BIOLOGICAL OVERVIEW

While extensive studies in several systems have made considerable progress in defining the general mechanisms that direct growth cone extension, much less is known of the mechanism that makes growth cones stop at a specific target layer underlying the formation of layer-specific connections. Misshapen (Msn) has been proposed to shut down Drosophila photoreceptor (R cell) growth cone motility in response to targeting signals linked to Msn by the SH2/SH3 adaptor protein Dock. To identify downstream targets of Msn in R cell growth cones, a genetic dissection was undertaken to search for second-site mutations that modify a Msn hyperactivation phenotype. bifocal (bif), a gene encoding a putative cytoskeletal regulator (Bahri, 1997), shows strong interaction with msn. Bif binds to F-actin in vitro (Sisson, 2000) and colocalizes with F-actin during development (Bahri, 1997). Bifocal is a component of the Msn pathway for regulating R cell growth cone targeting. Phenotypic analysis indicates a specific role for Bif to terminate R1-R6 growth cones. Biochemical studies show that Msn associates directly with Bif and phosphorylates Bif in vitro. Cell culture studies demonstrate that Msn interacts with Bif to regulate F-actin structure and filopodium formation. It is proposed that Bif functions downstream of Msn to reorganize actin cytoskeleton in decelerating R cell growth cone motility at the target region (Ruan, 2002).

To gain insight into the molecular mechanism that converts targeting signals into the arrest of axonal growth, genetic analysis has been used to investigate the control of photoreceptor (R cell) growth cone targeting in Drosophila. The Drosophila compound eye consists of ~800 ommatidia, each containing eight different R cell subtypes that project axons into one of two optic ganglia layers in the brain. R1-R6 axons innervate the most superficial layer, the lamina, whereas R7 and R8 project axons through the lamina into the deeper medulla layer. The establishment of this layer-specific R cell connectivity pattern begins at the late third-instar larval stage. The formation of each ommatidium in the third-instar eye-imaginal disc involves the sequential recruitment and differentiation of R cells; R8 differentiates first, followed by R2/5, R3/4, R1/6, and R7. R8 axons from each ommatidium migrate toward the most posterior end of the eye-imaginal disc and subsequently enter the optic stalk. After exiting the optic stalk, R8 axons project through the superficial lamina layer and terminate within the developing medulla in a topographic fashion. R1-R7 axons from the same ommatidium follow the route of the pioneer R8 axon until they encounter a layer of lamina glial cells (i.e., marginal glia). While R1-R6 growth cones stop extension in response to some unknown stop signals produced by marginal glial cells, R7 growth cones pass through the lamina into the medulla. About 4-5 days after the initial targeting into the lamina, R1-R6 growth cones undergo stereotyped rearrangements to establish synaptic connections with lamina neurons (Ruan, 2002).

Previous studies have implicated the SH2/SH3 adaptor protein Dreadlocks (Dock) in mediating signaling events in R cell growth cones. Mutations in the dock gene locus cause a failure for many R1-R6 growth cones to stop at the lamina, and also cause defects in R cell axonal fasciculation and growth cone expansion. Biochemical and genetic studies of Dock and its vertebrate homolog Nck suggest that Dock/Nck utilizes its SH2 and SH3 domains to recruit downstream effectors to tyrosine-phosphorylated activated proteins, thus inducing the rearrangements of the actin cytoskeleton in specifying growth cone decisions. Misshapen, a Ste20-like serine/threonine kinase, functions as a downstream effector of Dock in regulating R1-R6 growth cone targeting. Msn is the Drosophila homolog of Nck-interacting kinase (NIK) in vertebrates, that belongs to the GCK subfamily of the Ste20-like serine/threonine kinases. In addition to Msn, Dock also interacts with PAK-kinase (Dpak), a member of the Pak subfamily of the Ste20-like serine/threonine kinases, in regulating R cell growth cone guidance. Unlike Msn, Pak also has a binding site (i.e., CRIB) for activated small Rho family GTPases Rac and Cdc42, and has been shown to be regulated by the combined action of Dock and the guanine nucleotide exchange factor Trio. Unlike loss of dock, however, mutations in the trio locus do not affect R1-R6 growth cone targeting. Those observations suggest that Dock links multiple signals in R cell growth cones. Msn and Pak might function separately to induce distinct cytoskeletal changes in response to different Dock-linked upstream signals. Alternatively or additionally, Msn and Pak might cooperate to mediate at least some Dock-dependent signaling events (Ruan, 2002).

