SEARCH

SEARCH BY CITATION

Keywords:

  • Torso;
  • Drosophila;
  • review;
  • type III receptor tyrosine kinase;
  • embryonic development;
  • terminal system;
  • PDGF receptor;
  • c-Kit

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

The Torso receptor tyrosine kinase (RTK) is required for cell fate specification in the terminal regions (head and tail) of the early Drosophila embryo. Torso contains a split tyrosine kinase domain and belongs to the type III subgroup of the RTK superfamily that also includes the platelet-derived growth factor receptors, stem cell or steel factor receptor c-Kit proto-oncoprotein, colony-stimulating factor-1 receptor, and vascular endothelial growth factor receptor. The Torso pathway has been a model system for studying RTK signal transduction. Genetic and biochemical studies of Torso signaling have provided valuable insights into the biological functions and mechanisms of RTK signaling during early Drosophila embryogenesis. Developmental Dynamics 232:656–672, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Torso is a maternally contributed receptor tyrosine kinase (RTK), i.e., it is made by the nurse cells during oogenesis and deposited into the oocyte, which gives rise to the mature egg (Casanova and Struhl, 1989; Sprenger et al., 1989; reviewed by Duffy and Perrimon, 1994; Furriols and Casanova, 2003). Torso is the earliest known RTK to be activated during Drosophila embryogenesis. It is required for cell fate specification and pattern formation in Drosophila embryonic cells located in the anterior and posterior terminal regions of the embryo. Along with the anterior, posterior, and dorsoventral systems, the terminal signaling pathway is one of the four maternally contributed systems essential for setting up the initial coordinates of the egg, allowing progressive elaboration of the embryonic body plan. Torso is only transiently present in the early embryo and is not detected after gastrulation. In the early syncytial embryo, Torso is uniformly distributed on the membrane but is activated only at the two terminal regions by ligands whose production is spatially restricted to the termini (Casanova and Struhl, 1989; Sprenger et al., 1989; Stevens et al., 1990). Terminal cell fates, thus, are specified by localized activation of a uniformly distributed RTK (reviewed by St. Johnston and Nusslein-Volhard, 1992; Duffy and Perrimon, 1994; Furriols and Casanova, 2003).

Aside from its role in pattern formation, the Torso pathway has been a paradigm for genetically dissecting RTK signaling mechanisms. Early genetic studies of Torso signal transduction mainly focused on isolating components mediating Torso signaling and determining their sequence of action. Genetic screens, epistatic analyses, and mosaic analyses produced valuable information in this regard. Genetic studies of Torso signaling have demonstrated that Torso transduces signals mainly by means of the Ras–extracellular signal regulated kinase (ERK) signaling cassette (reviewed by Lu et al., 1993b; Duffy and Perrimon, 1994; Perrimon et al., 1995), but it also activates STAT, which plays a role in mediating a subset of Torso signals (Li et al., 2002, 2003). Biochemical studies have revealed that the signaling strategies used by Torso are highly conserved among all RTKs (Cleghon et al., 1996, 1998).

Several RTKs have been identified in Drosophila that play essential roles in distinct developmental processes. Besides Torso, two other well-studied Drosophila RTKs are the epidermal growth factor receptor (EGFR) and Sevenless. The Sevenless pathway, which signals by means of the Ras-ERK signaling cassette, is required for specifying the R7 photoreceptor cell fate during eye development (reviewed by Wassarman et al., 1995). The Drosophila EGFR homolog (also known as DER) acts through the same signaling cassette. It is used repeatedly by a variety of tissues during different stages of Drosophila development to regulate cell fate determination, cell proliferation, and cell migration (reviewed by Shilo, 2003). These RTKs share common signaling mediators and mainly use the canonical Ras-ERK pathway to transduce signals.

The Torso pathway specifies cell fate by inducing target gene expression. Two of the best-known Torso target genes are tailless (tll) and huckebein (hkb), which encode nuclear hormone receptor and zinc-finger transcription factors, respectively (Pignoni et al., 1990; Weigel et al., 1990). In the embryonic posterior, Torso signaling induces the expression of tll and hkb, specifying the posterior terminal cell fates. In the anterior region, however, the combined activities of Torso and the anterior morphogen Bicoid determine the proper formation of the head (reviewed by Schmidt-Ott, 2001; Ephrussi and St. Johnston, 2004). Thus, the correct specification of terminal cell fates depends on developmentally controlled activation of the Torso signaling pathway, the participation of additional factors/pathways, and the developmental history of the cell nuclei.

IDENTIFICATION OF TORSO

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

After fertilization, the early Drosophila embryo undergoes 13 rapid synchronous nuclear divisions without cytokinesis, resulting in a syncytium, with a single cell body encompassing all of the nuclei (Bate and Martinez Arias, 1993; Campos-Ortega and Hartenstein, 1997). Each synchronous nuclear division cycle takes only 10 min on average. At the end of the 13th cycle, cellularization takes place, resulting in a cellular blastoderm. By this stage, all cells have acquired positional coordinates according to their spatial localization in the embryo.

It had been postulated that the first steps of pattern formation and cell fate specification of a multicellular organism are determined by morphogen gradients that are set up by asymmetrical localization of maternal gene products in the early syncytium (Wolpert, 1996). Nusslein-Volhard and Wieschaus sought to identify systematically, by genetic screens, the maternal-effect genes required for patterning the Drosophila embryo (Schupbach and Wieschaus, 1986; Nusslein-Volhard et al., 1987). These genes encode products that are synthesized in the ovary of the mother and are directly deposited into the oocyte. One of the genes identified in these screens was torso.

Genetic screens designed to isolate mutations affecting embryonic patterning identified three groups of maternal genes whose functions are essential for specifying cell fates along the anterior/posterior axis and another group specifying fates along the dorsal/ventral axis of the embryo (reviewed by St. Johnston and Nusslein-Volhard, 1992; Nusslein-Volhard, 2004) (Ephrussi and St. Johnston, 2004). Genes of the Bicoid group control pattern formation of the anterior region, including part of the head and the thorax. The Nanos group specifies cell fates in the abdominal region. The nonsegmented regions of the tail and part of the head are controlled by the Torso, or terminal, group of genes. The fourth group of maternal genes, the dorsal class, is essential for dorsoventral patterning. Together these four groups of maternal gene products act in the early embryo to define four sets of positional coordinates and, thus, begin to establish the body plan of the animal (reviewed by St. Johnston and Nusslein-Volhard, 1992; Nusslein-Volhard, 2004; Ephrussi and St. Johnston, 2004).

As expected for a maternal gene, torso homozygous mutant flies are perfectly viable, but the females lay eggs that exhibit gross patterning defects—they are missing the head and tail structures, hence the name torso. Mutations in other members of the terminal group genes show similar embryonic patterning defects: the embryos exhibit normal patterning of central structures but are missing tissues derived from the acron and telson (Duffy and Perrimon, 1994). In Drosophila, the acron refers to the anterior structures including the labrum and head skeleton. The telson includes the last (eighth) abdominal segment, posterior spiracles, Filzkörper, and Malphigian tubules (Fig. 1).

thumbnail image

Figure 1. Expression patterns of the Torso target gene tailless and cuticle phenotypes of two classes of torso mutants. Loss-of-function (lof) mutations in torso eliminate posterior tailless (tll) expression and posterior structures; gain-of-function (gof) mutations result in expansion of tll expression and ectopic and enlargement of posterior structures at the expense of central elements. Note the loss of Filzkörper (arrow in wild-type [WT] embryo) and A8 in the torso lof mutant embryo and deletion of central denticle belts and enlargement and ectopic Filzkörper in torsoGOF embryo. Embryos are shown anterior to the left, posterior right. Dark blue stain in the left panels indicates tll mRNA, as detected by in situ hybridization. Darkfield photographs of embryonic cuticular preparations are shown in the central panels, and schematic representations of these cuticular patterns are shown on the right. Red color indicates tissues derived from the acron (left) and telson (right). These tissues include the head skeletons (open arrows), the eighth abdominal denticle belt (A8), and the Filzkörper (closed arrows). A1 to A8, abdominal denticle belts; T1 to T3, thoracic segments.

