the mouse β-globin locus, indicating similarities in 3D regulatory principles from flies to human (Ghavi-Helm, 2014).

The extent of 3D connectivity is surprising given the relative simplicity of the Drosophila genome. On average, each promoter-proximal element interacted with four distal promoters and two annotated enhancers, whereas each distal enhancer interacted with two promoters and three other enhancers. These numbers are probably underestimates, as 60% of interactions involved intragenic or intergenic fragments containing no annotated cis-regulatory elements. Despite this, the level of connectivity is similar to that recently observed in humans, where active promoters contacted on average 4.75 enhancers and 25% of enhancers interacted with two or more promoters. The multi-component contacts that were observed for Drosophila enhancers indicate topologically complex structures and suggest that, despite its non-coding genome being an order of magnitude smaller than humans, Drosophila may require a similar 3D spatial organization to ensure functionality (Ghavi-Helm, 2014).

Insulators, and associated proteins, are thought to have a major role in shaping nuclear architecture by anchoring enhancer-promoter interactions or by acting as boundary elements between topologically associated domains (TADs). Occupancy data from 0 to 12 h Drosophila embryos revealed a 50% overlap of interacting regions with occupancy of one or more insulator protein. Insulator-bound interactions are enriched in enhancer elements, suggesting that insulators may have a role in promoting enhancer-enhancer interactions. In contrast to mammalian cells, this study observed no association between insulator occupancy and the genomic distance spanned by chromatin loops, although there was a modest increase in average interaction strength. Conversely, 50% of interacting regions are not bound by any of the six Drosophila insulator proteins, suggesting that these 3D contacts are formed in an insulator-independent manner, or are being facilitated by neighbouring interacting regions (Ghavi-Helm, 2014).

If enhancer 3D contacts are involved in transcriptional regulation, then genes linked by interactions with a common enhancer should share spatio-temporal expression. For the four loci examined-pdm2, engrailed, unc-5 and charybde-this is indeed the case. For example, the pdm2 CE8012 enhancer interacts with both the pdm2 and nubbin (nub, also known as pdm1) promoters, located 2.5 and 47 kb away, respectively. Both genes, producing structurally related proteins, are co-expressed in the ectoderm, overlapping the activity of the pdm2 enhancer. Although there are examples of long-range interactions in Drosophila, often involving Polycomb response elements (PREs) and insulator elements, the vast majority of characterized enhancers are within 10 kb of their target gene, with few known to act over 50 kb. However, as investigators historically tested regions close to the gene of interest, characterized Drosophila enhancers are generally close to the gene they regulate. In contrast, although 4C cannot assess the full extent of short-range interactions, it provides an unbiased systematic measurement of the distance of enhancer interactions, far beyond 10 kb (Ghavi-Helm, 2014).

The distance distribution of all significant interactions reveals extensive long-range interactions within the ~180 megabase (Mb) Drosophila genome; 73% span >50 kb, with the median interaction-viewpoint distance being 110 kb. Two striking examples of long-range interactions are the unc-5 and charybde loci. The unc-5 promoter interacts with multiple regions, including a weak but significant interaction with the promoter of slit (sli), at a distance of >500 kb. These genes produce structurally unrelated proteins that are co-expressed in the heart, and are essential for heart formation (Ghavi-Helm, 2014).

A promoter-proximal element near the charybde (chrb) promoter has a strong interaction with the promoter of the scylla (scyl) gene, almost 250 kb away. Both genes are closely related in sequence and co-expressed throughout embryogenesis. These long-range interactions were confirmed by reciprocal 4C, using either the promoter of chrb or scyl, or an interacting putative enhancer as viewpoint. This interaction was further verified using DNA fluorescence in situ hybridization (FISH) in embryos. As a control, the distance was assessed between the chrb promoter (probe A) and an overlapping probe A' or a region on another chromosome (probe D), to determine the distances between regions very close or far away, respectively. Comparing the distance between the chrb and scyl promoters (probes A and B) showed a high, statistically significant co-localization, in contrast to the distance between the chrb promoter and a non-interacting region with equal genomic distance (probes A and C) (Ghavi-Helm, 2014).

The reciprocal 4C revealed several intervening interactions that are consistently associated with loops to both the scyl and chrb promoter. The activity was examined of two of these in transgenic embryos. Both interacting regions can function as enhancers in vivo, recapitulating chrb expression in the visceral mesoderm and nervous system (Ghavi-Helm, 2014).

When considering a 1-Mb scale around this region, the 4C interaction signal drops to almost zero just after the promoters of both genes. This 'contained block' of interactions is reminiscent of TADs, although the boundaries don't exactly match TADs defined at late stages of embryogenesis, which may reflect differences in the developmental stages used. However, the boundaries do overlap a block of conserved microsynteny between drosophilids spanning ~50 million years of evolution, suggesting a functional explanation underlying the maintained synteny. Expanding this analysis across all viewpoints, ~60% of interactions are located within the same TAD and the same microsyntenic domain as the viewpoint. In the case of the chrb and scyl genes, this constraint may act to maintain a regulatory association between a large array of enhancers, facilitating their interaction with both genes' promoters (Ghavi-Helm, 2014).

These examples, and the other 555 unique interactions >100 kb, provide strong evidence that long-range interactions are widely used within the Drosophila genome, potentially markedly increasing the regulatory repertoire of each gene. As enhancer-promoter looping can trigger gene expression, it follows that enhancer contacts should reflect the dynamics of transcriptional changes during development and therefore be temporally associated with gene expression. To assess this, looping interactions were directly compared between the two different time points and tissue contexts. Given the non-discrete nature of chromatin contacts, the quantitative 4C-seq signal was used to identify differential interactions based on a Gamma-Poisson model, and they were defined as having >2-fold change and false discovery rate <10% (Ghavi-Helm, 2014).

Despite the marked differences in development and enhancer activity between these conditions, surprisingly few changes were found in chromatin interaction frequencies, with ~6% of interacting fragments showing significant changes between conditions. Of these, 87 interactions were significantly reduced during mid-embryogenesis (6-8 h) compared to the early time point (3-4 h), and 90 interactions significantly increased. Similarly, 105 interactions had a higher frequency in mesodermal cells, compared to the whole embryo, and For example, a promoter-proximal viewpoint in the vicinity of the Antp promoter identified many interactions, two of which are significantly decreased at 6-8 h, although the expression of the Antp gene itself increases. For one region, the reduction in 4C interaction at 6-8 h corresponds to a loss in a H3K4me3 peak from 3-4 h to 6-8 h, suggesting that this 3D contact is associated with the transient expression of an unannotated transcript. The activity of the other interacting peak was examined in transgenic embryos, and it was shown to act as an enhancer, driving specific expression in the nervous system overlapping the Antp gene at 6-8 h. Along with the two enhancers discovered at the chrb locus, this demonstrates the value of 3D interactions to identify new enhancer elements, even for well-characterized loci like Antp (Ghavi-Helm, 2014).

A viewpoint in the vicinity of the Abd-B promoter interacted with a number of regions spanning the bithorax locus, three of which correspond to previously characterized Abd-B enhancers; iab-5, iab-7 and iab-8. The iab-7 and iab-8 enhancers are active in early embryogenesis, and have much reduced or no activity at the later time point. Notably, although the loop to those two enhancers is strong at the early time point, it becomes significantly reduced later in development, when both enhancers' activities are reduced. Conversely, the iab-5 enhancer contacts the promoter at a much higher frequency later in development, at the stage when the enhancer is most active. This locus therefore exhibits dynamic 3D promoter-enhancer contacts that reflect the transient activity of three developmental enhancers. It is interesting to note that in all loci examined, the dynamic contacts of specific elements are neighboured by stable contacts, as seen in the Antp and Abd-B loci. Dynamic changes, therefore, appear to operate in the context of larger, more-stable 3D landscapes (Ghavi-Helm, 2014).

Ninety-four per cent of enhancer interactions showed no evidence of dynamic changes across time and tissue context, which is remarkable given the marked developmental transitions during these stages. To investigate this further, enhancer-promoter interactions were examined of genes switching their expression state between time points or tissue contexts. The ap gene, for example, is not expressed at 2-4 h but is highly expressed during mid-embryogenesis (6-8 h). Despite the absence of expression, the interaction between the apME680 enhancer and the ap promoter is already present at 3-4 h, several hours before the gene's activation. To examine this more globally, differentially expressed genes, going either from on-to-off or off-to-on, were selected. Even for these dynamically expressed genes, there was no correlation with changes in their promoter-enhancer contacts. Similar 'stable' interactions were observed between tissue contexts. Genes predominantly expressed in the neuroectoderm at 6-8 h, for example, have interactions at the same locations in whole embryos and purified mesodermal nuclei at 6-8 h, despite the fact that they are not expressed in the mesoderm at this stage (Ghavi-Helm, 2014).

Pre-existing loops were recently observed in human and mouse cells, and suggested to prime a locus for transcriptional activation. However, why they are formed and how transcription is eventually triggered remains unclear. To investigate this, this study focused on the subset of genes that have both off-to-on expression and no evidence for differential interactions (20 genes; differentially expressed with stable loops (DS) genes). Despite changes in their overall expression, DS genes have similar levels of RNA polymerase II (Pol II) promoter occupancy at both time points. The presence of promoter-bound Pol II in the absence of full-length transcription is indicative of Pol II pausing. Using global run-on sequencing (GRO-seq) data to define a stringent set of paused genes, it was observed that most (75%) DS genes are paused (15 of 20 DS genes), and have a significantly higher pausing index. This percentage is significantly higher than expected by chance when sampling over all off-to-on genes, and is robust to using a strict or more relaxed) definition of Pol II pausing. This association is very evident when examining specific loci, showing Pol II occupancy, short abortive transcripts, and loop formation before the gene's expression. Taken together, these results indicate that 'stable' chromatin loops are associated with the presence of paused Pol II at the promoter (Ghavi-Helm, 2014).

To understand how transcription is ultimately activated, changes were examined in DNase I hypersensitivity at the promoter of DS genes. DNase I hypersensitivity is significantly increased at interacting promoters at the stages when the gene is expressed, suggesting that the recruitment of additional transcription factor(s) later in development might act as the trigger for transcriptional activation (Ghavi-Helm, 2014).

