BIOLOGICAL OVERVIEW

branchless is Drosophila's only known fibroblast growth factor (FGF), and as its name implies, is involved in branching morphogenesis. Branching morphogenesis is an essential part of development in organs such as the trachea and lungs. Before taking a closer look at the gene branchless, a brief overview of branching morphogenesis and the role of fibroblast growth factors in general: FGFs constitute a large family of peptide growth factors, with nine different mammalian FGF genes and four genes encoding FGF receptors (Johnson, 1993). FGFs function as mitogens, trophic factors and differentiation factors. FGFs are involved in several processes of branching morphogenesis: in lung morphogenesis, in branching of seminiferous tubules and as stimulators of angiogenesis.

The Drosophila FGF receptor Breathless (BTL), expressed in the developing tracheal system, promotes branching at several stages of tracheal development, and plays a permissive role in the formation of certain branches (Reichman-Fried, 1994 and 1995). A second FGF receptor, Fibroblast growth factor receptor 1, is necessary for heart formation and mesodermal cell migration (Beiman, 1996 and Gisselbrecht, 1996).

The Drosophila tracheal (respiratory) system is a tubular epithelial network that delivers oxygen to internal tissues. Trachea form from segmentally repeated clusters of tracheal precursor cells, which give rise to the tracheal system by cell migration and elongation. Each cluster invaginates from the ectoderm and forms an epithelial sac of about 80 cells. The six main (primary) branches of each sac begin to form as one or two lead cells from any of five or six positions migrate out in stereotyped directions, similar from segment to segment. A small number of cells follow the lead cells and organize into tubes as they migrate. Several hours later, secondary branches sprout from the primary branches. Subsequently secondary branches ramify into dozens of terminal branches, which are long cytoplasmic extension that form a lumen and transport oxygen directly to the fly's tissues. Each level of tracheal branching is controlled by a particular set of genes, which have provided molecular markers for the different branch types. For example, breathless is necessary for primary branching, pointed is required to form secondary branches (Samakovlis, 1996) and Serum response factor (Guillemin, 1996) regulates terminal branch formation (Sutherland, 1996).

In the absence of branchless, tracheal cells fail to migrate and branch. bnl is expressed not in the trachea but in ectodermal cells that overly the migrating and branching trachea. The striking feature of bnl expression is its spatial complexity, one of the most complex patterns known in the embryo. For example, at stage 11, just before tracheal branching begins, bnl expression appears in five small clusters of epidermal cells arrayed around the tracheal sac, at positions where the five primary tracheal branches will soon bud. As the primary branches grow by cell migration over the next 2 hours (stages 12 and 13), expression in the clusters decreases. This appears to occur in a specific spatial pattern: the bnl-expressing cells closest to or contacting the growing tracheal branches lose expression first, with the tracheal cells continuing to migrate toward the remaining bnl-expressing cells. Two more cell clusters begin expressing bnl as expression in the other clusters turns off, presaging the subsequent outgrowth of additional specific branches (Sutherland, 1996).

What regulates the complex expression pattern of bnl? The answer is not yet known, but most likely a conbination of segmentation and dorsal-ventral patterning genes act on the promoter sequence of bnl. The even more striking dynamic nature of bnl expression, with new patches developing bnl expression even as expression in older patches declines, suggests that there is a complex temporal scheme of bnl expression added on top of the spatial one, and that there is feedback between branching and bnl expression, resulting in the diminished expression overlying positions where branching has taken place (Sutherland, 1996).

In Drosophila, Branchless acts as a guidance molecule controlling tracheal cell migration, but does not act as a mitogen, a documented role of FGFs in vertebrates. As expected for a chemoattractant, the concentration of BNL is critical for migration: migration of tracheal branches is sensitive to the level of bnl expression. BNL acts not only on initial tracheal branching, but acts during formation of secondary and tertiary branching. At each stage in branching, different genes, each dependent on bnl function, are expressed in temporal sequence (Sutherland, 1996). What regulates the temporal progression of bnl dependent gene expression? As descrubed below, one of the critical regulatiors of bnl levels is the local oxygen supply.