Download figure to PowerPoint

TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Molecular investigations led to the discovery that Torso is an RTK of 923 amino acids that consists of an N-terminal extracellular region, a single transmembrane domain, and a C-terminal cytoplasmic tyrosine kinase domain (Casanova and Struhl, 1989; Sprenger et al., 1989). The extracellular domain, responsible for ligand binding, shows no homology to any known RTKs. Although full-length Torso is most homologous to the RET (REarranged during Transfection) proto-oncoprotein (reviewed by Jhiang, 2000), because the sequence homology includes the kinase domain and C-terminal tail (49% identical and 67% homologous, respectively), its split tyrosine kinase domain makes it structurally homologous to the type III RTKs (Fig. 2). Moreover, its kinase domain is significantly similar to those of the platelet-derived growth factor (PDGF) receptors and c-kit.

thumbnail image

Figure 2. Structure of receptor tyrosine kinases. Schematic outline of Torso and other receptor tyrosine kinases (RTKs). The type III RTK subfamily also includes c-Kit, CSF-1R (also known as c-FMS), and FMS-like tyrosine kinase-3 (FLT-3, not shown). These RTKs contain a kinase insert region with phosphotyrosine residues that serve as docking sites for downstream signaling molecules. This feature is not shared by other subfamily of RTKs, such as fibroblast growth factor receptor (FGF-R) and RET (REarranged during Transfection). The extracellular domain of Torso is not similar to the type III RTKs. PDGF-R, platelet-derived growth factor receptor; VEGF-R, vascular endothelial growth factor receptor; Ig, immunoglobulin.

Download figure to PowerPoint

A subgroup of RTKs (type III), including the PDGF receptors, c-Kit, CSF-1R (also know as c-FMS), FMS-like tyrosine kinase-3 (FLT-3), and vascular endothelial growth factor receptor (VEGF-R), share a common structural feature: the intracellular tyrosine kinase domain is split by an insert region comprising a stretch of approximately 100 hydrophobic amino acid residues (van der Geer et al., 1994; Hanks and Hunter, 1995). Torso shares this structural feature and, thus, belongs to this subgroup. In vertebrates, the type III RTKs are likely derived from a single ancestral RTK with a split tyrosine kinase domain for the following reasons. First, in mammals and pufferfish, genes for PDGFαR and PDGFβR are tandemly linked with c-Kit and CSF1R, respectively, and share a similar genomic organization, suggesting that these genes are duplicates of an ancestral RTK gene (Roberts et al., 1988; Williams et al., 2002). Second, extensive sequence homology is evident along the whole molecule among type III RTKs. PDGFβR is most similar to PDGFαR, C-Kit, and CSF-1R, with sequence homology of 59%, 51%, 48%, and 45%, respectively. It is also significantly homologous to FLT-3 and VEGF-R, although less similar in sequence. Outside of the type III RTK subgroup, the PDGF receptors are significantly homologous to the fibroblast growth factor (FGFR) receptors and RET, with sequence homology mainly restricted to the kinase domain. Third, certain signaling mechanisms are conserved within the type III subgroup of RTKs. For instance, many type III RTKs are capable of activating Ras-MAPK, STAT, PLCγ, and phosphatidylinositol 3-kinase (PI3-kinase) signaling (van der Geer et al., 1994). Although the role of Drosophila PI3K has not been evaluated with regard to Torso signaling, Torso is able to activate Ras and STAT92E (Li et al., 2003).

Torso and the PDGF receptors are not only similar at the structural level, but they also share certain functional or signaling specificities. It has been shown recently that a PDGFαR/Torso fusion protein can fulfill certain functional requirements of PDGFαR in knockin mice, whereas a similar fusion with mouse FGFR1 fails to rescue any aspect of the PDGFαR null phenotype (Hamilton et al., 2003). Although the combined tyrosine kinase domain of Torso is more homologous to that of vertebrate bFGFR1 (52% identical and 72% similar) than to that of PDGFαR (50% identical and 67% similar), results from the above in vivo experiments suggest that the structural similarity, i.e., the split kinase domain, is more important for the shared specificity between Torso and PDGFαR. This finding suggests that the specificity of RTK signaling lies more in the particular downstream signaling molecules or adaptor proteins with which the receptor can associate than in the sequence of its tyrosine kinase domain. From comparisons with kinases of known structure, it is believed that the insert forms a loop that protrudes from the globular kinase domain surface (reviewed by van der Geer et al., 1994). Indeed, the kinase insert regions of the PDGF receptors and Torso contain many phosphotyrosine residues and play important roles in recruiting adaptor proteins (see below; reviewed by Heldin and Westermark, 1999).

GENETIC STUDIES OF TORSO SIGNALING

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Identification of Torso Pathway Components by Genetic Screens

The functions of a gene can best be inferred from its loss-of-function phenotypes. Genetic screens by Nusslein-Volhard, Wieschaus, and others identified several loci, referred to as the terminal or torso group genes, because when mutated they share similar embryonic phenotypes (reviewed by Duffy and Perrimon, 1994). The torso group genes initially included torso (tor), torso-like (tsl), trunk (trk), female sterile (1) Nasrat [fs(1)N], female sterile (1) pole hole [fs(1)ph], tll, and hkb (Degelmann et al., 1986; Perrimon et al., 1986; Nusslein-Volhard et al., 1987; Klingler et al., 1988; Casanova and Struhl, 1989; Sprenger et al., 1989; Stevens et al., 1990; Weigel et al., 1990). Mutations in the first five loci are female sterile and in the latter two are zygotic lethal. Mutations in these genes result in similar embryonic patterning defects; the embryos are missing the tissues derived from the acron and telson but the embryo has perfect central elements. The genetic screens conducted by Nusslein-Volhard and Wieschaus were “saturated,” because multiple alleles were recovered for most genes isolated. However, such genetic screens could not easily identify essential genes (those mutations would cause lethality) whose maternal products affect embryonic patterning or genes required at multiple developmental stages or with multiple functions. To study the maternal functions of zygotically essential genes, Perrimon's group used a dominant female sterile mutation, ovoD1, to produce germline mosaic females and isolated X-linked zygotic lethals with specific maternal-effect phenotypes (Perrimon et al., 1989; see below). This study identified lethal (1) pole hole [l(1)ph] and corkscrew (csw) as novel members of the torso class judging by the similarity of their loss-of-function phenotypes to those of torso mutant embryos. Later molecular studies revealed that l(1)ph and csw encode the Drosophila homologs of Raf and the SH2 domain-containing tyrosine phosphatase (SHP2), respectively (Ambrosio et al., 1989b; Perkins et al., 1992).

Epistatic Analyses

Gain-of-function mutations are useful for determining the order of the components in a genetic pathway. Gain-of-function alleles of torso (torsoGOF) cause patterning defects that appear to be the “opposite” of its loss-of-function phenotypes (Klingler et al., 1988; Schupbach and Wieschaus, 1989; Szabad et al., 1989). These embryos exhibit overgrown and ectopic terminal structures at the expense of central elements (Fig. 1). The expansion of terminal cell fates is associated with ectopic and higher levels of expression of the target gene tll (Fig. 1). Genetic epistatic analyses—studies of phenotypes resulting from double mutants of torsoGOF and a loss-of-function allele of the terminal class—demonstrated that four of the terminal class genes act upstream of torso, whereas all the others function downstream. Specifically, mutations in tsl, trk, fs(1)ph, and fs(1)N do not affect torsoGOF phenotypes and, thus, were postulated to be involved in the production of Torso ligands, whereas mutations in genes such as csw, Draf, and Dsor1, suppress torsoGOF phenotypes, which is consistent with their being essential cytoplasmic transducers of Torso signals (Ambrosio et al., 1989b; Stevens et al., 1990; Perkins et al., 1992; Casanova and Struhl, 1993; Tsuda et al., 1993). Suppression of torsoGOF phenotypes by mutations in another gene routinely has been taken to implicate the gene in Torso signaling.

