InteractiveFly: GeneBrief

betpth: Biological Overview | Developmental Biology | Effects of Mutation | References

BIOLOGICAL OVERVIEW

beaten path codes for a secreted immunoglobulin domain protein that is required for subsets of motor axons to correctly defasciculate from other motor axons at specific choice points. Without this protein, motor axons do not properly enter their muscle target regions. The mutant phenotype resembles a result from previously published experiments where motor axon adhesivity is increased by overexpression of the cell adhesion molecule Fasciclin II (Lin, 1994 ). This similarity suggests that Beat may normally work to oppose Fas II-mediated adhesivity and allow defasciculation of otherwise adherent sets of motor axons. Genetic analysis supports this interpretation: when motoneurons have reduced quantities of Fas II, they have less need for beat function, as evidenced by suppression of the beat bypass phenotype double mutant embryos. A similar interaction is observed with the connectin gene, which encodes another cell adhesion protein expressed by a subset of motoneurons (Fambrough, 1996).

The beat phenotype and its apparent antiadhesive function sheds light on the motoneuron outgrowth mechanism. At choice points, motor axons selectively defasciculate from the common pathway and steer into their muscle target region. Several arguments suggest that defasciculation and steering are separable events. (1) The beat mutation affects all motoneuron defasciculation events, whereas mutations in genes involved in steering events might be expected to affect subsets of steering decisions. (2) The strong genetic interactions between beat and genes coding for cell adhesion proteins, both expressed in motoneurons, are not consistent with a role in steering, which is presumably a motoneuron-target muscle interaction. (3) Motoneurons would be expected to receive steering signals, not send them, and the secreted nature of Beat is not consistent with a receptor role. The common thread that runs through the phenotypes of the beat mutant and the misexpression experiments is that Beat interferes with cellular adhesive interactions. It seems more likely that Beat works through a receptor. One candidate receptor is the Fas II protein, with Beat binding to Fas II and blocking Fas II-mediated homophilic adhesion. Several lines of evidence argue against a direct interaction between Beat and Fas II: (1) ectopic expression of Beat on Fas II-expressing axon pathways in the CNS does not disrupt axon fasciculation in those pathways. If Beat functions as an antiadhesive for motor axons by directly binding to Fas II, then one would expect that such misexpression of Beat would lead to defasciculation of these Fas II-positive pathways. (2) Beat expression by muscles interferes with motoneuron-muscle interactions, and Fas II has not been strongly implicated in such interactions. Rather, the simplest interpretation of this result is that Beat interferes with specific interactions of motor axons with their target muscles, events that do not normally involve Fas II. (3) There exists a second gene, sidestep, that mutates to an identical phenotype as beat and thus is a candidate gene to encode the Beat receptor. (4) Cell coaggregation experiments using S2 cells expressing Fas II, and S2 cells expressing Beat tethered to the membrane with a GPI anchor, do not show any evidence of a binding interaction between Beat and Fas II (Fambrough, 1996).

Fasciclin II and Beaten path modulate intercellular adhesion in Drosophila larval visual organ development

The axonal adhesion molecule Fasciclin II and the secreted anti-adhesion molecule Beaten path carry out critical roles in the development of at least one set of sensory organs, the larval visual organs called Bolwig's organ (BOs). Instead of playing a role in axon defasciculation, in the development of Bolwig's organ, Beat appears to be involved in cell adhesion, a related phenomenon. In normal development, secretion of Beaten path by cells of the optic lobes (a portion of the brain that processes visual information sent from the eyes) allows the Fasciclin II-expressing larval visual organ cells to detach from the optic lobes as a cohesive cell cluster. Thus mechanisms guiding neuronal development may be shared between motoneurons and sensory organs, and adhesion and anti-adhesion are likely to be critical for early steps in the development of the larval visual system (Holmes, 1999).

The larval visual system (LVS) is a relatively simple sensory system composed of BOs: two clusters, each composed of 12 photoreceptor cells from which axons extend in a single fascicle to the brain. Development of the LVS differs in several significant ways from that of the motoneurons. Notably, the axon tracks of the Drosophila CNS and PNS develop through axonal outgrowth and migration, characterized by growth cone extension and selective fasciculation. In contrast, the initial connections of the pioneer axons of the LVS are established when the BOs are close to their target cells, the developing larval brain. Thus, it is neuronal cell bodies, the BOs, that move through a complex environment during LVS development, rather than the axons as in development of the CNS and PNS. It is currently unclear whether the anterior relocation of the BOs occurs as a passive consequence of head involution, or whether an active migratory process is involved. Preliminary analysis of two identified mutations, not enough anterior extension and out of place, which result in the failure of the BOs to attain their proper location, reveals that, despite failures in BO development, head involution appears to occur normally (Holmes, 1998a). The apparent genetic separation of BO movement and head involution suggests the processes may also be mechanistically separable. However, the developmental mechanism driving BO migration remains to be determined (Holmes, 1999).

The BOs arise from cells of the optic lobe placodes (OLs), ectodermal tissues that will go on to form the optic lobes of the adult fly. The BOs detach from the OLs and remain at the periphery of the embryo when the OLs invaginate (Green, 1993). As the BOs detach from the OLs, axons extend from the BOs and establish initial connections with the brain, which is adjacent to the OLs. Therefore, the axons need only navigate a relatively short distance to reach their targets in the brain at this early stage of development. The BO cells remain clustered for the remainder of development, during which they migrate anteriorly and the Bolwig's nerves (BNs) increase significantly in length. Therefore, proper development of the LVS requires that the BO precursors detach from the OLs, that the pioneer connections between the BOs and the brain be established and maintained, that axons properly fasciculate with the pioneer axons, and that the BOs migrate correctly to their final positions (Holmes, 1999).

By genetic analysis of LVS development, a mutant allele of beat has been identified that disrupts early stages of BO development (Holmes, 1998). Beat is expressed in a cluster of cells in the OLs before the BOs become distinct. In LVS development, as in motor neuron development, beat interacts genetically with fas II. A mutation in fas II also disrupts LVS development, resulting in a phenotype that is, in some ways, the opposite of that resulting from mutations in beat. Conversely, overexpression of fas II in the BO causes defects in BO development resembling those that result from mutations in beat. By directing expression of beat to either the BOs or the OLs, the LVS phenotype of beat mutations is reversed. Together, these results demonstrate that detachment of the BO precursors from the OL may involve interaction between Beat and Fas II. Thus, at least some of the mechanisms important for regulating intercellular interactions during motor axon development also function to guide development of sensory organs, suggesting that some of the general principles of neuronal development may overlap in these two different neuronal systems (Holmes, 1999).

Drosophila motor axons recognize and follow a Sidestep-labeled substrate pathway to reach their target fields

During development of the Drosophila nervous system, migrating motor axons contact and interact with different cell types before reaching their peripheral muscle fields. The axonal attractant Sidestep (Side) is expressed in most of these intermediate targets. This study shows that motor axons recognize and follow Side-expressing cell surfaces from the ventral nerve cord to their target region. Contact of motor axons with Side-expressing cells induces the down-regulation of Side. In the absence of Side, the interaction with intermediate targets is lost. Misexpression of Side in side mutants strongly attracts motor axons to ectopic sites. Evidence is provided that, on motor axons, Beaten path Ia (Beat) functions as a receptor or part of a receptor complex for Side. In beat mutants, motor axons no longer recognize Side-expressing cell surfaces. Furthermore, Beat interacts with Side both genetically and biochemically. These results suggest that the tracing of Side-labeled cell surfaces by Beat-expressing growth cones is a major principle of motor axon guidance in Drosophila (Siebert, 2009).

Wiring of the nervous system is a precisely controlled process that includes axon outgrowth, axon pathfinding, target recognition, and synapse formation. Outgrowing axons of the same subtype often join and migrate collectively toward their targets, a process called selective fasciculation. Upon arrival in the target region, a subset of growth cones must inevitably defasciculate from the main pathway to select a specific synaptic target. Advancing growth cones are believed to express appropriate receptors that detect and evaluate relevant guidance molecules presented by surrounding cells and tissues. Despite the discovery of several conserved regulators of axonal pathfinding, the spatiotemporal sequence of molecular events that steer growth cones is still unclear. How exactly do growth cones recognize and interpret guidance cues to make pathway decisions? Over the years, several models have been proposed to describe general features of axon navigation across species, including the chemoaffinity hypothesis, the guidepost cell hypothesis, the blueprint hypothesis, and the labeled pathways hypothesis (Siebert, 2009).