In summary, these data reveal extensive long-range interactions in an organism with a relatively compact genome, including pairs of co-regulated genes contacting common enhancers often at distances greater than 200 kb. Comparing enhancer contacts in different contexts revealed that chromatin interactions are very similar across developmental time points and tissue contexts. Enhancers therefore do not appear to undergo long-range looping de novo at the time of gene expression, but are rather already in close proximity to the promoter they will regulate. Within this 3D topology, highly dynamic and transient contacts would not be visible when averaging over millions of nuclei. As transcription factor binding is sufficient to force loop formation, these results suggest a model where through transcription factor-enhancer occupancy, an enhancer loops towards the promoter and polymerase is recruited, but paused in the majority of cases. The subsequent recruitment of transcription factor(s) or additional enhancers at preformed 3D hubs most likely triggers activation by releasing Pol II pausing. Such preformed topologies could thereby promote rapid activation of transcription. At the same time, as paused promoters can exert enhancer-blocking activity, the presence of paused polymerase within these 3D landscapes could safeguard against premature transcriptional activation, but yet keep the system poised for activation (Ghavi-Helm, 2014).

Transcriptional Regulation

The wide-ranging defects in dendrites and axons indicate that sequoia functions to regulate axonal and dendritic morphogenesis in most neurons. Alternatively, it is conceivable that sequoia regulates the expression of genes generally required for neuronal differentiation. To gain mechanistic insight into sequoia function, the transcript profiles in wild-type and sequoia mutant embryos were compared based on microarray analyses of over 3,000 genes or ESTs, corresponding to about 25% of the Drosophila genome. The vast majority of these genes show comparable expression levels, including genes for cytoskeletal elements, genes that specify neuronal cell fates, and genes generally required for neurite outgrowth such as cdc42. Interestingly, a small fraction of the genes/ESTs analyzed showed clearly distinct expression ratios in sequoia mutants. Of these, 93 (3.1%) different transcripts were reduced by at least one-third of the wild-type level, and 34 (1.1%) different transcripts were increased by at least 75% of the wild-type level. A number of genes that appear to be regulated by sequoia, directly or indirectly, correspond to genes implicated in the control of axon morphogenesis rather than neuronal fate. These include known genes such as connectin, frazzled, roundabout 2, and longitudinals lacking, in addition to novel molecules with homology to axon guidance molecules including slit/kekkon-1 and neuropilin-2. It is noteworthy that two of the genes showing increased transcript ratios, roundabout 2 and CG1435, a novel calcium binding protein, were both also identified in a gain-of-function screen affecting motor axon guidance and synaptogenesis. In addition to genes that have clearly been implicated in axon development based on previous studies or sequence similarity, microarray data reveal that other genes potentially regulated by sequoia include peptidases, lipases, and transporters, as well as novel zinc finger proteins. It should be noted that transcripts that are broadly expressed and increased or decreased in sequoia mutants may actually be altered to a greater extent within neurons, because sequoia likely functions cell autonomously and is only expressed in the nervous system (Brenman, 2001).

The pattern and level of expression of axon guidance proteins must be choreographed with exquisite precision for the nervous system to develop its proper connectivity. Previous work has shown that the transcription factor Lola is required for central nervous system (CNS) axons of Drosophila to extend longitudinally. Lola is simultaneously required to repel these same longitudinal axons away from the midline, and it acts, in part, by augmenting the expression both of the midline repellant, Slit, and of its axonal receptor, Robo. Lola is thus the examplar of a class of axon guidance molecules that control axon patterning by coordinating the regulation of multiple, independent guidance genes, ensuring that they are co-expressed at the correct time, place and relative level (Crowner, 2002).

The reduction of Robo expression seen in lola mutants is relatively modest (~40%). It is known, however, that a 50% diminution in Robo is sufficient by itself to cause some inappropriate midline crossing, and this effect is strongly enhanced by a simultaneous 50% reduction in Slit. Loss of lola causes a greater reduction than this in Slit levels. Thus, it is plausible that the change in Slit and Robo levels could account for much of the midline phenotype observed in embryos that bear strong lola mutations. But why are weaker lola alleles like lola1A4 able to cause extra midline crossing when their effect on target gene expression is presumably proportionately less? It is likely that regulation of Slit and Robo expression is only one part of the control of midline crossing by lola, and that a significant contribution to the phenotype is made by changes in the expression of other, interacting guidance genes that are also controlled by lola. For example, aspects of the lola midline phenotype resemble details of the axon pattern observed upon mutation of genes encoding receptor tyrosine phosphatases, suggesting that these are good candidates for potential lola effectors. Moreover, it is known that the Notch-dependent mechanism that promotes the alternative (longitudinal) trajectory of CNS axons also requires lola. The multiplicity of genes contributing to the midline/longitudinal axon growth decision underscores the need for a gene, like lola, to coordinate the expression of all these cooperating guidance factors. It is suggested that it is the combination of many quantitative effects, each individually modest, which together produce the profound effects of lola on axon patterning (Crowner, 2002).

Many questions remain from these studies. (1) Though Lola itself is a transcriptional regulator, it is not known whether robo and slit are direct Lola targets or whether Lola initiates a longer chain of events leading only indirectly to robo and slit. For example, Lola could regulate other genes that themselves control the stability of robo or slit RNA or protein, or the splicing or translation efficiency of these genes. Analysis of this issue will require unambiguous identification of the exact lola isoforms required for expression of robo and slit, and characterization of their DNA-binding specificities in combination with their appropriate dimerization partner(s). (2) Only the accumulation of Robo and Slit protein has been characterized in lola mutants, and not transcript levels. The inherent variability of whole-mount RNA in situ hybridization has prevented sufficiently precise quantification of robo and slit RNA levels for this purpose. Nonetheless, the observation that ectopic expression of lola 4.7 leads to ectopic expression of slit RNA strongly argues that lola is upstream of slit transcription, though it remains possible that Robo and Slit expression are also subject to lola-dependent regulation at some post-transcriptional level (Crowner, 2002).

Midline governs axon pathfinding by coordinating expression of two major guidance systems

Formation of the neural network requires concerted action of multiple axon guidance systems. How neurons orchestrate expression of multiple guidance genes is poorly understood. This study shows that Drosophila T-box protein Midline controls expression of genes encoding components of two major guidance systems: Frazzled, ROBO, and Slit. In midline mutant, expression of all these molecules are reduced, resulting in severe axon guidance defects, whereas misexpression of Midline induces their expression. Midline is present on the promoter regions of these genes, indicating that Midline controls transcription directly. It is proposed that Midline controls axon pathfinding through coordinating the two guidance systems (Liu, 2009).

To address how Mid activates expression of the three axon guidance genes, the binding sequence of Mid was determined using an in vitro binding site selection method. Mid-binding sequence was selected from a pool of random oligonucleotides using Mid protein affinity-purified from an embryonic extract. The consensus sequence deduced from the selected oligonucleotides was (G/A/T)NA(A/T)N(T/G)(A/G)GGTCAAG. This sequence was found in the upstream regions or an intron of slit, frazzled, and robo, and all of these sites were conserved among several Drosophila species. To determine whether Mid binds to these regions in vivo, chromatin immunoprecipitation (ChIP) was performed using anti-Mid antibody. In all three genes, Mid was present around the Mid-binding sites, but not on regions without the binding site. In contrast, a potential Mid-binding site 32-kb upstream of the commisureless gene, whose expression is not affected in mid mutants, was not occupied by Mid (Liu, 2009).

The importance of the Mid-binding sites in frazzled and slit was assessed by transgenic reporter assays. To test the role of the Mid site in frazzled, reporter genes were constructed that contain the transcription start site of frazzled and an upstream region including a wild-type Mid-binding site (fraPlacZ) or a mutated site (fraMPlacZ). Compared with the wild-type reporter gene, the reporter with a mutated binding site showed reduced expression levels (33% reduction). Thus Mid-binding site is indeed required for the proper expression of frazzled. Mutating the Mid-binding site in slit also caused a severe effect on slit expression. The lacZ expression in sliPlacZ is driven by the slit regulatory element and the endogenous promoter. While sliPlacZ with the wild-type binding site recapitulated the slit expression in the midline glia and lateral cells, base substitutions in the Mid-binding site in sliMPlacZ abolished the lacZ expression. It is possible that the Mid-binding site resides in an essential promoter element of slit, and hence, the base substitutions abolished slit transcription in all cells. However, the same results were obtained using sli4.5HHlacZ and sliM4.5HhlacZ in which the slit regulatory element is fused to a heterologous hsp70 promoter. Since mid was expressed in the lateral cells but not in midline glia, these results suggest that Mid-binding sites in slit control slit transcription via binding to multiple factors: Mid in lateral cells and unknown factor(s) in midline glia. Taken together, these results demonstrate a direct role for Mid in the regulation of frazzled and slit, and suggest that Mid governs the expression of multiple axon guidance genes through directly binding of the Mid sites in their regulatory regions (Liu, 2009).

This study has shown that Mid directly controls transcription of key components of the two major axon guidance systems: the Netrin/Frazzled system and the Slit/ROBO system. Because these two systems are considered to have opposing outputs, it is interesting that the expression of both systems are induced by the same transcription factor, Mid. Dynamic expression of Frazzled and ROBO is required for growth cones to simultaneously respond to both attractants and repellents, integrate these signals, and then respond to the relative balance of forces. These molecules also provide nonautonomous functions required for cell motility, such as mediating cell adhesion and promoting axon elongation. The coordination of axon guidance systems by Mid may thus ensure cooperative actions of multiple guidance molecules in growth cone dynamics, axonal adhesion, and elongation. The role of Mid in the transcriptional regulation of axon guidance might be a conserved function, because its orthologs of human, mouse, and zebrafish Tbx20 are also expressed in motor neurons (Liu, 2009).

Mechanosensilla in the adult abdomen of Drosophila: engrailed and slit help to corral the peripheral sensory axons into segmental bundles

The abdomen of adult Drosophila bears mechanosensory bristles with axons that connect directly to the CNS, each hemisegment contributing a separate nerve bundle. In this study the amount of Engrailed protein was altered and the Hedgehog signalling pathway was manipulated in clones of cells to study their effects on nerve pathfinding within the peripheral nervous system. It was found that high levels of Engrailed make the epidermal cells inhospitable to bristle neurons; sensory axons that are too near these cells are either deflected or fail to extend properly or at all. Attempts were made to find the engrailed-dependent agent responsible for these repellent properties. slit was found to be expressed in the P compartment and, using genetic mosaics, evidence is presented that Slit is the responsible molecule. Blocking the activity of the three Robo genes (putative receptors for Slit) with RNAi supported this hypothesis. It is concluded that, during normal development, gradients of Slit protein repel axons away from compartment boundaries - in consequence, the bristles from each segment send their nerves to the CNS in separated sets (Fabre, 2010).