Use of Germ-Line Clone Embryos to Study Signaling Mechanisms

Drosophila offers an advantageous genetic system for studying mechanisms of intracellular signal transduction. The Torso pathway operates during the early stages of embryogenesis to specify terminal cell fates. In the early embryo, all signaling components or other molecules are maternally contributed, i.e., they are made by the nurse cells during oogenesis and deposited into the oocyte, which gives rise to the mature egg. Importantly, for reasons not completely understood, many essential signaling molecules, such as Ras1 and Draf, appear dispensable for oogenesis, such that eggs completely lacking these molecules are produced and can proceed to develop. This finding is in contrast to the somatic tissues where these molecules are required for cell viability and/or division. This particular aspect of Drosophila development allows one to study the effects of systematic removal of one or more gene products on embryonic development. The Flp-mediated dominant female sterility (Flp-DFS) technique developed in the Perrimon laboratory allows the generation of large quantities of embryos derived from germ cells homozygous for one or more particular mutations (Perrimon et al., 1989; Chou and Perrimon, 1992; Perrimon et al., 1996). These germ-line clone (GLC) embryos are equivalent to knockout animals for particular maternal gene products. The technique provides a powerful tool for genetic screens as well as for dissection of signaling pathways in early embryos (reviewed by Perrimon et al., 1995). An example is provided in the following section.

Mechanisms of Raf Activation: Insights From GLC Embryos

Despite extensive research efforts, the mechanisms underlying Raf activation still remain unclear. The Torso signaling system provides unique advantages by virtue of the simplicity and accessibility of Drosophila genetics: There are single copies of genes encoding components of the Ras-ERK signaling cassette, and the gene products can be totally removed, singly or in combination, from the early embryo by using the Flp-DFS technique, a manipulation that presently cannot be performed in any other organism. Many such analyses have been successfully performed to answer questions pertaining to Ras/Raf signaling (reviewed by Perrimon et al., 1995).

Studies using Drosophila GLC embryos have provided mechanistic insights into the role of Ras in Raf activation. Unlike torso or Draf GLC embryos, in which no posterior tll expression can be detected, GLC embryos null for Ras1 still retain residual levels of tll expression in the posterior domain. Because Drosophila Ras2 is not redundant with Ras1, the data suggest that Raf can be activated by a Ras-independent mechanism (Hou et al., 1995). However, when the activities of several variant Draf proteins were examined in the complete absence of Ras1, it was found that all forms of Draf that had been believed previously to be “constitutively active” require Ras1 for activity (Li et al., 1998). These variant Draf proteins included N-terminally truncated Draf as well as the membrane targeted or both. Therefore, it appears that the role of Ras in Raf activation is not limited to membrane translocation of Raf through Ras-Raf association, a widely accepted model. Ras is also essential for the activation of an additional factor(s) that in turn activates Raf. Results from genetic analyses of Draf, thus, stand in contrast to studies on cultured cells, which have led to the proposal that the sole function of Ras in Raf activation is to bring it to the membrane (Leevers et al., 1994; Stokoe et al., 1994). Examination of a Ras null condition that can be achieved only in Drosophila GLC embryos has led to the conclusion that Ras has a novel second function, which is to activate a “Raf activator” that in turn activates Raf (Li et al., 1998; see Fig. 4).

thumbnail image

Figure 4. Mechanism of Torso signal transduction. Ligand (Trunk) binding triggers dimerization and autophosphorylation of Torso on multiple tyrosine residues. Phosphotyrosines located in the activation loop of the tyrosine kinase domain are essential for Torso enzymatic activity. Phosphotyrosines located outside of the kinase domain serve to recruit Csw and possibly DSHC and other adaptors or SH2-containing signaling molecules, which may include STAT92E. The adaptor molecules redundantly or synergistically recruit Son of Sevenless (Sos) to the membrane. Sos converts GDP-Ras1 to GTP-Ras1, linking Torso activation to the Ras1/Draf/Dsor1/Rolled signaling cassette. In addition, Torso is able to activate Draf by means of a Ras1-independent pathway (dashed line 1), and Ras1 plays a role in activating an unknown Draf activator (dashed line 2) in addition to binding to Draf. Thus, multiple signaling routes originating from Torso can converge on the Draf kinase, leading to expression of downstream target genes. In the posterior terminal region, Torso signaling induces target gene tailless (tll) and huckebein (hkb) mainly by derepression, counteracting repressor complexes, including Capicua (Cic) and Groucho (Gro) and possibly other proteins that bind to the regulatory regions of tll and hkb. Gro is a conserved transcription repressor that is recruited to specific genes by its DNA-binding partner Cic. After relief of repression, the expression of tll and hkb depends on participation of transcription activators that may or may not require input from Torso signaling. Gene regulation by Torso in the anterior terminal region is more complex (see text for detail) and is not depicted in this schematic drawing.

Download figure to PowerPoint

COMPONENTS OF THE TORSO PATHWAY

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Genetic studies have led to the conclusion that Torso and other known Drosophila RTKs share common signaling mediators and that they mainly use the canonical Ras-ERK pathway to transduce signals. Evidence that many of the same downstream components are required for transducing Torso signals was derived from examining their loss-of-function mutant phenotypes and epistatic relationships with torsoGOF mutations. For example, removal of Draf from early embryos, by means of germline mosaics, results in phenotypes identical to those of torso null mutations (Ambrosio et al., 1989b; Hou et al., 1995). Similar phenotypes were also observed for Dsor1 (Tsuda et al., 1993). These results suggest that Torso signals may be mediated entirely by the Ras1/Draf/Dsor1/Rolled signaling cassette. Table 1 lists components of the Torso signaling pathway isolated and/or confirmed by the similarity of their loss-of-function phenotypes to those of torso mutations or by the ability of the mutations to suppress torsoGOF phenotypes.

Table 1. Molecules Involved in Torso Signaling
Gene/proteinMolecule/motifVertebrate homologEffects of mutation on Torso signalingReference
  1. aNA, not analyzed; GOF, gain-of-function; HGF, hepatocyte growth factor.

