When an animal embryo is developing, nerve pathways grow to connect the left and right halves of the nervous system. These pathways allow coordination between the two sides of the body—which is important for tasks such as walking and swimming. However, in order for these pathways to be properly established, the activity of certain genes must determine whether each nerve fiber (or axon) will stay on one side of the body, or cross the midline to the other side.

It is not fully understood how genes, and the proteins they encode, interact with each other to regulate the crossing of the body's midline, but the process is known to involve proteins called Robo receptors. Robo receptors are a class of proteins found on nerve fibers. Most Robo receptors work to prevent nerve fibers from crossing the midline, but some proteins in this family—including Robo2—can also promote midline crossing. Based on previous studies, it was not clear how Robo2 could have such opposing effects on different sets of nerve cells.

Evans, Santiago et al. have now explored how Robo2 regulates the development of the nervous system of fruit fly embryos, and found that Robo2 promotes midline crossing by inhibiting the activity of a closely related protein called Robo1. Further experiments unexpectedly showed that Robo2 does not promote midline crossing in the cells in which it is produced. Instead, the Robo2 receptor instructs other Robo1-producing nerve cells to cross the midline. These findings reconcile the previous, seemingly paradoxical, observations about the activity of Robo2.

Following on from these findings, one important next step will be to determine exactly how Robo2 can inhibit the activity of Robo1, such that it no longer prevents nerve fibers from crossing the midline. Determining whether similar inhibitory interactions between Robo receptors are important for the development of other tissues in the fruit fly, or in other animals, is another challenge for the future.

The secreted Slit repellents and their Roundabout (Robo) receptors constitute a repulsive axon guidance system whose function is conserved across a wide range of animal taxa including vertebrates, planarians, nematodes, and insects (Brose and Tessier-Lavigne, 2000; Evans and Bashaw, 2012). Slits are normally expressed at the midline of the central nervous system (CNS), and axons expressing Robo receptors are thus repelled from the midline (Battye et al., 1999; Brose et al., 1999; Kidd et al., 1999). Prior to crossing the midline, commissural neurons in vertebrates and insects prevent premature responsiveness to Slit by regulating the expression and activity of Robo receptors through a variety of mechanisms (Evans and Bashaw, 2010a; Neuhaus-Follini and Bashaw, 2015). For example, the divergent Robo receptor Robo3/Rig-1 in vertebrates negatively regulates the activity of the Robo1 (Roundabout1) and Robo2 (Roundabout2) receptors in pre-crossing commissural axons in the spinal cord, thereby allowing midline crossing (Sabatier et al., 2004). In Drosophila, Commissureless (Comm) antagonizes Slit-Robo1 repulsion by preventing the trafficking of the Robo1 receptor to the growth cone, instead diverting newly synthesized Robo1 into the endocytic pathway (Kidd et al., 1998b; Keleman et al., 2002, 2005). As commissural axons approach the midline, Comm expression is high, allowing axons to cross the midline (Keleman et al., 2002). Once the midline is reached, Comm is down regulated, restoring Robo1-dependent Slit sensitivity and ensuring that commissural axons do not re-cross the midline. Accordingly, loss of Robos or Slits can cause axons to ectopically cross the midline, while loss of Comm or Robo3/Rig1 prevents commissural axons from crossing (Tear et al., 1996; Kidd et al., 1998a, 1999; Long et al., 2004; Sabatier et al., 2004).

In Drosophila, the three members of the Robo receptor family (Robo1, Robo2, and Robo3) cooperate to control multiple aspects of axon guidance during embryonic development, including midline repulsion of axons and the formation of longitudinal axon pathways at specific mediolateral positions within the nerve cord. Although Robo2 contributes to promoting midline repulsion, gain-of-function genetic experiments suggest that in some contexts Robo2 can also promote midline crossing (Rajagopalan et al., 2000; Simpson et al., 2000b). More recently, endogenous roles for robo2 in promoting midline crossing were identified during the guidance of foreleg gustatory neurons in the adult, as well as during the guidance of interneurons in the embryonic CNS (Mellert et al., 2010; Spitzweck et al., 2010). Robo2's pro-crossing role in the embryo is highlighted in frazzled and Netrin mutant backgrounds, in which midline attraction is partially compromised (Spitzweck et al., 2010). In the absence of Netrin-dependent midline axon attraction, loss of robo2 (but not robo1 or robo3) leads to a dramatic disruption in midline crossing that is far more severe than the complete loss of Netrins, indicating that robo2 likely acts in parallel to Netrin-Fra to promote midline crossing (Spitzweck et al., 2010).

In a complementary series of gain-of-function experiments using a panel of chimeric receptors comprising different regions of Robo1 and Robo2 fused together, we have previously shown that Robo2's ability to promote ectopic midline crossing correlates with the presence of the first and second immunoglobulin-like domains (Ig1 and Ig2) within its extracellular domain (Evans and Bashaw, 2010b). Consistent with these observations, replacing endogenous Robo2 by homologous recombination with chimeric receptors, in which the cytoplasmic domains of the Robo1 and Robo2 receptors were exchanged, reveals that the Robo2-1 chimeric receptor (containing the extracellular region of the Robo2 receptor) can rescue the commissural guidance defect observed in Netrin, robo2 mutants more effectively than the reciprocal chimeric receptor (Spitzweck et al., 2010). However, the mechanism by which Robo2 promotes midline crossing remains unclear. We can envision two alternative models that could account for Robo2's role in promoting midline crossing of commissural axons. First, Robo2 may act as an attractive receptor to signal midline attraction in response to a ligand produced by midline glia, analogous to Frazzled/Deleted in Colorectal Cancer (DCC)'s role in Netrin-dependent midline attraction. Indeed, a role for Robo2 in mediating attractive responses to Slit has been described in the context of muscle cell migration (Kramer et al., 2001). Alternatively, Robo2 may antagonize Slit-Robo1 repulsion by preventing Robo1 from signaling in response to midline-derived Slit, similar to the proposed role of Robo3/Rig-1 in pre-crossing commissural axons in the vertebrate spinal cord (Figure 1). Although Comm is an essential regulator of Robo1 activity in Drosophila, low levels of Robo1 escape Comm-dependent sorting and can be detected on commissural axons, raising the question of whether and how the activity of these Robo1 receptors is regulated (Kidd et al., 1998a).

