Nature Cell Biology
4, 798 - 805 (2002)
Published online: 23 September 2002; | doi:10.1038/ncb858
There is a Corrigendum (November 2002) associated with this Article.
Activation of the repulsive receptor Roundabout inhibits N-cadherin-mediated cell adhesionJinseol Rhee1, Najmus S. Mahfooz2, Carlos Arregui1, Jack Lilien1, Janne Balsamo1
& Mark F.A. VanBerkum21 Department of Biological Sciences, The University of Iowa, Iowa City, IA 52242, USA 2 Department of Biological Sciences, Wayne State University, Detroit, MI 48202, USA
Correspondence should be addressed to Jack Lilien jack-lilien@uiowa.eduThe formation of axon trajectories requires integration of local adhesive interactions with directional information from attractive and repulsive cues. Here, we show that these two types of information are functionally integrated; activation of the transmembrane receptor Roundabout (Robo) by its ligand, the secreted repulsive guidance cue Slit, inactivates N-cadherin-mediated adhesion. Loss of N-cadherin-mediated adhesion is accompanied by tyrosine phosphorylation of  -catenin and its loss from the N-cadherin complex, concomitant with the formation of a supramolecular complex containing Robo, Abelson (Abl) kinase and N-cadherin. Local formation of such a receptor complex is an ideal mechanism to steer the growth cone while still allowing adhesion and growth in other directions.The development of axon pathways is directed by many distinct, and potentially competing, receptor systems. Adhesion receptors provide the traction necessary to move the growth cone forward by coupling the external substrate with the internal cytoskeleton and secreted guidance cues provide directional information by either attracting or repelling the growth cone. We are just beginning to appreciate the complex interactions that occur in the growth cone, where information is integrated from many diverse guidance cues1. However, little attention has been given to analysis of how the function of specific adhesion receptors is integrated with directional information from secreted guidance cues. In this study, we demonstrate for the first time that binding of the secreted guidance cue, Slit, to its receptor, Robo, inactivates the adhesion molecule, N-cadherin.
N-cadherin mediates homophilic, calcium-dependent cell−cell adhesion2. In vertebrates N-cadherin, together with other cadherins, is distributed in a region-specific pattern throughout the central nervous system (CNS)3 and may be essential for the development of specific neuronal connections4. In vitro analyses indicate that N-cadherin supports neurite outgrowth when used as an artificial substrate5 or when present on the surface of cultured cells6,
7,
8,
9. Drosophila melanogaster N-cadherin mutants also have defects in axon fasciculation and tracking10. N-cadherin is associated with the actin-containing cytoskeleton, an association that is mediated by  -catenin bound to cadherin, and  -catenin, which functions as a link between -catenin and actin11. This association with actin is critical for the formation of stable adhesions and may be involved in providing the traction essential for growth cone motility.
Slit and Robo are also co-expressed in the CNS of many organisms12,
13. In vertebrates, distribution patterns, mutant phenotypes and in vitro analyses indicate that Slit−Robo interactions are involved in axon guidance decisions in many regions of the developing CNS14, including the brain15 and retina16,
17,
18. In Drosophila, an absence of Robo results in abnormal commissure formation: axons cross and recross the midline, forming the circular pathways for which the mutant is named19. Interactions between Slit and Robo are critical for determining the pattern of commissural neurons20,
21 through the initiation of signals that regulate remodelling of the actin cytoskeleton in the growth cone22. An additional way to determine directionality is to dictate what adhesion molecules remain coupled to the cytoskeleton and also where within the growth cone this occurs. Here, we demonstrate that activation of Robo results in the formation of a receptor complex between Robo and N-cadherin, uncoupling N-cadherin from its association with the cytoskeleton. Thus, directional information provided by a repulsive cue is directly converted into decreased traction in the growth cone.
Results
Activated Robo inhibits N-cadherin function.
To determine if Robo directly regulates the function of N-cadherin, full-length Drosophila Robo was expressed in mouse fibroblast L-cells stably expressing N-cadherin (LN cells). LN cells have been used extensively to dissect the function of N-cadherin, including its ability to couple to the actin cytoskeleton23. Several clones of stable transfectants were analysed for expression of full-length Robo (Robo/fl cells) by western blot using a monoclonal antibody, 13C9, which is specific to the extracellular domain of Drosophila Robo12 (Fig. 1a shows the results for one representative clone). N-cadherin-mediated adhesion is unimpaired in Robo/fl cells in the absence of Slit, as compared with mock-transfected LN cells (Fig. 1b). To determine the effect of Robo activation on N-cadherin function, we affinity purified Slit from Drosophila S2 cell conditioned medium24. When Robo/fl cells were treated with Slit, N-cadherin-mediated adhesion was reduced to levels detected in control LN cells treated with the NCD2 antibody, which blocks N-cadherin function (Fig. 1b; + NCD2).
