BIOLOGICAL OVERVIEW

The ability of neurons to alter their properties is fundamental to achieving the complex wiring required of nervous systems. At certain stages in development, projecting axons will respond to attractants while at other times the same axons change properties and become refractory to the same attractants that once served to direct their migratory behavior. An important region where these varying responses occurs is the midline of the central nervous system.

The midline of the insect central nervous system acts as a barrier to axon crossing. The trajectory of most growth cones in the nerve cord of insects extends toward the midline at some time during development, and a majority of these cross the midline, propelled by long-range chemoattractants. After crossing, the majority of these growth cones turn to project longitudinally, growing along or near the midline. Interestingly, these axons never cross the midline again. What mechanism prevents them from re-crossing the midline? What changes, if any, are there in neuronal properties as a result of a midline crossing, which then prevent additional crossing events? The existence of a mechanism to prevent subsequent crossing events is indicated by the mutation termed roundabout (robo), in which many growth cones that normally cross the midline only once appear to cross and recross multiple times. robo was isolated in a large-scale screen for mutations in which either too many or too few axons crossed the midline (Seeger, 1993). One possibility is that Robo is a growth cone receptor for a putative midline repellent and that Robo serves as a gatekeeper, preventing midline crossing by acting in a cell autonomous fashion. Alternatively, Robo may code for an unknown longitudinal attractant, ensuring that axons, once they have crossed the midline, never change course, thereby preventing a recrossing of the midline (Kidd, 1998a). Subsequent studies demonstrate that Slit functions as a midline repellent for the Robo receptor (Kidd, 1999).

Analysis by immunostaining of Robo expression on neurons demonstrates an alteration in the presence of Robo during the pathfinding process. Little or no Robo expression is observed on commissural growth cones as they extend toward and across the midline. However, as these growth cones turn to project longitudinally, their levels of Robo expression dramatically increase. Robo is expressed at high levels on all longitudinally projecting growth cones and axons. In contrast, Robo is expressed at nearly undetectable levels on commissural axons. Immunoelectron microscopy was used to examine Robo localization at high resolution. In stage 13 embryos, Robo is expressed at higher levels on growth cones and filopodia in the longitudinal tracts than on the longitudinal axons themselves. This localization is consistent with the model that Robo functions as a guidance receptor. The increased sensitivity of immunoelectron microscopy reveals the presence of very low levels of Robo protein on the surface of commissural axons. In addition, Robo-positive vesicles can be seen inside the commissural axons, possibly representing transport of Robo to the growth cone or internalization of Robo from the commissural axons. By reconstructing the path of single axons, using serial sections, it has been confirmed that Robo expression is greatly up-regulated after individual axons turn from the commissure into a longitudinal tract. The expression of Robo on noncrossing and postcrossing axons and its higher level of expression on growth cones and their filopodia suggest a model where Robo functions as an axon guidance receptor for a repulsive midline cue (Kidd, 1998a).

To test whether Robo functions in a cell-autonomous fashion, Robo was expressed in a subset of CNS neurons, including many of the earliest neurons to be affected by the robo mutation, such as pCC, vMP2, dMP2, and MP1. Expression of robo is sufficient to rescue these identified neurons in the robo mutant. In such rescued embryos the neuron pCC, which in robo mutants heads toward and crosses the midline, now projects ipsilaterally (to the same side) and does not cross the midline. When the same embryos are stained with the anti-Robo antibody, all Robo-positive axons are found not to cross the midline. In this directed expression, Robo was expressed in many of the axons in the pCC pathway, a medial longitudinal fascicle. In robo mutants, this axon fascicle freely crosses and circles the midline, joining with its contralateral pathway. When rescued by directed expression, this pathway largely remains on its own side of the midline, even though occasionally a few axons cross the midline. These experiments support the notion that Robo can function in a cell-autonomous fashion, that is robo directs the behavior of the neurons in which it is expressed (Kidd, 1998a).

While this model of Robo as a midline repellent remains a formal possibility, it is now thought unlikely, for at least two reasons: (1) in robo mutants, growth cones that abnormally cross the midline still tend to extend in the correct direction and project longitudinally the normal distance, and they display their normal patterns of selective fasciculation with the appropriate axons on both sides of the midline. (2) Other mutants that perturb longitudinal extension do not automatically lead growth cones to cross the midline. Crossing the midline is not the default pathway for abnormal longitudinal guidance cues. Rather, the midline appears to represent a repulsive barrier that is not easily penetrated, even when growth cones do not find their normal longitudinal pathways. These arguments lead to a preference for the model in which Robo functions as a guidance receptor for a midline repellent. Regardless of which model is correct, Robo clearly functions as a guidance receptor, and this receptor's major role is as a gatekeeper, controlling crossing and recrossing of the midline. Put succinctly, growth cones expressing high levels of Robo are prevented from crossing the midline; growth cones expressing very low levels of Robo are allowed to cross the midline (Kidd, 1998a).

Of special interest in understanding the function of roundabout are the dosage-sensitive interactions between robo and commissureless. In comm mutant embryos, commissural growth cones initially orient toward the midline but then fail to cross it and instead recoil and extend on their own side. Thus comm mutation has a complementary phenotype to that of robo; that is, too few axons cross the midline (Seeger, 1993).

comm encodes a novel surface protein expressed on midline cells. Comm is thought to be a midline attractant. As commissural growth cones contact and traverse the CNS midline, Comm protein is apparenty transferred from midline cells to commissural axons. The double mutant for robo and comm is indistinguishable from robo. The expression of Robo protein was examined in comm hypomorphic alleles. Normally, Robo is expressed at very low levels on commissural axons and at high levels on longitudinal axons. In mutant embryos, Robo expression in the longitudinal tracts appears as if it might be higher than normal. Interestingly, in comm hypomorphic alleles, the occasional thin commissures express Robo protein at levels that are higher than normally seen in the commissures and closer to what is typically seen in the longitudinal tracts. This result was the first hint that Comm protein might function by suppressing Robo expression on commissural axons, thus allowing axons to cross the midline, an event that occurs as a default state in the absence of both proteins (Kidd, 1998b).

To test the hypothesis that increased expression of comm might lead to a robo-like phenotype comm was expressed pan-neurally. A continuous range of robo-like phenotypes is generated under these conditions. The range of phenotypes reveals the comm-gain-of-function phenotype to be dosage-sensitive, since the severity increases in embryos expressing two copies of comm, as compared to those expressing only one. In wild-type embryos, the pattern of Robo protein expression begins in the neuroepithelium as well as in some lateral epidermal stripes, but is conspicuously absent from the midline region. In comm gain-of-function embryos, Robo expression in the neuroepithelium is greatly reduced or absent, while the epidermal expression outside the nervous system is maintained. This same pattern can be observed around the time when the first growth cones are extending. In wild-type embryos during stages 12 and 13, no Robo is seen at the midline, but there is a high level of Robo expression on ipsilaterally projecting growth cones and a significant level throughout the neuroepithelium. In contrast, in comm gain-of-function embryos, such growth cones lack Robo protein, and the neuroepithelium expresses greatly reduced levels of Robo (Kidd, 1998b).

