Tissue-autonomous immune response regulates stress signaling during hypertrophy

  1. Robert Krautz  Is a corresponding author
  2. Dilan Khalili
  3. Ulrich Theopold  Is a corresponding author
  1. Department of Molecular Biosciences, The Wenner-Gren Institute (MBW), Stockholm University, Sweden

Abstract

Postmitotic tissues are incapable of replacing damaged cells through proliferation, but need to rely on buffering mechanisms to prevent tissue disintegration. By constitutively activating the Ras/MAPK-pathway via RasV12-overexpression in the postmitotic salivary glands (SGs) of Drosophila larvae, we overrode the glands adaptability to growth signals and induced hypertrophy. The accompanied loss of tissue integrity, recognition by cellular immunity, and cell death are all buffered by blocking stress signaling through a genuine tissue-autonomous immune response. This novel, spatio-temporally tightly regulated mechanism relies on the inhibition of a feedback-loop in the JNK-pathway by the immune effector and antimicrobial peptide Drosomycin. While this interaction might allow growing SGs to cope with temporary stress, continuous Drosomycin expression in RasV12-glands favors unrestricted hypertrophy. These findings indicate the necessity to refine therapeutic approaches that stimulate immune responses by acknowledging their possible, detrimental effects in damaged or stressed tissues.

eLife digest

Tissues and organs work hard to maintain balance in everything from taking up nutrients to controlling their growth. Ageing, wounding, sickness, and changes in the genetic code can all alter this balance, and cause the tissue or organ to lose some of its cells. Many tissues restore this loss by dividing their remaining cells to fill in the gaps. But some – like the salivary glands of fruit fly larvae – have lost this ability.

Tissues like these rely on being able to sense and counteract problems as they arise so as to not lose their balance in the first place. The immune system and stress responses are crucial for this process. They trigger steps to correct the problem and interact with each other to find a common decision about the fate of the affected tissue. To better understand how the immune system and stress response work together, Krautz, Khalili and Theopold genetically manipulated cells in the salivary gland of fruit fly larvae. These modifications switched on signals that stimulate cells to keep growing, causing the salivary gland’s tissue to slowly lose its balance and trigger the stress and immune response.

The experiments showed that while the stress response instructed the cells in the gland to die, a peptide released by the immune system called Drosomycin blocked this response and prevented the tissue from collapsing. The cells in the part of the gland not producing this immune peptide were consequently killed by the stress response. When all the cells in the salivary gland were forced to produce Drosomycin, none of the cells died and the whole tissue survived. But it also allowed the cells in the gland to grow uncontrollably, like a tumor, threatening the health of the entire organism.

Mapping the interactions between immune and stress pathways could help to fine-tune treatments that can prevent tissue damage. Fruit flies share many genetic features and molecular pathways with humans. So, the next step towards these kinds of treatments would be to screen for similar mechanisms that block stress activation in damaged human tissues. But this research carries a warning: careless activation of the immune system to protect stressed tissues could lead to uncontrolled tissue growth, and might cause more harm than good.

Introduction

Immune and stress responses have evolved to protect the organism from both exogenous and endogenous stimuli (Eming, 2014; Adamo, 2017; Rankin and Artis, 2018). By sensing deviations from homeostasis and inducing compensatory mechanisms, immune and stress responses keep physiological parameters within tolerable limits (Hoffmann and Parsons, 1991; Vermeulen and Loeschcke, 2007).

Heat shock, radiation, starvation, toxic metabolites, hypoxia, or hyperoxia are all well-characterized stressors. They activate stress responses which can ultimately lead to the induction of programmed cell death (Lowe et al., 2004; Loboda et al., 2016). In Drosophila, the Keap1-Nrf2-, JNK-, and p38-signaling pathways are crucial for mounting these anti-stress-responses (Stronach and Perrimon, 1999; Fuse and Kobayashi, 2017). Immune responses on the other hand, like the Drosophila Toll and imd pathways, are typically activated by molecular structures exposed on the surfaces of pathogens (Medzhitov and Janeway, 2002; Kurata, 2004). These pathways coordinate the humoral and cellular immune system to eliminate intruding pathogens (Lemaitre and Hoffmann, 2007; Buchon et al., 2014).

Humoral immune responses in Drosophila are characterized by the production and secretion of large sets of effector molecules, most notably antimicrobial peptides (AMPs) like Drosomycin (Drs) (Imler and Bulet, 2005). AMPs not only target extrinsic threats in the form of intruding pathogens, but also react to intrinsic stimuli such as tumorigenic transformation with the possibility to induce apoptosis (Araki et al., 2019; Parvy et al., 2019). However, it remains poorly understood whether AMPs also have functions beyond promoting apoptosis when sensing and reacting to accumulating stress such as during wound healing and tumor formation.

Apart from their described individual roles, immune and stress pathways are proposed to be either induced successively or concomitantly dependent on the level of deviation from homeostasis (Chovatiya and Medzhitov, 2014; Ammeux et al., 2016). However, detailed characterization of wound healing and tumor models in Drosophila revealed a more complex picture (Park et al., 2004; Buchon et al., 2009; Meyer et al., 2014; Wu et al., 2015; Liu et al., 2015). Accordingly, immune and stress responses often neither occur separately, nor do they follow a simple linear cascade, but rather regulate each other via context-dependent mutual crosstalk (Wu et al., 2015; Liu et al., 2015; Fogarty et al., 2016; Pérez et al., 2017). One recurring motif throughout most of these models is the central role of the stress-responsive JNK-pathway and its frequent interaction with the Toll and imd immune pathways (Park et al., 2004; Rämet et al., 2002; Galko and Krasnow, 2004; Uhlirova et al., 2005; Igaki et al., 2006; Enomoto et al., 2015; Andersen et al., 2015). However, while JNK-signaling can function either in a tumor-promoting, anti-apoptotic or in a tumor-suppressive, pro-apoptotic manner depending on the context, Toll- and imd-signaling have only been shown to display a tumor-suppressing, pro-apoptotic role in Drosophila (Uhlirova et al., 2005; Igaki et al., 2006; Enomoto et al., 2015; Uhlirova and Bohmann, 2006; Cordero et al., 2010; Vidal, 2010).

These tumor-suppressive, pro-apoptotic functions of immune responses have been well characterized and attributed to the secretion of humoral factors or the recruitment of immune cells through the systemic immune system (Fogarty et al., 2016; Pérez et al., 2017; Babcock et al., 2008; Pastor-Pareja et al., 2008; Parisi et al., 2014; Hauling et al., 2014). In addition, during clonal cell competition in imaginal discs, Toll- and imd-signaling were implicated in the elimination of less fit cell clones by inducing apoptosis (Meyer et al., 2014). Importantly, the selective growth disadvantage of these less fit cells is thought to be a response to systemic infection (Germani et al., 2018) and it remains an open question whether genuine tissue-autonomous immune responses can contribute to adaptation of growth during wound healing and tumor formation.

The larval salivary gland (SG) of Drosophila is a powerful system to study adaptive growth control, because growth is not completely predetermined, but modulated by the nutritional status (Smith and Orr-Weaver, 1991; Britton and Edgar, 1998). In contrast to mitotically active tissues, growth in post-mitotic tissues like the larval SG is based on endoreplication and hypertrophy rather than on cell division (Edgar et al., 2014; Orr-Weaver, 2015). Modifying the underlying, tight growth regulation, for instance by continuous growth signaling via constitutively activated Ras/MAPK-signaling can easily lead to the accumulation of oxidative stress and DNA damage (Mason et al., 2004; Bartkova et al., 2005; Bartkova et al., 2006; Di Micco et al., 2006; Shim et al., 2013; Shim, 2015). However, the parameters defining the natural limit of growth adaptation and the buffering mechanisms in place to cope with prolonged or continuous stress remain poorly understood.

Here, we uncover a genuine tissue-autonomous immune response which directly regulates hypertrophic growth and adaptation to accumulating stress in larval SGs. By overexpressing a dominant-active form of the small GTPase Ras, RasV12, we induced hypertrophic growth. This activates a tissue-autonomous immune response which allows the hypertrophic gland to cope with the resulting stress through spatio-temporally regulated inhibition of the JNK-mediated stress response. We present evidence that tissue-autonomous expression of the AMP Drosomycin (Drs) is at the core of this inhibition: Drs directly interferes with JNK-signaling and inhibits JNK-dependent programmed cell death. This prevents recognition of the stressed tissue by the cellular immune response and allows survival and unrestricted growth of hypertrophic SGs.

Results

Local immune reaction accompanies RasV12-dependent hypertrophy

In order to identify buffering mechanisms that compensate for continuous stress, we made use of our previously published hypertrophy model in the SGs of Drosophila larvae (Hauling et al., 2014). We expressed a dominant-active form of Ras, RasV12, across the entire secretory epithelium of SGs throughout larval development by using the BxMS1096 enhancer trap (Figure 1—figure supplement 1A,C for 96/120 hr after egg deposition, AED). To further enhance RasV12-dependent hypertrophy, we combined RasV12-expression with RNAi-mediated knockdown of the cell polarity gene lethal (2) giant larvae (lgl; Figure 1—figure supplement 1C–H; Jacob et al., 1987; Strand et al., 1994). Their individual and cooperative role in tumor formation in mitotic tissues has been well characterized (Bilder et al., 2000; Pagliarini and Xu, 2003; Herranz et al., 2016). Cell- and tissue-morphology was assessed using Phalloidin staining (Figure 1A; Figure 1—figure supplement 1A,C) and nuclear morphology by DAPI (Figure 1B–C; Figure 1—figure supplement 1C–F).

Figure 1 with 1 supplement see all
RasV12-induced hypertrophy induces local and cellular immune responses.

(A) RasV12-glands and controls stained with Phalloidin (red) to monitor tissue integrity at 96 hr and 120 hr after egg deposition (AED). (B) Nuclei stained with DAPI (white) to visualize nuclear volume and disintegration at 120 hr AED in RasV12-glands. (C) Nuclear volume as quantified by z-stacks of DAPI-stained salivary glands (SGs) at 96 hr and 120 hr AED and averaged per gland. (D) Comparative transcriptome analysis of RasV12- vs. w1118-glands. Differential expression quantified as beta statistic with q-values by Wald test. Significantly differentially expressed genes (log2(beta) ≥1; q-value ≤0.05) highlighted in yellow. (E) Upper: Whole larvae with DrsGFP reporter (green) expressing RasV12 in glands or controls at 96 hr AED. Lower: RasV12- and control-glands with DrsGFP reporter (green) stained for hemocytes (anti-Hemese, red). Proximal and distal gland parts are indicated by ‘PP’ and ‘DP’. (F) Hemocyte attachment measured as ln(Hemese-area)/ln(SG-area) and separated by time and gland part. Insets: (A/E Lower) DAPI, (E Upper) brightfield. Scalebars: (A-B, E Lower) 100 µm, (E Upper) 500 µm. Boxplots in (C, F): lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student's t-tests (***p<0.001, **p<0.01, *p<0.05, n.s. p≥0.05).

At 96 hr AED, RasV12-expressing SG cells retained most of their normal morphology compared to w1118-control glands. However, their integrity and polarity were severely disrupted at 120 hr AED (Figure 1A; Figure 1—figure supplement 1C). Nuclei of RasV12-glands showed a continuous increase in volume at 96 hr and 120 hr AED (1.33 fold compared to w1118 controls at 96 hr AED; 5.66 fold at 120 hr AED; Figure 1C) and signs of nuclear disintegration at 120 hr AED implying the induction of programmed cell death (PCD; Figure 1B). Both loss of cell integrity and nuclear disintegration coincided temporarily (Figure 1A–B) and were exacerbated upon coexpression of l(2)glRNAi (Figure 1—figure supplement 1C–F). These results confirm our previous findings and demonstrate that continuous growth signaling in larval SGs leads to increased organ size accompanied with additional endocycles at 96 hr AED, both hallmarks of compensatory hypertrophy (Tamori and Deng, 2014). Furthermore, the additional, RasV12-induced endoreplications without obvious effect on tissue integrity imply an adaptability to excess growth signaling, whereas the subsequent collapse of nuclear integrity and cellular polarity at 120 hr AED delineate its limitations.