The peripheral sensory system of arthropods is segmented: neurons originate in sensilla in segmental groups in the epidermis and axons project from each of them to the corresponding segmental ganglion in the CNS. To achieve this, the nerves coming from each epidermal segment or compartment must not mix with nerves from neighbouring compartments. The epidermis of the fly and other arthropods is subdivided into a chain of anterior (A) and posterior (P) compartments, the P/A compartment boundary being the true segmental boundary. This segment boundary is recognised by neurons as they build the embryonic nervous system and is not crossed by peripheral sensory neurons in later stages. However, little is known of the molecular mechanisms responsible for this process (Fabre, 2010).

In the adult abdomen of Drosophila, the mechanoreceptive bristles are confined to a region of each A compartment; they develop de novo as sensory organ precursor cells that derive from the epidermal cells or `histoblasts' that proliferate during the pupal stage. Sensory organ precursor cells divide asymmetrically to generate a bristle and their associated neurons and supporting cells; the neurons then extend axons towards the CNS in an orderly manner (Fabre, 2008). These axons remain within their compartments of origin because they are oriented with respect to the body axes: within each A compartment, the more anteriorly situated bristle axons grow backwards, while the posteriorly situated bristle axons grow forwards, and thus both sets of axons meet to form a segmental nerve bundle in the middle of the A compartment (Fabre, 2010).

A and P compartments differ fundamentally: all the P cells but not the A cells, except for a6, express engrailed (en). The en gene encodes a homeodomain-containing transcription factor that induces hedgehog (hh) expression in P cells. Hh is a secreted morphogen that spreads into the A compartment, forming a U-shaped gradient that patterns cell fate and determines cell affinity. Only the epidermal cells of the A compartment produce Patched (Ptc) and Smoothened (Smo), proteins that act as receptors for Hh. Although the mechanosensory neurons are related to epidermal cells by lineage, it is not clear whether they retain all the compartmental properties of their origin (Fabre, 2010).

This study asked how En- and Hh-dependent information positions the neuronal cell bodies, affects the dendrites and influences the pathways followed by axons. To investigate this, cell identities wer altered by manipulating the relevant genes (en, hh, ptc and smo) within clones of cells, and effects on the neurons were examined. Strikingly, cells with P identity, but located within an A compartment, repel nearby neurons. This neuronal repulsion is not directly mediated by En or Hh, but indirectly by activating the expression of slit, a molecule previously implicated in neuronal pathfinding. Also, the response to Slit appears to be mediated by one or more of the Robo proteins. It is proposed that, during normal development, the secretion of Slit from P cells creates a Slit gradient in each A compartment that helps position neurons and orient axon outgrowth and thereby ensures segmental bundling of axons (Fabre, 2010).

In the wild-type fly (and perhaps therefore also in many other invertebrates), it was found that Sli is normally made in the P compartments, spreading forwards and backwards to repel neurons at the back and the front of the A compartments. As a consequence, the axons meet in the middle of the A compartments. Thus, En regulates sli expression to form a Sli gradient, the axons growing away from the source of Sli and down that gradient. Sli may also drive oriented nucleokinesis of the mechanosensory cell bodies away from the compartmental boundaries (Fabre, 2010).

Gradients of morphogens, such as Wingless (Wg), Hh and Decapentaplegic, can act at short or long range to specify cell identity and have also been implicated in axon pathfinding. Numerous studies have concluded that Hh can act as an axonal repellent or attractant, and that axons can respond directly to the gradient of Hh. Surprisingly, in the abdomen evidence is presented that Hh does not guide the mechanosensory neurons. No dependence on the Hh receptors Ptc or Smo is seen. This raises the possibility that some of the previously described effects of Hh might also be indirect. Indeed, in the zebrafish forebrain (Barresi, 2005), Hh acts to guide commissural and retinal axons indirectly by regulating sli expression (Fabre, 2010).

In vertebrates, En affects axon routing. In invertebrates, En modifies axon morphology via the expression of cell adhesion molecules such as Connectin and Neuroglian or the cell adhesion receptor Frazzled. In the cockroach cercus, En is essential for axonal pathfinding, perhaps acting directly on genes needed for guidance and synaptic recognition. There is a hypothesis that En acts directly: En protein has structural domains that could regulate nuclear export, secretion and cell-internalisation, processes also needed for axon pathfinding and target recognition. However, the current experiments in the fly abdomen point to a different conclusion. When smo- en- and ptc- en- clones were produced in the P compartment, mechanosensory axons traversed anterior cells of the P compartment and a6 cells in which en is expressed. Thus, it is unlikely that En itself repels axons in the abdomen of Drosophila. Evidence suggests instead that En drives the expression of sli autonomously, the effects of En on pathfinding being due to local gradients of Sli concentration. The behaviour of axons emanating from smo- en- and ptc- en- clones in the P compartment can be understood in this context: the clones are small and even though they do not themselves secrete Sli (because they are transformed into A cells), they nevertheless find themselves in a Sli gradient, high behind and lower in front. Axons leaving such clones behave as expected and grow down that gradient (Fabre, 2010).

The mode of action of Sli in neuronal and axonal repulsion has been studied in numerous systems. In vertebrates, a gradient of one or more of the three Slit genes can induce the arrest of growth cones, similar to that observed here with ptc- and en-expressing clones, which are ectopic sources of Sli (Fabre, 2010).

In other systems, Sli is received by one or more Robo receptors acting with Dock. In Drosophila, the three Robo genes are typically expressed in distinct but overlapping regions, but only Robo binds to Dock. This study presents evidence that Robo and the co-receptor Dock are expressed in the mechanosensory neurons and also that robo3 is expressed in the multidendritic neurons. Results with RNAi suggest that all three Robo genes are required for the normal fasciculation of the mechanosensory axons; the strongest effect was found with RNAi for robo2. Note that knockdown of any one of the Robo genes is unlikely to produce a very clear phenotype as they can partially substitute for each other (Fabre, 2010).

Flies carrying sli.lacZ suggest that sli is normally strongly expressed only at the back of the P compartments, raising the question of how its expression is controlled in the wild type. In the adult tergites, wg is expressed at the rear of each A compartment and Wg protein is thought to cross over the A/P border to form a gradient that patterns the P compartment. If so, and if a high concentration of Wg were to inhibit sli expression, then sli expression might be blocked in the anterior part of P (p3), but allowed in the posterior part of P. There are two other arguments supporting this hypothesis. First, wg is not (or is weakly) expressed in the most lateral tergite, which could explain why the band of sli expression is broader laterally and fills, or almost fills, the P compartment there. Second, wg is not expressed in the pleura, where sli.lacZ expression is ubiquitous. By contrast, in the sternites, wg is expressed and there sli.lacZ is confined to the P compartments. It could therefore be that ptc- as well as the en-expressing clones that are transformed towards P identity would not express wg themselves. Thus, when located far from the endogenous source of Wg they should escape repression and transform into p1, which is of extreme posterior P identity, and become sources of Sli, as observed (Fabre, 2010).

Sli might work with other guidance cues in the fly abdomen. In the Drosophila eye, disruption of the Sli/Robo mechanism disturbs the boundary between the lamina and the distal cell neurons. It has been suggested that the Fasciclin adhesion molecules also support the boundary: Fas3 is expressed in the region where distal cell neurons are found, and Fas2 is expressed by the photoreceptor axons that carry Hh to the lamina. It is suspected that Fas2 and Fas3 might contribute to corralling neurons inside of the A compartment by promoting axonal bundling to the APN (Fabre, 2010).

An individual axon might be pushed from behind by a chemorepellent, pulled from afar by a chemoattractant, and hemmed in by attractive and repulsive local cues. These signals constitute what Ramón y Cajal proposed to be an 'intelligent force' guiding axons. It is not easy to dissect out these various signals, this paper has documented one repulsive signal, Sli, that hems in neurons and helps bundle segmental sets of sensory neurons in an arthropod (Fabre, 2010).

Targets of Activity

Drosophila single-minded (sim) encodes a nuclear protein that plays a critical role in the development of the neurons, glia, and other nonneuronal cells that lie along the midline of the embryonic CNS. In sim mutant embryos the midline cells fail to differentiate properly into their mature CNS cell types and do not take their appropriate positions within the developing CNS. sim is required for midline expression of a group of genes including slit, Toll, rhomboid, engrailed, and a gene at 91F. The sim mutant CNS defect may be largely due to loss of midline slit expression. The snail gene is required to repress sim and other midline genes in the presumptive mesoderm (Nambu, 1990).

Protein Interactions

The availability of expressed Slit proteins enabled an examination of their interactions with Roundabout proteins. The interaction between Slit and Robo proteins was further demonstrated using soluble forms of the ectodomains of rRobo1, rRobo2, and dRobo1 fused to either the constant region (Fc) of the human immunoglobulin molecule or to alkaline phosphatase. rRobo1-Fc and rRobo2-Fc bind transfected cells expressing hSlit2, and Drosophila Robo bound cells expressing Drosophila Slit. These results indicate a high degree of specificity in the interaction between Slit and Robo proteins. In addition, in cross-species experiments it was found that the binding interactions are evolutionarily conserved. Thus, Drosophila Slit binds cells expressing either rRobo1 or rRobo2, and hSlit2 binds cells expressing dRobo1, although these interactions appeared weaker than those observed within species. Similarly, Robo1-Fc and Robo2-Fc bind cells expressing Drosophila Slit, and Drosophila Robo binds hSlit2-expressing cells. The amino-terminal LRRs of Drosophila and mammalian Slit proteins have homology to a number of ECM molecules, including the laminin-binding molecule biglycan. This prompted an examination of whether Slit proteins can also bind laminins. hSlit2, applied in conditioned medium from transfected cells, binds to a substrate with laminin 1, but not a substrate coated with fibronectin. Since Netrin proteins show homology to a portion of the laminin molecule, whether Netrin 1 and Slit2 can bind one another was examined. Netrin 1 binds to COS cells expressing hSlit2 in a pattern that was indistinguishable from binding observed with Robo1-Fc and Robo2-Fc (Brose, 1999).