What conclusions can be drawn from the results of comm overexpression? These are indictions that the normal function of comm is to down-regulate the low level of Robo expression present on commissural axons, thereby allowing them to cross the midline. Increasing levels of Comm in the CNS lead to more severe robo -like phenotypes, indicating a dosage sensitivity. This sensitivity to dosage is also reflected in the behavior of axons in robo heterozygotes, thus showing a parallel dosage sensitivity by either decreasing Robo or increasing Comm. Whereever Comm is high, Robo is low, and vice versa. Comm is expressed at the midline, and Robo expression is very low on commissural axons crossing the midline. In comm hypomorphic mutants, those few axons that do cross the midline now express higher levels of Robo protein. In comm gain-of function embryos, the overall levels of Robo are dramatically decreased wherever increased Comm expression coincides with Robo expression. Futhermore, once Comm disappears in older embryos, Robo protein expression begins to increase toward its normal levels. Thus, Comm appears to down-regulate Robo expression in a very tight fashion. It is suspected that it may do so by acting locally through cell-cell contact (Kidd, 1998b).

Slit is secreted by midline glia in Drosophila and functions as a short-range repellent to control midline crossing. Although most Slit stays near the midline, some diffuses laterally, functioning as a long-range chemorepellent. A combinatorial code of Robo receptors controls lateral position in the CNS by responding to this presumptive Slit gradient. Medial axons express only Robo, intermediate axons express Robo3 and Robo, while lateral axons express Robo2, Robo3, and Robo. Removal of robo2 (bearing the official FlyBase designation of leak) or robo3 causes lateral axons to extend medially; ectopic expression of Robo2 or Robo3 on medial axons drives them laterally. Precise topography of longitudinal pathways appears to be controlled by a combination of long-range guidance (the Robo code determining region) and short-range guidance (discrete local cues determining specific location within a region) (Simpson, 2000b).

Robo and Robo2 together play an early function in the control of midline crossing. robo continues to be expressed by all neurons, and Robo protein appears at high levels on axons either after they cross the midline, or from the outset if they never cross the midline. Robo2 is more dynamic in its pattern of expression. Initially, it is expressed by a wide range of neurons, including all of the early pioneer neurons whose axons do not cross the midline (e.g., pCC, MP1, dMP2, and vMP2). But during the period around late stage 13 in which these axons selectively defasciculate to form the medial pCC pathway and the intermediate MP1 pathway, the expression of Robo2 declines in many of these neurons. It is during this same period (late stage 13 to stage 14) that Robo3 begins to be expressed by a subset of neurons (Simpson, 2000b).

From stage 14 onward, as multiple longitudinal pathways form, all three Robos are expressed on some or all longitudinal axon tracts and are excluded from commissural axon tracts. Within the longitudinal tracts, their expression patterns differ dramatically. Robo is found on all longitudinal axon pathways. The second phase of Robo2 expression, and the only phase of Robo3 expression, have a common quality. Both are expressed on a subset of axons that extend in specific lateral positions of the developing CNS. Robo3 is expressed only on axons that extend in the outer two-thirds of the longitudinal pathways (the intermediate and lateral regions). A high level of Robo2 expression is restricted to axons that extend in the outer third of the longitudinal pathways (the lateral region), farthest from the midline and thus farthest from the source (Simpson, 2000b).

All three Robos show relatively tight boundaries. All three are absent from the commissures, and Robo3 and Robo2 are restricted to certain regions of the longitudinal pathways. The expression of Robo3 and Robo2 is not graded, but rather appears to form regional boundaries. While the high level of Robo2 is restricted to the lateral pathways, a lower level of Robo2 expression is detected on some of the intermediate pathways (the more lateral ones). The low level of Robo2 expression begins right in the middle of the intermediate Fas II pathway. This step-wise expression of Robo2 (from none on the medial portion of the intermediate pathways, to a low level on the lateral portion of the intermediate pathways, to a high level laterally) reveals further regional subdivisions of the longitudinal pathways (Simpson, 2000b).

The pattern of expression of Robo3 and Robo2 in individual identified neurons is consistent with their overall patterns of expression. For example, robo3 RNA is expressed in the MP1 neuron whose axon pioneers the intermediate Fas II pathway. robo3 is largely absent from pCC and other neurons whose axons pioneer the medial Fas II pathway. All of these neurons (e.g., MP1, pCC) transiently express robo2 when they are making the earlier decision not to cross the midline, but such expression declines by the time the medial and intermediate Fas II longitudinal pathways separate from one another. This is consistent with the presence of Robo2 only on lateral axons during later stages of development (Simpson, 2000b).

All three Robos are expressed on specific growth cones and filopodia. Commissural growth cones and axons are devoid of all three Robos. All ~150 longitudinal axons express Robo. The ~50 intermediate axons and the ~50 lateral axons express Robo3 (with one exception). The ~50 lateral axons express Robo2. Most lateral axons express Robo2 and Robo3, with one exception; the most lateral axon bundle of ~10 axons expresses high levels of Robo2, but is largely or completely devoid of Robo3 (Simpson, 2000b).

The mutant phenotypes of robo, robo2, and robo3 support the hypothesis that the Robos specify lateral position with respect to the midline. robo mutants show axons ectopically crossing and recrossing the midline. These axons are predominantly those of the innermost part of the longitudinal scaffold. When a robo mutant is examined with anti-Fas II (mAb 1D4), only the medial Fas II pathway crosses the midline. The intermediate and lateral Fas II pathways stay on their own side. One possible interpretation is that the intermediate and lateral expression of Robo3 and Robo2 keeps these axons from crossing the midline in a robo mutant. In a robo, robo2 double mutant, all axons go to the midline and do not leave it (and thus it looks like a slit mutant) (Simpson, 2000b).

robo2 loss-of-function mutations show occasional ectopic midline crossing, but, more prominently, they show abnormalities in lateral positioning. The most common phenotype as revealed with anti-Fas II staining is crossovers and 'braiding' between the intermediate and lateral Fas II pathways (and sometimes between the medial and intermediate Fas II pathways). Although superficially the axon scaffold looks relatively normal in a robo2 mutant (when visualized with mAb BP102, which labels all axons), the lateral positions of the longitudinal pathways are altered in the absence of Robo2 (Simpson, 2000b).

The role of Robo3 in lateral positioning was examined using RNA interference (RNAi). Injection of robo3 dsRNA causes the stage 16 embryo to have two Fas II longitudinal pathways instead of three. The intermediate pathway is missing, and the medial pathway is larger than normal. The lateral (normally Robo2 expressing) Fas II pathway appears normal. In the absence of Robo3, the medial and intermediate Fas II pathways fail to separate, and the intermediate (normally Robo3 expressing) pathway does not properly form (Simpson, 2000b).

The robo2, robo3 double mutant, generated by injecting robo3 dsRNA into a robo2 mutant, contains a large single Fas II longitudinal pathway. This single fascicle is thicker than wild-type Fas II pathways and is close to the midline in the normal location of the medial Fas II pathway. In the absence of both Robo3 and Robo2, it appears as if most (and in some cases all) of the Fas II axons selectively fasciculate into one Fas II pathway in the medial position. This suggests that Robo3 and Robo2 are required for the normal formation of the Robo3 expressing intermediate Fas II pathway and the Robo2/Robo3 expressing lateral Fas II pathway (Simpson, 2000b).

