Introduction

The tracheal system of Drosophila comprises three different parts: the main dorsal trunks, the smaller tracheal branches, and the terminal cells (1, 2). The tracheal terminal cells (TTCs) form the most distal part of the respiratory system. They are the site at which gas exchange takes place. TTCs fulfill an essential role by supplying all body organs with oxygen. To fulfill their task, even under ever-changing conditions, these cells possess a high degree of structural plasticity. For example, TTCs can sprout in response to local hypoxia (3), or in response to differences in nutrient availability (4); this process is analogous to angiogenesis in mammals. Sprouting is dependent on growth factors, as well as the hypoxia-inducible factor (HIF)-α homolog sima, again analogous to endothelial cells during angiogenesis in the mammalian vascular system (3, 5, 6).

The primary role of the proximal tracheal tubes is to conduct air to more distal parts. The epithelial cells lining these tubes can mount an effective immune response to infections (7, 8). This immune response is driven mainly by expression of several antimicrobial peptides (AMP) (9, 10). One out of the two major immune pathways operative in Drosophila, the immune deficiency (Imd) pathway, is solely responsible for responses in the larval airway epithelium; this is because the Toll pathway is not functional in these cells (8). The Imd signaling pathway is homologous to the human tumor necrosis factor-α (TNF-α) pathway, and converges on activation of NF-κB factors. Moreover, it is connected to c-Jun N-terminal kinase (JNK) signaling via transforming growth factor-β (TGF-β) activated kinase 1 (dTak1) (11). Infection of the airway system induces an immune response by the tracheal epithelium, which includes expression of canonical Imd and JNK target genes (9, 12). Chronic activation of tracheal immune signaling leads to marked structural changes in the epithelium (13). This tissue remodeling is mediated by JNK and its downstream transcription factor, forkhead box sub-group O (foxo), which is a as a terminal target of the epithelial immune system. FoxOs are also central proteins of the insulin signaling pathway, which also plays an important role in the structural plasticity of TTCs (14–16).

The present study shows that TTCs differ fundamentally from the rest of the tracheal epithelium. In the case of a natural infection of the tracheal system, these cells show only a negligible immune response. The very few TTCs that do show AMP expression are structurally impaired. We found that TTCs are immune privileged. The Imd pathway in TTCs is deactivated since the cells do not express the transmembrane peptidoglycan recognition protein (PGRP)-LCx. Chronic activation of the Imd signaling pathway in TTCs leads to JNK- dependent apoptosis. Therefore, we hypothesize that the immune privileged status of TTCs maintains the normal function of foxo, which acts as a regulator of structural plasticity.

Results

Tracheal infection revealed that AMP expression in terminal cells is rare

Drs-GFP larvae were infected with Erwinia carotovora for 24 h. These larvae express GFP under the control of the Drosomycin (Drs) promoter, which is activated by a natural infection of the trachea (9). GFP fluorescence was analyzed after 24 h, focusing on TTCs (Fig. 1). Fluorescence was visible in all parts of the tracheal system except for TTCs (Fig. 1A, A’, B, B’; dashed lines). A total of 169 larvae were analyzed. All larvae with fluorescence detectable in the dorsal branch (DB) were counted as GFP-positive (Fig. 1C). About 34% showed no GFP signal, a finding consistent with previous observations (10). The remaining larvae showed a GFP signal in the DB (65.7%); however, only 8.3% also showed fluorescence in TTCs (Fig. 1D). Most of these few fluorescent TTCs showed clear structural differences from TTCs without a GFP signal (Fig. 1E, E’). Although some cells had a normal structure (Fig. 1F, F’), most had shortened branches (Fig. 1E, E’, G, G’), showed signs of melanization (Fig. 1H, arrow), or were no longer air-filled (Fig. 1H, arrowhead). These observations suggest that an immune response in the TTC itself act on these cells.

