Control of Airway Tube Diameter and Integrity by Secreted Chitin-Binding Proteins in Drosophila

The transporting function of many branched tubular networks like our lungs and circulatory system depend on the sizes and shapes of their branches. Understanding the mechanisms of tube size control during organ development may offer new insights into a variety of human pathologies associated with stenoses or cystic dilations in tubular organs. Here, we present the first secreted luminal proteins involved in tube diametric expansion in the Drosophila airways. obst-A and gasp are conserved among insect species and encode secreted proteins with chitin binding domains. We show that the widely used tracheal marker 2A12, recognizes the Gasp protein. Analysis of obst-A and gasp single mutants and obst-A; gasp double mutant shows that both genes are primarily required for airway tube dilation. Similarly, Obst-A and Gasp control epidermal cuticle integrity and larval growth. The assembly of the apical chitinous matrix of the airway tubes is defective in gasp and obst-A mutants. The defects become exaggerated in double mutants indicating that the genes have partially redundant functions in chitin structure modification. The phenotypes in luminal chitin assembly in the airway tubes are accompanied by a corresponding reduction in tube diameter in the mutants. Conversely, overexpression of Obst-A and Gasp causes irregular tube expansion and interferes with tube maturation. Our results suggest that the luminal levels of matrix binding proteins determine the extent of diametric growth. We propose that Obst-A and Gasp organize luminal matrix assembly, which in turn controls the apical shapes of adjacent cells during tube diameter expansion.


Introduction
Tube length and diameter are major determinants of flow rates in tubular organs. The generation of appropriate tube dimensions during organ morphogenesis is critical for tissue function and animal homeostasis. Consequently an important challenge for research aiming to understand tubular organ development is to elucidate the acquisition of stereotyped dimensions and proportions in the branches of a tubular network. The major airways of the Drosophila tracheal network consist of a single epithelial cell layer and provide a simple system for the genetic dissection of tube size control. Tracheal tube expansion occurs without increases in cell numbers and experimental alterations in cell numbers do not affect tracheal tube dimensions. These results suggested that tube growth in the embryonic and larval trachea relies on cell rearrangements and cell shape adjustments to form tubes of specified sizes. The analysis of many mutants with selective tracheal tube overgrowth defects has elucidated some of the cellular mechanisms in epithelial tube size regulation [1][2][3][4][5][6][7].
A central role in tracheal tube size regulation has been ascribed to the structure and dynamic modifications of the apical extracellular matrix. Diametric tube expansion is preceded by the generation of a transient luminal cable composed of chitin fibrils. Chitin and associated proteins also assemble into a complex apical matrix, the taenidia, which is tightly juxtaposed with the apical surface of the epithelium. Whereas the luminal chitin cable is cleared from the tube before larval hatching to enable gas filing, the taenidial matrix remains and is thought to reinforce the larval network [8,9]. Mutations in genes involved in chitin biogenesis and assembly result in irregular diametric expansion leading to locally constricted and dilated tubes in the mutants [9][10][11]. In addition, these mutants show overelongated tracheal branches at the end of embryogenesis. This led to the hypothesis that the expanding luminal chitin filament coordinates the epithelial cell shape rearrangements during tube growth. The importance of luminal chitin in tracheal tube size control is further highlighted by the analysis of vermiform (verm) and serpentine (serp) mutants, which show luminal matrix defects and tube overelongation. verm and serp encode luminal putative chitin deacetylases, suggesting that the acetylated chitin matrix restricts cell extension and tube elongation [12,13]. Tube overelongation is also the hallmark phenotype for another group of mutants including genes involved in the assembly of septate epithelial junctions (SJs). Insect SJS are functionally similar to the vertebrate tight junctions controlling the paracellular traffic between epithelial cells. ''SJ mutants'' show impaired Verm and Serp secretion into the lumen indicating that a function of SJ proteins may be to facilitate the apical targeting of chitin modifying enzymes. An additional mechanism by which SJ proteins control tube size is, through the regulation of the subcellular localization of apical polarity proteins like Crumbs and aPKC [4], [14]. Recently, the non-receptor tyrosine kinase Src42A has been identified as a driver of tube elongation. src42A mutants show short tubes and Src42A overexpression can cause overelongation. It is suggested that Src42A activation promotes the axial extension of the tracheal cells along the longitudinal axis of the tubes to promote elongation [15,16]. Collectively, the results from several studies indicate that Src42A activation promotes elongation by controlling the apical cytoskeleton and cell junctions whereas the dynamic modifications of the apical ECM and suppression of Crb targeting to the apical membrane control the termination of tube elongation.
Despite the progress in identifying mechanisms of tube elongation, we know little about tracheal tube growth in the radial axis. A burst of epithelial secretory activity deposits luminal proteins in the tubes and precedes diametric expansion. Mutations affecting genes involved in the transport steps of secreted proteins from and to the endoplasmatic reticulum (ER) like sar1, show a strong reduction in tube diameter and defective luminal matrix [17][18][19][20][21]. There are several models attempting to explain the role of secretion in diametric tube expansion. One model postulates that the secretion burst of luminal proteins underlies the modification of the chitininous ECM. Unknown secreted proteins are thought to modify the biophysical properties of luminal chitin and cause it to swell to inflate the tube. Alternatively, the secretory burst may deliver the necessary apical membrane from the ER to the apical cell surface to meet the needs of tube diameter expansion. Finally, apical secretion may deliver regulators of osmotic pressure to the apical membrane and thereby generate a force for tube inflation. The later model involving the apical delivery of a tube-wide osmotic dilator seems unlikely since individual cells expressing a wild-type copy of Sec24 in sec24 mutants can support a local tube dilation in an otherwise stenotic tracheal tube [16]. Thus, although the importance of develop-mentally regulated apical secretion is at the heart of diametric tube expansion in the trachea, neither the secretory cargoes nor the cellular mechanisms underlying the secretion mediated tube expansion are known.
The aim of this work was to identify and characterize secreted proteins involved in diametric tube expansion. We present the analysis of obst-A and gasp mutants and show that both genes are required for larval growth and diametric tracheal tube expansion. They encode secreted luminal proteins with chitin binding domains and are required for intraluminal chitin filament assembly and taenidial integrity. We identify the Gasp protein as the unknown antigen recognized by the widely used antibody marker 2A12. The deletion of obst-A or gasp genes shows mildly stenotic tubes but the diametric expansion phenotype becomes exaggerated in the double mutants. Conversely, concurrent overexpression of both proteins in wild-type animals leads to localized tube dilations. Our analysis argues that Obst-A and Gasp modify the taenidial matrix to provide an expanding mold that determines airway tube expansion.