To characterize the mechanisms involved in the early phase of SG growth adaptation at 96 hr AED, we performed total RNA sequencing of complete RasV12-expressing and w1118-control SGs dissected at 96 hr AED prior to cellular and nuclear disintegration. The most significantly upregulated gene in RasV12-glands compared to their w1118-counterpart was Ras85D itself (q-value = 6.51⋅10−282), which validates the experimental set-up (Figure 1D). The most differentially expressed gene in turn was the AMP Drs, which indicates the activation of a local immune response in RasV12-glands. To evaluate this further, we employed a GFP reporter for Drs and observed strong induction in RasV12-glands, but not in any other larval tissue (Figure 1E; Figure 1—figure supplement 1B,G,I; Ferrandon et al., 1998). At 96 hr AED, the entire secretory epithelium of the SG expressed Drs with a strong tendency for increased induction in the proximal part (PP) closest to the duct (Figure 1—figure supplement 1B,G). At 120 hr AED Drs was almost exclusively expressed in the PP, an expression pattern that persisted until RasV12-expressing larvae belatedly puparate (Figure 1E; Figure 1—figure supplement 1B,G,I). To rule out artifacts derived from driver related unequal expression across the SG epithelium, we measured BxMS1096-driven RFP-expression across the longitudinal gland axis at 96 hr and 120 hr AED and normalized the acquired signal with driver-independent Phalloidin-fluorescence (Figure 1—figure supplement 1A; see ‘Materials and methods’). The average Drs-reporter signal already peaks at 96 hr AED in the middle of the gland epithelium, while BxMS1096-driven expression increases further until 120 hr AED. Hence, while BxMS1096-driven expression of RasV12 is crucial for activating Drs-expression, both are not qualitatively correlated (Figure 1—figure supplement 1A–B). In order to assess whether hemocytes were recruited as part of a parallel, cellular immune response, we stained glands with an antibody against the pan-hemocyte-marker Hemese. While RasV12-glands were completely devoid of hemocytes at 96 hr AED, at 120 hr AED they were recruited to the gland surface. However, hemocyte attachment was restricted to the distal, non-Drs expressing part (DP), rendering Drs expression and hemocyte attachment across the gland epithelium mutually exclusive (Figure 1E–F). Coexpression of l(2)glRNAi elevated the level of recruited hemocytes at 120 hr AED and pre-empted this recruitment to the DP already at 96 hr AED (Figure 1—figure supplement 1G–H).

We next investigated whether the change in nuclear volume as a marker for growth adaptation follows a similar proximal-distal-divide as Drs-expression and hemocyte attachment (Figure 1—figure supplement 1E). Nuclei in the DP of the SG at 96 hr AED showed a moderate volume increase upon RasV12-expression compared to distal w1118-control nuclei. However, after 120 hr AED distal nuclei had undergone a drastic increase in nuclear volume (6.28 fold compared to distal w1118-control nuclei) while nuclei in the PP of the SG displayed only a moderate increase in size compared to w1118-control nuclei, that did not increase over time. This indicates that nuclei in the DP of RasV12-glands undergo more rounds of endoreplication than their proximal counterparts coinciding with the decline of Drs-expression and an increase in hemocyte attachment in this part. Moreover, this difference also explains the marginal reduction in BxMS1096-driven expression from DP to PP (Figure 1—figure supplement 1A; PP shows 86% of DP-expression level).

Dorsal-dependent Drs expression is part of a genuine tissue-autonomous immune response

As barrier epithelia, the lumen of the SG forms part of a continuum with the exterior, exposing them to extrinsic stimuli including nutritional cues and pathogens (Andrew et al., 2000). Since systemic infection can modulate tissue growth, we sought to clarify whether the observed local immune response in the gland epithelium fulfills the criteria of a genuine tissue-autonomous immune response or was rather embedded in a wider systemic immune response (Germani et al., 2018). Therefore, we eradicated the majority of putative systemic infections, food-derived signals and pathogens or bacterial contamination by raising larvae with RasV12-glands under sterile conditions, on minimal medium or by bleaching embryos (Figure 2—figure supplement 1A–A’’; Shaukat et al., 2015; Kenmoku et al., 2017; Asri et al., 2019). None of these changes diminished the Drs expression, strongly indicating that Drs is indeed induced in a bona fide tissue-autonomous manner as a response to RasV12-dependent hypertrophic growth (Figure 2—figure supplement 1A–A’’; Colombani et al., 2005; Mirth et al., 2014).

We further sought to identify the upstream factors controlling Drs expression. The homeobox-like transcription factor Caudal is necessary for Drs expression in the female reproductive organs and the adult SG (Ferrandon et al., 1998; Ryu et al., 2004; Han et al., 2004). However, RNAi-lines directed against Caudal or its canonical interaction partner, Drifter (dfr / vvl), did not reduce Drs expression in RasV12-glands indicating that the regulation of Drs expression as part of the tissue-autonomous immune response is clearly different from its counterpart in the adult SG (Figure 2—figure supplement 2A; Junell et al., 2010).

In order to evaluate whether either Toll- or – as is the case for local infections – imd-signaling plays a role in Drs expression, we used the reproducible fluorescence pattern of the Drs-GFP reporter at 96 hr AED to assay RNAi-lines directed against canonical components of both pathways in RasV12-glands (Figure 2A; see ‘Materials and methods’ for scored phenotypes; Ferrandon et al., 1998; Tzou et al., 2000; Takehana et al., 2004; Wagner et al., 2009). Of the 11 tested RNAi-lines, most of which were previously published to cause phenotypes, only one targeting the NFκB-transcription factor Dorsal (dl) significantly reduced the fluorescence signal of the Drs-reporter (Figure 2A; Supplementary file 3). However, Drs expression is completely independent of the upstream modules of the two classical Drosophila immune pathways, Toll and imd.

Figure 2 with 3 supplements see all
Drs expression is part of a genuine tissue-autonomous immune response.

(A) Semi quantitative DrsGFP reporter assay to identify upstream effectors of Drs expression by RNAi in RasV12-glands. Schematic representation (left) of the Toll-/imd-pathways showing components with (green) and without an effect (red). The three distinguished phenotypes (right) were scored per replicate, means and standard deviations plotted (middle) and p-values for ‘None’-phenotype calculated via Dunn’s test based on Kruskal-Wallis rank sum test (***p<0.001, n.s. p≥0.05). Monitoring melanization in pupae (insets) confirmed RasV12-expression to avoid false positives (Hauling et al., 2014). (B) dl staining (red) and DrsGFP signal (green) in RasV12-glands with and without dl-knock-down. Dashed lines added to aid separating proximal part (PP) from distal part (DP). (C) In-situ hybridization for endogenous Drs in RasV12-glands hetero- or homozygous mutant for dl (dl15). Insets: (B) DAPI. Scalebars: (B–C) 100 μm.

The dl-Drs-relationship in RasV12-glands was further confirmed by the strong correlation of nuclear signals between dl-immunofluorescence and Drs-reporter intensities across PP and DP at 96 hr AED (Figure 2—figure supplement 2B; see ‘Materials and methods’). Moreover, at 96 hr AED dl was present in the entire secretory epithelium of both RasV12- and w1118-control glands. In contrast, at 120 hr AED its expression was solely confined to the PP (Figure 2B). This overlapped with the Drs-mRNA expression as determined by in situ hybridization (ISH; Figure 2—figure supplement 2C). Furthermore, both SG-specific knock-down or whole organism homozygous knock-out (dl15) of dl abolished the majority of Drs-expression in the DP of the gland at 96 hr AED and reduced it in the PP at both time points (Figure 2A–C; Figure 2—figure supplement 2C–D). On the contrary, RasV12-glands hetero- or homozygous mutant for Myd88 (Myd88KG03447) do not exhibit abolished Drs-mRNA-expression at 120 hr AED in line with our results from the reporter assay (Figure 2—figure supplement 2E).

To assess the extent to which Drs relies on dl for its expression, we measured the effect size on Drs expression of the employed dlRNAi-construct in RasV12-glands (Figure 2—figure supplement 3A). A reduction of Drs-expression to 6% in the DP of 96-hr-old dlRNAi;RasV12- compared to RasV12-glands indicates not only a strong reliance for Drs on dl in this part at 96 hr, but also a high efficiency of the dlRNAi-construct. However, a less pronounced decrease of Drs-expression in the PP at 96 hr and 120 hr AED to 27% and 52% upon dlRNAi-coexpression implies the presence of additional Drs-regulating factors. Putative candidates for this Drs-regulating role were identified by screening transcription factor binding sites amongst the up- and downregulated genes in RasV12-glands and all other acquired transcriptomes (Figure 2—figure supplement 3B; see Figure 3 for lglRNAi;RasV12, lglRNAi;RasV12 – PP). Of the six identified candidates, only one RNAi-line targeting the transcription factor Mef2 significantly reduces the expression of Drs as assayed by DrsGFP-reporter signal at 96 hr AED (Figure 2—figure supplement 3C). This is further emphasised by clusters of putative Mef2-binding sites just downstream of the Drs-locus and in the 5’-ends of the Dif- and dl-loci, implying both direct and indirect regulatory potential on Drs-expression (Figure 2—figure supplement 3D).

Figure 3 with 2 supplements see all
Hypertrophic RasV12-glands induce parallel immune and stress responses.

(A) Common and specific genesets significantly upregulated (log2(beta) ≥1; q-value ≤0.05) in either RasV12, lglRNAi;RasV12, or lglRNAi;RasV12-PP compared to w1118-glands. (B) PCA including all transcriptome replicates of all sequenced genotypes. (C) Left: Comparative transcriptome analysis between proximal part (PP) of lglRNAi;RasV12- and entire w1118-glands. Significantly differentially expressed genes (log2(beta) ≥1; q-value ≤0.05) and genes belonging to GO-term ‘immune response’ (GO:0006955) highlighted in yellow and blue. Right: Gene expression in RasV12, lglRNAi;RasV12 or lglRNAi;RasV12-PP compared to w1118-glands for immune genes significantly upregulated in the PP. Blue arrows indicate strongest expression in the PP for the indicated genes between all three groups. Missing bars indicate absence of expression values in the RNAseq data. (D) GO term enrichment among significantly upregulated genes in RasV12-glands including terms related to activation of JNK (green) and immune responses (red). Numbers in bars indicate amount of upregulated genes belonging to associated GO term. (E) qPCR results for canonical JNK target genes (log2-transformed, fold-change over Rpl32) at 96 hr after egg deposition (AED). Lower/upper hinges of boxplots indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR and bar represents median. Significance evaluated by Student's t-tests (***p<0.001). (F) TREGFP1b reporter (green) signal in RasV12- and control-glands at 96 hr and 120 hr AED. Scalebar: 100 µm.

Together, our results indicate that the spatio-temporal dynamics of Drs expression are strongly correlated to the decrease in endogenous dl expression, independent of canonical Toll- and imd-signaling. While dl expression is RasV12-independent, Drs is only expressed in the presence of dl during RasV12-induced hypertrophy. Alongside dl, Mef2 regulates Drs-activation emphasising the notion of non-canonical activation of this immune effector even further.

Hypertrophy in SGs induces parallel immune and stress responses

Between 96 hr and 120 hr AED, the decrease in Drs-expression (Figure 1E; Figure 2B–C) is correlated with deterioration of tissue integrity (Figure 1A–B) in the DP of hypertrophic RasV12-glands. Thus, we hypothesized that the tissue-autonomous immune response revealed by Drs-expression aids in preventing the collapse of nuclear as well as cellular integrity until 96 hr AED in the DP and due to its prolonged expression in the PP until 120 hr AED and beyond (Figure 1—figure supplement 1I).