Sprouting of the major tracheal branches is stereotyped and controlled by hard-wired developmental cues. Ramification of the fine terminal branches is variable and regulated by oxygen, and this process is controlled by a local signal or signals produced by oxygen-starved cells. Evidence is provided that the critical signal is Branchless (Bnl) FGF, the same growth factor that patterns the major branches during embryogenesis. During larval life, oxygen deprivation stimulates expression of Bnl, and the secreted growth factor functions as a chemoattractant that guides new terminal branches to the expressing cells. Thus, a single growth factor is reiteratively used to pattern each level of airway branching, and the change in branch patterning results from a switch from developmental to physiological control of its expression (Jarecki, 1999).

There were 68% more cytoplasmic extensions and mature branches in first instar larvae grown for 20 hr under 5% O2 compared to siblings grown under normal atmospheric oxygen (21% O2). Conversely, larvae grown under high oxygen tension (60% O2) had fewer branches than normoxic controls. Similar effects are observed at each larval stage and at different positions in the animal. The morphology of the branches is also affected by oxygen. Under high oxygen tension the terminal branches are shorter and straight with few side branches, whereas under the low oxygen condition they are long and tortuous with many side branches. Thus, oxygen is an important regulator of terminal branching, influencing the initial budding and the final number and morphology of the branches (Jarecki, 1999).

To determine if the effects of oxygen on terminal branching are globally regulated or mediated by local signals produced by oxygen-starved tissues, small regions of the larvae were deprived of their normal oxygen supply. This was accomplished by generating clones of blistered- tracheal cells, which are unable to form terminal branches, leaving the surrounding region without its normal tracheation. The blistered+ tracheal cells in neighboring segments grow into the detracheated region, sprouting 40% more terminal branches than normal, whereas blistered+ tracheal cells not bordering the clone are unaffected. To determine if it was the absence of terminal branches or just the absence of an oxygen supply that induced neighboring branches to grow, clones of synaptobrevin- tracheal cells that extend terminal branches but are unable to form a lumen and deliver oxygen to the target were tested. Even in the presence of the nonfunctional synaptobrevin- terminal branches, neighboring wild-type tracheal cells grow into the oxygen-starved region. This implies that it is lack of oxygen delivery to the region, not some other function of missing terminal branches, that causes neighboring tracheal cells to respond. It is concluded that terminal branching is regulated by local oxygen need, and that oxygen-starved cells produce a signal or signals that can attract tracheal branches from as far as one segment away (Jarecki, 1999).

When bnl expression is examined by in situ hybridization at larval stages of development, it is found that the gene turns back on by the first larval instar and continues to be expressed throughout larval life. However, in contrast to its highly restricted expression pattern in the embryo, the gene is broadly expressed in the larva, including all tissues that become heavily tracheated with terminal branches. In particular, the three most highly tracheated tissues—gut, muscles, and central nervous system (CNS)—all show generalized expression of bnl during the first and second larval instars when terminal branches are sprouting. Thus, bnl is expressed in the appropriate tissues at the right time to regulate terminal branching in the larva. Several tissues that are not tracheated, including the epidermis and salivary gland, do not express bnl, strengthening the correlation between bnl expression and terminal branching. The correlation is not absolute though, as several other tissues with few or no branches, including the imaginal discs, heart, and fat body, do express significant levels of BNL mRNA. However, in no case does a terminally tracheated tissue not express bnl (Jarecki, 1999).

The arborization pattern of terminal branches is remarkably complex. Each terminal tracheal cell sprouts dozens of branches that spread out and contact nearly every cell in the target. At first glance, the complex and variable nature of the branch pattern suggests that the process is highly random, with the constraint only that branches fill the available space. The results presented here support a model in which the final pattern, although variable, is not at all random. Instead, branching is precisely controlled during development to meet the oxygen needs of the target cells. A model is proposed in which each cell experiencing an oxygen debt senses the impending crisis and responds by upregulating expression of Bnl. Bnl FGF diffuses to nearby tracheal cells and stimulates new tracheal branches to form and grow toward each signaling source. This supplies oxygen to the hypoxic cells and shuts off the signal. The process is dynamic and repeats itself many times due to the constantly changing balance in cell oxygen need and supply. Over the course of development, most cells would become hypoxic, serve as an FGF signaling center, and receive an appropriate tracheal supply. Thus, the ultimate pattern of tracheal branches would reflect the complex history of the oxygen needs and Bnl expression pattern of the tissues (Jarecki, 1999).