 | | |
The activation of Robo by Slit may inhibit N-cadherin-mediated adhesion by directly interfering with extracellular cadherin−cadherin interactions. Alternatively, activation of Robo by Slit may generate signals that ultimately block N-cadherin function from within the cell. To distinguish between these possibilities, a construct encoding only the cytoplasmic domain of Drosophila or mouse Robo tagged with a haemagglutinin (HA) epitope was introduced into LN cells and stable cell lines were established, termed Robo/cdD and Robo/cdM, respectively. As with full-length Robo, several clones were examined for Robo/cd expression (Fig. 1a, right). Expression of the cytoplasmic domain of either Drosophila or mouse Robo (Fig. 1b) results in a loss of N-cadherin-mediated adhesion. Furthermore, expression of either cytoplasmic domain causes LN cells to become rounded and separated, a morphology that is consistent with a loss of N-cadherin-mediated adhesion (see Fig. 5a). Thus, expression of the cytoplasmic domain of Robo alone has the same effect as expression of full-length Robo after binding of Slit, indicating that it is constitutively active. This is consistent with observations that the cytoplasmic domain of Robo determines the functionality of receptor chimaeras25 and that it is sufficient to support an interaction with the netrin receptor DCC (deleted in colon cancer)26. Moreover, our results suggest that activated Robo affects N-cadherin function by interfering with its cytoplasmic domain.
| | Figure 5. Deletion of the CC3 region of the Robo cytoplasmic domain restores N-cadherin function. | | |  | a, N-cadherin is localized at points of cell−cell contact in LN cells and in Robo/cd- CC3 cells. In Robo/cd cells, no cell−cell contacts are formed and N-cadherin is found in a punctate, membrane-associated pattern. Cells were grown in 8-well culture slides for 24 h, fixed in paraformaldehyde and incubated with NCD2 before staining with Alexa-488 anti-rat secondary antibody. b, Adhesion of mock-transfected, Robo/cd, Robo/cd-CC3 or Robo/cd-CC1 cells to a substrate coated with Fc−N-cadherin. Robo/cd-CC3 cells adhere as well as control cells. c, LN, Robo/cd, Robo/cd-CC3 and Robo/cd-CC1 cell lysates were immunoprecipitated with NCD2 before western blotting with the indicated antibodies. In the absence of the CC3 domain in cytoplasmic Robo (Robo/cd-CC3), -catenin is found in association with N-cadherin, as in control cells.
 Full Figure and legend (57K) |
| |
The Robo-dependent loss of N-cadherin-mediated adhesion is not only restricted to transfected LN cells. Indeed, cells derived from the embryonic neural retina express Slit and Robo16,
17,
18 and are competent to form N-cadherin-mediated adhesions27. Embryonic chick neural retinal cells also extend neurites on a substrate composed of N-cadherin or anti-N-cadherin antibody5,
28. To test whether activated Robo affects N-cadherin-mediated adhesion in these cells, we introduced the Drosophila Robo cytoplasmic domain, expressed as a bacterial glutathione S-transferase (GST) fusion protein, into E8 chick neural retinal cells using the Chariot System (Active Motif). Immunolocalization of GST−Robo confirmed that the Robo cytoplasmic domain was effectively introduced into more than 90% of the cells (data not shown). The cells were then assayed for their ability to adhere to polylysine- or anti-N-cadherin-coated coverslips. Expression of the Robo cytoplasmic domain does not affect adhesion to polylysine or laminin (Fig. 1c; PLL/LM); however, it does result in a major reduction in the number of cells adhering to immobilized anti-N-cadherin antibody (Fig. 1c; NCD2). Decreased N-cadherin-mediated adhesion was also reflected in the smaller cell aggregates observed in suspension assays: from large clusters of up to 100 cells under control conditions to either single cells or cell pairs after introduction of cytoplasmic Robo (Fig. 1d). A similar loss of N-cadherin-mediated cell adhesion was observed in chick neural retina cells when endogenous Robo was activated by treatment with Drosophila Slit (Fig. 1e), consistent with the functional conservation of the Slit protein structure13.