To identify possible targets and effector mechanisms of the immune response that buffer the detrimental effects of hypertrophic growth, we analyzed the transcriptome data acquired for the w1118-control and RasV12-glands at 96 hr AED in further detail. In order to distinguish whether the differences between RasV12- and lglRNAi;RasV12-glands (Figure 1—figure supplement 1C–H) are of quantitative or qualitative nature and to characterize the PP with its persistent dl and Drs expression in depth, we also profiled transcriptomes of entire lglRNAi;RasV12-glands and solely the PP of lglRNAi;RasV12-glands at 96 hr AED (Figure 3—figure supplement 1A; see ‘Materials and methods’).

The sets of significant and differentially upregulated genes (i.e. log2(beta) ≥1; q-value ≤0.05) for all three conditions (i.e. RasV12 / lglRNAi;RasV12 / lglRNAi;RasV12-PP each normalized to w1118-controls) were intersected to determine common and specific genesets (Figure 3A). Notably, the two biggest genesets are differentially upregulated genes shared between all three conditions and genes specifically upregulated in the PP (Figure 3A). Furthermore, while all 6 RasV12- and lglRNAi;RasV12-replicates are in close proximity along the first two principle components in the PCA, the sets of replicates for w1118 and especially the PP are distant from the rest along PC2 emphasizing the distinctiveness of the PP compared to the rest of the gland (Figure 3B).

By screening for enriched gene ontology (GO) terms amongst the significantly upregulated genes in the PP of lglRNAi;RasV12-glands in comparison to the entire w1118-glands, we identified genes belonging to the GO-term ‘immune response’ (i.e. GO:0006955) as significantly enriched (Figure 3C; p-value=1.45⋅10−4). Detailed examination of the fold-enrichment of these genes across all three experimental groups (i.e. RasV12, lglRNAi;RasV12, lglRNAi;RasV12-PP) showed a preferential expression in the PP, too (17 of 30 genes highest expressed in PP; blue arrowheads in Figure 3C). In addition, 5 of the top 20 upregulated genes in the PP belong to this GO-term as well. Thus, the Drs-expression we observed using reporter lines serves as a proxy for a more complex, tissue-autonomous immune response in hypertrophic glands, especially in the PP. Nonetheless, Drs itself remains one of the top significantly, differentially upregulated genes across all three conditions (Figure 1D; Figure 3C; not shown for lglRNAi;RasV12). In fact, Drs-expression in the PP compared to the two conditions for entire glands is even further increased, confirming the strong tendency for proximal over distal Drs-GFP reporter activation (Figure 1E; Figure 2B; Figure 3C). Moreover, AttD and Drs are the only two AMPs that showed considerable expression in the two groups involving whole gland samples (i.e. RasV12, lglRNAi;RasV12). Strong expression of particular genes in the PP might occur as expression throughout the whole gland samples. Therefore, we tested AttD expression separately in PP and DP, which showed almost exclusive AttD-expression in the PP at 96 hr and 120 hr AED (Figure 3—figure supplement 1D). Similarly, CecA1 as the 4th highest expressed AMP in the PP-RNAseq was also confined exclusively to the PP as evaluated by ISH (Figure 3—figure supplement 1C,E). Thus, based on the RNAseq and additional evaluations, Drs appears to be the only AMP to be significantly expressed in the DP until 96 hr AED, while in parallel the PP shows a much broader immunocompetence.

A similar GO-term analysis amongst significantly, upregulated genes in entire RasV12- or lglRNAi;RasV12-glands revealed the enrichment of genes associated with GO-terms related to ‘growth’, ‘salivary gland development’ and ‘EGFR signaling’ consistent with the studied tissue and the induced RasV12-overexpression. Amongst the genes enriched in both RasV12- and lglRNAi;RasV12-glands were also sets significantly overlapping with GO-terms indicating activated JAK-STAT-signaling (data not shown). By using a construct reporting Stat92E-expression, we were able to confirm this geneset enrichment and show restriction of Stat92E-expression to the PP throughout development of RasV12-glands (Figure 3—figure supplement 1F; Bach et al., 2007). This may implicate JAK-STAT-signaling in the maintenance of immune competence in the PP and underlines the separation into PP and DP along the gland epithelium further. Moreover, the lack of unique, significantly enriched GO terms and thus distinct gene expression signatures for either RasV12- or lglRNAi;RasV12-glands at 96 hr AED excludes qualitative differences between these two genotypes as an explanation for their phenotypic differences (Figure 3—figure supplement 1B).

Importantly, we detected signatures of an activated JNK-cascade as well as cell death in RasV12- and lglRNAi;RasV12-transcriptomes indicating the presence of an activated stress response and confirming the stimulation of PCD as implied by nuclear disintegration at 120 hr AED (Figure 1B; Figure 3D; Figure 3—figure supplement 1B; GO:0006955; GO:0008219). To validate these signatures further, we performed qPCR for canonical JNK-targets at 96 hr AED and found that the expression of all tested genes was significantly increased compared to w1118-control glands (Figure 3E; Figure 3—figure supplement 2A). We used the TRE-GFP1b-reporter construct, which recapitulates JNK-activation by expressing GFP under the control of binding sites for JNK-specific AP-1 transcription factors (Figure 3F; Chatterjee and Bohmann, 2012). This not only confirmed JNK-signaling in RasV12-glands, but also uncovered its prevalence in the DP of these glands at 96 hr and 120 hr AED consistent with the transcriptome data (Figure 3F; Figure 3—figure supplement 2B). Moreover, the elevated fluorescence signal in the DP at 120 hr AED compared to 96 hr AED implies an increase in activation over time correlating with the collapse of nuclear and cellular integrity during this period.

In summary, the transcriptome analysis confirms our findings that RasV12-overexpression induces a strong tissue-autonomous immune response in the PP of the SG, beyond sole Drs expression. In contrast, the DP shows a striking increase in JNK-signaling which correlates with decreasing Drs expression and cellular and nuclear disintegration at 120 hr AED, consistent with the described role of JNK target genes in PCD.

Drs overexpression and JNK inhibition prevent RasV12-induced tissue disintegration

The increase in JNK-signaling in the DP of the SG as revealed by our transcriptome analysis coincided in space and time with the downregulation of Drs suggesting an interaction between the tissue-autonomous immune response and the stress response.

To test this assumption, we first overexpressed either Drs or a dominant negative form of the Drosophila Jun kinase jnk (basket) individually with RasV12 throughout the entire secretory epithelium of the SG (Figure 4A). Either Drs overexpression or JNK inhibition had a profound effect on the gland size, which was significantly increased at 120 hr AED, compared to RasV12 or UAS-dilution control mCD8::RFP;;RasV12 only (Figure 4B–C; Figure 4—figure supplement 1A).

Figure 4 with 1 supplement see all
Drs overexpression and JNK inhibition individually prevent tissue disintegration.

(A) Drs-specific in-situ hybridization identifies endogenous (RasV12, jnkDN;;RasV12) and exogenous (Drs, Drs,RasV12) Drs expression. (B) Collagen-GFP trap (vkgG00454, green) and Hemese staining (red) identify integrity of BM and hemocyte attachment to gland surface. Arrows indicate BM-free areas occupied and surrounded by hemocytes. Insets show DAPI (white) staining. (C) Size of salivary glands measured as outlined area in captured images of indicated genotypes at 96 hr and 120 hr after egg deposition (AED). (D) Hemocyte attachment at 96 hr and 120 hr AED represented as ln(Hemese-area)/ln(SG-area). (E) log2-transformed, Rpl32-normalized gene expression values for matrix metalloproteinases obtained by qPCR at 96 hr and 120 hr AED. (F) Hemocyte attachment at 120 hr AED in RasV12-glands hetero- and homozygous mutant for Mmp2 (Mmp2k00604). (G) Effect size of hetero- and homozygous mutant Mmp2-allele (Mmp2k00604) on hemocyte attachment to and compared to RasV12-glands. Mean of hemocyte attachment in RasV12-glands at 120 hr was set to ‘1’. (E) and (F) represent the same data points. Scalebars: (A–B) 100 µm. Boxplots in (C–G): lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student's t-tests (***p<0.001, **p<0.01, *p<0.05, n.s. p≥0.05).

Importantly, despite this size increase, glands of both genotypes (i.e. Drs,RasV12 / jnkDN;;RasV12) showed no morphological abnormalities and resembled control w1118- much more than RasV12-glands (Figure 4A–B). Strikingly, jnkDN;;RasV12- and Drs,RasV12-glands were completely devoid of attached hemocytes (Figure 4B,D). Given the basement membrane’s (BM) role in directly regulating organ morphology, the rescue of the gland shape upon coexpression of either Drs or jnkDN with RasV12 pointed towards changes in the integrity of the BM (Ramos-Lewis and Page-McCaw, 2019). In addition, previous reports suggested that hemocytes are only recruited to tissue surfaces upon tissue disintegration and when the integrity of the BM is lost (Kim and Choe, 2014). To trace the BM we used an endogenous GFP-trap in the viking gene, which encodes one subunit of CollagenIV (Figure 4B). Both, in the presence of Drs or by inhibiting jnk, the BM remained a continuous sheet surrounding the entire gland, whereas the BM on the surface of RasV12-glands was clearly disrupted.

Matrix metalloproteinases (Mmps) are likely candidates for executing the disruption of the BM and known target genes of the JNK-pathway (Uhlirova and Bohmann, 2006; Hauling et al., 2014; Stevens and Page-McCaw, 2012). In order to see whether hypertrophic glands express Mmps, we performed qPCRs for Mmp1 and Mmp2 at 96 hr and 120 hr AED on RasV12- and control w1118-glands. At 96 hr AED, RasV12-glands already exhibit increased Mmp1 expression (Figure 4E). However, Mmp2 only reaches a significant level of expression at 120 hr AED coinciding with the appearance of hemocyte attachment (Figure 4D–E). SG-wide, RasV12-independent overexpression of Mmp2, but not of Mmp1, caused opening of the BM and hemocyte attachment to the surface (Jia et al., 2015). In contrast, knock-out of Mmp2 reduces hemocyte recruitment significantly compared to RasV12-glands (Figure 4F–G; 33% in Mmp2-/+;RasV12; 34% in Mmp2-/-;RasV12). In turn, residual hemocytes still attached to the gland surface of Mmp2-/-;RasV12-glands were not activated (i.e. no filo- or lamellipodia, no spreading; Figure 4—figure supplement 1C). This indicates the necessity for a JNK-dependent expression of Mmp2 in the hypertrophic RasV12-glands prior to the recruitment of hemocytes to the tissue surface. However, the presence of residual hemocytes indicates additional cues such as reactive oxygen species (ROS) as recently suggested for neoplastic tumors (Diwanji and Bergmann, 2019). In fact, while RasV12-glands show accumulation of ROS in the DP, this accumulation is reverted upon Drs-overexpression or JNK-inhibition (Figure 4—figure supplement 1D). A parallel block of ROS-accumulation and Mmp2-dependent BM-degradation in Drs,RasV12 and jnkDN;;RasV12 could explain the absence of any attached hemocytes on the surface of these glands (Figure 4B,D).

Taken together, overexpression of Drs alone is sufficient to mimic the inhibition of the JNK-pathway in RasV12-glands: both lead to excess hypertrophic growth compared to RasV12-glands, but simultaneously prevent tissue disintegration and PCD. Prohibiting the disruption of the basal membrane reduces hemocyte recruitment and activation, thus preventing the cellular immune response from sensing hypertrophic growth. This in turn suggests that the endogenous Drs expression in RasV12-glands is seminal for maintaining nuclear and tissue integrity and thus part of the buffer mechanism to adapt to continuous growth signaling.