The labeled pathways hypothesis postulates that a small number of early differentiating neurons pioneer a stereotypic array of differentially labeled axonal pathways. These primary pathways are selectively recognized by subsequently developing growth cones. While follower axons migrate along a homogenous substrate toward their target region, pioneer axons have to interact with many different substrates, actively search for guidance cues, and interpret relevant guidance information. Pathfinding decisions for pioneer axons are greatly simplified if directional cues would be present in the substrate. However, the experimental evidence for the existence of such substrate pathways is scarce. It has been noted that projections of sensory axons in homeotic mutants of Drosophila involved the specific recognition of 'pre-existing trails.' Similarly, axons emanating from transplanted eyes or Mauthner neurons in Xenopus have been observed to project along defined tracks called 'substrate pathways'. Furthermore, it has been reported that chick optic axons are guided along a preformed adhesive pathway. Despite the appealing simplicity of substrate pathways, the molecular tags that provide directional information have not been identified (Siebert, 2009).

One possible molecule that could serve as a directional cue for Drosophila motor axons is the attractant Sidestep (Side), a transmembrane protein of the immunoglobulin superfamily that is dynamically expressed during embryogenesis but prominently enriched in muscles when motor axons arrive in their target areas. In side mutant embryos, motor axons fail to defasciculate at key choice points and hence bypass their muscle targets, suggesting that Side functions as a target-derived attractant. Interestingly, mutations in beaten path Ia (beat) lead to similar axon guidance phenotypes, and like Side, Beat has been shown to regulate axon defasciculation at choice points. Based on primary structure predictions, Beat is a secreted protein of the immunoglobulin superfamily that has been shown to function as an anti-adhesive factor on motor axons. Beat consists of two Ig domains and a Cysteine-rich C-terminal domain that shares similarities with cystine knots (Siebert, 2009).

This study shows that side encodes a candidate directional cue that steers Drosophila motor axons from the place of their birth to their peripheral targets. The spatiotemporal expression pattern of Side thus delineates a linear pathway for motor axons. High levels of Side are consistently found ahead of Beat-expressing growth cones. Contact with motor axons induces the down-regulation of Side and eliminates the pathway, preventing other outgrowing nerves from choosing the same route. Furthermore, biochemical and genetic evidence is provided that Beat functions as a membrane-associated receptor or part of a receptor complex for Side. Taken together, these findings support the concept of pre-existing pathways labeled with molecular markers that are recognized by a specialized subset of axons expressing the appropriate receptors (Siebert, 2009).

This work, provides evidence for a simple guidance mechanism that guides Drosophila motor axons to their target regions. During the period of axonal pathfinding, Side is dynamically expressed on different tissues but its temporal shift forms a spatial pattern as such that high levels are located ahead of motor axonal growth cones. Beat-expressing motor axons recognize and follow Side-labeled surfaces. In order to better visualize the complex spatial and temporal dynamics of axon guidance processes, a FasIIGFP^Mue397-based imaging assay was established that allowed analysis of the activity of growth cones in living wild-type and mutant embryos. These time-lapse observations revealed that in wild-type embryos, the ISN migrates continuously through the lateral body wall until it reaches a choice point near the dorsal trunk. As expected for migration along a substrate pathway, the ISN employs a relatively small growth cone that extends only few filopodia for steering. The observed growth rates are in good agreement with the growth rates of the RP2 axon in filleted, semiviable embryo preparations (Siebert, 2009).

The absence of the attractant Side strongly interferes with axonal growth and results in delayed arrival of the ISN in its dorsal target regions, frequently failing to innervate the dorsal-most muscles. The lack of Side-mediated attraction likely prevents the progression at a normal rate, causing the ISN to develop a complex growth cone that appears to actively search for guidance information. Since Side is a transmembrane protein, it is predicted to function as a contact attractant. The spatiotemporal expression pattern should therefore provide pathway information. The growth cone of the ISN follows Side-positive cell surfaces from its first emergence in the ventral nerve cord to its dorsal target region. Consequently, the tight association of motor axons and their substrates - e.g., sensory axons - is partially lost in side mutants. Since Side is expressed in all sensory clusters, it likely also prefigures the SNa and SNc pathways. Based on the positions of the segmental sensory clusters in the body wall, motor axons following Sidestep-labeled sensory axons are guided into the proximity of their target regions. At stage 15, motor axons reach the end of the sensory tracks and begin to defasciculate into the muscle fields that up-regulate Side at this developmental time point. In this respect, it is interesting to note that motor pathways terminate on the cell bodies of sensory neurons when deprived of their target muscles. The opposed migration of efferent motor axons and afferent sensory axons therefore provides a robust mechanism for the establishment of the basic neuromuscular connectivity pattern in Drosophila. In the brain of vertebrates, a similar mechanism controls the wiring of the thalamus and the cortex. Corticothalamic and thalamocortical fibers meet at a common intermediate target and continue to grow along each other in opposite directions (Siebert, 2009).

During its journey through the lateral body wall, the ISN completely ignores the nearby ventral and lateral muscle fields. At this stage, muscle fibers are not yet differentiated, and hence do not express endogenous Side. If these muscle precursors, however, are forced to express Side prematurely under control of Mef2-Gal4, the ISN diverts from its normal path and grows straight into the muscle field. The premature attraction drastically slows down the migration toward dorsal targets, leading to a permanent lack of NMJs on dorsal muscles. Thus, both gain and loss of Side cause migratory delays that result in the lack of NMJs on dorsal muscles but for different reasons. Ectopic expression of Side leads to excess attraction in wrong directions, whereas lack of Side leads to reduced attraction along the predestined path (Siebert, 2009).

Since mutations in both beat and side cause similar axon guidance phenotypes in embryos and innervation defects in larvae, it is assumed that the products of both genes might interact functionally. Several pieces of genetic evidence suggest that beat and side function in a common pathway: (1) the double mutant phenotype is similar to the respective single mutant phenotypes; (2) overexpression of Side in muscles of wild-type embryos leads to premature attraction of the ISN into ventral and lateral muscle fields. This gain-of-function phenotype is completely suppressed by coexpression of Beat. (3) Overexpression of Side in muscles of beat mutants renders motor axons unresponsive to ectopic Side; (4) endogenous Side is not down-regulated in beat mutants. In addition, Beat and Side interact in vitro. S2 cells expressing Beat and Side form large cell clusters in aggregation assays when individually transfected cell populations are mixed. Moreover, Beat coimmunoprecipitates with Side, supporting the idea that Beat interacts with Side or a Side-containing complex. The formation of cell aggregates further argues that Beat-Side interactions lead to the formation of heterophilic adhesion complexes (Siebert, 2009).

The spatiotemporal expression of Side appears to be strictly regulated. The levels of Side expression are highest in front of motor axonal growth cones. Side disappears from cell surfaces once these cells have been contacted by motor axons, indicating that motor nerves neutralize attractive surfaces and thereby disguise the path they are following. Motor nerve bundles that exit the CNS at a later time point are thus prevented from choosing the same route. In beat mutants, this regulatory mechanism appears to be nonfunctional. Side is constitutively expressed in peripheral nerves. Expression of exogenous Beat in post-mitotic neurons but not in muscles rescued the regulatory defects, suggesting that Beat induces the down-regulation of Side cell-autonomously. If Beat would be a secreted protein, one would expect that expression in muscles down-regulates Side on sensory axons. The secreted metalloprotease tolloid-related as well as the secreted TGF-β ligand Dawdle have been shown to rescue axonal guidance defects tissue independently. Biochemical data from transiently transfected S2 cells further support a cell-autonomous function for Beat. Beat was not secreted into the medium, and a fraction of it was associated with membranes. However, since Beat is not normally expressed in S2 cells the lack of a coreceptor or chaperone might prevent its secretion. Candidate coreceptors are the remaining members of the beat multigene family. Several family members are expressed in the ventral nerve cord, and might function together with Beat in the regulation of its subcellular localization and/or function. Further experiments will be necessary to determine the composition of Beat-containing complexes, and how they transduce guidance signals into the growth cone (Siebert, 2009).