Drosophila Strains
The obst-A null mutant was generated by flip FRT technique using two piggyBac elements (f02348, f03687). w 1118 was used as the wild-type strain. gasp (P{XP}d02290) mutant is described in Flybase. Crosses for ectopic expression using btl-GAL4, en-GAL4 and 69B-GAL4 drivers were performed at 25uC. In all experiments CyO, TM3 and FM7c balancer strains carrying GFP or lacZ transgenes were used as necessary to identify embryos with the desired genotypes.

Molecular Biology and Western Blotting
The UAS-obst-A transgene was generated by subcloning the insert of LD43683 into pUAST. The constructs used for protein expression in BL21 cells were the Melanotransferrin (MTf) cDNA fused to the 66HisTag [23] and the gasp cDNA fused to 66HisTag [17]. The pET28a plasmid vector was used as a negative control. Expressed proteins were analyzed by immunoblotting in a standard western blotting procedure.

Live Imaging
Embryos were dechorionated in 3% sodium hypo-chloride for 2-3 min, mounted in to a slide with a gas permeable membrane [17] and imaged with a CCD camera (AxioCam MRm) attached to an AxioImagerZ1 (Zeiss) microscope by using a 20x/0.75NA Plan-APOCHROMAT objective (Zeiss). Time-lapse movies were created from Z-stack projections using NIH ImageJ (http://rsb. info.nih.gov/ij/).

Morphometric Analysis and Statistics
To asses embryonic tracheal tube size (tube diameter or length), wild-type, obst-A, gasp and obst-A; gasp mutant embryos were stained with the apical markers Crumbs (Crb) or Uninflatable (Uif) and imaged by confocal microscopy (LSM510, Zeiss). Confocal digital micrographs (Maximum intensity projections) of tracheal meta-meres 4-6 were used to estimate diameter or length. Tube diameter in each embryo was calculated as the average of three different measurements within each metamere. Tube length was measured by tracing the apical localization of the Crb or Uif immunostainings. We used at least 6 embryos at stage 16.1 for the diameter and length measurements of each genotype. Larval body size measurements were done in anesthetized living larvae, imaged by wide field microscopy (Axioplan2, Zeiss). Measurements were conducted using the AxioVision 4.8.2 (Zeiss) software. Results are presented in bar graphs as the means 6 standard deviation unless otherwise indicated. In all cases a two-tailed distribution unpaired Student's t-test was used to estimate the statistical significance of the differences between wild-type and each mutant.

