Drosophila-Raf Acts to Elaborate Dorsoventral Pattern in the Ectoderm of Developing Embryos

Corresponding author: Linda Ambrosio, 3264 Molecular Biology Bldg., Iowa State University, Ames, IA 50011., lima{at}iastate.edu (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In the early Drosophila embryo the activity of the EGF-receptor (Egfr) is required to instruct cells to adopt a ventral neuroectodermal fate. Using a gain-of-function mutation we showed that D-raf acts to transmit this and other late-acting embryonic Egfr signals. A novel role for D-raf was also identified in lateral cell development using partial loss-of-function D-raf mutations. Thus, we provide evidence that zygotic D-raf acts to specify cell fates in two distinct pathways that generate dorsoventral pattern within the ectoderm. These functional requirements for D-raf activity occur subsequent to its maternal role in organizing the anterioposterior axis. The consequences of eliminating key D-raf regulatory domains and specific serine residues in the transmission of Egfr and lateral epidermal signals were also addressed here.

AS a member of the Ras/mitogen-activated protein kinase (MAPK) signaling cassette, the D-raf serine/threonine kinase plays an essential role in numerous developmental pathways in Drosophila. In the early embryo, D-raf proteins transmit a cell determination signal localized at the embryonic poles that depends on the activation of the Torso (Tor) receptor tyrosine kinase (RTK). Both D-raf and tor mRNAs are maternally synthesized and translated into proteins that specify terminal cellular fates within 3 hr after egg fertilization (reviewed by LU et al. 1993 ; DUFFY and PERRIMON 1994 ). Subsequently, along the ventral midline, a second RTK-generated signaling pathway, under the control of the Drosophila EGF receptor (Egfr), results in the specification of ventral ectodermal cell fates (reviewed by SCHWEITZER and SHILO 1997 ).

It has been anticipated, but not previously shown, that D-raf acts as an essential component for transmission of the Egfr-generated signal in ventral ectodermal cells of the embryo. Other embryonic Egfr pathways in which D-raf may function include those required for germband retraction, the development and viability of midline glial cells, and the secretion of ventral cuticle and denticles (CLIFFORD and SCHUPBACH 1992 ; RAZ and SHILO 1992 , RAZ and SHILO 1993 ; SCHOLZ et al. 1997 ). At later stages of the life cycle, it has been shown that D-raf functions downstream of Egfr in wing vein differentiation (DIAZ-BENJUMEA and HAFEN 1994 ), specification of photoreceptor cells (GREENWOOD and STRUHL 1999 ; HALFAR et al. 2001 ; YANG and BAKER 2001 ), and dorsoventral patterning of follicle cells in the ovary (BRAND and PERRIMON 1993 ).

It is known that the regulation of the Raf kinase family is complex, with a variety of proteins acting to control subcellular localization, conformational state, and ultimately the kinase activity of Raf molecules (for review and references within see KOLCH 2000 ). Raf family members share homology in three domains, conserved region 1 (CR1) that binds Ras, CR2 that binds 14-3-3, and CR3, the kinase region that also contains a 14-3-3 binding site. In mammals, Ras acts positively to position Raf-1 at the membrane, where it is subsequently activated (LEEVERS et al. 1994 ; STOKOE et al. 1994 ). In contrast, the phospho-binding protein, 14-3-3, has been implicated in both the positive and negative regulation of Raf-1 functioning as a "scaffold" that stabilizes both the active and inactive forms of the kinase (MICHAUD et al. 1995 ; MUSLIN et al. 1996 ).

In Drosophila, D-Ras acts to positively regulate the activity of D-raf in Tor, Egfr, and Sevenless signaling pathways (reviewed by DAUM et al. 1994 ; DUFFY and PERRIMON 1994 ). In addition, two isoforms of 14-3-3 ( and ) have been shown to enhance D-Ras signaling and are thought to operate through interaction with D-raf proteins (CHANG and RUBIN 1997 ; KOCKEL et al. 1997 ; LI et al. 1997 ). Since 14-3-3 has also been implicated in the negative regulation of D-raf (ROMMEL et al. 1997 ) it is likely that the 14-3-3 proteins can, in a manner equivalent to that found for Raf-1, bind to D-raf S388 and S743 to regulate activity. However, it is clear from genetic studies that other factors, including KSR (THERRIEN et al. 1995 , THERRIEN et al. 1996 ), PP2A (WASSARMAN et al. 1996 ), and CNK (THERRIEN et al. 1998 , THERRIEN et al. 1999 ) also serve to modulate the activity of D-raf (for review see STERNBERG and ALBEROLA-ILA 1998 ).

Signal transduction pathways mediate cellular responses such as growth and differentiation by eliciting a signal across the plasma membrane. The mechanisms that regulate Raf family members in different organisms in these signaling pathways during development remain unclear. Since the family of Raf proteins has been highly conserved during evolution, studies involving D-raf will have broad implications and better define how developmental cell fate choices are generated within the animal kingdom. In Caenorhabditis elegans the lin 45 raf kinase is required for vulva development (HAN et al. 1993 ). In Xenopus laevis Raf proteins have been shown to play a role in the induction of the embryonic mesoderm (MACNICOL et al. 1993 , MACNICOL et al. 1995 ; XU et al. 1996 ). Raf-1-deficient mice exhibit growth retardation and die at midgestation with anomalies in the placenta and in the fetal liver (NAUMANN et al. 1997 ; MIKULA et al. 2001 ), while B-Raf-deficient mice die in utero displaying defects in endothelial cell differentiation and survival (WOJNOWSKI et al. 1997 ).

Here we access the ability of D-raf molecules to act in signaling pathways activated subsequent to the establishment of terminal cell fates by the Tor RTK. We found that mutant forms of D-raf expressed using a heat-shock-driven transgene were variably stable and showed differences in the rescue of dorsoventral cuticle defects caused by loss of D-raf maternal and zygotic function. These data provide evidence for the hypothesis that D-raf plays a role in two distinct signaling pathways that direct maturation of the embryonic ectoderm, the Egfr pathway for ventral cell determination and a second novel pathway required for the specification of lateral cell fates.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Stocks, production of D-raf germline mosaics, and transgenic D-raf lines:
Egfr embryos were collected from parents heterozygous for the top^1P02allele (CLIFFORD and SCHUPBACH 1992 ) and the CyO, ß1 balancer chromosome that carried a lacZ gene under the control of the fushi-tarazu (ftz) promoter. To distinguish between wild-type, heterozygous, and homozygous Egfr embryos, the genotypes of the embryos were determined by following the expression pattern of the lacZ gene (see below). Embryos lacking the lacZ marker were homozygous lethal for the top^1P02 allele. To generate mosaic females that were homozygous for the D-raf^11-29 protein null allele (AMBROSIO et al. 1989A ; MELNICK et al. 1993 ) germline clones were made using the "FLP-DFS" technique of CHOU and PERRIMON 1992 . Both D-raf "null" (D-raf^-/Y) and paternally rescued D-raf "torso" (D-raf^-/+) progeny, derived from eggs that lacked maternal D-raf protein, suffered embryonic lethality (PERRIMON et al. 1985 ; AMBROSIO et al. 1989B ). To distinguish between these two embryo classes, females with D-raf germline clones were crossed with males of the genotype yw / Y; P[w⁺ ftz-ß-gal G2] / P[w⁺ ftz-ß-gal G2]. Embryonic genotypes were determined by following the expression pattern of the lacZ gene (see below). Embryos without the lacZ marker were referred to as D-raf null embryos because they lacked both maternal and zygotic D-raf protein and produced a small cuticular patch at the end of embryonic development. Their siblings, which expressed the lacZ gene, were referred to as paternally rescued D-raf torso embryos because they lacked only maternal D-raf gene activity and showed a terminal class phenotype in cuticular preparations.

