Echinoid regulates tracheal morphology and fusion cell fate in Drosophila


  • Caroline Laplante,

    1. Department of Biology, McGill University, Montréal, QC, Canada
    Current affiliation:
    1. Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, CT, 06511
    Search for more papers by this author
  • Sarah M. Paul,

    1. Department of Biochemistry, Molecular Biology and Cell Biology, Northwestern University, Evanston, Illinois
    Current affiliation:
    1. Department of Nutritional Science and Toxicology, University of California, Berkeley, Berkeley, CA 94720
    Search for more papers by this author
  • Greg J. Beitel,

    Corresponding author
    1. Department of Biochemistry, Molecular Biology and Cell Biology, Northwestern University, Evanston, Illinois
    • Dept. of Biochemistry, Molecular Biology and Cell Biology Hogan Hall, Rm. 2-100, 2205 Tech Drive, Northwestern University, Evanston, IL 60208-3500
    Search for more papers by this author
  • Laura A. Nilson

    Corresponding author
    1. Department of Biology, McGill University, Montréal, QC, Canada
    • Department of Biology, McGill University, Montréal, QC, H3A 1B1, Canada
    Search for more papers by this author


Morphogenesis of the Drosophila embryonic trachea involves a stereotyped pattern of epithelial tube branching and fusion. Here, we report unexpected phenotypes resulting from maternal and zygotic (M/Z) loss of the homophilic cell adhesion molecule Echinoid (Ed), as well as the subcellular localization of Ed in the trachea. edM/Z embryos have convoluted trachea reminiscent of septate junction (SJ) and luminal matrix mutants. However, Ed does not localize to SJs, and edM/Z embryos have intact SJs and show normal luminal accumulation of the matrix-modifying protein Vermiform. Surprisingly, tracheal length is not increased in edM/Z mutants, but a previously undescribed combination of reduced intersegmental spacing and deep epidermal grooves produces a convoluted tracheal phenotype. In addition, edM/Z mutants have unique fusion defects involving supernumerary fusion cells, ectopic fusion events and atypical branch breaks. Tracheal-specific expression of Ed rescues these fusion defects, indicating that Ed acts in trachea to control fusion cell fate. Developmental Dynamics 239:2509–2519, 2010. © 2010 Wiley-Liss, Inc.


Multicellular animals depend on networks of epithelial tubes to perform the essential task of transporting fluids and gases throughout their bodies. Efficient transport requires creating tubes with dimensions that are matched to the desired flow, and defects in tube-size control can lead to devastating diseases such as polycystic kidney disease (Boletta and Germino,2003). Efficient transport can also depend on the successful interconnection of tubes to create anastomosing networks or openings to the outside environment. Missing or ectopic interconnections can disrupt flow and cause life-threatening conditions such as aortic shunts. The central roles of epithelial tube networks to proper body functions underline the importance of understanding the molecular mechanisms that mediate tube-size control and branch interconnections.

The Drosophila tracheal system is one of the best characterized systems for studying morphogenesis of tubular networks, and serves as a combined pulmonary and vascular system to deliver oxygen to target tissues (reviewed by Uv et al.,2003; Affolter and Caussinus,2008). The tracheal system starts as 10 segmentally repeated clusters of ∼40 cells each on either side of the embryo. These cells invaginate, form sacs, and undergo a stereotyped pattern of branching. The major branches connect between segmental repeats, across the dorsal midline and, for the anterior three branches, across the ventral midline, in a process referred to as branch “fusion,” to create the final tracheal network (Samakovlis et al.,1996b). Finally, tubes expand their diameter and elongate to create a tubular network with characteristic dimensions (reviewed by Affolter and Caussinus,2008).

The fusion process is mediated by specialized cells, named fusion cells, located at the tip of each tracheal branch. The specification of a single fusion cell per branch tip involves a complex interplay of Wingless (Wg)/Wnt, fibroblast growth factor (FGF), Decapentaplegic (Dpp), and Notch signaling (Ikeya and Hayashi,1999; Steneberg et al.,1999; Chihara and Hayashi,2000; Llimargas,2000). The fusion cells extend actin-rich filopodia that lead migration in response to guidance cues and then recognize and adhere to each other (Tanaka et al.,2004). A critical element of the fusion process is the formation of adherens junctions between two fusion cells in a dynamic process that requires DE-cadherin, Armadillo (Arm)/β-catenin, and Polychaetoid (Tanaka–Matakatsu et al.,1996; Jung et al.,2006).

The length and diameter of the tracheal tubes are controlled in part by the septate junctions, which are invertebrate cell–cell junctions that function as diffusion barriers, analogous to vertebrate tight junctions (reviewed in Wu and Beitel,2004). However, septate junctions are located basal to adherens junctions, while tight junctions are apical of adherens junctions, and each contains distinct protein components. Embryos homozygous for mutations in septate junction components have overly long tubes causing them to adopt a convoluted appearance. One way septate junctions regulate tracheal tube length and diameter is by contributing to formation of a temporary luminal extracellular matrix. Organization of this matrix requires the secretion of Vermiform (Verm), a putative matrix-modifying protein (Luschnig et al.,2006; Wang et al.,2006), into the tracheal lumen, which depends on the septate junctions (Wang et al.,2006). The transient luminal matrix restricts tube elongation by an unknown mechanism (reviewed by Wu and Beitel,2004; Affolter and Caussinus,2008). Septate junctions have also recently been shown to regulate tracheal tube length through additional pathways involving apical/basal polarity genes (Laprise et al.,2009).

