A fat body-derived apical extracellular matrix enzyme is transported to the tracheal lumen and is required for tube morphogenesis in Drosophila

Organ systems are functionally coupled by the exchange of essential humoral factors, including hormones, nutrients and metabolites that coordinate organismal homeostasis. In vertebrates, the liver plays a central role in providing nutrients and proteins required by other organs, as well as detoxifying unfavourable metabolites (Arias, 2009). The fat body of insects is functionally equivalent to the mammalian liver and is the major storage site of lipids, proteins and sugars, thereby serving as an energy reservoir (Chapman, 1998).

In order for proteins synthesized in one organ to function in another, they must be released to the haemolymph or bloodstream, taken up by the target tissues and delivered to the correct cellular compartments. Transcytosis is a key process that facilitates protein translocation across epithelial barriers. In mammals, the polymeric immunoglobulin IgA receptor is transcytosed from the basal of the intestinal epithelium to the gut lumen (Apodaca et al., 1994; Rojas and Apodaca, 2002).

Insect extracellular matrix (ECM) consists of the polysaccharide chitin and an assortment of proteins required for exoskeleton assembly and tracheal development. For example, serpentine (Serp) is a chitin deacetylase required for the tracheal tube morphogenesis (Luschnig et al., 2006). Both the overexpression and loss of serp function result in trachea with excessively long and convoluted tubes (Luschnig et al., 2006; Wang et al., 2006). However, whether non-epithelial tissues contribute to tracheal tubulogenesis is unknown.

Here, we show evidence suggesting that the fat body serves as a source of Serp involved in tracheal development. Serp was expressed in the fat body and the secreted Serp in the haemolymph was transported across epithelial barriers into the tracheal lumen. Knock-down of non-tracheal serp expression resulted in an abnormal tracheal tube morphology during embryonic and larval stages. Thus, long-range transport of Serp through the haemolymph permits systemic control of respiratory tube morphogenesis.

The small GTPase Rab9 mediates Serp sorting and retrograde trafficking, and regulates tracheal tube geometry (Dong et al., 2013). The intracellular trafficking and luminal deposition of Serp is also influenced by Vps32/Shrub, a component of ESCRT III (endosomal sorting complex required for transport III) (Dong et al., 2014). We found abundant Serp protein in the fat bodies of rab9 and shrub mutants (Fig. 1A-C), while its level in the tracheal lumen of these mutants was reduced. Other luminal proteins did not show fat body accumulation in the mutant embryos (supplementary material Fig. S1A,B). To determine whether shrub functions cell-autonomously in the fat body, we expressed a dominant-negative Shrub GFP (Sweeney et al., 2006) using fat body-specific drivers (see Materials and Methods). Regions of partially collapsed tracheal tube were observed in 31% (n=50) of third-instar larvae (supplementary material Fig. S2), suggesting that fat body contributes to normal tracheal development.

The fat body-accumulated Serp in rab9 and shrub mutants could be derived from fat body cells or transported from another source through endocytosis. To examine these possibilities, we blocked clathrin-dependent endocytosis in the fat body using a dominant-negative form of Dynamin, shi K44A, using the ppl-GAL4 fat body driver. In control embryos, Serp was normally deposited in the tracheal lumen and it was not detected in fat body cells (Fig. 1D). However, in shi K44A-expressing embryos, Serp accumulated in intracellular puncta in the cells positive for the fat body marker Seven-up (Svp) (Fig. 1E). The level of Serp accumulation in the fat body cells was lower than in rab9 and shrub mutants (compare inset in Fig. 1E with Fig. 1B,C). Dynamin inhibition is known to interfere indirectly with exocytosis (Koenig et al., 1983). Thus, this result suggests that Serp accumulates in fat body cells with reduced endocytic and exocytic activity. To reduce exocytic activity, we abrogated protein trafficking from the endoplasmic reticulum to the Golgi using brefeldin A (BFA) (Misumi et al., 1986). BFA treatment resulted in strong Serp accumulation in the fat body (Fig. 1F,F′). These results indicated that Serp protein accumulation in the fat bodies of rab9 and shrub mutants was due to cell-intrinsic Serp expression.

The serp transcript is present in the pupal fat body (FlyBase), and we confirmed that the serp transcript was expressed in the tracheal system from stages 12 to 15, followed by strong expression in the epidermis by stage 16 (Luschnig et al., 2006) (supplementary material Fig. S3A). Low-level serp expression was also detected in the stage 16 fat body (yellow arrowheads in supplementary material Fig. S3B).

We also examined the fat body in mutants of the transcription factor Svp, which is required for fat body cell differentiation (Hoshizaki et al., 1994). Fat body cells in svp1 mutants accumulated Serp (Fig. 1G), suggesting that Svp function is required for fat body release of Serp. Collectively, these findings suggest that Serp protein is expressed in the fat body.