If Side is an instructive signal for Beat-expressing motor axons it should be possible to redirect their paths in a side mutant background; i.e., in the absence of endogenous Side but in the presence of exogenous Side on a defined tissue. Regardless of whether Side is expressed in trachea, muscles, or hemocytes in a side mutant background, motor axons head toward the ectopic source of Side. The growth cones find and recognize Side-expressing cell surfaces and adopt their route accordingly. In the most extreme case, motor axons strongly interacted with hemocytes, which are highly motile cells. Side therefore potently controls the path of motor axons. Although Side-mediated attraction is likely not the only mechanism to reach the muscle targets, these results, together with the high penetrance of sidestep mutant phenotypes, suggest that it is one of the major mechanisms. In this respect, it is important to note that Side is not required for motor axon outgrowth per se, but rather for the specification of the growth direction. Based on the experimental evidence, a model is proposed for the navigation of motor axons from the ventral nerve cord to their target area in Drosophila. Beat-expressing motor axon fascicles recognize, extend on, and subsequently mask a pre-existing, Side-labeled substrate pathway that determines their growth direction. In the absence of the labeled pathway, in side mutants, axonal migration is delayed or growth cones head into aberrant directions. In either case they will miss their targets. In beat mutants, the pathway is constitutively labeled but cannot be recognized, leading to similar phenotypes. Since Beat and Side are conserved in insects, similar guidance principles might occur in all organisms, in which the peripheral nervous system develops from sensory organ precursors (Siebert, 2009).

Deconstruction of the beaten Path-Sidestep interaction network provides insights into neuromuscular system development

An 'interactome' screen of all Drosophila cell-surface and secreted proteins containing immunoglobulin superfamily (IgSF) domains discovered a network formed by paralogs of Beaten Path (Beat) and Sidestep (Side), a ligand-receptor pair that is central to motor axon guidance. This study describes a new method for interactome screening, the Bio-Plex Interactome Assay (BPIA), which allows identification of many interactions in a single sample. Using the BPIA, four more members of the Beat-Side network were 'deorphanized'. Interactions were confirmed using surface plasmon resonance. The expression patterns of beat and side genes suggest that Beats are neuronal receptors for Sides expressed on peripheral tissues. side-VI is expressed in muscle fibers targeted by the ISNb nerve, as well as at growth cone choice points and synaptic targets for the ISN and TN nerves. beat-V genes, encoding Side-VI receptors, are expressed in ISNb and ISN motor neurons (Li, 2017).

Protein-protein interactions (PPIs) control a vast array of processes in metazoans, ranging from signal transduction and gene regulation within cells to signaling between cells via cell surface and secreted proteins (CSSPs). The strength of PPIs varies widely, from high-affinity interactions that create stable protein complexes to weak transient interactions. Defining global PPI patterns ('interactomes') has been the focus of much recent research. Progress has been made in generating high-throughput protein interaction data for a variety of organisms, including S. cerevisiae, C. elegans and D. melanogaster. Methods used to create interactomes include affinity purification/mass spectrometry (AP-MS) and the yeast two-hybrid assay (Y2H) (Li, 2017).

It is estimated that up to one sixth of human genes encode CSSPs. CSSPs control signaling from the extracellular milieu to cells and the flow of information between cells. Due to their importance and accessibility, CSSPs are often the targets for therapeutic agents, including humanized monoclonal antibody drugs such as checkpoint inhibitors, the non-Hodgkin's lymphoma drug Rituxan, and the breast cancer drug Herceptin. However, the biochemical properties of many CSSP interactions prevent them from being detected by commonly used techniques employed in high-throughput PPI screens, and CSSPs are underrepresented in global interactomes. There are several reasons for this. First, these proteins are usually glycosylated and have disulfide bonds, so they need to be expressed in the extracellular compartment. CSSP interactions between monomers are also often weak, with KDs in the μM range, making them difficult to capture due to their short half-lives. Lastly, insoluble transmembrane domains on cell surface proteins preclude their purification with standard biochemical techniques, which makes them difficult to study using methods such as AP-MS (Li, 2017).

Despite these difficulties, recent advances have been made in the study of global CSSP interaction patterns. Interactions among cell-surface proteins (CSPs) often occur between clusters of proteins on cell surfaces, and avidity effects (stronger binding due to clustering) can make these cell-cell interactions stable even when monomers bind only weakly. To facilitate detection of interactions among CSSP extracellular domains (ECDs) in vitro, several groups have taken advantage of avidity effects by attaching ECDs to protein multimerization domains and expressing ECD fusions as soluble secreted proteins. These methods have been shown to be effective, allowing detection of interactions that otherwise would not have been observable (Li, 2017).

Özkan (2013) scaled up the avidity-based approach, developing a high-throughput ELISA-like screening method, the Extracellular Interactome Assay (ECIA). The ECIA was used to define interactions among 202 Drosophila CSSPs, comprising all CSSPs within three ECD families. These were the immunoglobulin superfamily (IgSF), fibronectin type III (FNIII) and leucine-rich repeat (LRR) families. The ECIA utilized dimers as 'bait' and pentamers as 'prey'. It detected 106 interactions, 83 of which were previously unknown (Li, 2017).

The most striking finding from the ECIA interactome was that a subfamily of 21 2-IgSF domain CSPs, the Dprs, selectively interacts with a subfamily of 9 3-IgSF domain CSPs, the DIPs, forming a network called the 'Dpr-ome' (Özkan, 2013). Each Dpr and DIP that has been examined is expressed by a small and unique subset of neurons at each stage of development. One Dpr-DIP pair is required for normal synaptogenesis and influences neuronal cell fate. In the pupal optic lobe, neurons expressing a Dpr are often presynaptic to neurons expressing a DIP to which that Dpr binds in vitro. The Dpr-ome initially defined by the global interactome contained several 'orphans', proteins with no binding partner (Özkan, 2013). By expressing new versions of Dprs and DIPs, including chimeras, and using these to conduct a 'mini-interactome' analysis of the Dpr-ome, it was possible to find partners for all but one orphan. That protein, Dpr18, has changes to conserved binding interface residues and may lack the capacity to bind to any DIPs (Li, 2017).

The ECIA also identified a second IgSF network, formed among members of the Beaten Path (Beat) and Sidestep (Side) protein subfamilies. Beat-Ia and Side were identified by genetic screens for motor axon defects, and were later shown to have a ligand-receptor relationship. They control the projection of motor axons to muscle targets. Beat-Ia is expressed on motor axons, where it binds to Side, which is expressed on muscles. This binding causes motor axons to decrease their adhesion to each other, allowing them to leave their bundles and turn onto the muscle fibers. beat-Ia and side have strong motor axon phenotypes. In the absence of either protein, motor axons often remain in their fascicles and never leave to arborize on their target muscles (Li, 2017).

The ECIA detected the known Beat-Ia::Side interaction, and also uncovered other interactions between members of the Beat and Side subfamilies. Seven of the 14 Beats were found to bind to four of the eight Sides. The remaining Beats and Sides were still orphans with no binding partners in the other subfamily. The functions of the newly defined interactions between Beats and Sides were unknown. Most beat genes are expressed in embryonic neurons. Some Beats were genetically characterized using deletion mutations and RNAi, but loss of these Beats did not cause strong motor axon phenotypes. None of the other Side subfamily members had been examined (Li, 2017).

This paper describes the development of a new method for interactome screening, which is called the BPIA (Bio-Plex Interactome Assay). This method uses the 'Bio-Plex' system, which employs Luminex xMAP technology. This method detects binding of a prey protein to many bait proteins, each conjugated to a bead of a different color, in each assay well. For the ECIA, the number of assays required for the interactome screen was the square of the number of proteins examined, while with the Bio-Plex the number of assays could be equal to the number of proteins. In principle, then, the Bio-Plex might greatly speed up interactome screening, and might also be more sensitive, since the available signal-to-background ratio is much greater for the Bio-Plex than for the ECIA. As a test of the method, a Bio-Plex 200 was used to do a mini-interactome screen of the Beat-Side network. Based on the the fact that the Dprs and DIPs that were initially orphans were later shown to have binding partners, it was hypothesized that some of the orphan Beats should have Side partners, and vice versa. Consistent with this hypothesis, it was possible to deorphanize two more Beats and two Sides using the BPIA (Li, 2017).