Decreased adhesion among retina cells is also reflected in their ability to extend axons on a substrate of N-cadherin. Because purified Slit protein is a potent inhibitor of N-cadherin-mediated adhesion among retina cells when presented in solution, Slit was incorporated into agarose beads that were then placed on top of coverslips coated with either Fc−N-cadherin29 or the NCD2 antibody (refs 28,30). Single cells and small clusters of E8 neural retina cells were then seeded into the culture and incubated for 16−24 h in serum-free medium (Fig. 2a, b). We examined 10 coverslips for each condition, and each coverslip contained an average of 20 beads. In all cultures containing agarose beads alone, neurites extended randomly, contacting the area covered by the bead and also penetrating under the bead (Fig. 2c). In contrast, cells plated on coverslips with Slit-containing beads consistently extended neurites away from the Slit source, and in no case did neurites penetrate the area under the bead (Fig. 2d). This suggests that the presence of Slit does not allow maturation of nascent neurites at sites of highest Slit concentration. Together, these data suggest that Slit may cause a global collapse and retraction of growth cones at high concentrations and a local collapse under more restricted conditions.
| | |
Tyrosine phosphorylation of -catenin.
Tyrosine phosphorylation of -catenin reduces its affinity for N-cadherin31,
32 and results in disruption of the vital link to the actin-containing cytoskeleton23. In mock-transfected LN cells, -catenin was immunoprecipitated with N-cadherin from cell lysates prepared in neutral detergent (Fig. 3a; LN). This was also true for Robo/fl cells in the absence of Slit (Fig. 3a). However, -catenin was not detected in NCD2 immunoprecipitates from lysates of Robo/fl cells after treatment with Slit, or in NCD2 immunoprecipitates from lysates of Robo/cdD or Robo/cdM cells (Fig. 3a, top). In all cases, the amount of N-cadherin precipitated was similar (Fig. 3a, bottom). The loss of -catenin from N-cadherin immunoprecipitates occurred within 15 min of exposure to Slit. These data suggest that Slit-bound full-length Robo or the Robo cytoplasmic domain alone initiate signals that rapidly alter the ability of -catenin to associate with N-cadherin, resulting in a loss of the cadherin−actin connection, and thus cadherin function. Consistently, we observed an increase in the level of phosphorylated tyrosine residues on -catenin in all cases where expression of Robo resulted in a loss of N-cadherin function, that is, Robo/fl, Robo/cdM and Robo/cdD cells treated with Slit (Fig. 3b). Our results indicate that activation of Robo by Slit or expression of the constitutively active cytoplasmic domain of Robo alters N-cadherin-mediated adhesion by altering the balance of phosphorylated tyrosine residues on -catenin, which reduces the affinity of -catenin for cadherin31,
32 and consequently disrupts the cadherin−actin cytoskeleton linkage23. As observed in LN cells expressing full-length Robo, the association of -catenin with N-cadherin was lost in E8 retina cells exposed to Slit and there was an increase in the levels of phosphorylated tyrosine residues on -catenin (Fig. 3c). This was also true for retina cells after introduction of GST−Robo (Fig. 3c).
| | Figure 3. Activated Robo results in hyperphosphorylation of -catenin and dissociation of the cadherin−-catenin bond. | | |  | a, Confluent cell layers were washed free of serum and incubated in DMEM in the presence or absence of purified Drosophila Slit (1  g ml-1), for 30 min. The cell layers were then lysed, immunoprecipitated with NCD2 and analysed by western blotting with anti--catenin antibody. N-cadherin and -catenin co-immunoprecipitate in mock-transfected LN cells or in Robo/fl in the absence of Slit. No coprecipitation is observed in Robo/fl cells exposed to Slit or Robo/cd cells. b, Robo/fl cells incubated in the presence or absence of Slit, or Robo/cd cells, were lysed in RIPA buffer, immunoprecipitated with anti--catenin antibody and analysed by western blotting with an HRP-conjugated anti-phosphotyrosine antibody. c, E8 retina cells were incubated with or without Slit ( 1.4 g ml-1) for 15 min, or with GST or GST−Robo in the presence of the Chariot reagent. Cells were lysed in mild detergent and lysates were then immunoprecipitated with NCD2 and immunoblotted with anti--catenin antibody. Alternatively, cells were lysed in RIPA buffer, lysates immunoprecipitated with anti--catenin and immunoblotted with anti-phosphotyrosine antibody conjugated to HRP.
Full Figure and legend (45K) |
| |
Activated forms of Robo form a complex with N-cadherin.