Obst-A and Gasp are Expressed and Secreted into the Airways during Tube Expansion
We surveyed the BDGP gene expression database to identify genes that are expressed in the trachea and encode putative secreted proteins [25]. Several members of the Obstructor gene family were strongly expressed in the airways during the tube expansion period. This gene family is conserved among insects and can be divided into two subgroups, based on protein domain arrangement [26]. We focused on the Obstructor subgroup 1 because it contains genes with distinct tracheal expression patterns. obst-A, B, C and D are weakly expressed during early embryogenesis but their expression increases later, accompanying the initiation of cuticle formation in ectodermal tissues and diametric tube expansion in the airways [26]. obst-A (Fig. 1a) and obst-B are both expressed in the trachea and epidermis; gasp (obst-C) (Fig. 1b) shows expression in the tracheal system, foregut and salivary glands. Interestingly, gasp mRNA levels were homogeneous in all cells of the tracheal network, whereas obst-A mRNA was reduced or absent in cells located at regular intervals along the DT airways. The remaining of the tracheal cells expressed obst-A at high levels, suggesting that obst-A expression is differentially regulated in the DT cells. obst-D and obst-E are mostly detected in the midgut and epidermis. The expression of the subgroup 2 genes is not elevated during embryogenesis, arguing against a role of them in tube morphogenesis [26]. To examine the phylogenetic relationships among the obst family members, we constructed a phylogenetic tree based on the predicted protein sequences (Fig. 1c). obst-A and gasp are the most closely related genes expressed in the trachea. Alignment of the protein sequences revealed 37% identity. The predicted proteins contain a signal peptide at the amino terminus, followed by three chitin-binding type 2 domains (CBD2) (Fig. 1d). CBD2 is an extracellular domain that contains six conserved cysteines that probably form three disulphide bridges. These domains are connected with linker regions of similar length in both proteins. The features of the two proteins and the timing of their expression before the tube expansion prompted to explore Obst-A and Gasp function in the tracheal tubes.
We first generated a null obst-A D1 mutant (referred to as obst-A mutant in the remaining of the text) by the FLP-FRT method [27]. mRNA in situ hybridizations of mutant and wild-type embryos confirmed that obst-A was deleted since we could not detect any obst-A mRNA expression in the obst-A mutants (Fig. S1b). obst-A mutants showed larval lethality, which could be rescued by the expression of transgenic obst-A in the mutants using the ectodermal driver 69BGal4 indicating that the deletion does not affect any neighboring genes. For gasp, we used a transposon-induced mutation, which carries a P-element 66 bp upstream of the predicted translation initiation site of the gasp gene and is larval lethal. We analyzed gasp mutants using a polyclonal antiserum generated against recombinant Gasp [17]. We could not detect any Gasp protein in the gasp mutants suggesting that transposon insertion abolishes gasp expression in homozygous embryos (Fig.  S1d). During these experiments we also noticed that the Gasp protein and the unknown antigen recognized by the 2A12 antibody show a similar pattern of protein localization in the trachea. Initially during embryonic stages 13 and 14, both Gasp and the 2A12 antigen are predominantly cytoplasmic. At stage 15 they are both deposited into the lumen and by stage 16, we could only detect a luminal signal for both antigens. Surprisingly, when we stained gasp mutant embryos for both Gasp and the 2A12 antigen we did not detect any signal from either of these antibodies (Fig. 2b, c). Parallel staining for other luminal markers in the mutants did not reveal any notable changes in their localization (Fig 4c, g). These results implied that the antigen for 2A12 is either encoded by gasp or by another gene, whose expression is severely reduced in gasp mutants. To test this, we first expressed Gasp protein using the en-GAL4 driver in the dorsal part of the hindgut, where gasp is not normally expressed. We stained en.gasp embryos for Gasp (Fig. 2d), 2A12 (Fig. 2e) and Verm an unrelated chitinbinding protein (Fig. 2f). Ectopic Gasp was detected in the cells of the dorsal hindgut and in the hindgut lumen by both the polyclonal Gasp antiserum and by the 2A12 monoclonal antibody, but not by the Verm antiserum. This suggested that the epitope recognized by the 2A12 antibody is part of the gasp open reading frame. The Gasp antiserum produced a stronger signal in the hindgut lumen compared to 2A12 indicating that the guinea pig antiserum more readily recognizes additional epitopes on Gasp (Fig. 2d, e).
We further explored whether the 2A12 mAb recognizes recombinant Gasp protein expressed in bacteria. We prepared extracts from cultures expressing either a His-tagged version of Gasp, an unrelated His-tagged MTf protein [23] and a vector expressing only the tag as a control (Fig. 2g). We probed Western blots of the extracts with the Gasp antiserum, the 2A12 mAb and an antiserum against the His-tag. All three reagents detected the Gasp-His tagged protein, whereas MTf-His was only recognized by the anti-His-tag antibody. This indicates that recombinant   Gasp is the target of the 2A12 mAb and identifies the unknown antigen recognized by a widely used airway marker in Drosophila.