Ligand production    
 torso-likeSecreted; GlycoproteinNoneLoss of terminiStevens et al., 1990; Savant-Bhonsale et al., 1993
 fs(1)NasratCell surface moleculeNoneLoss of terminiJiménez et al., 2002
 fs(1) phCell surface moleculeNoneLoss of terminiJiménez et al., 2002
 trunkCystine knot motif; Sereted; LigandNoneLoss of terminiCasanova et al., 1995
Receptor    
 TorsoReceptor Tyrosine KinaseType III RTKLoss of terminiSprenger et al., 1989; Casanova and Struhl, 1989
Adaptors    
 DrkSH2-SH3 adaptor proteinGRB2Partial loss of terminiSimon et al., 1991; Hou et al., 1995
 DshcPhosphotyrosine interaction, SH2, PH-like domainSHCPartial loss of terminiLuschnig et al., 2000
 Dospleckstrin homology domain proteinGabPartial loss of terminiRaabe et al., 1996; Herbst et al., 1996
Enzymes    
 CorkscrewSH2 Protein-tyrosine phosphataseSH-PTP2 or SHP2Suppresses torso GOF phenotypePerkins et al., 1992
 SosGuanine nucleotide exchange factorSOSPartial loss of terminiSimon et al., 1991; Hou et al., 1995
 Src64BSrc family tyrosine kinaseSrcReduced Draf activityF. Xia & W.X. Li, unpubl. results
 PLC γPhospholipasePLCγNACleghon et al., 1996
Core signaling cassette    
 Ras1Small GTPaseRasLoss of terminiSimon et al., 1991; Hou et al., 1995; Lu et al., 1993
 Draf or l(1)phRaf serine/threonine kinaseRafLoss of terminiAmbrosio et al., 1989
 Dsor1MAP kinase kinase; dual specificity kinaseMEKLoss of terminiTsuda et al., 1993
 rolledMAP kinase; serine/threonine kinase ERKERKSuppresses torso GOF phenotypeBrunner et al., 1994; Li and Li, 2003
Cofactors/scaffolds    
 KsrSerine/threonine kinaseKSRLoss of terminiTherrien et al., 1995
 CNKPDZ, SAM, and PH domain proteinCNKLoss of Ras signalingTherrien et al., 1998
 14-3-3 ζPhosphoserine binding protein14-3-3ζLoss of posterior midgutLi et al., 1997; Kockel et al., 1997
 Hsp83Heat shock proteinHsp90Suppresses torso GOF phenotypeLi and Li, 2003; van der Straten et al., 1997
Negative regulators    
 D-RasGapRas GTPase activating proteinp120 Ras-GAPIncreased tll expressionCleghon et al., 1998
 Gap1Ras GTPase activating proteinGap1Increased tll expressionHou et al., 1995
 SproutyFAD/NAD(P)-binding domain proteinSprouty1/2Suppresses weak torso alleleCasci et al., 1999
 CapicuaHMG-box DNA-binding factorCicCauses torso GOF phenotypeJiménez et al., 2000
 GrouchoWD-repeat transcrioption corepressorTLEIncreased tll expressionParoush et al., 1997
 HrsHGF-regulated tyrosine kinase substrateHrsIncreased tll expressionLloyd et al., 2002
Target genes and protein    
 taillessNuclear hormone receptorTlxLoss of poserior terminiPignoni et al., 1990
 huckebeinZinc finger transcription factorSp1Loss of posterior midgutWeigel et al., 1990
 BicoidHomeo domain transcription factorPitx 1/2Loss of headDriever and Nusslein-Volhard, 1988
 orthodenticleHomeo domain transcription factorOtxHead defectsFinkelstein and Perrimon, 1990
Other    
 STAT92ESH2-containing transcription factorSTAT5suppresses torso GOF phenotypeLi et al., 2002

Ligands: Activation of Torso

Torso is uniformly distributed along the membrane of the early embryo but is activated in a spatially graded manner by diffusible ligands that are produced only at the terminal regions; their diffusion to other regions is impeded by interaction with Torso (reviewed by Furriols and Casanova, 2003; LeMosy, 2003). Trunk (Trk), a secreted protein synthesized by the oocyte during oogenesis (Schupbach and Wieschaus, 1986), has been proposed to be the Torso ligand (Casanova et al., 1995). Trk contains a cystine knot motif in its C-terminal region, a feature shared by certain growth factors and extracellular ligands. In particular, features of the Trk sequence resemble those of Spätzle (Spz), a secreted protein and proteolytically activated ligand for the dorsal–ventral patterning receptor Toll (Morisato and Anderson, 1994; Casanova et al., 1995). Such similarity suggests that Trk might also be cleaved by proteolysis to generate an active Torso ligand. Indeed, a C-terminal fragment of Trk but not the whole molecule is sufficient to activate Torso in the absence of the other three genes normally required for Torso activation, namely, torso-like, fs(1)Nasrat, and fs(1)ph (Savant-Bhonsale and Montell, 1993; Casali and Casanova, 2001). This finding has led to the hypothesis that Nasrat, Polehole, and Torsolike are involved in the proteolytic processing and activation of Trk.

Protein products encoded by fs(1)Nasrat and fs(1)ph accumulate at the oocyte surface during late stages of oogenesis and are essential for the cross-linking of vitelline membrane proteins, ensuring correct formation of the egg-shell as well as its integrity (Jimenez et al., 2000). In addition, Nasrat and Polehole are also essential for stabilizing Torsolike, which has been shown recently to be tethered to the inner surface of the vitelline membrane (Stevens et al., 2003). In the absence of Nasrat and Polehole, Torsolike is not efficiently localized to the poles of the early embryo (Jimenez et al., 2000; Stevens et al., 2003). Moreover, misexpression of Torsolike at different levels and in different genetic backgrounds suggests that Torsolike depends on Nasrat or Polehole not only for localization but also for activity; in addition, yet another unidentified factor at the poles is required for Torsolike function (Stevens et al., 2003; Fig. 3). Given all this complexity, evidence that Trk is actually cleaved proteolytically is still lacking, and the precise mechanism of how Nasrat, Polehole, and Torsolike cooperate to bring about Trk activation in conjunction with an unknown factor is still not clear.

thumbnail image

Figure 3. Activation of Torso. Nasrat and Polehole are present uniformly on the outer plasma membrane of the embryo. Torso-like is associated to the inner surface of the vitelline membrane in the anterior and posterior pole. Trunk is processed and activated only at the poles by collective actions of Torso-like, Nasrat/Polehole, and an unidentified factor(s). Processed C-terminal fragment of Trunk activates Torso. Torso in turn impedes the further diffusion of active Trunk molecules.

Download figure to PowerPoint

Cytoplasmic Mediators of Torso Signaling

After ligand binding, RTKs dimerize and undergo autophosphorylation on several tyrosine residues. This process serves two purposes. First, phosphorylation of certain tyrosine residues within the kinase domain is required for the full activation of its enzymatic activity. Second, and more importantly, autophosphorylation of tyrosine residues outside the kinase domain creates docking sites for signal transducers or adaptors that contain Src homology 2 (SH2) or phosphotyrosine binding (PTB) domains (reviewed by van der Geer et al., 1994; Pawson and Scott, 1997).

Adaptors.

Adaptor proteins play essential roles in mediating RTK signaling (reviewed by Pawson and Scott, 1997). The first adaptor identified in Drosophila was Downstream of receptor kinase (Drk), an SH2–SH3 adaptor protein and homolog of mammalian GRB2 and Caenorhabditis elegans Sem-5 (Clark et al., 1992; Lowenstein et al., 1992; Olivier et al., 1993; Simon et al., 1993). It was identified in a genetic screen for dosage-sensitive mediators of Sevenless signaling (Bonfini et al., 1992; Simon et al., 1993) and was later shown to be required for Torso signaling as well (Hou et al., 1995). Like GRB2, Drk binds to Son of Sevenless (Sos), a guanine nucleotide exchange factor for Ras, linking RTK activation to Ras signaling. Although Drk directly binds to Sevenless, its binding to other Drosophila RTKs is yet to be demonstrated (Raabe et al., 1996). In the case of Torso, Corkscrew (Csw) appears to mediate the association of Drk with the RTK (see below). Another adaptor identified in Drosophila in similar dosage-sensitive genetic screens is Daughter of Sevenless (DOS), a pleckstrin homology domain-containing protein and a substrate of Csw (Herbst et al., 1996; Raabe et al., 1996). Finally, a Drosophila SHC adaptor protein (DSHC), first identified in by sequence homology, has been shown to mediate signaling of a subset of Drosophila RTKs: dshc mutations affect signaling by Torso and EGFR but not Sevenless (Lai et al., 1995; Luschnig et al., 2000). These adaptors may all be necessary or may function redundantly in mediating Torso signaling.

Core signal transducers.

The Ras/Raf/MEK/ERK (Ras1/Draf/Dsor1/Rolled) signaling cassette appears to be the obligatory core signal transducer of all known Drosophila RTKs, including Torso, EGFR, Sevenless, and the FGFR homologs Heartless and Breathless (reviewed by Basler and Hafen, 1990; Perrimon, 1994). The Ras-ERK signaling cassette appears to be highly conserved evolutionarily throughout metazoans and plays a major role in the function of all known RTKs. Indeed, human oncogenic Ras and Raf-1 can function in Drosophila, causing phenotypes similar to those seen with RTK gain-of-function mutations (Brand and Perrimon, 1994; Casanova et al., 1994). In addition, Raf-1 can rescue a subset of Draf loss-of-function phenotypes and activated Draf can readily phosphorylate human MEK in vitro (W.X Li, unpublished observations).