To further understanding of Beat and Side function during embryonic development, this study characterized expression of several Beats and Sides, focusing primarily on Side-VI and the three Beat-Vs, which were the strongest interactors in both the ECIA and BPIA screens. The three beat-Vs exhibit differential expression in identified motor neurons, while side-VI is expressed at motor axon choice points and in a subset of target muscle fibers (Li, 2017).

In principle, the Bio-Plex system can allow 500 unique protein-protein (bait-prey) interaction pairs to be analyzed in a single well. In this method, capture of proteins from media with streptavidin-coupled beads bypasses the purification step for bait proteins. The assay is also compatible with the use of unpurified prey proteins, thereby reducing the workload for multiplexed screenings. The small size of the beads, the ability to probe multiple interactions simultaneously, and the small volume of the binding reactions all help reduce the amount of protein and reagents needed for the assay. It was possible to produce enough bait and prey proteins for the mini-interactome described in this study (a 23 x 23 matrix) with a single 10 cm dish transfection per protein (Li, 2017).

As a test of the system, the BPIA was used to examine interactions between the Drosophila Beat and Side IgSF protein subfamilies. Beat-Ia is a receptor on motor growth cones that recognizes Side expressed on muscles, and in the absence of Beat-Ia or Side motor axons fail to leave their bundles and arborize on their muscle targets. There are 14 Beat subfamily members and 8 Side subfamily members, but all of these proteins except Beat-Ia and Side itself were orphans until the global IgSF interactome revealed interactions between six other Beats and three Sides (Li, 2017).

In the Dpr-ome, the other IgSF network uncovered by the interactome, every Dpr protein likely to be capable of binding has an interaction partner in the DIP subfamily. Based on this, it is predicted that there should be additional interactions to be discovered within the Beat-Side subfamily network. Using the BPIA, three new interactions were found: Beat-VI::Side-II, Beat-Ic::Side-III, and Beat-Ic::Side. These results suggest that the BPIA is more sensitive than the ECIA. Like the ECIA, the BPIA should be able to find new receptor-ligand interactions even if proteins not previously known to have any interactions were tested. Of course, for both assays any candidate receptor-ligand pairs need to be confirmed as genuine using other methods. For the Beat-Side network, all three new interactions found by the BPIA, as well as the interactions between the three Beat-Vs and Side-VI found by the ECIA, were verified by SPR. This study also demonstrated that Beat-Vs and Side-VI interact using cell-based binding assays and binding to live-dissected embryos (Li, 2017).

There are still five Beats and two Sides that remain orphans. Since the structure of Beat-Side complexes is unknown, it cannot be determined whether these Beats and Sides are likely to be able to bind, but it is speculated that at least the three Beat-IIIs are likely to have Side partners. The Beat-II and Beat-V clusters each interact with a single Side partner, and perhaps the Beat-IIIs interact with one of the two orphan Sides. It is possible that changes in methodology, such as using more highly multimerized preys and/or baits, could increase sensitivity and allow detection of additional interactions (Li, 2017).

The expression patterns of side and beat genes were examined in order to obtain insights into their possible functions. Most sides are expressed in cells in the periphery as well as in the CNS, while most beats are expressed only by CNS neurons, including motor neurons. Beat-Ia::Side interactions are required for normal motor axon guidance, and highly penetrant motor axon defects in which muscles remain uninnervated are observed in mutants lacking either protein. By contrast, partial loss of function of beat-Ib, beat-Ic, both beat-IIs, or beat-VI causes motor axon defects with less than 20% penetrance. Genetic redundancy is a common theme in motor axon guidance , so it is not surprising that only low-penetrance defects are observed when Beat paralogs are not expressed. Given that Beat-Ia and Side both interact with other partners, it is perhaps remarkable that beat-1a and side have such strong phenotypes as single mutants (Li, 2017).

Beat-V and Side-VI also have redundant functions in motor axon guidance. side-VI is expressed in motor axon targets, including muscles 12 and 13 and interacts with the three Beat-Vs, at least two of which are expressed in subsets of motor neurons. Beat-V::Side-VI interactions produced the strongest signals in both the ECIA and BPIA. Sow-penetrance (~1/5 of stage 17 embryonic hemisegments affected) muscle 12 innervation defects were observed in side-VI insertion mutants or in deletion mutants lacking all three beat-V genes. There were also low-penetrance ISN guidance defects in both mutants. The fact that most muscle 12 s are innervated normally in beat-V or side-VI mutants indicates that, while Beat-V::Side-VI interactions may contribute to correct targeting of the RP5 axon to muscle 12, other cues must also be involved. Muscles 12 and/or 13 also express Wnt-4 (a repulsive ligand) and the LRR protein Capricious (Caps; probably an adhesion molecule), and low-penetrance RP5 targeting defects are observed in Wnt-4 and caps mutants. Perhaps muscle 12 is distinguished from other nearby muscles by a set of partially redundant cues, so that strong targeting phenotypes are not observed in any single mutant (Li, 2017).

Although Beat and Side paralogs may not be central to motor axon guidance, their expression patterns suggest that they could be important for determining synaptic connections within the CNS. side-VIII, encoding an orphan Side, is expressed in a small subset of embryonic CNS neurons. In the optic lobe of the pupal brain, an RNAseq analysis of two photoreceptors (R7 and R8) and five types of lamina neurons (L1-L5) revealed that beats and sides have highly specific expression patterns. For example, beat-VII is specific to L2, beat-VI to L5, beat-IIa to L3 (with lower levels in L4), and beat-IIIc to R8, being expressed at much higher levels in those cells relative to all other cells. side and side-III are specific to L3, side-II is specific to L1, side-IV is specific to L2, and side-V is specific to L5. Each of the L neuron types as well as R7 and R8 synapse with different sets of neurons in the medulla, a ten-layered structure that processes visual information from the retina and lamina. It has been observed that R and L neurons expressing specific Dprs often form synapses on medulla neurons expressing DIPs to which those Dprs bind in vitro. In a similar manner, perhaps some of the medulla neurons that are postsynaptic to L or R neurons expressing specific Sides or Beats express their in vitro binding partners, and these Beat-Side interactions might be important for synapse formation or maintenance (Li, 2017).

DEVELOPMENTAL BIOLOGY

Embryonic

Nonradioactive antisense RNA probes were generated to examine the distribution pattern of BEAT mRNA during embryonic development. No maternally loaded BEAT transcript is present in early embryos. Zygotic expression of beat begins in a subset of CNS cells at stage 12/5. While these particular cells could not be unambiguously identified at this stage, based on their positions it is likely that these are early born motoneurons. The number of expressing cells increases during germ band retraction, and by early stage 13 the expression level reaches its maximun. During this period the motoneuron growth cones exit the CNS and begin extending into the periphery. As development proceeds, the number of staining cells appears to remain constant, although cell body positions shift due to cell migrations. Counterstaining with antibodies such as MAb 22C10 (which stains a subset of CNS neurons including some motoneurons: see Futsch) and MAb BP102 (which stains all CNS axon tracts) indicates that all the identified motoneurons express BEAT mRNA. In addition, the number and positions of the other beat-expressing cells are consistent with these cells being the remaining motoneurons. Thus, beat is probably expressed by all motoneurons and exclusively by motoneurons. Expression persists at a high level through stage 14, the period during which the major peripheral motor nerve branches form. After stage 14 the expression level drops to a lower level, then remains constant through stage 17. A small number of cells of unknown function in the embryonic brain also express beat. beat is not expressed in imaginal discs. Both beat² and beat³ express mRNA normally. Among the embryonic motoneurons, BEAT mRNA is expressed at different levels. The RP1 and RP3 motoneurons, whose axons leave the ISN in the SNb, express beat at very high levels. In contrast, the aCC motoneuron, whose axon pioneers the full length of the ISN, expresses significantly lower levels (Fambrough, 1996).

To examine the distribution of Beat protein, a Glutathione-S-Transferase fusion protein was made with a fragment of the Beat protein (amino acids 245-318). Anti-Beat sera from mice recognizes a motoneuron-specific antigen in wild-type embryos but not in beat Df embryos or in beat³ mutant embryos. Beat is found around axons and growth cones, especially in choice point regions during the period of branch formation; the appearance of Beat expression is fuzzy in these localized regions, suggesting that Beat is secreted at choice points where motoneurons diverge. In the beat² mutant there is an abnormal accumulation of protein in the cell body, but some protein is found in axons and growth cones. The anti-Beat sera recognizes a single band of about 43 kDa on a Western blot of total protein from 10-15 hr embryos. This is in agreement with the predicted size of 44 kDa for the cleaved, secreted protein. (Fambrough, 1996).

