Remodelling of oxygen-transporting tracheoles drives intestinal regeneration and tumorigenesis in Drosophila

Abstract

The Drosophila trachea, as the functional equivalent of mammalian blood vessels, senses hypoxia and oxygenates the body. We show that the adult intestinal tracheae are dynamic and respond to enteric infection, oxidative agents, and tumors with increased terminal branching. Increased tracheation is necessary for efficient damage-induced intestinal stem cell (ISC)-mediated regeneration and sufficient to drive ISC proliferation in undamaged intestines. Gut damage or tumors induce Hif-1α/Sima, which stimulates tracheole branching via the FGF(Branchless/Bnl)/FGFR(Breathless/Btl) signalling cascade. Bnl/Btl signalling is required in the intestinal epithelium and the trachea for efficient damage-induced tracheal remodelling and ISC proliferation. Chemical or Pseudomonas-generated ROS affect the trachea directly and are necessary for branching and intestinal regeneration. Similarly, tracheole branching and the resulting increase in oxygenation are essential for intestinal tumor growth. We have identified a mechanism of tracheal-intestinal tissue communication, whereby damage and tumors induce neo-tracheogenesis in Drosophila, a process reminiscent of cancer-induced neoangiogenesis in mammals.

Introduction

The Drosophila trachea, a network of ramified oxygen-transporting tubes, is the functional equivalent of the mammalian vasculature^1,2. It encompasses highly plastic cells, the terminal tracheal cells (TTCs) that extend cytoplasmic tubular processes, the terminal branches (TBs) or tracheoles, towards oxygen-needy tissues. Developmental cues drive TTC formation, and oxygen regulates TB remodelling: low oxygen (hypoxia) induces, whereas increased oxygen (hyperoxia) suppresses tracheole formation^3–5. Therefore, the TTCs are oxygen sensors equivalent to the mammalian tip cells⁶. Hypoxia controls TB formation via the Hypoxia-inducible Factor-1α(Hif-1α)/Similar(Sima)⁴. Sima activates expression of non-tracheal FGF/Branchless(Bnl) and tracheal FGFR/Breathless(Btl) to orchestrate TTC remodelling in Drosophila larvae⁷, similar to Hif-1α in mammalian angiogenesis⁸. However, our knowledge about the adult tracheal maintenance and remodelling is limited.

The Drosophila adult midgut is an excellent system to study stem cell regeneration and tissue communication upon damage and aging^9–12. The midgut is an apicobasally-polarized epithelium comprised of intestinal stem cells (ISCs), and their progeny, the transient enteroblasts (EBs) and pre-enteroendocrine cells (pre-EEs), which differentiate to lumen-facing absorptive enterocytes (ECs) and secretory enteroendocrines (EEs), respectively^13–15. Visceral muscles (VM) ensheath the epithelium, support midgut peristalsis, supply a basement membrane and contribute signals to the ISC niche^16,17. Outside the VM lie the visceral tracheae that branch extensively and penetrate the VM to access the intestinal epithelium¹⁸ to supply it with oxygen internalized via the spiracles^19,20. Enteric infection damages the intestinal epithelium and induces ISC-mitoses, that facilitate regeneration^21–23. Ingestion of ROS-like chemicals (hydrogen peroxide (H₂O₂), Paraquat (PQ)), and tissue-damaging agents (bleomycin, dextran sulfate sodium) also induce ISC-mediated gut regeneration^24–26. Regeneration is controlled by paracrine niche signals activating conserved signalling cascades, including the JNK, Jak/Stat, Wnt/Wg, Hippo/Yki and EGFR ^10,27. The intestinal tracheae contribute niche signals for ISC maintenance¹⁸, respond to nutritional cues²⁸, and associate with metastatic tumors²⁹. However, a mechanistic study on the role of the trachea in gut regeneration and tumorigenesis has not yet been described.

We have investigated the signals mediating communication between the trachea and the intestinal epithelium. We show that the adult midgut-associated tracheae increase their terminal branching in response to infection, oxidative stress, and tumors. Increased tracheation is necessary for robust ISC regeneration and tumor growth. By comparing responses in normoxia and hypoxia, we show that the availability of oxygen, transported to the epithelium via the TTCs, is essential for ISC-mediated regeneration.

Results

Intestinal tracheal remodelling is necessary for ISC mitosis

The trachea actively regulates the metabolic status of adult tissues^18,28. To assess if damaging and regeneration-promoting agents induce intestinal tracheal branching, we subjected wild-type flies to different stresses known to induce ISC proliferation by damaging the intestinal epithelium^21–24 (Fig. 1a–l). We fluorescently-labeled the trachea either via the Gal4-UAS system³⁰ using the trachea-specific btl-Gal4>UAS-srcGFP³¹ or via the QF-QUAS system³² using the QF6>QUAS-mtdTomato that exhibits tracheal expression (Fig. 1; Extended Data Fig. 1). The flies were infected orally with pathogenic bacteria (Pseudomonas aeruginosa (P.a.), Pseudomonas entomophila (P.e.)) or fed with oxidative agents, such as H₂O₂ or PQ. All treatments led to significantly increased esg>GFP-positive midgut cells (Extended Data Fig. 1a) and ISC proliferation (Fig. 1h,j,l, and Extended Data Fig. 1b). Strikingly, Pseudomonas infection and oxidative agents also induced an increase in the midgut-associated tracheoles (Fig. 1a–f; Extended Data Fig. 1c–e). We quantified tracheal coverage using confocal images with fluorescently-labeled trachea or bright-field images of detergent-free midgut preparations. In the latter, the air maintained in the tubes offered contrast. Then, we measured the total length of the tracheae overlaying the midgut using NeuronJ (an ImageJ plugin) and we normalized to the gut surface. In some cases (e.g. Fig. 1k), we measured the number of TBs in specific midgut regions. Image quantification confirmed the imaging observations (Fig. 1g,i,k; Extended Data Fig. 1c). To exclude the possibility that extensive tracheal branching was caused by increased TTC numbers, we measured the TTCs in uninfected and P.a.-infected midguts. We found that infection did not alter the number of midgut TTCs (Fig. 1m). We did not observe any intestinal TTC undergoing mitosis either in control or in damaged conditions. Thus, direct epithelial damage caused by infection or ROS increases ISC proliferation and reorganizes the adult intestinal tracheoles.

Figure 1. Remodelling of visceral TTCs covering the midgut is associated with ISC proliferation.

a-f. The adult intestinal trachea is increased upon P.a. infection (a, b), administration of H₂O₂ (c, d) and P.e. infection (e, f). The intestinal TTCs were labeled with btl-Gal4>UAS-srcGFP in a-d and with QF6>QUAS-mtdTomato in e-f, and DAPI (blue) labeled all nuclei in a-f. The R5 midgut region was imaged in a-b, the R2 region in c-d and the R4-R5 region in e-f. Single channel images of the trachea are shown in a’-f’. g-l. Quantification of TTC branching and mitosis upon P.a. infection (g, n=10,9 and h, n=15,18), administration of H₂O₂ (i, n=10,9 and j, n=15,11) and P.e. infection (k, n=14,15 and l, n=7,15). m. Quantification of TTCs upon trachea-specific btl manipulation with or without P.a. infection (n=7 each). n. Tracheae (green) and cleaved Caspase-3 (red) upon trachea-specific btl knockdown. o. Quantification of mira-GFP-positive ISCs upon trachea-specific btl knockdown via trh-Gal4 (n=10,12). p-q. Quantification of TTC branching (p, n=7,9,10,10,10) and mitosis (q, n=35,12,8,10,10) upon trachea-specific btl manipulation with or without P.a. infection. r-u. Tracheae (green) of P.a.-infected R5 regions of the midgut in btl-Gal4>UAS-srcGFP control (r) and btl-Gal4>UAS-srcGFP, UAS-btl^DN-expressing (s), as well as control (t) versus btl-Gal4>UAS-srcGFP, UAS-λbtl-expressing flies (u). v. Quantification of midgut mitosis of flies heterozygous for btl pathway mutations with or without P. e. infection compared to controls (n=9,11,11,13,11,13,10,14,9,12). Scale bars: 75 μm in a-d, 60 μm in e-f, 37.5 μm in n, r-u. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

Since the FGFR/Btl pathway is integral to tracheal formation and remodelling during development^1,2, we reasoned that its manipulation might affect adult intestinal tracheole dynamics. To assess whether the FGFR/Btl is necessary in the trachea for maintenance and remodelling, we overexpressed a dominant-negative form of Btl, btl^DN, specifically in the adult trachea using btl^ts-Gal4. We noticed that files with trachea-specific btl inhibition had fewer intestinal TTCs in uninfected and P.a.-infected conditions (Fig. 1m). TTC apoptosis was evident by cleaved Caspase-3 staining (Fig. 1n). Importantly, trachea-specific btl knockdown did not alter the number of mira-GFP-positive ISCs (Fig. 1o) in uninfected midguts, confirming that the mitosis impairment was not due to fewer ISCs. Upon P.a. infection, flies expressing btl^DN in the adult trachea under btl^ts-Gal4 exhibited reduced tracheation relative to controls (Fig. 1r–s). When we overexpressed a constitutively-active form of Btl, λbtl, in the trachea, we found that it was sufficient to increase intestinal TTCs and TBs in uninfected flies (Fig. 1m,p,t–u). Therefore, trachea-specific increase and decrease of btl activity leads to increased and decreased tracheole density, respectively. To test whether intestinal tracheal remodelling is necessary for damage-induced regeneration, we assessed ISC mitosis in flies with trachea-specific manipulation of btl. We found that P.a.-infected flies expressing btl^DN in the trachea exhibited less mitosis, whereas uninfected flies expressing λbtl in the trachea exhibited increased mitosis (Fig. 1q). We confirmed the effects of btl on tracheogenesis and mitosis with two additional trachea-specific drivers (trh-Gal4 and dSRF-Gal4) (Extended Data Fig. 2). Furthermore, to underscore the necessity for the FGFR/Btl pathway in intestinal regeneration upon infection, we subjected heterozygote btl^MI03286, btl^f02864, bnl^MI01635 and bnl^e03115 mutants to P.e. infection and found significantly-decreased mitosis (Fig. 1v). These results indicate that FGFR/Btl receptor-mediated tracheal branching is critical for ISC regeneration.