Unlike robo mutants alone, robo, robo3 embryos (robo mutants with robo3 dsRNA) exhibit ectopic crossing of the intermediate Fas II pathway as well as the medial one. This resembles the addition of the two individual phenotypes: lack of Robo3 causes the intermediate pathway to join the medial pathway, and the lack of Robo allows this fused Fas II pathway to weave back and forth across the midline. In the robo, robo3 double mutant, the outer Fas II pathway remains on its appropriate side, presumably due to the presence of Robo2 (Simpson, 2000b).

These phenotypes support the model that Robo is the most important contributor to maintaining the Fas II pathways on the appropriate side of the midline, but that Robo2 and Robo3 determine the lateral position of these and other longitudinal pathways. Robo3 specifies the intermediate region and its pathways, while Robo2 specifies the lateral region and its pathways (Simpson, 2000b).

Overexpression of Robo2 supports the model that Robo2 levels contribute to the lateral position of axons. Overexpression of UAS-robo2 in all CNS axons using the elav-GAL4 driver results in a commissureless-like phenotype (i.e., appearing like the commissureless mutant). There are a number of other genetic combinations that result in a commissureless-like phenotype. These all look the same when examined with mAb BP102 that stains all axons: the commissures are missing. But the appearance of the three Fas II pathways differs, depending upon the genetic makeup of the embryo (Simpson, 2000b).

When Robo expression is increased on all axons, by either directly driving more Robo or in a comm mutant (in which Comm no longer downregulates Robo), three distinct Fas II pathways are still detected. This is true even when the Robo Y-F 'hyperactive' receptor is transgenically expressed on all axons. Under these various conditions, there is disorganization of longitudinal pathways, but the three Fas II pathways can generally be detected (Simpson, 2000b).

But when Robo2 is ectopically expressed on all axons, the lateral position of the pathways is disrupted, and as a result, all three Fas II pathways are bundled together into a single, thick tract. Ectopic Robo2, but not ectopic Robo or loss of Comm, is sufficient to override the endogenous positional information that specifies the locations of the three Fas II pathways (Simpson, 2000b).

Overexpression of Robo3 in all CNS axons (using the elav-GAL4 driver) results in a weakly commissureless-like phenotype. As with UAS-robo and UAS-robo2, the gain-of-function commissureless phenotype of Robo3 requires two copies of the UAS-robo3 reporter to generate commissureless segments. Some of the Fas II bundles are fused, but at least two distinct fascicles are still visible at this level of overexpression (Simpson, 2000b).

Expressing Robo2 or Robo3 in subsets of neurons whose axons normally extend in medial longitudinal pathways can drive these axons to assume more lateral positions. Although both can drive medial axons further laterally, Robo2 and Robo3 are not identical: when tested on the same axons, Robo2 drives medial axons further laterally than does Robo3 (Simpson, 2000b).

For example, in each abdominal hemisegment, three neurons express the transcription factor Apterous (Ap). These neurons normally extend their axons toward the midline, and then turn anteriorly on their own side close to the midline. These axons turn anteriorly in the medial region. When viewed at the light level with confocal microscopy, the Ap axons sometimes look like they are running just at the lateral edge of the medial Fas II bundle, and sometimes there is a little space between them (suggesting another axon or two interposed. This staining pattern was used to infer that the Ap axons run in a medial axon pathway just lateral to the medial Fas II tract (Simpson, 2000b).

When these axons ectopically express Robo2 under control of Apterous-GAL4 (Ap-GAL4), they move laterally and extend anteriorly in a specific location between the intermediate and the lateral Fas II pathways. In fact, they extend in the medial-most region of the endogenous Robo2 expression zone. The Ap axons from neighboring segments appear to pick the same lateral pathway, and to fasciculate together as they extend anteriorly from segment to segment. Two different GAL4 reporters were tested. Both drive different levels of Robo2 expression as indicated by their different strengths of pan-neural gain-of-function phenotypes. Nevertheless, both drive the Ap axons to the same lateral location. There is a second argument that supports this same conclusion. It is well known that the GAL4 expression system drives different levels of expression from a UAS transgene in different cells and segments of the same embryo. Yet, from cell to cell, segment to segment, and embryo to embryo, the Ap axons (expressing what is presumed to be variable levels of Robo2) always turn anteriorly in the same lateral location between the intermediate and lateral Fas II pathways (Simpson, 2000b).

This suggests that Robo2 reading of the Slit gradient drives axons to a rough lateral position, regardless of the precise level of Robo2, after which local cues determine which specific pathway a particular axon joins. Thus, Robo2 drives the Apterous axons to the lateral third of the scaffold, and regardless of the precise level of Robo2, this is sufficient to allow their final pathway choice to be precisely and uniformly dictated by some unknown but specific local cue. All of these experiments are done at different Robo2 levels within a relatively narrow range. It is conceivable that much higher levels of Robo2, or Robo3, might drive axons further laterally, even into more lateral zones (Simpson, 2000b).

Using the same Ap-GAL4 transgene to drive overexpression of Robo3 in the Ap axons leads to a different alteration in lateral position. Whereas ectopic Robo2 drives these axons quite far laterally to a position between the intermediate and lateral Fas II pathways, the ectopic expression of Robo3 drives them to an intermediate position, just medial to the intermediate Fas II pathway. Thus, Robo3 and Robo2 can both drive the medial Ap axons to more lateral positions, but they do so to different extents, as might be expected by their normal patterns of expression. Robo2 drives axons further laterally than does Robo3, relative to the intermediate Fas II bundle (Simpson, 2000b).

Using the same Ap-GAL4 transgene to drive overexpression of either Robo or the hyperactive Robo Y-F in the Ap axons leads to no alteration in lateral position. Even with more Robo, these axons still continue to extend in their normal medial location. These axons, and for that matter all longitudinal axons, normally express Robo, and it is inferred from this result that increasing the level of Robo does not alter the choice of lateral position by typical follower growth cones. The choice of lateral position by these axons is exquisitely sensitive to the presence of Robo2 and Robo3, but apparently not to the level of Robo. The ability of Robo2 to lateralize these axons does not depend on the presence of Robo: lateralization occurs in neurons that coexpress transgenic Robo2 and a Robo dominant negative receptor, or when Robo2 is expressed ectopically in a robo null mutant background (Simpson, 2000b).

The role of the Robo code in determining lateral position, and the potential interplay of this long-range guidance system were further analyzed using the presumptive Slit gradient with other local cues. Are all medial axons driven to the same lateral positions by Robo3 and Robo2? Or, alternatively, are they driven to cell-specific lateral positions? If the latter is the case, are their other cues that help predict the specific location? To this end, other GAL4 lines were used to drive the expression of various Robo family members in other subsets of axons (Simpson, 2000b).