The pattern of BEAT mRNA expression was examined in wild-type embryos. Fambrough (1996) noted that, in addition to being expressed in the motoneurons, beat is expressed in unidentified cells in the embryonic brain. In the brain, beat expression is first detected at stage 12 in a small cluster of cells located in the vicinity of the optic lobes (OLs). To determine the identity of the beat-expressing cells, the simultaneous expression of BEAT mRNA and glass-lacZ was examined. BEAT mRNA is never detected in Bolwig's organs (BOs) or Bolwig's nerves once these structures become distinguishable from the OLs. In stage-13 embryos, the pattern of beat expression, although distinct from the BO, is close to the developing BOs, consistent with its being expressed in the OLs. Expression of BEAT mRNA in the brain persists through stage 16, where beat is clearly expressed in the embryonic brain. Thus, beat is expressed in a subset of cells in the OL that are close to, but that do not overlap with, the BO (Holmes, 1998b).

EFFECTS OF MUTATION

In each abdominal hemisegment of the Drosophila embryo, an array of 30 muscle fibers is innervated by about 34 motoneurons in a highly stereotyped and cell-specific fashion. To begin to elucidate the molecular basis of neural specificity in this system, a genetic screen was conducted for mutations affecting neuromuscular connectivity. The focus was on 5 genes required for specific aspects of pathway (beaten path, stranded, and short stop) and target (walkabout and clueless) recognition. The different classes of mutant phenotypes suggest that neural specificity is controlled by a hierarchy of molecular mechanisms: motoneurons are guided toward the correct region of mesoderm, in many cases navigating a series of choice points along the way; they then display an affinity for a particular domain of neighboring muscles, and finally, they recognize their specific muscle target from within this domain. The intersegmental nerve (ISN) growth cones aften stop at or near the first choice point in beat embryos. ISN growth cones usually stop near the second dorsal choice point in stranded mutants, often just beyond contacts with one of the putative mesodermal guidepost cells: PT3. In short stop mutants, ISN growth cones stop at the second choice point, choosing to extend beyond PT2, another mesodermal guidepost cell, to explore dorsal sensory neurons rather than continuig to PT1 and the dorsal muscles. The location at which Segmental nerve b (SNb) growth curves stop in beat mutants approximately corresponds to the first major choice point in SNb pathfinding, where growth cones make extensive contacts with the surfaces of muscles 28 and 14 as they split away from the ISN and SNa. short stop mutant growth cones stop just beyond this first choice point at contacts with muscle 14 where the SNb must shift trajectory to grow between the ventral longitudinal muscles (internal) and the more external muscles 14 and 30. The location where SNb growth cones stop in stranded mutants corresponds to the second major choice point in SNb pathfinding. SNb growth cones in walkabout and clueless mutants extend into the ventral domain but fail to correctly recognize the appropriate target muscles, and instead arborize over a range of adjacent ventral muscle surfaces (Van Vactor, 1993).

The common element in all of these beat phenotypes is that axons remain adhered to other cell surfaces from which they should have disengaged. The lack-of-defasciculation phenotype is remarkably similar to that observed when the cell adhesion molecule Fas II is overexpressed on motor axons (Lin, 1994). This suggests that beat may provide an anti-adhesive function, opposing the adhesion promoted by such cell adhesion molecules as Fas II. In beat mutant embryos, the ultimate destination of most of the motor axons that have failed to defasciculate is not known, but in the case of the SNb they appear to continue along the ISN, often beyond the ventral muscle domain. SNb motor axons can be seen making ectopic contact with muscle 4, a target normally innervated by the ISN. The SNb axons can also make contact with muscles 12 and 13 by lateral sprouting from the common pathway. The penetrance of these defects is not 100%, even in beat null embryos. The percentage of segments in which the SNb fails to diverge from the ISN either completely (called full bypass), or partially (partial bypass) was quantitated. In deficiency embryos, transheterozygous for two overlapping deficiencies that completely remove the beat gene, 68% of segments show full bypass, and 28% partial bypass. In a control deficiency overlap, which removes an adjacent gene, Bic-C, but not beat, only 5% of segments show full bypass and 2% partial bypass. Transheterozygotes between two beat EMS-induced alleles show 38% full bypass, 28% partial bypass. These numbers may suggest that the EMS alleles are hypomorphic for beat. Alternatively, the presence of a large hemizygous region in the deficiency overlap may enhance the null condition and account for the low level of bypass phenotypes seen in the beat⁺ deficiency overlap (Fambrough, 1996).

To test the hypothesis that beat encodes an anti-adhesive function, genetic interactions were sought between beat and genes encoding CAMs expressed on motor axons. It was reasoned that if beat works to oppose adhesion, then a reduction in the amount of adhesion should at least partly compensate for the loss of beat. Thus, double mutants between beat and CAM genes should have a phenotype less severe than that of beat alone. To this end, flies were created carrying various alleles of the Fasciclin II (FasII) gene and the beat gene, being careful to keep the genetic background as nearly identical as possible, and the divergence of the SNb from the ISN was quantified using an antibody raised against the LBL protein to visualize the motoneurons. A series of FasII mutant alleles expressing various levels of the wild-type Fasciclin II protein was used: FasII^e93 expresses 100% of the wild-type level; FasII^e86 expresses about 50%, and FasII^e76 expresses about 10%. None of these FasII alleles has a major affect on SNb divergence. These FasII mutant alleles and wild-type control (FasII^e93) were generated by imprecise and precise excisions, respectively, of a single P element insert, and thus each genetic interaction is scored in an otherwise identical genetic background. While FasII^e93; beat² flies have a strong beat phenotype with 90% of segments having a divergence defect, FasII^e76; beat² flies show a significantly more wild-type phenotype, with only 41% of segments having a bypass phenotype. FasII^e86; beat² flies show a slight but not significant decrease in defects, with 86% of segments having a bypass phenotype, and the defects observed tend to be weaker. This interaction with FasII is not allele specific, since beat³ is also significantly suppressed by FasII^e76 (Fambrough, 1996).

Fas II is expressed on the surface of all motoneurons. Would beat have a similar genetic interaction with a gene encoding a different CAM that is normally expressed on the surface of only a subset of fasciculating motor axons? The connectin (conn) gene encodes a leucine-rich-repeat homophilic cell adhesion molecule expressed by SNa and SNc motoneurons, a few ISN motoneurons (that innervate muscles neighboring SNa targets), but no SNb motoneurons. The conn^Fvex238 allele was used: it expresses about 5% of the normal level of mRNA. conn^Fvex238 has no major effect on divergence of the SNc from the SNa. The conn^Fvex238 allele was combined with the beat² and beat³ alleles and the percentage of segments in which SNc fail to diverge from SNa was scored. As a control, the percentage of segments in which SNb fails to diverge from ISN was also scored. In these segments, conn is not expressed and thus is unlikely to be involved. While conn^Fvex238 does not significantly affect SNb divergence from ISN, it does very significantly affect the divergence of the SNc from the SNa. For beat², the percentage of mutant segments drops from 82% to 48% in a conn^Fvex238 background, while for beat³ the percentage of mutant segments drops from 77% to 44% in a conn^Fvex238 background. This result indicates that beat interacts with multiple cell adhesion systems in a way consistent with it encoding an antiadhesive function (Fambrough, 1996).

An examination was performed to see whether ectopic expression of Beat disrupts interneuron and sensory neuron pathways, paying special attention to the Fas II-positive CNS longitudinal pathways. No defects are observed in these or other CNS or sensory pathways. These results suggest that Beat protein cannot lead to the defasciculation of all axon fascicles that express Fas II, and that the beat function appears to be highly specific to motoneurons (Fambrough, 1996).

