BIOLOGICAL OVERVIEW

Son of sevenless (Sos) is a dual specificity guanine nucleotide exchange factor (GEF) that regulates both Ras and Rho family GTPases and thus is uniquely poised to integrate signals that affect both gene expression and cytoskeletal reorganization. Sos is recruited to the plasma membrane, where it forms a ternary complex with the Roundabout receptor and the SH3-SH2 adaptor protein Dreadlocks (Dock) to regulate Rac-dependent cytoskeletal rearrangement in response to the Slit ligand. Intriguingly, the Ras and Rac-GEF activities of Sos can be uncoupled during Robo-mediated axon repulsion; Sos axon guidance function depends on its Rac-GEF activity, but not its Ras-GEF activity. These results provide in vivo evidence that the Ras and RhoGEF domains of Sos are separable signaling modules and support a model in which Robo recruits Sos to the membrane via Dock to activate Rac during midline repulsion (Yang, 2006).

Correct wiring of the nervous system depends on precisely coordinating the distribution and activity of a diverse set of axon guidance cues and their neuronal receptors. Studies of both invertebrate and vertebrate nervous systems have begun to define the signaling mechanisms that function downstream of guidance receptors to regulate growth cone steering and motility. The Rho family of small GTPases (Rac, Rho, and Cdc42) have emerged as central regulators of actin cytoskeletal dynamics in neurons and have been implicated in diverse axon guidance receptor signaling pathways. Increasing evidence indicates that the positive and negative regulators of the Rho GTPases (GEFs and GAPs) can couple axon guidance receptors to the Rho GTPases to regulate actin dynamics in the growth cone. For example, activation of RhoA downstream of the Eph receptor is mediated by the Rho family GEF 'Ephexin' while Eph-dependent activation of Rac is mediated by another Rho family GEF, Vav (Yang, 2006 and references therein).

Drosophila Robo is the founding member of a conserved group of repulsive guidance receptors of the immunoglobulin (Ig) superfamily and consists of an ectodomain with five Ig domains and three fibronectin type III repeats, a single transmembrane domain, and a long cytoplasmic tail that contains four blocks of conserved cytoplasmic (CC) sequences (CC0, CC1, CC2, CC3). Robo is required to prevent axons from inappropriately crossing the CNS midline in both invertebrates and vertebrates, and it has also been implicated in controlling cell migration in other cell types. In Drosophila, mutations in robo and its midline-expressed ligand Slit result in too many axons crossing and staying at the midline. Several proteins that regulate the actin cytoskeleton, including the cytoplasmic tyrosine kinase Abelson (Abl) and its substrate Enabled (Ena), contribute to the Robo signaling pathway in Drosophila and C. elegans. In addition, genetic interaction and biochemical experiments in Drosophila and biochemical experiments in mammalian cell culture indicate that activation of Slit-Robo signaling leads to activation of Rac and Rho, and inactivation of Cdc42 (Fan, 2003; Fritz, 2002; Matsuura, 2004; Wong, 2001; Yang, 2006 and references therein).

It is clear from studies of Slit-mediated neural precursor cell migration in rats that inactivation of Cdc42 by Robo is mediated by Slit-Robo GAP (SrGAP1) (Wong, 2001). However, how Slit leads to the activation of Rac in either Drosophila or vertebrate systems is still unknown. Recent work in Drosophila suggests that the SH3-SH2 adaptor protein Dock may play a role in recruiting Rac to the Robo receptor. Slit stimulation recruits Dock and p21-activated kinase (Pak) to the Robo receptor, and Pak is a downstream target of Rac. It has been proposed that Dock recruits Pak to specific sites at the growth cone membrane, where Pak, activated by Rac, regulates the recycling and retrograde flow of actin filaments (Fan, 2003; Hing, 1999). Despite these observations, it still remains unclear how Rac is activated in this context. One possible mechanism would be by negative regulation of a Rac-specific GAP(s) upon Slit stimulation. Indeed, a genome-wide analysis in Drosophila has identified a Rac-specific GAP, CrossGAP/Vilse (CrGAP), which interacts directly with the CC2 motif of Robo (Hu, 2005; Lundstrom, 2004). Overexpression of crGAP mimics the robo mutant phenotype, which suggests that it plays a negative role in Slit-Robo signaling. However, crGAP/vilse mutants do not have major midline axon guidance defects; in fact, loss of crGAP/vilse actually leads to mild robo-like defects (Lundstrom, 2004). Thus, it would appear that downregulating crGAP alone in the Robo signaling pathway is not sufficient to lead to activation of Rac (Yang, 2006 and references therein).