FGF/FGFR controls tracheal remodelling and ISC proliferation

The Drosophila FGF, Bnl, is essential for the directional migration of terminal tracheal cells during embryonic development, and in larvae upon hypoxia^3,4,33. To test whether Bnl is involved in damage-induced intestinal tracheogenesis, we examined bnl induction upon different damages. Using bnl-Gal4>UAS-srcGFP, we monitored bnl expression in the adult midgut. We found that bnl was expressed in midgut epithelial cells and P.a., H₂O_2, or PQ feeding robustly increased the number of bnl-expressing cells (Fig. 2a–d; Extended Data Fig. 3). To confirm that bnl is induced transcriptionally, we performed RT-qPCR in midguts of wild-type w¹¹¹⁸ flies at the same time points used for reporter imaging. We found that bnl mRNA was induced by more than 1.5-fold by P.a. and PQ, but not H₂O₂ (Fig. 2e). The lack of bnl mRNA induction by H₂O₂ in whole midgut tissue is possibly due to the regional effect of H₂O₂ evident by the bnl reporter (Extended Data Fig. 3), which can affect the RT-qPCR sensitivity. To identify the specific cell types with bnl expression, we assessed co-localization of bnl-Gal4>UAS-srcGFP with esg-lacZ, Dl-lacZ, Su(H)-lacZ, and Prospero, which report progenitors (ISCs and EBs), ISCs, EBs and EEs, respectively. We also identified mature ECs by their large size. We found that bnl was expressed and induced in all the midgut epithelial cell types (Fig. 2a–d). To assess whether Bnl is involved in tracheal remodeling and mitosis, we knocked-down and overexpressed bnl in adult intestinal progenitors or ECs using esg^ts-Gal4 or Myo1A^ts-Gal4 (Fig. 2f–m; Extended Data Fig. 4a–d), respectively. Progenitor- and EC-specific bnl knockdown reduced tracheal coverage in P.a.- or P.e.-infected midguts (Fig. 2f,j; Extended Data Fig. 4a–b). Furthermore, the ISC proliferation rate upon P.a. (Fig. 2g,k) or P.e. infection (Extended Data Fig. 4c–d) was reduced, suggesting that bnl is necessary for regeneration. To test whether bnl overexpression is sufficient to induce tracheal remodelling in baseline conditions, we overexpressed bnl in intestinal progenitors or ECs. bnl overexpression was sufficient to induce tracheal remodelling (Fig. 2h,l), and it was also sufficient to induce ISC mitoses (Fig. 2i,m). Notably, the viability and number of progenitors upon progenitor-specific bnl knockdown remained unchanged (Fig. 2n–q; Extended Data Fig. 4e). Collectively, our data suggest that damage-induced bnl expression in the midgut epithelium is necessary for ISC proliferation and that bnl is sufficient for TTC remodelling and induction of ISC proliferation in the absence of damage.

Figure 2. The FGF/Bnl is expressed in the midgut epithelium and controls ISC proliferation and TTC remodelling.

a-b. Control (a) and P.a.-infected (b) bnl-Gal4>UAS-srcGFP midguts (R5 region) with esg-lacZ-labeled (red) ISCs and EBs, and DAPI (blue) staining all the nuclei. c-d. Control (c) and H₂O₂-fed (d) bnl-Gal4>UAS-srcGFP midguts (R2 region) with Prospero-labeled (red) EEs, and DAPI (blue) staining all the nuclei. Yellow arrowheads indicate bnl-positive ECs (big cells) and white arrowheads indicate bnl-positive esg-positive progenitors (a-b) and EEs (c-d). e. RT-qPCR for bnl mRNA expression in w¹¹¹⁸ whole adult midguts upon different stresses. The average and standard deviation of n=3 biological expreriments are plotted. f-g. Quantification of TTC branching (f, n=12,12,10,10,10,10) and mitosis (g, n=15,11,12,12,13,15) upon progenitor-specific bnl knockdown with or without P.a infection. h-i. Quantification of TTC branching (h, n=6 each) and mitosis (i, n=12,14) upon progenitor-specific bnl overexpression. j-k. Quantification of TTC branching (j, n=9,10,9,9,10,10) and mitosis (k, n=15,12,13,13,7,10) upon EC-specific bnl knockdown with or without P.a infection. l-m. Quantification of TTC branching (l, n=6 each) and mitosis (m, n=9,11) upon EC-specific bnl overexpression. n-p. Midguts (R5 region) of control (n) and esg^ts-Gal4 progenitor-specific bnl knockdown (o-p), stained for ISCs+EBs (green), cleaved Caspase-3 (red) and Topro-3 (blue). q. Quantification of esg-positive ISCs+EBs in control and progenitor-specific bnl knockdown (n=9,10,9). Scale bars: 37.5 μm in a-d, 75 μm in n-p. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

We did not observe btl outside the trachea or bnl outside the intestinal epithelium (Fig. 1a–d, Fig. 2a–d). Nevertheless, because the reporters might not precisely highlight the cells expressing the genes of interest, we tested btl and bnl requirements in the intestinal epithelium and the trachea, respectively. To assess if btl is necessary in intestinal progenitors or ECs, we inactivated btl by overexpressing btl^DN or btl^RNAi under esg^ts-Gal4 and Myo1A^ts-Gal4, respectively. We found that tracheal branching and ISC mitosis were impaired upon btl inhibition in the epithelium upon infection (Fig. 3a–d; Extended Data Fig. 4a–b,f–g). Importantly, the number of esg-positive progenitors remained unchanged upon progenitor-specific btl downregulation (Extended Data Fig. 4e,h). Thus, intestinal epithelial btl is necessary for mitosis and tracheal remodelling upon infection.

Figure 3. Damage-induced TTC remodelling and ISC mitosis require epithelial FGFR/Btl and tracheal FGF/Bnl.

a-d. Quantification of TTC branching (a, n=10 each, and c, n=10,10,10,9) and mitosis (b, n=31,27,25,30 and d, n=29,23,26,23) upon progenitor-specific (a-b) and EC-specific (c-d) btl knockdown with or without P.a. infection. e-f. Quantification of TTC branching (e, n=10 each) and mitosis (f, n=32,23,27,24,28,27) upon trachea-specific btl knockdown via btl^ts-Gal4 with or without P.a. infection. g-h. Quantification of TTC branching (g, n=7,9,7,10) and mitosis (h, n=10,10,9,8) upon trachea-specific btl knockdown via dSRF^ts-Gal4 with or without P.e. infection. i-j. Quantification of TTC branching (i, n=12,11) and mitosis (j, n=8,11) upon progenitor-specific λbtl overexpression. k-l. Quantification of TTC branching (k, n=12,11) and mitosis (l, n=12,8) upon EC-specific λbtl overexpression. m-n. Quantification of TTC branching (m, n=4,6) and mitosis (n, n=28,24) upon trachea-specific λbtl overexpression via btl^ts-Gal4. The flies were uninfected in i-n. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

To assess if bnl functions in the trachea, we silenced it in the adult trachea via the btl^ts-Gal4 using two independent RNAi lines. We found that tracheal bnl was necessary for damage-induced tracheogenesis and mitosis (Fig. 3e–f). Interestingly, trachea-specific bnl knockdown using dSRF^ts-Gal4 showed that tracheal bnl was required for infection-induced ISC mitosis, but had no effect on tracheal remodelling (Fig. 3g–h). Thus, bnl is necessary in the intestinal trachea for mitosis and damage-induced tracheole remodelling.

To test if epithelial btl is sufficient to induce ISC mitosis and tracheole remodelling, we overexpressed λbtl in progenitors or ECs using the esg^ts-Gal4 and Myo1A^ts-Gal4. Neither progenitor- nor EC-specific λbtl expression affected tracheal remodelling and ISC mitosis (Fig. 3i–l). To assess if tracheal bnl is sufficient, we overexpressed bnl under btl^ts-Gal4. We found that trachea-specific bnl overexpression was not sufficient to induce mitosis or tracheal remodelling (Fig. 3m–n). Thus, neither epithelial btl nor tracheal bnl are sufficient to induce mitosis and tracheal remodelling, indicating the requirement for other factors.