Fig. 1:

Tracheal terminal cells do not express the Imd receptor PGRP-LCx

We wondered whether constitutive immune activation within the entire tracheal system would exclude TTCs, as expected from the results of the infection experiments. To ensure that all parts of the tracheal epithelium, including TTCs, were exposed to this activating stimulus, we ectopically expressed the secreted (and intracellular) Imd pathway receptor PGRP-LE using the ppk4-Gal4 driver. AMP expression (a readout of the immune response) was visualized by simultaneous expression of GFP-tagged Drs (Fig. 2). A strong immune response was observed in the tracheal tissue, from the dorsal trunks down to the smaller branches (Fig. 2A– C). While expression of Drs was consistent throughout most parts of the tracheal system, it was completely absent from the most distal parts (Fig. 2B–E). Closer observation of the TTCs revealed a distinct breakpoint in the GFP signal, which occurred at the proximal end of the TTCs. This was true for TTCs attached to the outer cuticle (Fig. 2D) as well as for cells attached to the intestine (Fig. 2E). Activation of the Imd pathway may not be fully functional in these cells. To activate the Imd pathway successfully, secreted PGRP-LE has to bind to transmembrane PGRP-LC, meaning that only cells expressing PGRP-LC can be activated (17, 18). Therefore, to investigate expression of PGRP-LC in the tracheal system, and specifically in TTCs, we expressed Gal4 under the control of the PGRP-LC promoter to drive concurrent GFP expression (Fig. 2F, G). While GFP was visible throughout the entire tracheal system, it was absent from all TTCs associated with the cuticle (Fig. 2F, F’) and intestine (Fig. 2G, G’). To demonstrate that GFP expression can be visualized in TTCs, and that the lack of a signal was not an artifact, expression of GFP in TTCs was driven by the tracheal driver btl-Gal4 (btl-Gal4; UAS-GFP) (Fig. 2H, I). TTCs on the dissected cuticle (Fig. 2H) and on the intestine (Fig. 2I) expressed GFP. Thus, the data indicate that lack of expression of the Imd pathway receptor PGRP-LC by TTCs is the reason for the difference between TTCs and cells within the rest of the tracheal system.

Fig. 2:

Ectopic activation of PGRP-LCx in the tracheal system leads to death of TTCs

To further investigate the effects of immune responses on the tracheal system, we expressed PGRP-LCx exclusively in TTCs (PGRP-LCx^OE; Fig. 3). We found that Imd-mediated immune activation was restricted to these cells, as indicated by concurrent expression of GFP. Compared with the widely branched and diversified TTCs in control samples (Fig. 3A), PGRP-LCx-expressing TTCs showed a striking phenotype characterized by small and shrunken cells (Fig. 3B). Measurement of TTC branches in PGRP-LCx-expressing larvae revealed a significant reduction in the number and length of branches compared with the controls (Fig. 3C, D). While cells in the control had about 14 branches, with a total length of about 1550 µm, TTCs in PGRP-LCx^OE insects had seven branches at the most, and the length did not exceed 514 µm. This means that the branched surface of TTCs expressing PGRP-LCx was reduced to less than one-third of that in controls. To investigate whether this cellular phenotype has an impact on physiological oxygen supply, we exposed larvae to low oxygen levels (2–3 %). Leaving the lawn under hypoxic conditions is the normal response of larvae, and this response can be monitored over time. A significantly higher percentage of larvae with TTCs overexpressing PGRP-LCx showed the escape phenotype after 15–25 minutes of exposure (Fig. 3E). Activation of the Imd pathway by ectopic expression of PGRP-LCx clearly impairs TTCs, not only with respect to induction of the shrunken phenotype, but also to the functionality of these cells (i.e., the enhanced response to hypoxic conditions). To elucidate whether this shrinking is part of an apoptotic cell program, we analyzed expression of two apoptotic factors: head involution defective (hid) and reaper (rpr) (19, 20). We targeted expression of these factors to TTCs using the DSRF-Gal4 driver line. Larvae expressing hid; rpr showed a strong response to hypoxia by leaving the lawn at the very early larval stages; these larvae did not survive (Fig. 3F). Since PGRP LCx-expressing larvae survived and did not show this extreme phenotype, we examined Tak1 (a downstream component of the Imd pathway). Expression of Tak1 in TTCs also resulted in early larval death, similar to expression of hid; rpr (Fig. 3F). Dissection of hid; rpr-expressing insects revealed a complete absence of TTCs, with only rudimentary cell remnants connected to the intestine (Fig. 3G, H). By contrast, overexpression of PGRP-LCx in TTCs resulted in similar cellular alterations, although the cells remained alive. Nevertheless, the overall phenotype was strikingly similar to that of apoptotic terminal cells (Fig. 3I).