Robo may alter -catenin tyrosine phosphorylation and N-cadherin function through the formation of a supramolecular complex containing the appropriate kinase, or through the triggering of a diffusible signalling cascade. To determine if Robo and N-cadherin become associated in a complex, Robo/fl or Robo/cd cells were assayed for the presence of Robo in immunoprecipitates of the N-cadherin complex, and conversely, for N-cadherin in immunoprecipitates of the Robo complex (Fig. 4). Immunoprecipitates of cell lysates expressing activated Robo (Robo/fl and Slit, but not Robo/fl in the absence of Slit or Robo/cd) with the NCD2 antibody contained Robo (Fig. 4a). A complex between N-cadherin and full-length Robo after Slit treatment or with the cytoplasmic domain of Robo was also detected when the immunoprecipitates were prepared using an anti-Robo antibody (Fig. 4b). These data indicate that binding of Slit resulted in a change in the cytoplasmic domain of Robo that allowed it to interact with N-cadherin, either directly or through additional components of the complex.
| | |
Abl is a component of the Robo−N-cadherin complex.
The protein tyrosine kinase Abl is a candidate for providing a physical link between Robo and cadherin, as well as altering the balance of phosphorylated tyrosine residues on -catenin and thus altering the N-cadherin−cytoskeleton linkage. Abl contains several protein interaction domains33 and binds to one of four conserved regions in the cytoplasmic domain of Robo (termed CC3), possibly regulating Robo function34. In Robo-transfected LN cells, Abl coprecipitated with full-length Robo independently of Slit and was always found in association with cytoplasmic Robo (Fig. 4c, d). As expected, this association was lost in LN cells expressing cytoplasmic Robo lacking the Abl binding domain CC3 (Robo/cd-CC3 cells; Fig. 4d). In contrast, deletion of the CC1 domain of Robo, which contains a potential Abl phosphorylation site34, did not affect its association with Abl (Fig. 4d).
Abl was co-immunoprecipitated with N-cadherin and Robo after Robo activation (Fig. 4e; Robo/fl cells after treatment with Slit, and Robo/cdD and Robo/cdM cells;). Conversely, immunoprecipitation of LN cell lysates expressing Robo/fl with anti-Abl and then immunoblotting with anti-N-cadherin also resulted in coprecipitation of N-cadherin, but, as above, only after treatment with Slit (data not shown). Consistent with a role for Abl in formation of the complex, Robo/cd-CC3 did not co-immunoprecipitate with N-cadherin (Fig. 4f), nor did Abl bind to N-cadherin in the absence of the Robo CC3 domain (Robo/cd-CC3; Fig. 4e). In contrast, deletion of the CC1 domain had no effect on formation of the Robo−Abl−N-cadherin complex (Robo/cd-CC1, Fig. 4e, f).
Complex formation between activated Robo, Abl and N-cadherin correlates with loss of N-cadherin-mediated adhesion. In monolayer culture, LN cells expressing Robo/cd do not form islands of closely apposed cells with cadherin-rich cell−cell boundaries; however, N-cadherin is still present at the cell surface in discrete puncta (Fig. 5a). In contrast LN cells expressing Robo/cd-CC3 seem to be identical to control LN cells, showing islands of closely apposed cells with the continuous presence of N-cadherin at cell−cell boundaries (Fig. 5a). Furthermore, Robo/cd-CC3-expressing cells were able to form N-cadherin-mediated adhesions in our standard assay, whereas Robo/cd and Robo/cd-CC1-expressing cells were not (Fig. 5b). In addition, -catenin was associated with N-cadherin in cells expressing Robo/cd-CC3, but not in cells expressing either Robo/cd-CC1 or intact Robo/cd (Fig. 5c), suggesting that loss of N-cadherin function is caused by a loss of the actin connection.
To examine the role of the Robo CC3 domain in cells expressing endogenous Robo and functional N-cadherin, we designed a peptide that mimics the CC3 domain coupled to the antennapedia permeation sequence35,
36 and introduced it into E8 retina cells. The efficacy of the peptide was confirmed by its ability to compete for the association of Abl with Robo in E8 retina cells (Fig. 6a). The peptide completely abolished the ability of Slit to inhibit N-cadherin mediated adhesion (Fig. 6b) and prevented formation of the Robo−Abl−N-cadherin complex (Fig. 6c).
| | |
Although Abl is an essential component of the Robo−Abl−N-cadherin complex, at least one other component seems to be required. When Robo/cd attached to glutathione−Sepharose beads was incubated with lysates from LN cells prepared in mild detergent34, N-cadherin specifically associated with Robo (Fig. 6d). However, using purified fusion proteins in an in vitro binding assay we did not detect an association between the cytoplasmic domain of Robo and N-cadherin, even when Abl was present (data not shown). Thus, Abl and at least one additional intermediate component are likely to mediate the association between the cytoplasmic domains of Robo and N-cadherin in LN cells.
The role of Abl in the Robo−Abl−N-cadherin complex.