Frazzled (Fra) is the DCC-like Netrin receptor in Drosophila that mediates attraction; Roundabout (Robo) is a Slit receptor that mediates repulsion. Both ligands, Netrin and Slit, are expressed at the midline; both receptors have related structures and are often expressed by the same neurons. To determine if attraction versus repulsion is a modular function encoded in the cytoplasmic domain of these receptors, chimeras were created carrying the ectodomain of one receptor and the cytoplasmic domain of the other and their function in transgenic Drosophila was tested. Fra-Robo (Fra's ectodomain and Robo's cytoplasmic domain) functions as a repulsive Netrin receptor; neurons expressing Fra-Robo avoid the Netrin-expressing midline and muscles. Robo-Fra (Robo's ectodomain and Fra's cytoplasmic domain) is an attractive Slit receptor; neurons and muscle precursors expressing Robo-Fra are attracted to the Slit-expressing midline (Bashaw, 1999).

In Drosophila, the same midline cells normally secrete both Netrins and Slit. Growth cones can simultaneously respond to both ligands in a cell-specific fashion. Some growth cones express high levels of Fra and low levels of Robo, and they extend toward and across the midline. Other growth cones appear to express high levels of both receptors, and they can extend toward the midline, but they do not cross it. Growth cones can dramatically change their levels of Robo expression; once they cross the midline, growth cones increase their level of Robo, a change that prevents them from crossing the midline again. Such complex and dynamic behavior requires growth cones to be able to simultaneously respond to both attractants and repellents and to integrate these signals and respond to the relative balance of forces. Introducing a chimeric receptor into this finely tuned system leads to dramatic phenotypes. Adding a receptor that responds to Netrin as a repellent leads to a comm-like phenotype in which too few axons cross the midline. Adding a receptor that responds to Slit as an attractant leads to the opposite robo- or slit-like phenotypes, in which too many axons cross the midline or remain at the midline, respectively. These phenotypes are dose dependent, suggesting that by adding more chimeric receptor, the relative balance can be tipped and in tis way the growth cone's response is selectively controlled. This striking dosage sensitivity raises the possibility of using these phenotypes as the basis for genetic suppressor screens to identify signaling components that function downstream of attractive and repulsive guidance receptors (Bashaw, 1999).

Another finding of this study is that the signal transduction machinery for attraction and repulsion downstream of these receptors appears to be present in all neurons, and probably in all migrating muscle precursors as well. All neurons expressing either Fra-Robo or Robo-Fra appear to behave the same, regardless of their environment: if they express Fra-Robo, they stay away from the midline; if they express Robo-Fra, they extend toward the midline. No other factor appears to intrinsically commit one growth cone or another to only one kind of response. The same is true for migrating muscle precursors. Normally, many of them express Robo and migrate away from the Slit-expressing midline. However, given the opportunity (by transgenic expression of Robo-Fra), they clearly contain the full machinery for the opposite response. In all these transgenic experiments, the growth cone or muscle response always correlated with the level of receptor (Brashaw, 1999 and references).

The finding that the cytoplasmic sequence determines the response of a guidance receptor raises a number of interesting questions. Attraction might lead to a local change favoring actin polymerization over depolymerization, while repulsion might lead to the opposite change. But is guidance that simple? The cytoplasmic sequences of five different families of repulsive guidance receptors are now known: UNC-5s, Eph receptors, Neuropilins, Plexins, and Robos. Interestingly, they appear to share little if any sequence similarity to one another in their cytoplasmic domains. It is possible, of course, that they bind different adapter proteins that converge on the same repulsive motility machinery. But it is equally likely that not all repulsion is the same and that different classes of repulsive receptors mediate different types of responses in the growth cone. It could be that what is lumped together under the term 'repulsion' actually represents several molecularly distinct mechanisms that negatively influence local growth cone behavior. Just what these different cytoplasmic domains do, and how many different types of repulsion exist, awaits future investigation (Bashaw, 1999 and references).

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, 2000).

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, 2000).

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, 2000).

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, 2000).

The Drosophila ARF6-GEF Schizo controls commissure formation by regulating Slit

The CNS of bilateral symmetric organisms is characterized by intensive contralateral axonal connections. Genetic screens in Drosophila have identified only a few genes required for guiding commissural growth cones toward and across the midline. Two evolutionarily conserved signaling molecules, Netrin and Slit, are expressed in the CNS midline cells. Netrin acts primarily as an attractive signaling cue, whereas Slit mediates repulsive functions. A detailed analysis is provided of the Drosophila gene schizo, which is required for commissure formation. schizo leads to a commissural phenotype reminiscent of netrin mutant embryos. Double-mutant analyses indicate that Netrin and Schizo act independently. The schizo mutant phenotype can be suppressed by either expressing netrin in the CNS midline cells or by a reduction of the slit gene dose, indicating that the balance of attractive and repulsive signaling is impaired in schizo mutants. Overexpression of the schizo RNA in the CNS midline using the GAL4/UAS system leads to a slit phenocopy, suggesting that schizo primarily antagonizes Slit signaling. This is further supported by cell type-specific rescue experiments. The schizo gene generates at least two proteins containing a conserved Sec7 and a pleckstrin homology domain (PH) characteristic for guanine nucleotide exchange factors (GEF) acting on ARF GTPases, which are known to regulate endocytosis, In support of the notion that schizo regulates Slit expression via endocytosis, it was found that blocking endocytosis leads to a schizo-like phenotype. It is thus proposed that the balance of the two signaling cues Netrin and Slit can be regulated, controlling membrane dynamics (Önel, 2004).

Only four zygotically active genes were found in a screen for mutations affecting commissure formation (frazzled, weniger, schizo and the netrin gene complex). Two EMS-induced schizo mutants (schizoC1-28 and schizoU112) were initially identified. Subsequently two P-element induced schizo alleles (schizol(3)3 and schizoP244) were identified. All these alleles led to a reduction in the number of commissural fibers crossing the CNS midline. Interestingly, the anterior commissures were affected more prominently. Not all neuromeres were equally affected and the strongest defects were generally observed in abdominal segments A1-A4. All CNS midline cells formed in normal number in the absence of schizo function. However, as generally observed in mutants affecting formation of commissures, the midline glial cells migrated out laterally along the few remaining commissural fibers. In addition to the commissural phenotype, defects in the longitudinal connectives were noted (Önel, 2004).

The most prominent function of schizo is its role in commissure development. Two major signaling cascades are known to control axonal growth across the midline. They are initiated by the signaling molecules Netrin and Slit, which are both secreted by the CNS midline glial cells in the Drosophila embryo. First genetic interaction studies of schizo and frazzled or schizo and netrin function demonstrate a much stronger commissural phenotype in double mutants than embryos mutant only for schizo, frazzled or netrin. The commissural phenotypes of the double-mutant embryos suggest that schizo is not acting within the Netrin signaling pathway but may be required for a Netrin-independent attractive pathway. Alternatively, schizo may be necessary for suppressing the perception or the generation of a repulsive signal normally generated by the CNS midline cells (Önel, 2004).

The main axonal repulsive signal is encoded by slit. Slit is an LLR protein secreted by the CNS midline glial cells. schizo function appears to be required to downregulate repulsive signaling, either by affecting the generation of active Slit protein or by preventing signaling in the commissural growth cones; the mutant schizo phenotype could be explained by an upregulated Slit signaling. Thus, one might expect that the schizo commissural phenotype could be suppressed by a concomitant reduction in the dose of slit function. slit-/+; schizo–/– as well as robo-/+; schizo–/– embryos were generated, and in both cases a suppression of the schizo CNS phenotype was observed. Thus schizo might be required to negatively regulate Slit signaling (Önel, 2004).

If schizo is indeed a negative regulator of slit function, an increase of schizo gene dose should result in a decrease of active Slit signaling. One might thus be able to enhance the mutant slit phenotype by using a schizo gene duplication. Following mapping of schizo to the base of the left arm of chromosome 3 a chromosomal translocation of the corresponding part of the third chromosome to the Y chromosome (Tp(3;Y)A81) was used to generate embryos with three copies of schizo. In an otherwise wild-type background, this triplication of the schizo region did not result in an abnormal CNS phenotype. However, when the schizo translocation was placed in a heterozygous slit–/+ background, a slit-like phenotype was observed that was never detected in heterozygous slit embryos (Önel, 2004).

schizo maps to chromosome region 78A/B between the genes poils aux pattes and knockout. To identify the schizo gene in this chromosomal interval P-element-induced schizo alleles were used. The chromosomal insertions of the P-elements in l(3)3 and P224 were determined by inverse PCR and Southern analyses and the results suggested that schizo corresponds to CG32434. The lethality associated with the P-element-induced l(3)3 schizo mutation could be reverted by precise excision of the P-element. Mutant l(3)3 embryos displayed a schizo phenotype with reduced commissures and defective fasciculation in the longitudinal connectives. Subsequent sequencing of cDNA clones LP01489, RE44556 and GH10594 isolated by the BDGP showed that the schizo locus encompasses 41 kb of genomic DNA. At least two different promoters direct the expression of two isoforms of 1325 amino acids (SchizoP1) and 1313 amino acids (SchizoP2) in length. Verification of the cloning of schizo was obtained by genetic rescue experiments. These deduced schizo proteins correspond to the Iso1 and Iso2 variants of the loner gene, which was recently identified in a screen for mutations affecting mesoderm development. By contrast Chen (2003) GH10594 was found to be entirely contained within the LP01489 sequence and no evidence was found for a third schizo protein isoform (Önel, 2004).

The deduced Schizo proteins share three conserved sequence modules. In the N-terminal region there is a so-called IQ domain, which is predicted to interact with calmodulin. Within the C-terminal third of the protein a Sec7 domain is directly adjacent to a PH-domain. Proteins characterized by such a domain signature are generally acting as guanine nucleotide exchange factors (GEFs). The Anopheles homolog is about 90% identical. The closest human homologs are EFA6, being 32% identical to Schizo, lacking the IQ domain, and ARF-GEP100 showing a 40% identity to Schizo. Both human proteins were shown to act as ADP ribosylation factor 6 (ARF6)-GEFs suggesting that schizo might have a similar function (Önel, 2004).