Fig. 3:

Drosophila death caspase-1 (Dcp-1) is orthologous to mammalian Caspase-3, and is an established marker of apoptosis (21). Immunohistochemical analysis using an anti-Dcp-1 antibody (Fig. 3J, K) revealed a lack of Dcp-1 in the control; however, strong GFP staining of TTCs was visible (Fig. 3J-J’’). TTCs expressing PGRP-LCx were positive for cleaved Dcp-1, whereas counterstaining for GFP revealed reduced expression (Fig. 3K-K’’). These data clearly indicate that TTCs are undergoing apoptosis.

PGRP-LCx-induced terminal cell death is mediated by JNK signaling

The Drosophila Imd signaling pathway, which classically leads to activation of the Nf-κB factor Relish, is homologous to the TNF-α signaling pathway in mammals. Similar to the TNF-α pathway, the Imd pathway branches out to the JNK signaling pathway; the branch point is Tak1 (Fig. 4A). To test whether the Nf-κB or the JNK branch is necessary to translate PGRP-LCx activation in TTCs into triggering of apoptosis, we expressed these downstream components in flies, which were then phenotyped. The number and length of the terminal branches were measured and compared with those in control TTCs, and with the phenotype produced by PGRP-LCx expression (Fig. 4B–F). Ectopic expression of an activated Relish allele had no effect on the TTC branching phenotype (Fig. 4B, E, F). Expression of a constitutively active form of dJNKK, hemipterous (hep^CA), resulted in a phenotype similar to that caused by expression of the upstream receptor PGRP-LCx (Fig. 4C, E, F). Branching of TTCs was reduced significantly, and did not differ significantly from that of TTCs with PGRP-LCx expression. Overexpression of the wild-type JNK, basket (bsk^OE), also resulted in reduced branching when compared with the control (Fig. 4D–F). These results suggest that JNK activation is responsible for the TTC phenotype. To further support this hypothesis, the cells were stained for the Nf-κB factor Relish (Rel) and for the phosphorylated JNK (pJNK) basket (Fig. 4G, H).

Fig. 4:

Staining of the transcription factor Relish was observed throughout intestinal tissue, as well as in TTCs, from controls, and in TTCs expressing PGRP-LCx; however, we observed no translocation of Relish to the nucleus (Fig. 4G, arrow). In contrast to controls, PGRP-LCx- expressing cells showed distinct nuclear staining by an anti-pJNK antibody (Fig. 4H, arrow).

Moreover, nuclear pJNK was also observed in the nuclei of surrounding intestinal cells. This suggests that intestinal cells experience some additional JNK-mediated stress.

JNK-mediated TTC damage can be rescued by AP-1 and foxo depletion

The Drosophila JNK signaling pathway induces apoptosis (22, 23). The canonical transcription factors (TFs) AP-1, foxo and Ets21C act downstream of Tak1 and JNK to initiate transcription of target genes (Fig. 5A). Activation of each of these factors can lead to expression of apoptotic genes (24–27). To evaluate whether one of these TFs is responsible for the PGRP-LCx^OE-induced phenotype, we silenced each of them. This experimental approach required the use of different controls with the PGRP-LCx^OE cassette located on different chromosomes (Fig. 5F, G; red and blue). First, we used a dominant-negative form of Tak1 (Tak1^DN) and bsk (bsk^DN) together with PGRP-LCx^OE to confirm that PGRL-LC-induced apoptosis of TTCs is mediated by JNK. Concurrent expression of PGRP-LCx^OE with either Tak1^DN or bsk^DN rescued the apoptotic phenotype (Fig. 5B, F, G). Fos and Jun proteins form homo- or heterodimers that act as AP-1 transcription factors, also a canonical target of the JNK pathway. To target AP-1, we focused on the Drosophila Fos ortholog kayak (kay). We co-expressed a dominant-negative kay allele (kay^DN) together with PGRP-LCx^OE in TTCs. Although kay^DN rescued the PGRP-LCx^OE phenotype with respect to the number and length of branches, it was not complete (Fig. 5C, F, G; red). Co-expression of foxo^RNAi rescued the PGRP-LCx^OEphenotype to the same level as kay^DN (Fig. 5D, F, G; red). RNAi of Ets21C failed to rescue the PGRP-LCx^OE phenotype (Fig. 5E, F, G; blue); rather, it worsened the phenotype with respect to the number and length of branches.