The use of deletion constructs and the CC3 peptide mimetic confirmed the key structural role of Abl in formation of the Robo−N-cadherin complex. The tyrosine kinase inhibitor, AG957, which belongs to a family of inhibitors of Abl kinase activity37, was used to assess the role of Abl catalytic activity in complex formation and loss of N-cadherin-mediated adhesion. The inhibitor was used at 10 M, a concentration that completely inhibits Abl activity in vitro and in vivo37. The inhibitor completely reversed the effect of Slit treatment (Fig. 7a); however, it did not prevent formation of the Robo−Abl−N-cadherin complex (Fig. 7b). Furthermore, AG957 blocked in vitro phosphorylation of -catenin by purified Abl (Fig. 7c). These results indicate that Abl catalytic activity is not essential for formation of the Robo−Abl−N-cadherin complex, but is crucial for loss of N-cadherin-mediated adhesion and tyrosine phosphorylation of -catenin.
Discussion
Our data indicate that activation of Robo by Slit results in loss of the functionally critical association of actin with N-cadherin. Without this connection, cadherin-mediated adhesion is compromised and the forces necessary for traction, and thus axon extension, are lost. Loss of N-cadherin-mediated adhesion and neurite outgrowth is mediated by the formation of a complex between Robo, Abl and N-cadherin, with concomitant phosphorylation of -catenin on tyrosine residues. Tyrosine phosphorylation of -catenin reduces its affinity for cadherin31,
32, resulting in loss of the critical N-cadherin−actin connection. Abl most probably has two related roles in the regulation of N-cadherin function: first, Abl bound to the CC3 region of the Robo cytoplasmic tail functions as part of the linkage in bringing Robo and N-cadherin together; second, Abl catalytic activity is essential for phosphorylation of -catenin. We cannot be sure that Abl phosphorylates -catenin directly; it may initiate a cascade that ultimately results in tyrosine phosphorylation of -catenin by some other kinase. However, we do show that Abl can phosphorylate -catenin in vitro, and Abl has been shown to phosphorylate the closely related Armadillo protein  -catenin38. Additionally, we believe that other components are essential for formation of the Robo−Abl−N-cadherin, as we are unable to reconstruct the complex in vitro from purified components. A diagrammatic representation of the critical components and effectors in formation of the Robo−Abl−N-cadherin complex and silencing of N-cadherin is presented in Fig. 8.
| | |
At least one other guidance cue, the extracellular chondroitin sulfate proteoglycan, neurocan, contributes to the regulation of N-cadherin-mediated adhesion and axon growth extension among neural retina cells39. Neurocan lines many axon tracts in the CNS40 and is expressed within the amacrine layer in the retina39. Binding of neurocan to its receptor initiates a signal that also results in enhanced phosphorylation of tyrosine residues on -catenin and loss of cadherin function39. This suggests that the uncoupling of N-cadherin from its actin connection may be a more general phenomenon that allows a rapid response to repulsive cues, altering the direction of growth cone extension. Regardless of how general this may be, our identification of the mechanism of cross talk between activated Robo and N-cadherin fills an important gap in our understanding of how growth cone traction is related to secreted guidance cues.
Activation of Robo by Slit also results in 'silencing' of the Netrin receptor Frazzled/DCC26. In contrast to the Robo−N-cadherin complex, formation of the Robo−Frazzled/DCC complex is mediated by a direct interaction between the CC1 cytoplasmic region of Robo and the P3 region of Frazzled/DCC. Traditionally, the equilibrium between attractive and repulsive cues has been thought to direct an axon to its target. However, the ability of Robo to silence Frazzled/DCC has led to the idea that some guidance cues function in a hierarchical fashion; that is, certain cues may have a higher priority that is manifested by their ability to directly silence the receptors of competing guidance cues. Our results suggest that Robo is a dominant member within such a hierarchy, as it functions to silence both long-range attractive cues and local adhesive interactions that provide traction. It will be important to determine if attractive cues compete with the ability of Robo to silence N-cadherin.
