- Email: [email protected]
Drosophila Gastrulation: Identification of a Missing Link
Current Biology, Vol. 14, R480–R482, June 22, 2004, ©2004 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cub.2004.06.016
Drosophila Gastrulation: Identification of a Missing Link Maria Leptin1 and Markus Affolter2
Drosophila gastrulation serves as a model system to elucidate the genetic and molecular mechanisms involved in morphogenetic movements. The ligand of the FGF receptor Heartless, which is involved in mesoderm movement, has now been isolated and shown to be a link between a morphogen gradient and cell behavior.
Genetic screens occasionally fail to uncover a missing developmental regulatory gene, and in many cases this is due to functional redundancy. In cases in which the molecular nature of the gene product can be anticipated (for example, a ligand of a known receptor), the the genome sequence helps the search and in most cases leads to the identification of the missing component. However, even in organisms for which the full genome sequence is available a candidate gene may not be evident, and this is usually the point where speculations about novel mechanisms emerge. The ligand of Heartless (Htl), a fibroblast growth factor (FGF) receptor in Drosophila, has been a case in point: despite extensive genetic analysis, it was not identified in standard genetic screens and did also not show up when the Drosophila genome was searched using bioinformatics. A number of explanations were put forward to account for the ‘missing-in-action’ factor, ranging from the incompleteness of the genome sequence to activation of the receptor by other cell surface molecules or even ligand-independent receptor activation. Two recent papers [1,2] put an end to these entertaining beer-hour discussions, as they report the identification of two FGF8-like ligands required for the activation of Heartless. Although the papers report the identification of the same two genes, the experimental approaches that led to their isolation were very different. Gryzik and Müller [2] set out to find genomic regions containing genes involved in mesoderm morphogenesis, which includes the Htl-dependent spreading of the mesodermal cell layer during the early stages of gastrulation. One such region showed a phenotype resembling that of htl mutants. Painstaking genetic dissection, including the production of sets of defined, small overlapping deletions in the genome, narrowed the region down to a stretch that contained 14 genes. Two neighboring genes were expressed at the right time and place to be candidates for the Htl-ligand. Most satisfyingly, these two genes encode functionally redundant FGF8-like molecules.
1Institute
of Genetics, University of Cologne, Weyertal 121, D50931 Cologne, Germany. E-mail: [email protected] 2Biozentrum of the University of Basel, Klingelbergstrasse 70, CH-4053 Basel, Switzerland. E-mail: [email protected]
Dispatch
These results highlight the use of defined deletions in the hunt for novel gene functions, to overcome certain cases of functional redundancy (if the genes in question are close to each other), or to look for subtle phenotypes which put limitations on the number of individual mutants that can be carefully scored. The success of Gryzik and Müller [2] depended on exploiting a new method to generate precisely defined chromosomal deletions in the Drosophila genome, which is currently used in a large-scale approach to generate a genome-wide collection of such deletions [3,4]. The same two fgf8-like genes were isolated in a totally different approach by Stathopoulos et al. [1]. For a number of years the group of Michael Levine has been analyzing the genetic cascade involved in the determination of cell fates along the dorsal–ventral axis of the early Drosophila embryos. Cell fates along this axis are controlled by a nuclear gradient of the transcription factor Dorsal, and the Levine group [5] has used expression profiling to identify genes regulated by various levels of the Dorsal morphogen. These studies aimed at defining a detailed developmental pathway leading to the establishment of the dorsal–ventral body axis in the fly embryo. The FGF receptor Htl is activated by the transcription factor Twist in the ventral cells that represent the anlage of the mesoderm and that contain the highest nuclear concentration of Dorsal. Microarray screens identified a gene called Neu4 that is directly activated by low levels of the