Fig. 5:

Because depletion of the TFs foxo and kayak rescued the TTC phenotype induced by PGRP- LCx overexpression, we next tested whether upregulation of either induces a similar TTC phenotype (Fig. 5H–K). Overexpression of foxo led to a slightly reduction in the number and length of the branches (Fig. 5 H–J), while overexpression of a combination of kay and Jra, which drives concurrent overexpression of both AP-1 components, led to a severe reduction in number and length of the branches, to levels below that induced by PGRP-LCx overexpression (Fig. 5 H–J, K). Therefore, we asked whether foxo and AP-1 are present and active in TTCs. To do this, we used an AP-1 responsive TRE-RFP reporter line to detect AP-1 activity in TTCs (28). We observed AP-1 activity in wild-type TTCs, and even stronger activity upon overexpression of PGRP-LCx (Fig. 5L, M). When we used a Gal4 line with a foxo promoter, we observed strong foxo promoter activity in all TTCs (Fig. 5N). Thus, both TFs appear to be expressed and functional in TTCs.

Foxo controls TTC branching under normal conditions and under conditions of oxygen deprivation

A previous study reported a role for foxo in maintaining homeostasis of tracheal epithelial cells (13). One of the main features that ensures full functionality of TTCs under changing conditions is the ability to adapt to local hypoxia to maintain the supply of oxygen to target tissues; therefore, we subjected wild-type larvae and larvae with foxo-RNAi in TTCs to mild hypoxia (5% O₂) and measured their branching ability (Fig. 6). As expected, wild-type TTCs responded with increased branching (Fig. 6 A, B, E). The foxo^RNAi TTCs already had an increased number of branches under control conditions (21% O₂; Fig. 6C–E), and the number of branches did not increase further after oxygen deprivation; this suggest that foxo is required for appropriate responses to hypoxic conditions (Fig. 6D, E).

Fig. 6:

In summary, the data suggest that TTCs differ from the other parts of the tracheal system in terms of immune pathway activation (Imd/JNK); this is a mechanism that circumvents cell death and prevents impairment of functionality (Fig. 7A). The Imd pathway receptor PGRP- LCx does not appear to be expressed in TTCs, indicating that no downstream signaling takes place (Fig. 7B). Ectopic expression of PGRP-LCx exclusively in these cells leads to JNK- mediated cell death; however, all downstream components must be present and functional.

Fig. 7:

We have shown that depletion of the JNK-associated TFs AP-1 and foxo can rescue the PGRP-LCx-mediated phenotype. Both are present under physiological conditions, and their overexpression can induce a severe TTC phenotype. In addition, the universal transcription factor foxo, which is not only associated with JNK signaling, plays a role in branching of TTCs under normal and hypoxic conditions, indicating its importance for TTC homeostasis.

Discussion

Activation of the immune system is a double-edged sword in that although it fights infection, it can also damage the organism’s tissues (29); therefore, appropriate regulation and restriction of the immune system is of prime importance. Tight regulation of the immune system is essential when delicate, sensitive tissues are involved, or when immune-relevant signaling pathways are also needed for other cellular functions. Tracheal TTCs appear to fall into both categories; they are susceptible due to their specific structure but require foxo signaling to maintain structural plasticity. Here, we found that TTCs, in contrast to the other epithelial cells in the trachea, barely respond to bacterial infection. Immune-privileged status might protect TTCs from damage caused by a strong immune response, a strategy that also protects other organ systems (30–32). Here, we found that the entire tracheal epithelium expressed the transmembrane receptor PGRP-LC, but not TTCs. We propose that the lack of PGRP-LCx expression by TTCs is the reason for their lack of an immune response. This hypothesis is supported by the observation that targeted overexpression of PGRP-LC in TTCs induced an immune response, suggesting that all other pathway components are active in these cells. Since targeted overexpression of PGRP-LC leads to degeneration and, ultimately, apoptosis of these cells, this finding is particularly interesting. Targeted induction of apoptosis in TTCs by hid and rpr induced a similar, but more severe, phenotype; indeed, these larvae die after complete ablation of the cells.