To further assess beat function, Beat was ectopically expressed in the mesoderm. Such flies display major motoneuron outgrowth defects. Staining with anti-Beat sera shows that these embryos express high levels of Beat protein in muscles. Several types of defects are observed in such embryos. SNb axons can diverge from the common pathway, but often stall at the edge of muscles 28 and 14. The SNb can also bypass the ventral choice point. The distal ISN shows abnormal exploration of the dorsal muscle region, sometimes crossing the segment boundary and fasciculating with the ISN from the adjacent segment. Defects are also observed in the other nerve branches. These phenotypes suggest that expression of Beat on muscles disrupts motoneuron-muscle interactions (Fambrough, 1996).

In beat mutant embryos the cells of the Bolwig's organs (BOs) form, but the larval visual system (LVS) morphology is severely disrupted and increased numbers of photoreceptor cells are apparent from the earliest stages of LVS development. In stage-13 embryos, in addition to photoreceptor cells at the normal location of the BO clusters, extra photoreceptor cells are dispersed between the two clusters. Axonal projections from these cells suggest that they are neuronal cells. The optic lobes (OLs) begin to invaginate during stage 13, and in beat mutants the abnormal glass-expressing cells (a lacZ reporter gene driven by the glass regulatory elements was used to identify BOs) are often located near the invaginating optic lobes. Precise quantitation of the number of extra photoreceptor cells in beat mutants is difficult with currently available reagents. However, using an antibody to Kruppel, which stains the nuclei of the BOs, it is concluded that there are more photoreceptor cells in beat mutants than in wild-type. Frequently in beat mutant embryos, a BO fails to separate properly, suggesting that mutations in beat may affect this early step of LVS development (Holmes, 1999).

The proposed role of Beat in motoneuron development is to promote defasciculation of axons expressing Fas II (Fambrough, 1996). Because Fas II is expressed in the BOs, Bolwig's nerves (BNs) and OLs, it was important to determine whether Fas II is also required for their development. Fas II expression in the OL begins at least as early as stage 12 before the BOs become morphologically distinct. Fas II expression in the BOs, the BNs and the optic lobes persists for the remainder of larval visual system development. Embryos hemizygous for the fas II ^e76 mutation, which is reported to result in greater than 90% reduction in Fas II protein expression, were examined. In a majority of fas II ^e76 embryos, the LVS is morphologically defective, although the defects are more subtle than those seen in beat mutant embryos. Disruption of BO development can be seen as early as stage 13 in fas II ^e76 embryos, in which unclustered photoreceptor cells are located near the normal BOs. Defects seen at later stages in development of fas II ^e76 mutants include BOs that are smaller than normal, which could result from fewer than 12 cells in the BO cluster. In a subset of later stage fas II ^e76 embryos the BOs do not migrate properly, and nodules are present along the nerve in a subset of mutant embryos. Head involution appears normal in all fas II ^e76 mutants. It is noteworthy that mutations in fas II ^e76 and beat disrupt BO relocation without apparently affecting head involution. The defects in BO migration could be due to disruption of the coordination of BO movement and head involution, or they could reflect a role for these molecules in BO migration. Especially intriguing are embryos in which a small cluster of photoreceptor cells has continued to migrate beyond the point at which the majority is located (Holmes, 1999).

The observation that mutations in beat and fas II disrupt development of the LVS in opposing directions raises the possibility that beat and fas II interact genetically. Considering that Beat is an anti-adhesion molecule expressed in the OLs, and Fas II is an adhesion molecule expressed in the BOs, BNs and OLs, it was reasoned that interactions between Beat and Fas II may facilitate development of the BOs. Interaction between Beat and Fas II in LVS development would be consistent with their roles in motoneuron development (Fambrough, 1996). If a role for Beat in the OLs is to downregulate Fas II-mediated adhesion, reducing the levels of Fas II might suppress the severity of the beat mutant phenotype. To test these possibilities, a stock homozygous for both fas II and beat mutants was constructed that also carries the glass-lacZ reporter. Using anti-beta-galactosidase to examine the larval visual system (LVS) it was found that in 67% of the double-mutant embryos the BOs and BNs develop normally. In the 33% of double-mutant embryos in which development is abnormal, the defects most closely resemble those seen in fas II single mutants: the BOs are smaller than wild-type BOs or fail to migrate properly. In double-mutant embryos in which the beat mutant phenotype is suppressed, development appears to occur normally from the outset and continues in a normal manner through later stages. Ectopic photoreceptor cells are never detected in double-mutant embryos, which is a feature of beat mutant embryos. Greater than 65% of embryos homozygous for single mutations in either beat or fas II exhibit defects in LVS development. In contrast, in the majority of fas II;beat double-mutant embryos the LVS develops normally, and the defects observed in the double mutants are relatively minor. Therefore, the double mutant combination suppresses the phenotypes of the single mutants. This striking result supports a model in which relative levels of expression of the adhesion molecule and the anti-adhesion molecule are important for normal development: concurrent reduction of both restores normal LVS development in the double mutants (Holmes, 1999).

To further explore the importance of relative levels of Beat and Fas II, an experiment was carried out to determined whether elevating expression of Fas II in the BOs would affect LVS development. Increasing expression of Fas II in the BO prescursor cells could strengthen adhesion between the BOs and OLs. This could in turn result in interference with the coordinated separation of the BOs from the OLs. In the majority of stage-13 embryos in which Fas II is overexpressed, extra photoreceptor cells are seen distributed between the two normal BO clusters. These embryos very closely resemble beat minus embryos of the same stage. Despite the similarity of defects in early stages of LVS development, later stages are less affected by Fas II overexpression, when compared to development in the absence of Beat. Overexpression of Fas II appears to have only minimal effects on anterior migration of the BOs. The BOs are located in the normal position in greater than 90% of stage-15 Fas II overexpressing embryos. Subtle disruptions in the morphology of the BOs and BNs do result from overexpression of Fas II, but these are neither as frequent nor as severe as defects caused by mutations in beat. The similarity of defects in early LVS development resulting from reduced expression of Beat and overexpression of Fas II bolsters the proposal that interaction between Beat and FasII is important for separation of the BOs from the OLs. That the effects of these two alterations in adhesion are different in later stages of LVS development may reflect differing roles of the molecules at different times (Holmes, 1990).

Axon pathfinding and target choice are governed by cell type-specific responses to external cues. In the Drosophila embryo, motorneurons with targets in the dorsal muscle field express the homeobox gene even-skipped and this expression is necessary and sufficient to direct motor axons into the dorsal muscle field. Motorneurons projecting to ventral targets express the LIM homeobox gene islet, which is sufficient to direct axons to the ventral muscle field (Thor, 1997). Thus, even-skipped complements the function of islet, and together these two genes constitute a bimodal switch regulating axonal growth and directing motor axons to ventral or dorsal regions of the muscle field (Landgraf, 1999).

How do transcriptional regulators such as Eve and Islet direct patterns of axonal growth? The phenotype observed in ectopic Eve embryos (fusion of the main nerve trunks and failure of secondary nerve branching) is similar to, though more severe than, phenotypes produced in embryos where general interaxonal adhesion is increased either by overexpression of the homophilic CAM Fas II or by removal of its antagonist Beaten path (Beat). In such embryos, the two main nerve trunks (SN and ISN) form, but secondary nerves fail to branch off. A test was carried out to see if ectopic Eve increases interaxonal adhesion by downregulating the antiadhesive neural CAM antagonist Beat. No significant changes were seen in the overall pattern or relative levels of BEAT mRNA expression in ectopic Eve embryos. The expression patterns of the major neural CAMs Fas II, Fas III, and Connectin were examined in ectopic Eve or eve mutant embryos, but no changes in their expression patterns were detected. To test if Eve might regulate the expression of another (as yet unidentified) neural CAM, it was reasoned that beat might antagonize interaxonal adhesion mediated by such a CAM, just as beat antagonizes adhesion mediated by Fas II and Connectin. When Eve and beat are ectopically co-expressed, the ectopic Eve phenotype of excessive axonal fasciculation is partially rescued. Thus, Eve directs motor axons to the dorsal region of the muscle field by suppressing expression of the ventrally directing islet gene and by promoting adhesion to the ISN (Landgraf, 1999).