It is proposed that Dock-linked stop signals activate Msn, which in turn induces the rearrangements of actin cytoskeleton leading to the shutdown of R1-R6 growth cone motility at the lamina. While studies on vertebrate Paks have provided some molecular details about the link between the activation of Pak and the changes in actin cytoskeleton, it has remained unclear how Msn/NIK and other GCK family members regulate cytoskeletal changes. bif, like msn, is required for the proper targeting of R1-R6 growth cones. Msn binds directly to Bif and phosphorylates Bif in vitro. Genetic interaction between msn and bif indicates that Bif functions downstream of Msn to regulate actin reorganization. Msn is also capable of modulating Bif-induced cytoskeletal changes in cultured cells. It is proposed that Bif links the activation of Msn and the changes in actin cytoskeleton to decelerate R1-R6 growth cone motility at the target region (Ruan, 2002).

A genetic approach was undertaken to search for genes encoding other components of the Dock-Msn signaling pathway. A msn gain-of-function phenotype was generated by overexpressing Msn in R cell growth cones. In wild-type, after exiting the optic stalk, R1–R6 growth cones migrate over a distance of ~20 µm within the lamina, then stop extension and expand significantly in size to form the lamina plexus, while R7 and R8 growth cones migrate through the lamina into the medulla. In larvae overexpressing msn, however, many R1–R6 growth cones terminate before reaching the lamina plexus, a phenotype in marked contrast to that in msn loss-of-function mutants in which many R1–R6 growth cones failed to stop at the lamina layer. Overexpression of msn also disrupts the regular array of R7 and R8 growth cones in the medulla. It was reasoned that if a gene functions downstream of msn, then reducing the dosage of this gene by half would decrease the level of signaling through the Msn pathway, thereby suppressing the Msn hyperactivation phenotype. Thus, a screen for modifiers of this msn hyperactivation phenotype might lead to the identification of other components of the Dock-Msn signaling pathway (Ruan, 2002).

To determine the feasibility of this approach, an examination was carried out to see if reducing the dosage of other genes in the genome would dominantly modify the pretarget termination phenotype in flies overexpressing msn. Analysis of deficiency lines shows that reducing the dosage of a number of cytological regions could dominantly modify the msn overexpression phenotype, indicating that this msn gain-of-function genetic background is indeed sensitive to the dosage of other genes. This approach was undertaken to examine the potential interaction between msn and a set of genes that had been previously implicated in regulating cytoskeletal changes in Drosophila. Interestingly, it was found that reducing the dosage of bif largely suppresses the pretarget msn hyperactivation phenotype. Suppression was observed using two different bif alleles R38 (~56%, n = 34) and R47 (~75%, n = 20). In contrast, reducing the dosage of bsk, a gene that encodes the fly homolog of C-Jun N-terminal kinase and has been shown previously to function downstream of Msn to regulate dorsal closure in early embryos, showed no effect. This result argues against the idea that msn and bsk interact similarly in R cell growth cones (Ruan, 2002).

Among other genes examined, reducing the dosage of cdc42 (~10%, n = 34 hemispheres) or disabled (dab) (~30%, n = 12 hemispheres) enhances the msn overexpression phenotype. In eye-brain complexes showing enhanced phenotype, R cell axons terminate within the optic stalk or the eye disc, which is never observed in wild-type larvae overexpressing msn. No modification of the msn overexpression phenotype is observed by reducing the dosage of either dpak, Rho1, all three Rac genes Rac1-Rac2-Mtl, or chickadee (chic), which encodes the fly homolog of profilin (n = 26 hemispheres) (Ruan, 2002).