Sos.

Sos is a key link from RTK to Ras. In a dosage-sensitive genetic screen, Sos, together with Ras1, was among the first molecules identified as performing crucial early steps in signaling by the Sevenless RTK (Simon et al., 1991). Genetic studies showed Sos is also essential in mediating EGFR and Torso signaling (Rogge et al., 1991; Lu et al., 1993a). Its function as a direct Ras activator was initially suggested by its sequence similarity to yeast CDC25, a guanine nucleotide exchange factor (GEF) bound to Ras (Jones et al., 1991; Rogge et al., 1991; Simon et al., 1991). Further genetic and biochemical studies demonstrated that Sos is indeed an activator of Ras1 and functions to convert GDP-bound Ras1 to the GTP-bound, or active, form in response to RTK activation (Bonfini et al., 1992; Olivier et al., 1993; Simon et al., 1993).

Ras.

There are two Ras homologs in the fly genome: Ras1 (also known as Ras85D due to its cytogenetic location) and Ras2 (Ras64B). Ras1 is 75% identical to the oncogenic mammalian Ki/Ha ras (p21ras) and is considered the only Ras homolog in the fly genome that functions in the Ras-ERK signaling cassette. Ras2, on the other hand, is homologous to mammalian R-ras and does not activate Draf (Fortini et al., 1992; Lu et al., 1993a). The function of Ras1 in mediating Drosophila RTK signaling was first suggested by results of genetic screens for dosage-sensitive modifiers of RTK variants (Simon et al., 1991; Doyle and Bishop, 1993) and was later supported by genetic and biochemical analyses (Lu et al., 1993a; Hou et al., 1995). Embryos devoid of Ras1 function, produced by the germline clone technique, almost completely abolish Torso signaling, as evidenced by phenotypes identical to those of torso loss-of-function mutations in the vast majority (approximately 80%) of these embryos. However, residual tll expression and posterior structures can be detected in the rest of approximately 20% of Ras1 mutant embryos.

Raf.

Draf is the sole Raf homolog in the fly genome and is 61% and 56% homologous to mammalian B-Raf and Raf-1 (C-Raf), respectively. Draf was first isolated as a lethal mutation, l(1)ph, whose maternal function is essential for the proper determination of terminal cell fates (Perrimon et al., 1985; Ambrosio et al., 1989a). Draf plays a central role in Torso signal transduction, because embryos lacking Draf gene activity have phenotypes identical to those of torso null embryos, resulting in the complete absence of posterior tll expression; weaker Draf alleles exhibit reduced levels of tll expression (Ambrosio et al., 1989b; Melnick et al., 1993; Sprenger et al., 1993).

MEK.

Genetic screens for mutations that suppress the lethality of a hypomorphic allele of Draf identified mutations in DownstreamofRaf1 (Dsor1; Tsuda et al., 1993; Lu et al., 1994). Dsor1 turned out to encode a Drosophila MEK1/2 homolog and the Dsor1 mutations isolated in the above screens were activating mutations that caused constitutive activity of the kinase (Tsuda et al., 1993; Lu et al., 1994). Loss of function Dsor1 mutations reverted the suppression, and these null mutations produced torso loss-of-function phenotypes, demonstrating that Dsor1 mediates Torso signaling and functions downstream of Draf (Tsuda et al., 1993; Lu et al., 1994).

ERK.

The Drosophila ERK homolog is encoded by the rolled (rl) gene (Biggs and Zipursky, 1992; Biggs et al., 1994; Brunner et al., 1994). Mutant alleles of rl have been isolated repeatedly in genetic screens for genes required for signaling by activated forms of Ras1, Draf, or Torso (Dickson et al., 1996; Karim et al., 1996; Li et al., 2000; Therrien et al., 2000; Li and Li, 2003). A dominant mutation in rolled that resulted in a partially constitutive activation of Rolled/ERK was isolated in a genetic screen for dominant mutations that enable Sevenless signaling in the absence of activation of the Sev RTK (Brunner et al., 1994). This mutation also causes embryonic defects similar to those of torsoGOF mutations and is suppressible by tll mutations (Brunner et al., 1994), suggesting that it is required for transducing Ras1 and Draf signals in general.

Cofactors and Scaffold Proteins

Genetic screens in Drosophila designed to isolate genes essential for Draf activation have identified several genes that encode important modulators such as scaffold proteins, chaperones, and cofactors. These include Kinase Suppressor of Ras (KSR), Connector enhancer of KSR (CNK), Hsp83, and 14-3-3, to name just a few (Therrien et al., 1995; Chang and Rubin, 1997; Kockel et al., 1997; Li et al., 1997; van der Straten et al., 1997; Therrien et al., 1998). The identification of these modulators underscores the complexity in the propagation of signals through the Ras-ERK signaling cassette, as well as the importance of protein–protein interactions (see reviews by Morrison, 2001; Dhillon and Kolch, 2002; Chong et al., 2003; Morrison and Davis, 2003).

HOW DOES TORSO TRANSDUCE SIGNALS?

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

A combination of genetic and biochemical studies have suggested that Torso signaling shares common features with signaling of other RTKs but also exhibits unique properties. As with all RTKs, Torso presumably dimerizes after binding to extracellular ligands. This presumption is consistent with the finding that all three torsoGOF mutations are due to point mutations in the extracellular or transmembrane domains, presumably causing ligand-independent dimerization (Sprenger and Nusslein-Volhard, 1992). Dimerization of the Torso receptor causes its autophosphorylation on multiple tyrosine residues. These phosphotyrosine residues increase its tyrosine kinase activity or serve as specific docking sites for proteins with SH2 domains.

Autophosphorylation on tyrosine residues appears to be an essential step in initiating the signal transduction cascade after RTK activation. When purified Torso protein is allowed to autophosphorylate in vitro, only tyrosine phosphorylation can be detected, and two tyrosine residues (Y630 and Y918) contribute to most of the detectable phosphorylation (Cleghon et al., 1996). When mutant Torso with phenylalanine substitutions of these residues were analyzed in vivo, it was found that pY630 plays a positive role, whereas pY918 plays a negative role in mediating Torso signals, suggesting that Torso activity is modulated by compensatory mechanisms (Cleghon et al., 1996). Phosphorylation of Y630 provides a docking site for the tyrosine phosphatase Csw, an SHP2 homolog (Perkins et al., 1992; Cleghon et al., 1996), and the negative site pY918 serves to recruit the Ras GTPase-activating protein (RasGAP), which negatively regulates Ras activity (Cleghon et al., 1996, 1998; Feldmann et al., 1999). Thus, the strength of the Torso signal is modulated by opposing actions of Csw and RasGAP, contributing to precise control of cell fate specification and pattern formation. Because many vertebrate RTKs bind to both RasGAP and Csw/SHP2, this modulation may be a conserved regulatory mechanism.

The SH2-containing phosphotyrosine phosphatase SHP2 in vertebrates can associate with several RTKs and plays a positive role in transducing RTK signals (Lechleider et al., 1993; Tang et al., 1995). However, the precise molecular mechanisms of Csw/SHP function in transducing RTK signaling remain unclear, as several substrates of Csw/SHP2 have been reported. These substrates include RTK itself (Cleghon et al., 1998; Agazie et al., 2003), the adaptor protein Gab/DOS (Herbst et al., 1996; Raabe et al., 1996), the C-terminal Src Kinase (Csk) regulator PAG/Cbp (Zhang et al., 2004), and a negative RTK regulator Sprouty (Spry; Hanafusa et al., 2004). Csw specifically dephosphorylates pY918 of Torso, a negative pY site capable of binding RasGAP (Cleghon et al., 1998). Similarly, dephosphorylation of Y992 of EGFR by SHP2 prevents the recruitment of RasGAP to the receptor, thus increasing the half-life of activated Ras (Agazie et al., 2003). In a separate report, it has been shown that SHP2 may dephosphorylate a conserved RTK inhibitor Spry, preventing its inhibitory association with Grb2 (Hanafusa et al., 2004). The phosphatase activity of SHP2, therefore, appears essential for its role as a positive mediator of RTK signaling. Deletion of the catalytic site in the phosphatase domain in Xenopus SHP2 results in a dominant-negative mutant (Tang et al., 1995). Similarly, expression of a catalytically dead Csw resulted in dominant-negative phenotypes during Drosophila embryogenesis (M. Melnick and L. Perkins, personal communication). Mutations in human SHP2 that cause Noonan syndrome are associated with significantly increased basal phosphatase activity (Fragale et al., 2004).