Drosophila glypican Dally-like acts in FGF-receiving cells to modulate FGF signaling during tracheal morphogenesis

Studies in Drosophila have shown that heparan sulfate proteoglycans (HSPGs) are involved in both breathless (btl)- and heartless (htl)-mediated FGF signaling during embryogenesis. However, the mechanism(s) by which HSPGs control Btl and Htl signaling is unknown. This study shows that dally-like (dlp, a Drosophila glypican) mutant embryos exhibit severe defects in tracheal morphogenesis and show a reduction in btl-mediated FGF signaling activity. However, htl-dependent mesodermal cell migration is not affected in dlp mutant embryos. Furthermore, expression of Dlp, but not other Drosophila HSPGs, can restore effectively the tracheal morphogenesis in dlp embryos. Rescue experiments in dlp embryos demonstrate that Dlp functions only in Bnl/FGF receiving cells in a cell-autonomous manner, but is not essential for Bnl/FGF expression cells. To further dissect the mechanism(s) of Dlp in Btl signaling, the role of Dlp was analyzed in Btl-mediated air sac tracheoblast formation in wing discs. Mosaic analysis experiments show that removal of HSPG activity in FGF-producing or other surrounding cells does not affect tracheoblasts migration, while HSPG mutant tracheoblast cells fail to receive FGF signaling. Together, these results argue strongly that HSPGs regulate Btl signaling exclusively in FGF-receiving cells as co-receptors, but are not essential for the secretion and distribution of the FGF ligand. This mechanism is distinct from HSPG functions in morphogen distribution, and is likely a general paradigm for HSPG functions in FGF signaling in Drosophila (Yan, 2007).

There are three main important findings in this work. First, Dlp was identified as an essential molecule required for tracheal development. Dlp is required for Btl-mediated tracheal branching during embryogenesis while both Dlp and Dally are involved in the formation of air sac tracheoblasts in the wing disc. Second, the data show that other HSPGs cannot replace Dlp for Btl signaling during embryogenesis and that both Dlp and Dally are not essential for Htl-mediated mesodermal cell migration. These data demonstrate that different FGFs may require different HSPGs to execute their effective signaling activities during development. Third and most importantly, strong evidence is provided that Dlp controls Btl signaling only in FGF-receiving cells in both embryonic and larval tracheal systems. This mechanism of HSPG activity in FGF signaling is very different from its roles in regulating the signaling activities of morphogens including Wnt, Hh and Dpp. Together, these new findings further define novel mechanisms and the specificities of HSPGs in FGF signaling during development (Yan, 2007).

Extensive biochemical and cell culture studies suggest that HSPGs are the part of the FGF/FGFR signaling complex. However, the mechanisms of HSPGs in FGF signaling during development are less known. Embryos mutant for two HSPG biosynthesis enzymes, sgl and sfl, exhibit defects in both Btl- and Htl-mediated FGF signaling. An important issue remaining to be solved is which HSPG core proteins are involved in these signaling events. The data in this work provide strong evidence that Dlp is the key molecule required for Btl signaling during embryonic tracheal development, while both Dlp and Dally are involved in the Btl mediated air sac tracheoblasts formation in the wing disc. The results provide several novel insights into the specificity of individual HSPG in FGF signaling. First, Dlp is involved in Btl signaling, but not in Htl signaling. These findings indicate that different FGF/FGFR complexes may require different HSPGs for their signaling activities. Second, Dlp is highly active and specific for Btl signaling; overexpression of the other three Drosophila HSPGs fail to rescue tracheal defects in dlp embryos. The specific activity of Dlp in Btl signaling could be due to the Dlp protein core or the HS GAG chains attached to the Dlp core protein. In this regard, it is especially surprising that Dally, which has 22% identity with Dlp and also bears a GPI anchor, cannot rescue tracheal phenotypes associated with dlp embryos. As Dlp is involved in several other signaling pathways such as Hh, it is unlikely that Dlp core protein interacts with the ligands directly. In this regard, it is worthwhile to note that ectopic expression of Dally also fails to rescue Hh signaling in dlp embryos. It is proposed that Dlp may have unique HS GAG chains that might provide high and specific activity for ligands such as Bnl and Hh (Yan, 2007).