Hif-1α/Sima controls tracheogenesis and ISC proliferation

Sima, the Drosophila Hif-1α, controls Btl activity and tracheal remodelling in response to hypoxia in larvae⁴, and is induced upon epithelial damage and tumors^34,35. In mammalian cells, Hif-1α activity is controlled by ROS, produced by mitochondria in moderate hypoxia, and can inhibit the PHD/Hph proteins that promote Hif-1α degradation^36–38. In normoxia, Hif-1α activity can be controlled by exogenous ROS, e.g. H₂O₂, in cultured mammalian cells³⁹. To monitor midgut Sima activity, we reared flies carrying the hypoxia-inducible reporters⁴⁰ lactate dehydrogenase (ldh)-Gal4>UAS-nlsGFP or ldh-Gal4>UAS-srcGFP in normoxia upon different damages (Fig. 4a–d; Extended data Fig. 5). These Hif-1α/Sima-responsive reporters were expressed weakly in midgut epithelial cells and the trachea in baseline conditions. They were strongly induced in the epithelium and the trachea upon P.a. infection or oxidative damage (Fig. 4a–d; Extended data Fig. 5). To assess whether ldh was induced transcriptionally, we performed RT-qPCR in midguts of wild-type w¹¹¹⁸ flies under the same conditions. We found that P.a. and ROS induced ldh mRNA (Fig. 4m). Co-localization experiments showed that ldh was activated in ISCs, EBs and ECs, and rarely in EEs. Importantly, ldh-Gal4>UAS-srcGFP highlighted the terminal branches of the intestinal tracheae (Fig. 4a–d). Sima is, thus, induced in the intestinal epithelium and trachea in response to infection and oxidative stress, suggesting that it may act in both tissues to control tracheal remodelling and intestinal proliferation.

Figure 4. The Hif-1α/Sima is expressed in the midgut epithelium and the visceral trachea, and is necessary and sufficient for ISC proliferation and TTC remodelling.

a-b. Uninfected (a) and P.a.-infected (b) ldh-Gal4>UAS-srcGFP midguts (R5 region) with Prospero-labeled (red) EEs. c-d. Control (c) and H₂O₂-fed (d) midguts (R2 region) with Su(H)-lacZ-labeled (red) EBs. Purple arrowheads indicate ldh-positive Su(H)-positive EBs and white arrowheads indicate ldh-negative Prospero-positive EEs. e-f. Quantification of TTC branching (e, n=5,7,6,6) and mitosis (f, n=14,13,18,14) with or without P.a. infection in trachea-specific sima knockdown in the background of heterozygous sima^KG. g-h. Quantification of TTC branching (g, n=15,16) and mitosis (h, n=10,8) upon trachea-specific sima overexpression in baseline conditions. i-j. Quantification of TTC branching (i, n=18,18,8,7) and mitosis (j, n=24,22,11,11) with or without P.a. infection upon EC-specific sima knockdown in the background of heterozygous sima^KG. k-l. Quantification of TTC branching (k, n=8,10) and mitosis (l, n=9,7) upon trachea-specific sima overexpression in baseline conditions. m. RT-qPCR for ldh mRNA expression in w¹¹¹⁸ whole adult midguts upon different stresses. n=3 biological experiments. n. RT-qPCR analysis of btl, bnl and ldh upon EC-specific sima overexpression. n=3 biological experiments. o. RT-qPCR analysis of Dl, wg, upd, upd3, spi, vn, krn, and egr upon EC-specific sima overexpression. n=3 biological experiments. The average and standard deviation are plotted in m-o. Scale bars: 37.5 μm in a-d. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided in f-k, m-o and U-tested in e,l): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

To assess whether Sima induction is necessary for tracheal remodelling and ISC mitoses, we performed tissue-specific loss- and gain-of-function tests. To achieve effective sima silencing, we inactivated sima by tissue-specific RNAi in a heterozygous sima^KG07607 mutant background⁴¹, in the tracheal cells or ECs. Upon P.a. infection, both tracheal coverage and ISC mitosis were reduced by sima depletion (Fig. 4e–f,i–j; Extended data Fig. 6). To assess if sima is sufficient for tracheal remodelling and intestinal regeneration in baseline conditions, we overexpressed sima⁴ specifically in adult tracheal cells using btl^ts-Gal4, and in progenitors and ECs using esg^ts-Gal4 and Myo1A^ts-Gal4, respectively. Overexpression of sima was sufficient to induce tracheogenesis and ISC proliferation without infection (Fig. 4g–h,k–l). Interestingly, sima gain-of-function increased expression of the bnl and btl transcripts, suggesting that both are sima targets (Fig. 4n). Morevover, EC-specific sima overexpression led to increased expression of ISC mitosis regulators, such as genes encoding the Notch ligand Delta (Dl), the Jak/Stat ligand Unpaired 3 (Upd3), and the EGFR ligands Spitz (Spi) and Vein (Vn) (Fig. 4o). Thus, Sima controls midgut TTC remodelling and damage repair acting both in the epithelium and the trachea.

Increased tracheoles induce ISC mitosis only in normoxia

In addition to bacteria and ROS, hypoxia is known to activate Hif-1α, and to affect remodelling of oxygen-carrying blood vessels and trachea, in mammals and insects, respectively^42,43. To assess if oxygen availability affects intestinal tracheogenesis, we subjected wild-type flies to hypoxia (5% O₂) for 12 hrs and compared them with flies reared in normoxia (21% O₂). We found that intestinal tracheal coverage, but not TTC number, was significantly increased in hypoxia-treated flies (Fig. 5a–c, g). Contrary to our expectation, however, we found that intestinal mitosis of wild-type flies was significantly reduced in hypoxia (Fig. 5d). Importantly, hypoxia did not cause progenitor apoptosis (Fig. 5e–f), or reduction in esg-positive cells (Fig. 5h). Hypoxia induced bnl in epithelial cells, and sima in the epithelium and the trachea (Fig. 5i–l). In addition, flies subjected to hypoxia for 4 hrs exhibited increased midgut transcription of bnl and ldh. Midgut btl was induced only after 12 hrs of hypoxia by approximately 2-fold (Fig. 5q). Hypoxia, therefore, induces intestinal tracheal remodelling, but it does not promote ISC mitosis, uncoupling the two processes.

Figure 5. Hypoxia induces TTC branching, but not ISC proliferation.

a-b. Brightfield images of the midgut TTCs (R5 region) of wild-type (w¹¹¹⁸) uninfected flies reared in normoxia (21% O₂) vs. hypoxia (5% O₂). c-d. Quantification of TTC branching (c, n=10 each) and mitosis (d, n=11,13) in normoxia vs. hypoxia. e-f. esg-Gal4>UAS-srcGFP (ISCs+EBs, green) midguts (R5 region) of flies reared in normoxia (e) and hypoxia (f) stained for cleaved Caspase-3 (red) and Topro3 (blue) staining all nuclei. g-h. Quantification of TTCs (g, n=18, 14) and esg-lacZ+ progenitors (h, n=12,13) in normoxia and hypoxia. i-j. Fluorescence images showing normoxia- (i) and hypoxia-reared (j) bnl-Gal4>UAS-srcGFP (Bnl reporter, green) midguts with Dl-lacZ (red) labeling the ISCs and DAPI (blue) staining all nuclei. Yellow arrowheads indicate bnl-positive and Dl-positive progenitors and white arrowheads indicate bnl-negative and Dl-positive progenitors. k-l. Fluorescence images showing normoxia- (k) and hypoxia-reared (l) ldh-Gal4>UAS-srcGFP (Sima reporter, green) midguts with esg-lacZ (red) labeling the ISCs and EBs and DAPI (blue) staining all nuclei. Yellow arrowheads indicate bnl-positive and esg-positive progenitors and white arrowheads indicate bnl-negative and esg-positive progenitors. m-n. Brightfield images of the midgut TTCs (R5 region) of wild-type (w¹¹¹⁸) P.a.-infected flies reared in normoxia (21% O₂) vs. hypoxia (5% O₂). o-p. Quantification of TTC branching (o, n=6,11,6,8) and mitosis (p, n=19,14,11,11) of uninfected and P.a.-infected flies in normoxia and hypoxia. q. RT-qPCR analysis of btl, bnl and ldh in normoxia vs. hypoxia. The average and standard deviation of n=3 biological experiments are shown. r-s. Quantification of TTC branching (r, n=12,10,10,17) and mitosis (s, n=13,10,14,17) of untreated vs. H₂O₂-fed flies in normoxia and hypoxia. All scale bars: 75 μm. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided, and U-tested for o): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

To assess the relationship between infection and hypoxia, which induce tracheal remodelling but affect ISC mitoses oppositely, we infected flies with P.a. while subjecting them to hypoxia. Although tracheal branching was increased in treated flies (Fig. 5m–o), hypoxia significantly suppressed mitosis upon infection (Fig. 5p). To exclude the possibility that hypoxia affected the damaging capacity of bacteria, we tested tracheal branching and mitosis of H₂O₂-treated flies upon hypoxia. Regardless of the increased tracheal coverage, hypoxia suppressed H₂O₂-induced mitoses too (Fig. 5r–s). Therefore, increased tracheae induce ISC proliferation only in the presence of oxygen.