Figure 1

Download asset Open asset

Here, we show that in addition to its cell-autonomous role in midline repulsion, Robo2 acts non-autonomously to promote midline crossing by inhibiting canonical Slit-Robo1 repulsion and offer insights into the molecular and cellular mechanisms underlying this activity of Robo2. We find that the cytoplasmic domain of Robo2 is dispensable for its pro-crossing role, suggesting that Robo2 does not transduce a midline attractive signal, and that Robo2 over-expression can suppress comm mutants, supporting a model in which Robo2 antagonizes Slit-Robo1 repulsion. Moreover, Robo2 can bind to Robo1 in Drosophila embryonic neurons, and this biochemical interaction, like Robo2's pro-crossing role, correlates with the presence of Ig1 and Ig2. Surprisingly, we observe that Robo2 is able to promote midline crossing of axons non-cell autonomously when mis-expressed in midline cells, and we further show that Robo2 is expressed in midline glia and neurons during the early stages of commissure formation. Finally, we find that restoring Robo2 expression in midline cells can rescue midline crossing of axons in robo2, fra double mutants and that this rescue activity is dependent on Ig1 and Ig2. Together, our results indicate that Robo2 acts non-autonomously to bind to Robo1 and prevent Slit-Robo1 repulsion in pre-crossing commissural axons. This model accounts for Robo2's seemingly paradoxical roles in both promoting and inhibiting midline crossing and explains how the small amount of Robo1 present on pre-crossing commissural axons might be prevented from responding to Slit.

The midline attractive ligand Netrin and its receptor Frazzled (Fra) are the only known attractive ligand-receptor pair in Drosophila, yet many commissural axons still cross the midline in the absence of attractive Netrin-Frazzled signaling (Kolodziej et al., 1996; Mitchell et al., 1996). It has recently been demonstrated that the Robo family receptor Robo2 acts independently of Netrin and Fra to promote midline crossing, through an as yet unknown mechanism (Spitzweck et al., 2010). In robo2, fra double mutants, midline crossing of commissural axons is severely compromised, leading to thin or absent commissures, a phenotype that is qualitatively and quantitatively more severe than loss of fra alone (Figure 1). This phenotype can be observed by staining the entire axon scaffold with anti-HRP antibodies (Figure 1A–D) or by labeling a subset of commissural axons using eg-GAL4 (Garbe et al., 2007; O'Donnell and Bashaw, 2013) in robo2, fra double mutants (Figure 1F–I). To quantify the midline crossing defects, we scored the number of segments in which the EW axons, which normally cross the midline in the posterior commissure, fail to cross (Figure 1, histogram). We find that in robo2, fra double mutants approximately 70% of EW axons fail to cross the midline, compared to around 30% in fra mutants. Analysis of cell fate markers including Eg, even-skipped and zfh1 revealed no gross differences in segmentation and neuronal differentiation in robo2, fra double mutants, and although the cell bodies of the EW neurons were sometimes displaced, they were easily identifiable (data not shown). Importantly, restoring Robo2 expression by introducing one copy of an 83.9 kb robo2 bacterial artificial chromosome (BAC) transgene that includes the entire 40 kb robo2 transcription unit in this background significantly rescues the EW axon crossing defects (Figure 1E,J), confirming that this is a robo2-dependent phenotype.

Robo2's pro-crossing activity does not require its cytoplasmic domain

If Robo2 were to act as a midline attractive receptor (Figure 1, model 1), its cytoplasmic domain would likely be required for midline attraction. To test whether the Robo2 cytoplasmic domain contributes to its pro-crossing activity, we tested whether a truncated Robo2 receptor lacking its cytoplasmic domain (Robo2∆C) could promote midline crossing when mis-expressed in embryonic neurons. We found that, as with full-length Robo2, pan-neural mis-expression of Robo2∆C (with elav-GAL4) produced strong ectopic crossing of FasII-positive axons in the embryonic CNS (Figure 2). Indeed, the Robo2∆C mis-expression phenotype was stronger than full-length Robo2. In contrast, pan-neural over-expression of Robo1 did not generate ectopic crossing (Figure 2). In these experiments, all UAS-Robo transgenes are expressed from the same genomic insertion site in order to ensure that they are expressed at similar levels. Importantly, the pro-crossing activity of Robo2∆C is unlikely to be caused solely by a dominant-negative effect of the truncated receptor, as a similarly truncated form of Robo1 (Robo1∆C) has a qualitatively weaker ectopic crossing phenotype when combined with elav-GAL4 (Figure 2). Robo2∆C expression, unlike Robo1∆C, leads to ectopic crossing of all of the ipsilateral FasII axon bundles and also results in many segments exhibiting a slit-like phenotype (Figure 2). Due to the strong phenotypic effects of the targeted insertion lines of Robo1∆C and Robo2∆C, we also compared the phenotypes generated by lower levels of expression of the two truncated receptors using standard UAS inserts and observed that Robo2∆C is significantly more potent at driving ectopic midline crossing than comparable levels of the Robo1∆C receptor (Figure 3). Together these observations indicate that the pro-crossing activity of Robo2 is independent of the cytoplasmic domain and argue against the idea that Robo2 promotes midline crossing by signaling attraction.