Robo seems to be a multifunction guidance receptor that regulates axon trajectories through a variety of mechanisms. It functions as a repulsive cue for axons extending from fragments of neural retina embedded in a collagen matrix16,
17,
18. Under these conditions, adhesive traction is not provided by cadherins, but presumably by integrins. Because we do not see a direct effect of activated Robo on 1-integrin function, it is possible that the effects of Slit on integrin are indirect. Rho GTPase activating proteins that interact with Robo (srGAPs) may be among the critical effectors functioning downstream of guidance cues to regulate actin assembly, and therefore axon outgrowth41. Interestingly, srGAP interacts with the CC3 region of the cytoplasmic domain of Robo, the same region that binds Abl, suggesting that the two mechanisms may be incompatible. Guanine nucleotide exchange factors (GEFs) may also be critical to the mechanisms underlying repulsive cues in Drosophila42,
43. This suggests that Robo may determine the balance between GAPs and GEFs, which affects the state of actin assembly. Furthermore, genetic and biochemical analysis of the role of Robo in axon guidance at the midline of the Drosophila ventral nerve cord suggest that Enabled (Ena) and Abl function downstream of Robo34 and may also be critical effectors in regulating the dynamics of actin assembly/disassembly44. Both of these putative mechanisms for Robo-mediated repulsion are based on altering the dynamics of actin assembly/disassembly. This is in contrast to the mechanism reported here, which is based on uncoupling of the actin network from the adhesion receptor providing traction.
Axon guidance in the vertebrate neural retina and the Drosophila ventral nerve cord have been used repeatedly for analysis of Robo and N-cadherin function. N-cadherin is the primary cadherin in the developing neural retina and is required for proper development27 and axon extension45. Slit and Robo are expressed in the developing retina, in the ganglion cell layer and in the region of the optic chiasm16,
17,
18. Mice null for Slit1 and Slit2 have defects in retinal ganglion cell axon trajectories46 and mutations in the Zebrafish orthologue of Robo, astray, also resulted in abnormal projections of retinal ganglion cells47,
48. In the Drosophila ventral nerve cord, Robo is expressed throughout the longitudinal connectives12,
13,
20,
21, whereas N-cadherin is present on axons within both the commissures and longitudinal connectives10. Loss-of-function mutations in both CadN and the Drosophila homologue of -catenin, armadillo (arm), cause defects in axon bundling within the longitudinal connectives and, at least for arm, these defects are enhanced by abl loss-of-function mutations. Thus, in contrast to our model, these interactions suggest that Abl stabilizes adherens junctions, possibly by ensuring that Arm remains within the Cadherin complex10,
49,
50. However, a key feature of our model is localized cis interactions between Robo and N-cadherin. Thus, Robo may recruit Abl to inhibit N-cadherin adhesion during exploration of some regions (for example, the embryonic midline) and defects in this process would be expected to be re-enforced in both CadN and arm mutants. Within the longitudinal connective, Abl may indeed function to stabilize N-cadherin-mediated adhesion and it is defects in this process that are reflected in the CadN and arm mutant phenotypes.
The accumulated data suggest that Robo activates multiple parallel pathways during axon guidance. It has the potential to affect growth cone traction by directly affecting the coupling of cadherin to the cytoskeleton, directly affecting the growth cone motile apparatus through modulation of actin assembly/disassembly and inactivating the attractive cue Netrin. When and where each of these mechanisms is brought into play will undoubtedly depend on the specific developmental situation. The challenge ahead lies in dissecting the role of each of these many molecular interactions in the context of the intact tissue or organism.
Methods
Antibodies
A synthetic peptide corresponding to amino acids 1186−1204 of the mouse Robo was used to generate a rabbit polyclonal antibody (Biosynthesis Inc., Lewisville, TX). Monoclonal anti-Drosophila Robo 13C9 (ref. 12) was a kind gift from Corey Goodman (UC Berkeley, CA). NCD2 is an anti-chick N-cadherin antibody originally developed by M. Takeichi (Kyoto, Japan). This antibody has been used extensively in our lab as a substrate for N-cadherin-dependent cell adhesion and neurite outgrowth assays28,
30. The monoclonal rat anti-HA high-affinity antibody was purchased from Roche (Indianapolis, IN). Anti-Abl antibody was from BD PharMingen (San Diego, CA) anti--catenin and HRP-conjugated anti-phosphotyrosine antibodies were from BD Transduction Laboratories (San Diego, CA). HRP-conjugated secondary antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies conjugated to magnetic beads for use in immunoprecipitations were from PerSeptive Biosystems (Farmingham, MA). Alexa−Fluor 488 goat anti-rat IgG was from Molecular Probes (Eugene, OR).
Cloning and expression of Drosophila and mouse Robo
The cDNA encoding the cytoplasmic domain of Drosophila was obtained from the full-length cDNA (a gift from M. Seeger, Ohio State University, Columbus, OH) by PCR using a 5' primer corresponding to the sequence just C-terminal of the transmembrane domain and a 3' primer designed to introduce an HA tag at the C terminus. The cytoplasmic domain of mouse Robo was amplified by PCR from adult mouse brain cDNA and an HA tag was introduced at the C-terminal end. The PCR products were sequenced and introduced into the mammalian expression vector pcDNA3.1(+)zeo (Invitrogen, Carlsbad, CA). To generate CC3 and CC1 Robo cDNA, recombinant PCR was used to remove the CC3 (PPPPVPPP) or CC1 (PTPYATT) domain from mouse cytoplasmic Robo. PCR products were sequenced and subcloned into pcDNA3.1(+)zeo. LN cells were transfected with the pcDNA3.1(+)zeo constructs and stable cell clones selected with 1 mg ml-1 Zeocin (Invitrogen).