ROS induce ISC proliferation via the tracheal niche

Enteric infection phenocopies oxidative stress with regards to intestinal regeneration and tracheal remodelling. Pseudomonas are extracellular pathogens damaging tissues via secreted virulence factors. The P.a. quorum-sensing MvfR pathway is a master regulator of virulence. Genes involved in phenazine production, including phzS and phzM, control pyocyanin, a ROS-generating secreted virulence factor that provides the blue hue in the P.a. cultures^44,45. To assess the role of pyocyanin in intestinal regeneration and tracheogenesis, we fed wild-type flies with bacterial culture supernatants from wild-type P.a. (strain PA14) and MvfR, PhzM and PhzS mutants (isogenic to PA14⁴⁶). All mutants led to significantly-reduced tracheation (Fig. 6a–g) compared to PA14, accompanied by significant reductions in ISC mitoses (Fig. 6h). Since pathogenic bacteria induce ROS via the Dual Oxidase (Duox)^22,47, we assessed the requirement for Duox in P.a.-induced tracheal remodelling and ISC mitosis. To assess if epithelial or tracheal Duox affect regeneration, we silenced Duox in the ECs or the trachea via Myo1A^ts-Gal4 and btl^ts-Gal4, respectively. We found that Duox was necessary in the epithelium and the trachea for induction of tracheal remodelling and mitosis (Fig. 6i–l). Thus, bacterial pyocyanin and host Duox are necessary for tracheal remodelling and intestinal mitosis upon infection.

Figure 6. Stress-responsive increased TTC remodelling is ROS dependent.

a-e. Brightfield images of the midgut TTCs (R5 region) of wild-type (w¹¹¹⁸) flies fed on LB (a, negative control) and P.a. culture supernatants of the virulent wild-type PA14 strain (b, positive control), the quorum-sensing mutant MvfR (c), and the phenazine mutants PhzS (d) and PhzM (e). f. Schematic of the P.a. quorum-sensing and phenazine production pathway. g-h. Quantification of TTC branching (g, n=10 each) and midgut mitosis (h, n=12,11,11,11,10) upon P.a. culture supernatant feeding. i-j. Quantification of TTC branching (i, n=10,10,8,10) and midgut mitosis (j, n=22,23,10,21) upon EC-specific Duox knockdown with or without P.a. infection. k-l. Quantification of TTC branching (k, n=10,10,8,10) and midgut mitosis (l, n=42,32,36,36) upon trachea-specific (via btl^ts-Gal4) Duox knockdown with or without P.a. infection. m-o. Quantification of midgut mitosis (m, n=37,30), TTC number (n, n=10 each) and TTC branching (o, n=7,10) upon trachea-specific (via dSRF^ts-Gal4) Cat knockdown in uninfected flies. p. Quantification of midgut mitosis upon trachea-specific (via dSRF^ts-Gal4) Cat and Sod1 overexpression in P.e.-infected flies (n= 12,13,15). q-r. Quantification of TTC branching (q, n=7,9,10,10) and midgut mitosis (r, n=15,11,8,11) upon trachea-specific (via btl^ts-Gal4) Cat and Sod1 overexpression with or without P.a. infection. All scale bars: 75 μm. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

ROS induce ISC mitoses and cell turnover in the fly intestine^48,49. However, no report has indicated any direct effects of ROS on the intestinal trachea. We found that bacterial ROS-generators and chemical ROS affected the intestinal trachea (Fig. 1a–l, Fig. 6a–h and Extended data Fig. 1) via Sima. To address the role of tracheal ROS in damage-induced tracheal remodelling and mitosis, we genetically modulated ROS levels or activity directly in the trachea and measured tracheation and ISC proliferation. Trachea-specific knockdown of the antioxidant Catalase A (Cat) in uninfected conditions led to increased ISC mitosis, accompanied by increased tracheole branching without differences in TTC numbers (Fig. 6m–o). This suggests that ROS exist in tracheal branches and affect the epithelium. When antioxidants, Cat and Superoxide dismutase 1 (Sod1), were overexpressed in the adult trachea, the infection-induced ISC proliferation was severely restricted (Fig. 6p,r). Trachea-specific overexpression of Cat and Sod1 also led to significantly reduced tracheae (Fig. 6q). Our results suggest that flies remodel their intestinal tracheoles in response to ROS-mediated tissue damage that induces Sima in both the midgut epithelium and the trachea. In normoxia, increased tracheogenesis increases oxygen supply, which is necessary for efficient ISC regeneration.

ISC-derived tumors induce and require tracheal arborization

Neoangiogenesis is a hallmark of mammalian cancer⁵⁰. While differentiation of tumor cells into trachea has been suggested^34,51 and tumor-trachea association has been reported²⁹ previously, no study has examined the role of the Drosophila trachea in tumorigenesis. To investigate whether Drosophila ISC-derived tumor development associates with increased tracheal branching (“neo-tracheogenesis”), we generated MARCM clones expressing the Ras^V12 oncogene in the adult midgut. We observed increased tracheole density in the tumor compared to the surrounding tissue (Fig. 7a). Tracheal length measurements of intestines carrying control versus Ras^V12 MARCM clones indicated increased tracheation in the Ras^V12 (Fig. 7b). To test the trachea requirements for tumor growth, we transplanted progenitor-derived GFP- or RFP-labeled Ras^V12Notch^RNAi tumor masses into recipients with mCherry- or GFP-labeled trachea, respectively, and found that wild-type recipient tracheae were recruited to the transplants (Fig. 7c–d). Furthermore, we transplanted Ras^V12Notch^RNAi tumor masses into wild-type recipients or recipients with trachea-specific btl knockdown. We found that 85% of the transplants survived and grew in wild-type recipients. However, only 45% survived in the trachea-specific btl^RNAi-expressing recipients, and the tumor masses grew substantially smaller (Fig. 7e). Furthermore, we generated flies bearing ISC-derived tumors by expressing a dominant negative form of the Notch tumor suppressor (Notch^DN) or activated forms of the Ras GTPase (Ras^V12 or Ras^Q13) (Fig. 7; Extended data Fig. 7–8) in adult intestinal progenitors. In all genotypes we observed increased tracheation either by fluorescent tracheal labeling (Fig. 7f–g, k–l, Extended data Fig. 7–8) or by brightfield microscopy (Fig. 7a–b, Fig. 8a). To achieve simultaneous labeling of the tumors and the trachea, we combined Gal4/Gal80^ts/UAS^30,52 with QF/QUAS³² to visualize the tumors with eGFP and the trachea with membrane-targeted mtd-Tomato. We performed time-course analysis for the Notch^DN and the Ras^V12 tumors to simultaneously assess their development and tracheation. We found that Notch^DN caused slow-progressing, regional tumorigenesis, whereby tumors developed gradually and were clearly visible in the R4a (P1) region of the midgut^53,54 after 4 days of induction (Extended data Fig. 7). The Ras^V12-expressing tumors developed rapidly. Their growth was noticeable 1 day post-induction (Fig. 7g), and eGFP-positive cells almost covered the midgut 3 days after tumor initiation (Extended data Fig. 8). In both genotypes, tumor growth correlated with an expansion of TTC branching. Quantification of the intestinal trachea (tracheal branch length normalized to the midgut surface in the R4a region for Notch^DN and throughout the midgut for Ras^V12) showed significant and gradual increases in the tumorous midguts (Fig. 7f–h, Extended data Fig. 7a–d; e-g; Extended data Fig. 8). Further, the tumor-bearing midguts encompassed more mitotic cells (Fig. 7i, Extended data Fig. 7g). Similarly, double Ras^V12Notch^RNAi tumors exhibited increased tracheation (number of tertiary tracheal branches in R5), and mitoses (Fig. 8a,c–d). These results suggest that the trachea is induced by the tumor and is necessary for its growth, and that tracheal branches are needed for this process.