maternal Dorsal gradient in broad lateral stripes flanking the mesodermal anlage [5]. Initially, no function could be predicted for this gene, as only a small fragment had been isolated, which showed no clear homology to genes with known functions, and the gene annotation programs failed to detect the entire coding region. However, a combination of sophisticated computational and molecular methods led to the identification of the encoded protein as a FGF8-like molecule and in the discovery of a neighboring sister gene presumably generated by gene duplication [1]. Many receptor tyrosine kinases are expressed broadly and activated locally by spatially restricted ligands. Because FGF receptors control aspects of directed cell migration or cell spreading [6], it is important that receptors and ligands not be expressed in the same cells. This would impede migration because the spatial information that is built into the distribution of ligand and receptor would be lost. In contrast to other receptor tyrosine kinases, the expression of the two FGF receptors Breathless (Btl) and Htl, along with their downstream signaling component Downstream-of-FGFreceptor (Dof) is restricted to defined cells and tissues in the fruit fly [7–10]. Expression of the ligand Branchless (Bnl), which is specific for Btl, is under dynamic transcriptional control, and ligand and receptor are largely expressed in non-overlapping, complementary groups of cells [11]. Indeed, misexpression of the ligand or expression of constitutively active FGF receptors
Current Biology R481
A
B
D
C
Dorsal
Dorsal
Twist Snail Heartless Dof
FGF8 Mesoderm
FGF8 Ectoderm Current Biology
Figure 1. Expression patterns and hierarchy of genes involved in the FGF-dependent morphogenesis of the mesoderm. The expression patterns are shown on cross-sections of embryos in successive stages of gastrulation. The drawings do not accurately reflect the RNAs’ first appearance, but rather the logical succession of the hierarchy of gene activities. The ventral side of the embryo is at the bottom. (A) The amount of Dorsal protein in the nuclei of the embryo is highest on the ventral side (bottom) and declines in a gradient toward the dorsal side of the embryo. (B) Genes that define the mesoderm under control of the Dorsal gradient: Twist (red) is directly activated by Dorsal and is expressed in a gradient with the highest level in the most ventral cells. Snail (blue) is also activated weakly by Dorsal, but the combination of Dorsal and Twist is needed for full expression and the definition of the sharp boundary at the edge of the mesoderm. (C) The FGF receptor Heartless and its signal transduction substrate Dof (both yellow) are expressed in the central portion of the invaginated mesoderm, which is about to flatten out onto the ectoderm. The ligand for the receptor, Fgf8 (green), is expressed in the ectoderm. (D) Genetic hierarchy of genes shown in A–C. Left: The highest level of Dorsal, found in the prospective mesoderm, is able to activate genes such as twist and snail, which have low affinity binding sites in their promoters, as well as those with higher affinity binding sites. The promoter of htl also has high affinity binding sites that can respond directly to Dorsal, but in the embryo, htl critically depends on Twist [15]. The transcription factor Twist acts as an activator promoting its own transcription as well as that of snail, htl and dof. Snail is a repressor that is predicted to block the expression of the Fgf8 genes in the mesoderm. Right: low levels of Dorsal, as present in the prospective ectoderm, can activate only those genes with high affinity binding sites. Thus, Fgf8 is expressed, but Twist and Snail are not, and therefore Htl and Dof are not activated and Fgf8 expression is not blocked.
leads to prominent defects in cell migration [12], illustrating the importance of the restricted, complementary expression patterns. The work of Stathopoulos et al. [1,5] provides a beautiful example of how the complementary expression pattern of Htl and its two FGF8-like ligands is achieved during development (Figure 1). htl, dof and the two Htlligand-encoding genes appear to be direct targets of the transcription factor Dorsal. High levels of Dorsal participate with Twist in activating htl and dof ventrally, i.e. in the mesoderm, while the two FGF ligands are activated laterally by lower levels of Dorsal, and are therefore expressed more broadly along the dorsal–ventral axis.