Herein, we show that A) TTCs respond to a strong immune response by undergoing apoptosis, which leads to the death of the animal; and B) that immune activation is prevented by a lack of expression of the central receptor of the Imd pathway, PGRP-LCx. To better understand the first finding, we needed to elucidate the mechanism of apoptosis induction on the one hand and to identify mechanisms that still protect against infection on the other. Activation of the Imd pathway by ectopic expression of PGRP-LCx throughout the trachea leads to JNK-dependent meta- and hyperplasia of the trunk cells (13). The JNK pathway, as well as activation of its canonical TFs, is associated with both proliferation and apoptosis. The intertwining of the Imd pathway with the JNK pathway occurs at the level of Tak1 and is a general organizational principle within these signaling pathways. This architecture means that strong activation of the Imd pathway in these cells also triggers the multifaceted JNK pathway. Downstream of JNK (bsk in Drosophila), we found that foxo and AP-1 are both necessary and sufficient for the apoptotic phenotype in TTCs. AP-1 is a classical TF that is activated by JNK signaling. The reason underlying the incomplete rescue is unknown but might be due to a specific role for foxo in this context.

We also demonstrated that depleting foxo rescues the apoptotic phenotype, at least in part. Although the effects triggered by foxo depletion or foxo overexpression were statistically significant, they were certainly not as strong as those triggered by AP-1. This difference in efficiency is because the tools used to manipulate foxo (RNAi and overexpression) are much weaker than those used to manipulate AP-1 signaling. For example, in the latter case, a dominant-negative form of kay and a fully active heterodimer of Jra and kay were available; in the case of foxo, the overexpression allele does not give rise to an activated version of foxo, and RNAi is generally less effective than dominant-negative alleles. In the tracheal cells of the dorsal trunk, and the primary and secondary branches, foxo plays a role in tracheal remodeling downstream of Imd pathway activation (13). It seems reasonable then that foxo would also be activated in TTCs downstream of Imd signaling, which would in turn disrupt branching control and thus TTC functionality, as well as the ability to respond to environmental changes. Our central hypothesis derived from these observations is as follows: since foxo is essential for structural plasticity, and this property is critical for the survival of the organism, foxo should not be activated by other signals such as those from the Imd pathway. Foxo is activated not only by JNK signaling but also is a canonical member of the insulin signaling pathway. Linneweber and colleagues showed that insulin signaling, and especially the insulin receptor, is necessary for TTCs to respond (i.e., with reduced branching) to changing nutritional conditions such as low protein concentrations in food (15). Moreover, different insulin receptor alleles affect the size of TTCs (16). Here, we showed that foxo is also part of this signaling pathway and that it is (presumably) required to translate different insulin signaling activities into structural changes in TTC branching. A very similar scenario for FoxO signaling is operative in endothelial cells (ECs). Insulin and VEGF signaling in human ECs targets FoxO to control its transcriptional activity directly via PI3K/Akt signaling (35, 36).

Following our discussion of why it is sensible and necessary to omit this vital part of epithelial immunity from TTCs, we will focus on alternative strategies that operate to ensure that the cells are protected against infection. The strong immune response of the tracheal epithelium may be sufficient to protect the trachea, making an immune response in distant TTCs unnecessary. This strong and efficient immune response in the airway epithelia includes contact structures that connect the tracheal system with the outside world: the dorsal trunks and the primary and secondary branches (8, 37–39). TTCs reside at the very ends of these systems, so potential pathogens would have to travel a long distance and escape powerful immune responses before reaching the TTCs. In addition, the diameters of 0.1–1 µm are so small that most bacteria will not gain access (3).