Since Rho GTPases are directly activated by GEFs, it was asked whether Slit-dependent upregulation of a Rac-specific GEF leads to the activation of Rac. Among the 22 Rho family GEFs in the Drosophila genome, Sos is a good candidate to play this role for the following reasons. First, sos is among eight RhoGEFs that are enriched in the Drosophila embryonic central nervous system (Hu, 2005). Second, sos was previously shown to genetically interact with slit during midline guidance (Fritz, 2000); no other GEF has been shown to genetically interact with slit or robo. Third, mutations in sos partially suppress the commissureless mutant phenotype, where elevated robo function results in a complete absence of axon commissures, suggesting that sos functions in the robo pathway (Fritz, 2000). Finally, mammalian Sos is a Rac-specific GEF and it directly binds to Nck, the mammalian homolog of Drosophila Dock (Hu, 1995; Nimnual, 1998; Okada, 1996; Yang, 2006 and references therein).

Sos was identified in Drosophila as a GEF for Ras in the sevenless signaling pathway during the development of the Drosophila compound eye, where it activates the Ras signaling cascade to determine R7 photoreceptor specification (Bonfini, 1992; Simon, 1991). Studies in mammalian cell culture demonstrated that Sos functions as a GEF for both Ras and Rac in the growth factor-induced receptor tyrosine kinase (RTK) signaling cascade (Nimnual, 2002; Nimnual, 1998). Upon RTK activation, the SH3/SH2 adaptor protein Grb2/Drk recruits Sos to autophosphorylated receptors at the plasma membrane, where Sos activates membrane-bound Ras. In a later event downstream of RTK activation, Sos is thought to be targeted to submembrane actin filaments by interaction with another SH3 adaptor, E3b1(Abi-1), where Sos activates Rac (Innocenti, 2002; Innocenti, 2003; Scita, 1999; Scita, 2001). Whether the activation of Rac by Sos is strictly dependent on prior activation of Ras remains controversial, nor is it clear how Sos coordinates the activity of its two GEF domains in vivo (Yang, 2006).

Evidence is provided that Sos functions as a Rac-specific GEF during Drosophila midline guidance. Sos is enriched in developing axons, and sos exhibits dosage-sensitive genetic interactions with slit and robo. Strikingly, genetic rescue experiments show that the Dbl homology (DH) RhoGEF domain of Sos, but not its RasGEF domain, is required for its midline guidance function. Biochemical experiments show that Sos physically associates with the Robo receptor through Dock in both mammalian cells and Drosophila embryos. Furthermore, Slit stimulation of cultured cells results in the rapid recruitment of Sos to membrane Robo receptors. These results provide a molecular link between the Robo receptor and Rac activation, reveal an independent in vivo axon guidance function of the DH RhoGEF domain of Sos, and support the model that Slit stimulation recruits Sos to the membrane Robo receptor via Dock to activate Rac-dependent cytoskeletal changes within the growth cone during axon repulsion (Yang, 2006).