Transgenic lines with modified D-raf genes were generated using standard P-element transformation (SPRADLING and RUBIN 1982 ). Modified D-raf cDNAs as described in BAEK et al. 1996 were inserted into the polylinker site of the pCaSpeR-hs vector (THUMMEL and PIRROTTA 1992 ) using EcoRI and XbaI. Females with D-raf^11-29 germline clones were crossed to yw / Y; P[w⁺ D-raf^modified] / P[w⁺ D-raf^modified] transgenic males. In the case of the D-raf^S743A gene, transgenic males were heterozygous for the D-raf^S743A insertion on the TM2 balancer chromosome. Thus, only one-half of the D-raf embryos inherited the D-raf^S743A gene. This was considered in Western blots to determine D-raf^S743A protein concentration and in the phenotypic analysis of embryonic expression and cuticular patterns.

Rescue of Egfr embryos by central mRNA microinjection:
D-raf^WT (1 µg/µl) and D-raf^tor4021 (0.5 µg/µl) mRNA injections were as described in BAEK and AMBROSIO 1994 and BAEK et al. 1996 with the following modifications: embryos from top^1P02/+ parents were collected for 30 min, aged 20 min (Nuclear Cycle 10), and then processed for central injection. We used chi-square analysis (² = ) and one degree of freedom and found that deviation from the expected genetic ratio of 3:1 was significant for those embryos injected with D-raf^tor4021 mRNA (P value of 0.90). As a control for the injection procedure, embryos from heterozygous Egfr parents were injected with D-raf^WT mRNA. Deviation from the expected 3:1 ratio was not statistically significant in this case, with a P value of 0.90.

Phenotypic characterization of Egfr, D-raf, and transgenic D-raf embryos:
In situ hybridizations were performed as described in TAUTZ and PFEIFLE 1989 . Antisense digoxigenin probes were prepared from plasmids containing the otd (FINKELSTEIN and PERRIMON 1990 ), rho (BIER et al. 1990 ), or dpp (IRISH and GELBART 1987 ) cDNAs. For double-labeling experiments to distinguish between wild-type, heterozygous, and homozygous Egfr embryos or D-raf torso and null embryos, antibodies were used at 1:1000 and 1:500 for the anti-ß-gal primary (Sigma, St. Louis) and goat-anti-mouse secondary (Jackson ImmunoResearch, West Grove, PA) antibodies, respectively. Immunocytochemistry was performed as described in PERKINS et al. 1996 . For cuticular analysis, unhatched embryos were prepared according to ASHBURNER 1989 . Embryos were photographed with a Zeiss Axioscope microscope using phase contrast or Nomarski optics.

In gene expression studies to measure the distance between dpp stripes, embryos were placed in 7% gelatin:63% glycerol mounting medium, were rolled onto their dorsal side, and then were covered with a coverslip. Measurements were performed on a Macintosh IIci computer using the public National Institutes of Health (NIH) Image program. Chi-square analysis (1 d.f. and a 95% confidence interval) showed that the separation between lateral dpp stripes was significantly different in homozygous Egfr and D-raf null embryos when compared with wild-type or D-raf torso embryos.

To assay for phenotypic rescue in D-raf null embryos by paternally inherited D-raf^modified transgenes, embryos were collected for 1 hr, allowed to develop at 25° for 2.75 hr, and then heat-shocked at 37° for 0.5 hr. Embryos were transferred to 25° to continue development.

Western analysis:
Western analysis was performed as described in RADKE et al. 1997 with each sample containing 100 embryos. Eggs were collected over a 1-hr period from females with D-raf^11-29 germline clones after mating with yw / Y; P[w⁺ D-raf^modified] / P[w⁺ D-raf^modified] males. The embryos produced by these females were devoid of maternal D-raf protein. For non-heat-shocked samples, embryos were aged at 25° for 4.5 hr and then collected for processing, representing the 5-hr time point. For heat-shocked samples, embryos were aged for 3 hr and then heat-shocked for 30 min at 37°. These embryos were allowed to recover for 1 or 6 hr before processing, representing the 5- and 10-hr samples, respectively. Since expression of each D-raf^modified gene was regulated by the hsp70 promoter, the accumulation of 90-kD or truncated D-raf proteins was observed only after heat shock, with the exception of those lines with leaky transgenic expression. Densitomeric analysis was performed using the NIH Image program (developed at the National Institutes of Health and available from the internet by anonymous FTP from zippy.nimh.nih.gov or on floppy disk from the National Technical Information Service, Springfield, VA, part no. PB95-500195GEI) with individual molecular weight contributions of the truncated D-raf proteins considered in this analysis.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

D-raf acts downstream of Egfr for the establishment of ventral ectodermal cell fates:
In the Drosophila embryo, Egfr activity is required to instruct a field of cells that lie on either side of the ventral furrow to adopt a ventral ectodermal fate. It is from this neuroectodermal cell population that the ventral nervous system and epidermis arise. At later times, Egfr functions in germband retraction and cuticle formation. Embryos that develop without Egfr activity fail to form ventral cuticular structures and show the "faint little ball" phenotype (Fig 1B). We used a constitutively active form of the D-raf protein, D-raf^tor4021, to bypass the requirements for Egfr function in embryos that lacked Egfr gene activity. For the generation of hyperactive D-raf^tor4021-proteins, the extracellular and transmembrane domains of the torso RTK gene were fused to the D-raf kinase domain. Chimera D-raf^tor4021proteins were shown to act independently of sevenless RTK gene function in developing photoreceptor cells and exhibited gain-of-function effects in the Tor signaling pathway (DICKSON and HAFEN 1994 ; BAEK et al. 1996 ).