Obst-A and Gasp Control Cuticle Morphogenesis and Larval Size
Obst-A and other proteins involved in the assembly of chitinous extracellular matrices are essential for larval growth and cuticle integrity [28]. We first examined the anticipated role of Gasp in larval size control and cuticle formation. We analyzed the size and cuticle structure of obst-A and gasp mutants at the first instar larval stage (Fig. 3). Both obst-A and gasp single mutant larvae are shorter compared to heterozygote siblings. The body length of obst-A larvae was reduced by 8% and the length of gasp mutants was 14% shorter compared to the wild-type. Given the similarities of the encoded proteins and the resemblance of the expression patterns of the two genes we generated and analyzed obstA; gasp double mutants. obstA; gasp larvae were even shorter than obstA or gasp single mutants. Their body length was reduced by 25% compared to the wild-type larval length (Fig. 3a-e). The width of the mutants was not significantly different than the width of wild-type larvae (Fig. 3f). The increased severity of the body length defect in double mutants suggests that Obst-A and Gasp have overlapping additive roles in regulating larval length. We compared cuticle preparations of first instar larvae of the three mutant genotypes to wild-type preparations to assess the potential roles of role of obst-A and gasp in epidermal differentiation (Fig. 3g-j). We did not detect any obvious patterning defects or dorsal closure and head involution phenotypes in the single mutants or in obst-A; gasp animals. However, in contrast to the wild-type and single mutant preparations the cuticles of obst-A; gasp double mutants were severely bloated. This cuticle defect in the double mutants suggests a synergistic function for Obst-A and Gasp in chitin fibril assembly and exoskeleton integrity, which subsequently defines the shape and growth of the larva.