In this report, we demonstrate a role for the homophilic cell adhesion protein Echinoid (Ed) in tracheal development. Ed is an immunoglobulin-domain-containing cell adhesion molecule that facilitates Notch signaling (Ahmed et al.,2003; Escudero et al.,2003; Rawlins et al.,2003a) and antagonizes epidermal growth factor receptor (EGFR) signaling (Bai et al.,2001; Rawlins et al.,2003b; Spencer and Cagan,2003), and has essential roles in assembly of actomyosin structures during epithelial development (Wei et al.,2005; Laplante and Nilson,2006; Lin et al.,2007). Here we report defects in tracheal morphology and fusion in embryos lacking maternal and zygotic Ed (edM/Z embryos). edM/Z embryos exhibit a convoluted tracheal phenotype that is typically associated with septate junction or luminal matrix defects. Unexpectedly, however, we found that such embryos have intact septate junctions (SJs) and exhibit normal accumulation of Vermiform, a matrix-modifying protein, in the tracheal lumen and that Ed does not localize to SJs. Moreover, although a convoluted phenotype is typically associated with longer trachea, tracheal length is not increased in edM/Z mutants; instead, a combination of reduced intersegmental spacing and deep epidermal grooves appears to contribute to the convolution of the trachea. We also show that edM/Z mutants have unique fusion defects and exhibit supernumerary fusion cells, ectopic fusion events, and atypical branch breaks. These fusion defects can be rescued by tracheal-specific expression of Ed, indicating that Ed acts in trachea to control fusion cell fate.


Ed Is Uniformly Expressed in Trachea Cells

Ed plays crucial roles during epithelial morphogenesis, including dorsal closure of the embryonic epidermis and formation of the eggshell dorsal appendages (Laplante and Nilson,2006; Lin et al.,2007). Because several genes are required for both dorsal closure and tracheal morphogenesis (e.g., mummy and coracle; Fehon et al.,1994; Lamb et al.,1998; Behr et al.,2003; Schimmelpfeng et al.,2006; Caussinus et al.,2008; Laprise et al.,2009) and because Ed is a known binding partner of Neuroglian (Nrg), a septate junction component essential for trachea morphogenesis (Genova and Fehon,2003; Islam et al.,2003), we investigated the role of Ed in tracheal development. We detected Ed in the tracheal cells from stage 10, before invagination of the tracheal placodes, through the end of embryogenesis (Fig. 1A–C and S1A,A′; Supp. Fig. S1A, which is available online). Ed is detectable in the fusion cells before and during fusion (Supp. Fig. S1 and Fig. 1D,D′; fusion cells are marked by nuclear Dysfusion (Dys) staining) and after fusion occurs (Fig. 1E; asterisk marks the fusion cells). Ed expression levels among the tracheal cells, including the fusion cells, appear uniform during invagination, branching, and fusion of the tracheal repeats into a continuous tubular network (Fig. 1A–E). Unlike Arm, Ed expression is not apparently increased in the fusion cells compared with the other cells of the tubular network (Fig. 1D and Supp. Fig. S1C,C′; compare Arm and Ed).

Figure 1.

edM/Z embryos exhibit convoluted trachea and aberrant branch fusion. A–C: Echinoid (Ed) is expressed in the trachea throughout embryogenesis. Wild-type embryos stained for Ed at stage 10 (A, dotted line indicates tracheal placode), stage 11 (B), and stage 16 (C) trachea. D,D′: Wild-type stage 14 embryo stained for Ed (white; D) and Dys (red; merge in D′), a nuclear fusion cell marker. Ed is detectable at the membranes of the fusion cells soon after they contact (arrow in D points to Ed at the fusion cell membranes at the recent fusion site). E: Wild-type stage 16 embryo stained for Ed and Armadillo (Arm). Ed and Arm are expressed in the fusion cells (asterisk marks the fusion cells). F,G: Wild-type (F) and edM/Z(G) stage 16 embryos stained for the tracheal luminal marker 2A12. edM/Z mutant embryos exhibit a convoluted trachea. Large arrows point to the dorsal trunk in both F and G. Small arrow in G points to absent dorsal anastomoses. H–H″: Dorsal trunk of a wild-type stage 16 embryo stained for Arm (H) and wheat germ agglutinin (WGA; H′), merged channels shown in H″(Arm, white; WGA, green). The fusion cells appear as tightly packed triple ring structures (arrows). I: Dorsal trunk of a stage 16 edM/Z embryo stained for Arm. The fusion cells (arrow) appear disorganized and the fusion site thicker. J–J″: Dorsal trunk of a stage 16 edM/Z embryo stained for Arm (J) and WGA (J′), merged channels shown in J″(Arm, white; WGA, green). The embryo exhibits a break in the dorsal trunk (asterisk). The triple ring fusion cell structure is visible at one end of the break (arrow). K: Stage 16 edM/Z embryo stained for WGA exhibits a break in the lateral trunk (arrow). L–L″: Dorsal trunk of a stage 16 edM/Z embryo stained for Arm (L) and WGA (L′), merged channels shown in L″(Arm, white; WGA, green). An ectopic branch emerges from a fusion site along the dorsal trunk (arrowhead) and fuses with an adjacent transverse connective (arrow).