Figure 2

Download asset Open asset

Figure 3

Download asset Open asset

Robo2's pro-crossing activity does not strictly depend on Slit binding

Slit is the canonical ligand for Robo family receptors, and all three Drosophila Robos can bind to the single Drosophila Slit (Howitt et al., 2004). To test whether Robo2's pro-crossing activity depends on its ability to bind Slit, we deleted the canonical Slit-binding domain (the first immunoglobulin-like domain: Ig1) from Robo2. As predicted by previous in vitro binding studies using Drosophila Robo1 (Brose et al., 1999; Fukuhara et al., 2008), we found that deleting Ig1 from Robo2 prevented Slit binding in cultured Drosophila cells (Figure 4 and Figure 4—figure supplement 1). Pan-neural over-expression of Robo2 produces a phenotype in which some axons are repelled from the midline, and some axons ectopically cross the midline, reflecting Robo2's two opposing activities in regulating midline crossing. As expected, deleting the Ig1 domain prevents Robo2 from signaling midline repulsion in vivo, both broadly in all neurons (when expressed pan-neurally with elav-GAL4) (Figure 5) and in a subset of commissural neurons (the EW neurons, labeled by eg-GAL4) (Figure 4), confirming that Slit binding is required for Robo2-mediated repulsion. In contrast, we found that the Robo2 receptor lacking Ig1 retained a partial ability to promote ectopic midline crossing of FasII-positive axons, indicating that the pro-crossing activity of Robo2 does not strictly depend on its ability to bind Slit (Figure 5). Notably, the ectopic crossing phenotype produced by Robo2∆Ig1 mis-expression was significantly weaker than that caused by mis-expression of full-length Robo2 (Figure 5). This result suggests that the Ig1 domain contributes to, but is not strictly required for, promotion of midline crossing by Robo2.

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 5

Download asset Open asset

Robo2 can promote crossing non-autonomously

Midline crossing is strongly reduced in robo2, fra double mutants, and pan-neural mis-expression of Robo2 can promote ectopic midline crossing. However, it is unclear whether Robo2 acts autonomously or non-autonomously to promote midline crossing. The ectodomain-dependent nature of Robo2's pro-crossing activity and our pan-neural mis-expression assays do not distinguish between these possibilities. Indeed, elav-GAL4 is transiently expressed in midline glia as well as post-mitotic neurons, preventing us from ruling out a non-neuronal contribution to the observed mis-expression phenotypes. Notably, we have never observed a clearly cell-autonomous pro-crossing phenotype caused by Robo2. We have previously described a series of experiments in which we mis-expressed full-length, truncated, or chimeric receptor variants of Robo1 and Robo2 in a subset of ipsilateral neurons (the apterous neurons, labeled by ap-GAL4) (Evans and Bashaw, 2010b). In contrast to the very different phenotypes caused by pan-neural mis-expression of these two truncated receptors (where Robo2∆C is much more potent at inducing midline crossing than Robo1∆C), Robo1∆C and Robo2∆C induce similar low levels of ectopic crossing when expressed in the apterous neurons (Figure 6A). We interpret this as a cell-autonomous dominant-negative effect of these truncated receptors.

Figure 6

Download asset Open asset

Full-length Robo2 is unable to autonomously promote midline crossing of the apterous axons (Evans and Bashaw, 2010b). Instead, Robo2 mis-expression redirects apterous axons to lateral regions of the neuropile. In the course of examining this lateral positioning activity of Robo2, we mis-expressed Robo2 in a second class of longitudinal interneurons: those labeled by Hb9-GAL4. Intriguingly, we observed two distinct phenotypes in embryos where Hb9-GAL4 drives Robo2 expression. First, Hb9-positive axons were shifted to more lateral positions within the neuropile. Second, Hb9-negative FasII-positive axons ectopically crossed the midline (Figure 6B). These results suggest that Robo2 can autonomously specify the lateral position of Hb9-positive axons while non-autonomously instructing FasII axons to cross the midline. We note that Hb9-GAL4 expression initiates earlier than ap-GAL4 and includes a larger number of neurons, including some located near the CNS midline (such as the RP motor neurons), suggesting the possibility that early midline-proximal expression of Robo2 accounts for the non-autonomous effect observed with Hb9-GAL4.

To more explicitly test whether Robo2 can promote midline crossing non-autonomously, we used slit-GAL4 to drive Robo2 expression in midline glia and neurons. We found that mis-expression of Robo2 or Robo2∆C in midline cells caused many FasII-positive axons which do not express slit-GAL4 to ectopically cross the midline, confirming that Robo2 can act non-autonomously to promote midline crossing of axons, and that this effect does not depend on the cytoplasmic domain (Figure 7). We observed a significantly milder effect with mis-expression of Robo1, suggesting that Robo2's non-cell autonomous activity is not solely a consequence of Slit titration (Figure 7). Moreover, this non-cell autonomous activity of Robo2 appears to be Ig1/Ig2-dependent, as Robo1^R2Ig1+2 but not Robo2^R1Ig1+2 promoted strong ectopic midline crossing when expressed using slit-GAL4 (Figure 7). In addition, expression of Robo2 variants missing either Ig1 or Ig2 with slit-Gal4 did not result in any ectopic midline crossing (Figure 7). The requirement for both Ig1 and Ig2 in this context contrasts with our findings with pan-neural mis-expression, in which Robo2∆Ig1 retained some pro-crossing activity. However, it is worth noting that the phenotype generated by elav-GAL4 mis-expression of Robo2 is stronger than that generated by slit-GAL4, perhaps because slit-GAL4 is expressed in a much smaller number of cells.

Figure 7

Download asset Open asset

robo2 is expressed in midline glia and neurons during commissure formation

Robo2 can promote midline crossing when expressed in a subset of embryonic neurons and glia, and endogenous robo2 contributes to midline crossing of commissural axons. During embryogenesis, robo2 expression is dynamically regulated: it is broadly expressed in neurons during early stages of CNS development, including transient expression in a number of ipsilateral pioneer neurons, and later becomes restricted to neurons whose axons form longitudinal pathways in the lateral regions of the neuropile (Simpson et al., 2000a). To gain additional insight into Robo2's role in promoting midline crossing of commissural neurons, we examined robo2 mRNA and protein expression in embryos during the early stages of axon path finding, when the first commissural axons are crossing the midline (stages 12–13). Using fluorescent mRNA in situ hybridization, we were able to detect robo2 mRNA expression in cells labeled by slit-GAL4 in late stage 12 embryos, around the time that pioneer commissural axons are crossing the midline (Figure 8B). robo2 mRNA expression persists through the end of stage 13, but is no longer detectable by stage 14; thus, midline expression of robo2 coincides with the time when most commissural axons are crossing the midline (Figure 8, Figure 8—figure supplement 1). Moreover, a robo2-GAL4 enhancer-trap insertion is expressed in midline glia at this time, as detected by anti-GFP staining in robo2-GAL4, UAS-TauMycGFP embryos (Figure 8A). Expression of UAS-HARobo2 with robo2-Gal4 and detection of transgenic Robo2 with anti-HA reveals an expression pattern that closely resembles the endogenous pattern of Robo2 protein, suggesting that the robo2-GAL4 faithfully reports Robo2 expression (data not shown). In addition, we could detect weak expression of Robo2 protein produced by an HA-tagged knock-in allele of robo2 (Spitzweck et al., 2010) in a subset of slit-GAL4 expressing cells at stage 12, confirming that Robo2 protein is produced in midline cells during the stages of commissural axon path finding, and raising the possibility that Robo2 endogenously acts in these cells to promote midline crossing of commissural axons (Figure 8B).