View larger version (123K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Cuticular preparations of Egfr and D-raf mutant embryos. (A) A wild-type embryo with inverted head skeleton (h), posterior filzkörper (f), and ventral abdominal denticle bands. (B) View of an Egfr homozygote devoid of denticle setae and lacking anterior, posterior, and ventral cuticular structures. (C) Partial rescue of an Egfr homozygote after central injection with D-raftor4021 mRNA. There was some restoration of cuticular structures, including ventral denticles (arrow). For D-raf embryos that developed without maternal D-raf activity two phenotypes were observed: (D) A D-raf torso embryo with a truncated head skeleton (h), seven abdominal denticle bands, and missing tail structures and (E) a D-raf null embryo that produced a small patch of cuticle with few distinguishing features. Expression of transgenic D-raf proteins often resulted in partial rescue of the D-raf null phenotype with embryos showing either the "imperfect torso" or "null with denticles" phenotype. (F) A representative of the "imperfect torso" embryonic class with robust cuticle and denticle bands approximately two-thirds the width of a wild-type band. (G) A representative of the "null with denticles" embryonic class has faint cuticle and denticle belts approximately one-third the width of a wild-type band. Typically each denticle band consisted of four or five rows of shortened setae similar to those setae that lie at the lateral edges of denticle belts from wild-type embryos indicative of a deletion in the central most pattern with a corresponding expansion of more ventral lateral elements. High-power-magnification views of denticle belts from (H) wild-type, (I) "imperfect torso," and (J) "null with denticles" embryos.

We tested whether this activated D-raf protein would act independently of Egfr to rescue the embryonic lethality associated with homozygous mutations in the Egfr gene (Table 1). In the case of our noninjected control, 25% of the embryos derived from heterozygous Egfr parents (Egfr^-/+) failed to hatch, showed the faint little ball phenotype, and were homozygous for the Egfr mutation. We used D-raf^WT mRNA as a control for the injection procedure and found that after injection 27% of the embryos from heterozygous Egfr parents failed to hatch. These embryos showed the Egfr mutant phenotype at 24 hr. When D-raf^tor4021 mRNA was injected into the central region of embryos collected from heterozygous Egfr parents, all aspects of defective Egfr signaling were rescued for some of the mutant Egfr embryos. Of the 258 embryos that received injection, 217 (84%) hatched out of their egg cases as larvae, while 41 (16%) remained within their eggshells. Thus, we observed an increase in embryonic hatching and suppression of Egfr-induced lethality after injection of D-raf^tor4021 mRNA. We calculated and found a statistically significant difference in hatching rate for embryos that had received D-raf^tor4021 mRNAs when compared to those that did not (see MATERIALS AND METHODS). We also found partial rescue of the Egfr phenotype in unhatched embryos that had received D-raf^tor4021 mRNAs with ventral cuticular structures observed (Fig 1C). We concluded that constitutively active D-raf^tor4021 molecules could bypass the requirement for Egfr activity in the embryo and direct cells of the embryonic ectoderm to adopt a ventral fate. These results showed that D-raf participates downstream of Egfr in developing embryos.

Specification of ectodermal cell fates in D-raf null embryos:
Once we had found that an activated form of the D-raf protein could suppress the effects of a loss-of-function Egfr allele, we reasoned that embryos lacking maternal and zygotic D-raf activity would exhibit an Egfr-like phenotype. We expected that these embryos would also show defects associated with the loss of maternal D-raf function in Tor signaling. To determine whether the identities of cells in the ventral ectoderm were dependent on D-raf activity, marker gene expression patterns and cuticles produced by D-raf embryos were compared to those of wild-type and Egfr embryos. To generate these D-raf embryos, mosaic D-raf females were produced whose eggs lacked maternal D-raf proteins (see MATERIALS AND METHODS). Once fertilized, these eggs gave rise to two classes of embryos (PERRIMON et al. 1985 ; AMBROSIO et al. 1989B ): the first class was composed of the paternally rescued D-raf torso embryos (D-raf^-/+) that had inherited a wild-type D-raf gene from their fathers, were defective in Tor RTK signaling and were missing head and tail structures at 24 hr (Fig 1D). These D-raf torso embryos lacked maternal but not zygotic D-raf activity. The second phenotypic class was composed of the D-raf null embryos (D-raf^-/Y) whose exoskeletons consisted of what appeared to be a small patch of dorsal cuticle (Fig 1E). These embryos lacked maternal and zygotic D-raf activity throughout development. We anticipated that this D-raf null embryonic class would exhibit the phenotypic characteristics consistent with defective Egfr signaling, a consequence of defective D-raf protein activity.

First, to determine whether the establishment of ventral cell identity by the maternal dorsal gene system occurred normally in D-raf embryos we assayed the accumulation of rhomboid (rho) mRNAs between 4 and 6 hr (stages 9–12) of development (BIER et al. 1990 ). As visualized by in situ hybridization, a column of cells 2–3 wide on either side of the ventral midline showed the accumulation of rho mRNAs (data not shown). This temporal and spatial pattern of rho expression was observed in all embryos in our D-raf collections with each embryo a member of either the D-raf torso or null class. An equivalent rho expression pattern was observed in wild-type and Egfr embryos. Thus, the initial step in the establishment of ventral cell identity, by dorsal and other maternal genes that act to define the dorsoventral embryonic axis, was not perturbed when these events took place in the absence of maternal or zygotic D-raf activity.

To determine whether EGR-receptor signaling occurred normally in D-raf embryos, expression of the orthodenticle (otd) gene was monitored (RAZ and SHILO 1993 ). In wild-type control embryos, at 6 hr (stage 11) otd mRNAs accumulated in cells adjacent to the ventral midline and in the head (Fig 2A). In embryos lacking Egfr activity, otd expression occurred only in those cells within the embryonic head (Fig 2B). In D-raf embryo collections, we observed two patterns of embryo staining with approximately one-half of the embryos showing otd expression in cells along the ventral midline and in the head (Fig 2C). For the remaining D-raf embryos, the accumulation of otd mRNAs was observed only in the head, similar to Egfr embryos (Fig 2D).

View larger version (156K): [in this window] [in a new window] [Download PPT slide]   Figure 2. The accumulation of otd mRNAs at 6 hr and dpp mRNAs at 10 hr in Egfr and D-raf embryos. (A) otd mRNAs accumulated in two head placodes and a ventral midline stripe (arrow) in wild-type and heterozygous Egfr embryos. (B) Homozygous Egfr embryos lacked expression of otd mRNA along the ventral midline. (C) D-raf torso embryos showed head and ventral midline accumulation of otd mRNAs and were twisted. (D) Twisted D-raf null embryos lacked ventral otd stripe expression. (E) At 10 hr the accumulation of dpp mRNA was in two lateral stripes for wild-type and heterozygous Egfr embryos. (F) Homozygous Egfr embryos showed a reduced distance between lateral dpp stripes. (G) D-raf torso embryos had a wild-type pattern of dpp mRNA accumulation. (H) D-raf null embryos showed little separation between lateral dpp stripes and were twisted.

To distinguish between torso and null embryos in our D-raf collections, we used a ftz-ß-gal marker gene located on the paternal X chromosome. Males with the ftz-ß-gal gene were allowed to fertilize eggs from mosaic females that lacked D-raf activity (data not shown). In this double-labeling experiment, embryos that showed a ftz pattern of ß-gal expression were assigned to the D-raf torso class. These embryos also displayed a wild-type pattern of otd expression. In those D-raf null embryos lacking ß-gal expression, otd mRNAs were detected only in cells of the head, similar to Egfr embryos.