Aside from dephosphorylating a substrate, Csw/SHP2 may serve as an adaptor protein for recruiting downstream signaling molecules to the RTK. Indeed, it has been shown that after binding to Torso, Csw is phosphorylated by Torso on Y666, which subsequently can be bound by Drk, a Growth factor receptor bound-2 (Grb2) homolog (Cleghon et al., 1998). Ras activation by many RTKs is associated with the recruitment of a Grb2/Drk-Sos complex by means of an autophosphorylated pYXNX motif (Songyang et al., 1993). However, Torso lacks the consensus Drk-binding motif YXNX and does not appear to bind directly to Drk (Cleghon et al., 1996). Phosphorylation of CSW on Y666 effectively creates a Drk-binding site pYTNI, allowing recruitment of the Drk-SOS complex to Torso, which leads to Ras activation (Cleghon et al., 1998).

Because mutation of the major positive phosphotyrosine pY630 causes a moderate decrease but does not completely eliminate Torso signaling, efforts were made to identify additional minor pY sites. Four additional tyrosine residues (Y644, Y698, Y767, and Y772) were identified and analyzed by mutagenesis and transgenic analysis (Gayko et al., 1999). Taken together with analyses of the two major pY sites, it became clear that, upon activation, Torso is autophosphorylated on multiple tyrosine residues located in both the catalytic domain (activation loop) and noncatalytic regions (the kinase insert and C-terminal tail). Analysis of transgenic Torso molecules with tyrosine to phenylalanine substitutions demonstrates that the activation loop tyrosines (Y767 and Y772) are essential for Torso enzymatic activity (Gayko et al., 1999). The C-terminal tail tyrosine (Y918) negatively regulates Torso activity (Cleghon et al., 1996; Gayko et al., 1999). In the kinase insert region of Torso, any single mutation of the four tyrosines (Y630, Y644, Y656, and Y698) has little or no effect on Torso signaling, while simultaneously eliminating all four tyrosine residues (quadruple mutant) completely abolishes Torso signaling (Gayko et al., 1999). These results indicate that multiple tyrosine residues outside of the kinase domain play a synergistic role in mediating Torso signaling, and that Torso recruits downstream molecules by means of the pYs located in the kinase insert region. Downstream signaling molecules/adaptors other than Csw that bind to these pY residues are yet to be identified.

Although Csw is the only positive mediator so far identified that directly binds to Torso, null mutations in csw do not completely abolish Torso signaling (Perkins et al., 1996). Indeed, csw mutant embryos exhibit intact Filzkörper and significant expression levels of the target gene tll (Perkins et al., 1992, 1996). Thus, it is likely that other adaptor proteins or signaling mediators are required to transduce Torso signals. Studies of mammalian RTKs have demonstrated the involvement of several SH2-domain containing adaptors that are recruited to pYs after RTK activation. In addition to Csw/SHP2 and Drk/Grb2 mentioned above, such adaptors include SHC (Lai et al., 1995; Luschnig et al., 2000), NCK/DOCK (Garrity et al., 1996; Rao and Zipursky, 1998), Crk (reviewed by Feller, 2001), and Grb7 family proteins (reviewed by Han et al., 2001). Of these adaptors, only SHC has been analyzed and shown to play a role in Torso signaling.

SHC can be recruited to an activated RTK and lead to Ras activation (Rozakis-Adcock et al., 1992). SHC contains two pY-binding motifs, an N-terminal PTB domain and an SH2 domain in the C-terminus. Studies of mammalian SHC have demonstrated that SHC binds to the NPXpY motif of the EGF receptor by means of its PTB domain and is in turn phosphorylated by EGFR on a specific tyrosine, which provides a binding surface for the SH2 domain of Grb2 (Rozakis-Adcock et al., 1992). Although DSHC binds to EGFR by a similar mechanism, it does not contain a Grb2-binding motif and does not interact with Drk (Lai et al., 1995; van der Geer et al., 1995). Genetic analysis of the dshc gene has demonstrated that DSHC is required for signaling by Torso and EGFR but not by the Sevenless RTK (Luschnig et al., 2000). Significantly, double-mutant analysis demonstrates that three adaptors, DOS, DRK, and DSHC, function in parallel to transduce Torso signals. Thus, mutations in each of the adaptor genes only partially affect Torso signaling, but removal of any two of them simultaneously severely reduces Torso signaling, as assayed by posterior terminal structures and tll expression (Luschnig et al., 2000). Although direct interaction with Torso has not been demonstrated for DSHC, these results indicate an intriguing redundancy among adaptor proteins and suggest that RTKs may use different combinations of adaptors to regulate signaling output qualitatively and quantitatively.

Based on genetic and biochemical studies of Torso, the following model can be proposed for Torso signal transduction (Fig. 4). Upon ligand-induced dimerization, Torso becomes autophosphorylated on multiple tyrosine residues. Two phosphotyrosines located in the activation loop (Y767 and Y772) are essential for Torso enzymatic activity. Four phosphotyrosines in the kinase insert region and the C-terminal tail serve as docking sites for Csw and possibly DSHC and other adaptors or SH2-containing signaling molecules, which may include STAT92E. These molecules redundantly or synergistically recruit Sos to the membrane. Sos converts GDP-Ras1 to GTP-Ras1, leading to activation of the Ras1/Draf/Dsor1/Rolled signaling cassette. In addition, Torso is able to activate Draf by means of a Ras1-independent pathway, and Ras1 plays a role in activating an unknown Draf activator, in addition to binding to Draf. Thus, multiple signaling routes originating from Torso can converge on the Draf kinase, leading to the expression of downstream target genes (Fig. 4).

OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Studies of the mammalian PDGF receptors have shown that they are able to associate with a large number of SH2 domain-containing proteins upon ligand stimulation (reviewed by Heldin and Westermark, 1999; Tallquist and Kazlauskas, 2004). In addition to numerous adaptor proteins that lead to activation of the Ras-ERK signaling pathway, many SH2-containing enzymes that lead to activation of distinct intracellular signaling cascades are implicated. These enzymes include phosphatidylinositol 3′-kinase (PI3K), phospholipase C (PLC)-γ, STAT proteins, SHP2, GAP, and the Src family of tyrosine kinases (reviewed by Heldin and Westermark, 1999; Tallquist and Kazlauskas, 2004). Mouse knockin experiments replacing the endogenous PDGF receptors with molecules in which the intracellular domain has been swapped with another kinase or has specific tyrosine to phenylalanine substitutions (F series) have been carried out, and the results suggest that many of the downstream signaling pathways are essential for mediating the full range of biological functions of the PDGF receptors (reviewed by Tallquist and Kazlauskas, 2004). For instance, when the PI3K binding site was disrupted, severe developmental defects were observed for PDGFαR and partial defects for PDGFβR, suggesting that PDGFαR relies more on PI3K signaling than PDGFβR does (Heuchel et al., 1999; Tallquist et al., 2000; Klinghoffer et al., 2002).

Drosophila Torso can associate with RasGAP and a mammalian PLCγ protein in vitro by means of the same phosphotyrosine (pY918; Cleghon et al., 1996, 1998). Because mutating Y918 results in up-regulation of Torso signaling, similar to mutations in RasGAP (Cleghon et al., 1996, 1998), the significance of pY918 binding to PLCγ, which normally functions positively in PDGFR signaling, remains to be investigated. A Drosophila PLCγ homolog with two SH2 domains has been isolated (Emori et al., 1994). However, mutational analysis especially with regard to Torso signaling has not been carried out.