Drs inhibits JNK-signaling

The strong correlation between loss of Drs and the increase in JNK-signaling in the DP of RasV12-glands between 96 hr and 120 hr AED indicated an active interaction between Drs and the JNK-pathway, which prompted us to resolve their hierarchy by epistatic analysis.

Coexpression of jnkDN with RasV12 did not deplete Drs in the PP of 120-hr-old glands, since neither the fluorescence signal of the Drs-GFP reporter nor staining for endogenous Drs-mRNA via ISH showed any effects (Figure 4A; Figure 5—figure supplement 1A). This excludes a direct regulation of Drs by JNK-signaling and also negates indirect reduction of Drs-expression in the DP due to JNK-induced PCD at 120 hr AED (see Figure 6 for more details). qPCR for Drs in RasV12- and jnkDN;RasV12-glands confirmed these results further (Figure 5—figure supplement 1B). In contrast, qPCR in hypertrophic RasV12-glands showed a significant reduction in expression of JNK target genes upon coexpression of Drs at 96 hr and 120 hr AED (Figure 5B; Figure 5—figure supplement 2C,D), which could not be reciprocated by coexpressing mCD8::RFP as a UAS-dilution control (Figure 5—figure supplement 3A) and which was in line with a decrease in activated jnk (Figure 5C,E; Figure 5—figure supplement 3B–C) and TRE-GFP1b signal in Drs,RasV12- compared to RasV12-glands (Figure 5B,D,F). Thus, the overexpression of the AMP Drs actively and tissue-autonomously inhibits the JNK-dependent stress response in hypertrophic RasV12-glands beyond 96 hr AED.

Figure 5 with 3 supplements see all
Drs overexpression inhibits JNK-activation.

(A) Schematic representation of the JNK-pathway including read-outs (green) employed to track its activation. (B) qPCR results for canonical JNK-target genes (log2-transformed, fold-change over Rpl32) at 120 hr after egg deposition (AED). (C) Quantification of activated jnk by staining with phosphorylation-sensitive antibody normalized for the salivary gland (SG) size per gland at 120 hr AED. (D.) Quantification of TREGFP1b signal normalized for SG size per gland indicating JNK-dependent transcriptional activation at 120 hr AED. (E) Visualization of phosphorylated jnk in RasV12-glands with and without Drs-coexpression. (F) Distribution of TREGFP1b reporter signal in RasV12-glands in the presence or absence of coexpressed Drs. Insets: (E–F) DAPI. Scalebars: (E–F) 100 µm. Boxplots in (B–D): lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student's t-tests (***p<0.001).

In addition, we used an efficient RNAi-line against Drs, which in combination with RasV12 reduced Drs-expression significantly compared to its expression in RasV12-glands as shown by qPCR and ISH (Figure 5—figure supplement 2A–D; 8% residual Drs-expression at 96 hr and 4% at 120 hr in DrsRNAi;RasV12- compared to RasV12-glands). All significantly upregulated JNK-target genes in RasV12-glands apart from upd2 were further increased upon knockdown of Drs at 96 hr AED (Figure 5—figure supplement 2C). However, this effect ceased at 120 hr AED (Figure 5—figure supplement 2D) consistent with a lack of change in activated jnk (Figure 5—figure supplement 2F,H; Figure 5—figure supplement 3B–C) and TRE-GFP1b-signal (Figure 5—figure supplement 2E,G), as well as the canonical loss of Drs expression in the DP of RasV12-glands (Figure 1E). This result demonstrates that until 96 hr AED the endogenously expressed Drs has the same inhibitory effect on the JNK-pathway as the Drs-overexpression has at 120 hr AED.

Drs prevents cell death in RasV12-glands

The capacity to induce apoptosis is a well-established function of the JNK-pathway in Drosophila (Uhlirova et al., 2005; Igaki et al., 2006; Enomoto et al., 2015; Uhlirova and Bohmann, 2006; Cordero et al., 2010). Combined with the identification of ‘cell death’ as a significantly enriched term in the GO-analysis, we hypothesized that the observed nuclear disintegration in RasV12-glands after 96 hr AED was a consequence of the JNK-dependent induction of PCD (Figure 1B–C; Figure 1D–F; Figure 3D; Figure 3—figure supplement 1B).

In mitotic tissues, the apoptotic inducer head involution defective (hid) is inhibited by Ras/MAPK-signaling (Bergmann et al., 1998; Kurada and White, 1998). However, in RasV12-glands hid expression gradually increased from 96 hr to 120 hr AED coinciding with the increase in nuclear disintegration (Figure 6A). Coexpression of Drs with RasV12 significantly decreased hid expression at both time points, while Drs-knock-down in RasV12-glands increased hid expression even further already at 96 hr AED. Together with the reduction in hid expression upon JNK-inhibition in jnkDN;;RasV12-glands, Drs emerges as a negative regulator of the apoptotic inducer hid by inhibiting JNK-signaling (Figure 5—figure supplement 1B).

Figure 6 with 1 supplement see all
Drs inhibits programmed cell death.

(A) Hid expression as measured by qPCR and plotted as log2-transformed values normalized to Rpl32-expression in RasV12-glands with in- (Drs) or decreased (DrsRNAi) Drs expression. (B) Schematic representation of the JNK-pathway including used methods of interference (green). (C) Nuclear volumina derived from z-stacks of DAPI-stained salivary glands (SGs) and averaged per gland at 96 hr and 120 hr after egg deposition (AED). (D) DAPI-stained (white) SG nuclei to indicate nuclear size and disintegration. (E–F) RasV12-glands with increased Drs expression stained with anti-CC3-antibody to detect Dronc activation (red) and corresponding quantifications of detected signal normalized for SG size. Insets in (E) show DAPI and scalebars in (D–E) represent 100 µm. Lower/upper hinges of boxplots in (A, C, F) indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student’s t-tests (***p<0.001).

To evaluate the induction of PCD in RasV12-glands, we either inhibited JNK-signaling at the level of jnk activation or coexpressed the caspase-inhibitor p35 with RasV12 and examined nuclear volume and integrity (Figure 6B–D). p35 inhibits Drice and thus blocks PCD at the level of effector caspase activation (Hay et al., 1994; Meier et al., 2000; Hawkins et al., 2000). While both interventions successfully blocked nuclear disintegration, the additional rounds of endoreplication as observed in RasV12-glands were not significantly suppressed (Figure 6C–D). Crucially, Drs-coexpression in RasV12-glands phenocopies the inhibition of JNK-signaling and effector caspases both in terms of restoring nuclear integrity as well as the persistence of excess endoreplications, in spite of a trend of overall nuclei size reduction similar to p35;RasV12-glands.

Last, to validate the inhibition of PCD by Drs, we monitored Dronc activity using the cleaved caspase 3 (CC3) antibody. In fact, the strongest Dronc activity occurred in cells that also displayed heavy disintegration of nuclei, confirming the relation between caspase activation and nuclear disintegration as part of PCD (Figure 6—figure supplement 1B). However, in Drs,RasV12-glands Dronc was significantly less activated compared to RasV12-glands, a phenotype which we showed to be independent of UAS-dilution effects (Figure 6E–F; Figure 6—figure supplement 1A–D; Fan and Bergmann, 2010). Given the strong reduction of Dronc activity upon knock-down of hid in RasV12-glands, the inhibitory effect of Drs on PCD is most likely to a large extent dependent on hid-inhibition (Figure 6—figure supplement 1E–F).

Taken together, we show that JNK-dependent induction of PCD is a consequence of the collapse in adaptation to RasV12-induced hypertrophy. Importantly, overexpression of Drs in RasV12-glands prevents PCD by blocking full JNK-activation. This is in contrast to recent observations where AMPs, including Drs, act pro-apoptically aiding in the elimination of tumor cells and limiting tumor size (Araki et al., 2019; Parvy et al., 2019).

Drs inhibits the JNK-feedback loop

Various Drosophila models for tissue transformation have shed light on the functional separation of the JNK-pathway into an upstream kinase cascade leading to jnk-activation and a downstream feedback-loop converging again on jnk activation (Figure 7A; Fogarty et al., 2016; Pérez et al., 2017; Shlevkov and Morata, 2012; Muzzopappa et al., 2017).

Figure 7 with 1 supplement see all
Drs inhibits the JNK-feedback loop.

(A) Schematic representation of the JNK-feedback loop including levels of interference (blue) and used read-outs of its activity (green). hid-overexpressing glands with or without coexpressed Drs were quantified for (B) activated jnk via a phosphorylation sensitive antibody and (C) activated Dronc via the CC3-antibody. (D) Activated jnk (pJnk) was quantified and normalized to tissue size in RasV12-glands upon knockdown of Dronc. (E) Salivary glands solely overexpressing hid with or without Drs-coexpression were stained for activated jnk (green) and activated Dronc (red). (F) Activated jnk (red) was detected in RasV12-glands with or without Dronc-knock-down at 120 hr AED. Insets: (E–F) DAPI. Scalebars: (E–F) 100 µm. Boxplots in (B–D): lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student's t-tests (***p<0.001, **p<0.01).

Our results revealed a negative regulatory impact of Drs on JNK-signaling in RasV12-glands, but did not determine which part of the pathway is targeted by Drs. To answer this question, we uncoupled the feedback loop from the upstream kinase cascade by solely overexpressing hid (Figure 7A). Irrespective of the differing models for the signal propagation downstream of Dronc, this initiator caspase remains seminal for establishing the actual feedback with jnk (Fogarty et al., 2016; Shlevkov and Morata, 2012). Thus, we stained glands for activated Dronc and jnk after a pulse of hid-overexpression during the larval wandering stage (Figure 7A). As expected, hid-expressing glands showed highly elevated levels of activated Dronc, but also jnk, which indicates the presence of feedback activation. Moreover, both phenotypes were reversed upon coexpression of Drs during the hid-expression pulse (Figure 7B–C,E).

To clarify that Drs operates in a similar fashion during RasV12-induced hypertrophic growth, we stained RasV12-glands (Figure 6E–F) for activated jnk in the presence and upon knock-down of Dronc (Figure 7D,F; Figure 7—figure supplement 1A–B). In fact, the absence of a signal for activated jnk in DroncRNAi;RasV12- compared to strong RasV12-glands confirms the presence of a genuine feedback-regulation as part of JNK-signaling in SGs. Thus, Drs seems to inhibit JNK-signaling downstream of hid and upstream of Dronc and jnk, emphasizing an inhibition of the JNK-feedback loop rather than the initial kinase cascade.

Discussion

Tissue-autonomous vs. cellular immune response mediated via JNK-signaling

By overexpressing RasV12 in larval SGs, we made use of and overrode the gland’s ability to adapt to growth signals. Activating constitutive Ras/MAPK-signaling allowed us to identify local immune and stress responses as part of a buffering mechanism to compensate the accumulation of stress and decipher natural limits of growth adaptation (Hauling et al., 2014). Remarkably, in this context, the local immune response inhibits the parallel stress response, an effect that to our knowledge has not been described before. Central to this inhibition is the AMP Drs, which directly impinges on the JNK-pathway and thereby subsequently also on inducing PCD, a function completely opposite to previous observations in other tissues (Araki et al., 2019; Parvy et al., 2019). These results also indicate an unprecedented role for an AMP as a signal transducer enabling tissue-autonomous crosstalk between immune and stress pathways. Dependent on the extent of JNK-inhibition by the local, Drs-dependent immune response and thus the integrated decision of both on the state of the tissue’s homeostasis, the gland epithelium attracts hemocytes as part of a wider, cellular immune response. By virtue of inhibiting JNK-signaling, tissue-autonomous and cellular immune responses antagonize each other, a balance that determines continuous hypertrophic growth or its restriction (Figure 8). The latter has far reaching implications for therapeutic approaches that need to consider the adverse effects that stimulating immune responses might have on tissue growth after damage and under stress.