Obst-A and Gasp are Required for Apical Matrix Integrity in the Airways
The remarkable intensification of the cuticle shape defect in the double mutants prompted us to examine the phenotypes of single and obst-A; gasp double mutants in the tracheal airways. A distinguishing feature of a functional tracheal respiratory network is the presence of gas in the larval tubes. Similar to wild-type controls, the airways of single obst-A and gasp mutant embryos became gas-filled at the end of embryogenesis (Fig. 3a-c). However, double mutant embryos showed a fully penetrant gasfilling defect (Fig. 3d). Additionally, gasp mutants showed regions of local tube collapse (Fig 3c), which could be rescued by reexpression of Gasp in the trachea of 89% of the mutants using the btlGal4 driver (n = 28). A similar tube collapse phenotype was detected in the tubes of obst-A; gasp mutants (Fig 3d). This phenotype is common among mutants with defective epithelial integrity. Therefore, we analyzed the epithelial apical cell membrane and adherens (AJ) and septate junction (SJ) integrity by staining single and double mutants for Crumbs, DE-cadherin and MTf [23]. Neither the single nor the obst-A; gasp mutants show notable defects in the intensity or localization of these markers suggesting that epithelial integrity is not affected in the double mutants (Fig. S2). To identify the origin of the tube collapse phenotype in obst-A; gasp embryos we analyzed the luminal chitin filament and the apical taenidial lining of the tubes.
Chitin is detected inside the dorsal trunk tubes of wild-type embryos at stage 14. It forms a cable of fibrils, which expands in diameter concurrently with the radial growth of the lumen before it disappears late at stage 16 [9]. We examined the transient luminal chitin cable and its deposition by using a fluorescent chitin-binding probe (ChtB) (Fig. 4). Both obst-A and gasp single mutants showed a decreased and irregular staining intensity of luminal chitin compared to wild-type embryos (Fig. 4b, c). In obst-A;gasp double mutants the staining of the intraluminal matrix was even more diffuse compared to the single mutants and to wild-type embryos (Fig. 4d). This suggested that the putative chitin binding proteins Obst-A and Gasp control the assembly or stability of the extracellular matrix. We used an antiserum against the secreted chitin deacetylase, Verm to examine if the extracellular matrix defect may be due to aberrant modifications of the chitin polysaccharides [13]. Luminal Verm staining was reduced in both obst-A and gasp single mutants and was further diminished in the double mutant (Fig. 4e-h). Importantly, the transcript levels of Verm were not affected in obst-A or gasp mutants compared to wild-type embryos indicating that Gasp and Verm control the luminal levels of Verm post-transcriptionally (Fig. S3).
To further test the potential structural role of Obst-A and Gasp in chitin assembly, we analyzed the chitin in the taenidia, the cuticular ridges that line and reinforce the airway tubes (Fig. 5). Transmission electron microscopy (TEM) reveals three layers of cuticle in wild-type embryos: the envelope layer, the proteinaceous epicuticle and the chitin-rich procuticle (Fig. 5a). TEM analysis of obst-A; gasp double mutants revealed a grossly irregular shape in the taenidial folds (Fig. 5b, c). The outermost layers of the envelope, and the dark epicuticle appeared normal in the mutants. However, the chitin-rich space between the epicuticle and the apical epidermal surface was disorganized and granular with amorphous composition compared to the uniform procuticle layer of the wildtype embryos.
Thus, the reduced staining intensity of the luminal chitin cable and the defects in the procuticle layer and the irregular taenidial folds in obst-A; gasp double mutant embryos suggest a direct function of Obst-A and Gasp in chitin fibril organization and maintenance.

Obst-A and Gasp Drive Diametric Tube Expansion
A characteristic phenotype in many mutants with disrupted apical extracellular matrix is the presence of overelongated or cystic tubes. To assess the impact of Obst-A and Gasp function in tube shapes and size control we assessed tracheal tube dimensions in single and double mutant embryos at stage 16. We stained obst-A and gasp single mutants for the apical membrane markers Crumbs and Uninflatable (Uif) [29] (Fig. 6a-d). Because the airway tubes are tapered we evaluated dorsal trunk (DT) length and diameter in three central consecutive tracheal metameres to obtain a representative view (n = 6). obst-A and gasp single mutants showed a small but significant reduction in airway diameter. In obst-A mutants we found an average 5% reduction in tube diameter, whereas in gasp mutants DT diameter was reduced by 10% compared to wild-type embryos. In obst-A; gasp double mutants the airway diameter was decreased by 15% in comparison to wild-type (Fig. 6e). By contrast we did not detect any significant changes in the length of the DT obst-A; gasp double mutants at stage 16 (Fig.  S4). However, a portion of the obst-A and obst-A; gasp embryos also showed some tube tortuosity at stage 17, presumably due to the reduction in the luminal levels of Verp or Serp. This suggests that the primary role of the luminal proteins Gasp and Obst-A is to facilitate radial tube expansion. This function likely involves their role in the assembly of the chitin cable or the taenidial lining, which could provide a distending scaffold force that stretches the apical surfaces of the epithelial cells.
To further test this hypothesis, we overexpressed Gasp-GFP alone or Gasp-GFP together with Obst-A or an unrelated secreted protein ANF-GFP in the trachea of wild-type embryos and assessed their ability to induce tube dilations (Fig. 7a-c). We monitored tube shape changes by live imaging, taking advantage of Gasp-GFP or ANF-GFP, which labeled the tracheal lumen. Obst-A, Gasp-GFP double overexpression caused local dilations in the DT tubes in 58% of the analyzed embryos (Fig. 7d). In these embryos the tubes were only partly cleared from luminal protein and collapsed during the subsequent step of liquid clearance (Movie S2). Gasp-GFP expression alone had only a minor effect in tube dilation and overexpression of the exogenous luminal marker ANF-GFP did not cause any phenotypes (Movie S1 and Fig. 7a). These results suggest that a concurrent increase in the luminal levels of both Obst-A and Gasp in the tracheal tubes can directly stimulate tube diameter expansion, presumably by modifying the structure of the luminal chitin. The subsequent tube collapse defect indicates that the excess of luminal chitin binding proteins interferes with the biophysical properties of the matrix and leads to airway collapse during the process of liquid clearance.