Immunoprecipitation assays
Cell layers were washed in PBS and incubated for 30 min in lysis buffer (2% NP-40 in PBS containing Complete mini protease inhibitor cocktail (Roche) and 1 mM sodium orthovanadate). Cell lysates were cleared by centrifugation at 15000g for 10 min and aliquots containing equivalent amounts of protein were incubated with the appropriate antibody before addition of secondary antibody conjugated to magnetic beads. To determine phosphorylation of -catenin, cells were lysed in RIPA buffer (PBS containing 0.1% SDS, 0.1% sodium deoxycholate, 1% NP-40, 50 mM sodium fluoride, 1 mM sodium orthovanadate and Complete mini protease inhibitor cocktail). The SDS concentration was reduced to 0.05% with RIPA buffer without SDS and the cleared lysates were immunoprecipitated with an anti--catenin antibody.
Adhesion assays
96-well plates coated with poly-l-lysine (Sigma, St Louis, MO) were incubated with anti-N-cadherin antibody NCD2 (20 g ml-1 in PBS; 200 l per well) overnight at 4 °C. The wells were washed with PBS and blocked with 2% BSA for 1 h at 37 °C. Alternatively, 96-well plates pre-coated with mouse anti-IgG (Bio-Coat; BD Biosciences, Billerica, MA) were incubated with purified N-Cadherin−Fc (chick N-cadherin ectodomain fused with the Fc fragment of the mouse IgG2b29), diluted in PBS containing 2% BSA. Cells growing to confluence were harvested with trypsin in the presence of calcium, resuspended in DMEM in the presence or absence of Slit and equal numbers of cells were aliquoted into the precoated wells. When neural retina cells were used, E8 single cells were prepared by trypsin dissociation in the presence of calcium, as previously described30. After 1 h, non-adherent cells were washed and bound cells were quantified using the CellTiter 96 aqueous non-radioactive cell proliferation assay kit (Promega, Madison, WI). Adhesion in the presence of soluble NCD2 was measured as a negative control.
Neurite outgrowth assay
Agarose beads containing Slit were prepared by mixing approximately 10 l of low-melting agarose (1% in PBS) at 39 °C with 8 l of Fc−N-cadherin or NCD2 (1 mg ml-1) and 2 l of PBS (Control) or Slit preparation (14 mg ml-1). This mix was spotted quickly in small droplets (0.5 l each; approximately 500 m in diameter) on Fc−N-cadherin- or NCD2-coated coverslips. Approximately 20 beads were spotted per coverslip. After gelling at 4 °C for 2 min, coverslips with beads were incubated with F12 medium for 15 min at 37 °C; the medium was then replaced by F12 medium containing additives (1% insulin/transferrin/selenium and 0.5% gentamycin) containing a suspension of E7 retina cells and small cell clusters. Neurite outgrowth was evaluated after 16 h.
Received 19 February 2002; Accepted 21 August 2002; Published online: 23 September 2002.
REFERENCES
-
Yu, T. W. & Bargmann, C. I. Dynamic regulation of axon guidance. Nature Neurosci. Suppl. 4, 1169–1176 (2001). | Article | ChemPort |
-
Tepass, U., Truong, K., Godt D., Ikura, M. & Peifer, M. Cadherins in embryonic and neural morphogenesis. Nature Rev. Mol. Cell Biol. 1, 91–100 (2000). | Article | PubMed | ISI | ChemPort |
-
Redies, C. Cadherins in the central nervous system. Prog. Neurobiol. 61, 611–648 (2000). | Article | PubMed | ISI | ChemPort |
-
Ranscht, B. Cadherins: molecular codes for axon guidance and synapse formation. Int. J. Develop. Neurosci. 18, 643–651 (2000). | Article | ISI | ChemPort |
-
Bixby, J. L. & Zhang, R. Purified N-cadherin is a potent substrate for the rapid induction of neurite outgrowth. J. Cell Biol. 110, 1253–1260 (1990). | PubMed | ISI | ChemPort |
-
Bixby, J. L., Lilien, J. & Reichardt, L. F. Identification of the major proteins that promote neuronal process outgrowth on Schwann cells in vitro. J. Cell Biol. 107, 353–361 (1988). | PubMed | ISI | ChemPort |
-
Neugebauer, K. M., Tomaselli, K. J., Lilien, J. & Reichardt, L. F. N-cadherin, NCAM, and integrins promote retinal neurite outgrowth on astrocytes in vitro. J. Cell Biol. 107, 1117–1187 (1988).