The observation that Msn hyperactivation phenotype is sensitive to the dosage of bif, together with previous reports showing the link between Bif and the actin cytoskeleton, raises the interesting possibility that Bif functions downstream of Msn to regulate cytoskeletal changes in R cell growth cones (Ruan, 2002).

How could upstream stop signals be relayed through Dock and Msn to Bif? While the molecular nature of upstream regulators of Dock in R cell growth cones remains unknown, it has been shown that Dock functions downstream of the guidance receptor Dscam, a member of immunoglobulin superfamily, in larval photoreceptor growth cones. Dock mediates growth cone signaling through recruiting Pak to activated Dscam. It is speculated that Dock relays stop signals to Msn similarly in the adult visual system to regulate R1–R6 growth cone targeting. Dock and Msn form an in vivo complex at both larval and adult stages. It is speculated that when R1–R6 growth cones reach the lamina, stop signals produced by the intermediate target (i.e., lamina marginal glia) activate their growth cone receptors, which subsequently recruit the Dock-Msn complex. Consequently, this might bring Msn in proximity to Bif or directly stimulate the activity of Msn through a conformational change, leading to increased phosphorylation on Bif (Ruan, 2002).

While Msn associates directly with Bif and phosphorylates Bif in vitro, it is not known if Msn and Bif constitutively associate in R cell growth cones or if the association between them is transient and dependent on stop signals from the target region. Given that Bif is predominantly associated with the plasma membrane, it is speculated that the recruitment of Dock-Msn complex by activated growth cone receptors might relocate Msn from cytoplasm to plasma membrane, thus bringing Msn and Bif together. The formation of such a signaling complex could allow Msn to regulate the function of Bif through phosphorylation, or relocate Bif into a specific region within the growth cone to initiate downstream events. Testing these speculations awaits the identification of upstream stop signals and R cell growth cone receptors (Ruan, 2002).

How does the interaction between Msn and Bif regulate the changes in growth cone cytoskeleton? One possible scenario is that Bif, activated by Msn, induces the redistribution of F-actin within the growth cone, leading to the withdrawal of the growth cone leading edge. Consistently, cell culture studies show that Msn can reorganize Bif-induced actin fibers and reduce the number and the length of filopodia-like structures. Several studies have also demonstrated that the arrest of growth cone extension in vitro could be achieved through growth cone collapse, which is due at least in part to the loss of actin bundles at the leading edge of the growth cone. While the fact that R cell growth cones expand significantly in size upon reaching the target region argues against a mechanism involving the collapse of the whole growth cone, it remains possible that the initial termination involves partial growth cone collapse at the leading edge after exposure to stop signals. Since R cell growth cone morphology remains normal in msn and bif mutants, the view is favored that the interaction between Msn and Bif regulates the reorganization of actin filaments in spatially restricted domains within the growth cone without affecting the general structure of growth cone cytoskeleton (Ruan, 2002).

That Bif colocalizes with F-actin and can promote actin polymerization in cultured cells, together with the report that immobilized F-actin could pull down Bif from fly lysates (Sisson, 2000), suggest strongly that Bif associates either directly or indirectly with F-actin filaments. Such association may stabilize actin filaments, thus contributing to the dramatic increase in the level of F-actin observed in cultured cells. Whether Bif also plays a similar role in promoting actin polymerization in R cells remains unclear since loss of bif affects neither growth cone outgrowth nor the amount of F-actin in R cell bodies and growth cones. One possible explanation is that other functionally redundant proteins maintain the level of F-actin in the absence of Bif. It is speculated that Bif may have at least two activities in R cell growth cones. Bif may be functionally redundant with other proteins to promote actin polymerization. Additionally, it may also play a role in restructuring F-actin in terminating R cell growth cones. The latter activity of Bif may resemble that of the Dictyostelium actin binding protein Severin. Severin and its mammalian homolog Gelsolin can bind to F-actin and fragment actin filaments. Interestingly, both Severin and Gelsolin are also phosphorylated by members of the GCK family of Ste20-like kinases in vitro. It is speculated that phosphorylation of Bif by Msn might directly increase such activity of Bif, thus inducing the shortening and aggregation of F-actin filaments leading to growth cone termination (Ruan, 2002).