Figure 8 with 1 supplement see all

Download asset Open asset

Midline expression of Robo2 rescues the commissural defects in fra, robo2 mutants

Robo2 can promote midline crossing non-autonomously, and endogenous robo2 expression can be detected in slit-GAL4-expressing cells as well as in contralateral and ipsilateral neurons during the initial stages of commissure formation (Figure 8 and data not shown). Our ability to partially rescue midline crossing in robo2, fra double mutants with the robo2 BAC confirms that this is a robo2-specific phenotype, but does not address in which cells robo2 acts to instruct commissural axons to cross the midline. To address this question, we attempted to rescue robo2's endogenous pro-crossing activity by restoring robo2 expression in restricted subsets of cells in robo2, fra double mutants. We first expressed Robo2 in the commissural EW neurons (using eg-GAL4) in robo2, fra double mutants. We found that neither full-length Robo2 nor Robo2ΔC can rescue midline crossing when expressed autonomously in the EW neurons (Figure 9), suggesting that Robo2 may not act cell autonomously to promote midline crossing. We next attempted to rescue midline crossing in robo2, fra double mutants by expressing Robo2 using slit-GAL4. Strikingly, we found that driving Robo2 expression in these cells significantly restores posterior commissure formation, as measured by anti-HRP staining (Figure 8C–G). Furthermore, this effect is dependent on Ig1 and Ig2 (Figure 8C–G). Cell-type specific loss of function experiments will be necessary to confirm the site of Robo2's endogenous activity, and our attempts to recapitulate the robo2, fra phenotype by over-expression of RNAi transgenes have so far been unsuccessful, likely because of the well known difficulty of achieving sufficient knockdown in embryonic stages. Nevertheless, our results suggest that Robo2 promotes midline crossing non-cell autonomously, and may act in midline glia and neurons, where it is expressed during the stages of commissural axon path finding, to promote midline crossing of commissural axons.

Figure 9

Download asset Open asset

Robo2 can antagonize Slit-Robo1 repulsion

Robo2 can promote midline crossing of axons independently of its cytoplasmic domain and its Slit-binding Ig1 domain, suggesting that Robo2 does not promote crossing by acting as an attractive signaling receptor or solely by titrating Slit. Does Robo2 antagonize Slit-Robo1 repulsion through another mechanism? In order to test this hypothesis, we took advantage of comm mutants, which provide a genetic background in which hyperactive Slit-Robo1 signaling completely prevents midline crossing. In comm mutants, endogenous Robo1 is inappropriately trafficked to the growth cone plasma membrane in pre-crossing commissural axons, triggering premature Slit repulsion and preventing commissure formation. We reasoned that if Robo2 antagonizes Slit-Robo1 repulsion, then Robo2 mis-expression might restore midline crossing in comm mutant embryos. Indeed, pan-neural mis-expression of Robo2 with elav-GAL4 significantly restored commissure formation in comm mutant embryos (Figure 10). To specifically test the Ig1/Ig2-dependence of Robo2's pro-crossing activity in this assay, we mis-expressed our Ig1+2 chimeric receptors (Robo1^R2Ig1+2 and Robo2^R1Ig1+2) with elav-GAL4 in comm mutant embryos. We found that pan-neural mis-expression of Robo1^R2Ig1+2 in comm mutant embryos strongly suppressed the commissureless phenotype and restored midline crossing of many axons, as assayed by anti-HRP antibody staining, while mis-expression of Robo2^R1Ig1+2 had a much milder effect (Figure 10). These results suggest that Robo2 promotes midline crossing in an Ig1/Ig2-dependent manner by antagonizing canonical Slit-Robo1 repulsion.

Figure 10

Download asset Open asset

We were also able to suppress the comm mutant phenotype by over-expressing Robo2 using slit-GAL4, and this effect was fully dependent on both Ig1 and Ig2 of Robo2 (Figure 10). This is consistent with our observations that Robo2 can act non-cell autonomously to promote ectopic midline crossing (Figure 7) and rescue midline crossing defects (Figure 8) in an Ig1/2-dependent manner. Of note, the suppressive effect of Robo2 expression in comm mutants is much greater when expressed in midline cells than when expressed pan-neurally (Figure 10). This is likely because when expressed pan-neurally, in addition to its pro-crossing activity, full-length Robo2 also has repulsive activity. In contrast, when expressed in midline cells, Robo2 would be unable to act as a repulsive receptor.