edM/Z Embryos Exhibit Aberrant Tracheal Morphology and Branch Fusion Defects

Zygotic mutant edF72 or edlF20 embryos exhibit no obvious trachea defects (data not shown), presumably due to the maternal contribution of ed. Embryos lacking both maternal and zygotic ed contributions (edM/Z; see the Experimental Procedures section) exhibit highly convoluted tracheal tubes compared with those of wild-type embryos (Fig. 1F,G). In addition to this convoluted morphology, the tracheal networks of edM/Z embryos also exhibit abnormal branch fusion. In wild-type embryos the tracheal repeats fuse into a seamless tubular network with fusion cells adopting the characteristic triple ring structure at fusion points (Fig. 1F,H–H″), but edM/Z embryos frequently exhibit breaks along the dorsal and lateral trunks (Fig. 1J,K) and in dorsal anastomoses (Fig. 1G). Furthermore, the fusion cell triple ring structure appears thicker and less organized in edM/Z embryos (Fig. 1I, compare with H). We observed that 56% (n = 55 embryos) of the embryos exhibit dorsal and/or lateral trunk breaks and/or missing dorsal branch anastomoses, suggesting defects in the fusion process that interconnects tracheal branches into a continuous tubular network (see section on fusion cells below for more detail). Ectopic branch fusions were also observed in 24% of embryos (n = 25; Fig. 1L–L″). These phenotypes were characterized in edF72 and were also observed in embryos maternally and zygotically mutant for edlF20, an independently isolated allele of ed (data not shown). Thus, lack of Ed causes tracheal morphology and fusion defects. Importantly, the combination of convoluted tubes and aberrant fusions present in edM/Z mutants has not been so far reported in mutants that affect trachea development, suggesting that Ed plays unique roles in tracheal morphogenesis.

Septate Junctions Are Functional in edM/Z Embryos

Embryos mutant for nrg, which encodes an adhesion molecule and essential septate junction component, have convoluted trachea that resemble those of edM/Z embryos (Genova and Fehon,2003). Although Ed has been reported to colocalize with adherens junctions in follicle, epidermal, and tracheal cells (Escudero et al.,2003; Wei et al.,2005; Laplante and Nilson,2006), Ed has also been shown to bind to Nrg (Genova and Fehon,2003; Islam et al.,2003). We therefore investigated the subcellular localization of Ed in the trachea and determined whether septate junctions are affected in edM/Z embryos. Immunostaining for Ed, the septate junction markers Nrg, Discs Large (Dlg), and Coracle (Cora), and the adherens junction marker Armadillo (Arm), revealed complete colocalization between Ed and the adherens junction marker Arm (Fig. 2A–A″), but no apparent colocalization between Ed and septate junction markers (data not shown, Fig. 2C–C″,E–E″). Thus Ed appears to be excluded from the septate junctions in the trachea. Moreover, the localization of Nrg, Dlg, and Cora appears normal in edM/Z embryos, suggesting that there are no apparent defects in septate junction organization or composition (Fig. 2D,F, and data not shown).

Figure 2.

Septate junctions are functional in edM/Z mutant embryos. A–A″: Wild-type embryo stained for Echinoid (Ed; A) and Armadillo (Arm; A′), merge shown in A″(Ed, green; Arm, red). Ed and Arm appear to colocalize at the adherens junctions of trachea cells. Arrows point to fusion cells. B:edM/Z embryo stained for Arm. Fusion cell triple ring structures (arrows) appear disorganized. C–C″: Wild-type embryo stained for Ed (C) and Coracle (Cora; C′), merge shown in C″(Ed, green; Cora, red). D:edM/Z embryo stained for Cora. E–E″: Wild-type embryo stained for Ed (E) and Discs Large (Dlg; E′), merge shown in E″(Ed, green; Dlg, red). F:edM/Z embryo stained for Dlg. G–G″: Wild-type embryos stained for Vermiform (Verm). Verm is detectable in the cytoplasm of stage 14 trachea cells (G), it is secreted into the lumen of the trachea at stage 15 (G′) and its secretion is complete at stage 16 (G″). H–H″:edM/Z embryo stained for Verm. Verm is localized normally at all stages. I,I′:edM/Z embryo injected with Texas-Red conjugated 10 kDa dextran dye. The dye is excluded from the trachea (dashed lines outline the lumen). J,K: Stage 16 wild-type (J) and edM/Z(K) embryos stained with WGA to highlight the luminal chitin bundle.

To test the functionality of septate junctions in edM/Z embryos, we performed a standard dye diffusion assay (Lamb et al.,1998; Paul et al.,2003). In wild-type embryos, a 10 kDa fluorescent dextran dye injected into the hemocoel is excluded from the trachea by the diffusion barrier function of the septate junctions (Paul et al.,2003). However, in embryos mutant for septate junction components, the diffusion barrier is compromised and the dye readily diffuses into the tracheal lumen (Paul et al.,2003). Dye injected into edM/Z embryos was successfully excluded from the lumen of the trachea (Fig. 2I,I′; n = 6). Consistent with the above immunostaining data, the septate junctions of edM/Z embryos are functional and fulfill their role as diffusion barriers.