As was shown previously, between 4 and 7 hr (stages 9–11) of development wild-type and Egfr embryos accumulated decapentaplegic (dpp) mRNAs in cells that formed two lateral stripes, when embryos were viewed ventrally (CLIFFORD and SCHUPBACH 1992 ; RAZ and SHILO 1993 ). We observed a similar pattern of dpp mRNA accumulation in D-raf mutant embryos at this developmental stage (data not shown). However, CLIFFORD and SCHUPBACH 1992 showed that the ventral distance between dpp stripes becomes smaller in Egfr embryos as they develop. We recorded and compared the distance between lateral dpp stripes in wild-type, Egfr, and D-raf embryos at 10 hr (stage 13) of development (Table 2). For wild-type embryos the average stripe distance was 0.111 units. In our collection of Egfr embryos, 75% showed an average dpp lateral stripe distance of 0.118 units, similar to wild type (Fig 2E). This phenotypic class contained embryos that were heterozygous mutant (Egfr^-/+) or wild type with respect to the Egfr gene. In the remaining 25% of the embryos the average dpp stripe distance was reduced to 0.075 units as anticipated for homozygous mutant Egfr embryos (Fig 2F).

Two phenotypic classes of D-raf embryos were also distinguished on the basis of a statistically relevant difference in dpp stripe distance (Fig 2G and Fig H). In approximately one-half of the embryos the average dpp lateral stripe distance was 0.120 units, with the remaining embryos showing an average separation of 0.064 units (Table 2). We speculated that this second phenotypic class contained the D-raf null embryos (see MATERIALS AND METHODS). To test this idea, the marker ftz-ß-gal X chromosome was again employed in a double-labeling experiment to distinguish between D-raf torso and null embryos (data not shown). As anticipated, it was the male D-raf null embryonic class that showed the decrease in distance between lateral dpp stripes, indicative of a loss in ventral cell fates.

On the basis of our analysis of rho, otd, and dpp gene expression patterns in D-raf null embryos, we concluded that ventral ectoderm cells were specified incorrectly in the absence of D-raf activity. This loss resulted in the production of a mature D-raf null exoskeleton that was severely reduced in size and devoid of ventral structures, consistent with the Egfr embryonic phenotype. However, when we compared the distance between lateral dpp stripes in Egfr (0.075 units) and D-raf null (0.064 units) embryos, it was smaller in D-raf null embryos. In addition, after cursory inspection, the size of the exoskeleton patch produced by D-raf null embryos appeared smaller than that from Egfr embryos (Fig 1B and Fig E). We speculated that these differences could be biologically significant and expanded our analysis to address this potentially interesting finding.

The role of D-raf in the embryonic ectoderm:
To better understand the role that D-raf plays in the ectoderm and to access its regulation in various developmental pathways we utilized partial loss-of-function alleles of D-raf generated in vitro (BAEK et al. 1996 ). D-raf shares homology with family members in CR1 that contains D-ras binding motifs; CR2, a region rich in serine and threonine residues; and the CR3 kinase domain (Fig 3A). CR1 is thought to exhibit positive control in the regulation of the D-raf protein via its interaction with D-Ras, while CR2 appears to be involved in the negative regulation of the molecule (HOU et al. 1995 ; BAEK et al. 1996 ). We tested whether conserved subdomains, CR1 and CR2, or putative phosphorylation sites, serine 388 or 743, were essential for the activity of D-raf in the embryo or involved in its positive or negative regulation. These modifications of D-raf often resulted in decreased D-raf activity. Thus, by expressing partial loss-of-function D-raf alleles in D-raf null embryos we were successful in deciphering the role D-raf plays in developing embryos.

View larger version (24K): [in this window] [in a new window] [Download PPT slide]   Figure 3. A linear representation of D-raf proteins expressed in D-raf embryos and their accumulation as visualized by Western analysis. (A) The wild-type D-raf protein shows three regions of homology with Raf family members: conserved region 1 (CR1), CR2, and CR3. D-rafK497M proteins contained the amino acid substitution, lysine to methionine, at residue 497. Amino acid substitutions to alanine were also generated at serine 388 (D-rafS388A) and serine 743 (D-rafS743A). The D-raf315 protein was missing the first 315 amino acids of D-raf, while the D-raf445 protein lacked residues 1–445. (B) Accumulation of wild-type and D-rafmodified protein in embryos at 5 and 10 hr of development, with (+) or without (-) heat shock. The hsp 70 promoter was used to regulate D-rafmodified gene expression and controlled by heat shock. The first two sets of lanes show the accumulation of D-raf proteins from wild-type (WT) and D-raf11-29 embryos that lacked maternal D-raf proteins. In D-raf11-29 embryo collections only embryos of the D-raf torso class produce zygotic D-raf protein. The accumulation of this zygotic D-raf protein was very low with almost none detected here. For the D-raf transgenic lanes with embryos derived from homozygous D-raf11-29 germlines the accumulation of 90-kD D-rafWT, 60-kD D-raf445, and 38-kD D-raf315 proteins was detected. (C) Accumulation of D-raf proteins with amino acid substitutions at 5 and 10 hr of embryonic development with (+) and without (-) heat shock. High levels of these 90-kD D-raf proteins accumulate at 5 hr after heat induction.

Using a structure-function strategy, we generated several modified forms of the D-raf protein (Fig 3A). The D-raf^WT and D-raf^K497M genes were constructed as positive and negative controls, respectively, with the D-raf^WT allele a full-length copy of a D-raf cDNA (SPRENGER et al. 1993 ). D-raf^K497M lysine 497, which was shown to be critical for D-raf protein kinase activity and likely involved in ATP binding, was replaced with a methionine (SPRENGER et al. 1993 ; BAEK et al. 1996 ). The N-terminal and CR1 deletion mutation, D-raf³¹⁵, was likely to show a partial loss-of-function in D-raf null embryos. For the D-raf⁴⁴⁵ mutation both positive (CR1) and negative (CR2) control elements were lost, and we predicted that this form of D-raf would act in a manner similar to wild type or, on the basis of its structural similarity to oncogenic forms of Raf-1, show a gain-of-function effect in the embryo. Of the five phosphorylation sites identified for the human Raf-1 kinase, two were conserved in the D-raf protein (MORRISON et al. 1993 ). BAEK et al. 1996 generated serine to alanine substitutions at these sites and showed that S388 (CR2) played a negative role while S743 (CR3) was involved in the positive control of D-raf in the Tor pathway. We predicted that the D-raf^S388A and D-raf^S743A proteins would show similar phenotypic consequences for developing cells in the embryo.