Binding of Torso to PI3K has not been demonstrated. PI3K homologs have been identified at four loci in Drosophila (Flybase). Mutational analysis did not reveal any interaction with torso, although domain swap experiments in mice indicated that the PDGFαR/Torso fusion protein could effectively signal through the Ras-ERK pathway but only weakly through the PI3K pathway, which may explain the partial rescue of PDGFαR null phenotypes (Hamilton et al., 2003). Thus, it appears that Torso does not signal through the PI3K pathway effectively, if at all.

Drosophila STAT, STAT92E, is activated by Torso and appears to mediate a subset of Torso signals (Li et al., 2002, 2003). Mutations of stat92E suppress torsoGOF phenotypes (Li et al., 2002). STAT92E activation, as revealed by antibodies specific for pSTAT92E, is detected in early embryos in a pattern consistent with that of Torso activation domains (Li et al., 2003). The association of Torso with STAT92E has been detected in embryo extracts or with purified proteins (Li et al., 2002; V. Cleghon, personal communication), although the pYs on Torso responsible for STAT92E association remain to be elucidated.

TURNING OFF TORSO SIGNALING

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Feedback Inhibition

Negative feedback by means of induction of inhibitors is a common strategy for turning off a signaling pathway. It is important that RTK signaling be tightly controlled during development. Uncontrolled RTK signaling has been linked to carcinogenesis in humans (reviewed by Porter and Vaillancourt, 1998; Robertson et al., 2000). In Drosophila, several inhibitors of the EGFR and FGFR pathways, such as Argos, Kekkon, and Sprouty (Spry), have been identified (reviewed by Freeman, 2000; Rebay, 2002). Among these inhibitors, Spry appears to be a general inhibitor of all RTK pathways, including Torso (Casci et al., 1999). First identified as a negative regulator of the FGFR during tracheal development in the fly embryo, Spry was shown to regulate other RTKs as well, including Torso, Sevenless, and EGFR (Hacohen et al., 1998; Casci et al., 1999). In contrast to an earlier proposal that Spry is extracellular, biochemical analysis has demonstrated that Spry is an intracellular membrane-bound protein and binds to two intracellular components of the Ras pathway: Drk and Gap1 (Casci et al., 1999). Indeed, human Spry proteins have been isolated and shown to function by means of a negative feedback mechanism downstream of RTK signaling: RTK signaling induces phosphorylation and translocation of Spry to the membrane, where it binds to the adaptor protein Grb2 and inhibits the recruitment of the Grb2-Sos complex, allowing for rapid negative feedback control (Hanafusa et al., 2002).

Receptor Internalization and Degradation

Signaling by growth factor receptors is modulated by endocytosis, during which activated and autophosphorylated receptors are directed by a ubiquitination-mediated process to lysosomes, where they are degraded (reviewed by Marmor and Yarden, 2004). Torso is similarly degraded after activation, and the Hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs) is required for regulating endosomal sorting of Torso and EGFR (Lloyd et al., 2002). Animals mutant for hrs exhibit impaired vesicle trafficking and fail to degrade active Torso and EGF receptors, leading to prolonged signaling and patterning defects (Lloyd et al., 2002).

Ligand Trapping

Torso is uniformly distributed along the syncytial membrane and is activated by diffusible ligands that are produced at the ends of the embryo. Studies of Torso signaling have suggested that the limited extent of Torso activation is due to binding of limited amount of ligand by the abundantly expressed Torso receptor thereby impeding ligand diffusion into the central region of the embryo, a phenomenon termed “ligand trapping” (Sprenger and Nusslein-Volhard, 1992; Casanova and Struhl, 1993). Indeed, in mutant embryos that fail to accumulate Torso protein at one or both poles, inappropriate activation of more centrally located Torso receptors occurs, leading to ectopic terminal structures in the central region. Thus, membrane receptors not only transduce signals upon activation by their ligands, but also spatially limit their activation by sequestering the ligands (Sprenger and Nusslein-Volhard, 1992; Casanova and Struhl, 1993).

BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

Cell Fate Specification Mediated by Transcription Factors

The Torso pathway specifies terminal cell fates by regulating target gene expression. The Torso target genes tll and hkb mediate cell fate specification by Torso in the posterior terminal region where they are expressed in overlapping domains, with tll more broadly expressed, reflecting a differential threshold requirement for Torso activity (Pignoni et al., 1990; Weigel et al., 1990). The posterior expression of tll and hkb is solely dependent on Torso, and the extent of their expression domains reflects the strength of Torso signaling. Indeed, tll is turned on in all cells in the posterior, from 0% to approximately 15% of the egg length (EL), except in the pole cells localized in the posterior-most tip. The pole cells are primordial germ cells that give rise to future germ cells. These two Torso target genes are not expressed in torso loss-of-function mutants; their expression domains are expanded in gain-of-function mutants. Analysis of gap genes normally expressed in the adjacent domains indicates that, in torso loss-of-function mutants, the cells normally fated to become telson are now respecified as abdominal cells (Klingler et al., 1988; Strecker et al., 1989). Thus, the number of subterminal (abdominal) cells expands in torso loss-of-function mutants at the expense of terminal cells. Conversely, in torso gain-of-function mutants, the tll and hkb expression domains expand toward the central region (see Fig. 1). This expansion abolishes specification of central cell fates in the embryo. Indeed, overexpression of tll alone is sufficient to repress the expression of central gap genes and to promote ectopic terminal cell fates (Steingrimsson et al., 1991). Thus, the Torso pathway, mediated by its target genes, is instrumental in cell fate determination in the posterior terminal domain of the embryo.

In the anterior region, however, Torso and Bicoid (Bcd), a homeodomain transcription factor and morphogen essential for development of anterior body segments (Driever and Nusslein-Volhard, 1988), play antagonistic as well as cooperative roles in regulating gene expression (Pignoni et al., 1992; reviewed by Schmidt-Ott, 2001; Ephrussi and St. Johnston, 2004). For instance, tll and hkb are regulated by both Torso and Bcd; in the absence of Torso signaling, the anterior tll expression domain expands (see Fig. 1), and this expansion is dependent on Bcd, suggesting that the activity of Bcd is normally inhibited by Torso in the anterior-most region (Pignoni et al., 1992). Indeed, Torso signaling down-regulates Bcd transcriptional activity in the anterior region by modifying Bcd or its binding partner (Bellaiche et al., 1996; Janody et al., 2001). This regulation requires components of the Torso signaling cascade but is independent of its target genes tll and hkb (Ronchi et al., 1993). Thus, although Torso is essential for head development, its target genes tll and hkb appear largely dispensable in this process, as head structures form in tll hkb double mutants (Strecker et al., 1986; Weigel et al., 1990). In addition to Bcd itself, several genes essential for head formation are coregulated by Torso and Bcd. These genes include hunchback (hb) and orthodenticle/ocelliless (otd/oc; Finkelstein and Perrimon, 1990). Thus, gene regulation during head formation is complex and involves multiple pathways.

Torso Signaling Derepresses Target Gene Expression

Studies of how the Torso pathway induces the expression of its target genes tll and hkb in the posterior pole have revealed that the main modus operandi is by relief of repression (reviewed by Furriols and Casanova, 2003). Analysis of the tll promoter has identified torso response elements, tor-RE, that appear to mediate binding of repressors that are present throughout the embryo, because mutating these putative repressor-binding sites causes derepression of a reporter gene in all regions of the embryo (Liaw et al., 1995). One ubiquitous repressor that is involved in repressing tll and hkb is Groucho (Gro; Paroush et al., 1997). Gro is an evolutionarily conserved corepressor that can be recruited by its DNA-binding partners to the promoters of a variety of genes, leading to their repression (reviewed by Courey and Jia, 2001). A DNA-binding protein that might recruit Gro to the promoters of tll and hkb is Capicua (Cic). Studies of mutant phenotypes and protein distribution of Cic in the presence or absence of Torso signaling indicate that it might be directly modulated and down-regulated by ERK, allowing derepression of tll and hkb expression (Jimenez et al., 2000). It should be noted that other repressors might also participate in the process of tll repression. Searches for proteins that bind to tll promoter sequences, including the tor-RE, have identified proteins such as GAGA, NTF-1, and Tramtrack, which might be components of repressor complexes involved in repression of tll and hkb (Liaw et al., 1995; Chen et al., 2002; Fig. 4).