The mechanisms underlying immune privilege vary. The same is true for the mechanisms that dampen the immune response. Here, we found that the dampening of the immune response was almost complete (based on non-expression of the proximal receptor within the canonical signaling pathway controlling epithelial immunity). A similar strategy was reported in a subpopulation of intestinal epithelial cells, called enterocytes that do not express the Imd receptor PGRP-LCx (40). It was hypothesized that this prevents chronic MAMP-mediated activation of innate epithelial immune responses in these cells, which are constantly exposed to gut bacteria. Organs such as the liver, kidney, lymphoid organs, brain, and lung are immunotolerant (41). Relevant immune pathways that induce apoptosis are dampened to different extents. In Drosophila epithelia, this dampening applies primarily to the Imd pathways because Toll signaling is not functional in these structures (8, 42). Interestingly, the mammalian homolog of the Imd pathway, the TNF-α pathway, also invokes these protective strategies. This can be seen in ECs, which are the first cells to meet pathogens in the circulation and share some characteristics with TTCs. For example, they express several PRRs, as well as secrete pro-inflammatory cytokines to initiate immune responses (43). To prevent EC death and apoptosis, protective genes and negative feedback loops become active when immune pathways are triggered (44). For example, VEGF/VEGFR signaling is immunosuppressive (45) because it inhibits TNF-α induced apoptosis of ECs; VEGF inhibits secretion of TNF in a concentration-dependent manner (46). For example, during angiogenesis and VEGFR activation, ECs express the anti-apoptotic gene surviving, which reduces caspase-3 activity and inhibits TNF-α-mediated apoptosis (47). The anti-apoptotic gene A20, when expressed in ECs, protects them from TNF-α-induced cell death, and inhibits inflammatory responses triggered by NF-κB activation (48). Trabid, the equivalent of A20 in flies, is a negative regulator of the Imd signaling pathway (49). TNF-α signaling, which induces death of various cell types, is a classical apoptotic pathway. TNF-α-induced apoptosis is highly regulated in ECs, which resist TNF-α-induced and physiological inflammatory apoptosis via TAK1 (50), which also regulates necroptosis and metastasis of ECs (51).

The branching process in TTCs is comparable with that observed during angiogenesis, in which the outgrowth of new capillaries is regulated by VEGF/VEGFR signaling. Angiogenesis is influenced by events that induce survival or apoptosis in ECs; it can only be maintained by EC survival or inhibition of apoptosis mediated; for example, by growth factors such as VEGF/VEGFR and PI3K/Akt (52). Expression of the Ig-family member CD31 by ECs prevents cell death and renders them immune privileged (53). When the TNF-α pathway is activated in ECs, CD31 is also activated; this in turn activates the Erk/Akt pathway and subsequent exclusion of FoxO3 from the nucleus, leading ultimately to inhibition of apoptosis in vitro. FoxO factors play a central role in controlling cell fate (i.e., apoptosis or proliferation). Constitutively active FOXO3a promotes EC apoptosis by downregulating protective factors, whereas dominant-negative FoxO protects ECs from apoptosis (54). In line with this observation, loss of mFoxO1 from murine ECs leads to uncontrolled overgrowth and hyperplasia. Moreover, FoxO controls endothelial quiescence by reducing glycolysis and mitochondrial respiration (55). Active FoxO inhibits EC migration, whereas its silencing has the opposite effect (56). By coincidence, we discovered a similar role for foxo in Drosophila larval TTCs. We showed that foxo controls the branching of TTCs. Overexpression of foxo led to reduced branching, whereas foxo depletion increased TTC branching. We hypothesize that TTCs need to avoid activation of the Imd pathway because they depend on foxo to maintain full functionality in response to changing environmental conditions. The exact role of foxo in TTCs, and the signaling pathways involved, remain to be elucidated. Our study highlights the importance of Drosophila TTCs as a model for human ECs and thus may provide information that is useful for angiogenesis- related research.

The data presented herein demonstrate how immune privilege is necessary to maintain the functionality of specific cell types. In the case of TTCs, which should mount potent immune responses, the architecture and interconnectivity of the Imd (TNF-α) and JNK pathways mean that cell fate is inevitably linked to foxo. This is, however, incompatible with the role of foxo in controlling the structure and functionality of these cells under changing conditions. For these reasons, switching off the Imd pathway is the only solution to this dilemma.