Figure 7. Midgut ISC-derived tumors induce TTC branching required for tumor growth.

a. Midgut Ras^V12 MARCM clone (green) and corresponding brightfield image (a’) of the trachea. b. Quantification of TTC branching in tumorous Ras^V12 clone-bearing midguts vs control clone-bearing midguts (n=26,14). c. Schematic of the transplant experimental process. d. Fluorescently-labeled midgut tumor pieces from a donor fly are transplanted to a recipient fly with fluorescently-labeled trachea. Images showing GFP+ green tumors (first two images) recruiting CD8-mCherry marked tracheae from the recipient. The third image shows RFP+ (red) tumor mass overlayed with CD8-GFP+ (green) recipient trachea. e. Table showing comparison of the growth ability and survival of transplanted tumors in wild-type or trachea-defective (ppk-Gal4>UAS-btl^RNAi) recipients. f-g. The R5 region of control esg-Gal4 UAS-eGFP tub-Gal80^ts (f) and esg-Gal4 UAS-eGFP tub-Gal80^ts>UAS-Ras^V12-tumor (g) bearing midguts with concomitant expression of QF6>QUAS-mtdTomato (red) to label the trachea reared for 1 day at 29°C. DAPI (blue) was used to label all midgut nuclei. f’- g’, and f”-g” show the individual channels for the eGFP and the Tomato-labeled trachea, respectively. h-i. Quantification of TTC branching (h, n=10,11,9,9,10,10) and midgut mitosis (i, n=20 each) in control and Ras^V12-tumor bearing flies during a time-course analysis at 1, 3, and 5 days post-tumor induction. j. qRT-PCR analysis of btl, bnl, ldh, puc and upd3 upon Ras^V12 tumor progression. The average and standard deviation of n=3 biological experiments are plotted. k-l. Control genotype and Ras^V12 tumor reversal (3 days at 29°C followed by 7 days at 18°C). k’-k” and l’-ll” show the individual channels for eGFP and the Tomato-labeled trachea, respectively. m-n. Ras^Q13 tumor mitosis in the presence or knockdown of sima (m, n=10 each), and bnl (n, n=10,8,7). o-p. Quantification of TTC branching (o, n=10,10,10,10,10,11) and midgut mitosis (p, n=20 each) in control and Ras^V12 following reversal of the tumors after 1, 3, and 5 days post tumor initiation. Data are presented as mean values ± SD. All scale bars: 75 μm in f-g, k-l. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

Figure 8. Tumor-induced increased tracheal coverage, oxygen supply, and ROS are necessary for tumor growth.

a-d. Brightfield images of the TTCs (a) and fluorescent images of the GFP-positive (b) progenitor or progenitor derived tumor cells in control midguts, midguts bearing esg-Gal4 UAS-eGFP tub-Gal80^ts>UAS-Ras^V12 UAS-Notch^RNAi tumors, and esg-Gal4 UAS-eGFP tub-Gal80^ts>UAS-Ras^V12 UAS-Notch^RNAi co-expressing UAS-bnl^RNAi. Quantification of TTC branching (c, n=7,10,11) and mitosis (d, n=7,13,10) in control, Ras^V12 Notch^RNAi and Ras^V12 Notch^RNAi bnl^RNAi tumor-bearing flies. e. Percent survival of control, Ras^V12 Notch^RNAi and Ras^V12 Notch^RNAi bnl^RNAi tumor-bearing flies (n=160,78,120). f-h. Images showing growth of Ras^V12 Notch^RNAi tumors (f), quantification of TTC branching (g, n=8,9,9,9) and midgut mitosis (h, n=13,8,7,12) in control and Ras^V12 Notch^RNAi tumor-bearing flies in normoxia vs. hypoxia. i-k. Images showing tumorous midguts (R5 region) with progenitor derived Ras^Q13 tumors (green) with Cat and Sod1 overexpression or btl knockdown. Quantification of TTC branching (j, n=11,10,10) and midgut mitosis (k, n=15,13,13) in the same genotypes. Scale bars: 200 μm in a, 100 μm in b and 75 μm in f and i. l. Model of the intestinal trachea-midgut communication during damage-induced regeneration. Data are presented as mean values ± SD. The Kaplan-Meier method was used to test significance in e. For all others, a two-sided t-test was used: ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

To assess whether tumors induce btl, bnl and sima similar to infection and oxidative damage, we performed RT-qPCR on tumorous midguts and compared them to wild-type controls. We found that Ras^V12 tumors upregulated, as expected, damage-induced genes, such as puckered (puc) and upd3, and that btl, bnl and ldh (the Sima activity reporter) were also upregulated (Fig. 7j). To assess the requirement for sima, we silenced sima in the tumor cells, and found reduced tumor mitoses (Fig. 7m). To assess the requirement for bnl, we silenced bnl in the tumor cells, wherein tumor size, mitoses, and tracheal coverage were found reduced (Fig. 7n and Fig. 8a–d). Further, we performed survival experiments on tumor-bearing flies with or without bnl^RNAi in the tumors. We found that bnl silencing in the tumors significantly increased survival of the tumor-bearing flies (Fig. 8e). Collectively, these results suggest that the activation of sima and bnl in intestinal tumors induces tracheal outgrowth, which enhances tumor growth.

To assess whether induction of tracheal remodelling is a reversible feature of growing tumors, we discontinued Ras^V12 expression 1, 3 and 5 days after induction by shifting the flies from 29°C to 18°C (Fig. 7k–l). We found that Ras^V12 tumor growth and associated tracheation were reversed by switching Ras^V12 expression off for 7 days, 1, 3, or 5 days post-oncogene induction (Fig. 7o–p; also compare to Fig. 7h–i). Since increased tracheal branching was associated with increased ISC proliferation only in normoxia, we hypothesized that tumor growth would be reduced in hypoxia. To test this, we reared tumor-bearing flies under hypoxia. Both tumor size and ISC proliferation were reduced by hypoxia, although no significant difference in intestinal tracheal coverage was observed between normoxia and hypoxia (Fig. 8f–h). Thus, during tumorigenesis, the visceral trachea expands TBs to accommodate the extra metabolic needs of the growing tumors and to sustain tumor mitoses. In hypoxia, however, due to the limited oxygen availability to the tumor, regardless of expanded tracheation, tumor growth is reduced.

Furthermore, to assess the involvement of ROS in tumor growth and tracheal remodelling, we inhibited ROS in the tumor by overexpressing Cat and Sod1. We found that tumor growth, mitosis and tracheole branching were reduced (Fig. 8i–k). Finally, to test if btl has a non-tracheal role in the growing tumors, as previously suggested⁵¹, we inactivated btl by overexpressing btl^DN in the Ras^V12 tumorous midguts. We found that btl is necessary for tumor growth, mitosis and tracheole branching (Fig. 8i–k). Thus, ROS and btl are necessary for tumor mitosis and tracheole remodelling.

Discussion

Angiogenesis is the process by which oxygen-carrying blood vessels are remodeled physiologically⁵⁵. This process can go awry during inflammatory diseases and is a recognized cancer hallmark^50,56. Studies in cell culture and vertebrate models have established the link between tumor growth and angiogenesis^57–59 that implicates hypoxia and the Hif-1α pathway^60,61. Despite the identification of several tumor angiogenesis regulators, including VEGF, PDGF, and FGF^58,62, we know little about the environmental factors contributing to their induction and whether they are induced in non-tumor contexts. Here we show that infection, oxidative damage and tumors induce intestinal tracheole branching in Drosophila adults that supports regenerative and tumor growth. Independent evidence for the role of tracheogenesis in regeneration of the damaged Drosophila intestine is also presented in an accompanying study⁶³.

During development, branching morphogenesis is triggered by cell-cell communication between Btl-expressing tracheae and Bnl-expressing cells¹. Not much is known, however, about adult tracheogenesis, or the role of non-tracheal Btl functions in tracheogenesis. We know that insulin/oxygen-sensitive neurons control adult intestinal tracheal plasticity in response to nutrition, and that tracheal arborization affects metabolism, e.g. lipid storage²⁸. Whether neurons also regulate damage-induced tracheal branching and ISC proliferation requires further inquiry. Remarkably, hypoxic Drosophila wing disc tumor cells stabilize Sima, induce bnl, while a subset of them express the btl transcriptional inducer, Trachealess⁶⁴, and differentiate into tracheae in ways reminiscent of mammalian angiogenesis³⁴. Also, Sox21a mutant midgut tumors induce btl unconventionally in the tumor⁵¹. Furthermore, a tracheal role in tumor oxygenation has been suggested from ISC-derived Apc-Ras tumors expressing Snail, which encompass migratory cells that envelope the intestinal trachea, using it as a scaffold²⁹. We show that the trachea actively regulates tumor growth and is an essential part of niche modification by the tumor for its growth and survival.

We found that tracheal btl and epithelial bnl are necessary and sufficient for tracheogenesis and ISC mitosis. Surprisingly, we also found that epithelial btl and tracheal bnl are necessary for damage-induced regeneration, thus, uncovering an unconventional role for Btl. Of note, we observed an inconsistency in the tracheal bnl knockdown: bnl^RNAi driven by btl^ts-Gal4 affected both mitosis and tracheogenesis, but bnl^RNAi driven by dSRF^ts-Gal4 only impaired mitosis. These results could be explained by the different strength of the drivers: btl-Gal4 is expressed broadly and strongly in all the tracheae, whereas dSRF-Gal4 is a TTC-specific driver, weaker than btl-Gal4. Strikingly, overexpression of btl in the epithelium and bnl in the trachea were not sufficient to induce tracheole remodelling or ISC mitosis, suggesting that these sources are playing secondary roles. Tracheal Btl is abundant, and its sensitivity to the variable-level and multi-source epithelial Bnl directly controls tracheal branching, and consequently, ISC mitosis. On the other hand, epithelial Btl and tracheal Bnl are present at lower levels in the intestine to fine-tune ISC mitosis and, in turn, indirectly impinge on tracheal remodelling.