Discussion
We have identified Gasp as the unknown antigen of the 2A12 mAb, which has been the major luminal marker for the analysis of tracheal airways in the last 20 years. The 2A12 antigen together with the apical protein Piopio represent a group of endogenous secreted cargoes that selectively rely on the formin protein, Diaphanous for their secretion [30]. On the other hand, Verm and Serp exemplify luminal proteins that require alternative targeting routes involving the septate junctions and endosomal compartments for their luminal accumulation [13,31]. The identification of the 2A12 antigen further enables the investigation of the targeting signals determining the luminal deposition of tube size regulators (Gasp) and other apical cargos.
Our genetic analysis and the predicted protein sequences strongly argue that both Obst-A and Gasp are required for chitin extracellular matrix assembly and integrity. Both the chitin staining and the levels of the putative chitin deacetylase Verm are reduced in obst-A and gasp mutants. This is in accord with biochemical and histological analysis of Obst-A, which has been shown to bind to chitin and other chitin binding proteins and modifying enzymes [28].
We showed that the Gasp and Obst-A duo present the first apical ECM proteins directly involved in diametric tube expansion in the Drosophila airways. Mutants lacking both proteins show a significant decrease of 15% in DT diameter compared to wildtype. Zygotic null sar1 mutants show a 30% reduction in tube diameter and a severe defect in ER integrity and the apical secretion of all ECM proteins tested [17]. This comparison suggests that Obst-A and Gasp account for a substantial subset of the secreted proteins that directly drive diametric expansion. This is further supported by the tube dilation defects generated by the concurrent overexpression of both proteins in the tracheal epithelium. However, neither the overexpression of Obst-A nor Gasp alone could generate a notable diametric increase in the tubes of wild-type embryos. This suggests that the luminal amount of the two proteins need to reach a critically high level to drive tube expansion. Alternatively, and despite the similarities in amino acid sequence and domain organization, either of the two proteins may influence chitin assembly differently and they may both be needed to achieve the optimal structural modifications. Petkau et al. noticed a late defect of tube overelongation in obst-A mutants. We noted that this phenotype appears at later stages compared to the diametric expansion phenotype of obst-A and obst-A; gasp mutants. We therefore argue that the tube overelongation defect is a secondary function of the tracheal Obst family members and is likely attributed to the reduction of the luminal levels of Verm and Serp.
How does chitin modification in the lumen lead to diametric tube expansion? Our data suggest that Obst and Gasp organize luminal chitin fibrils into an expanding structure that generates an intra-luminal mechanical tension forcing the apical surface of the underlying epithelial cells to expand. This dilation mode depending on local assembly and swelling of the luminal matrix also operates in the hindgut, where instead of chitin it involves mucin-type glycoproteins [32]. In the tracheal tubes, the secretion of chitin-binding proteins and modifying enzymes would lead to their direct association with the luminal chitin in the vicinity of their secretion. Thereby, secreted tube size regulators, like Obst-A and Gasp would exert the local dilation effects observed around secretion-defective sec24 mutant cells overexpressing sec24 [19].
The Obstructor gene family is conserved across invertebrate species ranging from nematodes to arthropods and the family members are expressed in variety of tubular organs and the epidermis [26]. Our genetic analysis of obst-A and gasp suggests central roles of obst genes in the morphogenesis of the majority of the identified animal species. Additionally, the analysis of of obst-A and gasp mutants argues for the important role of apical extracellular matrix assembly in tube dilation. Glycan-rich apical matrices line the surface of several tubular organs in vertebrates and may similarly participate in shaping the final tube dimensions during their morphogenesis [33][34][35]. Number of embryos used for measurement of each genotype n = 6. The y-axis represents the length in micrometers. No significant difference was detected (p.0.05) in comparison between wild-type embryos with different mutants (two-tailed distribution unpaired Student's t-test).

(TIF)
Movie S1 Wide-field, time-lapse movie showing the DT of a wild-type embryo expressing btl.Gasp-GFP at late stage 16. Images were acquired every 10 min and for a period of 90 min. (MOV) Movie S2 Wide-field time-lapse movie showing the DT of a wild-type embryo expressing btl.Gasp-GFP; Obst-A at late stage 16. The trachea develops tube dilations in the dorsal trunk and finally collapses. Images were acquired every 10 min and for a period of 90 min. (MOV)