Serp does not normally accumulate in the fat body. Thus, we postulated that fat body-derived Serp is secreted into the haemolymph, taken up by tracheal cells and delivered to the tracheal lumen. To test this hypothesis, we used a fusion protein containing the chitin-binding domain of Serp (Serp-CBD-GFP), which is a secreted luminal-marker (Luschnig et al., 2006). Serp-CBD-GFP was expressed using twi-GAL4. The co-expressed cytoplasmic TagRFP marker showed that the twi-GAL4 driver was active in the mesodermal, but not tracheal, cells (Fig. 2A). By contrast, Serp-CBD-GFP was present in the mesoderm and in the tracheal lumen (Fig. 2A,A′). Serp-CBD-GFP also accumulated at the basal surface (Fig. 2B). We next expressed Serp-CBD-GFP using driver ppl-GAL4 and found its accumulation in the tracheal lumen (Fig. 2C). By contrast, a functionally inert secreted protein marker, ANF-GFP (Rao et al., 2001; Tsarouhas et al., 2007), with a similar molecular weight and diffusion characteristics as Serp-CBD-GFP (Dong et al., 2014) was absent from the tracheal lumen (magenta arrowhead in Fig. 2D,D′).

These experiments suggest that Serp translocates to the tracheal lumen. Transcytosis could be the mechanism for Serp translocation; however, it is also possible that fat body-derived Serp reaches the tracheal tube through protein diffusion before trachea acquires epithelial barrier function in the middle of tracheal tubulogenesis (stage 15) (Paul et al., 2003). To distinguish these two possibilities, we induced expression of Serp-CBD-GFP in mesodermal cells surrounding the trachea at stage 15 using infrared laser-induced local heat shock (Kamei et al., 2009; Miao and Hayashi, 2014), and found GFP signal in the tracheal lumen at stage 16 (Fig. 2E, supplementary material Movie 1). This indicates that Serp-CBD-GFP can reach the tracheal lumen across epithelial barriers through transcytosis.

Next, we asked about the role of chitin in Serp transport. Full-length Serp fusion protein (Serp-GFP) expressed by ppl-GAL4 accumulated in the tracheal lumen and the basement membrane of tracheal cells (Fig. 2F). When the experiment was repeated in mutants of chitin synthase, krotzkopf verkehrt (kkv) (Tonning et al., 2005), no Serp accumulation in the tracheal lumen was detected (Fig. 2G,G′), suggesting that chitin synthesis is essential for retention of Serp in the tracheal lumen.

To determine whether Serp transport from the haemolymph to the tracheal lumen plays an active morphological role, we expressed full-length Serp in serp mutants. Expression of Serp by the trachea driver (btl-GAL4) rescued the over-elongated tube phenotype (Fig. 3A,B). Importantly, when Serp was expressed by ppl-GAL4 or twi-GAL4 in serp mutant embryos, similar levels of rescue effect were observed (Fig. 3C-E). These results indicated that the haemolymph-derived Serp is functional.

Finally, we asked whether a non-tracheal source of Serp is required for normal tracheal development by UAS-Serp RNAi driven by ppl-GAL4. In a wild-type genetic background, serp knockdown with ppl-GAL4 reduced Serp levels in the tracheal lumen, but did not cause any defect in tube morphology (Fig. 4A,C,E,F; ppl-GAL4 is not expressed in the tracheal cells). Similarly, reducing the serp gene dose by half using a heterozygous serp^RB had no effect on tracheal tube morphology (Fig. 4B,F). However, the combination of fat body-specific serp knockdown with reduced gene dose resulted in abnormally elongated tube (Fig. 4B,D,F), suggesting that Serp expressed in the fat body and trachea functions redundantly, and that fat body-derived Serp is only essential for tube morphogenesis when the trachea-derived product is limiting.

We here found that the apical ECM (aECM) component Serp is produced not only by tracheal cells lining the lumen, but also by mesodermal fat body cells. We propose that Serp synthesized in the fat body contributes to its accumulation in the tracheal lumen, based on the following evidence. First, serp transcripts were detected in fat body cells. Second, serp knockdown by ppl-GAL4 significantly decreased Serp protein levels in the tracheal lumen (Fig. 4E). Third, inhibition of the vesicle secretory pathway in embryos using BFA resulted in the intracellular accumulation of Serp in the fat body. Serp protein was also detected in the fat body of third-instar larvae (Stefan Luschnig, personal communication). We also found that expression of Serp by the mesodermal driver functionally complemented serp mutant tracheal phenotype, and low-level serp mRNA was detectable in internal tissues, including gut, raising the possibility that other internal organs also serve as sources of Serp.

Serp produced in the fat body is secreted into the haemolymph, taken up by tracheal cells, and then transcytosed to the lumen. We have previously shown that secreted luminal Serp is recycled in the tracheal cells and transported back to the lumen via retrograde trafficking (Dong et al., 2013). Once Serp is internalized from the basal side, it may be transported to the recycling pathway and then released into the lumen, where it is retained by binding to chitin (supplementary material Fig. S4). Thus, through retrograde recycling and transcytosis pathways, sufficient levels of Serp are maintained in the tracheal lumen, thereby facilitating the dynamic remodelling of the aECM that is essential for the regulation of tube morphogenesis.