A genetic screen led to the identification of the beaten path (beat Ia) gene in Drosophila. Beat Ia contains two immunoglobulin (Ig) domains and appears to function as an anti-adhesive factor secreted by specific growth cones to promote axon defasciculation. A family of 14 beat-like genes has been identified in Drosophila. In contrast to beat Ia, four novel Beat-family genes encode membrane-bound proteins. Moreover, mutations in each gene lead to much more subtle guidance phenotypes than observed in beat Ia. Genetic interactions between beat Ic and beat Ia reveal complementary functions. The data suggest a model whereby Beat Ic (and perhaps other membrane-bound family members) functions in a pro-adhesive fashion to regulate fasciculation, while Beat Ia (the original secreted Beat) functions in an anti-adhesive fashion to regulate defasciculation (Pipes, 2001).

The degree of primary amino acid sequence identity among the Beats ranges from 24% to 72% across the Ig domains. In general, homology is highest at the N-termini and drops significantly towards the C-termini. A phylogenetic tree constructed from CLUSTALW alignments of the homologous regions of the Beats reliably divides the family into seven distinct classes. The original beat was renamed beat Ia for clarity (Pipes, 2001).

Eleven of the Beats reside in four chromosomal clusters of two to three genes each. Without exception, Beats that are more closely related to each other at the level of primary amino acid sequence are also located in the same cluster. Therefore, it is likely that Beats located in a local cluster are the products of local gene duplications. All of the Beats are expressed in the adult head: fragments of each of them can be amplified by PCR out of a head-specific cDNA library. Therefore, none of the Beats is a pseudogene (Pipes, 2001).

Full-length cDNAs have been isolated for beat Ib, beat Ic, beat IIa and beat VI. The predicted protein structures of these new Beats are significantly different from that of Beat Ia, which is predicted to be a dimerized, secreted molecule: these new Beats are predicted to be membrane-anchored (Pipes, 2001).

In situ hybridization of embryos shows that most of the Beats are transcribed in subsets of neurons in the embryonic CNS. beat IIIa, beat IIIb, beat IV and beat VII do not show detectable embryonic expression. Interestingly, the class II Beats, beat IIa and beat IIb, are also expressed in a subset of the ventral oblique muscles, making them the only Beats with detectable embryonic expression in a non-neural tissue. Of course, these muscles are targets for specific motor axons. The class V Beats all appear to be transcribed in large, ventral neurons in the CNS, possibly the VUM neurons. Many of the Beats that are expressed in the embryo show highly restricted and specific expression patterns; none of them is expressed as broadly as beat Ia, which is expressed in all motoneurons during the stages of axon outgrowth (Pipes, 2001).

In further contrast to beat Ia, loss-of-function mutations in several of the other Beats have only partially expressive defects in axon guidance. The extensive genetic reagents available to study genes in the Adh region of chromosome arm 2L, where the class I Beats reside, were used to generate animals that lack beat Ib and beat Ic. Extensive analyses of animals lacking beat Ib fail to reveal any significant defect in embryonic or larval neural development, and altering the levels of Beat Ib does not appear to modify the severity of other axonal phenotypes. beat Ic mutants display a strong phenotype in the development of the transverse nerve and a mild phenotype in the development of the ISNb (Pipes, 2001).

Double-stranded (ds) RNA interference was used to examine the loss-of-function phenotypes of several other Beats. Embryos injected with beat VI ds RNA have abnormal ISNb development in ~13% of hemisegments, which is similar to the 10% of abnormal hemisegments observed in buffer-injected control embryos. Similarly, only 17% of hemisegments in embryos co-injected with beat IIa and beat IIb ds RNAs display defective ISNb phenotypes. None of these treatments caused significant defects in the development of any other major motor nerve (Pipes, 2001).

The lack of a strong phenotype in single-gene mutants of genes that belong to large families is hardly surprising. There may well be functional redundancy between the members of the beat gene family, or it may be that each of the new Beats plays only a small role in generating axonal specificity, secondary to stronger mechanisms such as Fas II-mediated axon adhesion or Semaphorin-mediated repulsion. If the latter is true, an effect of the loss of one of the new Beats might be seen when examining a sensitized genetic background (Pipes, 2001).

Given that the new Beats, as predicted membrane-anchored IgCAMs, most closely resemble other proteins that function as axonal CAMs, the beat Ia mutant was used as a sensitized background to test beat Ic for adhesive activity in vivo. If Beat Ic functions in an adhesive fashion, then the loss of beat Ic should suppress the axonal hyper-adhesivity phenotype in beat Ia mutants. Motor axon defasciculation phenotypes were examined in beat Ic, beat Ia double mutants. Using a chromosomal deficiency overlap that deletes both beat Ic and beat Ia, it was found that loss of beat Ic does indeed improve the beat Ia mutant phenotype, from only 42% ISNb defasciculation in a beat Ia null background to 63% ISNb defasciculation in a beat Ic, beat Ia double mutant. Furthermore, the ISNd defasciculation phenotype in beat Ia mutants is improved from 14% to 32% in this double mutant combination. Significantly, other nerves that exhibit defasciculation errors in the absence of Beat Ia do not appear to be suppressed in the same way (Pipes, 2001).

In wild type, beat Ic mRNA is first transcribed in the early (stage 9) embryo in a subset of cells that invaginate to form the cephalic furrow and germ band. This expression ceases by the end of germ band retraction, and no beat Ic transcript is visible in embryos of stage 12/1. Transcription of beat Ic commences again in a small subset of cells at stage 13, shortly after the onset of beat Ia expression. It is impossible to identify these cells definitively at this stage because of cell movements, but some of the beat Ic-positive cells at this stage also stain with mAb 1D4 (anti-Fas II), which labels all motoneurons. The number of cells expressing beat Ic mRNA increases gradually but steadily to include a subset of the identified motoneurons. This restricted expression pattern is in contrast to that of beat Ia, which appears to be expressed in all embryonic motoneurons (Pipes, 2001).

The elav-GAL4 enhancer, which drives expression in all neurons, in conjunction with a UAS-beat Ic transgene, was used to alter the restricted pattern of Beat Ic expression found in wild type. Using two copies of driver and two copies of reporter to misexpress Beat Ic in all neurons causes defects in ISNb development. Of the ISNb axons that show a phenotype in this treatment, many can be seen stalling in tangles as they enter their target muscle field. Some ISNb motor axons appear to be more sensitive to Beat Ic levels than others. The RP3 motor axon stalls 23% of the time in this treatment, while RP1 only stalls 5% of the time. This phenotype appears to be dose sensitive, since reducing the levels of misexpressed Beat Ic by using only one copy of the elav-GAL4 driver reduces the frequency of these errors from 23% to 11% (Pipes, 2001).

In normal development, the transverse nerve forms from the fasciculation of the axons of a sensory nerve (the lateral bipolar dendrite or LBD neuron) and a motor nerve (the transverse motor nerve or TMN). These two growth cones grow along the process of a specialized glial cell called the dorsal median cell (DMC), thought to be crucial for the proper development of the transverse nerve. At mid- to late-stage 15, these two growth cones find each other on the surfaces of the ventral interior muscles, fasciculate, and then both continue growing – the sensory axon following the efferent motor axon projection into the CNS and the motor axon following the LBD dorsal projection to a dorsal muscle target (Pipes, 2001).

Embryos that lack beat Ic often display errors in the development of the TN. Whereas in wild type, the TMN and LBD axons are tightly fasciculated by stage late 16, beat Ic mutants of the same age are often found to be bifurcated and exploring the surface of the ventral muscles. This effect appears to be due to the loss of beat Ic in the TMN, since beat Ic mRNA is not detected in the LBD cell body, but can be seen in a cell located just outside the longitudinal axon tracts, below the level of the axons, the location of the TMN cell body (Pipes, 2001).

Alterations in Beat Ic expression in these two neurons also result in TN fasciculation defects. The elav-GAL4 driver drives relatively high levels of transgene expression in both the TMN and the LBD. Using this driver to misexpress a UAS-beat Ic transgene results in an increase in the level of Beat Ic in the TMN and simultaneously misexpresses Beat Ic in the LBD neuron. This manipulation results in increased stalling and ectopic exploration of the ventral muscle surfaces by the TN (Pipes, 2001).

Attempts were made to rescue the TN phenotype in beat Ic^-/- mutants by re-supplying Beat Ic protein to all neurons using the elav-GAL4 driver described above. If the elav-GAL4 driver perfectly recapitulates the beat Ic expression pattern, then one would expect complete rescue of the beat Ic mutant phenotype. However, elav-GAL4 drives expression in both the TMN and in the LBD neuron, whereas beat Ic is normally only expressed in the former. In spite of this limitation, TN fasciculation is improved in beat Ic mutants by elav-GAL4-driven misexpression of UAS-beat Ic: from 44% of wild type to 56% of wild type (Pipes, 2001).