Using P-element-mediated transformation, we generated Drosophila lines that contained an insertion of the D-raf^WT, D-raf^K497M, D-raf³¹⁵, D-raf⁴⁴⁵, D-raf^S388A, or D-raf^S743A gene on either the second or third chromosome. Each of these modified D-raf genes were paternally introduced into D-raf embryos lacking maternal D-raf protein (see MATERIALS AND METHODS). We also monitored the level and stability of D-raf proteins produced by expression of each paternally inherited D-raf^modified gene. In this assay 100 embryos were collected for each sample and processed for Western analysis (Fig 3B and Fig C). Since the expression of each D-raf^modified gene was under the control of the hsp70 promoter, samples were processed from non-heat-shocked or heat-shocked embryos at 5 and 10 hr of development. We found that these D-raf^modified proteins were variably stable and in D-raf null embryos showed differences in the rescue of dorsoventral cuticular defects caused by the loss of D-raf maternal and zygotic function. We organized our results on the basis of degree of phenotypic rescue that was observed in D-raf null embryos with the activity of D-raf^WT > D-raf^S388A > D-raf⁴⁴⁵ > D-raf^S743A > D-raf³¹⁵ > D-raf^K497M and these findings are presented below.

Rescue of the D-raf null phenotype by paternally inherited D-raf^WT proteins: We assayed the accumulation of D-raf protein in D-raf embryos that had inherited the D-raf^WT gene. For these embryos the accumulation of D-raf proteins after heat induction was approximately twofold greater than that found in wild-type embryos at 5 hr (Fig 3B). At 10 hr, the level of the D-raf^WT protein was unchanged. We also determined the effect of D-raf^WT proteins on otd and dpp gene expression patterns in D-raf embryos. As anticipated, induction of the D-raf^WT gene resulted in 100% of the D-raf null class showing wild-type ventral otd stripe expression and a normal pattern of dpp expression (Table 3). We also examined embryonic cuticles at 24 hr to assess the ability of the D-raf^WT gene to promote signaling in the late-stage Egfr pathway responsible for epidermal differentiation and the final cuticular pattern (CLIFFORD and SCHUPBACH 1992 ; RAZ and SHILO 1993 ). Of these D-raf null embryos that had inherited the D-raf^WT gene, 99% developed cuticles indistinguishable from their D-raf torso sisters (Table 4). Thus, all ectodermal signaling pathways dependent on D-raf activity could be fully restored in null embryos by expression of the D-raf^WT gene.

The consequence of D-raf^S388A and D-raf⁴⁴⁵ expression in D-raf null embryos: Fig 3C shows the quantity of serine to alanine substituted D-raf proteins generated by the D-raf^S388A gene. When compared with the expression of the D-raf^WT gene at 5 hr, an equivalent accumulation of D-raf^S388A protein was observed, with a slight reduction at 10 hr. In our phenotypic analysis, 84% of these D-raf^S388A expressing D-raf null embryos showed rescue of Egfr-induced otd expression in ventral cells and the distance between dpp stripes appeared normal (Table 3). By the completion of embryonic development, 97% of the D-raf null embryos showed the torso phenotype, while the remaining 3% showed a composite "imperfect torso" phenotype (Table 4). In addition to showing head and tail defects associated with the torso phenotype, embryos of the "imperfect torso" class were twisted and had denticle bands of reduced width, indicative of partial loss of signaling in ventral cells that depend on the Egfr pathway for development (Fig 1F and Fig I). Since all of the D-raf null embryos showed some phenotypic rescue by D-raf^S388A, we concluded that serine 388 was not essential for the function of D-raf in the ectoderm. Instead it was likely that S388 plays a negative role in the regulation of D-raf similar to its function in Tor signaling (BAEK et al. 1996 ).

For D-raf null embryos that inherited the D-raf⁴⁴⁵ gene, 52% showed rescue of the Egfr-induced otd expression pattern (Table 3). This was approximately one-half the percentage rescued by the D-raf^WTgene, although the quantity of truncated 38-kD D-raf protein in these embryos was equivalent to that observed for D-raf embryos expressing the D-raf^WT gene at 5 hr (Fig 3B). For the human Raf-1 protein, removal of CR1 and CR2 resulted in unregulated kinase activity (BONNER et al. 1985 ; RAPP et al. 1988 ). We assessed whether the D-raf⁴⁴⁵ protein acted ectopically to create a wide ventral otd stripe, but all of the otd stripes were of wild-type width (data not shown). When dpp mRNA patterns were analyzed in D-raf⁴⁴⁵ expressing null embryos the distance between lateral stripes in the third thoracic segment at 10 hr was similar to those that had inherited the D-raf^WT gene (Table 3).

In the analysis of 24-hr cuticular patterns 52% of the D-raf⁴⁴⁵ embryos were rescued and showed the torso phenotype (Table 4). For the remaining embryos, partial rescue was observed with signaling by the D-raf⁴⁴⁵ protein defective in the determination of the ventral ectoderm. Of these embryos, 18% showed the "imperfect torso" phenotype and 30% showed the "null with denticles" phenotype (Fig 1G and Fig J). These "null with denticles" embryos were twisted, had faint cuticles with narrow denticle bands, and were phenotypically similar to Egfr embryos homozygous for intermediate defective alleles of Egfr (CLIFFORD and SCHUPBACH 1992 ; RAZ and SHILO 1992 , RAZ and SHILO 1993 ). Overall, we found that signal transmission by D-raf⁴⁴⁵ was less reliable when compared with D-raf^WT, although the D-raf⁴⁴⁵ protein had the potential to rescue all aspects of the embryonic D-raf null phenotype.

The consequence of D-raf^S743A and D-raf³¹⁵ expression in D-raf null embryos: Analysis of D-raf embryos expressing the D-raf^S743A gene was somewhat complicated by the insertion of D-raf^S743A on the TM2 balancer chromosome. Thus, only one-half of the D-raf null embryos fertilized by D-raf^S743A transgenic males inherited the D-raf^S743A gene. We determined the amount of D-raf^S743A protein that accumulated in D-raf embryos with the D-raf^S743A gene and found that it was 1.5-fold greater than that observed for those embryos that had inherited the D-raf^WT gene (Fig 3C). Although greater levels of this modified D-raf protein accumulated in D-raf null embryos expressing the D-raf^S743A gene, otd stripe expression was not observed (Table 3). Also, the distance between lateral dpp stripes in these D-raf^S743A embryos was diminished when compared with wild type, but not to the degree observed for embryos expressing the D-raf³¹⁵ or D-raf^K497M genes, as presented below. Thus, the specification of ventral cell fates at the midline requires the positive regulation of the D-raf protein at serine 743.

Accordingly, 99% of the D-raf null embryos expressing the D-raf^S743A gene showed the "imperfect torso" phenotype (Table 4). To better assess the pattern deletions generated by the loss of epidermal cell fates in these D-raf^S743A embryos we scored epidermal sensory organs that develop in ventral and lateral domains of the embryo. The separation between Keilin's organs and ventral black dots on the ventral surface was measured. Also, to determine whether patterning in lateral cells was normal for these embryos the distance between ventral and dorsal black dots was recorded. When compared with wild type, D-raf^S743A embryonic cuticles showed a decreased distance between Keilin's organs and ventral black dots (Table 5 and Fig 4B). A decrease in the distance between ventral and dorsal black dot material was also observed (Fig 4E). This later finding proved very informative for it led to the hypothesis that a novel pathway, dependent upon the D-raf protein, was operating for signal transmission in cells undergoing lateral epidermal development. It appears that cell fate specification in the ventralmost ectoderm via the EGR receptor and proper development of a subpopulation of lateral cells requires an optimal level of D-raf activity that was not achieved by the D-raf^S743A protein.