In addition to this well-known barrier function, septate junctions have the poorly understood role of mediating the secretion of Verm, a putative matrix-modifying enzyme, into the tracheal lumen (Wang et al.,2006). This function is critical as failure to produce or secrete Verm causes excessive trachea elongation and consequently a convoluted tracheal morphology (Luschnig et al.,2006; Wang et al.,2006). The gradual luminal accumulation of Verm in edM/Z embryos is indistinguishable from that of wild-type embryos suggesting that septate junctions are functional with respect to Verm secretion (Fig. 2G–H″). Embryos mutant for wurst and convoluted also exhibit convoluted tube phenotypes despite normal Verm secretion and luminal accumulation, due to defective organization of the luminal matrix (Behr et al.,2007; Swanson et al.,2009). Confocal analysis using wheat germ agglutinin (WGA) staining to highlight the luminal chitin bundle showed no obvious defects in luminal chitin organization in edM/Z embryos (Fig. 2J,K). Thus, in contrast to other characterized mutations that cause a convoluted tracheal phenotype, the lack of Ed does not disrupt trachea septate junction function or result in an obvious disorganization of the luminal matrix, suggesting that Ed causes convolution of the trachea by a distinct and as yet undescribed mechanism.

Tracheal Length Is Independent of Ed Expression

In embryos mutant for septate junction components, the convolution of the dorsal trunk occurs because the length of the dorsal trunk is increased relative to that of the embryo, which remains the same (Beitel and Krasnow,2000; Paul et al.,2003). We therefore tested whether the convolution of ed mutant trachea resulted from an increase in tube length. We measured the path length of the dorsal trunk of stage 16 embryos between the transverse connectives of tracheal segments 4 and 6 (denoted “Path4-6”; Supp. Fig. S2). Surprisingly, despite the convolutions of the trachea, there was no significant difference between the lengths of Path4-6 in edM/Z and wild-type embryos (74.3 ± 2.5μm vs. 79.7 ± 2.4μm, respectively, length of path ± standard error of the mean (SEM), P= 0.15), indicating that the convoluted appearance of the edM/Z embryonic trachea is unlikely to be caused by increased tube length (Fig. 3A).

Figure 3.

Intersegmental distance and epidermal segment groove morphogenesis influences the morphology of the underlying trachea. A: Graph of mean Path4-6± SEM for wild-type, edM/Z(ed) and edM/Z; UAS-Echinoid (Ed) embryos (ed; UAS-Ed). This measurement follows the path of the tracheal tube. B: Graph of mean Dist4-6± SEM for wild-type, edM/Z(ed) and edM/Z; UAS-Ed embryos (ed; UAS-Ed)(**P < 0.001). This measurement describes the straight-line distance between embryo segments. C:edM/Z embryo (dorsal view) expressing the UAS-Ed construct in the trachea under the control of btl-GAL4. Specific expression of UAS-Ed in the trachea of edM/Z embryos does not rescue the convoluted phenotype (arrows point to the dorsal trunks). D–D″:edM/Z embryo stained for wheat germ agglutinin (WGA) seen at three different focal depths (D, surface view of the epidermis; D′, underlying trachea; D″, deeper view than D′). The surface view of the embryo shows the deep intersegmental grooves of the epidermis (D; arrows mark deep grooves). The bending of the underlying dorsal trunk can correlate with the grooves pushing inward (D′ and D″; arrows point to bends in the dorsal trunk coinciding with the deep epidermal grooves). E–E″: Embryo expressing DiaCA and β-Gal in paired expression pattern stained for β-Gal (E) and WGA (E′), merge shown in E″(β-Gal, white; WGA, green; arrow points to dorsal trunk). F–F″: Embryo expressing DiaCA and β-Gal in engrailed expression pattern stained for β-Gal (F) and WGA (F′), merge shown in F″(β-Gal, white; WGA, green; arrow points to dorsal trunk).

To test directly whether Ed was required in the trachea to control tracheal morphology, we generated edM/Z embryos in which UAS-Ed was specifically expressed in the trachea under the control of the breathless(btl)-GAL4 driver (henceforth designated as edM/Z; btl>Ed embryos). In these embryos, Ed was expressed in the trachea and amnioserosa, but was not detected in other epithelial tissues (Fig. 3C and data not shown). The transgenic Ed was detectable at the membrane at levels comparable to that of endogenous Ed (Fig. 3C), but did not rescue the aberrant convoluted morphology phenotype of the trachea (Fig. 3C, arrows) and did not alter the length of the Path4-6 segment compared with wild-type (Fig. 3A; 80.0 μm ± 2.7 μm, n = 5; P= 0.92 compared with wild-type length). Thus, Ed regulates tracheal morphology nonautonomously and the convoluted appearance of the trachea in edM/Z mutants results from the lack of Ed from tissues other than the trachea.

Deep Epidermal Segment Grooves and Short Intersegmental Distances Deform the Trachea in edM/Z Embryos

We investigated how a requirement for Ed in other tissues could impact the appearance of the trachea in edM/Z embryos. First, we observed that edM/Z embryos exhibit defects in the process of head involution, in which the epidermis is drawn toward the anterior to cover head segments as they move into the interior of the embryo (Laplante and Nilson,2006; Lin et al.,2007). We reasoned that a failure of head involution could reduce intersegmental spacing, because the epidermis would not be stretched over the anterior structures, and that a reduction in intersegmental spacing could contribute to a convoluted appearance of the trachea because the same length of tracheal tube would be confined to a shorter distance. In wild-type embryos the straight-line distance between the transverse connective of tracheal segment 4 to six along the dorsal trunk (Dist4-6; Supp. Fig. S2) was 71.7 μm ± 2.4 μm (n = 9 embryos; Fig. 3B). We measured the Dist4-6 of edM/Z embryos to be 60.8 μm ± 1.7 μm (n = 6 embryos; Fig. 3B), which is significantly shorter than wild-type (P < 0.001 edM/Z vs. wild-type; Fig. 3B). Thus, in edM/Z and edM/Z; btl>Ed embryos, a dorsal trunk of wild-type length is confined to a shorter distance, accounting for its convoluted path. The Dist4-6 of edM/Z; btl>Ed embryos was 64.5 μm ± 5.2 μm, which is similar to edM/Z embryos and also significantly shorter than wild-type (P < 0.001; Fig. 3B), indicating that expression of Ed in the trachea is not sufficient to rescue this defect.