Tissue-autonomous antagonizes cellular immune response via JNK-inhibition in RasV12-glands .

dl-mediated Drs expression inhibits JNK activation in the entire RasV12-gland until 96 hr after egg deposition (AED). At 120 hr AED, dl and thus Drs expression is reduced to the PP, derepressing full JNK-activation in the DP. Consequently, JNK-dependent expression of hid and Mmp2 participate in stimulating nuclear and tissue disintegration, which eventually leads to the interference with tissue growth and integrity by the cellular immune response.

Drs is expressed as part of a genuine tissue-autonomous immune response

The SGs of Drosophila larvae are an integral part of its gastrointestinal system and the lumen of the mature glands forms a continuum with the exterior. As such the glands are constantly exposed to microbial and pathogenic influences, which predestines them to act as a dedicated immunological barrier epithelium. However, raising larvae with RasV12-glands under strictly sterile conditions corroborated the authenticity of the immune response as tissue-autonomous and with high likelihood independent of exogenous or systemically distributed pathogenic stimuli. This is further emphasized by the dependency of Drs expression on the tissue-specific overexpression on RasV12 and tissue-autonomous manipulations (i.e. Drs or jnkDN overexpression) leading to the inhibition of JNK-activation and hemocyte recruitment. The activation of immune effectors in the absence of exogenous or endogenous pathogens is also consistent with the definition of sterile inflammation and aligns with finding in other models (Shaukat et al., 2015; Kenmoku et al., 2017; Asri et al., 2019). However, Drs is embedded into a wider spatio-temporally regulated activation of immune effectors and immune mechanisms providing immunocompetence to the PP until 120 hr AED and thus constituting a genuine immune response (Figure 3C–D; Figure 3—figure supplement 1B–C). Follow-up studies will determine the mode of their activation under physiological and sterile conditions and thus clarify their nature in the context of inflammation and immunity.

Apart from dl and Mef2 neither the homeobox transcription factor Caudal as in the adult glands nor any other of the canonical members belonging to the Toll- and imd-pathway are involved in Drs expression (Hauling et al., 2014; Ferrandon et al., 1998; Ryu et al., 2004). Under wild-type conditions, dl is expressed throughout the entire gland epithelium until 96 hr AED, but only remains expressed in the PP at 120 hr AED (Figure 2B). This spatio-temporal expression pattern in turn determines to a large extent the expression of Drs in RasV12-glands essentially separating the gland at 120 hr AED into an immunocompetent duct-proximal and a stress-responsive duct-distal compartment (Figure 2B; Figure 2—figure supplement 3A; Figure 8). Due to the absence of Drs in the DP at 120 hr AED, stress-responsive JNK-signaling finally exceeds a critical threshold leading to elevated Mmp2- and hid-expression. This stimulates the onset of PCD and opening of the basal membrane as an important prerequisite for attachment of hemocytes that are subsequently recruited to the surface of the SGs (Figure 4B,D; Figure 8; Hauling et al., 2014). To the contrary, dl continues to induce Drs in the PP of RasV12-glands until 120 hr AED in line with the complete absence of hemocyte recruitment to this part of the gland (Figure 1F; Figure 1—figure supplement 1G–H). However, Drs becomes less dependent on dl between 96 hr and 120 hr AED. While its precise role in Drs-regulation especially in relation to dl remains to be elucidated, Mef2 might influence Drs-expression directly or indirectly parallel to ceasing dl-influence (Figure 2—figure supplement 3B–C).

Drosomycin impinges on JNK feedback activation

In depth analysis of wound regeneration and tumor formation has shed light on the intricate architecture of the JNK-pathway and its signal propagation (Santabárbara-Ruiz et al., 2015; Pinal et al., 2018). This has led to the discovery of feedback- or self-sustenance-loops as part of JNK-signaling (Fogarty et al., 2016; Pérez et al., 2017; Shlevkov and Morata, 2012; Muzzopappa et al., 2017). Similarly, several lines of evidence validated the existence of a complex quantitatively and qualitatively regulated activation of the JNK-pathway in RasV12-glands. In fact, the strong reduction of jnk activation upon knock-down of the initiator caspase Dronc in the RasV12-background at 120 hr AED indicates that a feedforward-loop is also part of the propagation of JNK-activation in hypertrophic glands (Figure 7D,F; Figure 7—figure supplement 1A–B). Consistently, even sole hid overexpression in SGs led to Dronc as well as jnk activation emphasizing the presence of this feedback-regulation as part of the JNK-pathway further (Figure 7B–C,E).

In general, feedback loops serve as a predestined platform to integrate additional signals via crosstalk with other pathways to dynamically modulate the originally transmitted signal and thus either amplify or weaken the response (Kholodenko, 2006; Antebi et al., 2017). Here, we describe a novel mode of signal attenuation by the tissue-autonomous immune response that rather than eliminating an exogenous stimulus directly interferes with the signal propagation in the JNK-feedback loop. Fundamentally, this interaction is part of an emerging picture of crosstalk between immune and stress responses involved in organ growth and maintenance in Drosophila (Wu et al., 2015; Liu et al., 2015; Parisi et al., 2014; Hauling et al., 2014).

Drs regulates the signal propagation of the intracellular module of the JNK-pathway and throughout our experiments only directly Drs-expressing cells prevented RasV12-induced cell disintegration and PCD. However, it remains an outstanding question whether Drs functions exclusively cell-autonomously and intracellularly or whether secreted Drs operates in an autocrine manner too. Clonal analysis and rescue experiments will serve this purpose in the future. Further work on the effector mechanism of Drs will also elucidate further details about the components of the JNK-feedback loop in- or directly regulated by Drs. This will also contribute to mapping the manifold modes of immune-stress-crosstalk in Drosophila and find general patterns among them beyond a sole dependency on the specific context.

Drs promotes hypertrophic growth and inhibits PCD

As our Drs,RasV12-experiments indicated, under conditions of continuous growth and therefore chronic stress induction, the ability to suppress the JNK-pathway in a Drs-dependent manner supports continuous, hypertrophic growth of RasV12-glands and the survival of the tissue. Moreover, a prolongation of Drs-expression beyond its endogenous decrease inhibits the induction of PCD and the recognition of the cellular immune response and thus renders the hypertrophic gland unchallenged.

In fact, while Ras/MAPK-signaling in mitotic tissues crucially suppresses apoptosis by downregulating hid expression, this effect is revoked in hypertrophic RasV12-glands (Bergmann et al., 1998; Kurada and White, 1998). Although the reason for this difference remains to be elucidated, it is in fact Drs which operates as the inhibitor of apoptotic inducers in hypertrophic glands. This function is central to the suppression of PCD in RasV12-glands (Figure 6A).

This differs fundamentally from the pro-apoptotic function of AMPs, which was recently described for two tumor models. In both, disc (discs large, dlg) (Parvy et al., 2019) and leukemic tumors (mxcmbn1) (Araki et al., 2019), AMPs were shown to target tumor cells and limit tumor size by inducing apoptosis. In addition to the differences between the tumorous tissues (i.e. proliferative discs and lymph glands), the fact that Drs is induced locally in SGs and acts tissue-autonomously may explain its different activities.

Hypertrophic SGs are a remarkable system to discover buffer mechanisms

Being incapable of cell proliferation, damaged postmitotic tissues cannot rely on regenerative cell plasticity like imaginal discs or stem cell-derived tissue regeneration as in the Drosophila adult midgut (Ohlstein and Spradling, 2006; Herrera et al., 2013; Herrera and Morata, 2014; Schuster and Smith-Bolton, 2015; Ahmed-de-Prado and Baonza, 2018). Instead, they need to cope with endogenous and exogenous influences via elaborate mechanisms to prevent or buffer detrimental consequences.

Here, we show that continued RasV12-expression in the larval SG overrides the dependency on nutritional cues and stimulates excess endoreplications that eventually have damaging effects on the tissue integrity by inducing elevated levels of stress. Uninterrupted induction of endoreplications has its natural limits in every system, even in SGs that are already polyploid. Eventually, continuous induction of more endocycles is challenged by nutritional restrictions in synthesizing more DNA, replication stress, spatial limitations in the gland nuclei and continuously more unsynchronized metabolic turnover. Hence, in spite of the anti-apoptotic function RasV12 conveys in mitotic tissues, unrestricted stimulation of excess endoreplications ultimately leads to cell death. Since this characterizes the final collapse of tissue homeostasis, it also allows to study the extent of buffering capacity conveyed by immune and stress-responsive signaling. While JNK-signaling has been shown to be seminal during development (i.e. tissue morphogenesis) (Agnès et al., 1999; Zeitlinger and Bohmann, 1999), its activation in RasV12-glands appears however rather likely to be due to its function in relaying stress signals, emphasizing the presence of stress as part of hypertrophic growth further.

Thus, hypertrophic RasV12-glands constitute an outstanding system to study the involvement of tissue-autonomous immune and stress-induced responses to buffer deviation from homeostasis.