The molecular identification of schizo allowed for the determination of the expression pattern throughout development. schizo expression is already detected in the unfertilized egg, indicating a prominent maternal contribution. schizo expression stays almost uniform until the end of stage 10. Within the developing nervous system, expression can be noted in the CNS midline cells. In addition, schizo expression can be detected in the epidermis and the visceral mesoderm (Önel, 2004).

The commissural schizo phenotype does not allow the deduction of cell type in which Schizo normally acts. To test the cell-type requirement the GAL4 system was used and UAS-schizoP1 and UAS-schizoP2 transgenic flies were established. Expression of the different schizo proteins was directed in the CNS midline cells of mutant schizo embryos using the sim-GAL4 or sli-GAL4 driver strains. In both cases expression could rescue the schizo mutant CNS phenotype indicating that Schizo acts in the midline glial cells, which express both Slit and Netrin (Önel, 2004).

Genetic data indicate that schizo impairs Slit signaling in the CNS. This was further supported by overexpression of schizo. Whereas expression of schizo (P1 or P2) in all CNS midline cells of wild-type embryos did not evoke an abnormal phenotype, the same expression of schizo in heterozygous slit mutant embryos was able to induce a mild slit phenocopy. These results were similar to the ones obtained using chromosomal translocations, supporting the notion that Schizo acts in the CNS midline by regulating the level of Slit expression (Önel, 2004).

To reduce Slit activity, schizo might suppress exocytosis of Slit-containing vesicles, or it might promote endocytosis of Slit-containing vesicles from the membrane. Work from vertebrate tissue culture models has shown that Arf-GEFs such as Schizo can activate endocytosis. To test whether endocytosis might be relevant for commissure formation a dominant negative Shibire protein was expressed; this efficiently blocks endocytosis specifically in the CNS midline cells using the sim-GAL driver. In about 60% of such embryos a schizo phenocopy was observed. When higher levels of the dominant negative Shibire protein were expressed using the rho-GAL driver all embryos developed a schizo phenocopy, suggesting that endocytosis participates in the regulation of Slit function in the CNS midline cells. To further support the notion that Schizo induces endocytosis of Slit, the negative Shibire protein was expressed in a heterozygous slit mutant background. This indeed led to clear suppression of the Shibire-induced phenotype. Schizo and its vertebrate homologs exert at least part of the function through the small GTPase Arf6. Arf6 mRNA is supplied maternally and is expressed ubiquitously during embryonic development. To determine whether Schizo acts via Arf6 to control endocytosis of Slit by the midline glial cells, a dominant negative Arf6 construct (Chen, 2003) was expressed. Following expression in the midline cells using the sim-GAL driver no mutant phenotype was observed. Following expression of higher levels of Arf6DN using the rho-GAL driver, about 15% of the embryos developed a schizo phenocopy, supporting the notion that Schizo acts via Arf6-regulated endocytosis to control the level of Slit expression on the midline glial cells (Önel, 2004).

Thus genetic and molecular data support a model in which schizo negatively regulates the expression of Slit in the CNS midline cells. This study shows that schizo acts in a rather similar way to commissureless; however, rather than affecting the Roundabout receptor, schizo appears to act on the expression of the Slit ligand. First it was found that the triplication of the schizo gene interfers with slit function and that reduction of slit expression in schizo mutant embryos rescues the schizo mutant phenotype. Finally, expression of a schizo transgene in the Slit-expressing CNS midline cells (1) was able to rescue the schizo mutant phenotype and (2) could induce a slit phenocopy when expressed in wild-type embryos. The deduced nature of the Schizo protein suggests that it affects Slit expression by post-transcriptional mechanisms (Önel, 2004).

Guanine-nucleotide exchange factors (GEFs) help to convert the inactive GDP-bound form of small GTPases into a GTP-bound active form. Schizo is a new Sec7 domain containing GEF, which shows 40% homology to human Arf-GEP100. Arf-GEP100 localizes to endosomal membranes and promotes GDP/GTP exchange on ARF6. The small GTPase ARF6 is a plasma membrane-localized protein and functions in the regulation of membrane ruffling, cell motility, aspects of endocytosis and exocytosis, membrane recycling, reorganization of the cortical actin cytoskeleton and activation of phospholipase D. In Drosophila, Arf6 is remarkably well conserved, being more than 96% identical to the human counterpart (Önel, 2004).

One aspect that might hint at how Schizo regulates Slit expression is the role of ARF6 in endocytosis and exocytosis. The function of ARF6 in endocytosis is twofold. It either regulates clathrin-mediated endocytosis at the apical surface of polarized epithelial cells or it is able to regulate non-clathrin-mediated endocytosis and the recycling pathway in non-polarized cells. ARF6 has also been postulated to play a role in Ca2+-activated dense core vesicle (DCV) exocytosis by regulating phosphatidylinositol(4,5) biphosphate (PIP2). Overexpression of a UAS-ARF6 construct in midline glia cells does not result in a schizo-like phenotype, whereas expression of a dominant negative form of Arf6 results in a phenocopy of several phenotypes associated with the schizo mutant (Chen, 2003). This suggests that Arf6 might also be involved in the regulation of Slit expression (Önel, 2004).

In-vivo Slit and Netrin are both expressed by the same CNS midline cells and their expression needs to be in an intricate balance. The importance of this balance and not the individual expression levels is highlighted by the fact that it is possible to rescue the schizo mutant phenotype by both increased Netrin expression or reduced Slit expression. Within the midline glia, however, Schizo appears to primarily affect Slit expression either by inducing its endocytosis and subsequent degradation or by blocking exocytosis and thus release of Slit (Önel, 2004).

The latter case would suggest that Slit and Netrin are brought to the membrane of the midline glial cells in distinct vesicle populations, whereas the former case would require a specific membrane receptor for the Slit protein expressed by the CNS midline glia. Given the fact that the secreted Slit protein is found at very high levels at the midline glial cell membrane, this appears probable. Moreover, expression of a dominant negative Shibire protein in the midline glia leads to a schizo phenocopy. shibire encodes the Drosophila dynamin and is required for endocytosis and a block of shibire function leads to a block of endocytosis, which might result in higher levels of Slit expression. Thus, regulation of membrane dynamics appears crucial in controlling the function of the signaling molecule Slit (Önel, 2004).

Heparan sulfate proteoglycan syndecan promotes axonal and myotube guidance by slit/robo signaling

Biochemical studies suggest that axon guidance activity requires cell-surface heparan sulfate to promote binding of mammalian Slit/Robo homologs. Drosophila Syndecan, a heparan sulfate proteoglycan (HSPG), is required for proper Slit signaling. Slit, the ligand for the Roundabout (Robo) receptors, is secreted from midline cells of the Drosophila central nervous system (CNS). It acts as a short-range repellent that controls midline crossing of axons and allows growth cones to select specific pathways along each side of the midline. In addition, Slit directs the migration of muscle precursors and ventral branches of the tracheal system, showing that it provides long-range activity beyond the limit of the developing CNS. Syndecan (Sdc) mutations have been generated; they affect all aspects of Slit activity and cause robo-like phenotypes. sdc interacts genetically with robo and slit, and double mutations cause a synergistic strengthening of the single-mutant phenotypes. The results suggest that Syndecan is a necessary component of Slit/Robo signaling and is required in the Slit target cells (Steigemann, 2003).

Genetic assays provide a sensitive means of detecting an in vivo interaction between different components in a pathway, but they do not show that the association is direct. Thus, a biochemical assay was developed to determine whether Sdc binds to Slit and/or Robo in cellular extracts in which all three proteins are endogenously expressed. Immunoprecipitation of either Slit or Robo and subsequent detection with anti-Sdc antibodies reveals that Sdc associates with both Slit and its receptor, suggesting the possibility of a ternary complex. This association is specific because no Sdc is trapped by nonspecific IgG or N-Cadherin antibodies that successfully IP other signaling molecules. Thus, Sdc participates directly in a complex with Slit and Robo (Johnson, 2004).

Homozygous sdc mutants are semilethal and show identical phenotypes in the CNS and in the muscle pattern. In order to unambiguously demonstrate that the lack of sdc activity is responsible for the mutant phenotype observed, Sdc-RA was panneurally expressed using the GAL4/UAS system in sdc mutant individuals. The neural phenotype of the mutants was rescued, indicating that the mutant phenotype was caused by the lack of Sdc and that the transgene provides functional Sdc activity (Steigemann, 2003).

In order to examine the possible defects in axonal guidance and muscle patterning, sdc mutant embryos were stained with both Fasciclin II (FasII; mAB 1D4) antibodies, which label three longitudinal axon tracts at each side of the midline, and anti-Mhc antibodies, which visualize the muscle pattern. The results show that the lack of sdc activity causes phenocopies of robo and robo2 mutants; i.e., it affects both midline guidance of axons and the establishment of the muscle pattern. The defects in CNS axon guidance were strikingly similar to robo2 mutants but less pronounced than in robo mutants. The muscle and CNS phenotypes were also weaker than in slit mutants, in which signaling through all Robo receptors is impaired (Steigemann, 2003).

In order to link embryonic Sdc requirement genetically to Slit/Robo signaling, it was next asked whether sdc mutations can enhance loss-of-function slit and robo phenotypes. It was found that the number of ventral muscles, which cross the midline dorsal of the CNS in homozygous sdc and robo2 single mutants, is significantly increased in double mutant combinations of sdc and robo2, resulting in a muscle phenotype indistinguishable from slit and homozygous robo, robo2 double mutants. In the CNS, the FasII-expressing longitudinal fascicles of robo2, sdc double mutants converged into a single thick axon bundle at the ventral midline, resembling the effects seen with slit mutants. Similar observations were obtained with the mAb BP102 against all CNS axons, showing strongly condensed fascicles in robo2, sdc double mutant embryos. The synergistic strengthening of both the muscle and the CNS phenotypes in robo2, sdc double mutants, which are similar to a weak slit mutant phenotype, indicates that only some Slit-derived repellent activity is received along the midline. In contrast to robo2, the robo mutant phenotype was not significantly enhanced by the simultaneous lack of sdc. The data suggest that Robo can, in part, compensate for the lack of Robo2 and vice versa and that Robo is more sensitive to reduced Sdc-dependent Slit activity than Robo2 (Steigemann, 2003).