To bring about the non-complementary expression of ligand and receptor, Dorsal activates transcription of the Snail repressor in the ventral mesoderm. Snail binding sites have been identified in an enhancer of the fgf8-like genes, and it is likely that Snail represses their transcription in the ventral cells. These results provide the logic of a gene regulatory cascade initiated from the graded distribution of a ligand, followed by differential activation of a receptor (Toll), the graded nuclear accumulation of a transcription factor (Dorsal), and the specification of regional cell fate. Gastrulation in the Drosophila embryo can thus be described as a manifestation of discrete threshold readouts of the Dorsal gradient. It remains to be determined whether FGF ligands act as true chemoattractants, or rather as motogens inducing motility per se. In many cases where cursory observations suggest chemotaxis, other mechanisms could be operating to determine the direction of movement, such as polarity in the substratum, release from attachments, or plain constraints in the availability of space into which movement is possible. A role for one Drosophila FGF as a chemoattractant is supported by studies of Branchless/FGF (Bnl/FGF), the ligand of the second Drosophila FGF receptor, Btl. In the developing tracheal system, Bnl/FGF is expressed dynamically in cells surrounding the tracheal placodes, thus prefiguring the direction of migration of the tracheal tip cells [11]. The expression of Bnl/FGF in ectopic positions induces tracheal cell migration toward such sites, both in the embryo and in the larva [11,13]. In addition, live imaging has shown that activation of both Drosophila FGF receptors leads to cytoskeletal changes that are typical for migrating cells and that Bnl induces the formation of directed filopodia [2,13,14]. The Htl ligands might act both in a permissive and in an instructive manner; permissive for the early spreading of the mesodermal cells and instructive for the efficient guidance of the movement of the mesoderm toward the dorsal ectoderm. The late modulation in the levels and expression patterns of the two novel Htl ligands (Figure 1D) is consistent with such a view [1], and the identification of these ligands now allows us to address this issue directly by manipulating ligand expression experimentally. References 1. Stathopoulos, A., Tam, B., Ronshaugen, M., Frasch, M., and Levine, M. (2004). pyramus and thisbe: FGF genes that pattern the mesoderm of Drosophila embryos. Genes Dev. 18, 687–699. 2. Gryzik, T., and Müller, H.A. (2004). FGF8-like1 and FGF8-like2 Encode Putative Ligands of the FGF Receptor Htl and Are Required for Mesoderm Migration in the Drosophila Gastrula. Curr. Biol. 14, 659–667. 3. Parks, A.L., Cook, K.R., Belvin, M., Dompe, N.A., Fawcett, R., Huppert, K., Tan, L.R., Winter, C.G., Bogart, K.P., Deal, J.E., et al. (2004). Systematic generation of high-resolution deletion coverage of the Drosophila melanogaster genome. Nat. Genet. 36, 288–292. 4. Ryder, E., Blows, F., Ashburner, M., Bautista-Llacer, R., Coulson, D., Drummond, J., Webster, J., Gubb, D., Gunton, N., Johnson, G., et al. (2004). The DrosDel collection: a set of P-element insertions for generating custom chrosomomal aberrations in Drosophila melanogaster. Genetics, in press. 5. Stathopoulos, A., Van Drenth, M., Erives, A., Markstein, M., and Levine, M. (2002). Whole-genome analysis of dorsal-ventral patterning in the Drosophila embryo. Cell 111, 687–701. 6. Dormann, D., and Weijer, C.J. (2003). Chemotactic cell movement during development. Curr. Opin. Genet. Dev. 13, 358–364. 7. Beiman, M., Shilo, B.Z., and Volk, T. (1996). Heartless, a Drosophila FGF receptor homolog, is essential for cell migration and establishment of several mesodermal lineages. Genes Dev. 10, 2993–3002.
Dispatch R482
8. Klämbt, C., Glazer, L., and Shilo, B.Z. (1992). breathless, a Drosophila FGF receptor homolog, is essential for migration of tracheal and specific midline glial cells. Genes Dev. 6, 1668–1678. 9. Gisselbrecht, S., Skeath, J.B., Doe, C.Q., and Michelson, A.M. (1996). heartless encodes a fibroblast growth factor receptor (DFR1/DFGF-R2) involved in the directional migration of early mesodermal cells in the Drosophila embryo. Genes Dev. 10, 3003–3017. 10. Vincent, S., Wilson, R., Coelho, C., Affolter, M., and Leptin, M. (1998). The Drosophila protein Dof is specifically required for FGF signaling. Mol. Cell 2, 515–525. 11. Sutherland, D., Samakovlis, C., and Krasnow, M.A. (1996). branchless encodes a Drosophila FGF homolog that controls tracheal cell migration and the pattern of branching. Cell 87, 1091–1101. 12. Lee, T., Hacohen, N., Krasnow, M., and Montell, D.J. (1996). Regulated Breathless receptor tyrosine kinase activity required to pattern cell migration and branching in the Drosophila tracheal system. Genes Dev. 10, 2912–2921. 13. Sato, M., and Kornberg, T.B. (2002). FGF is an essential mitogen and chemoattractant for the air sacs of the Drosophila tracheal system. Dev. Cell 3, 195–207. 14. Ribeiro, C., Ebner, A., and Affolter, M. (2002). In vivo imaging reveals different cellular functions for FGF and Dpp signaling in tracheal branching morphogenesis. Dev. Cell 2, 677–683. 15. Shishido, E., Higashijima, S., Emori, Y., and Saigo, K. (1993). Two FGF-receptor homologues of Drosophila: one is expressed in mesodermal primordium in early embryos. Development 117, 751–761.