The vertebrate gene that shows the most homology to Beat is the BL1 gene of mice and humans, but this gene is predicted to encode a protein with a significantly divergent domain structure: three Ig domains, followed by a transmembrane domain and a cytoplasmic tail. The worm gene with the greatest homology to Beat is F12F3.3, the protein product of which is predicted to resemble titins, with dozens of tandem repeats of FN3 and Ig domains. Hence, it seems that the Beat Ig domains predate the arthropod divergence, but since that time these domains have been swapped around as discrete modules onto many different proteins in both lineages. It will be interesting to examine the genomic sequences of other dipterans to see if the expansion in beat gene number was an invention of Drosophila (Pipes, 2001).

Why have the Beats undergone such tremendous expansion in the Drosophila lineage? It may be that some Beat family-regulated developmental process required very fine modulation of signal over time and space. One way to provide this fine modulation would be by duplicating a beat gene and giving it a slightly different signaling or expression property. In support of this notion, many of the Beats are expressed in similar patterns in the embryo: class I Beats are all broadly expressed in ventral nerve cord cells and motoneurons late in embryogenesis; class II Beats are expressed in the ventral oblique muscles and are the only Beats expressed in muscles; class III Beats are all weakly expressed in the embryo; and class IV Beats are all expressed at similar levels by large, ventral cells in the CNS. All of these Beats may play critical roles during some later developmental stage, such as the larval or the adult nervous system (Pipes, 2001).

The fact that the homology among the Beats is highest at the N termini may reflect a binding interaction common to all of the Beats that takes place primarily at the N-terminal immunoglobulin domain (Pipes, 2001).

Loss-of-function mutations in beat Ic partially suppress the hyper-adhesive motor axon phenotype in beat Ia mutants. In a beat Ia mutant, motor axons fail to defasciculate. Simultaneously removing beat Ic allows some of them to defasciculate. Beat Ic therefore appears to function in wild type to promote motor axon adhesion and fasciculation. But Beat Ic does not appear to be a homophilic adhesion molecule: S2 cells expressing high levels of Beat Ic do not aggregate, and beat Ic is only expressed in a limited subset of neurons, not broadly enough to mediate adhesion among all of the axons in the ISN, ISNb and ISNd. Therefore, the function of Beat Ic in axon adhesion is presumably mediated by heterophilic binding to an unknown Beat receptor protein. This Beat receptor could, in turn, bind to all the various members of the Beat family with differing avidities. In this model, a broadly expressed Beat receptor could provide varying levels of adhesion among different subsets of axons, depending on which membrane-anchored Beat they express (Pipes, 2001).

A role for Beat Ic in heterophilic adhesion could explain the loss- and gain-of-function phenotypes of beat Ic in the formation of the transverse nerve (TN). The TN forms as two axons meet around the ventral muscles: the TMN axon extends out from the CNS, and the LBD sensory axon extends in towards the CNS. beat Ic is expressed in the TMN but not in the LBD. In the absence of Beat Ic, these axons do not properly fasciculate. However, when extra Beat Ic is transgenically expressed by both neurons, a defasciculation phenotype is observed. This result is most easily explained by the presence of a putative Beat Ic receptor on either the LBD axons, or some other guiding cell. Saturating this receptor with excess ligand might lead to a gain-of-function phenotype that appears in certain respects similar to the loss of function. These data lead to a model in which Beat Ic and its putative receptor are part of a guidance system that normally drives the TMN axon to fasciculate with the LBD axon (Pipes, 2001).

Despite extensive mutant screens of much of the Drosophila genome, a defasciculation phenotype has never been described for the motor axon projection. No mutant has been described in which the motor axons prematurely defasciculate in the periphery. Even in embryos with severe reductions in the axonal levels of the homophilic cell adhesion molecule Fas II, motor nerves adhere to each other properly as they grow out into the periphery. It is inferred from this that what keeps motor axons fasciculated as they extend into the periphery are multiple adhesion molecules, possibly from a redundant family of multiple genes. Whatever the nature of this non-Fas II adhesion mechanism, it is regulated in large part by secretion of Beat Ia at defasciculation choice points. Embryos that lack Beat Ia show severe phenotypes in which motor axons fail to defasciculate and instead continue to extend along the major motor nerves, tightly fasciculated with other motor axons. The other Beat genes in the fly have single-gene phenotypes that are much less severe, ranging from no observable phenotype (beat Ib) to moderate TN fasciculation defects (beat Ic). These other Beat genes are also all membrane anchored, either by a transmembrane domain or a GPI-linkage (Pipes, 2001).

These other Beats represent obvious candidates for being the Beat Ia-regulated axonal adhesion proteins. Given that Beat Ia is a secreted protein that resembles cell adhesion molecules, it is proposed that Beat Ia functions as a competitive inhibitor of axon adhesion mediated by transmembrane Beat family members binding to a Beat receptor. Secretion of Beat Ia at specific points would thus allow fine alterations in the level of axonal adhesivity. Testing this model in the future will require the identification of the putative Beat family receptor(s) (Pipes, 2001).

REFERENCES

Search PubMed for articles about Drosophila beaten path

Bazan, J. F. and Goodman, C. S. (1997). Modular structure of the Drosophila Beat protein. Curr. Biol. 7(6): R338-9. PubMed ID: 9197253

Fambrough, D. and Goodman, C. S. (1996). The Drosophila beaten path gene encodes a novel secreted protein that regulates defasciculation at motor axon choice points. Cell 87(6): 1049-58. PubMed ID: 8978609

Green, P., Hartenstein, A. Y. and Hartenstein, V. (1993). The embryonic development of the Drosophila visual system. Cell Tissue Res. 273: 583-598. PubMed ID: 8402833

Holmes, A. L., Raper, R. N. and Heilig, J. S. (1998). Genetic analysis of Drosophila larval optic nerve development. Genetics 148: 1189-1201. PubMed ID: 9539434

Holmes, A. L. and Heilig, J. S. (1999). Fasciclin II and Beaten path modulate intercellular adhesion in Drosophila larval visual organ development. Development 126(2): 261-272. PubMed ID: 9847240

Landgraf, M., et al. (1999). even-skipped determines the dorsal growth of motor axons in Drosophila. Neuron 22(1): 43-52. PubMed ID: 10027288

Li, H., Watson, A., Olechwier, A., Anaya, M., Sorooshyari, S. K., Harnett, D. P., Lee, H. P., Vielmetter, J., Fares, M. A., Garcia, K. C., Ozkan, E., Labrador, J. P. and Zinn, K. (2017). Deconstruction of the beaten Path-Sidestep interaction network provides insights into neuromuscular system development. Elife 6. PubMed ID: 28829740

Ozkan, A. and Berberoglu, H. (2013). Cell to substratum and cell to cell interactions of microalgae. Colloids Surf B Biointerfaces 112: 302-309. PubMed ID: 24004676

Lin, D. M. and Goodman, C. S. (1994). Ectopic and increased expression of Fasciclin II alters motoneuron growth cone guidance. Neuron 13: 507-523. PubMed ID: 7917288

Mushegian, A. R. (1997). The Drosophila Beat protein is related to adhesion proteins that contain immunoglobulin domains. Curr. Biol. 7(6): R336-8. PubMed ID: 9197251

Pipes, G. C. T., et al. (2001). The Beat generation: a multigene family encoding IgSF proteins related to the Beat axon guidance molecule in Drosophila. Development 128: 4545-4552. 11714679

Siebert, M., Banovic, D., Goellner, B. and Aberle, H. (2009). Drosophila motor axons recognize and follow a Sidestep-labeled substrate pathway to reach their target fields. Genes Dev. 23(9): 1052-62. PubMed Citation: 19369411

Thor, S. and Thomas, J.B. (1997). The Drosophila islet gene governs axon pathfinding and neurotransmitter identity. Neuron 18: 397-409. PubMed ID: 9115734

Vactor, D. V., et al. (1993). Genes that control neuromuscular specificity in Drosophila. Cell 73(6): 1137-53. PubMed ID: 8513498

date revised: 15 April 2020

Home page: The Interactive Fly © 2017 Thomas Brody, Ph.D.