The 15J2-GAL4 line drives expression in the dMP2 and vMP2 neurons (and variably in a few other neurons). These two neurons normally express Fas II, and normally extend in the medial Fas II pathway. Ectopic expression of Robo2 in these neurons leads to a bimodal phenotype. The dMP2 and vMP2 axons always appear to extend in a Fas II pathway, but they now pick either the intermediate or lateral Fas II pathways. These axons are never found medially, are often found in the intermediate Fas II pathway, and occasionally are found in the lateral Fas II pathway. These are distinctly different locations from where the Ap neurons are driven by Robo2 expression. It is sometimes difficult to determine which pathway the dMP2 and vMP2 axons are in (i.e., intermediate vs. lateral) because the two pathways intertwine in these gain-of-function embryos. Comparing the two experiments, Robo2 drives the Ap neurons to a non-Fas II pathway, while it drives dMP2 and vMP2 (which normally follow the medial Fas II pathway) into either the intermediate or lateral Fas II pathway (Simpson, 2000b).

It is concluded that precise topography requires more than just the Robo code. Robo3 and Robo2 expression define specific lateral regions. Ectopic Robo3 drives axons into the intermediate region, while ectopic Robo2 drives them even further laterally. But axons respond in a cell-specific fashion. Ectopic Robo2 drives the three Ap axons to between the intermediate and lateral Fas II pathways, while it drives the dMP2 and vMP2 axons into either the intermediate or lateral Fas II pathway. The control of location appears irrespective of level, since the result is consistent in spite of the different levels of expression generated by different Robo2 reporter lines, and by the variability in expression as driven with the GAL4 system. How can this precision be explained (Simpson, 2000b)?

Many models for topographic specificity involve the notion of two opposing gradients, either both of the same sign (i.e., both either attractive or repulsive) in the opposite orientation, or both of different signs in the same orientation. Such models are very attractive to explain certain aspects of sensory maps in the brain. However, such models need not apply to all topographic projections. Thus far, no evidence has been found for a second gradient working in concert with the repulsive Slit gradient. The most parsimonious model is that precise topography in the medial-lateral axis of the Drosophila CNS requires two opposing forces: long-range repulsion and short-range attraction. Cues might exist, for example, that mark the boundary of the neuropil. But in terms of location with the neuropil, all that is required is an opposing force to the Slit gradient -- it need not be a long-range gradient itself. Discrete local cues would be sufficient. Clearly, the long-range repulsion is controlled by the Slit gradient and the Robo code. It is proposed that the opposing force is short-range attraction as controlled by discrete local cues, one of which is Fasciclin II. In this way, the Robo code specifies the lateral region, while local cues specify precise location within that region (Simpson, 2000b).

The strongest support of this model involves the specification of the three major Fas II pathways. Fas II is a homophilic cell adhesion molecule expressed on axons that fasciculate together in three major longitudinal pathways: one medial, one intermediate, and one lateral. Growth cones expressing Fas II and Robo pick the medial Fas II pathway. Growth cones expressing Fas II, Robo3, and Robo pick the intermediate Fas II pathway. Presumably, the attraction of the medial Fas II pathway is insufficient to balance the repulsion mediated by Robo3. Growth cones expressing Fas II, Robo2, Robo3, and Robo pick the lateral Fas II pathway. In this case, it is not until they contact the lateral Fas II pathway that the Fas II-mediated attraction is stronger than the Robo2-mediated repulsion. Removal of Robo3 leads to only two Fas II pathways in which the intermediate pathway is missing, and instead the medial pathway is twice as thick. Ectopic expression of Robo2 in the dMP2 and vMP2 neurons, which normally extend in the medial Fas II pathway, drives their axons into either the intermediate or lateral Fas II pathway. Specificity is determined by the combination of Fas II and the particular Robo family members (Simpson, 2000b).

It is proposed that other pathways are specified by other pathway labels. For example, two pathways -- one medial and the other lateral -- express Connectin, another homophilic cell adhesion molecule. Growth cones expressing Connectin and Robo pick the medial Connectin pathway, while growth cones expressing Connectin and Robo2 (and presumably Robo and Robo3) pick the lateral Connectin pathway. Removal of Robo2 leads to only one fused medial Connectin pathway (Simpson, 2000b).

How do Robos read and respond to the Slit gradient? How are Robo3 and Robo2 different from Robo? How do Robo3 and Robo2 specify lateral position? Why does Robo3 drive axons into the intermediate region, while Robo2 drives them into the lateral region? Robo3 and Robo2 must differ from one another in either their ectodomains (and thus have different abilities to read the Slit gradient), or in their cytoplasmic domains (and thus have different abilities to signal), or both. What are the key differences that allow them to drive axons to different lateral regions? Both of these receptors (Robo3 and Robo2) differ from Robo in some quality of their signaling, either having some additional output or missing some output found in Robo. Their cytoplasmic domains are quite different from Robo, but what differences are key for determining lateral position (Simpson, 2000b)?

It will be of interest to determine to what extent different chimeric receptors and mutated receptors can drive lateralization. Preliminary collaborative results suggests that it should be possible to separate the functions of the various ectodomains and cytoplasmic domains (Simpson, 2000b).

Drosophila neurexin IV interacts with Roundabout and is required for repulsive midline axon guidance

Slit/Roundabout (Robo) signaling controls midline repulsive axon guidance. However, proteins that interact with Slit/Robo at the cell surface remain largely uncharacterized. This study reports that the Drosophila transmembrane septate junction-specific protein Neurexin IV (Nrx IV) functions in midline repulsive axon guidance. Nrx IV is expressed in the neurons of the developing ventral nerve cord, and nrx IV mutants show crossing and circling of ipsilateral axons and fused commissures. Interestingly, the axon guidance defects observed in nrx IV mutants seem independent of its other binding partners, such as Contactin and Neuroglian and the midline glia protein Wrapper, which interacts in trans with Nrx IV. nrx IV mutants show diffuse Robo localization, and dose-dependent genetic interactions between nrx IV/robo and nrx IV/slit indicate that they function in a common pathway. In vivo biochemical studies reveal that Nrx IV associates with Robo, Slit, and Syndecan, and interactions between Robo and Slit, or Nrx IV and Slit, are affected in nrx IV and robo mutants, respectively. Coexpression of Nrx IV and Robo in mammalian cells confirms that these proteins retain the ability to interact in a heterologous system. Furthermore, the extracellular region of Nrx IV was shown to be sufficient to rescue Robo localization and axon guidance phenotypes in nrx IV mutants. Together, these studies establish that Nrx IV is essential for proper Robo localization and identify Nrx IV as a novel interacting partner of the Slit/Robo signaling pathway (Banerjee, 2010).

Recent studies have established strong expression of Nrx IV at the interface between neurons and midline glie (MG) that underlies the adhesive interactions between these cells in maintaining ML cytoarchitecture. This function of Nrx IV is Wrapper-dependent. It is believed that the axon guidance function of Nrx IV is Wrapper independent based on the following observations. First, the juxtaposition of strong neuronal Nrx IV and glial Wrapper in the ML, together with the absence of axon guidance phenotype in wrapper mutants strongly suggest that Nrx IV localization in ML neurons does not contribute to the axon guidance function of Nrx IV. This is further supported by the failure to rescue the axon guidance phenotypes in nrx IV mutant by expressing Nrx IV in all ML neurons and glia using sim-Gal4::UAS-nrx IV. Therefore, a significant contribution of the axon guidance phenotype seen in nrx IV mutants is unlikely to come from ML neurons or glia. Interestingly, the MG ensheathment defects seen in nrx IV mutants are rescued by expression of Nrx IV in neurons alone and not MG. Furthermore, no changes in Robo localization, or crossing of Robo positive axons were seen in wrapper mutants, thus providing additional evidence that alterations in MG/neuronal architecture in wrapper mutants does not significantly contribute to the axon guidance phenotype. Of note, older stage 16 robo and slit mutants show considerable disorganization of MG, as revealed by immunostaining with anti-Wrapper antibodies; however, these glial phenotypes are thought to be secondary to their axon guidance phenotypes. In addition, Contactin (Cont) and Nrg, two well-established binding partners of Nrx IV at SJs, do not contribute to Nrx IV in its CNS axon guidance function, as revealed by Fas II immunostaining of cont and nrg mutants. Based on these observations, it is believed that Nrx IV is a multifunctional protein that functions in a cell type specific manner. Therefore, the axon guidance function of Nrx IV is to ensure proper localization and stability of Robo in the lateral CNS soma and axons during ML axon repulsion (Banerjee, 2010).