The results imply that sdc, slit, and robo are components of the same genetic circuitry. This proposal was tested by genetic means, asking whether the gene activities interact in vivo. Loss of only one copy of sdc led to the development of a normal muscle pattern, whereas the simultaneous absence of one copy of both slit and sdc in slit/+, sdc/+ double heterozygous embryos caused an increase in the number of longitudinal transverse muscles. Furthermore, the number of FasII-expressing inner fascicles that cross the midline is increased (3.3%) as compared to slit heterozygous embryos (0.6%;). More clearly, homozygous sdc mutant embryos, which also lack one copy of either robo, robo2, or slit, show an enhanced axonal guidance defect with multiple midline crossings of the fascicles (72%, 47%, and 95%, respectively), a phenotype very similar to the robo mutant. These results establish that Sdc acts in the same genetic circuitry as Slit and the Robo receptor family and represents a critical component of the Slit/Robo signaling pathway (Steigemann, 2003).

Distinct protein domains and expression patterns confer divergent axon guidance functions for Drosophila Robo receptors

The orthogonal array of axon pathways in the Drosophila CNS is constructed in part under the control of three Robo family axon guidance receptors: Robo1, Robo2 and Robo3. Each of these receptors is responsible for a distinct set of guidance decisions. To determine the molecular basis for these functional specializations, homologous recombination was used to create a series of 9 'robo swap' alleles: expressing each of the three Robo receptors from each of the three robo loci. The lateral positioning of longitudinal axon pathways was shown to rely primarily on differences in gene regulation, not distinct combinations of Robo proteins as previously thought. In contrast, specific features of the Robo1 and Robo2 proteins contribute to their distinct functions in commissure formation. These specializations allow Robo1 to prevent crossing and Robo2 to promote crossing. These data demonstrate how diversification of expression and structure within a single family of guidance receptors can shape complex patterns of neuronal wiring (Spitzweck, 2010).

The midline guidance cue Slit is thought to act through each of three different Robo family receptors to help form the orthogonal axonal pathways of the Drosophila ventral nerve cord. Each of the three Robos has a distinct role in forming these projections. Robo1 is primarily required to prevent longitudinal axons from crossing the midline. Robo2 has a minor role in preventing longitudinal axons from crossing, and, as this study has shown, also facilitates the crossing of commissural axons. Finally, Robo3 may also help prevent some longitudinal axons from crossing, but its major function is to direct the formation of the intermediate longitudinal pathways (Spitzweck, 2010).

The goal of this study was to assess whether these functional specializations reflect structural differences in the Robo proteins themselves or differences in robo gene regulation. To this end, gene targeting was used to replace the coding region of each robo gene with that of each other robo, creating a series of robo swap alleles. It was found that commissure formation relies on the unique structural features of both Robo1 (to prevent crossing) and Robo2 (to promote crossing). In contrast, lateral positioning of longitudinal axons does not rely on structural differences between the Robo proteins, but rather differences in robo gene expression (Spitzweck, 2010).

In the longitudinal pathways, axons are organized into discrete and stereotyped fascicles. In part, this requires selective fasciculation mediated by contact-dependent attractive or repulsive surface proteins that 'label' specific axon fascicles. This includes the FasII protein which was exploited in this study as a marker. In addition to these pathway labels, the lateral pathways are also segregated into three broad zones according to the distinct combination of Robo receptors they express. Loss- and gain-of-function genetic experiments have shown that these Robo proteins are instructive in lateral pathway selection and, hence, define a 'Robo code' (Spitzweck, 2010).

A popular model for lateral pathway selection posits that the three Robo proteins have distinct signaling properties, and that they position axons on a lateral gradient of their common ligand Slit. In this model, the Robo proteins are assumed to differ in either their affinity for Slit, the strength of their 'repulsive output,' or both. However, direct evidence for a role of Slit in lateral pathway is still lacking, and alternative models have to be considered. One such possibility is that the Robo proteins might act instead as homophilic adhesion molecules. In such a model, the Robo proteins might operate in a manner similar to other pathway labels such as FasII, but over broader zones. Regardless of whether they invoke a role for Slit, homophilic adhesion, or some other unidentified ligand, all models presented to date have assumed that there must be critical structural differences in the Robo proteins. These structural differences would form the basis of a combinatorial Robo code for lateral pathway selection (Spitzweck, 2010).

The current data demonstrate that this cannot be the case. Lateral positioning does not rely on structural differences between the Robo proteins. This is particularly clear for the distinction between the medial and the intermediate zones, which relies entirely on the selective expression of Robo3 on intermediate axons. This study found, however, that lateral positioning of these axons works surprisingly well even when Robo3 protein is replaced by either Robo1 or Robo2. Although some minor disruption in specific pathways cannot be excluded, the overall structure of the longitudinal pathways appears normal in these embryos. Notably, this includes the formation of the intermediate FasII pathway and the projections of the Sema2b axons, both of which were diagnostic for Robo3's role in lateral positioning. Thus, at least for the medial and intermediate axons, the only relevant differences between the Robos are in their patterns of gene expression. The 'Robo code' is not a protein code; it is a gene-expression code (Spitzweck, 2010).

At first glance, this result is difficult to reconcile with the previously published gain-of-function experiments. In these experiments, the various Robo proteins were expressed from GAL4/UAS transgenes in specific neurons (the Ap neurons). These Ap neurons normally express only Robo1 and hence project ipsilaterally in the medial zone. In both reports, expression of Robo3 shifted these axons into the intermediate zone, as expected, but expression of Robo1 did not. Why might Robo1 be able to replace the endogenous Robo3 in the swap experiments, but not the transgenic Robo3 in these gain-of-function studies? A trivial but unsatisfying explanation is that this was merely an artifact of the GAL4/UAS system. Prior to the advent of site-specific transgenesis, it was notoriously difficult to control for the varying expression levels from different transgene insertions, which rarely match endogenous levels. More interesting possibilities are that the discrepancy may reflect differences resulting from assaying the behavior of neurons that normally express Robo3 versus those that don't, or perhaps a 'community effect' that results from manipulating an entire cohort of neurons, not just a single neuron. In this regard it is also important to note that the Ap axons are likely to be follower axons for their specific pathway, not pioneers. Whatever the reason for this discrepancy, the substitution of the robo1 coding region into the robo3 locus is presumably the more physiologically relevant assay (Spitzweck, 2010).

How might differences in robo gene expression explain lateral positioning? One possibility is that it is only the total Robo levels that are important, with higher levels sending axons further laterally on the presumptive Slit gradient. This model fits with the results of 'supershifting' experiments, in which additional copies of the Robo3 transgene displaced the Ap axons even further from the midline. It is also supported by mathematical modeling of the Robo code. This model still invokes a role for the Slit gradient, for which there is admittedly no direct evidence. Alternatively, lateral pathway selection might rely on critical differences in the precise spatial and temporal pattern of expression, rather than differences in total Robo levels (Spitzweck, 2010).

It has long been appreciated that Robo1 is the primary receptor through which Slit repels longitudinal axons to prevent them from crossing the midline. Midline crossing errors occur in every segment of robo1 mutants, but are relatively rare in both robo2 and robo3 mutants. This study has shown that this unique function of Robo1 relies on differences in both gene regulation and protein structure. Specifically, Robo1 cannot exert its midline repulsion function when expressed in the pattern of robo2 or robo3, nor can Robo2 or Robo3 prevent midline crossing when expressed in the manner of robo1 (Spitzweck, 2010).

By examining a series of chimeric receptors consisting of distinct parts of Robo1 and Robo3, this critical and unique function of Robo1 in midline repulsion was mapped to a region of the cytoplasmic domain containing the CC1 and CC2 motifs. This conclusion is broadly consistent with previous studies that have examined Robo1 deletion mutants lacking specific CC motifs, in this case in a pan-neuronal transgenic rescue assay. Although there are subtle differences that may reflect the use of chimeric receptors versus single domain deletions, and the consequences of expressing them under the control of endogenous versus heterologous gene regulatory elements, the two studies together strongly suggest that the proline-rich CC2 motif is the critical structural determinant of Robo1's unique capability of preventing midline crossing. This domain is thought to serve as a docking site for a number of factors that contribute to Slit-dependent repulsion through Robo1, including Enabled, the Rac GTPase activating protein Vilse/CrGAP, and the SH2-SH3 adaptor Dock, the latter recruiting in turn the Rac guanine nucleotide exchange factor Sos and p21 activated kinase. CC2 is also the most broadly conserved of the cytoplasmic domains in Robo1, with the insect Robo2 and Robo3 proteins being the only known Robo receptors that lack CC2. The lack of CC2 in Robo2 and Robo3 cautions against the inference that the distinct guidance functions of these two receptors are necessarily mediated by repulsive signaling in response to activation by Slit (Spitzweck, 2010).

Indeed, this study has presented evidence that Robo2 can even act in opposition to Robo1 to promote crossing. It is assumed that Robo2 normally exerts this positive function autonomously in commissural neurons, acting in parallel to Netrin-Frazzled signaling to allow midline crossing. Two models are envisioned to account for the positive role of Robo2 in midline crossing. In one scenario, Robo2 transduces an attractive signal that promotes crossing, possibly in response to its midline ligand Slit. Such a model has previously been proposed for Robo2 in the guidance of ganglionic tracheal branches. Alternatively, Robo2 might promote crossing by antagonizing the repulsive function of Robo1, thus mediating an 'anti-repulsion' rather than an 'attraction' signal. Formally, this model is analogous to the role of Comm in Drosophila, and of Robo3/Rig-1 in vertebrates. Preliminary data are more consistent with this latter scenario (Spitzweck, 2010).