We show that upon midgut damage, both Btl and Bnl are regulated by Sima. This is reminiscent of the effects of Sima in hypoxic larvae⁴. However, in the normoxic gut the trigger for Sima is ROS. ROS are byproducts of oxygen metabolism within cells and have a role in regeneration and injury response in Drosophila and mammals^49,65–67. In mammals, ROS regulate angiogenesis in a tissue-specific manner: they suppress angiogenesis in diabetic mice, but enhance placental angiogenesis during pregnancy^68,69. Epithelial tumors induce ROS-mediated signalling to proliferate in mammals and Drosophila^49,70,71. Pathogens and commensals induce intestinal ROS-producing Duox and Nox, respectively, as a defense mechanism^72–74. Upon P.e. infection, intracellular ROS produced in ECs by Nox are necessary for intestinal regeneration⁴⁹. Duox is highly expressed in barrier epithelia, including the Drosophila trachea⁷⁵, but its tracheal role in intestinal growth has recently emerged⁷⁶. We show that Duox is required in both the trachea and the epithelium for tracheogenesis and regeneration upon infection. Overall, our data suggest that exogenous and intestinal ROS act coordinately in the trachea and the epithelium for regeneration.

We propose that intestinal homeostasis is maintained by the availability of oxygen supply to the midgut epithelium through the tracheoles. Tracheal branching is dynamically regulated by ROS- and oxygen-dependent signalling emanating from the epithelium. Thus, a feed-forward loop of ROS-induced tracheal branching increases oxygen supply to the epithelium. We suggest that increased oxygen availability drives ISC mitosis by impinging on key damage/infection- and tumor-induced signals, including Upds and EGFs (Fig. 8l).

Materials and Methods

All Drosophila and Pseudomonas strains used in this study are listed in Supplementary Table 1. Commercial reagents, chemicals and antibodies are listed in Supplementary Table 2.

Fly stocks and maintenance

Drosophila stocks were maintained at 18°C on a 12:12 hour light:dark cycle on standard fly food containing propionic acid (Sigma) and Tegosept (Genesee). P.a. and hypoxia experiments were conducted at the University of Cyprus, and P.e. experiments and tumor transplants were conducted at the German Cancer Research Center (DKFZ)-Center for Molecular Biology (ZMBH) and the Huntsman Cancer Institute.

The following Gal4 lines were used for tissue and cell-type specific expression: yw; btl-Gal4 UAS-srcGFP ⁷⁷, w; btl-Gal4 UAS-srcGFP tub-Gal80^ts/CyO (this work), yw; esg-Gal4^NP5130 UAS-srcGFP ⁷⁷, w; esg-Gal4 UAS-GFP tub-Gal80^ts13, w; Su(H)-Gal4 UAS-CD8-GFP tub-Gal80^ts20/CyO (this work; Su(H)-Gal4 was a gift from S. Hou ¹⁵), w; Myo1A-Gal4 UAS-GFP/CyO Su(H)-Gal80 tub-Gal80^ts/TM6C (this work), w; Myo1A-Gal4 UAS-mcherry tub-Gal80^ts/CyO (this work), w; trh-Gal4^14D03 (BDSC# 47463), w; dSRF-Gal4/CyO (from BDSC# 25753), w; dSRF-Gal4 UAS-eGFP ⁷⁸, w; esg-Gal4 UAS-GFP tub-Gal80^ts/CyO; QF6 QUAS-mtdTomato/TM6B (this work; QF6 was a gift from C. Potter ³²), w; Myo1A-Gal4 UAS-nls-mCherry tub-Gal80^ts; QF6 QUAS-mCD8GFP (this work; ³²), w; btl-Gal4 UAS-srcGFP tubGal80^ts; sima^KG/T(2;3) (this work), w; Myo1A-Gal4 UAS-nls-mCherry tub-Gal80^ts; sima^KG/T(2;3) (this work), w; esg-Gal4 UAS-eGFP tub-Gal80^ts/CyO; UAS-Ras1A^Q13/TM6C (this work); w; ppk-Gal4/CyO (BDSC# 32078), w; UAS-CD8::GFP.L^LL5 UAS-CD8::GFP.L² (BDSC# 5137), w; mira-GFP/CyO; trh-Gal4^14D03-Gal4/TM6B (this work, mira-GFP was a gift from A. Bardin ⁷⁹).

The following UAS strains were used: yw; UAS-btl^DN80, w; UAS-λbtl/TM3,Sb ⁸⁰, w UAS-λbtl (BDSC# 29046), w; UAS-bnl.S^A1.1 (from BDSC# 64231), yv; UAS-sima^V10,RNAi (BDSC# 26207), w; UAS-simaB (BDSC# 9582) ⁴⁰, w; UAS-Cat^RNAi (VDRC# 6283), w; UAS-Sod1.A (BDSC# 24750), w; UAS-Cat.A (BDSC# 24621), w; UAS-bnl^RNAi1 (VDRC# 101377), w; UAS-bnl^RNAi2 (VDRC# 5730), yv; UAS-bnl^RNAi (BDSC# 34572), w; UAS-Ras85D^V12 (BDSC# 4847) and w; UAS-Notch^DN781, w UAS-Notch^RNAi (BDSC# 7078), w; UAS-Duox^RNAi82.

The following mutants and reporter lines were used: w¹¹¹⁸ (BDSC# 3605), sima^KG07607 (BDSC# 14640, isogenized in CP lab), yw; btl^MI03286 (BDSC# 36229), btl^f0286483, yw; bnl^MI01635/TM3 (BDSC# 34228), bnl^e0311583, w; UAS-src-GFP; bnl-Gal4^NP2211 (this work, originating from KyotoSC# 112825), w; ldh-Gal4 UAS-nls-GFP and w; ldh-Gal4 UAS-src-GFP (this work, ldh-Gal4 was a gift from P. Wappner ⁴⁰), yw; esg-lacZ⁶⁰⁶ (BDSC# 10359), ry⁵⁰⁶ Dl-lacZ⁰⁵¹⁵¹/TM3 (BDSC# 11651) and w; Su(H)-lacZ^GBE (BDSC# 83352). For crosses involving driver lines with “tubulin-Gal4, Gal80^ts” were set up at 18°C, F1 adult progeny were collected within two days of emerging from the pupae. These flies were transferred to 29°C for the expression of the transgene including RNAi construct of interest. Flies were used within 10 days of emerging for all the stress experiments. Flies were transferred to new food every 2 days except for weekends. Fly stocks used for the experiments are listed in Supplementary Table 1.

Control and Ras^V12 MARCM clones ⁸⁴ were generated by crossing yw hs-FLP tub-Gal4 UAS-nls-EGFP; FRT82B tubGal80 ⁸⁵ to w; FRT82B arm-lacZ (BDSC# 7369) and w; UAS-Ras^V12; FRT82B arm-lacZ (this work), respectively. In short, 3-day-old female flies reared at 25°C were heat shocked at 37°C in a circulating water bath for 60 min, once a day for 2 sequential days. Then, they were maintained at 25°C for one week until dissection.

Oral administration of bacteria and chemicals

Female mated mature flies were used for all feeding assays.

P. aeruginosa:

P.a. feeding was performed as previously described²³. Briefly, a single colony from the P. aeruginosa strain PA14 was grown at 37°C to OD_600nm=3, corresponding to 5×10⁹ bacteria/ml. Female mature flies of the desired genotype were starved for 5 hours and added in groups of 10 per fly vial containing a cotton ball at the bottom impregnated with 5 ml of 0.5ml PA14 OD_600nm=3, 1ml 20% sucrose and 3.5ml dH₂O. For uninfected control, 5ml of 1ml sucrose 20% and 4ml dH₂O was used. Flies were incubated for 48 hours at 25°C or 29°C (for all experiments utilizing the Gal4-UAS system, unless otherwise noted).

P. entomophila:

P.e. was cultured in LB with rifampicin (100μg/ ml, Sigma-Aldrich, USA) at 29°C. Initial 50 ml overnight culture was transferred to 150 ml media for further 24 hours at the same temperature. P.e. was spun-down, mixed with 5% sucrose, yeast-extract and bacterial supernatant. Flies starved for 4 hours with water were fed with this P.e. mix. 5% sucrose was fed to the control groups at the same time.

H₂O₂:

30 % H₂O₂ solution (EMSURE) was stored at 4°C. 500 μl of 1% H₂O₂ with 4% sucrose solution was pipetted onto a sterile cotton ball at the bottom of an empty vial, and 4% sucrose solution without H₂O₂ was used as control. Flies were transferred to treatment vials and incubated for 48 hours at 25°C before dissection, unless otherwise noted.

Paraquat:

PQ powder (Sigma) was diluted in ddH₂O and stored as a stock of 100mM at 4°C. A 10mM PQ solution prepared fresh in 4% sucrose was used as test food and 4% sucrose solution without PQ as a control. Flies were transferred to the empty fly food vial containing only one 23mm Whatman paper disc soaked with 250μl of test or control food and incubated for 24 hours at 25°C, unless otherwise noted.