Drosophila Gastrulation: Identification of a Missing Link Maria Leptin1 and Markus Affolter2
Drosophila gastrulation serves as a model system to elucidate the genetic and molecular mechanisms involved in morphogenetic movements. The ligand of the FGF receptor Heartless, which is involved in mesoderm movement, has now been isolated and shown to be a link between a morphogen gradient and cell behavior.
Genetic screens occasionally fail to uncover a missing developmental regulatory gene, and in many cases this is due to functional redundancy. In cases in which the molecular nature of the gene product can be anticipated (for example, a ligand of a known receptor), the the genome sequence helps the search and in most cases leads to the identification of the missing component. However, even in organisms for which the full genome sequence is available a candidate gene may not be evident, and this is usually the point where speculations about novel mechanisms emerge. The ligand of Heartless (Htl), a fibroblast growth factor (FGF) receptor in Drosophila, has been a case in point: despite extensive genetic analysis, it was not identified in standard genetic screens and did also not show up when the Drosophila genome was searched using bioinformatics. A number of explanations were put forward to account for the ‘missing-in-action’ factor, ranging from the incompleteness of the genome sequence to activation of the receptor by other cell surface molecules or even ligand-independent receptor activation. Two recent papers [1,2] put an end to these entertaining beer-hour discussions, as they report the identification of two FGF8-like ligands required for the activation of Heartless. Although the papers report the identification of the same two genes, the experimental approaches that led to their isolation were very different. Gryzik and Müller [2] set out to find genomic regions containing genes involved in mesoderm morphogenesis, which includes the Htl-dependent spreading of the mesodermal cell layer during the early stages of gastrulation. One such region showed a phenotype resembling that of htl mutants. Painstaking genetic dissection, including the production of sets of defined, small overlapping deletions in the genome, narrowed the region down to a stretch that contained 14 genes. Two neighboring genes were expressed at the right time and place to be candidates for the Htl-ligand. Most satisfyingly, these two genes encode functionally redundant FGF8-like molecules.
1Institute
of Genetics, University of Cologne, Weyertal 121, D50931 Cologne, Germany. E-mail: [email protected] 2Biozentrum of the University of Basel, Klingelbergstrasse 70, CH-4053 Basel, Switzerland. E-mail: [email protected]
Dispatch
These results highlight the use of defined deletions in the hunt for novel gene functions, to overcome certain cases of functional redundancy (if the genes in question are close to each other), or to look for subtle phenotypes which put limitations on the number of individual mutants that can be carefully scored. The success of Gryzik and Müller [2] depended on exploiting a new method to generate precisely defined chromosomal deletions in the Drosophila genome, which is currently used in a large-scale approach to generate a genome-wide collection of such deletions [3,4]. The same two fgf8-like genes were isolated in a totally different approach by Stathopoulos et al. [1]. For a number of years the group of Michael Levine has been analyzing the genetic cascade involved in the determination of cell fates along the dorsal–ventral axis of the early Drosophila embryos. Cell fates along this axis are controlled by a nuclear gradient of the transcription factor Dorsal, and the Levine group [5] has used expression profiling to identify genes regulated by various levels of the Dorsal morphogen. These studies aimed at defining a detailed developmental pathway leading to the establishment of the dorsal–ventral body axis in the fly embryo. The FGF receptor Htl is activated by the transcription factor Twist in the ventral cells that represent the anlage of the mesoderm and that contain the highest nuclear concentration of Dorsal. Microarray screens identified a gene called Neu4 that is directly activated by low levels of the maternal Dorsal gradient in broad lateral stripes flanking the mesodermal anlage [5]. Initially, no function