Robo2 binds to Robo1 in vivo and the interaction depends on Ig1 and Ig2

As shown above, Robo2 is able to antagonize Slit-Robo1 repulsion in an Ig1/Ig2-dependent manner. One possibility is that Robo2 may form an inhibitory receptor–receptor complex with Robo1 to prevent it from signaling midline repulsion in response to Slit. If this is the case, we reasoned that we might be able to detect a physical interaction between Robo2 and Robo1 in embryonic protein extracts. To test this idea, we mis-expressed epitope-tagged forms of Robo1 and Robo2 in Drosophila embryonic neurons with elav-GAL4 and looked for physical interactions by co-immunoprecipitation (Figure 11). We found that Robo1-myc and HA-Robo2 co-immunoprecipitated from embryonic lysates when both were expressed in embryonic neurons (Figure 11A). Interactions were also observed between Robo1 and the closely related Robo3 receptor but not with a similarly tagged and structurally related Fra receptor (Figure 11A). As we would predict from our gain of function experiments, Robo2's ability to bind to Robo1 is independent of its cytoplasmic domain (Figure 11, Figure 11—figure supplement 1). Importantly, Robo2's ability to bind Robo1 depends on the Ig1–Ig2 region of Robo2, as Robo1^R2Ig1+2 was readily co-immunoprecipitated with Robo1, while binding between Robo1 and the reciprocal receptor Robo2^R1Ig1+2 was only weakly detected (Figure 11B). Consistently, deleting both of the Ig1 and Ig2 domains from Robo2 results in a diminished interaction with Robo1 in vivo and in vitro (Figure 11, Figure 11—figure supplement 1). The fact that not all binding is eliminated when both Ig1 and Ig2 are deleted suggests that other regions of the Robo2 receptor may contribute to the interaction with Robo1. Thus, we see a correlation between the presence of the Ig1 and Ig2 domains, a biochemical interaction with Robo1, and pro-crossing activity in the Robo2 receptor. These observations suggest that Robo2 may promote midline crossing through inhibitory interactions with Robo1, likely mediated at least in part by the Robo2 Ig1 and Ig2 domain. Of note, the Ig2 domain is essential for Robo2's pro-crossing activity but is not required for the interaction with Robo1 or Slit, suggesting the existence of an Ig2-specific activity that is distinct from the ability to bind Robo1 or Slit. One possible mechanism that could explain how receptor–receptor interactions could prevent Robo1 signaling is through blocking the access of Slit to the Ig1 region of Robo1. Deleting Robo1's Ig1 domain does not significantly attenuate the interaction with Robo2, but further experiments will be necessary to determine if Robo2 interferes with Robo1's interaction with Slit (Figure 11, Figure 11—figure supplement 1).

Figure 11 with 1 supplement see all

Download asset Open asset

Our biochemical experiments examining receptor–receptor interactions when the Robo receptors are expressed in all neurons do not distinguish between cis and trans interactions. As we observed that Robo2 is able to non-cell autonomously inhibit Robo1 repulsion and promote midline crossing, we reasoned that we might be able to detect physical interactions between Robo1 and Robo2 receptors when they are presented in trans. We tested this prediction by transfecting Drosophila cultured S2R+ cells with either Robo1-myc or HA-tagged Robo2 and assaying for physical interactions by co-immunoprecipitation. Although we detected strong interactions between Robo1 and Robo2 in co-transfected cells, we could not detect interactions in cells that were transfected separately and mixed together (Figure 11, Figure 11—figure supplement 1 and data not shown). However, when we mixed the membrane lysates of cells that were transfected separately, we observed that Robo1 readily co-immunoprecipitated Robo2, in an Ig1-2 dependent manner (Figure 11). These data suggest that physical interactions can occur between Robo1 and Robo2 receptors that are expressed in different cells and are consistent with the possibility of a physical interaction occurring across cell membranes in vivo. It is important to recognize that binding detected with mixed cell lysates could occur in either cis or trans. Future work should more rigorously evaluate the potential for trans interactions.

Robo2's Ig2 domain is required for its endogenous activity in promoting midline crossing

Our data are consistent with a non-autonomous requirement for Ig1 and Ig2 of Robo2 in antagonizing Robo1 to promote midline crossing. However, the genetic data supporting this model arise from gain of function and rescue experiments using GAL4/UAS over-expression. In order to more rigorously address the endogenous requirement for Robo2 in promoting midline crossing, we generated modified BACs and evaluated the ability of either wild-type Robo2 or Robo2∆Ig2 to restore midline crossing in robo2, fra double mutants, when expressed under robo2's endogenous control elements. As Ig1 is required for both Robo2's pro-crossing activity and for its repulsive signaling output, the Robo2ΔIg2 variant provides a more specific reagent for testing our model. Therefore, we modified the original Robo2 BAC by recombineering to insert wild-type Robo2 cDNA or Robo2∆Ig2 cDNA, and introduced these BAC transgenes into robo2, fra double mutants. We determined the rescuing activity of each BAC through two assays: first, by scoring midline crossing of EW axons labeled by eg-GAL4, and second, by analyzing commissure formation in embryos stained with anti-HRP to label all axons (Figure 12).

Figure 12

Download asset Open asset

We found that the ability of the Robo2 BAC to rescue midline crossing defects in robo2, fra double mutants was strongly impaired by deleting the Ig2 domain. In the EW crossing assay, one copy of the Robo2 FL cDNA BAC provides a significant rescue of robo2, fra double mutants at stage 16, whereas one copy of the Robo2ΔIg2 BAC has no effect (Figure 12A–D). Of note, removing one allele of robo2 enhances midline crossing defects in fra mutants, explaining in part the incomplete rescue (Figure 12). In addition, it is likely that the Robo2 BAC does not contain all of the regulatory elements required for robo2's pro-crossing function, as one copy of the BAC does not restore EW crossing back to the levels of fra mutants heterozygous for robo2 (Figure 12).

We also assessed the ability of the BAC transgenes to rescue midline crossing defects when analyzing all axons using anti-HRP. By this method, we see a robust rescue in posterior commissure (PC) formation in robo2, fra double mutant embryos with one copy of the Robo2 cDNA BAC compared to controls (Figure 12E–H). In contrast, the Robo2ΔIg2 BAC provides a much weaker rescue (Figure 12G). The partial rescue by the Robo2ΔIg2 BAC in this assay suggests that the severe fra, robo2 phenotype is due to the combined requirement for multiple activities of Robo2, including one that is Ig2-independent. Nevertheless, these data unambiguously reveal an endogenous requirement for Robo2's Ig2 domain during commissural axon guidance. Importantly, the Robo2∆Ig2 BAC fully rescues Robo2's repulsive activity at the midline (data not shown), further demonstrating that the Ig2 domain is specifically required for Robo2 to successfully promote midline crossing, but not for other known activities of the Robo2 receptor. Taken together, these results demonstrate a requirement for Robo2's Ig2 domain in promoting midline crossing when expressed under its endogenous control elements and strongly support the model that Robo2 promotes midline crossing of commissural axons by antagonizing repulsion through an Ig1/Ig2-mediated inhibitory interaction with Robo1.