P. aeruginosa mutants supernatant administration (MvfR, PhzS, PhzM):

The mutant strains MvfR, PhzS and PhzM were taken from a P.a. library designed by F. Ausubel⁴⁶. Cell-free supernatants of PA14, MvfR, PhzS and PhzM were prepared by centrifugation and filtering of the supernatant with 0.2 μm filters. Fresh cell-free supernatant solutions were prepared daily for 2 days. Flies were transferred to the vials with the wet cotton balls impregnated with 5 ml of a solution composed of 80% LB medium or supernatant and 20% sucrose.

Dissections and Immunohistochemistry

Adult midgut dissections and immunohistochemistry were performed as previously described²³. Following dissections on Sylgard (VWR) plates in 1x PBS (130mM NaCl, 70mM Na₂HPO₄, 30mM NaH₂PO₄). Dissected midguts were fixed in 4% Formaldehyde (Polysciences) in 1x PBS for 30 minutes at room temperature. After 3 rinses with PBS, the tissues were incubated in blocking PBT solution (1x PBS, 0.5% BSA, 0.2% TritonX-100) at room temperature for at least 20 minutes. Addition of the primary antibody in PBT was followed by incubation for 16 hours at 4°C. Primary antibody was washed 3 times for 10 min at room temperature with PT (1x PBS, 0.2% TritonX-100). Secondary antibodies and DAPI (Sigma) are added for 1–2 hours at room temperature. Excess antibodies are washed in PT (as before) and the midguts are mounted in Vectashield (Vector). Primary antibodies: rabbit-anti-pH3 (Millipore 1:4000), rabbit anti-cleaved-Caspase-3 (#9661 Cell Signaling Technology 1:400), mouse-anti-Prospero (DSHB 1:100), mouse anti-β-gal (Promega 1:500), chicken-anti-GFP (Invitrogen 1:1000), rabbit anti-GFP (1:3000; Invitrogen) DAPI (Sigma, 1:3000 of 10mg/ml stock) (staining of the nuclei). Secondary antibodies against mouse, rabbit or chicken conjugated to Alexa Fluor 488 and 555 (Invitrogen) were used at 1:1000.

RT-qPCR

Dissected midguts were collected in Qiazol (Qiagen). Total RNA was extracted from 20 midguts per strain per condition per biological replicate using the Qiazol protocol. 800 ng of total RNA were freed from genomic DNA using the RQ1 RNase-Free DNase Kit (Promega). Reverse transcription was performed using 145,4ng of the total DNase-treated RNA using the TaKaRa PrimeScript RT Master Mix Kit. qPCR amplification was performed using gene specific primers with the following amplification program: 95°C for 30 seconds (initial denaturation), 40 cycles of 95°C for 10 seconds (denaturation), 60°C for 30 seconds (annealing), 65°C for 30 seconds (extension) and 65°C for 1 minute (final extension). Primer sequences for each gene are shown in Table 1. The expression of the genes of interest was normalized to the expression levels of two reference genes, rpl32 and gapdh1 using the 2-ΔΔCt method. Data were analyzed using the Bio-Rad CFX Manager 3.1 program and graphs were prepared in Excel. For all experiments at least 3 biological replicates were used to calculate the average and the standard deviation. The primer pairs used for qPCR analyses are listed in Supplementary Table 3.

Midgut Tumorigenesis

Tumors (Ras^V12, Ras^Q13, Notch^DN, Ras^V12 Notch^RNAi) were induced in the midgut intestinal progenitor population via the Gal4-UAS-Gal80^ts system^30,52. Tumor-generating crosses were reared at 18°C, and emerging adult progeny were maintained at 18°C to mature for 5–7 days. After maturation, mated adult females were transferred to 29°C to induce transgenes. For Ras^V12 tumors, flies were induced for 1, 3 and 5 days, for Notch^DN tumors flies were induced for 4, 7 and 10 days. For Ras^Q13 tumors flies were induced for 1 and 2 days. For Ras^V12 Notch^RNAi tumors, flies were induced for 7 days. For MARCM tumors, the crosses were reared at 25°C and mature 3–5 days old adult female progeny with the correct genotype were heat-shocked for 1 hour at 37°C in a water bath. A time-course analysis (3, 7, 10, 14 days) was performed to identify the time point when large clones were readily visible. Large close were obtained 14 days after heat shock. Images were acquired in a Zeiss Axioscope A1 to show GFP fluorescence of the clone and bright-field image of the trachea.

Survival assay

Flies of the appropriate genotypes were collected within 1–2 days after hatching at 18°C. The flies were transferred to 29°C and the number of dead flies were counted every 2 days while changing the food. The Kaplan-Meier log-rank test was used to assess statistical significance and the data were plotted using Graphpad Prism 9.

Tumor transplants

Midguts bearing control progenitor cells and tumors were dissected out and cut into pieces. Cut pieces were aspirated into a hypodermic syringe and carefully injected into the abdomen of the anesthetized recipient flies. Recipient flies with transplanted midgut pieces were incubated at 29°C until observation was complete. The percentage of tumor growth was calculated by counting the number of flies bearing tumors compared to the number of flies injected with the tumor pieces. The percentage of distal colonization was calculated by counting the number of tumor-bearing flies showing tumors above the thorax region (including the head) away for the site of injection in the abdomen to the number of flies injected with the tumor pieces.

Hypoxia

Drosophila strains were reared in normoxic (21% O₂) and hypoxic conditions (5% O₂) for 12 hours. For the hypoxic conditions the Modular Incubator Chamber (MIC-101, Billups – Rothenberg Inc.), was used connected to a cylinder that contained a mixture of gasses with 4% O₂. The flies for both normoxic and hypoxic conditions were transferred into new vials containing the oral experimental condition and were covered with net (instead of cotton). The chamber was moved into a 29°C incubator overnight (12 hrs). For all the combined hypoxia-infected and hypoxia-tumor conditions, hypoxia was the last step, meaning that the flies before dissection were reared in low oxygen tension. The flies that were reared in hypoxia were dissected first to reduce contact with the atmospheric oxygen.

Image acquisition and analysis

Full midgut images (as those used in Extended Data Fig. 1 and 3) were acquired with a Leica MZ16F Fluorescent Stereomicroscope equipped with GFP and DAPI filters. Stacks of optical sections were acquired using the Leica TCS SP2 DMIRE2 Scanning Confocal Microscope using a 40x oil objective (zoom 1x, average 4). Images to be compared were acquired using the exact same settings. Bright-field images were acquired using the fluorescent microscope Zeis Axioscope A.1 at 20x magnification with the Axiovision Software.

Measurements of tracheal length were performed using the NeuronJ Plugin of ImageJ 1.34s software (ImageJ – Plugins – NeuronJ. File – Open image file). The monitoring of the tracheal system can take place in two independent ways. The first relies on the quantification of the fluorescence intensity of the trachea and the other on the quantification of the tracheal branches. For trachea quantification, images were acquired using a Leica TCS SP2 DMIRE2 confocal microscope and the LCS software, at 40Χ magnification, zoom1, image size: 1024×1024, 0.5 μm step size, sequential scan mode DAPI/RFP, ~18–20 serial sections for both the anterior and the posterior areas of the midgut. Importantly, to achieve accurate comparisons, all the images were acquired under the same GFP laser settings. Numerous intestines were photographed from the anterior (R2 region), the posterior (R5) region and occasionally from the middle (R3, R4a) midgut (specified throughout the text). Using the NeuronJ tool (Add tracings), an ImageJ Plugin for Neurite Tracing and Analysis found in the ICY software (http://icy.bioimageanalysis.org/), we highlighted each tracheal branch along on the surface of the midgut. We avoided highlighting the large, thick external tracheal vessels that do not adhere to the gut surface. Then, we summed the full length of the branches (Measure tracings). Normalization to the surface of the midgut was performed for each image before plotting. We tested this image quantification method and compared it to other methods. We found that this was the most accurate method for intestinal tracheal measurements, and thus, we used it for all subsequent experiments. Graphs were generated using Graphpad Prism 9 by calculating the average and standard deviation values. Bright-field images from the Zeiss Axioscope A.1 were also used to measure the tracheal length using the NeuronJ Plugin. In the Edgar lab, the images of the midguts with trachea (Fig. 1e, f, Fig. 7b) were taken at the same magnification with a 40x water objective using a Leica SP8 confocal microscope and the LAS X software. Bright-field DIC images (Fig. 1k, Fig. 7a,c) were taken with a Leica DM microscope and grouped anonymously for quantifying the phenotypes. The tracheal branches were categorized into primary, secondary, and tertiary branches according to their thickness with thickest being the primary trachea. No significant differences in the number of primary or secondary trachea were observed across the groups. The number of tertiary tracheal branches per field was counted manually from the images.

The numbers of pH3-positive cells were counted under the fluorescent microscope (Zeiss Axioscope A.1 and Nikon Eclipse Ti) at 20x magnification along the whole midgut. Graphs were generated using Graphpad Prism 9 by calculating the average and standard deviation values. Image panles were compiled using Photoshop 2020 (Adobe).

Statistics and Reproducibility

All experiments (fluorescence micrographs, brightfield imaging of the trachea, and mitosis assessement) were repeated at least three times with similar results. Representative images are shown for microscopy images.