Immune surveillance theory

According to the immune surveillance theory, the immune system has evolved to reduce the risk of somatic cells accumulating cancerous mutations (Burnet, 1957; Burnet, 1970). In order to reduce the danger of cell-transformation, cells express or expose molecules upon recognition of stress or damage during transformation. These markers are sensed by the immune system, which in turn eradicates the potentially harmful cells (Jung et al., 2012; Vantourout et al., 2014; Schmiedel and Mandelboim, 2018). Given the absence of exogenous or endogenous pathogenic agents, the activation of immune effectors (i.e. Drs) in the here presented system is most likely a consequence of sensing Danger or Damage Associated Molecular Patterns during hypertrophic gland overgrowth (Matzinger, 1994; Seong and Matzinger, 2004). However, the immune surveillance theory remains controversial, since tumor-associated inflammation was also shown to promote rather than suppress tumor growth (Balkwill and Mantovani, 2001; Mantovani et al., 2008). Our model bridges the gap between these two opposing views, since the effects of the tissue-autonomous and cellular immune responses appear to be antagonistic regarding the regulation of JNK-activation and thus ultimately PCD. In fact, only the integration of the various stress and immune mechanisms in hypertrophic RasV12-glands allows a concerted decision to eradicate a putatively dangerous cell via inducing PCD or not. Given the evolutionary conservation from insects to mammals of signaling pathways that govern growth control (Edgar, 2006), it is likely that mechanisms to detect and counteract a loss in regulation of these pathways, such as stress and immune pathways, are similarly conserved between both phyla.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Genetic reagent (D. melanogaster)UAS-dlRNAiBloomington36650
Genetic reagent (D. melanogaster)UAS-l(2)glRNAiVDRC109604/KKRives-Quinto et al., 2017
Genetic reagent (D. melanogaster)UAS-imdRNAiVDRC101834/KKBosch et al., 2005
Genetic reagent (D. melanogaster)UAS-FaddRNAiVDRC100333/KK
Genetic reagent (D. melanogaster)UAS-keyRNAiVDRC100257/KK
Genetic reagent (D. melanogaster)UAS-RelRNAiVDRC108469/KKCammarata-Mouchtouris et al., 2020
Genetic reagent (D. melanogaster)UAS-spzRNAiVDRC105017/KKPanettieri et al., 2020
Genetic reagent (D. melanogaster)UAS-TlRNAiVDRC1000788/KKAlpar et al., 2018
Genetic reagent (D. melanogaster)UAS-cadRNAiVDRC49562/KK
Genetic reagent (D. melanogaster)UAS-Stat92ERNAiVDRC106980/KKRecasens-Alvarez et al., 2017
Genetic reagent (D. melanogaster)UAS-DroncRNAiVDRC100424/KKKale et al., 2015
Genetic reagent (D. melanogaster)UAS-DroncRNAiVDRC23035/GDFlorentin and Arama, 2012
Genetic reagent (D. melanogaster)UAS-Myd88RNAiVDRC25402/GDLi et al., 2020
Genetic reagent (D. melanogaster)UAS-pllRNAiVDRC2889/GDWu et al., 2015
Genetic reagent (D. melanogaster)UAS-DifRNAiVDRC30578/GDWu et al., 2015
Genetic reagent (D. melanogaster)UAS-DifRNAiVDRC30579/GDWu et al., 2015
Genetic reagent (D. melanogaster)UAS-DrsRNAiVDRC2703/GD
Genetic reagent (D. melanogaster)UAS-dfrRNAiS. Certel
Genetic reagent (D. melanogaster)UAS-hidRNAiVDRC8269/GDNagata et al., 2019
Genetic reagent (D. melanogaster)UAS-foxoRNAiVDRC107786/KKMcLaughlin et al., 2019
Genetic reagent (D. melanogaster)UAS-grhRNAiVDRC33680/GD
Genetic reagent (D. melanogaster)UAS-Mef2RNAiBloomington38247Zhao et al., 2020
Genetic reagent (D. melanogaster)UAS-Nrf2RNAiVDRC101235/KKBrock et al., 2017
Genetic reagent (D. melanogaster)UAS-Nrf2RNAiVDRC108127/KKBrock et al., 2017
Genetic reagent (D. melanogaster)UAS-Sox14RNAiVDRC107146/KKWang et al., 2020
Genetic reagent (D. melanogaster)DrsGFPW.-J. Lee
Genetic reagent (D. melanogaster)TRE-GFP1bD. Bohmann
Genetic reagent (D. melanogaster)UAS-DrsB. Lemaitre
Genetic reagent (D. melanogaster)UAS-CecA1B. Lemaitre
Genetic reagent (D. melanogaster)dl15Y. Engström
Genetic reagent (D. melanogaster)Myd88KG03447Y. Engström
Genetic reagent (D. melanogaster)UAS-hidM.Suzanne
Genetic reagent (D. melanogaster)UAS-Mmp2#4A. Page-McCaw
Genetic reagent (D. melanogaster)UAS-Mmp1APM1037A. Page-McCaw
Genetic reagent (D. melanogaster)UAS-Mmp1APM3099A. Page-McCaw
Genetic reagent (D. melanogaster)UAS-RasV12Bloomington4847
Genetic reagent (D. melanogaster)BxMS1096Bloomington8860
Genetic reagent (D. melanogaster)UAS-jnkDNBloomington6409UAS-bskDN
Genetic reagent (D. melanogaster)tubP-Gal80tsBloomington7108
Genetic reagent (D. melanogaster)Mmp2k00604Bloomington10358
Genetic reagent (D. melanogaster)10xStat92E-GFPBloomington26197
Genetic reagent (D. melanogaster)UAS-p35.HBloomington5072
Genetic reagent (D. melanogaster)UAS-mCD8::mRFPBloomington27400
Genetic reagent (D. melanogaster)vkgG00454FlytrapRef. 100; CollagenIV
AntibodyAnti-Hemese (mouse monoclonal)István AndóH2IF(1:5)
AntibodyAnti-pJNK (mouse monoclonal)Cell Signaling TechnologyCat#:9255IF (1:250)
AntibodyAnti-cleaved caspase 3 (rabbit polyclonal)Cell Signaling TechnologyCat#:9661IF(1:200)
AntibodyAnti-dorsal (mouse monoclonal)DSHB7A4-39IF(1:50)
AntibodyAnti-mouse-IgG-Alexa546 (goat polyclonal)ThemoFisher ScientificCat#: A-11030IF(1:500)
AntibodyAnti-rabbit-IgG-Alexa568 (goat polyclonal)ThemoFisher ScientificCat#: A-11011IF(1:500)
Recombinant DNA reagentDrs (cDNA clone)DGRCLP03851
Recombinant DNA reagentCecA1 (cDNA clone)DGRCIP21250
Sequence-based reagentoligo(dT)16-primerThermoFisher ScientificCat#:8080128
Sequence-based reagentDrs_FThis paperqPCR primersgaggagggacgctccagt
Sequence-based reagentDrs_RThis paperqPCR primersttagcatccttcgcaccag
Sequence-based reagentAttD_FThis paperqPCR primersgtttatggagcggtcaacg
Sequence-based reagentAttD_RThis paperqPCR primerstctggaagagattggcttgg
Sequence-based reagentTIMP_FThis paperqPCR primersaacagagcgtcatggcttca
Sequence-based reagentTIMP_RThis paperqPCR primerstcacaccaaaacaggtggca
Sequence-based reagentUpd1_FThis paperqPCR primerscgggtgatcgcttcaatc
Sequence-based reagentUpd1_RThis paperqPCR primersctgcggtactcccgaaag
Sequence-based reagentUpd2_FThis paperqPCR primersaagttcctgccgaacatgac
Sequence-based reagentUpd2_RThis paperqPCR primersatccttgcggaacttgtactg
Sequence-based reagentUpd3_FThis paperqPCR primersactgggagaacacctgcaat
Sequence-based reagentUpd3_RThis paperqPCR primersgcccgtttggttctgtagat
Sequence-based reagentHid_FThis paperqPCR primerstctacgagtgggtcaggatgt
Sequence-based reagentHid_RThis paperqPCR primersgcggatactggaagatttgc
Sequence-based reagentRpr_FThis paperqPCR primersgatcaggcgactctgttgc
Sequence-based reagentRpr_RThis paperqPCR primersactgtgactcccgcaagc
Sequence-based reagentGrim_FThis paperqPCR primersatcgatgaccatgtcggagt
Sequence-based reagentGrim_RThis paperqPCR primerscgcagagcgtagcagaagat
Sequence-based reagentMMP1_FThis paperqPCR primersgtttccaccaccacacagg
Sequence-based reagentMMP1_RThis paperqPCR primersgcagaggcgggtagatagc
Sequence-based reagentMMP2_FThis paperqPCR primerstttcgatgcggacgagac
Sequence-based reagentMMP2_RThis paperqPCR primersgccacgttcagaaaattggt
Sequence-based reagentPUC_FThis paperqPCR primerscgtcatcatcaacggcaat
Sequence-based reagentPUC_RThis paperqPCR primersaggcggggtgtgtttctat
Sequence-based reagentRPL32_FThis paperqPCR primerscggatcgatatgcta
Sequence-based reagentRPL32_RThis paperqPCR primerscgacgcactctgttg
Peptide, recombinant proteinRNase-free DNase IThermoFisher ScientificCat#:EN0521
Peptide, recombinant proteinSuperscriptIIIThermoFisher ScientificCat#:18080-093
Peptide, recombinant proteinPhalloidin-546Molecular probesCat#:A22283
Commercial assay or kitKAPA SYBR FAST qPCR Master Mix (2x) kitKapa Biosystems; Sigma-AldrichKR0389, v9.13
Commercial assay or kitRNAqueous-Micro KitThermoFisher ScientificCat#:AM1931
Commercial assay or kitRNAqueous KitThermoFisher ScientificCat#:AM1912
Commercial assay or kitExperion RNA StdSens Reagents and SuppliesBio-RadCat#:700-7154
Commercial assay or kitExperion RNA StdSens ChipsBio-RadCat#:700-7153
Chemical compound, drugVancomycinSigma-AldrichCat#:V2002
Chemical compound, drugMetronidazoleSigma-AldrichCat#:M3761
Chemical compound, drugNeomycinSigma-AldrichCat#:N1876
Chemical compound, drugCarbenicillinSigma-AldrichCat#:C1389
Chemical compound, drugSodium Hypochlorite solutionFisher ScientificCat#:10401841
Chemical compound, drugDAPISigma-AldrichD9542
Software, algorithmImageJFiji contributorsv1.52nhttps://imagej.nih.gov/ij/
Software, algorithmZen softwareZeissBlue edition
Software, algorithmkallistoBray et al., 2016v0.44.0https://pachterlab.github.io/kallisto/
Software, algorithmsleuthPimentel et al., 2017v0.30.0https://pachterlab.github.io/sleuth/
Software, algorithmGOstatsFalcon and Gentleman, 2007v2.48.0https://github.com/Bioconductor/GOstats/
Software, algorithmAnnotationDbiPagès et al., 2020v1.44.0https://github.com/Bioconductor/AnnotationDbi/
Software, algorithmorg.Dm.eg.dbCarlson, 2019v3.7.0http://bioconductor.org/packages/org.Dm.eg.db/
Software, algorithmRNAseq_sg_
analysis.Rmd
This paperhttps://github.com/robertkrautz/sg_analysis/; Krautz, 2021; copy archived at swh:1:rev:82c91040d3434b215c04cfce11cf73f70300e099
Software, algorithmRNAseq_motif_
enrichment.Rmd
This paperhttps://github.com/robertkrautz/sg_analysis/
Software, algorithmscanner.ijmThis paperhttps://github.com/robertkrautz/sg_analysis/
Software, algorithmRcisTargetAibar et al., 2017v1.6.0https://www.bioconductor.org/packages/release/bioc/html/RcisTarget.html
Software, algorithmbiomaRtDurinck et al., 2009v2.42.1https://bioconductor.org/packages/release/bioc/html/biomaRt.html
Software, algorithmBiostringsPàges et al., 2020v2.54.0https://bioconductor.org/packages/release/bioc/html/Biostrings.html
Software, algorithmPWMEnrichStojnic and Diez, 2020v4.22.0https://www.bioconductor.org/packages/release/bioc/html/PWMEnrich.html

Fly husbandry and stocks

Request a detailed protocol

All crosses were reared on standard potatomash/molasses medium under tempered conditions (see 'Staging') in a 12 hr dark/12 hr light-cycle. Drs-GFP (W.-J. Lee), UAS-dfrRNAi (S.Certel), TRE-GFP1b (D. Bohmann), UAS-Drs and USA-CecA1 (B. Lemaitre), dl15 and Myd88KG03447 (Y. Engström), UAS-hid (M.Suzanne), UAS-Mmp2#4, UAS-Mmp1APM1037, and UAS-Mmp1APM3099 (A. Page-McCaw) were kind gifts from the indicated donors. The following stocks were provided by the Bloomington Drosophila Stock Center: UAS-RasV12 (4847), BxMS1096 (8860), UAS-jnkDN (UAS-bskDN; 6409), tubP-Gal80ts (7108), Mmp2k00604 (10358), 10xStat92E-GFP (26197), UAS-p35.H (5072), and UAS-mCD8::mRFP (27400). RNAi-lines were sourced from the Bloomington Stock Center and the Vienna Drosophila Resource Center and are detailed in Supplementary file 3. The generation of the CollagenIV flytrap line vkgG00454 was described in Morin et al., 2001. Please, see Supplementary file 1 for a complete list of experimental crosses.

Staging

Request a detailed protocol

Virgins were collected for 3–7 days before setting crosses. Initially, crosses were kept on standard food without antibiotics for 48 hr at 25°C. Eggs were collected for 6 hr (Immunohistochemistry and qPCR) or for 2 hr (RNASeq) at 25°C. When necessary, precollections were performed for 2 hr at 25°C prior to the actual collection. Egg collections were incubated for 24 hr at 29°C or for hid experiments 48 hr at 18°C and replicates of 24 hatched 1st instar larvae were afterwards transferred to vials with 3 ml standard food supplemented with Neomycin (0.1 mg·ml−1, Sigma-Aldrich, N1876), Vancomycin (0.1 mg·ml−1, Sigma-Aldrich, V2002), Metronidazol (0.1 mg·ml−1, Sigma-Aldrich, M3761) and Carbenicillin (0.1 mg·ml−1, Sigma-Aldrich, C1389). After incubation at 29°C for another 72 hr (96 hr AED) or 96 hr (120 hr AED), 3rd instar larvae were prepared for dissection or pictures were taken of whole larvae with a Leica MZ FLIII Fluorescence Stereomicroscope. For hid experiments, transferred larvae were incubated for 197 hr at 18°C, shifted to 29°C for 12 hr and finally dissected. To exclude microbial contamination and maintain germ-free conditions, BxGal4;;DrsGFP/UAS-RasV12-eggs were dechorionated with a 50% Sodium Hypochlorite solution (Fisher Scientific, 10401841) immediately after collection and transferred to vials with apple-agar supplemented with Nipagin, Propionic acid and the same antibiotics as above. Larvae were then analyzed at 24 hr and 48 hr AED.