could be predicted for this gene, as only a small fragment had been isolated, which showed no clear homology to genes with known functions, and the gene annotation programs failed to detect the entire coding region. However, a combination of sophisticated computational and molecular methods led to the identification of the encoded protein as a FGF8-like molecule and in the discovery of a neighboring sister gene presumably generated by gene duplication [1]. Many receptor tyrosine kinases are expressed broadly and activated locally by spatially restricted ligands. Because FGF receptors control aspects of directed cell migration or cell spreading [6], it is important that receptors and ligands not be expressed in the same cells. This would impede migration because the spatial information that is built into the distribution of ligand and receptor would be lost. In contrast to other receptor tyrosine kinases, the expression of the two FGF receptors Breathless (Btl) and Htl, along with their downstream signaling component Downstream-of-FGFreceptor (Dof) is restricted to defined cells and tissues in the fruit fly [7–10]. Expression of the ligand Branchless (Bnl), which is specific for Btl, is under dynamic transcriptional control, and ligand and receptor are largely expressed in non-overlapping, complementary groups of cells [11]. Indeed, misexpression of the ligand or expression of constitutively active FGF receptors
Current Biology R481
A
B
D
C
Dorsal
Dorsal
Twist Snail Heartless Dof
FGF8 Mesoderm
FGF8 Ectoderm Current Biology
Figure 1. Expression patterns and hierarchy of genes involved in the FGF-dependent morphogenesis of the mesoderm. The expression patterns are shown on cross-sections of embryos in successive stages of gastrulation. The drawings do not accurately reflect the RNAs’ first appearance, but rather the logical succession of the hierarchy of gene activities. The ventral side of the embryo is at the bottom. (A) The amount of Dorsal protein in the nuclei of the embryo is highest on the ventral side (bottom) and declines in a gradient toward the dorsal side of the embryo. (B) Genes that define the mesoderm under control of the Dorsal gradient: Twist (red) is directly activated by Dorsal and is expressed in a gradient with the highest level in the most ventral cells. Snail (blue) is also activated weakly by Dorsal, but the combination of Dorsal and Twist is needed for full expression and the definition of the sharp boundary at the edge of the mesoderm. (C) The FGF receptor Heartless and its signal transduction substrate Dof (both yellow) are expressed in the central portion of the invaginated mesoderm, which is about to flatten out onto the ectoderm. The ligand for the receptor, Fgf8 (green), is expressed in the ectoderm. (D) Genetic hierarchy of genes shown in A–C. Left: The highest level of Dorsal, found in the prospective mesoderm, is able to activate genes such as twist and snail, which have low affinity binding sites in their promoters, as well as those with higher affinity binding sites. The promoter of htl also has high affinity binding sites that can respond directly to Dorsal, but in the embryo, htl critically depends on Twist [15]. The transcription factor Twist acts as an activator promoting its own transcription as well as that of snail, htl and dof. Snail is a repressor that is predicted to block the expression of the Fgf8 genes in the mesoderm. Right: low levels of Dorsal, as present in the prospective ectoderm, can activate only those genes with high affinity binding sites. Thus, Fgf8 is expressed, but Twist and Snail are not, and therefore Htl and Dof are not activated and Fgf8 expression is not blocked.
leads to prominent defects in cell migration [12], illustrating the importance of the restricted, complementary expression patterns. The work of Stathopoulos et al. [1,5] provides a beautiful example of how the complementary expression pattern of Htl and its two FGF8-like ligands is achieved during development (Figure 1). htl, dof and the two Htlligand-encoding genes appear to be direct targets of the transcription factor Dorsal. High levels of Dorsal participate with Twist in activating htl and dof ventrally, i.e. in the mesoderm, while the two FGF ligands are activated laterally by lower levels of Dorsal, and are therefore expressed more broadly along the dorsal–ventral axis.