We also noted that edM/Z embryos have abnormally deep epidermal grooves, which appeared to distort the underlying tracheal network and thus could contribute to convolution of tracheal tubes. (Fig. 3D–D″; Lin et al.,2007). To test experimentally whether deep segment grooves can affect the morphology of the underlying trachea, we expressed a constitutively active version of the formin Diaphanous (DiaCA) in the striped expression profile of the paired (prd) or engrailed (en) genes, resulting in the formation of deep epidermal grooves (Fig. 3E–F″, and Homem and Peifer,2008). In such embryos, the epidermis appears to impinge upon the tracheal dorsal trunk, and the underlying trachea is more convoluted than wild-type (Fig. 3E′,F′, arrows; 3/3 stage 16 embryos for each driver; compare with Fig. 1F). The phenotype is more severe with the prd-GAL4 expression driver than en-GAL4, perhaps due to wider expression within each stripe and deeper grooves (compare Fig. 3E–E″ with F–F″). Although the grooves do not always align precisely with the bends in the trachea, this observation is consistent with the hypothesis that deep grooves can contribute to the convoluted phenotype. Based on these observations, we propose that the reduced intersegmental spacing in the epidermis of edM/Z embryos, together with their abnormally deep segmental grooves can explain why the trachea of edM/Z mutant embryos appear convoluted despite their wild-type length.

Ed Restricts the Number of Fusion Cells per Branch Tip

Embryos lacking Ed exhibit interruption in their dorsal and lateral tubes suggesting branch fusion failures. Quantification of this phenotype showed that 30% of fusion sites exhibited breaks in the lateral trunk (n = 90). As with characterized fusion mutants such as escargot or fear of intimacy (Samakovlis et al.,1996b; Tanaka–Matakatsu et al.,1996; Van Doren et al.,2003), we also observed breaks near fusion sites in the dorsal trunk, but at a lower frequency (3.1% of dorsal trunk fusion sites [n = 522]). Expression of Ed in the trachea of edM/Z embryos rescues both the lateral and dorsal trunk defects, reducing lateral trunk breaks from 30% to 3.7% (n = 90 and 27, respectively) and dorsal trunk breaks from 3.1% to 0% (n = 522 and 90, respectively). Ed expression in the trachea is, therefore, essential for proper fusion of tracheal repeats.

Despite the colocalization of Ed with adherens junctions, Ed is not required for adherens junction formation in fusion cells because adjacent pairs of fusion cells in edM/Z embryos created structures similar to the characteristic “triple rings” of adherens junctions at the dorsal trunk fusion points (Fig. 1I, arrows). Surprisingly, and in contrast to known fusion mutants, triple rings were also present at sites adjacent to breaks (Fig. 1J–J″ arrows). Furthermore, in many of the ectopic fusions there appeared to be increased Arm labeling at the fusion sites compared with wild-type (see Fig. 1I, arrows, and 2B arrows). Although this phenotype could be caused by an increase in Arm expression levels, we asked whether it might reflect the presence of an increased number of fusion cells. To test for fusion cell specification defects, we stained edM/Z embryos for the marker Dys, a basic helix–loop–helix (bH–L–H) transcription factor specifically expressed in tracheal fusion cells (Jiang and Crews,2003,2006). Wild-type embryos have one Dys-expressing cell per branch tip resulting in two Dys-expressing cells per fusion site after the branch tips have fused (Fig. 4A,C–C″; average number of Dys-expressing cells per fusion site = 2.03 ± 0.02, n = 162). In contrast, edM/Z embryos have an average of three Dys-expressing cells per fusion site (Fig. 4A,D–D″, average number of Dys-expressing cells per fusion site = 3.05 ± 0.14; n = 209; P < 0.00001 wild-type vs. edM/Z). More specifically, in wild-type embryos, 95% of fusion sites contained two fusion cells, 4% contained three fusion cells and 1% contained a single fusion cell (n = 162 fusion sites; Fig. 4B). In edM/Z embryos, only 34% of fusion sites contained two fusion cells per fusion site, 37% contained three, 28% of fusion sites contained between four and seven fusion cells, and only 1% contained a single fusion cell (n = 209 fusion site; Fig. 4B). Expression of UAS-Ed in the trachea of edM/Z embryos rescued the number of Dys-expressing cells to two per fusion site (Fig. 4A,E-E′; rescued edM/Z average number of Dys-expressing cells = 2.21 ± 0.06; n = 81) and restored a wild-type distribution of fusion cell number (Fig. 4B). Ed is, therefore, a critical negative regulator of fusion cell fate in the developing trachea. Furthermore, our data show that Ed acts autonomously in the trachea to control fusion cell fate.

Figure 4.