What activates the Torso target genes tll and hkb? It has been postulated that, once the repression is relieved, activators ubiquitously present in the embryo are responsible for activating tll and hkb. These activators may or may not be regulated by Torso signaling. STAT92E appears to meet the criteria for such an activator, although loss of STAT92E only minimally affects tll expression in otherwise wild-type embryos (Li et al., 2002). It is possible that multiple activators are collectively responsible for the induction of tll and hkb. Further investigation is needed to clarify this issue.

Germ Cell Development

Of interest, Torso activation in the posterior region of the early embryo correlates temporally and spatially with the formation of primordial germ cells (PGCs), which are also known as “pole cells” arising at the posterior pole (reviewed by Wylie, 1999, 2000; Starz-Gaiano and Lehmann, 2001), raising the question of whether Torso plays a role in germ cell development. Recently, several groups reported interesting observations along this line. While one group has shown that Torso signaling is required for the initial division and migration of germ cells, other groups demonstrate that its activity has to be suppressed in early germ cells for establishing or maintaining transcriptional silencing (Li et al., 2003; Deshpande et al., 2004; Martinho et al., 2004).

Recent work in our lab has found that Torso activates both the Ras-ERK pathway and STAT92E in the terminal regions. Coactivation of these two branches of signaling components appears to be essential for the initial division and migration of primordial germ cells (Li et al., 2003). Embryos mutant for stat92E or Ras1 have fewer PGCs, and these cells migrate slowly, errantly, and fail to coalesce. Conversely, overactivation of these molecules causes supernumerary PGCs, their premature transit through the gut epithelium, and ectopic colonization (Li et al., 2003). Moreover, pole cells in torso mutant embryos appear immobile and, when isolated and put into medium, exhibit slow movements, suggesting cell-autonomous defects in cytoskeletal reorganization (Li et al., 2003). A requirement for an RTK in Drosophila PGC development is analogous to the situation in the mouse, in which the RTK c-Kit is required, suggesting a conserved molecular mechanism governing PGC behavior in flies and mammals.

Of interest, whereas STAT92E activation is prominently detected in pole cells, ERK phosphorylation is attenuated in pole cells, compared with neighboring somatic cells in the posterior region. Recently, it has been shown that, in Drosophila, the noncoding RNA encoded by polar granule component (pgc) is essential for silencing transcription in germ cells (Deshpande et al., 2004; Martinho et al., 2004). In the absence of pgc and loss of transcriptional silencing, tll is expressed in the pole cells, where it is normally repressed. These observations can be interpreted in two different ways. First, the Torso pathway is normally inhibited in pole cells. Support for this idea comes from the detection of an increased level of dpERK in pgc RNAi pole cells (Deshpande et al., 2004). Second, pgc represses transcription in pole cells independent of Torso signaling. This idea is supported by the demonstration that genes whose expression is independent of Torso signaling are also ectopically expressed in pgc mutant pole cells (Martinho et al., 2004). Moreover, ectopic expression of normally silent genes in pole cells can be induced by overexpression of Bcd, and their expression depends on appropriate activation of Torso signaling (Deshpande et al., 2004); thus, it is unlikely that Torso signaling is absent in pole cells.

Thus, there may be different requirements for Torso signaling for different aspects of germ cell development, or different downstream effectors may be required at different times. For instance, when the function of Torso signaling in inducing gene expression is inhibited, its effects on cell migration and division by means of inducing actin-cytoskeleton reorganization remain essential for the early stages of germ cell development. The precise mechanisms of Torso signaling in Drosophila germ cell development require further investigation.

PERSPECTIVES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

An important unresolved question regarding signal transduction by the Torso RTK through the Ras-MAPK cassette is how this highly conserved signaling cassette specifically regulates gene expression. Great efforts have been made to resolve the issue of how an activated MAPK results in expression of different genes in different contexts. Differential gene expression that depends on the strength and/or duration of MAPK activation has been observed (Marshall, 1995; Greenwood and Struhl, 1997; Li and Perrimon, 1997; Ghiglione et al., 1999). However, it has become clear that strength or duration of MAPK activation alone cannot account for the great variety of genes that are precisely controlled in different cells and at different stages of an animal's life (reviewed by Tan and Kim, 1999; Simon, 2000). Rather, integration of multiple signaling pathways and of tissue-specific transcription factors is required for determining particular cell fates and behaviors during Drosophila development. For example, the generation of a subset of muscle and cardiac progenitors during Drosophila embryogenesis requires integration of RTK, TGFβ, and Wnt signaling pathways (Flores et al., 2000; Halfon et al., 2000; Xu et al., 2000). In contrast, in the early Drosophila embryo, a gain-of-function mutant of Torso is capable of engaging the STAT and Dpp pathways in regulating target gene expression (Li et al., 2002; Li and Li, 2003).

Another important question is why RTK overactivation, which has generally been assumed to be quantitative in nature, should lead to qualitative changes in cellular properties. RTK overactivation, caused either by oversupply of ligands or by mutations that result in ligand-independent constitutive activation, has been linked to many cancers and other human diseases (reviewed by Porter and Vaillancourt, 1998; Robertson et al., 2000). It has been proposed that a constitutively activate RTK hyperactivates a canonical downstream signal transduction pathway, such as the Ras-MAPK signaling cassette, and that the qualitative changes in gene expression are determined by the variation in signaling duration and/or intensity (Marshall, 1995; Greenwood and Struhl, 1997; Li and Perrimon, 1997; Sewing et al., 1997; Woods et al., 1997; Ghiglione et al., 1999). Alternatively, high RTK signaling intensity could lead to activation of alternative signaling pathways that are not engaged under normal physiological conditions. Studies of Torso signaling have demonstrated that STAT92E is differentially required by wild-type and gain-of-function Torso molecules. While wild-type Torso can function nearly normally without STAT92E, gain-of-function mutant Torso is not able to exert its full biological effect without the participation of STAT92E (Li et al., 2002). Thus, specific responses to RTK signaling are determined or influenced by multiple pathways or signaling networks.

Signaling networks, resulting from the intertwining of multiple signaling pathways, have increasingly emerged as the rule rather than the exception in controlling complex biological processes such as cell proliferation and differentiation, during normal development or tumor progression. Specific biological responses often appear to result from such interactions. Moreover, signaling pathways may engage and modulate the activities of one another at multiple levels, such that perturbation of one pathway will have an impact on intracellular signaling networks as a whole. Thus, the precise effects of a signaling pathway on a variety of specific developmental programs depend on combinatorial inputs from signaling networks.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES

I apologize to those whose work was not cited. I thank Drs. Louise Silver-Morse, Dirk Bohmann, and the anonymous reviewers for helpful comments on the manuscript; all members of the Li lab for helpful discussions; and the National Institutes of Health for grant support.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. IDENTIFICATION OF TORSO
  5. TORSO IS A TYPE III RTK WITH A SPLIT TYROSINE KINASE DOMAIN
  6. GENETIC STUDIES OF TORSO SIGNALING
  7. COMPONENTS OF THE TORSO PATHWAY
  8. HOW DOES TORSO TRANSDUCE SIGNALS?
  9. OTHER INTRACELLULAR PATHWAYS MEDIATING TORSO SIGNALING
  10. TURNING OFF TORSO SIGNALING
  11. BIOLOGICAL FUNCTIONS OF THE TORSO PATHWAY
  12. PERSPECTIVES
  13. Acknowledgements
  14. REFERENCES