Material and methods

Fly lines and husbandry

Flies were raised on corn meal medium at 25°C. Tracheal humoral immune responses were tracked using Drs-GFP (BDSC_55707), which allows measurement of the GFP-tagged promotor activity of the antimicrobial peptide gene Drosomycin. The expression pattern of PGRP-LCx was analyzed by crossing PGRP-LCx-Gal4 (BDSC_77776) flies with UAS-GFP (BDSC_52262) flies. Tracheal driver lines ppk4-Gal4 (57) and btl-Gal4; UAS-GFP (Leptin Group, EMBL Heidelberg), or TTC driver line UAS-GFP; DSRF-Gal4 (58)), were crossed to the corresponding responder lines UAS-PGRP-LCx(III) (Kathryn Anderson, New York), UAS-PGRP-LC(II) (BDSC_30918), UAS-PGRP-LE (Dipt.-lacZ,UAS-Flag-PGRP-LE/CyO, Shoichiro Kurata, Sendai), UAS-Rel68 (BDSC_55777), UAS-Tak1 (BDSC_58810), UAS- hid;rpr (UAS-hid; rpr; Christian Wegener, Würzburg), UAS-bsk^OE (BDSC_9310), TRE-RFP attP40 (Chatterjee & Bohmann 2012), UAS-bsk^DN (BDSC_44801), UAS-kay^DN (BDSC_7214), UAS-Ets21C^RNAi (BDSC_39069), UAS-foxo^RNAi (BDSC_27656), UAS-kay^OE (BDSC_7213), and UAS-Jra^OE (BDSC_7216), or w¹¹¹⁸ (for control crossing, BDSC_5905).

Infection experiments

Natural infection of larvae was performed as described (10) (9). The gram-negative bacterium Erwinia carotovora (Ecc-15, 2141) was cultured overnight at 30°C in LB broth. The bacteria were pelleted by centrifugation for 20 min at 3200 × g, resuspended in PBS, and absorbance at OD₆₀₀ measured. Next, 200 µl of the bacteria solution (OD₆₀₀ = 160) was dropped into a vial containing developing 2^nd or early 3^rd instar larvae (3 days after egg laying) in standard cornmeal medium. After 24 h, the 3^rd instar larvae were used for microscopy. The larvae were heat killed in a drop of glycerol at 70°C for 20–25 s and arranged on a microscopy slide with the dorsal side facing up. GFP expression in the tracheal DB and TTC between the dorsal trunks was analyzed.

Branching analysis

The 3^rd instar larvae were washed in PBS and deposited on a slide in a drop of glycerol. The larvae were heat killed at 70°C for 20–25 s and arranged dorsal side up before microscopy. Images of the TTCs were taken using the GFP channel (at 10× or 20× magnification). The right TTC from the 3^rd dorsal segment was chosen for analysis. Images were taken by an Axio Imager.Z1 (Zeiss, Munich, Germany) with ApoTome in Z-stack mode to capture all TTC branches. Measurement of the number and length of TTC branches was undertaken using the ImageJ plugin NeuronJ (59). The branches were measured, as described previously (60).

Immunohistochemistry and microscopy

For microscopic and immunohistochemical analyses of intestinal TTCs, the gut of 3rd instar larvae (along with the connected TTCs) was dissected. As an alternative, a fillet dissection for the cuticle-attached TTCs was used for microscopy. The tissue was dissected in PBS, fixed with 4% PFA, and blocked with 5% normal goat serum (Sigma-Aldrich, Munich, Germany) before overnight incubation at 4°C with an anti-GFP antibody (1:200, DSHB, Iowa City, USA). Cells were also stained with anti-Dcp-1 (1:200, Cell Signaling Technology, Danvers, USA), anti-pJNK (Promega, V7931), and anti-Relish-C (DSHB, 21F3). Images were taken with an Axio Imager.Z1 (Zeiss, Munich, Germany) with ApoTome in the Z-stack mode.

Hypoxia sensitivity assay

The 3^rd instar larvae were washed with PBS and placed into a new food vial with a scratched surface; 20 larvae were used per replicate. The vials were incubated for at least 10 min until all larvae were buried in the food, and then deposited in a sealed desiccator. Nitrogen gas was introduced until the O₂ concentration reached 2–5%. This O₂ level was maintained for 25 min. The number of larvae outside of the food was counted every 5 min.

Acknowledgements

We would like to thank Britta Laubenstein and Christiane Sandberg for their excellent technical assistance.

Additional information

Funding

This work was supported by Christian-Albrechts University, Kiel, as part of the Leibniz Campus EvoLung (TR, JB), by the International Max-Planck Research School for Evolutionary Biology (RR), and by the Deutsche Forschungsgemeinschaft (DFG) (as part of the CRC 1182 (project C2))(TR).

Author contributions

Conceptualization: JB, TR

Methodology: RR, JB

Investigation: JB, RR, JH

Visualization: JB, RR, JH

Funding acquisition: TR

Project administration: TR

Supervision: TR, JB

Writing – original draft: JB, TR

Writing – review & editing: TR, JB, RR

Competing interests

The authors declare no conflict of interest.