The fat body supplies essential proteins required by other organs, such as yolk proteins for oogenesis (Telfer, 1961; Roth and Porter, 1964; Pan et al., 1969; Hagedorn and Fallon, 1973), collagen IV for imaginal discs (Pastor-Pareja and Xu, 2011) and xanthine dehydrogenase for eye pigmentation (Reaume et al., 1989). Proteins synthesized and/or stored in the fat body of ants were found in the cuticle and eggshell (Roma et al., 2006, 2008). Our findings suggest that Serp should be added to the increasing repertoire of proteins that the fat body supplies to other organs. The transport of Serp from the fat body is unique in that the protein reaches the tracheal lumen and regulates tracheal tube length. Thus, our results, taken with earlier findings, support the hypothesis that the fat body plays essential roles in the development of multiple organs through the production and secretion of numerous proteins.

shrub⁴/CyO and UAS-Shrub-GFP were gifts from Fenbiao Gao (Sweeney et al., 2006); UAS-Serp-CBD-GFP, UAS-Serp-GFP and serp^RB were gifts from Stefan Luschnig (Luschnig et al., 2006); UAS-ANF-GFP was a gift from Christos Samakovlis (Tsarouhas et al., 2007); rab9⁵⁶ and rab9¹⁹⁹ have been described previously (Dong et al., 2013); btl-Moesin-RFP was a gift from Markus Affolter (Ribeiro et al., 2004); UAS-shi K44A, svp¹, twi-GAL4 and kkv¹ was from the DGRC in Kyoto; ppl-GAL4 (Colombani et al., 2003) and cg-GAL4 (FBti0027802) were gifts from Takashi Nishimura (RIKEN CDB, Kobe, Japan). The serp RNAi (v15466) strain was from the Vienna RNAi Center.

The primary antibodies were: mouse monoclonal anti-Svp [1:10, a gift from Yasushi Hiromi (Kanai et al., 2005)]; guinea pig anti-Uif [1:500, a gift from Robert E. Ward (Zhang and Ward, 2009)]; and rabbit anti-Serp (1:300, a gift from Stefan Luschnig). The chitin-binding probe (CBP) (1:50) was prepared from a bacterial expression construct according to the protocol provided by Yinhua Zhang (New England Biolabs).

The serp probe was designed as described previously (Luschnig et al., 2006). The sequence was cloned into pBluescript II KS and then digested with XhoI or EcoRI to create the template for in vitro transcription using T7 and T3, respectively, to produce sense and antisense probes. Whole-mount wild-type embryos were fixed using 4% paraformaldehyde in PBS. In situ hybridization was conducted using DIG-labelled sense and antisense probes, following a standard protocol. After hybridization, stage 16 embryos were cut transversely with needles, and then serp mRNA expression was analysed.

Serp-CBD-GFP was digested by EcoRI and XbaI from UAS-Serp-CBD-GFP construct (Luschnig et al., 2006), and then was inserted into EcoRI and XbaI-digested pCaSpeR-hs vector. HS-Serp-CBD-GFP plasmid was injected into y¹ w^63C1; delta2-3 embryos to obtain a transgenic line. The trachea-surrounding tissues of embryos carrying HS>eGFP, HS>Serp-CBD-GFP and btl>Moesin-RFP (Fig. 1E) or HS>Serp-CBD-GFP alone (supplementary material Movie 1) were illuminated by infrared laser for transient induction of Hsp70 promoter (Kamei et al., 2009; Miao and Hayashi, 2014).

The embryos were dechorionated and incubated with heptane for 30 s, and then washed in a PBS buffer containing 0.5 mM CaCl₂ and 0.5 mM MgCl₂ for 30 s. The embryos were then incubated in 20 μg/ml brefeldin A in PBS buffer for 180 min. Brefeldin A was from Cell Signaling (catalogue number 9972) and was dissolved in DMSO.

Images were acquired using a laser-scanning confocal microscope (FV1000, Olympus) equipped with 60× oil- and water-immersion objectives. Malma software (Kagayaki Kato, unpublished) was used to adjust the images and to measure the fluorescent intensity. The dorsal trunks (DTs) were measured in fixed embryos (5 to 12) using CBP staining and analysed with Fiji-Image J. The expression level of Serp in different genotypes was quantified as anti-Serp fluorescence intensity normalized by the level of CBP staining. Student's two-tailed unpaired t-test with equal variance was used to assess statistical significance.

We thank S. Luschnig, C. Samakovlis, Y. Hiromi, F. Gao, T. Nishimura, M. Affolter, the Bloomington, Kyoto and Vienna Stock Centers for generously providing antibodies and fly stocks; and Y. Zhang for the plasmid and protocol for CBP preparation. We also thank T. Nishimura for discussion and suggestions, S. Luschnig for sharing unpublished data, and Hayashi's lab members for critical readings of the manuscript.