In this manuscript, we have described a role for Robo2 in promoting midline crossing through inhibition of Slit-Robo1 repulsion. Loss of function experiments point to an endogenous requirement for Robo2 in promoting midline crossing. Additional genetic analyses indicate that Robo2 can antagonize Robo1 in the absence of its cytoplasmic domain and that this inhibitory effect can be generated by non-cell autonomous expression of Robo2. These observations, together with the demonstration that Robo2 variants that promote midline crossing are potent suppressors of comm mutants, support the model that Robo2 inhibits Slit-Robo1 repulsion, rather than acting as a receptor that promotes midline axon attraction. Biochemical and gain of function genetic analyses show that Robo2 can bind to Robo1 in vivo through its Ig1 and Ig2 domains and that this binding interaction correlates with Robo2's pro-crossing activity. Furthermore, cell type specific rescue experiments and analysis of Robo2 mRNA and protein expression are consistent with a requirement for Robo2 in midline cells and support an endogenous requirement for Robo2's Ig2 domain in promoting midline crossing. Taken together, the data in this manuscript support the model that Robo2 expressed in cells other than commissural neurons acts to inhibit Robo1 receptor activity through extracellular domain binding interactions, and that this activity ensures the precise execution of midline guidance (Figure 13). Our model reconciles two previously confounding observations: one, that a small amount of Robo1 protein is detectable on commissural axons as they cross the midline, yet this pool of Robo1 is unable to signal midline repulsion; and two, that the single known isoform of Robo2 can act to both promote and inhibit midline crossing.

Figure 13

Download asset Open asset

1. 1. Ahn VE
   2. Chu ML
   3. Choi HJ
   4. Tran D
   5. Abo A
   6. Weis WI

   (2011) Structural basis of Wnt signaling inhibition by Dickkopf binding to LRP5/6

   Developmental Cell 21:862–873.

   https://doi.org/10.1016/j.devcel.2011.09.003

   • Google Scholar
2. 1. Bashaw GJ
   2. Kidd T
   3. Murray D
   4. Pawson T
   5. Goodman CS

   (2000) Repulsive axon guidance: Abelson and enabled play opposing roles downstream of the roundabout receptor

   Cell 101:703–715.

   https://doi.org/10.1016/S0092-8674(00)80883-1

   • Google Scholar
3. 1. Battye R
   2. Stevens A
   3. Jacobs JR

   (1999)

   Axon repulsion from the midline of the Drosophila CNS requires slit function

   Development 126:2475–2481.

   • Google Scholar
4. 1. Brose K
   2. Bland KS
   3. Wang KH
   4. Arnott D
   5. Henzel W
   6. Goodman CS
   7. Tessier-Lavigne M
   8. Kidd T

   (1999) Slit proteins bind Robo receptors and have an evolutionarily conserved role in repulsive axon guidance

   Cell 96:795–806.

   https://doi.org/10.1016/S0092-8674(00)80590-5

   • Google Scholar
5. 1. Brose K
   2. Tessier-Lavigne M

   (2000) Slit proteins: key regulators of axon guidance, axonal branching, and cell migration

   Current Opinion in Neurobiology 10:95–102.

   https://doi.org/10.1016/S0959-4388(99)00066-5

   • Google Scholar
6. 1. Chen S
   2. Bubeck D
   3. Macdonald BT
   4. Liang WX
   5. Mao JH
   6. Malinauskas T
   7. Llorca O
   8. Aricescu AR
   9. Siebold C
   10. He X
   11. Jones EY

   (2011) Structural and functional studies of LRP6 ectodomain reveal a platform for Wnt signaling

   Developmental Cell 21:848–861.

   https://doi.org/10.1016/j.devcel.2011.09.007

   • Google Scholar
7. 1. Coleman HA
   2. Labrador JP
   3. Chance RK
   4. Bashaw GJ

   (2010) The Adam family metalloprotease Kuzbanian regulates the cleavage of the roundabout receptor to control axon repulsion at the midline

   Development 137:2417–2426.

   https://doi.org/10.1242/dev.047993

   • Google Scholar
8. 1. del Alamo D
   2. Rouault H
   3. Schweisguth F

   (2011) Mechanism and significance of cis-inhibition in Notch signalling

   Current Biology 21:R40–R47.

   https://doi.org/10.1016/j.cub.2010.10.034

   • Google Scholar
9. 1. Evans TA
   2. Bashaw GJ

   (2010a) Axon guidance at the midline: of mice and flies

   Current Opinion in Neurobiology 20:79–85.

   https://doi.org/10.1016/j.conb.2009.12.006

   • Google Scholar
10. 1. Evans TA
    2. Bashaw GJ

    (2010b) Functional diversity of Robo receptor immunoglobulin domains promotes distinct axon guidance decisions

    Current Biology 20:567–572.

    https://doi.org/10.1016/j.cub.2010.02.021

    • Google Scholar
11. 1. Evans TA
    2. Bashaw GJ

    (2012) Slit/Robo-mediated axon guidance in Tribolium and Drosophila: divergent genetic programs build insect nervous systems

    Developmental Biology 363:266–278.

    https://doi.org/10.1016/j.ydbio.2011.12.046

    • Google Scholar
12. 1. Fan X
    2. Labrador JP
    3. Hing H
    4. Bashaw GJ

    (2003) Slit stimulation recruits Dock and Pak to the roundabout receptor and increases Rac activity to regulate axon repulsion at the CNS midline

    Neuron 40:113–127.

    https://doi.org/10.1016/S0896-6273(03)00591-9

    • Google Scholar
13. 1. Fukuhara N
    2. Howitt JA
    3. Hussain SA
    4. Hohenester E

    (2008) Structural and functional analysis of slit and heparin binding to immunoglobulin-like domains 1 and 2 of Drosophila Robo

    The Journal of Biological Chemistry 283:16226–16234.

    https://doi.org/10.1074/jbc.M800688200

    • Google Scholar
14. 1. Garbe DS
    2. O'Donnell M
    3. Bashaw GJ

    (2007) Cytoplasmic domain requirements for Frazzled-mediated attractive axon turning at the Drosophila midline

    Development 134:4325–4334.