Additional support for tissue specific functions of Nrx IV comes from recent reports that Drosophila cardiac development uses a non-canonical role of Nrx IV to maintain cardiac integrity, by coupling with G-protein signaling. Both Robo and Slit have previously been shown to control cardiac cell polarity and morphogenesis. Since the embryonic heart cells lack SJs, these findings further underscore the fact that Nrx IV and Robo/Slit coordinate diverse roles in different tissues involving multiple molecular partners. Thus, Nrx IV functions in Slit/Robo axon guidance pathway independently of its other known partners, such as Cont, Nrg and Wrapper (Banerjee, 2010).

The findings strongly support the existence of Nrx IV, Robo and Slit as a molecular complex. Although Nrx IV function appears interwoven with Robo and Slit, the phenotypes displayed by nrx IV mutants do not completely phenocopy either slit or robo mutants, suggesting that Nrx IV plays a modulatory role in Slit/Robo signaling. The biochemical analyses suggest that Robo is required for Nrx IV stability, as the levels of Nrx IV are significantly reduced in robo mutants. Slit, on the other hand, showed a modest stimulatory effect on Robo and Nrx IV association and expression levels, further confirming that these three proteins are functionally interlinked. Furthermore, the Slit/Robo complex is less efficiently immunoprecipitated from nrx IV mutant. Thus, while loss of Nrx IV does not abolish interactions between Robo and Slit, it could potentially affect proper functioning of the Robo/Slit signaling complex. Similarly, reduced association of Slit and Nrx IV in robo mutants suggests that Robo is also important for efficient complex formation between these three proteins (Banerjee, 2010).

Both Nrx IV and Robo are transmembrane proteins that colocalize in longitudinal axons. Most of the known Nrx IV interacting proteins, such as Cont, Nrg and Wrapper belong to Ig superfamily of cell adhesion molecules (CAM). Therefore, it is conceivable that Nrx IV associates with Robo (an Ig CAM) in neurons which do not express detectable levels of Cont or Wrap. Furthermore, nrx IV mutant phenotypes resemble those of robo mutant, and Nrx IV interacts with Robo/Slit, suggesting that Nrx IV functions in the Robo/Slit pathway (Banerjee, 2010).

Recent studies support a model where Slit stimulation recruits cytoplasmic Sos to Robo receptor via Dock to activate Rac-dependent cytoskeletal changes within the growth cone during repulsion. This study shows that Nrx IV and Robo retain their ability to colocalize and interact when co-expressed in a heterologous system, and indicate that Slit is dispensable for their interaction. In addition, slit RNAi experiments in S2 cells reveal that Nrx IV and Robo associate in the absence of Slit. However, Slit stimulation of nrx IV/robo cotransfected CHO cells caused enhanced colocalization of Nrx IV and Robo in intracellular compartments and membrane ruffles, further supporting a functional relationship between Nrx IV and Robo/Slit. Together, these in vivo and in vitro findings indicate that Nrx IV and Robo interact in the absence of Slit, and in the presence of Slit ligand the molecular interactions between Nrx IV and Robo are strengthened. Formation of this larger molecular complex at the axonal surface thus ensures proper ML axon guidance (Banerjee, 2010).

The phenotypic similarities, dose-dependent genetic interactions and the in vivo biochemical data suggest that Nrx IV acts as a modulator in Slit/Robo signaling pathway. One of the key reasons for this conclusion is the fact that the axon guidance phenotypes in nrx IV mutants is rescued by the expression of the extracellular region of Nrx IV (Nrx IV^mycΔCT), where as the phenotype is not rescued by the intracellular region of Nrx IV (Nrx IV^mycΔNT). For Nrx IV to act as an independent signal transducer, it would need an intact cytoplasmic region. Since the axon guidance phenotypes and Robo localization are both rescued by Nrx IV^mycΔCT, the downstream signaling controlling axon repulsion is controlled by Robo or an as yet unidentified protein. Therefore, the data support a role for Nrx IV in the proper localization and stabilization of Robo at axonal membrane, where it interacts with Slit, to regulate downstream axon guidance signaling. Although the exact domains regulating Nrx IV-Robo interactions are unknown at this point, it is predicted that they occur via the Ig or FNIII domains, as these domains regulate Nrx IV-Cont interactions. The data rule out the possibility that Nrx IV interactions with Robo occur via a large cytoskeletal scaffolding complex, as Nrx IV^mycΔCT lacks the cytoplasmic region. This led to the conclusion that Nrx IV and Robo interact in cis through their extracellular regions, and therefore eliminate the possibility of a parallel Nrx IV signaling pathway (Banerjee, 2010).

One of the interesting in vivo observations is the association between Nrx IV and Slit still occurs in robo null mutant embryos, indicating that Slit/Nrx IV can interact in the absence of Robo. Although it remains to be seen if Nrx IV and Slit can associate in the absence of all Drosophila proteins in a heterologous system, based on the existing findings it is tempting to speculate that Nrx IV may function as a co-receptor for Slit, and together with Robo, they stabilize the complex to ensure proper presentation or retention at the axonal surface. A similar role has been assigned to Sdc, which is thought to be critical for the fidelity of Slit repellent signaling, as sdc mutants exhibit consistent defects in ML axon guidance. Thus, a multitude of mechanisms seem to operate at the axonal surface and growth cones to ensure that axons reach their correct targets. The Slit-independent interactions between Nrx IV and Robo, but seemingly enhanced colocalization and interactions in the presence of Slit, point to an interesting mechanism where signaling molecules use accessory proteins to ensure their proper localization and stability. This mechanism ensures checks and balances at several molecular levels to allow navigating axons to reach their final destinations. Very few proteins have been implicated in Slit/Robo signaling at the axonal surface, and additional yet unidentified proteins may be involved. With the identification of Nrx IV as an essential component of the Slit/Robo complex, new insights into this highly sophisticated molecular pathway are opened, and may allow for future studies aimed at identifying the modulatory proteins that coordinate and/or control axon guidance (Banerjee, 2010).

teiresias, a Fruitless target gene encoding an immunoglobulin-superfamily transmembrane protein, is required for neuronal feminization in Drosophila

This study aims at identifying transcriptional targets of FruitlessBM (FruBM), which represents the major isoform of male-specific FruM transcription factors that induce neural sexual dimorphisms. A promoter of the axon-guidance factor gene robo1 carries the 16-bp palindrome motif Pal1, to which FruM binds. A genome-wide search for Pal1-homologous sequences yielded ~200 candidate genes. Among these, CG17716 potentially encodes a transmembrane protein with extracellular immunoglobulin (Ig)-like domains similar to Robo1. Indeed, FruBM overexpression reduced CG17716 mRNA and protein expression. In the fru-expressing mAL neuron cluster exhibiting sexual dimorphism, it was found that CG17716 knockdown in female neurons completely transformed all neurites to the male-type. Conversely, CG17716 overexpression suppressed male-specific midline crossing of fru-expressing sensory axons. CG17716 was renamed teiresias (tei) based on this feminizing function. It is hypothesized that Tei interacts with other Ig superfamily transmembrane proteins, including Robo1, to feminize the neurite patterns in females, whereas FruBM represses tei transcription in males (Sato, 2020).