Pair comparisons were statistically-tested with a two-sided Student’s t-test. The U-test was used in cases where the number of points was less than 10, as specified in individual figure legends.

Data availability

All data supporting the findings of this study are available from the corresponding authors on reasonable request.

Code availability

No custom algorithms or software were generated for this study.

Extended Data

Extended Data Fig. 1. Infection and oxidative damage increase esg>GFP+ cells in the midgut and associate with increased TTC branching.

a. Adult midgut intestinal progenitors labeled with esg^NP5130-Gal4>UAS-srcGFP were imaged in unchallenged conditions (4% sucrose) and upon oral P.a. infection (48hrs), and feeding with H₂O₂ (48hrs) and PQ (24hrs). DAPI (blue) in the upper panels stains all midgut nuclei. The bottom panels show the GFP-labeled progenitors separately. P.a. and PQ expanded the intestinal progenitors with a posterior midgut bias, whereas H₂O₂ exhibited an anterior midgut bias. b-c. Quantification of midgut mitosis (b, n=10 each) and TTC branching (c, n=7,6) in PQ-treated flies. d-e. Posterior midgut (R4) images of btl-Gal4>UAS-srcGFP flies in baseline conditions (sucrose) and upon PQ feeding. DAPI (blue) staining all the nuclei. Single channel images of the GFP are shown in d’-e’. f-g. Posterior midgut images of QF6>QUAS-mtdTomato flies in baseline conditions exhibit tracheal expression of the reporter. Midgut epithelial ECs with low expression of the reporter are visible is zoomed image (g). Single channel images of the Tomato are shown in f’-g’. Scale bars: 300 μm in a, 75 μm in d-g. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided for b, and U-tested for c): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

Extended Data Fig. 2. The FGFR/Btl is necessary and sufficient for midgut TTC branching and ISC mitosis.

a-b. Brightfield images of the tracheae of P.a. infected R5 regions of the midgut in trh-Gal4 control (a) and trh-Gal4>UAS-btl^DN (b). c-d. Brightfield images of the tracheae of uninfected R5 regions of the midgut in trh-Gal4 control (c) and trh-Gal4>UAS-λbtl (d). e-g. Quantification of TTCs (e, n=10,11,10,9,10), TTC branching (f, n=10,11,10,8,10), and midgut mitosis (g, n=8,6,10,9,6) upon trh-Gal4-driven btl manipulation with or without P.a. infection. h-i. Brightfield images of the tracheae of P.a.-infected R5 regions of the midgut in dSRF-Gal4 control (h) and dSRF-Gal4>UAS-btl^DN (i). j-k. Brightfield images of the tracheae of uninfected R5 regions of the midgut in dSRF-Gal4 control (j) and dSRF-Gal4>UAS-λbtl (k). l-n. Quantification of TTCs (l, n=10,10,8,8,11), TTC branching (m, n=10,10,8,8,11), and midgut mitosis (n, n=11,9,12,12,9) upon dSRF-Gal4-driven btl manipulation with or without P.a. infection. All scale bars: 75 μm. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

Extended Data Fig. 3. Infection and oxidative damage induce FGF/bnl in the midgut epithelium.

Adult midgut bnl-expressing cells labeled with the reporter bnl-Gal4>UAS-srcGFP were imaged in unchallenged conditions (4% sucrose) and upon oral P.a. infection (48hrs), feeding with H₂O₂ (48hrs) and PQ (24hrs). DAPI (blue) in the upper panels stained all midgut nuclei. The bottom panels show the GFP-labeled bnl expressing cells separately. P.a. and PQ induced the reporter throughout the midgut, whereas H₂O₂ exhibited an anterior midgut bias. Scale bar: 300 μm.

Extended Data Fig. 4. Btl/Bnl signaling in the epithelial cells is necessary for efficient tracheal remodeling and mitosis in response to infection.

a-b. Quantification of TTC branching upon progenitor- (a) and EC-specific (b) silencing of bnl (bnl^RNAi3) and btl (btl^RNAi) (a, n=10,8,5,10,7,7 and b, n=10,7,9,10,9,8). c-d. Quantification of midgut mitosis upon progenitor- (c) and EC-specific (d) silencing of bnl (bnl^RNAi3) (c, n=6,8,12,12 and d, n=9,9,11,13). e. Quantification of esg+ progenitors as a percent of total number of cells in the posterior regions of the midgut upon progenitor-specific knockdown of btl (btl^RNAi) and bnl (bnl^RNAi3) (n=12,15,15). f-g. Quantification of midgut mitosis upon progenitor- (f) and EC-specific (g) silencing of btl (btl^RNAi) (f, n=8,13,11,13 and g, n=12,11,11,16). h. Quantification of esg+ progenitor cells/total number of cells in the posterior midgut upon progenitor-specific knockdown of btl (btl^DN, n=9,9). Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ns, not significant, * 0.01<p≤0.05, ** 0.001<p≤0.01 and *** p≤0.001.

Extended Data Fig. 5. Infection and oxidative damage activate Hif-1α/Sima in the midgut epithelium and the visceral TTCs.

Hif-1α/Sima activation was monitored via the ldh-Gal4>UAS-nlsGFP reporter expression in the adult midgut epithelium and the intestinal trachea of the R5 region in unchallenged flies (sucrose) and upon P.a. and PQ treatment (a, c, e), and of the R2 region in unchallenged flies (sucrose) and upon H₂O₂ feeding (b, d). Epithelial sections (a-d) and trachea surface sections (a’-d’) of the same midguts were imaged. DAPI (blue) in a-d and a’-d’ stains all the nuclei. a”-d” and a”’-d”’ correspond to separated channels for reporter expression in the epithelium and the intestinal trachea, respectively. The ldh-Gal4>UAS-nlsGFP reporter was expressed in cells of the midgut epithelium and in the midgut TTCs in baseline conditions in the anterior (R2 in b, b’) and posterior (R5 in a, a’) midgut. P.a. (c, c’), H₂O₂ (d, d’) and PQ (e, e’) induced the reporter in the epithelium and the trachea at varying degrees. All images were acquired at the same confocal settings as their respective controls. Scale bar: 75 μm.

Extended Data Fig. 6. Hif-1a/Sima is necessary in the intestinal epithelium and the trachea for TTC branching.

a-b. Brightfield images of the midgut TTCs (R5 region) upon trachea-specific (via btl-Gal4) sima knockdown in the background of heterozygous simaKG in baseline conditions. c-d. Bright-field images of the midgut TTCs (R5 region) upon EC-specific (via Myo1A-Gal4) sima knockdown in the background of heterozygous simaKG in baseline conditions. e-f. Bright-field images of the midgut TTCs (R5 region) upon trachea-specific (via btl-Gal4) sima knockdown in the background of heterozygous sima^KG in P.a.-infected conditions. g-h. Bright-field images of the midgut TTCs (R5 region) upon EC-specific (via Myo1A-Gal4) sima knockdown in the background of heterozygous sima^KG in P.a.-infected conditions. The images correspond to examples of those quantified for Fig. 4e,i. Scale bar: 75 μm.

Extended Data Fig. 7. Time-course analysis of Notch^DN progenitor-derived midgut tumors.

a-d. The R4a region of control (reared for 4 days at 18°C) (a) and tumorous midguts (reared for 4, 7 and 10 days at 29°C) (b-d) of the esg-Gal4 UAS-eGFP tub-Gal80^ts>UAS-Notch^DN genotype with concomitant expression of QF6>QUAS-mtdTomato (red) to label the trachea. DAPI (blue in a-d) is used to label all midgut nuclei and Prospero (a”-d”) labels the EEs. a’-d’, a”-d” and a”’-d”’ correspond to the individual channels for eGFP, Prospero and Tomato-labeled trachea, respectively. Scale bars: 75 μm. e-g. Quantification of TTC branching in the R4a of control (Notch^DN uninduced) and Notch^DN-expressing midguts (e, n=4,8,4,6,4,7), in the Notch^DN tumor-region vs. neighboring non-tumor area on the same image (f, n=6,6,4,4,6,6), and midgut mitosis of control (Notch^DN uninduced) and Notch^DN-expressing midguts (g, n=20 each) during a time-course analysis at 4, 7, and 10 days post-tumor induction. Scale bar: 75 μm. Data are presented as mean values ± SD. Statistical significance (t-tested, two-sided): ** 0.001<p≤0.01 and *** p≤0.001.

Extended Data Fig. 8. Time-course analysis of Ras^V12 progenitor-derived midgut tumors.

a-d. The R5 region of control esg-Gal4 UAS-eGFP tub-Gal80^ts (reared for 1 day at 29°^C) and esg-Gal4 UAS-eGFP tub-Gal80^ts>UAS-Ras^V12-tumor bearing midguts (reared for 1, 3 and 5 days at 29°C) with concomitant expression of QF6>QUAS-mtdTomato (red) to label the trachea. DAPI (blue in a-d) was used to label all midgut nuclei. a’-d’ and a”-d” correspond to the individual channels for the eGFP and the Tomato-labeled trachea, respectively. Scale bar: 75 μm.