The biosynthesis of HS GAG chains is determined by the HSPG protein core in which the GAG attachment sites and other protein parts such as the N-terminal cystenine-rich domain control both quantity and quality of the attached GAG chains. Detailed structure and functional studies of Dlp will further help to define specific requirements of the core protein or GAG attachment sites in FGF signaling. Furthermore, the unique GAG chains may be modified by specific enzymes. In this regard, it is particularly important to note that 6-O sulfation of HS is critical for Btl signaling, as Drosophila heparan sulfate 6-O-sulfotransferase is specifically expressed in embryonic tracheal system and is required for Btl signaling during embryogenesis. Recent study has shown that the overall sulfation level is more important than strictly defined HS fine structures for FGF signaling in some developmental contexts. In this regard, it is suggested that Dlp may be the optimal substrate for sulfation enzymes during embryogenesis. Therefore, the activity of Dlp in FGF signaling during embryogenesis cannot be replaced by other HSPGs including Dally, Syndecan and Perlecan (Yan, 2007).

Although Dlp is essential for Btl signaling during embryogenesis, both Dally and Dlp are involved in Btl signaling in air sac tracheoblast cells. Similarly, previous studies have shown that both Dally and Dlp are involved in regulating Wg, Hh and Dpp distribution in the wing disc. The different functions of the same HSPG in embryos and discs may reflect temporal and developmental stage dependent regulation of HSPG functions (Yan, 2007).

While it is well established that HSPGs can regulate FGF signaling by facilitating FGF/FGFR interaction, it is unknown whether HSPGs can also control FGF distribution, thereby modulating FGF signaling. This is a particularly important issue as in many developmental contexts FGF ligand is produced in one type of cell and acts on other cells to initiate its biological activity. One important finding of this work is that HSPGs control tracheal morphogenesis by regulating FGF signaling only in FGF-receiving cells, but not by regulating the secretion or distribution of FGF ligand in its producing cells and surrounding cells. Several important results support these conclusion: (1) dlp mutant embryos can suppress the phenotype of overexpressing Bnl in the tracheal cells. (2) Ectopic expression of Dlp in tracheal cells, rather than FGF expression cells, can effectively restore tracheal defects associated with dlp embryos. (3) Embryos rescued by prd-Gal4/UAS-dlp in dlp backbround is very similar to btl mutant embryos rescued by prd-Gal4/UAS-btl-GFP. (4) HSPGs are required for FGF signaling in its receiving cells in the air sac, but are dispensable in the columnar epithelial layer which includes FGF producing cells and other surrounding cells. Detailed analyses thus demonstrate the specific and distinct requirement of HSPGs in FGF signaling during tracheal development. Moreover, embryonic and larval data together suggest this is likely a general mechanism for HSPG function in FGF signaling in Drosophila (Yan, 2007).

Two major models are proposed for the role of HSPGs in FGF signaling. In one model, low affinity HS/GAG chains on the cell surface limit the diffusion of FGF ligand, thereby increasing its local concentration and the probability that it will interact with high-affinity FGFRs. In the second model, HSPGs facilitate the dimerization or oligomerization of FGF ligands thereby inducing receptor clustering and signal transduction. The experimental data cannot exclude either of these mechanisms. However, the results are in favour of the second case, since it is shown that HSPGs are not required in FGF concentration gradient in FGF producing cells, but are essential in FGF-receiving cells. Finally, a recent study showed that dynamin-mediated vesicle internalization is a crucial step to regulate FGF signaling in Drosophila tracheal system. Mutants in awd (abnormal wing disc) or shi (shibire), which encodes for a nucleoside diphosphate kinase and Drosophila dynamin, respectively, have increased levels of Btl in tracheal cell surface, increased FGF signaling activity and ectopic tracheal branching. In this regard, HSPGs may control FGF signaling by stabilizing the FGF/FGFR complex from degradation or internalization in FGF receiving cells. Further experiments using HSPG and awd/shi double mutant are needed to test this possibility (Yan, 2007).