These data support the idea that Sos provides a direct molecular link between the Robo receptor and the activation of Rac during Drosophila midline guidance. Genetic interactions between sos, robo, dock, crGAP/vilse, and the Rho family of small GTPases strongly suggest that Sos functions in vivo to regulate Rac activity during Robo signaling. Genetic rescue experiments indicate that sos is required specifically in neurons to mediate its axon guidance function. Furthermore, genetic data establish that, in the context of midline axon guidance, the Ras-GEF and Rac-GEF activities of Sos can be functionally uncoupled. Biochemical experiments in cultured cells and Drosophila embryos show that Sos is recruited into a multiprotein complex consisting of the Robo receptor, the SH3-SH2 adaptor protein Dock, and Sos, in which Dock bridges the physical association between Robo and Sos. Finally, experiments in cultured cells support the idea that Slit activation of Robo can recruit Sos to the submembrane actin cytoskeleton to regulate cell morphology. Together, these results suggest a model in which Slit stimulation recruits Sos to the Robo receptor via Dock to regulate Rac-dependent cytoskeletal changes within the growth cone during axon repulsion (Yang, 2006).

Based on previous work implicating rac in Robo repulsion, as well as in vitro studies demonstrating that Sos exhibits GEF activity for Rac, but not Rho or Cdc42, Rac seemed the most likely Sos substrate. However, rho has also been implicated in mediating Robo repulsion (Fan, 2003; Fritz, 2002), and genetic interactions between sos and dominant-negative Rho have been interpreted to suggest that Sos could act as a GEF for Rho. This question was investigated further, and two types of genetic evidence have been presented that suggest that indeed Rac is the favored substrate of Sos. First, ectopic expression experiments in the eye reveal interactions exclusively between sos and rac. Second, genetic interaction experiments using loss of function mutations in rac and rho (rather than the more problematic dominant-negative forms of the GTPases) reveal strong dose-dependent interactions between sos and rac, but not sos and rho during midline axon guidance. Together, these observations argue in favor of Rac as the primary in vivo Sos substrate. Nevertheless, the possibilities that Sos also contributes to Rho activation and that the combined activation of Rac and Rho is instrumental in mediating the Robo response cannot be excluded (Yang, 2006).

Previous studies have demonstrated that Slit stimulation of the Robo receptor leads to a rapid increase in Rac activity in cultured cells. However, the mechanism by which Rac is activated downstream of Robo was not clear. This study provides direct genetic and biochemical evidence that Sos is coupled to the Robo receptor through the Dock/Nck SH3-SH2 adaptor, where it can regulate local Rac activation. Studies in cultured mammalian cells have highlighted the importance of distinct Sos/adaptor protein complexes in controlling the subcellular localization and substrate specificity of Sos. In the context of Rac activation, the E3b1 (Abi-1) adaptor has been shown to play a critical and rate-limiting role in Sos-dependent Rac activation and subsequent formation of membrane ruffles (Innocenti, 2002). Could Sos regulation of Rac activity during Robo repulsion be similarly limited by the availability of specific adaptor proteins? It is interesting to note in this context that overexpression of dock does not lead to ectopic axon repulsion, suggesting that Dock may not be limiting for Robo signaling. However, although dock mutants do have phenotypes indicative of reduced Robo repulsion, their phenotype is considerably milder than that seen in robo mutants, raising the possibility that there may be additional links between Robo and Sos (Yang, 2006).

A number of studies in cultured mammalian cells have suggested that Rac activation induced by activated growth factor receptors requires the prior activation of Ras. For example, PDGF-induced membrane ruffling can be promoted or inhibited by expression of constitutively active or dominant-negative Ras, respectively. However, other studies have suggested that in Swiss 3T3 cell lines RTK activation of Rac is Ras independent. In addition, the observation that Ras activation and Rac activation display very different kinetics, with Rac activation persisting long after Ras activity has returned to basal levels, has been used to argue against an obligate role for Ras in Rac activation (Innocenti, 2002). In this study, using a genetic rescue approach, whether the ability of Sos to activate Rac during axon guidance in an intact organism requires its Ras-GEF function was directly tested. Genetic data indicate that the RasGEF domain of Sos is dispensable for axon guidance, while the DH RhoGEF domain is strictly required. This observation argues strongly in favor of the model that in vivo Sos activation of Rac does not strictly require Sos activation of Ras (Yang, 2006).