Ed restricts the number of fusion cells per branch tip. A: Graph of mean number of Dys-expressing cells per fusion site for each genotype tested (**P < 0.00001). B: Table of distribution of number of Dys-expressing cells per fusion site for each genotype tested (wild-type, blue; edM/Z, red; edM/Z;btl>Ed, yellow). C–C″: Wild-type embryo stained for Echinoid (Ed; C) and Dys (C′). Merged images in C″(Ed, red; Dys, white). D–D″:edM/Z embryo stained for wheat germ agglutinin (WGA; D) and Dys (D′). Merged images in D″(WGA, green; Dys, white). E–E″:edM/Z embryo expressing the UAS-Ed construct in the trachea stained for Ed (E) and Dys (E′). Merged images in E″(Ed, red; Dys, white).

Finding both fusion defects and supernumerary fusion cells in edM/Z mutants was unexpected, as fusion breaks are more typically associated with a loss of fusion cell fate. We therefore considered the possibility that the deep segmental grooves sterically block outgrowing tracheal branches. As noted above, expressing ed specially in the tracheal system rescued the tracheal fusion defects. Furthermore, in embryos with deep segmental grooves caused by expression of a constitutively active form of Dia under the control of en or prd, no LT breaks were observed (0/18 for en>DiaCA and 0/18 for prd>DiaCA) compared with 30% of LT being broken in edM/Z mutants. Similarly, DiaCA-induced deep grooves were only very infrequently associated with DT breaks (0/45 for en>DiaCA, 2/70 for prd>DiaCA). Thus deep segmental grooves are unlikely to cause fusion failures. We also investigated the possibility that despite the increased average number of fusion cells in edM/Z mutants, some branch tips might nonetheless have no fusion cell thus be unable to fuse. Quantification of fusion cell number in fusion branches demonstrates this is not the case; the frequency of fusion sites with one or zero Dys-expressing cells was not increased in edM/Z mutant embryos compared with wild-type embryos, and none of the wild-type or edM/Z embryos observed lacked fusion cells (Fig. 4B; n = 162 and 209, respectively). Thus, lack of Ed results in a significant increase in the number of fusion cells that are specified per branch tip, and tracheal breaks are not caused by a loss of fusion cell fate. Moreover, we noticed that four of six dorsal trunk breaks examined had more than one fusion cell at one of the two ends of the break. This observation suggests that the presence of multiple fusion cells at the tip of a branch can interfere with proper fusion events. Furthermore, because two of the six dorsal trunk breaks had the normal one Dys-expressing cell on each branch, our observations also suggest that Dys-expressing cells in edM/Z mutants do not always correctly execute the fusion cell fate and are unable to mediate branch fusion. Thus, Ed is required for both specifying and executing the fusion cell fate.


Lack of Ed Reveals Roles for Nontracheal Tissues in Regulation of Tracheal Morphology

The convoluted trachea in embryos lacking Ed bear a striking resemblance to those in embryos lacking septate junction components or proteins required for creating or modifying the tracheal luminal matrix that restricts tracheal tube lengthening during late embryogenesis (Beitel and Krasnow,2000; Behr et al.,2003; Paul et al.,2003; Llimargas et al.,2004; Tonning et al.,2005; Luschnig et al.,2006; Moussian et al.,2006). Unexpectedly, however, we did not detect any defects in the septate junctions and extracellular luminal matrix pathways and, in contrast to other known mutations that cause convoluted trachea, we found that the trachea in edM/Z mutants are the same length as in wild-type. Instead, the convoluted tracheal phenotype in edM/Z mutants appears to be a consequence of the requirement for Ed in other tissues and/or processes, because expressing an Ed transgene specifically in the trachea of edM/Z mutants did not rescue the convoluted tube morphology. For example, we observed that Ed is required for head involution, and that the failure of head involution decreases the spacing between epidermal segments during development and thus contributes to the convolution of the trachea. Furthermore, edM/Z mutants have abnormally deep segmental grooves, and we show that such deep grooves can deflect the paths of underlying trachea and, therefore, may also contribute to the convoluted phenotype. Together these effects phenocopy the well-described septate junction and tracheal matrix convoluted tube phenotypes and provide further support for the notion that the convolution of the trachea is based on the link between the length of the trachea and the intersegmental space and, therefore, the overlying epidermis. Tubes become convoluted either when longer tubes accommodate a wild-type intersegmental distance or when tubes of wild-type length are forced to fit in a shorter body length.

Importantly, although multiple papers have recently identified mutations that cause elongated trachea, the actual length of the trachea in some of these papers was not quantified. Our results with ed illustrate the importance of establishing whether trachea that appear long do in fact have increased length.

Ed Regulates Fusion Cell Number and Loss of ed Creates a Unique Fusion Phenotype

Embryos lacking Ed frequently exhibit breaks in their dorsal and lateral trunks, as well as ectopic branch fusion events, two seemingly opposite phenotypes. We show that Ed is expressed in the tracheal cells throughout trachea development and is required to restrict the number of fusion cells per branch tip (Samakovlis et al.,1996b). We observed on average three, and in some cases up to seven, fusion cells per tracheal segment. Ectopic branches associated with extra fusion sites suggest that at least some of the extra fusion cells are competent to migrate, recognize and fuse with other trachea cells. Therefore, the extra fusion cells are capable of executing the functions associated with their determined fusion cell fate even if they are not positioned at the tip of the branch.