-
Tomaselli, K. J., Neugebauer, K. M., Bixby, J. L., Lilien, J. & Reichardt, L. F. N-cadherin and integrin: Two receptor systems that mediated neuronal process outgrowth on astrocyte surfaces. Neuron 1, 33–43 (1988). | PubMed | ISI | ChemPort |
-
Matsunaga, M., Hatta, K., Nagafuchi, A. & Takeichi, M. Guidance of optic nerve fibres by N-cadherin adhesion molecules. Nature 334, 62–64 (1988). | Article | PubMed | ISI | ChemPort |
-
Iwai, Y. et al. Axon patterning requires DN-cadherin, a novel neuronal adhesion receptor, in the Drosophila embryonic CNS. Neuron 19, 77–89 (1997). | PubMed | ISI | ChemPort |
-
Wheelock, M. J., Knudsen, K. A. & Johnson, K. R. Membrane–cytoskeleton interactions with cadherin cell adhesion proteins: roles of catenins as linker proteins. Curr. Top. Mem. 43, 169–185 (1996). | ISI | ChemPort |
-
Kidd, T. et al. Roundabout controls axon crossing of the CNS midline and defines a novel subfamily of evolutionarily conserved guidance receptors. Cell 92, 205–215 (1998). | PubMed | ISI | ChemPort |
-
Brose, K. et al. Slit proteins bind Robo receptors and have an evolutionarily conserved role in repulsive axon guidance. Cell 96, 795–806 (1999). | PubMed | ISI | ChemPort |
-
Holmes, G. P. et al. Distinct but overlapping expression patterns of two vertebrate slit homologues implies functional roles in CNS development and organogenesis. Mech. Dev. 79, 57–72 (1998). | Article | PubMed | ISI | ChemPort |
-
Bagri, A. et al. Slit proteins prevent midline crossing and determine the dorsoventral position of major axonal pathways in the mammalian forebrain. Neuron 33, 233–248 (2002). | PubMed | ISI | ChemPort |
-
Erskine, L. et al. Retinal ganglion cell axon guidance in the mouse optic chiasm: expression and function of Robos and Slits. J. Neurosci. 20, 4975–4982 (2000). | PubMed | ISI | ChemPort |
-
Niclou, S. P., Jia, L. & Raper, J. A. Slit2 is a repellent for retinal ganglion cell axons. J. Neurosci. 20, 4962–4974 (2000). | PubMed | ISI | ChemPort |
-
Ringstedt, T. et al. Slit inhibition of retinal axon growth and its role in retinal axon pathfinding and innervation patterns in the Diencephalon. J. Neurosci. 20, 4983–4991 (2000). | PubMed | ISI | ChemPort |
-
Seeger, M., Tear, G., Ferres-Marco, D. & Goodman, C. S. Mutations affecting growth cone guidance in Drosophila: genes necessary for guidance toward or away from the midline. Neuron 10, 409–426 (1993). | PubMed | ISI | ChemPort |
-
Kidd, T., Bland, K. S. & Goodman, C. S. Slit is the midline repellent for the Robo receptor in Drosophila. Cell 96, 785–794 (1999). | PubMed | ISI | ChemPort |
-
Rajagopalan, S., Nicolas, E., Vivancos, V., Berger, J. & Dickson, B. J. Crossing the midline: roles and regulation of Robo receptors. Neuron 28, 767–777 (2000). | PubMed | ISI | ChemPort |
-
Murray, M. J. & Whitington, P. M. Effects of roundabout on growth cone dynamics, filopodial length, and growth cone morphology at the midline and throughout the neuropile. J. Neurosci. 19, 7901–7912 (1999). | PubMed | ISI | ChemPort |
-
Lilien, J., Balsamo, J., Arregui, C. & Xu, G. Turn-off, drop-out: functional state switching of cadherins. Develop. Dyn. 224, 18–29 (2002). | Article | ISI | ChemPort |
-
Rothberg, J. M., Jacobs, J. R., Goodman, C. S. & Artavanis-Tsakonas, S. Slit: an extracellular protein necessary for development of midline glia and commissural axon pathways contains both EGF and LRR domains. Genes Dev. 4, 2169–2187 (1990). | PubMed | ISI |
|