View larger version (123K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Cuticular preparations of D-raf embryos with D-rafWT, D-raf315, or D-rafS743A expression. Views from the ventral or lateral surface of the third thoracic and first abdominal denticle regions are shown. (A) Expression of the D-rafWT transgene resulted in full rescue of D-raf null embryos including Keilin's organs and ventral and dorsal black dots. Expression of the D-rafS743A or D-raf315 gene resulted in partial rescue of the D-raf null phenotype. Both (B) D-rafS743A and (C) D-raf315 embryos showed pattern deletions in the ventral epidermis, which was most severe for embryos expressing the D-raf315 protein. In this case, deletion of Keilin's organs was accompanied by an expansion of the remaining cellular fates and gave rise to enlarged ventral black dots. For this D-raf315 embryo the naked cuticular region between the enlarged ventral black dots and the first abdominal denticle belt appeared narrow. This was not typical of most D-raf315 embryos and was likely an artifact of embryo twisting. Lateral views of D-raf null embryos with D-raf transgene expression: (D) A wild-type organization of ventral and dorsal black dot material was observed after expression of the D-rafWT transgene. For embryos that developed with (E) D-rafS743A or (F) D-raf315 proteins, the distance between ventral and dorsal black dots was reduced. Deletion of lateral cuticle was most extreme for embryos with D-raf315 proteins. High-magnification views of the lateral surface from the D-raf315 embryo shown in F with ventral and dorsal black dots from the second (G) and third (H) thoracic segments. (I) We hypothesize that D-raf acts to specify cellular fates in two distinct ectodermal domains of the embryo. Together with the EGF receptor, the D-raf protein acts to determine cell fates in the ventral ectoderm. Laterally, D-raf acts with an unknown receptor to elicit lateral ectodermal identities. a1, abdominal denticle belt one; D, dorsal black dot; K, Keilin's organ; V, ventral black dot.

Rescue of epidermal patterning defects was further diminished in D-raf null embryos that expressed the D-raf³¹⁵ gene. Using Western analysis we found that the D-raf³¹⁵ protein migrated as an 60-kD band detected at a level equivalent to that of the 90-kD D-raf^WT protein at 5 hr (Fig 3B). Approximately 80% of this D-raf³¹⁵ protein was present at 10 hr. When D-raf null embryos that inherited the D-raf³¹⁵ gene were assayed for otd and dpp stripe expression, ventral otd expression was not observed and the distance between lateral dpp stripes was much reduced when compared with embryos expressing the D-raf^WT gene (Table 3). Thus, a substantial decrease in the output of the Egfr-induced signal was detected. By the completion of development, 83 (81%) of the expected 102 D-raf null embryos with D-raf³¹⁵ protein showed cuticles with the "null with denticles" phenotype (Table 4).

We also identified epidermal sensory organs in D-raf null embryos expressing the D-raf³¹⁵ gene and made note of their relative positions (Table 5). Significantly, an absence of Keilin's organs was recorded and a corresponding expansion in the size of ventral black dot material was observed (Fig 4C). The distance between these enlarged ventral dots was substantially reduced when compared with wild-type embryos (Table 5). A reduction in the distance between ventral and dorsal black dot sensory organs was also observed (Fig 4, F–H). This finding again implicates D-raf in a pathway required for the development of lateral cells. Thus, by reducing the ability of the D-raf protein to act in signaling we have verified its role in the Egfr pathway and have also uncovered its function in a novel pathway involved in lateral cell development.

The consequence of D-raf^K497M proteins in developing D-raf null embryos: As anticipated, D-raf-dependent pathways were not rescued when D-raf null embryos expressed the kinase defective D-raf^K497M gene. For these embryos the accumulation of D-raf^K497M proteins after heat induction was 2-fold greater than that found in D-raf^WT embryos at 5 hr (Fig 3C). However, by 10 hr this level was dramatically reduced 0.75-fold, indicating that the K497M modification renders the D-raf protein unstable. Induction of the kinase defective D-raf^K497M gene did not restore wild-type otd or dpp expression patterns in D-raf null embryos (Table 3). When assessed after completion of embryogenesis, those embryos that had inherited the D-raf^K497M gene showed the D-raf null phenotype (Table 4 and Table 5). Thus, the kinase activity of the D-raf protein proved essential in those embryonic cells that utilize D-raf for signal transmission.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The role of D-raf in embryonic dorsoventral patterning:
Along the dorsoventral egg perimeter, embryonic cell fates are first established by the dorsal maternal patterning system (reviewed in ANDERSON 1998 ). Elaboration of this pattern in the ectoderm is dependent upon the Dpp protein in dorsal cells and the EGR receptor in ventral cells (SCHWEITZER and SHILO 1997 ; PODOS and FERGUSON 1999 ). The Dpp ligand directs nuclei to initiate programs for the development of extraembryonic aminoserosa and dorsal epidermis. In the ventral domain, the EGF receptor acts to determine cells of the neuroectodermal region that give rise to the ventral nerve cord and the final cuticular pattern. The specification of lateral fates that comprise the remaining 20% of the ectoderm occurs in cells that lie between dorsal and ventral domains and is initiated by an unknown mechanism. These cells secrete cuticular hairs, denticles, or naked cuticle similar to those exoskeletal structures produced by dorsal and ventral cells.

A variety of genetic approaches were employed here to define the role that D-raf plays in the development of the embryo. Using a constitutively active D-raf protein we found that D-raf acts downstream of the Egfr for the specification of ventral ectodermal cell fates. We have also discovered that D-raf plays a second role in a novel pathway that is required for lateral cell development. In particular the D-raf^S743A and D-raf³¹⁵ alleles generated in vitro proved useful in defining the function of D-raf in cells of the lateral epidermis. We hypothesize that this novel pathway acts to specify cells of the lateral ectoderm subsequent to instructions received by nuclei from the dorsal maternal gene product. Thus, dorsoventral patterning in the embryo is likely dependent on the activity of three zygotic signaling pathways with Dpp that acts in dorsal cells, Egfr that directs cells in the ventral ectoderm, and a novel RTK pathway that specifies lateral cell fates.