Reproductive success in many male animals relies on their behavioral performance during mating attempts with a female1. The neural circuitry that controls mating behavior has thus evolved under the strong pressure of sexual selection. Such an evolutionary drive led to the development of sexual dimorphisms in the neural circuitry and neurons that compose this circuitry1. Drosophila melanogaster is an outstanding model organism highly amenable to genetic dissection of complex traits, including mating behavior2. In this organism, the transcription factor gene fruitless (fru) plays a key role in organizing the sexually dimorphic circuitry for mating behavior by specifying sex-specific neuronal structures during development (Sato, 2020).

Among the four promoters identified in the fru gene, the most distal one (the P1 promoter) is dedicated to sex-specific functions ascribed to the male-specific translation of P1-derived mRNAs. As a result of sex-specific splicing of the fru-P1 primary transcript, only male mRNAs have a long ORF encoding male-specific functional fru proteins, called FruM. These proteins are translated as five isoforms, FruAM, FruBM, FruCM, FruDM, and FruEM, which share a common BTB domain in their N-terminus, but which each have a unique C-terminus. The most prevalent isoform is FruBM, which has two zinc-finger motifs in the C-terminus, and forms a complex with chromatin regulators such as the TIF1 homolog Bonus (Bon), histone deacetylase 1 (HDAC1), and heterochromatin protein 1a (HP1a), to bind to more than 100 target sites on the genome for transcriptional regulation of downstream genes. The best-characterized FruBM target is robo1, which encodes a transmembrane protein belonging to the immunoglobulin superfamily with an axon-guidance role in the developing nervous system. FruBM binds to a 42-bp promoter segment of the robo1 gene named the FruBM Response Obligatory Sequence (FROS), which contains a characteristic palindrome sequence of 16 nucleotides (Pal1). Partial deletion of Pal1 in the robo1 gene impairs neural sexual differentiation and mating behavior in male flies, indicating that Pal1 is pivotal for FruBM in order to execute its sexual functions (Sato, 2020).

Many of the fru-expressing neurons are sexually dimorphic, including neurons composing the mAL cluster. The mAL cluster displays sexual dimorphisms in the following four respects: (i) the number of neurons in the cluster (~5 in females vs. ~30 in males); (ii) the absence (females) or presence (males) of a neuronal subset carrying the ipsilateral neurite; (iii) the branched (females) vs. unbranched (males) tip structure of the descending contralateral neurite; and (iv) the focal (females) or expanded (males) terminal arbor distribution of the ascending contralateral neurite. While these four types of sexual dimorphisms of mAL neurons are established in a fru-dependent manner, they are also regulated separately and independently from each other. For example, Hunchback knockdown in male mAL neurons feminizes only the descending contralateral neurite, whereas robo1 knockdown feminizes only the ipsilateral neurite. On the other hand, loss of the cell-death genes, reaper, grim and head involution defective, increases the number of cells composing the mAL cluster from ~5 to ~30 in females, but has no effect in males. Therefore, FruBM likely interacts with different cofactors for transcription regulation, and/or FruBM acts on different targets to establish each of the four sexually dimorphic features of mAL neurons. With the aim of obtaining additional FruBM transcriptional targets for the establishment of neural sexual dimorphisms, this study examined the effects of knocking down the genes harboring a Pal1-homologous motif on mAL neuron structures. teiresias (tei), encoding another immunoglobulin-superfamily member with a transmembrane domain, was identified. Remarkably, knockdown of tei in female mAL neurons completely masculinized all three neurite characteristics, leaving the cell number unaffected. tei knockdown also masculinized fru-expressing sensory neurons and mcALa interneurons in females. It is proposed that the Tei protein functions as a common receptor and interacts with a second receptor, e.g., robo1, which confers the ligand specificity on the heteromeric receptor complex, producing distinct neuronal populations that are diversified in their sex-specific structures (Sato, 2020).

tei, a new transcriptional target of FruBM, was identified based on the in silico search for genes harboring the FruBM-binding consensus sequence and subsequent analysis of gene knockdown effects on sexually dimorphic neurite structures. Tei is a putative transmembrane protein carrying multiple immunoglobulin-like repeats in the extracellular portion. This structure is very much alike that of Robo1, the product of the best-characterized FruBM target gene. Indeed, tei knockdown in female mAL neurons induced the male-specific ipsilateral neurite as did robo1 knockdown, revealing a functional similarity between tei and robo1 in terms of the feminizing effect on this particular structure. Nonetheless, tei was distinctly different from robo1 in that tei is also required for the female-typical shaping of ascending as well as descending contralateral neurites, in which robo1 has no role. Robo1 has been shown to operate as a receptor for the secreted ligand Slit, activating several signal transduction pathways depending on the developmental context in both invertebrates and vertebrates. Promiscuous interactions of Robo1 with other transmembrane receptors in cis have been implicated as a basis for the versatile performances of Robo1 in multiple developmental contexts. The lack of the cytoplasmic portion in Tei might suggest that Tei contributes to the ligand specificity whereas the specificity of intracellular signal transduction following heteromeric receptor activation is determined by Robo1 or another Tei partner that carries the C-terminal cytoplasmic domain. It is therefore plausible that Tei is one of these transmembrane receptors that cooperate with Robo1, and likely initiates together an intracellular biochemical cascade which ultimately inhibits the male-specific ipsilateral neurite from forming in an mAL neuron. Perhaps Tei associates with transmembrane receptors other than Robo1 in specifying the sex-specific structure in the ascending and descending contralateral neurites of an mAL neuron. Systematic knockdown in mAL neurons of genes encoding putative transmembrane receptors with immunoglobulin-like domains other than Robo protein members might lead to the identification of additional Tei partner proteins that are involved in the sex-type specification of the contralateral neurites (Sato, 2020).