If edM/Z embryos do not have a shortage of fusion cells, why do they have branch breaks? Breaks do not appear to result from steric obstruction by epidermal grooves because expressing Ed specifically in the tracheal system restores normal fusion. Furthermore, fusion breaks were not caused by deep epidermal grooves in embryos misexpressing DiaCA in the epidermis. Rather, the breaks are associated with fusion sites, suggesting they are failed fusion events. Failure in branch fusion typically results from lack of fusion cells or the inability to up-regulate the expression of the cell adhesion molecule DE-cadherin in the fusion cell (Tanaka–Matakatsu et al., 1996; Cela and Llimargas,2006). However, absence of fusion cells at branch tips was never observed in edM/Z embryos and all the breaks analyzed had at least one but more often multiple fusion cells at their ends. In addition, enriched levels of Arm, the DE-cadherin binding partner and, therefore, marker of the cell–cell adhesion complex, was always detected in the fusion cells of edM/Z embryos, even when fusion of the branches failed. Thus, edM/Z mutants exhibit a previously unreported defect in branch fusion. How can an excess number of fusion cells result in lack of fusion? One possibility is that extra fusion cells undergo intra-branch fusion; adjacent fusion cells on the same branch adopt characteristic triple-ring shape and up-regulate their level of Arm expression, therefore, “fusing” together. In doing so, they make the tip cell of that branch unavailable for fusion with the neighboring branch and thus cause inter-branch fusion failure. In edM/Z mutants, some of the supernumerary fusion cells maybe sufficiently competent as fusion cells that they can mediate intra-branch fusion events. This would be a previously undocumented mode of branch fusion failure.

Alternatively, perhaps there is there a self vs. nonself recognition system that allows fusion cells to recognize when they have contacted cells from other segments and that normally block fusion cells on the same branch from fusion with each other. An equivalent, but opposite phenomenon occurs during dorsal closure when epidermal cells from one segment selectively adhere to cells of the same segment and minimize contact with epidermal cells of adjacent segments as they reach across the amnioserosa (Millard and Martin,2008). The unique presence of intra-branch fusion events might then indicate a role for Ed in mediating segmental identity recognition.

Ed Does Not Act Like Dpp or Wg in Specifying Fusion Cell Number

Multiple cells in growing branches have the potential to become fusion cells. However, as branch development proceeds in wild-type embryos, the interactions between FGF, Wg/Wnt, Dpp and Notch signaling pathways restrict fusion cell fate, ultimately selecting a single fusion cell at the tip of each branch (Ikeya and Hayashi,1999; Steneberg et al.,1999; Chihara and Hayashi,2000; Llimargas,2000). Similar to edM/Z mutants, either loss- or gain-of-function of the above factors results in expression of fusion cell fate markers in multiple cells per branch, suggesting that Ed may function in one or more of these pathways. Direct comparison of the phenotypes of edM/Z mutants with embryos defective for the Dpp, Wg, and Notch pathways is complicated by the fact that most analyses of tracheal fusion have been done in dorsal branches, which in edM/Z mutants are physically prevented from fusing by the failure of dorsal closure (Laplante and Nilson,2006), but comparison of other phenotypic features suggests however that Ed is not involved in Dpp or Wnt pathways. Wg also contributes to fusion cell fate specification, but while overexpression of Wg causes supernumerary fusion cells, it also converts many visceral branch cells into dorsal trunk cells, a phenotype not observed in edM/Z mutants. Also, despite the presence of excess fusion cells in Wg overexpression embryos, fusion defects were not reported (Chihara and Hayashi,2000; Llimargas,2000). Also, while activation of the Dpp pathway can specify more fusion cells and cause ectopic branch fusions, activation of the Dpp pathway also converts many presumptive dorsal trunk cells into dorsal branch cells that migrate dorsally (Wappner et al.,1997; Steneberg et al.,1999), which is not observed in edM/Z embryos.

Fusion Phenotypes of edM/Z Embryos Resemble But Are Distinct From Those of Notch Mutants

The trachea phenotypes of edM/Z and Notch mutant embryos share similarities. In both edM/Z embryos and in Notch temperature sensitive mutant embryos incubated at restrictive temperature after early embryogenesis, supernumerary fusion cells are observed without changing the underlying branch identities of the tracheal cells and branches frequently fail the fusion process (Ikeya and Hayashi,1999; Steneberg et al.,1999). In addition, in both edM/Z and Notch mutant embryos, ectopic branch fusions are observed, presumably as a consequence of excess fusion cells finding inappropriate partners to fuse with (Ikeya and Hayashi,1999; Steneberg et al.,1999). These similarities are consistent with previous reports that Ed contributes positively to Notch signaling in other tissues (Ahmed et al.,2003; Escudero et al.,2003; Rawlins et al.,2003a) and suggest that Ed regulates fusion cell specification by means of its influence in the control of the Notch signaling pathway.

However, the nature of branch breaks in edM/Z embryos suggests that Ed could have roles distinct from the Notch pathway. In Notch mutants, the incomplete fusion of branches is associated with and may result from a failure to up-regulate DE-cadherin expression in the fusion cells (Ikeya and Hayashi,1999). In contrast, edM/Z embryos exhibit enriched Arm protein in the fusion cells that undergo successful inter-branch fusion and even in those that undergo intra-branch fusion events. This observation strongly suggests that the up-regulation of adherens junctions in the fusion cells is independent of Ed function and, therefore, the fusion defects are unlikely to be due to adherens junction defects. Further studies will unravel the link between Ed and the Notch pathway during fusion cell fate determination.