It is clear that subcellular localization plays a major role in regulating Sos activity and that different protein complexes containing Sos exist in different locations in the cell. This study has shown that activation of the Robo receptor by Slit triggers the recruitment of Sos to Robo receptors at the plasma membrane. Biochemical data argue that the adaptor Dock/Nck is instrumental in bridging this interaction, and given the diverse interactions between Dock/Nck and guidance receptors, it seems likely that Dock/Nck could fulfill this role in many guidance receptor contexts. This bridging function of Dock/Nck and guidance receptors is analogous to the role of Grb2 for growth factor receptors only insomuch as it brings signaling molecules to the receptor—the mechanism of interaction is distinct, since it is mediated through SH3 domain contacts rather than SH2/phosphotyrosine interactions. These observations suggest that there may be an additional pool of Sos that can function in a distinct adaptor protein/guidance receptor complex to regulate cell morphology in response to extracellular guidance cues (Yang, 2006).

Is regulating subcellular localization the only mechanism by which Sos activity is controlled? This seems unlikely. Indeed, a recent study has implicated tyrosine phosphorylation of Sos by Abl as an additional mechanism to activate the Rac-specific GEF activity of Sos in vertebrate cell culture models (Sini, 2004). This raises the intriguing possibility that Abl may fulfill a similar role for Robo signaling. This is a particularly appealing idea given the well-documented genetic and physical interactions between Robo and Abl. Indeed, sos and abl exhibit dose-dependent genetic interactions during midline axon guidance. A clear genetic test of whether Abl activates the Rac-GEF activity of Sos downstream of Robo may be complicated by the fact that Abl appears to play a dual role in Robo repulsion: both increasing and decreasing abl function lead to disruptions in Robo function. Nevertheless, it should be possible in the future to generate mutant versions of Sos that are refractory to Abl activation and to test whether these alterations disrupt the Sos guidance function. It will also be of great interest to determine whether the redistribution of Sos can also be observed in response to guidance receptor signaling in navigating growth cones, and if so, then what changes in actin dynamics and growth cone behavior are elicited (Yang, 2006).

GENE STRUCTURE

cDNA clone length - 5625

Bases in 5' UTR - 386

Exons - 7

Bases in 3' UTR - 448

PROTEIN STRUCTURE

Amino Acids - 1596

Structural Domains

A genetic screen was conducted for mutations that decrease the effectiveness of signaling by a protein tyrosine kinase, the product of the Drosophila sevenless gene. These mutations define seven genes whose wild-type products may be required for signaling by sevenless. Four of the seven genes also appear to be essential for signaling by a second protein tyrosine kinase, the product of the Ellipse gene (Egfr). The putative products of two of these seven genes have been identified. One encodes a ras protein. The other locus encodes a protein that is homologous to the S. cerevisiae CDC25 protein, an activator of guanine nucleotide exchange by ras proteins. These results suggest that the stimulation of ras protein activity is a key element in the signaling by Sevenless and Ellipse and that this stimulation may be achieved by activating the exchange of GTP for bound GDP by the Ras protein (Simon, 1991).

The Son of sevenless (Sos) gene functions in signaling pathways initiated by the Sevenless and Epidermal growth factor receptor tyrosine kinases. The Sos gene has now been isolated and sequenced. Its product is a 1595-amino acid protein similar to the CDC25 protein in Saccharomyces cerevisiae, a guanine nucleotide exchange factor that activates Ras. These results imply a role for the ras pathway in Drosophila neuronal development (Bonfini, 1992).

date revised: 2 January 2023

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

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