In addition to Msn, it has been shown recently that Bif also interacts with a Protein phosphatase 1 at 87B (Helps, 2001). Bif binds directly to PP1 and modulates its phosphatase activity in vitro. The interaction between Bif and PP1 is proposed to be essential for morphological changes in R cell bodies later in the development. Interestingly, in human T lymphocytes, PP1 has been shown to reactivate cofilin by dephosphorylation. Similarly, PP1 might also interact with Bif in R cell growth cones to dephosphorylate Bif and/or other actin binding proteins phosphorylated by Msn, thus promoting the effective remodeling of actin cytoskeleton after the initial targeting of R cell growth cones (Ruan, 2002).

In conclusion, Bif functions downstream of Msn to regulate R1–R6 growth cone targeting. These results suggest a role for Bif to induce the remodeling of actin cytoskeleton in R cell growth cones in response to a Dock-linked stop signal. Future studies will identify proteins that function upstream of Dock and Msn (i.e., stop signals and growth cone receptors). The Dock-Msn-Bif pathway is unlikely to be the only signaling pathway for R1–R6 growth cone targeting. In dock null mutants, many R1–R6 axons still terminate correctly, indicating the involvement of additional signaling mechanism for R cell growth cone targeting. Indeed, mutations in the brakeless gene disrupted R1–R6 growth cone targeting more severely than either loss of dock, msn, or bif. Dissection of the brakeless pathway will identify additional key players required for R cell growth cone targeting (Ruan, 2002).

GENE STRUCTURE

bif encodes several transcripts, the most predominant form of which is ~4.5 kb long. Their expression is differentially regulated during development, with relatively stronger expression in embryos, pupae, and adults and weaker expression in larvae. In the adult head, a larger transcript of ~7 kb is the predominant species. bif cDNA clones were isolated by screening embryonic cDNA libraries. They are all related to each other by restriction mapping and differ only in the extent of their respective 5' and 3' regions. The two largest clones recovered contain inserts of ~4.6 and ~5.2 kb (referred to henceforth as the bif cDNAs); these have been sequenced and encode the same open reading frame. Each of these two cDNAs spans >80% of the genomic region between the 3' end of the DPTP10D transcription unit and the Is(2)P842 insertion site. The major difference between the two species is at their respective 3' termini, where different polyadenylation sites have been utilized. These two bif cDNAs are each comprised of six exons; they contain the same 3,189-bp open reading frame, which starts with an ATG at nucleotide 305 in exon 2 and ends with a TGA at nucleotide 3493 in exon 5 (Bahri, 1997).

cDNA clone length - 5197

Bases in 5' UTR - 304

Exons - 6

Bases in 3' UTR - 1569

PROTEIN STRUCTURE

Amino Acids - 1063

Structural Domains

bifocal encodes a putative novel protein of 1,063 aa that has no striking homology to any known proteins in the databases searched. bif is rich in serine/threonine (19.8%) and contains a stretch of 15 glutamine residues at its N terminus (aa 66 to 71). Six of the total of 10 tyrosine residues are concentrated in the C terminus of the bif protein (aa 916 to 1018). The Prosite search predicted a possible tyrosine phosphorylation in the sequence 908KSTKEVPDY (Bahri, 1997).

date revised: 18 February 2003

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

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