Over the past several years, extensive studies in Drosophila and other model systems have established the essential roles of HSPGs in developmental signaling pathways including Wg, Hh and Dpp. In Drosophila embryo and wing imaginal disc, HSPGs are involved in the transport of morphogens including Wg, Hh and Dpp by a restricted diffusion mechanism. Narrow stripes of clones mutant for HSPGs can impede the movement of morphogens to further cells. However, in all of these cases, the first mutant cells adjacent to the morphogen source can still transduce signals arguing that HSPGs are not essential for morphogen signaling activity, but rather control the distributions or local concentrations of morphogens. The novel results from this work point out a major difference for a role of HSPGs in FGF signaling from their roles in morphogen signaling, as removal of HSPGs (dally-dlp or sfl) from FGF receiving cells can effectively block FGF signaling. Although the graded FGF activity may play an essential role in tracheal morphogenesis, the data from this work argue that the main function of HSPGs in FGF signaling is not to regulate the distribution of FGF ligand. Consistent with the different roles of HSPGs in FGF and morphogen signaling, it was found that Dlp acts cell-autonomously in FGF signaling while it functions non-autonomously in Hh signaling in embryos. These results suggest that Bnl transportation may be different from morphogen movement in the epithelial cells of the wing pouch. Indeed, morphogen molecules diffuse through the same layer of cells, columnar epithelial cells, while FGF is transported between different layers of tissues, from columnar epithelia to tracheoblasts. Moreover, leading air sac cells are always in close proximity with underlying columnar epithelia. They also extend multiple filopodia toward ligand gradient and presumably actively pursue the FGF ligands while wing disc morphogens including Wg, Hh and Dpp need to transport many cell diameters from their sources to reach their receiving cells. Studies in vertebrate also suggest that a graded distribution of FGF8 protein can be generated by the decay of fgf8 mRNA and this RNA gradient is translated into a protein gradient. In this case, no active transport mechanism is required to form a FGF gradient. In mammalian limb and lung development different FGFs are often expressed in different layers of cells, such as epithelium and mesenchyme, and signal through each other. It is interesting to determine whether HSPGs function similarly in these systems as in Drosophila (Yan, 2007).

GENE STRUCTURE

Genomic length - 28 kb

Bases in 5' UTR - 216

Exons - 3

Bases in 3' UTR - 213

PROTEIN STRUCTURE

Amino Acids - 770

Structural Domains

Like most FGFs, the BNL protein contains a signal peptide at the N-terminus, suggesting that it is secreted. BNL is unusual in that it contains large domains flanking the FGF domain that are not found in the mammalian proteins (Sutherland, 1996).

A 99 residue segment of BNL (residues 260-358) is 30-40% identical to human FGFs 1, 2, and 9 as well as other vertebrate FGFs. Within this region, 17 of the 23 residues that are highly conserved among vertebrate FGFs are also conserved with BNL. Two introns in the BNL FGF domain are in the same positions as the introns in mammalian FGF genes (Sutherland, 1996).

The three-dimensional structures of two members of the FGF family, bovine acidic FGF and human basic FGF, have been crystallographically determined. These structures contain 12 antiparallel beta strands organized into a folding pattern with approximate threefold internal symmetry. Topologically equivalent folds have been previously observed for soybean trypsin inhibitor and interleukins-1 beta and -1 alpha. Beta-sheet strand 10, implicated in receptor and heparin binding by FGF, is adjacent to the site of an extended sequence insertion in several oncogene proteins of the FGF family (Xhu, 1991).

date revised: 2 Jan 97  

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

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