Drs reporter assay

Request a detailed protocol

Replicates of Drosophila larvae (n = 24 larvae·replicate−1; N > 6 replicates) at 96 hr AED were screened for Drs-GFP reporter signals. Three phenotypes were distinguished: (1.) ‘Full’ SG pattern includes GFP signal across the entire PP and GFP+-cells with reduced intensity in the DP. (2.) ‘Partial’ SG is GFP+ throughout the PP, but less pronounced in DP with reduced signal intensity and fewer GFP+-cells. (3.) ‘None’ phenotype lacks signal throughout the entire gland. Distribution of phenotypes were scored per replicate and significance calculated for the ‘None’-phenotype via Dunn’s test after performing the Kruskal-Wallis rank sum test.

qPCR

Request a detailed protocol

Total RNA of dissected SGs was isolated with the RNAqueous-Micro Kit (ThermoFisher Scientific, AM1931) and from whole larvae with the RNAqueous Kit (ThermoFisher Scientific, AM1912) according to the instructors manual. Residual genomic DNA was digested with RNase-free DNaseI (ThermoFisher Scientific, EN0521) and cDNA reverse transcribed with SuperscriptIII (ThermoFisher Scientific, 18080–093) while using oligo(dT)16-primer (ThermoFisher Scientific, 8080128). Quality of all prepared totalRNA-extractions was evaluated on a 5 mM Guanidinium Thiocyanate-agarose gel, for optimization purposes on a BioRad Experion system (RNA StdSens Assay, 7007153, 7007154) and totalRNA for sequencing was run on a 2100 Bioanalyzer Instrument (Agilent Technologies, 5067–1511). qPCR reactions were set as technical triplicates with KAPA SYBR FAST qPCR Master Mix (KR0389, v9.13) including 200 nM final concentration of forward and reverse primers and run on a Rotor-Gene Q 2plex HRM machine (9001550). See supplementary file 2 for list of all used qPCR primers. The number of biological replicates acquired and assayed in the experiment represented in the individual figures are denoted underneath the respective graphs.

RNASeq library preparation and analysis

Request a detailed protocol

To avoid variability and thus confounding influences among the various RNASeq sample groups, we controlled rigorously for age of female parents, larval density to avoid larval crowding, age differences as well as developmental age itself and bacterial influences by using axenic culture conditions. For each genotype, three biological replicates were dissected with 30 pairs of SGs for the three groups including whole glands and 60 pairs of PPs for the corresponding group to acquire 5 µg of total RNA. No power or sample size calculations were performed. Replicate numbers were determined by the available resources and funds.

Poly(A)-containing mRNA molecules from totalRNA-samples were purified with oligo(dT)-magnetic beads, subsequently fragmented and cDNA synthesized with random primers using the TruSeq RNA Sample Preparation Kit v2. Adapter ligation and PCR-amplification precede cluster formation with a cBot cluster generation system. All samples were sequenced on a HiSeq 2500 Illumina Genome Sequencer as PE50. Reads were pseudoaligned with kallisto (v0.44.0) to a transcriptome index derived from all Drosophila transcript sequences of the dmel release r6.19. Subsequent analysis of transcript abundances was performed in R with sleuth (v0.30.0) including principal component analysis for dimensionality reduction, statistical and differential expression analysis based on the beta statistic derived from the wald test. Enriched gene ontology terms were identified by calculating hypergeometric p-values via the GOstats (v2.48.0) R package. Gene IDs were converted via the AnnotationDbi (v1.44.0) package and the reference provided with the org.Dm.eg.db (v3.7.0) database and geneset intersections visualized as UpSetR plot (v1.3.3). Please, see the accompanied R markdown (i.e. ‘RNAseq_sg_analysis.Rmd’) deposited on GitHub for details (https://github.com/robertkrautz/sg_analysis). The data is available under the accession number GSE138936 on the NCBI Gene Expression Omnibus.

Detection of enriched motifs and Mef2 visualization

Request a detailed protocol

Significantly, differentially up- (i.e. q < 0.001 and b > 1) or downregulated (i.e. q < 0.001 and b < −1) genesets were determined for all three sample populations by subsetting q- and b-values as determined by sleuth with the latter indicating fold-change over expression in BxGal4>w1118-control crosses. The RcisTarget R package (v1.6.0) together with the accompanied file including all dm6 motif rankings (i.e. ‘dm6-5kb-upstream-full-tx-11species.mc8nr.feather’) were used for motif enrichment analysis in the various genesets and adding annotations. Enriched motifs were manually screened for functionally related motif groups (e.g. ‘NFkB’ with dl-, Dif-, Rel-relaetd motifs) and motifs plotted according to their affiliation to these groups.

The sequence of the ‘dm6’ genome assembly was acquired as fasta-file from the Ensembl database and nucleotide sequence frequencies across the genome calculated via customized functions including all major chromosomes. In parallel, promoter sequences for all ‘dm6’ genes were collected via the biomaRt package (v2.42.1) and converted into a DNAStringSet with the Biostrings package (v2.54.0). A curated list of all ‘dm6’ transcription factor motifs as position probability matrices was obtained from the Cis-BP database and converted to position weight matrices (PWMs) with the help of the PWMEnrich R package (v4.22.0) by using the above cacluated nucleotide sequence frequencies. Based on these PWMs and all promoter sequences, background scores for all ‘dm6’ transcription factor motifs across all promoters of the ‘dm6’ genome were quantified. Sequences for the wider Drs-locus and the shared Dif-/dl-loci environments were obtained via biomaRt and the enrichment of the Mef2-motif ‘M08214_2.00_Mef2’ compared to all background scores was calculated via PWMEnrich. Mef2 motif enrichment scores were plotted as a function of the location of the detected motif in the screened loci (i.e. Dif/dl; Drs) with the help of the novel, devised function geom_bar_rastr(), which servers as an extension to the ggrastr R package. For details please follow the entire computational workflow as provided in ‘RNAseq_motif_enrichment.Rmd’ deposited on GitHub (https://github.com/robertkrautz/sg_analysis).

Immunohistochemistry

Request a detailed protocol

PPs of staged larvae were inverted in PBS, unnecessary organs removed and samples fixed in 4% paraformaldehyde for 20 min. Subsequently, samples were washed three times for each 10 min in PBS or PBST (1% TritonX-100). Blocking was performed with 0.1% BSA in PBS (H2) or 5% BSA in PBST (anti-CC3, anti-pJNK and anti-dl). Samples were then incubated in primary antibodies dissolved in blocking buffer for 12 hr at 4°C or 1 hr at room temperature (RT). Primary antibodies comprised rabbit anti-cleaved caspase 3 (1:200, Cell Signalling Technology, 9661), mouse anti-pJNK (1:250, Cell Signalling Technology, 9255), mouse anti-dl (1:50, DSHB, 7A4), and mouse anti-Hemese (1:5; gift from István Andó). After washing as prior to blocking, secondary antibodies were applied together with DAPI (1:500; Sigma-Aldrich; D9542) and when necessary Alexa Fluor 546 Phalloidin (1:500, ThermoFisher Sceintific, A22283) in blocking buffer for 1 hr at RT. As secondary antibodies goat anti-Mouse-IgG-Alexa546 (H+L; 1:500; ThermoFisher Scientific; A-11030) and goat anti-Rabbit-IgG-Alexa568 (H+L; 1:500; ThermoFisher Scientific; A-11011) were employed. Final washing as prior to blocking preceded dissection of the samples in PBS and separation of SGs. Tissues were mounted in Fluoromount-G (SouthernBiotech) and analyzed with a Zeiss LSM780 confocal microscope. Images were extracted with Zen software (Blue edition) for further processing either in Adobe Photoshop CS5 Extended (v12.0.4 × 64) or Inkscape (v0.92).

ROS-staining

Request a detailed protocol

H2DCFDA (D399, Thermo Fisher Scientific) was reconstituted in DMSO just prior to the experiment. This stock solution was further diluted to 10 µM in PBS. Freshly dissected glands were incubated in the 10 µM- working solution for 30 min in the dark. Three 5-min wash steps in PBS ensured removal of excess H2DCFDA-dye. Glands were mounted with PBS and confocal pictures obtained with the Zeiss LSM780 microscope system.

Hemocyte- / pJNK- / CC3- quantification and size measurement

Request a detailed protocol

Pictures of stained glands were taken with a Zeiss Axioplan two microscope equipped with an ACHROPLAN ×4 lens and a Hamamatsu ORCA-ER camera (C4742-95). Images were extracted with AxioVision software (v40V 4.8.2.0) and analyzed with ImageJ. Cumulated area of hemocyte attachment, pJNK or CC3 fluorescence per SG or per SG part was filtered in the Red-channel and gland size determined by outlining glands, PPs or DPs with the 'Polygon selection'-tool. The distribution of hemocytes in Drosophila can be approximated by a natural logarithm, which required transformation of hemocyte attachment- and SG-areas before calculating ratios (Sorrentino, 2010). Normality across all samples of a particular genotype and where necessary separated by gland part (i.e. proximal or distal) or time (i.e. 96 hr and 120 hr AED) was evaluated with the Shapiro-Wilk-test and by bootstrapping via the fitdistrplus R package (v1.0.11). Significant differences between experimental groups were determined for pairwise comparisons via the Student’s t-test after validating un-/equal variance via the Bartlett’s test. Data and statistical analysis was performed in R. No power or sample size calculations were performed. All experiments were performed at least twice, while preparing and acquiring meaurements for all specimens and replicates of the respective experimental set-up. The amount of replicates was determined as a trade-off between reproducibility (maximum number) and feasibility of the experimental procedure to ensure same treatment and timing for all specimens. No measurements were categorized as outliers or excluded. Effect sizes were calculated by normalizing the effect and the reference samples by the mean of the reference samples, which anchors the reference mean at ‘1’ and reports all deviations from it in procent.

Nuclei volume

Request a detailed protocol

Z-stacks of entire SGs were captured with a Zeiss LSM780 confocal microscope for DAPI signal. Obtained stacks were further processed in ImageJ via the '3D Objects Counter'-plugin (v2.0.1.). Proximal and distal compartments were defined with the 'Polygon selection'-tool. Transfer of the region of interest to all z-stack slices, signal thresholding, object identification and volume determination were integrated in a macro workflow. Data were plotted as average nucleus volume for all nuclei of individual glands or gland parts. Representative sections of individual z-stack slices showing the transition between proximal and distal compartments were cropped and added for illustration. Statistics were performed similar to the analysis of Hemocyte-, pJNK- and CC3- quantifications.

Drs-dl-correlation

Request a detailed protocol

BxGal4;;DrsGFP>RasV12-glands were stained for dl (mouse anti-dl-Ab; 1:50, DSHB, 7A4) and genomic DNA (DAPI), whereafter z-stacks were obtained on a Zeiss LSM780 confocal microscope. Proximal and distal gland compartments were delineated as regions of interest (ROI) in ImageJ with the ‘Polygon selection’-tool and nuclei therein defined via their DAPI-signal and the ‘3D Objects counter’-ImageJ plugin. Transfer of the 3D-nulei outlines across the dl- and Drs-channels via the ImageJ ‘RoiManager 3D’ (v3.96) allowed quantifying an accumulated intensity of the corresponding signals per nucleus. Data analysis was conducted in R including calculation of linear regression and Pearson coefficients for correlations between nuclei-wide Drs- and dl-signal for individual gland compartments and visualization of all nuclei data points after log10-normalization for all five screened samples.