Three factors are now known that promote midline crossing: Comm, Netrin-Frazzled, and Robo2. Of these, only Comm appears to be instructive. Comm is expressed in commissural but not ipsilateral neurons, and is both necessary and sufficient for crossing. In contrast, both Frazzled and Robo2 are permissive: they are expressed in both commissural and ipsilateral neurons, and are required but not sufficient for crossing. They are also partially redundant and independent, as crossing is severely disrupted only when both are eliminated. A conceptual model for midline crossing proposes a bistable switch created by the mutual inhibition between high Robo1 levels and midline crossing: high Robo1 levels prevent crossing due to repulsive signaling, whereas crossing the midline leads to clearance of Robo1 protein from the midline axon segment. In such a model, the permissive factors (Frazzled and Robo2) may act to ensure the appropriate balance between midline attraction and midline repulsion, bringing this feedback loop into the dynamic range at which the instructive factor (Comm) can operate. In principle, any one of the three factors--Comm, Robo2, or Frazzled--could have taken on the instructive role. Comm has evidently done so in Drosophila. To the extent that a similar feedback loop operates in mice, the instructive role may have fallen in this species to the Robo2 analog, Robo3 (Spitzweck, 2010).

Functional diversity of Robo receptor immunoglobulin domains promotes distinct axon guidance decisions

Recognition molecules of the immunoglobulin (Ig) superfamily control axon guidance in the developing nervous system. Ig-like domains are among the most widely represented protein domains in the human genome, and the number of Ig superfamily proteins is strongly correlated with cellular complexity. In Drosophila, three Roundabout (Robo) Ig superfamily receptors respond to their common Slit ligand to regulate axon guidance at the midline: Robo and Robo2 mediate midline repulsion, Robo2 and Robo3 control longitudinal pathway selection, and Robo2 can promote midline crossing. How these closely related receptors mediate distinct guidance functions is not understood. This study reports that the differential functions of Robo2 and Robo3 are specified by their ectodomains and do not reflect differences in cytoplasmic signaling. Functional modularity of Robo2's ectodomain facilitates multiple guidance decisions: Ig1 and Ig3 of Robo2 confer lateral positioning activity, whereas Ig2 confers promidline crossing activity. Robo2's distinct functions are not dependent on greater Slit affinity but are instead due in part to differences in multimerization and receptor-ligand stoichiometry conferred by Robo2's Ig domains. Together, these findings suggest that diverse responses to the Slit guidance cue are imparted by intrinsic structural differences encoded in the extracellular Ig domains of the Robo receptors (T. A. Evans 2010).

In the Drosophila embryonic central nervous system (CNS), Robo receptors are expressed in overlapping domains that divide the longitudinal axon connectives into three broad zones: axons occupying the medial zone express Robo, axons in the intermediate zone express Robo and Robo3, and axons in the most lateral zone express Robo, Robo2, and Robo3. Loss of robo2 shifts lateral axons to intermediate positions, whereas loss of robo3 shifts intermediate axons to medial positions. Conversely, ectopic expression of Robo2 or Robo3 in medial axons forces them to select more lateral pathways, whereas increased levels of Robo do not. The 'Robo code' model posits that a combinatorial code of Robo receptor expression determines the lateral position of CNS axons. To test whether a combinatorial code is necessary, the ability was tested of Robo2 and Robo3 to shift apterous axons in embryos deficient for various combinations of robo genes; removing endogenous robo or robo3 was found not to affect Robo2's ability to shift apterous axons laterally. Indeed, UAS-Robo2 was sufficient to direct the apterous axons to the lateral edge of the connectives even in robo3, robo double mutant embryos. Similarly, removal of robo2 or robo had little or no effect on the ability of UAS-Robo3 to redirect the apterous axons to more lateral pathways. Thus, it is the individual expression of Robo2 and Robo3 that dictates lateral positions of CNS axons, not a combinatorial Robo code (T. A. Evans 2010).

Robo2 and Robo3 dictate the lateral position of axons in the Drosophila CNS, a role that is not shared by Robo. What is the basis for this differential activity? All three receptors have similar ectodomains with five immunoglobulin (Ig) domains and three fibronectin (Fn) III repeats, whereas their cytoplasmic domains are more divergent. In particular, Robo2 and Robo3 both lack two conserved motifs (CC2 and CC3) that mediate interactions with several downstream effectors and are required for Robo's midline repulsive function, leading to the speculation that distinct Robo functions are directed by their cytoplasmic domains. To determine whether the functional difference between Robo2-Robo3 and Robo is due to a qualitative difference in cytoplasmic signaling, a set of chimeric receptors was assayed for their ability to induce lateral shifting in the medial apterous axons (T. A. Evans 2010).

First, the cytoplasmic domain of Robo was replaced with that of Robo2 or Robo3 (Robo1:2 and Robo1:3). Neither of these receptor variants was able to reposition the apterous axons. In contrast, when the cytoplasmic domains of Robo2 or Robo3 were replaced by that of Robo, the resulting chimeric receptors (Robo2:1 and Robo3:1) exhibited lateral positioning activity similar to full-length Robo2 and Robo3. These results reveal that the lateral positioning activities of Robo2 and Robo3 are specified by their ectodomains. Importantly, the cytoplasmic domains of Robo2 and Robo3 are not dispensable for lateral positioning activity, because receptors without any cytodomains are unable to redirect the apterous axons laterally. Because Robo cytoplasmic domains are functionally interchangeable for longitudinal pathway selection, any required intracellular events must be mediated by cytoplasmic sequences that are common to Robo, Robo2, and Robo3 (T. A. Evans 2010).

To dissect the structural basis underlying the differential activities of Robo receptor extracellular domains, the relative contributions of Robo2's Ig and Fn domains were examined by generating a more restricted set of domain swaps between Robo and Robo2. Exchanging all five Ig domains between Robo and Robo2 completely swapped their lateral positioning activities. These results reveal that Robo2's ability to position axons is specified entirely by its Ig domains. However, the Fn repeats are not completely dispensable for lateral positioning activity because Robo2 variants lacking these elements displayed reduced activity. Thus, when combined with Robo2's five Ig domains, the Fn repeats and cytoplasmic domain of Robo can act permissively to facilitate lateral pathway choice (T. A. Evans 2010).

The five Ig domains of Robo2 are necessary and sufficient to functionally distinguish it from Robo in the context of longitudinal pathway choice. To subdivide the ectodomains of Robo and Robo2 further, the presumptive Slit-binding region (Ig1) was targeted. Initially Ig1 and Ig2 were swapped together, because some evidence suggested that Ig2 could contribute to Slit binding of human Robo receptors. Robo variants possessing the first and second Ig domains of Robo2 (Robo1R2I1+2) displayed activity comparable to full-length Robo2. However, the converse swap revealed that Robo2 still retained its activity even when its Ig1+2 was replaced with those of Robo (Robo2R1I+2). These results reveal a bipartite contribution to Robo2's lateral positioning activity from (at least) two genetically separable elements located within Ig1+2 and Ig3-5, respectively (T. A. Evans 2010).

Next whether Ig1 and Ig3 together could be responsible for dictating the lateral positioning activity of Robo2 was tested. Replacing Ig1 or Ig3 of Robo with those of Robo2, alone (Robo1R2I1 and Robo1R2I3) or in combination (Robo1R2I1+3), was sufficient to confer Robo2-equivalent activity to Robo. Importantly, replacing Ig1-3 of Robo2 with the corresponding domains of Robo eliminated its lateral positioning activity, demonstrating that the Ig1-3 region is both necessary and sufficient to functionally distinguish Robo1 and Robo2 in the context of longitudinal pathway choice (T. A. Evans 2010).

Ig1 and Ig3 of Robo2 can independently specify its ability to redirect medial axons to more lateral pathways. Further, the lateral positioning activities of chimeric receptors containing Ig1 or Ig3 of Robo2 were indistinguishable in the apterous neuron assay. To determine whether these receptors could also influence longitudinal pathway choice in a broader context, the effects were assayed of pan-neuronal misexpression of selected chimeric receptors on lateral positioning of FasII-positive axon pathways (T. A. Evans 2010).

In wild-type embryos or elavGAL4;UAS-Robo embryos, three major FasII-positive tracts were detectable on either side of the midline. Pan-neuronal misexpression of Robo2, in contrast, disrupted longitudinal pathway formation such that the intermediate FasII pathway was absent in nearly all segments. Notably, this effect appeared to depend solely on Ig3 of Robo2, because it was recapitulated by UAS-Robo2R1I1+2 and UAS-Robo1R2I3, but not by UAS-Robo1R2I1+2 or UAS-Robo2R1I1-3. These observations draw a functional distinction between the activities of Ig1 and Ig3 of Robo2 and suggest that these two domains regulate longitudinal pathway choice via distinct mechanisms (T. A. Evans 2010).

Because the Slit-binding Ig1 contributes to Robo2's lateral positioning activity, it is possible that Robo2 regulates longitudinal pathway selection in response to Slit. If so, then removing slit or disrupting its interaction with Robo2 should reduce or eliminate Robo2's lateral positioning activity. Therefore, the effects of Robo2 misexpression in apterous axons were examined in a slit mutant background. In the absence of Slit, the entire axon scaffold collapsed at the midline, and even high levels of ectopic Robo2 could not force the apterous axons laterally. This may indicate a direct requirement for Slit or instead reflect the inability of Robo2-expressing apterous axons to move outside of the collapsed axon scaffold (T. A. Evans 2010).

Whether Robo2 could reposition axons without its Slit-binding region was examined next. To ensure complete disruption of Slit binding, both the first and second Ig domains were deleted from Robo2; Robo2ΔIg1+2 was completely unable to reposition the apterous axons. Deleting these two domains did not interfere with expression or localization of Robo2 . Together, these results provide evidence that Robo2-directed lateral positioning is dependent on interactions with Slit; however, it is noteed that in addition to disrupting Slit binding, deletion of Ig1 and Ig2 would also disrupt other potentially important functions of these domains. Genetic analysis of the role of robo3 in the regulation of lateral chordotonal axon arborization within the CNS also supports Slit-dependent control of lateral position by Robo receptors (T. A. Evans 2010).

Interestingly, pan-neuronal misexpression of Robo2 results in phenotypes that are inconsistent with a strictly repulsive function for Robo2. At the highest levels of overexpression, Robo2 prevents all midline crossing. However, moderate levels of Robo2 overexpression lead to ectopic midline crossing, suggesting that in some contexts Robo2 can promote midline crossing. Perhaps Robo2, like the divergent Robo receptor Rig-1/Robo3 in vertebrates, can antagonize Slit-Robo repulsion (T. A. Evans 2010).