    https://doi.org/10.1242/dev.012872

    • Google Scholar
15. 1. Ghiglione C
    2. Amundadottir L
    3. Andresdottir M
    4. Bilder D
    5. Diamonti JA
    6. Noselli S
    7. Perrimon N
    8. Carraway IK

    (2003) Mechanism of inhibition of the Drosophila and mammalian EGF receptors by the transmembrane protein Kekkon 1

    Development 130:4483–4493.

    https://doi.org/10.1242/dev.00617

    • Google Scholar
16. 1. Gilestro GF

    (2008) Redundant mechanisms for regulation of midline crossing in Drosophila

    PLOS ONE 3:e3798.

    https://doi.org/10.1371/journal.pone.0003798

    • Google Scholar
17. 1. Howitt JA
    2. Clout NJ
    3. Hohenester E

    (2004) Binding site for Robo receptors revealed by dissection of the leucine-rich repeat region of Slit

    The EMBO Journal 23:4406–4412.

    https://doi.org/10.1038/sj.emboj.7600446

    • Google Scholar
18. 1. Kao TJ
    2. Kania A

    (2011) Ephrin-mediated cis-attenuation of Eph receptor signaling is essential for spinal motor axon guidance

    Neuron 71:76–91.

    https://doi.org/10.1016/j.neuron.2011.05.031

    • Google Scholar
19. 1. Keleman K
    2. Rajagopalan S
    3. Cleppien D
    4. Teis D
    5. Paiha K
    6. Huber LA
    7. Technau GM
    8. Dickson BJ

    (2002) Comm sorts robo to control axon guidance at the Drosophila midline

    Cell 110:415–427.

    https://doi.org/10.1016/S0092-8674(02)00901-7

    • Google Scholar
20. 1. Keleman K
    2. Ribeiro C
    3. Dickson BJ

    (2005) Comm function in commissural axon guidance: cell-autonomous sorting of Robo in vivo

    Nature Neuroscience 8:156–163.

    https://doi.org/10.1038/nn1388

    • Google Scholar
21. 1. Kidd T
    2. Bland KS
    3. Goodman CS

    (1999) Slit is the midline repellent for the robo receptor in Drosophila

    Cell 96:785–794.

    https://doi.org/10.1016/S0092-8674(00)80589-9

    • Google Scholar
22. 1. Kidd T
    2. Brose K
    3. Mitchell KJ
    4. Fetter RD
    5. Tessier-Lavigne M
    6. Goodman CS
    7. Tear G

    (1998a) Roundabout controls axon crossing of the CNS midline and defines a novel subfamily of evolutionarily conserved guidance receptors

    Cell 92:205–215.

    https://doi.org/10.1016/S0092-8674(00)80915-0

    • Google Scholar
23. 1. Kidd T
    2. Russell C
    3. Goodman CS
    4. Tear G

    (1998b) Dosage-sensitive and complementary functions of roundabout and commissureless control axon crossing of the CNS midline

    Neuron 20:25–33.

    https://doi.org/10.1016/S0896-6273(00)80431-6

    • Google Scholar
24. 1. Klein DE
    2. Nappi VM
    3. Reeves GT
    4. Shvartsman SY
    5. Lemmon MA

    (2004) Argos inhibits epidermal growth factor receptor signalling by ligand sequestration

    Nature 430:1040–1044.

    https://doi.org/10.1038/nature02840

    • Google Scholar
25. 1. Klein DE
    2. Stayrook SE
    3. Shi F
    4. Narayan K
    5. Lemmon MA

    (2008) Structural basis for EGFR ligand sequestration by Argos

    Nature 453:1271–1275.

    https://doi.org/10.1038/nature06978

    • Google Scholar
26. 1. Kolodziej PA
    2. Timpe LC
    3. Mitchell KJ
    4. Fried SR
    5. Goodman CS
    6. Jan LY
    7. Jan YN

    (1996) Frazzled encodes a Drosophila member of the DCC immunoglobulin subfamily and is required for CNS and motor axon guidance

    Cell 87:197–204.

    https://doi.org/10.1016/S0092-8674(00)81338-0

    • Google Scholar
27. 1. Kramer SG
    2. Kidd T
    3. Simpson JH
    4. Goodman CS

    (2001) Switching repulsion to attraction: changing responses to slit during transition in mesoderm migration

    Science 292:737–740.

    https://doi.org/10.1126/science.1058766

    • Google Scholar
28. 1. Kraut R
    2. Zinn K

    (2004) Roundabout 2 regulates migration of sensory neurons by signaling in trans

    Current Biology 14:1319–1329.

    https://doi.org/10.1016/j.cub.2004.07.052

    • Google Scholar
29. 1. Long H
    2. Sabatier C
    3. Ma L
    4. Plump A
    5. Yuan W
    6. Ornitz DM
    7. Tamada A
    8. Murakami F
    9. Goodman CS
    10. Tessier-Lavigne M

    (2004) Conserved roles for Slit and Robo proteins in midline commissural axon guidance

    Neuron 42:213–223.

    https://doi.org/10.1016/S0896-6273(04)00179-5

    • Google Scholar
30. 1. Mellert DJ
    2. Knapp JM
    3. Manoli DS
    4. Meissner GW
    5. Baker BS

    (2010) Midline crossing by gustatory receptor neuron axons is regulated by fruitless, doublesex and the Roundabout receptors

    Development 137:323–332.

    https://doi.org/10.1242/dev.045047

    • Google Scholar
31. 1. Mitchell KJ
    2. Doyle JL
    3. Serafini T
    4. Kennedy TE
    5. Tessier-Lavigne M
    6. Goodman CS
    7. Dickson BJ

    (1996) Genetic analysis of Netrin genes in Drosophila: Netrins guide CNS commissural axons and peripheral motor axons

    Neuron 17:203–215.

    https://doi.org/10.1016/S0896-6273(00)80153-1

    • Google Scholar
32. 1. Neuhaus-Follini A
    2. Bashaw GJ

    (2015) Crossing the embryonic midline: molecular mechanisms regulating axon responsiveness at an intermediate target

    Wiley Interdisciplinary Reviews. Developmental Biology 4:377–389.