Fusion Cell Specification by Ed Is Distinct From Epidermal Boundary Formation

We and others have previously demonstrated that an actomyosin cable is formed at the interface between Ed-expressing and Ed-nonexpressing epithelial cells (Wei et al.,2005; Laplante and Nilson,2006; Lin et al.,2007). However, such an interface does not appear to contribute to the regulation of fusion cell number, because Ed expression in the trachea appeared uniform (Fig. 1A–C) and Ed was clearly visible in the fusion cells (Fig. 1D,E). Moreover, while juxtaposition of Ed-expressing and -nonexpressing cells has been shown to play a role in morphogenesis, differential Ed expression has not been shown to mediate cell fate specification.

Concluding Remarks

This work uncovers an unexpected, and previously unreported, nonautonomous mechanism by which tracheal morphogenesis can be disrupted to produce a convoluted tracheal phenotype. Instead of increasing tracheal length, convolutions result from the combined effects of decreased intersegmental spacing and deep epidermal segment grooves. In addition to these effects on morphogenesis, we also show that Ed regulates the number of fusion cells determined per branch, consistent with a potential role for Ed as a positive regulator of Notch activity. Interestingly, however, the production of these supernumerary fusion cells is associated with both defective and ectopic branch fusion events, suggesting that Ed may have a novel role in fusion cell regulation.


Drosophila Strains

The strains used in this work were w; edF72 P[FRT]40A/CyO P[twi-GAL4] P[UAS-GFP] (Laplante and Nilson,2006), w; edlF20 P[FRT]40A/CyO P[Act-GAL4] P[UAS-GFP] (de Belle et al.,1993), P[btlGAL4]/TM (Shiga et al.,1996), w; edF72 P[FRT]40A P[UAS-Ed]/CyO, P[ovoD1] P[FRT]40A (Bloomington; Chou and Perrimon,1996), P[UAS-DiaCA] (Homem and Peifer,2008), y w; P[en2.4-Gal4]e16E (Bloomington), w; P[prd-Gal4]/TM3,Sb (Bloomington). edM/Z embryos were generated by mating mosaic females bearing germline clones homozygous for edF72 or edlF20 to males heterozygous for edF72 or edlF20 and a GFP-marked balancer chromosome. edM/Z embryos were recognized by the absence of GFP, while the sibling embryos rescued paternally by a wild-type copy of ed were used as “wild-type” controls. Both the edF72 and edlF20 alleles bear nonsense mutations early in the coding region, suggesting that they are likely to be null alleles (Laplante and Nilson,2006).

Generation of Inducible Ed Expression Constructs

Transgenes were generated by polymerase chain reaction (PCR) amplification from cDNA RE66591 (Drosophila Genome Resource Center [DGRC]) and inserted in the pENTRY vector (Invitrogen). The resulting clones were sequenced (Génome Québec Innovation Center) and then recombined into the destination vector pTWH (DGRC). Forward primer: 5′-CACCCGTGTGTGCGAACAACAACTCAG-3′. Reverse primer: 5′-CTAGACAATAATCTCGCGTATG-3′.

Immunohistochemistry, Dye Exclusion, and Microscopy

Embryos were collected 24 hr after egg deposition at room temperature (∼22°C), dechorionated in 50% bleach and heat fixed as described previously (Wieschaus and Nusslein–Volhard,1986). For Neuroglian and 2A12 staining, samples were formaldehyde fixed as described previously (Samakovlis et al.,1996a). The antisera used were Ed (1:1,000, Laplante and Nilson,2006), Coracle (1:500; gift of R. Fehon), Discs Large (1:50; Developmental Studies Hybridoma Bank [DSHB]), Neuroglian (1:500; gift of M. Horsch), Vermiform (1:1,000; gift from S. Luschnig), Dysfusion (1:400; gift of S. Crews), Armadillo (1:100; DSHB), α-Catenin (1:100; DSHB), dp-ERK (1:2000; Sigma) and 2A12 (1:5; DSHB). Alexa-conjugated secondary antibodies were preblocked against fixed wild-type embryos and used as described previously (Laplante and Nilson,2006). Alexa 488- WGA (Invitrogen Molecular Probes) was used at a final concentration of 1:500, and embryos were stained for 1 hr at room temperature. All images, excepted for the embryo injection images, are confocal micrographs taken on a Zeiss LSM510 using sub-saturation conditions controlled by percent laser output, camera gain and exposure time. Texas Red-conjugated 10 kDa dextran was injected into embryos as described previously (Lamb et al.,1998).

Imaging and Length Measurements

Confocal micrographs were analyzed with Volocity image analysis software (Improvision). To accurately count of the Dys-positive cells in the DTs of stage 16 embryos, and to in particular to avoid confusion with cerebellar branch fusion cells, Dys-positive cells were marked and counted while scrolling through z-stacks. Second, Dys-positive cells were also counted in reconstructed three-dimensional (3D) images, which were rotated to obtain unambiguous views of tracheal branches and cells. The length of the tracheal dorsal trunk between transverse connective four and six, denoted “Path4-6,” was obtained by measuring the length of a line traced following the 3D path of the dorsal trunk between the middles of the points where transverse connectives four and six intersected the dorsal trunk (Supp. Fig. S2). The linear distance between the intersections of the dorsal trunk from the transverse connective four to six, denoted “Dist4-6,” was obtained by measuring the length of a straight line drawn between these points (Supp. Fig. S2).


We thank Beili Hu for the injection of the construct. C.L. was funded by The Terry Fox Foundation of the National Cancer Institute of Canada (NCIC). S.M.P. received Postdoctoral support by a NIH Lung Biology Training Grant, G.J.B. was funded by the NIH and a Northwestern University alumni grant. L.N. was supported by The Canada Research Chairs program.