To bring about the non-complementary expression of ligand and receptor, Dorsal activates transcription of the Snail repressor in the ventral mesoderm. Snail binding sites have been identified in an enhancer of the fgf8-like genes, and it is likely that Snail represses their transcription in the ventral cells. These results provide the logic of a gene regulatory cascade initiated from the graded distribution of a ligand, followed by differential activation of a receptor (Toll), the graded nuclear accumulation of a transcription factor (Dorsal), and the specification of regional cell fate. Gastrulation in the Drosophila embryo can thus be described as a manifestation of discrete threshold readouts of the Dorsal gradient. It remains to be determined whether FGF ligands act as true chemoattractants, or rather as motogens inducing motility per se. In many cases where cursory observations suggest chemotaxis, other mechanisms could be operating to determine the direction of movement, such as polarity in the substratum, release from attachments, or plain constraints in the availability of space into which movement is possible. A role for one Drosophila FGF as a chemoattractant is supported by studies of Branchless/FGF (Bnl/FGF), the ligand of the second Drosophila FGF receptor, Btl. In the developing tracheal system, Bnl/FGF is expressed dynamically in cells surrounding the tracheal placodes, thus prefiguring the direction of migration of the tracheal tip cells [11]. The expression of Bnl/FGF in ectopic positions induces tracheal cell migration toward such sites, both in the embryo and in the larva [11,13]. In addition, live imaging has shown that activation of both Drosophila FGF receptors leads to cytoskeletal changes that are typical for migrating cells and that Bnl induces the formation of directed filopodia [2,13,14]. The Htl ligands might act both in a permissive and in an instructive manner; permissive for the early spreading of the mesodermal cells and instructive for the efficient guidance of the movement of the mesoderm toward the dorsal ectoderm. The late modulation in the levels and expression patterns of the two novel Htl ligands (Figure 1D) is consistent with such a view [1], and the identification of these ligands now allows us to address this issue directly by manipulating ligand expression experimentally. References 1. Stathopoulos, A., Tam, B., Ronshaugen, M., Frasch, M., and Levine, M. (2004). pyramus and thisbe: FGF genes that pattern the mesoderm of Drosophila embryos. Genes Dev. 18, 687–699. 2. Gryzik, T., and Müller, H.A. (2004). FGF8-like1 and FGF8-like2 Encode Putative Ligands of the FGF Receptor Htl and Are Required for Mesoderm Migration in the Drosophila Gastrula. Curr. Biol. 14, 659–667. 3. Parks, A.L., Cook, K.R., Belvin, M., Dompe, N.A., Fawcett, R., Huppert, K., Tan, L.R., Winter, C.G., Bogart, K.P., Deal, J.E., et al. (2004). Systematic generation of high-resolution deletion coverage of the Drosophila melanogaster genome. Nat. Genet. 36, 288–292. 4. Ryder, E., Blows, F., Ashburner, M., Bautista-Llacer, R., Coulson, D., Drummond, J., Webster, J., Gubb, D., Gunton, N., Johnson, G., et al. (2004). The DrosDel collection: a set of P-element insertions for generating custom chrosomomal aberrations in Drosophila melanogaster. Genetics, in press. 5. Stathopoulos, A., Van Drenth, M., Erives, A., Markstein, M., and Levine, M. (2002). Whole-genome analysis of dorsal-ventral patterning in the Drosophila embryo. Cell 111, 687–701. 6. Dormann, D., and Weijer, C.J. (2003). Chemotactic cell movement during development. Curr. Opin. Genet. Dev. 13, 358–364. 7. Beiman, M., Shilo, B.Z., and Volk, T. (1996). Heartless, a Drosophila FGF receptor homolog, is essential for cell migration and establishment of several mesodermal lineages. Genes Dev. 10, 2993–3002.
Dispatch R482
8. Klämbt, C., Glazer, L., and Shilo, B.Z. (1992). breathless, a Drosophila FGF receptor homolog, is essential for migration of tracheal and specific midline glial cells. Genes Dev. 6, 1668–1678. 9. Gisselbrecht, S., Skeath, J.B., Doe, C.Q., and Michelson, A.M. (1996). heartless encodes a fibroblast growth factor receptor (DFR1/DFGF-R2) involved in the directional migration of early mesodermal cells in the Drosophila embryo. Genes Dev. 10, 3003–3017. 10. Vincent, S., Wilson, R., Coelho, C., Affolter, M., and Leptin, M. (1998). The Drosophila protein Dof is specifically required for FGF signaling. Mol. Cell 2, 515–525. 11. Sutherland, D., Samakovlis, C., and Krasnow, M.A. (1996). branchless encodes a Drosophila FGF homolog that controls tracheal cell migration and the pattern of branching. Cell 87, 1091–1101. 12. Lee, T., Hacohen, N., Krasnow, M., and Montell, D.J. (1996). Regulated Breathless receptor tyrosine kinase activity required to pattern cell migration and branching in the Drosophila tracheal system. Genes Dev. 10, 2912–2921. 13. Sato, M., and Kornberg, T.B. (2002). FGF is an essential mitogen and chemoattractant for the air sacs of the Drosophila tracheal system. Dev. Cell 3, 195–207. 14. Ribeiro, C., Ebner, A., and Affolter, M. (2002). In vivo imaging reveals different cellular functions for FGF and Dpp signaling in tracheal branching morphogenesis. Dev. Cell 2, 677–683. 15. Shishido, E., Higashijima, S., Emori, Y., and Saigo, K. (1993). Two FGF-receptor homologues of Drosophila: one is expressed in mesodermal primordium in early embryos. Development 117, 751–761.