    https://doi.org/10.1002/wdev.185

    • Google Scholar
33. 1. Niehrs C

    (2006) Function and biological roles of the Dickkopf family of Wnt modulators

    Oncogene 25:7469–7481.

    https://doi.org/10.1038/sj.onc.1210054

    • Google Scholar
34. 1. O'Donnell MP
    2. Bashaw GJ

    (2013) Src inhibits midline axon crossing independent of Frazzled/Deleted in Colorectal Carcinoma (DCC) receptor tyrosine phosphorylation

    The Journal of Neuroscience 33:305–314.

    https://doi.org/10.1523/JNEUROSCI.2756-12.2013

    • Google Scholar
35. 1. Özkan E
    2. Carrillo RA
    3. Eastman CL
    4. Weiszmann R
    5. Waghray D
    6. Johnson KG
    7. Zinn K
    8. Celniker SE
    9. Garcia KC

    (2013) An extracellular interactome of immunoglobulin and LRR proteins reveals receptor-ligand networks

    Cell 154:228–239.

    https://doi.org/10.1016/j.cell.2013.06.006

    • Google Scholar
36. 1. Rajagopalan S
    2. Vivancos V
    3. Nicolas E
    4. Dickson BJ

    (2000) Selecting a longitudinal pathway: robo receptors specify the lateral position of axons in the Drosophila CNS

    Cell 103:1033–1045.

    https://doi.org/10.1016/S0092-8674(00)00207-5

    • Google Scholar
37. 1. Sabatier C
    2. Plump AS
    3. Le M
    4. Brose K
    5. Tamada A
    6. Murakami F
    7. Lee EY
    8. Tessier-Lavigne M

    (2004) The divergent Robo family protein rig-1/Robo3 is a negative regulator of slit responsiveness required for midline crossing by commissural axons

    Cell 117:157–169.

    https://doi.org/10.1016/S0092-8674(04)00303-4

    • Google Scholar
38. 1. Santiago C
    2. Labrador JP
    3. Bashaw GJ

    (2014) The homeodomain transcription factor Hb9 controls axon guidance in Drosophila through the regulation of Robo receptors

    Cell Reports 7:153–165.

    https://doi.org/10.1016/j.celrep.2014.02.037

    • Google Scholar
39. 1. Simpson JH
    2. Bland KS
    3. Fetter RD
    4. Goodman CS

    (2000a) Short-range and long-range guidance by Slit and its Robo receptors: a combinatorial code of Robo receptors controls lateral position

    Cell 103:1019–1032.

    https://doi.org/10.1016/S0092-8674(00)00206-3

    • Google Scholar
40. 1. Simpson JH
    2. Kidd T
    3. Bland KS
    4. Goodman CS

    (2000b) Short-range and long-range guidance by slit and its Robo receptors. Robo and Robo2 play distinct roles in midline guidance

    Neuron 28:753–766.

    https://doi.org/10.1016/S0896-6273(00)00151-3

    • Google Scholar
41. 1. Spitzweck B
    2. Brankatschk M
    3. Dickson BJ

    (2010) Distinct protein domains and expression patterns confer divergent axon guidance functions for Drosophila Robo receptors

    Cell 140:409–420.

    https://doi.org/10.1016/j.cell.2010.01.002

    • Google Scholar
42. 1. Tear G
    2. Harris R
    3. Sutaria S
    4. Kilomanski K
    5. Goodman CS
    6. Seeger MA

    (1996) Commissureless controls growth cone guidance across the Cns midline in Drosophila and encodes a novel membrane protein

    Neuron 16:501–514.

    https://doi.org/10.1016/S0896-6273(00)80070-7

    • Google Scholar
43. 1. Venken KJ
    2. Carlson JW
    3. Schulze KL
    4. Pan H
    5. He Y
    6. Spokony R
    7. Wan KH
    8. Koriabine M
    9. de Jong PJ
    10. White KP
    11. Bellen HJ
    12. Hoskins RA

    (2009) Versatile P[acman] BAC libraries for transgenesis studies in Drosophila melanogaster

    Nature Methods 6:431–434.

    https://doi.org/10.1038/nmeth.1331

    • Google Scholar
44. 1. Walsh DW
    2. Godson C
    3. Brazil DP
    4. Martin F

    (2010) Extracellular BMP-antagonist regulation in development and disease: tied up in knots

    Trends in Cell Biology 20:244–256.

    https://doi.org/10.1016/j.tcb.2010.01.008

    • Google Scholar
45. 1. Yang L
    2. Bashaw GJ

    (2006) Son of sevenless directly links the Robo receptor to rac activation to control axon repulsion at the midline

    Neuron 52:595–607.

    https://doi.org/10.1016/j.neuron.2006.09.039

    • Google Scholar
46. 1. Yang L
    2. Garbe DS
    3. Bashaw GJ

    (2009) A frazzled/DCC-dependent transcriptional switch regulates midline axon guidance

    Science 324:944–947.

    https://doi.org/10.1126/science.1171320

    • Google Scholar
47. 1. Yaron A
    2. Sprinzak D

    (2012) The cis side of juxtacrine signaling: a new role in the development of the nervous system

    Trends in Neurosciences 35:230–239.

    https://doi.org/10.1016/j.tins.2011.12.003

    • Google Scholar

Author details

1. Timothy A Evans

   1. Department of Neuroscience, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
   2. Department of Biological Sciences, University of Arkansas, Fayetteville, United States

   Contribution

   TAE, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Contributed equally with

   Celine Santiago

   For correspondence

   evanst@uark.edu

   Competing interests

   The authors declare that no competing interests exist.

2. Celine Santiago

   Department of Neuroscience, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States

   Contribution

   CS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Contributed equally with

   Timothy A Evans

   Competing interests

   The authors declare that no competing interests exist.

3. Elise Arbeille

   Department of Neuroscience, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States

   Contribution

   EA, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

4. Greg J Bashaw

   Department of Neuroscience, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States

   Contribution

   GJB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   For correspondence

   gbashaw@mail.med.upenn.edu

   Competing interests

   The authors declare that no competing interests exist.

• 2,760

  views

• 503

  downloads

• 70

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.