Determination of lateral cellular fates:
The lateral epidermis consists of two narrow strips of tissue on the left and right sides of the embryo extending from the anterior head to the posterior tail region. For the meta- and meso-thoracic regions this lateral tissue gives rise to epidermal cuticular structures that form between dorsal and ventral black dot sensilla (according to CAMPOS-ORTEGA and HARTENSTEIN 1985 ). Along the circumference of each abdominal segment these two regions of lateral cuticle can be subdivided into dorsolateral and ventrolateral domains. Normally in late-stage embryos the dorsolateral region is characterized by numerous discontinuous rows of long slender hairs that have a pattern similar to that found for region b of the dorsal epidermis (CAMPOS-ORTEGA and HARTENSTEIN 1985 ). These dorsolateral hairs are most similar in size and morphology to a subset of dorsal hairs, the 4° hairs (HEEMSKERK and DINARDO 1994 ). The ventrolateral domain is characterized by a segmental organization of naked cuticle alternating with two to three sparse rows of denticles similar to those found in the ventral belts although not as strongly pigmented.

Several findings have indicated that a novel pathway acts in the determination of lateral ectodermal cell fates and were consistent with a role for D-raf in this pathway. IRISH and GELBART 1987 found that embryos that developed in the absence of dpp and dorsal activity were lateralized. Mutations in the Drosophila dCREB-A gene are also important for defining lateral embryonic regions. In the absence of dCREB-A gene function, embryos showed development of only lateral epidermal structures (ANDREW et al. 1997 ). dCREB-A encodes a transcription factor that is required in both Dpp and Egfr signaling cascades. dpp is a member of the TGFß family, while spitz is a TGF family member and potentiates Egfr signaling in the ventral ectoderm (for review see SCHWEITZER and SHILO 1997 ; PODOS and FERGUSON 1999 ). Two consequences of lateral cell induction were also identified: activation of the MAP kinase protein and expression of the msh gene encoding a homeodomain protein product (SKEATH 1998 ; YAGI et al. 1998 ; VON OHLEN and DOE 2000 ). Using D-raf proteins with partial function we have found that D-raf also participates in the development of the lateral epidermis most likely to specify cellular fates in the lateral ectoderm.

Is there a RTK receptor responsible for triggering the activation of the D-raf protein and MAP kinase in cells of the lateral ectoderm? One possible candidate is the insulin receptor (Inr) gene product. It was noted that occasionally embryos developing without zygotic Inr activity showed a decrease in denticle belt width (FERNANDEZ et al. 1995 ). To better address the potential role of Inr signaling in lateral ectodermal patterning, a phenotypic analysis of Inr mutant embryos derived from eggs lacking maternal contributions of Inr would likely be informative. Alternatively, the generation of protein null alleles of Inr may help to better define the function of this RTK receptor in the embryo. In mammalian systems, mitogenic signaling by insulin in fetal rat, brown adipocyte, and primary cultures involves the activation of Ras and Raf-1 proteins (VALVERDE et al. 1996 ). Insulin was also known to trigger an increase in Raf-1 activity in several cell lines that expressed large numbers of insulin receptors (BLACKSHEAR et al. 1990 ; KOVACINA et al. 1990 ).

A gradient of D-raf activity specifies ventral ectodermal cell fates:
The Raf-MEK-MAP kinase cascade acts in a variety of cells to transmit RTK-generated signals during Drosophila development. Here we show that the protein kinase activity of D-raf was required to elicit distinct ventral cell fates specified by the EGR receptor in early embryos. Using partial loss-of-function mutations in D-raf, cell fates normally specified by high levels of Egfr activity were lost while those that required lower receptor activity appeared normal or were expanded. Similarly, in Tor RTK signaling different levels of D-Ras, D-raf, or corkscrew activity were shown to have distinct transcriptional and morphological consequences in embryonic head and tail development (BAEK et al. 1996 ; GREENWOOD and STRUHL 1997 ; RADKE et al. 1997 ; GHIGLIONE et al. 1999 ).

How is a graded pattern of cell types within a developmental field generated by an RTK receptor? It has been hypothesized that the main function of the Raf-MEK-MAPK phosphorylation cascade is to amplify RTK-initiated signals. In this case, the quantity of activated Raf, MEK, and MAPK molecules is directly proportional to the number of receptor molecules activated, in the absence of feedback mechanisms. This information is then translated into position-dependent gene expression patterns that lead to morphological changes and cellular development. In this model, the quantity of activated RTK receptors defines the determined state of the cell. However, a number of studies in Drosophila reveal the existence of parallel signaling pathways emanating from a receptor during embryonic development (HOU et al. 1995 ; RAABE et al. 1995 ; HERBST et al. 1996 ). To extend the amplification hypothesis, the Raf-MEK-MAP kinase cascade may also act to integrate signals received from these parallel pathways and ultimately define precise transcriptional outcomes using a multistep mechanism. In mammalian cells, Raf-1 is regulated by a variety of inputs including the enzymatic function of PKC, Src, and Jnk kinases that upregulate activity (for review see MORRISON 1994 , MORRISON 1995 ; MORRISON and CUTLER 1997 ). Autophosphorylation also plays a role in regulating Raf-1, as well as binding to Ras, 14-3-3, KSR, hsp90, and p50 proteins. In addition, PKA, Atk (PKB), and phosphatases have been implicated in the downregulation of Raf-1 function (COOK and MCCORMICK 1993 ; WU et al. 1993 ; HAFNER et al. 1994 ; ROMMEL et al. 1999 ; ZIMMERMANN and MOELLING 1999 ).

Here we address the consequences of eliminating key D-raf regulatory domains or specific serine residues that might act to integrate distinct signaling pathways in the Egfr pathway for ventral cell determination. In general, signal transmission was less reliable for D-raf proteins that lacked the negative regulatory site S388 (D-raf^S388A) or the regulatory sequences CR1 and CR2 associated with the N-terminal one-half of the molecule (D-raf⁴⁴⁵). However, both proteins showed the potential to transmit the highest level of ventral signal. This phenomenon was perhaps indicative of an important role played by the D-raf protein in the assembly of multiprotein complexes with components derived from parallel pathways. The full-length wild-type D-raf molecule, which contains several conserved motifs, may serve to bring parallel-signaling components together. Thus, the structural integrity of the D-raf protein may be important for the efficiency of complex assembly or its stability. In this model only complete and stable-signaling complexes achieve the highest level of signal output. We speculate that in the case of D-raf^S388A and more often for D-raf⁴⁴⁵ proteins complete signaling complexes were not built, leading to the phosphorylation of fewer D-MEK molecules, decreased signal output, and fewer cell fate choices specified within the Egfr developmental field.

In contrast, the Egfr signal was severely compromised when transmitted by either D-raf^S743A or D-raf³¹⁵ proteins. The range of cell types specified by these mutant D-raf molecules was dramatically reduced from the wild type. In both cases, the establishment of cell fates that require the highest level of Egfr activity was consistently lost. Serine 743 may be important for the formation of D-raf dimers or oligomers as has been suggested for Raf-1 (FARRAR et al. 1996 ; LUO et al. 1996 ; MORRISON and CUTLER 1997 ). This type of complex may be essential for the generation of the highest level of ventral signal. In embryos that developed with D-raf³¹⁵ proteins, cell fates were generated that required substantially lower levels of Egfr activity. We speculate that the wild-type D-raf protein undergoes release from negative regulation imparted by the CR2 domain via its N-terminal and CR1 sequences. In the case of the D-raf³¹⁵ protein, maintenance of the negative regulatory function of CR2 severely limited the ability of D-raf molecules to activate D-MEK. These results point to a multistep process in the generation of active D-raf molecules with multiple upstream factors acting in parallel. The highest level of D-raf signal was generated when all inputs were received. In the absence of one or several interactions the signaling potential of the D-raf protein was reduced, but not abolished.