Despite its striking effects on neurite structures, tei knockdown had no effect on the sexually dimorphism in the number of mAL neurons. This contrasts with the effects of fru loss-of-function mutation, which reduces the number of cells composing the mAL cluster in males from ~30 to ~5, the cell number typical of the mAL cluster in a wild-type female. Removal of three major cell-death genes, hid, grim and rpr, in female mAL neurons increased the number of cells composing the female mAL cluster up to 29 (the mean cell number was 19). This observation led to the proposition that cell death is the primary cause of the smaller number of cells composing the mAL cluster in females. Moreover, the sex difference in the cell number of fru-expressing neurons has recently been attributed to a difference in proliferation in addition to a difference in cell death: male neuroblasts produce more cells than female neuroblasts, so that males have a larger number of cells even after blocking cell death. These considerations lead to the supposition that FruM proteins separately control the neurite sex-type and the cell number sex-type. In this study it was noted that the Pal1-homologous motif was not found in or around any of the hid, grim, and rpr loci, suggesting they are not direct transcriptional targets of FruM proteins. Alternatively, transcription of these genes might be regulated by a FruM isoform other than FruBM. In fact, the mAL cell number was shown to be affected by the fruB2 (null for FruEM) as well as the fruC1 (null for FruBM) mutation (Sato, 2020).

The 16-bp Pal1 core motif is located within the 42-bp FruBM-binding site, FROS. This motif was first identified in the robo1 gene by a combination of reporter assays, electromobility-shift assays, and CRISPR-Cas9-mediated in vivo mutagenesis in conjunction with phenotypic characterizations of single-cell clones of mAL neurons and mutant-fly behavior. Using different strategies, two other research groups determined the binding consensus sequences for FruAM and FruEM in addition to FruBM. A previous study used the DNA adenosine methyltransferase identification (DamID) method, in which a Fru-fused bacterial methylase methylates the DNA around the genomic region to which the fru moiety of the fusion protein binds. One of the reported FruBM-binding motifs in the DamID-based search revealed a 4/6 match when compared with Pal1, yet robo1 was not obtained as a putative FruBM target in that study. Another study used in vitro screening of oligo-DNAs with random sequences for binding to fru zinc-finger motifs fused to glutathione S-transferase, revealing a binding consensus sequence for each isoform, but none of these had a similarity to Pal1. Of note, CG17716 (tei) has been included, together with another ~1400 genes, in a list of potential FruBM targets having a deduced FruBM-binding sequence. Reliable determination of consensus sequences for the binding of fru isoforms and rigorous identification of fru transcriptional targets are indispensable for obtaining a more complete picture of the molecular mechanisms underlying sex-specific circuit formation (Sato, 2020).

Glial-secreted Netrins regulate Robo1/Rac1-Cdc42 signaling threshold levels during Drosophila asymmetric neural stem/progenitor cell division

Asymmetric stem cell division (ASCD) is a key mechanism in development, cancer, and stem cell biology. Drosophila neural stem cells, called neuroblasts (NBs), divide asymmetrically through intrinsic mechanisms. This study shows that the extrinsic axon guidance cues Netrins, secreted by a glial niche surrounding larval brain neural stem cell lineages, regulate NB ASCD. Netrin-Frazzled/DCC signaling modulates, through Abelson kinase, Robo1 signaling threshold levels in Drosophila larval brain neural stem and progenitor cells of NBII lineages. Unbalanced Robo1 signaling levels induce ectopic NBs and progenitor cells due to failures in the ASCD process. Mechanistically, Robo1 signaling directly impinges on the intrinsic ASCD machinery, such as aPKC, Canoe/Afadin, and Numb, through the small GTPases Rac1 and Cdc42, which are required for the localization in mitotic NBs of Par-6, a Cdc42 physical partner and a core component of the Par (Par-6-aPKC-Par3/Bazooka) apical complex (de Torres-Jurado, 2022).

A precise regulation of ASCD is critical in development, tissue homeostasis, and tumorigenesis. Extrinsic signals from specialized microenvironments, the niches, importantly contribute to that regulation by promoting the stem cell fate in the daughter cell that receives those signals, whereas the other daughter cell enters a differentiation program. Intriguingly, the ASCD of some stem cells, including the Drosophila CNS stem cells called NBs, the subject of this study, seems to depend exclusively on intrinsic regulatory mechanisms. Then, are those stem cells completely independent of their surrounding environment? Also, how general is the requirement of niches and the signals secreted from them for maintaining the stem cell fate in an ASCD (de Torres-Jurado, 2022)?

This study shows that the ASCD of Drosophila NBs, a traditional paradigm for studying intrinsic ASCD regulatory mechanisms, do also depend on extrinsic cues secreted by a glial niche, which are in close contact with those neural stem cell lineages in the larval brain. However, these extrinsic signals are not required to maintain the stem cell fate. They ultimately impact on the regulation of intrinsic factors to induce differentiation in one daughter cell, repressing the self-renewal 'basal state' in this cell. Different studies have shown the possibility of growing in culture isolated larval NBs, which are able to form crescents and divide asymmetrically without any additional extrinsic signal. However, most of these experiments were performed using central brain type I NB lineage (NBI) NBs, which do not require Fra and Robo1 signaling for their correct development, or included type II NB lineage (NBII,) but only particular markers were analyzed (i.e., Baz and Pon). This is relevant as, for example, it was observed that the localization of some ASCD regulators were not affected without extrinsic signals (i.e., Insc, Mira, and Brat). In the experiments in culture, some of the latter regulators are frequently used. It is also key to properly quantify the cases of crescent formation at metaphase, as the phenotypes are never fully penetrant and can be even totally rescued at telophase. For example, in the system used in this study, aPKC showed about 50% localization failures at metaphase, implying that there are NBIIs that show aPKC crescents. Nevertheless, finding that glia-secreted cues are specifically required in NBII lineages was indeed very intriguing. NBII lineages are larger than NBI, as they undergo an additional proliferation phase through INPs and hence are more prone to induce tumor-like overgrowth when ASCD fails. Thus, additional levels of regulation might have evolved in these lineages to ensure the correct division of the NB and INPs to avoid overgrowths. This issue will be further examined in the future (de Torres-Jurado, 2022).

This work has unveiled a novel function for the axon guidance cues Netrins and their Fra/DCC-like receptor in regulating self-renewal versus differentiation in neural stem and progenitor cells of larval NBII lineages. The cortex glial niche that surrounds those lineages secretes Netrins, which modulate Robo1 signaling threshold levels through Fra and Abl kinase in stem and progenitor cells. Whereas Robo1 signaling is activated by its ligand Slit, also secreted by the glial niche, Abl kinase represses this signaling, and these balanced Robo1 signaling levels appear to be critical for the cell-fate commitment of the daughter cell prone to differentiate. The cortex glia dynamically undergoes remodeling through larval stages; only at late third instar larval stages (L3) the glia chamber enwrapping each NB lineage forms completely. This study already observed the presence of the secreted ligands Slit and NetA in the cortex glia at L2, suggesting that these cues are being secreted from the glia since the reactivation of dormant NBs at early L2 (de Torres-Jurado, 2022).

Slit-Robo signaling regulates progenitor cell proliferation in the mammalian CNS and promotes the terminal asymmetric division of a differentiation-committed cell in the Drosophila CNS. Likewise, in mammary stem cells, Robo1 favors their asymmetric mode of cell division. Slit-Robo signaling is also required in other stem or progenitor cells to regulate their lineage specification, identity, or their adhesion/anchoring to the niche. No role for Netrin-Fra/DCC signaling has been previously described in all those contexts (de Torres-Jurado, 2022).