The panel of chimeric receptors was used to map this activity of Robo2. All of the receptor variants that contain Ig2 of Robo2 promoted midline crossing when misexpressed with elavGAL4, whereas those that contain regions of Robo2 apart from Ig2 did not. Thus, the promidline crossing activity of Robo2 is conferred by Ig2. Interestingly, rather than being excluded from the crossing portions of axons like all other Robo receptor variants, Robo2 proteins that promoted midline crossing were expressed strongly on crossing axons. This localization to crossing axons was not shared by any of the Robo3 or Robo3-Robo1 receptors (T. A. Evans 2010).

Although the mechanism of Robo2's procrossing function cannot be addressed at this time, the fact that it is dependent on Ig2 alone suggests that it is probably not due to Robo2 binding Slit and sequestering Slit away from endogenous Robo. It is also noted that this crossing activity does not correlate with lateral positioning activity, because some variants with strong lateral positioning activity (e.g., Robo2R1I1+2, Robo1R2I1+3, Robo1R2I1, and Robo1R2I3) do not promote ectopic midline crossing. It will be interesting to determine whether Robo2 in Drosophila promotes midline crossing through inhibition of Robo or, alternatively, whether it mediates midline attraction in certain contexts. If, like Rig-1/Robo3, Robo2 acts as an antirepellent, it is likely to achieve this function through a distinct mechanism because Rig-1/Robo3's antirepellent function is specified by its cytoplasmic domain (T. A. Evans 2010).

Because Robo2's Ig domains control lateral positioning, one possibility is that Robo2 may have a higher affinity for Slit, encouraging Robo2-expressing axons to seek out positions farther down the Slit gradient. To test this possibility, the Ig domain-containing portions of the Robo and Robo2 ectodomains were purified, and their affinities for the Robo-binding domain of Slit (Slit D2) were compared with surface plasmon resonance (SPR). It was found that Robo2 does not exhibit a higher Slit affinity than Robo; instead, the Ig1-5 region of Robo binds Slit D2 around 4-fold as strongly as the equivalent region of Robo2. Thus, the functional distinction between Robo and Robo2 for longitudinal pathway choice is not increased Slit affinity of Robo2. Furthermore, these observations suggest that the promidline crossing activity of Robo2 does not result from greater Slit affinity (T. A. Evans 2010).

Apart from modest affinity differences, a second distinction was observed between the Slit binding profiles of Robo and Robo2. When tested against a constant amount of immobilized Slit, the maximum equilibrium binding response for Robo was approximately half of that for Robo2. Thus, at equilibrium, the same amount of Slit can bind twice as much Robo2 as Robo, suggesting a difference in receptor-ligand stoichiometry. Size-exclusion chromatography (SEC) confirmed that the Ig1-5 fragment of Robo is almost exclusively monomeric in solution, whereas Robo2 Ig1-5 appears almost exclusively as a dimer. These experiments were performed in the absence of Slit, indicating that the observed multimerization of Robo2 is at least partially ligand independent. However, the differences in maximum Slit binding response in the SPR experiments indicate that the multimerization states of Robo and Robo2 remain distinct even upon Slit binding (T. A. Evans 2010).

To determine which region(s) of Robo2 are responsible for dimerization and whether the observed differences in receptor multimerization correlate with the two distinct lateral positioning activities observed in vivo, equivalent Ig1-5 fragments derived from the chimeric receptors Robo1R2I1+2 and Robo2R1I1+2 were examined via SEC. These reciprocal chimeric receptors contained distinct portions of Robo2 and exhibited distinct large-scale effects on FasII tract formation. The Robo2R1I1+2 receptor fragment (containing Ig3-5 of Robo2) was found to exhibit Robo2-like Slit-independent dimerization, whereas the Robo1R2I1+2 fragment (containing Ig1+2 of Robo2) did not. Thus, ectodomain-dependent dimerization of Robo2 correlates with its ability to influence large-scale longitudinal pathway choice by FasII-positive axons and may account for Ig3's contribution to the lateral positioning activity of Robo2 (T. A. Evans 2010).

How do closely related axon guidance receptors, responding to a common ligand, generate diverse and, in some cases, opposing guidance outcomes? This study has shown that the differential roles of the Robo receptors in directing longitudinal pathway choice are determined by structural differences between receptor ectodomains. In addition, evidence is provided that a second function of Robo2 to promote midline crossing also depends on structural features of its ectodomain. It is concluded that the diversification of Robo receptor axon guidance activities is facilitated by the functional modularity of individual receptor ectodomains. Although the importance of guidance receptor cytoplasmic domains in controlling guidance decisions has been known for a decade, the results reveal that Robo receptor Ig domains play an important part in the functional diversification of this ancient and evolutionarily conserved guidance receptor family (T. A. Evans 2010).

Slit and Receptor tyrosine phosphatase 69D confer spatial specificity to axon branching via Dscam1

Axonal branching contributes substantially to neuronal circuit complexity. Studies in Drosophila have shown that loss of Dscam1 receptor diversity can fully block axon branching in mechanosensory neurons. This paper reports that cell-autonomous loss of the receptor tyrosine phosphatase 69D (RPTP69D) and loss of midline-localized Slit inhibit formation of specific axon collaterals through modulation of Dscam1 activity. Genetic and biochemical data support a model in which direct binding of Slit to Dscam1 enhances the interaction of Dscam1 with RPTP69D, stimulating Dscam1 dephosphorylation. Single-growth-cone imaging reveals that Slit/RPTP69D are not required for general branch initiation but instead promote the extension of specific axon collaterals. Hence, although regulation of intrinsic Dscam1-Dscam1 isoform interactions is essential for formation of all mechanosensory-axon branches, the local ligand-induced alterations of Dscam1 phosphorylation in distinct growth-cone compartments enable the spatial specificity of axon collateral formation (Dascenco, 2015).

This study reports on a molecular mechanism regulating Dscam1 activity in growth cones and provides insight in the regulation and spatial specificity of axon collateral formation. Biochemical and genetic results are consistent with the molecular model that the specificity of mechanosensory (ms)-axon branching arises from a spatially restricted change of Dscam1 phosphorylation in growth cone (Dascenco, 2015).

Previous studies on the function of Dscam1 have established the model that isoform-specific homophilic Dscam1-Dscam1 interactions trigger repulsion between sister dendrites. This controls for regular spacing of sister dendrites in a process termed neurite self-avoidance. In addition, cell-intrinsic and isoform-specific interactions have also been shown to be important in sensory axons for growth-cone sprouting and branching. Importantly, for both of these functions, it is thought that Dscam1 signaling is primarily dependent on and initiated by homophilic binding of matching isoforms present on sister neurites. The results reported in this study provide evidence that Dscam1-Dscam1 interactions in axonal growth cones are subject to branch-specific modulation by extrinsic cues. Binding of the ligand Slit to Dscam1 can locally enhance cis-interactions with the receptor tyrosine phosphatase RPTP69D as well as the dephosphorylation of Dscam1. Although homophilic Dscam1 interactions can be considered to play an initial permissive role in all neurite-neurite interactions in a sprouting growth cone, the spatial restriction of an extrinsic Dscam1 ligand likely initiates functional disparity of Dscam1 signaling across different growth-cone compartments (Dascenco, 2015).

The biochemical data support the notion that RPTP69D directly dephosphorylates Dscam1 at specific cytoplasmic tyrosines. Three candidate tyrosines were identified for the regulation of Dscam1 phosphorylation: Y1857, Y1890, and Y1981. Two of the tyrosine residues, Y1857 and Y1890, are part of consensus SH2-binding sites and therefore are likely involved in regulating recruitment of SH2-domain-containing adaptor molecules. Given that these mutations diminish the Dscam1 GOF effects, it seems reasonable to speculate that they are required for downstream signaling and/or receptor turn-over or trafficking. Surprisingly, the single Y1981F mutation causes strong dominant interference with axon branching where the phenotypic effects are qualitatively indistinguishable from a loss of Dscam1 isoform diversity, which is thought to increase the probability of matching isoform interactions (i.e., GOF activity). The primary amino acid sequence surrounding Y1981 does not reveal any distinct signaling motif. However, in silico 3D protein modeling based on structural predictions suggests that phosphorylation of Y1981 could directly result in structural changes of the Dscam1 cytoplasmic domain and thereby influence Dscam1 activity (Dascenco, 2015).

Biochemical results suggest that Slit can enhance Dscam1-RPTP69D complex formation and Dscam1 dephosphorylation. Furthermore, Slit-N can directly bind to the N-terminal Ig domains of Dscam1 (Ig1-4) with an affinity comparable to that of other guidance cue/receptor interactions, suggesting that Slit-N can function as a bona fide Dscam1 ligand. Numerous studies have shown that the repellent as well as the branch-promoting function of vertebrate Slit require the function of Robo receptors. The current results show that for the formation of specific axon collaterals of Drosophila ms-neurons, Slit functions via Dscam1 in a Robo1-3-independent pathway (Dascenco, 2015).

Slit is one of the best characterized 'axon-repellent' cues and also contributes to axon branching. Imaging single ms-axons and growth-cone branching, this study found that in Slit mutant animals, only filopodia or micropodia with a midline-directed growth direction are reduced, consistent with a positive role of Slit in promoting the extension of specific branches. In contrast, branch-point initiation in ms-neurons is likely independent of Slit or RPTP69D (Dascenco, 2015).

Given that high Slit protein concentrations are likely only encountered by filopodia- or micropodia-like extensions that reach the midline proximity, the Slit-Dscam1-RPTP69D interactions are likely only occurring in a spatially restricted sub-compartment of the branching growth cone. It is envisioned that the Dscam1-RPTP69D interactions in ms-axons constitute a molecular selection process, which depends on Dscam1-RPTP69D complex formation in a subset of axonal processes that encounter sufficient Slit protein. As a result, Dscam1 dephosphorylation by RPTP69D is increased locally and triggers a response by either promoting axon-branch extension or blocking repulsion (Dascenco, 2015).

The loss of only a subset of axon branches in RPTP69D/Slit mutants suggests that there are multiple molecular control pathways accounting for the selection of different axon collaterals or the extension of the main axon shaft. Although this study has focused on RPTP69D and Slit, it is most likely that other co-receptors and extracellular cues control the activity of Dscam1 in growth cones (Dascenco, 2015).

slit: Biological Overview | Evolutionary Homologs | Developmental Biology | Effects of Mutation | References

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

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