	ACKNOWLEDGMENTS

We thank Trudi Schüpbach for the top^1P02/CyO stock, Leslie Pick for the yw; P[w⁺ ftz-ß-gal G2] stock, Masaharu Go for the P[ry+; ftz/lacC]/ CyO stock, and Robert Finklestein (otd) and Ethan Bier (rho) for their contributions of marker gene DNA. We also thank an anonymous reviewer and Kelly Foreman for helpful comments on the manuscript. National Science Foundation grants IBN-9513837 and IBN-998228 to L.A. supported this research. Predoctoral support for K.J. was by National Science Foundation grant DIR-913595.

Manuscript received October 6, 2000; Accepted for publication August 2, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AMBROSIO, L., A. P. MAHOWALD, and N. PERRIMON, 1989a  Requirement of the Drosophila raf homologue for torso function. Nature 342:288-291[Medline].

AMBROSIO, L., A. P. MAHOWALD, and N. PERRIMON, 1989b  l(1)pole hole is required maternally for pattern formation in the terminal regions of the embryo. Development 106:145-158[Abstract].

ANDERSON, K. V., 1998  Pinning down positional information: dorsal-ventral polarity in the Drosophila embryo. Cell 95:439-442[Medline].

ANDREW, D. J., A. BAIG, P. BHANOT, S. M. SMOLIK, and K. D. HENDERSON, 1997  The Drosophila dCREB-A gene is required for dorsal/ventral patterning of the larval cuticle. Development 124:181-193[Abstract].

ASHBURNER, M., 1989 Drosophila: A Laboratory Handbook. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

BAEK, K.-H. and L. AMBROSIO, 1994  An efficient method for microinjection of mRNA into Drosophila embryos. Biotechniques 17:1024-1026[Medline].

BAEK, K.-H., J. R. FABIAN, F. SPRENGER, D. K. MORRISON, and L. AMBROSIO, 1996  The activity of D-raf in torso signal transduction is altered by serine substitution, N-terminal deletion and membrane targeting. Dev. Biol. 175:191-204[Medline].

BIER, E., L. Y. JAN, and Y. N. JAN, 1990  rhomboid, a gene required for dorsoventral axis establishment and peripheral nervous system development in Drosophila melanogaster. Genes Dev. 4:190-203. (erratum: Genes Dev. 4: 680–681).[Abstract/Free Full Text].

BLACKSHEAR, P. J., D. M. HAUPT, H. APP, and U. R. RAPP, 1990  Insulin activates the Raf-1 protein kinase. J. Biol. Chem. 265:12131-12134[Abstract/Free Full Text].

BONNER, T. I., S. B. KERBY, P. SUTRAVE, M. A. GUNNELL, and G. MARK et al., 1985  Structure and biological activity of human homologs of the raf/mil oncogene. Mol. Cell. Biol. 5:1400-1407[Abstract/Free Full Text].

BRAND, A. and N. PERRIMON, 1993  Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118:401-415[Abstract].

CAMPOS-ORTEGA, J. A., and V. HARTENSTEIN, 1985 The Embryonic Development of Drosophila melanogaster. Springer-Verlag, New York/Berlin.

CHANG, H. C. and G. M. RUBIN, 1997  14–3-3 epsilon positively regulates Ras-mediated signaling in Drosophila. Genes Dev. 11:1132-1139[Abstract/Free Full Text].

CHOU, T. B. and N. PERRIMON, 1992  Use of a yeast site-specific recombinase to produce female germline chimeras in Drosophila embryogenesis. Genetics 131:643-653[Abstract].

CLIFFORD, R. J. and T. SCHÜPBACH, 1992  The torpedo (DER) receptor tyrosine kinase is required at multiple times during Drosophila embryogenesis. Development 115:853-872[Abstract].

COOK, S. J. and F. MCCORMICK, 1993  Inhibition by cAMP of Ras-dependent activation of Raf. Science 262:1069-1072[Abstract/Free Full Text].

DAUM, G., I. EISENMANN-TAPPE, H. W. FRIES, J. TROPPMAIR, and U. R. RAPP, 1994  The ins and outs of Raf kinases. Trends Biochem. Sci. 19:474-480[Medline].

DIAZ-BENJUMEA, F. D. and E. HAFEN, 1994  The sevenless signalling cassette mediates Drosophila EGF receptor function during epidermal development. Development 120:569-578[Abstract].

DICKSON, B. and E. HAFEN, 1994  Genetics of signal transduction in invertebrates. Curr. Opin. Genet. Dev. 4:64-70[Medline].

DUFFY, J. B. and N. PERRIMON, 1994  The Torso pathway in Drosophila: lessons on receptor tyrosine kinase signaling and patterning formation. Dev. Biol. 166:380-395[Medline].

FARRAR, M. A., I. ALBEROL, and R. M. PERLMUTTER, 1996  Activation of the Raf-1 kinase cascade by coumermycin-induced dimerization. Nature 383:178-181[Medline].

FERNANDEZ, R., D. TABARINI, N. AZPIAZU, M. FRASCH, and J. SCHLESSINGER, 1995  The Drosophila insulin receptor homolog: a gene essential for embryonic development encodes two receptor isoforms with different signaling potential. EMBO J. 14:3373-3384[Medline].

FINKELSTEIN, R. and N. PERRIMON, 1990  The orthodenticle gene is regulated by bicoid and torso and specifies Drosophila head development. Nature 346:485-488[Medline].

GHIGLIONE, C., N. PERRIMON, and L. A. PERKINS, 1999  Quantitative variations in the level of MAPK activity control patterning of the embryonic termini in Drosophila. Dev. Biol. 205:181-193[Medline].

GREENWOOD, S. and G. STRUHL, 1997  Different levels of Ras activity can specify distant transcriptional and morphological consequences in early Drosophila embryos. Development 124:4879-4886[Abstract].

GREENWOOD, S. and G. STRUHL, 1999  Progression of the morphogenetic furrow in the Drosophila eye: the roles of Hedgehog, Decapentaplegic and the Raf pathway. Development 126:5795-5808[Abstract].

HAFNER, S., H. S. ADLER, H. MISCHAK, P. JANOSCH, and G. HEIDECKER et al., 1994  Mechanism of inhibition of Raf-1 by protein kinase A. Mol. Cell. Biol. 14:6696-6703[Abstract/Free Full Text].

HALFAR, K., C. ROMMEL, H. STOCKER, and E. HAFEN, 2001  Ras controls growth,