Ultimately, a transcriptional control has been pointed out as the most common way of action by which Robo signaling regulate the above-mentioned cellular processes. This study shows a novel, transcription-independent mechanism by which Robo1 signaling regulates ASCD. Robo1 signaling would be required to activate the small GTPases Rac1 and Cdc42 by repressing its inhibitor RhoGAP93B as well as by recruiting the Dock-Pak complex, which, through Pak, can also bind activated forms of both Rac1 and Cdc42 Rac1 and Cdc42 downregulation directly impacted on the intrinsic machinery that modulate ASCD in neural stem and progenitor cells. Specifically, compromising those small GTPases led to defects in the localization of the ASCD regulators, Par-6, aPKC, Cno, and Numb, and the concomitant formation of ectopic NBs (eNBs) within brain neural lineages, a phenotype that recapitulates that of Slit-Robo1 signaling impairment. The overexpression of robo1 caused similar defects than the loss of robo1. It was also a similar phenotype than for the loss of fra, the downregulation of Abl, or the expression of a kinase dead form of Abl (unable to repress Robo1). Hence, based on all those experiments, a working model proposes that Netrin-Fra signaling would be modulating, through Abl kinase, the Robo1 signaling threshold levels necessary to regulate in turn the correct activity of the small GTPases Rac1 and Cdc42. In fact, and according to this, the expression of Rac1V12 within NBII lineages caused the formation of eNBs and led to defects in the localization of aPKC in NB and progenitor cells, a similar phenotype than that observed after overexpressing robo1 in these NBII lineages. It would be interesting to determine whether this novel function of Netrin-Fra/DCC signaling regulating ASCD is also conserved in vertebrates, and whether Robo/Rac1-Cdc42 signaling threshold levels in the above-mentioned contexts are also critical for and dependent on Netrin-DCC signaling, as was have found in Drosophila (de Torres-Jurado, 2022).

GENE STRUCTURE

Exons - 18+

PROTEIN STRUCTURE

Amino Acids - 1395

Structural Domains

robo encodes an axon guidance receptor that defines a novel subfamily of immunoglobulin superfamily proteins, which have been found to be highly conserved from fruit flies to mammals. Robo protein is a member of the Ig superfamily; Robo's ectodomain contains five Ig-like repeats followed by three fibronectin (FN) type-III repeats. The protein also begins with a putative signal sequence; in the middle it contains a transmembrane domain, and it ends with a large, 457 amino acid cytoplasmic domain (Kidd, 1998a).

The presence of five Ig and three FN domains, a transmembrane domain, and a long (457 amino acid) cytoplasmic region suggests that Robo may be a receptor and signaling molecule. The netrin receptor DCC/Frazzled/UNC-40 has a related domain structure, with 6 Ig and 4 FN domains and a similarly long cytoplasmic region. Another protein with a "5 + 3" organization is CDO (Kang, 1997). However, CDO is only distantly related to Robo (15%-33% amino acid identity between corresponding Ig and FN domains) (Kidd, 1998a).

The alignment of the Robo family proteins reveals that the first and second Ig domains are the most highly conserved portion of the extracellular domain. The cytoplasmic domains are highly divergent except for the presence of three highly conserved motifs. The consensus for the first motif is PtPYATTxhh, where x is any amino acid and h is I, L, or V. The presence of a tyrosine in the center of the motif suggests that it could be a site for phosphorylation. The other two motifs consist of runs of prolines separated by one or two amino acids. All three of these conserved sites could function as the binding sites for cytoplasmic adapter proteins, which would transmit signals generated by ligand binding (Kidd, 1998a).

The predicted Robo2 and Robo3 proteins consist of 1406 and 1342 amino acids, respectively. Robo family receptors identified to date in diverse species are characterized by an extracellular domain consisting of 5 immunoglobulin and 3 fibronectin type III repeats, and a cytoplasmic domain without any known catalytic activity but containing four short conserved motifs called CC0, CC1, CC2, and CC3. The extracellular domains of Robo2 and Robo3 are typical for the family, being 37% and 33% identical to Robo, respectively, and 49% identical to each other. In contrast, the cytoplasmic domains of Robo2 and Robo3 are unusual in that they lack both the CC2 and CC3 motifs. In Robo, these two motifs are required to prevent inappropriate midline crossing. They are thought to constitute binding sites for various cytoplasmic signaling molecules, including Enabled (Ena) and the Abl tyrosine kinase. Ena interacts primarily with the CC2 motif, and genetic data suggest that it contributes positively to Robo signaling. Abl, on the other hand, binds strongly to CC3 and negatively regulates Robo function. In vitro, Abl phosphorylates Robo on three tyrosine residues, including one in each of the CC0 and CC1 motifs. Both of these tyrosine residues are conserved in Robo2 and Robo3 (Rajagopalan, 2000).

robo2 and robo3 were initially detected by searching the Drosophila genomic database for sequences similar to robo. Homology and protein prediction programs identified two paralogs of robo. The robo2 and robo3 genes are quite close to each other and face in opposite directions on the left arm of chromosome 2 at location 22A; robo is located on the right arm of chromosome 2 at location 58F (Simpson, 2000a).

The robo2 cDNA encodes a 1540 amino acid protein with the same domain structure as Robo. The extracellular region contains five immunoglobulin-like (Ig) domains and three fibronectin type III (Fn) domains, followed by a single-pass transmembrane domain and a 450 amino acid cytoplasmic domain. The homology between Robo and Robo2 is highest in the extracellular region, ranging from 52%-53% (in the first two Ig domains) to 29% and dropping to less than 23% in the cytoplasmic region. Robo2 lacks two of the four conserved cytoplasmic motifs that Robo shares with its orthologs in other species. The first two of these motifs (CCO and CC1), which are tyrosine phosphorylation sites, are maintained, but the second two motifs (CC2 and CC3), a proline-rich Enabled binding motif and another polyproline stretch, are missing. In Robo2, the proline-rich Enabled binding motif (CC2) is replaced by a polyglutamine repeat. Interestingly, although several of the mammalian Robo orthologs contain all four conserved cytoplasmic motifs, a more divergent Robo family member, Rig-1, appears to lack CC1 (Simpson, 2000a).

Robo3 resembles Robo2 more closely than it does Robo; it too lacks two of the four cytoplasmic motifs. The genomic organization of robo2 and robo3 is very similar. Both have large first introns (23 and 17 kb, respectively), while the first intron in robo is less than 1 kb. The intron and exon sizes are similar, and which exons code for particular domains is also conserved, suggesting that robo2 and robo3 may be the result of a recent duplication (Simpson, 2000a).

Examination of the vertebrate databases shows that the identified homologs have cytoplasmic domain organization more like Drosophila Robo than like Robo2 or Robo3, indicating that Robo is closer to the common ancestor with vertebrates. Robo2 and 3 resemble each other and Robo more closely than they resemble any of the mammalian Robos or the Caenorhabditis elegans Sax3. Examination of the exon-intron boundaries within the coding regions of the three robo genes suggests that robo2 and robo3 may be the result of a recent duplication event. This phylogeny holds when the complete protein sequences of the homologs are compared, as well as when the extracellular domains are aligned and when only the most highly conserved first Ig domain is used. The Robo receptor family is related to other neural adhesion and guidance molecules like DCC/Frazzled and Neuroglian but is a distinct subgroup of the Ig superfamily (Simpson, 2000a).

date revised: 19 June 2024

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

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