Driver expression strength

Request a detailed protocol

DrsGFP-reporter or Phalloidin-stained, GFP-glands were captured with the ‘Zeiss Axioplan two microscope / ACHROPLAN 4x lens / Hamamatsu ORCA-ER camera (C4742-95) / AxioVision software (v40V 4.8.2.0)’-system. ROIs were added manually to the imaged glands to outline their position in ImageJ with the ‘Polygon selections’-tool. Feret diameter and angle were measured to delineate the longitudinal axis of each gland. For each position along the longitudinal axis, the corresponding rectangular line region intersecting with the glands ROI is determined and the encapsulated average fluorescence intensity measured. This workflow is implemented as ImageJ-macro (i.e. ‘scanner.ijm’) for screening of entire picture batches and available on GitHub (https://github.com/robertkrautz/sg_analysis). The resulting data was further analyzed in R to bin and align measurements along the longitudinal axis. Measurements were further averaged and normalized per bin across all screened glands. Given the statistically limited amount of observations (i.e. 200–350 bins), a loess function was used for smoothing the mean during visualization as implemented in geom_smooth() (i.e. span = 0.05).

Probe synthesis and in situ hybdridization

Request a detailed protocol

cDNAs for the Drs (LP03851) and CecA1 (IP21250) loci were obtained from DGRC and plasmids transformed into DH5α according to the supplier’s instructions. Extracted plasmids were linearized (EcoRI 20 U/µl, 10x EcoRI buffer, BSA 20 mg/ml in ddH2O), purified with phenol:chloroform:isoamyl alcohol (24:24:1/125:24:1) and precipitated (3M sodium acetate, ice-cold 100% Ethanol) for 30 min. In vitro transcription (5x transcription buffer, 1 µg template DNA, 10x Digoxigenin-11-UTP rNTP; Roche; 11277073910, 20U Ribolock; Fermentes; EO0381, T7 polymerase; ThermoFisher Scientific; EP0111) was performed for 3 hr at 37°C. The remaining DNA template was digested with DNase I for 15 min at 37°C and digestion terminated with ammonium acetate (DEPC-H2O:7.8M NH4Ac:100% Ethanol – 1:0.5:3). The transcribed RNA was precipitated, purified and diluted in hybridization buffer (6.57 mg/ml Torula RNA; Sigma-Aldrich; R6625, 65% deionized formamide, 6.5x Saline sodium citrate [SSC], 65 µg/ml heparin, 0.1% Tween-20).

SGs were dissected in PBS, fixed in paraformaldehyde (4% PFA in PBSTw: 0.1% Tween-20 in PBS) for 40 min at room temperature (RT), washed four times in PBSTw for 5 min and transferred to Methanol. Permeabilization was performed for 1 hr in Ethanol, followed by washing once in Methanol for 5 min and twice in PBSTw for each 5 min. The samples were treated with Proteinase K (30 µg/ml; ThermoFisher Scientific; EO0491) for 6 min, washed twice with glycine (2 mg/ml) for 1 min, fixed for 20 min in 4% PFA and washed four times in PBSTw for each 5 min.

Prior to hybridization, glands were pre-treated with hybridization buffer (HYB: 5 mg/ml Torula RNA, 50% deionized formamide, 5x SSC, 50 µg/ml heparin, 0.1% Tween-20) for 1 hr before incubation with either CecA1- or Drs-probe overnight, both at 60°C (in HYB with 3% Dextran). To avoid unspecific binding, samples were stringently washed at 60°C once with HYB-wash solution (50% formamide, 2x SSC, and 0.1% Tween-20) for 30 min, twice with 2x SSCTw, and trice with 0.2 x SSCTw for 20 min. At RT, samples were washed once with PBSTw for 5 min, blocked with sheep serum (8% in PBSTw; Sigma-Aldrich; S2263) for 1 hr and incubated with alkaline phosphatase coupled to anti-digoxigenin-antibody (1:4000 in PBSTw with 8% sheep serum; Roche; 1093274) at 4°C. Washing proceeded four times with PBSTw for each 15 min, twice with TNTw buffer (Tris-NaCl-Tween – 0.1M:0.1M:0.1%; pH 8) for each 15 min and twice with alkaline phosphatase buffer (100 mM Tris-HCL pH 9.5, 100 mM NaCl, 50 mM MgCL2, 0.1% Tween-20) for each 5 min. Staining was perfomed with BCIP (5-bromo-4-chloro-3-indolyl-phosphate 4-toluidine salt: 0.175 mg/ml; Roche; 11383221001), NBT (4-nitro-blue-tetrazolium chloride: 0.3375 mg/ml; Roche; 11383213001) and levamisole (0.001 M) in alkaline phosphatase buffer. Stained samples were washed with alkaline phosphatase buffer for 5 min, TNTw for 15 min and PBSTw for 15 min. Washed samples were mounted in Fluoromount G (SouthernBiotech) and capturesd with Leica Firecam (v3.4.1) using a Leica DFC300x FX digital color camera coupled to a Leica MZ16 microscope. All experiments were performed at least twice including all specimens dissected from the entire set of replicates. The amount of replicates was determined as a trade-off between reproducibility and the amount of samples that could be processed with specific equipment of limited supply (i.e. sieves for sample transfer between buffers).

Data availability

All sequencing data has been deposited at NCBI GEO under the record GSE138936.

The following data sets were generated
    1. Krautz RK
    2. Khalili DK
    3. Theopold T
    (2019) NCBI Gene Expression Omnibus
    ID GSE138936. Transcriptomes of whole hypertrophic Drosophila salivary glands and separated proximal gland parts.

References

    1. Agnès F
    2. Suzanne M
    3. Noselli S
    (1999)
    The Drosophila JNK pathway controls the morphogenesis of imaginal discs during metamorphosis
    Development 126:5453–5462.
    1. Asri RM
    2. Salim E
    3. Nainu F
    4. Hori A
    5. Kuraishi T
    (2019)
    Sterile induction of innate immunity in Drosophila melanogaster
    Frontiers in Bioscience 24:1390–1400.
    1. Britton JS
    2. Edgar BA
    (1998)
    Environmental control of the cell cycle in Drosophila nutrition activates mitotic and endoreplicative cells by distinct mechanisms
    Development 125:2149–2158.
    1. Hay BA
    2. Wolff T
    3. Rubin GM
    (1994)
    Expression of baculovirus P35 prevents Cell-Death in Drosophila
    Development 120:2121–2129.
  1. Book
    1. Hoffmann AA
    2. Parsons PA
    (1991)
    Evolutionary Genetics and Environmental Stress
    Oxford, New York: Oxford University Press.
  2. Software
    1. Pagès H
    2. Carlson M
    3. Falcon S
    4. Li N
    (2020) AnnotationDbi
    AnnotationDbi: Manipulation of SQLite-Based Annotations in Bioconductor.
  3. Software
    1. Pàges H
    2. Aboyoun P
    3. Gentleman R
    4. DebRoy S
    (2020) Biostrings, version 2.54.0
    Biostrings: Efficient Manipulation of Biological Strings.
    1. Smith AV
    2. Orr-Weaver TL
    (1991)
    The regulation of the cell cycle during Drosophila embryogenesis: the transition to polyteny
    Development 112:997–1008.
    1. Zeitlinger J
    2. Bohmann D
    (1999)
    Thorax closure in Drosophila involvement of fos and the JNK pathway
    Development 126:3947–3956.

Decision letter

  1. Bruno Lemaitre
    Reviewing Editor; École Polytechnique Fédérale de Lausanne, Switzerland
  2. Utpal Banerjee
    Senior Editor; University of California, Los Angeles, United States
  3. Mark Austin Hanson
    Reviewer; École Polytechnique Fédérale de Lausanne, Switzerland

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Krautz et al. demonstrate that a tissue-autonomous innate immune response regulates hypertrophic growth using a Drosophila salivary gland model. A key finding is that the antimicrobial peptide Drosomycin inhibits the Jun-Kinase apoptotic feedback loop, permitting continued hypertrophic growth.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Tissue-autonomous immune response regulates stress signalling during hypertrophy" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that the decision was to reject the manuscript while allowing re-submission. In this way, you are more than two months to submit a revised version that address all the point of the reviews. Importantly, you submission will be considered as a new submission but is likely to reviewed by the same reviewers. At this stage, the reviewers that important control are lacking to ascertain the solidity of the claim. If you feel that you will not be able to address the reviewer's comments, you can decide to transfer the manuscript to another journal.

Reviewer #1:

This work proposes that tissue-autonomous Drosomycin (Drs) negatively regulates JNK signalling in a hypertrophic salivary gland (SG) model. The flow of experiment ideas is thorough, but lacking critical control treatments. There appears to be a spatial bias in the BxGal4 expression model used. Additionally, there is no demonstration that this is Drs-specific, as the authors never test for the effect of Gal4 dilution by involving additional UAS constructs such as other AMPs or a control like UAS-GFP. As the paper relies entirely on a BxGal4>UAS model, alternate Gal4 and UAS control treatments are essential to avoid spatial bias and the effect of multiple UAS diluting RasV12 production and consequent hypertrophy.

List of substantive concerns:

1) The authors use a BxGal4 model to drive SG expression throughout this manuscript. The assumption seems to be that BxGal4 is a pan-SG driver, and that differences in the proximal and distal parts of the SG arise due to differences in Drs expression in these regions. However it is revealed in Figure 2B that BxGal4 does not efficiently drive dorsal-RNAi in the proximal part (PP) of the SG, as this tissue retains abundant dorsal expression. In Figure 4A there is a weakening of Drs signal in situ in the BxGal4>Drs at the PP, indicating BxGal4 is not efficiently driving Drs in the PP. Finally, the PCA analysis of RNAseq data shows totally independent clustering of the PP transcriptome from whole SGs; no BxGal4>w1118 PP control is given for comparison.

Thus, throughout the manuscript, the PP is very likely a region with significantly reduced RasV12 expression, readily explaining most differences between the sick and apoptotic DP and the (relatively) healthy PP. For instance, this would readily explain differences in hemocyte attachment between the DP and PP. This may also imply a Drs response is seen in the PP solely because this relatively healthy tissue remains immune competent for longer than the DP.

Finally, Drs expression in the SG increases between the L3 and prepupa stages of development (FlyAtlas). It is unclear how BxGal4>RasV12 might affect SG developmental patterning, and indeed the model relies on rapid overgrowth of the tissue. Moreover there is an expected Drs response in immune-responsive tissues to sterile injury or stress. It is likely that SG remodelling during RasV12-induced stress leads to corresponding programmed cell death and similarly induction of NF-κB. This interpretation is congruous with the findings of Araki et al. (2018) and Parvy et al. (2019) that suggest NF-κB effectors cooperate in clearing tumorous tissue and responds to cell death signals.

2) Throughout the manuscript, the authors combine additional UAS constructs without controlling for the dilution of their Gal4 drive. This affects almost every figure in the latter half of the manuscript, and the effect of this dilution is apparent in Figure S5C, where Drs levels in BxGal4>Drs alone are 1-2 fold higher than BxGal4>Drs,RasV12 double drive. If RasV12 is diluted in combined UAS lines, this would promote a delayed phenotype, resulting in an apparent rescue effect if you simply compare equal time points. But it would be expected that continued monitoring to 144h would reveal the same apoptotic effects as the lone BxGal4>RasV12 tissue at 120h, similar to how BxGal4>RasV12 at 96h displays less apoptosis than 120h.

There is one argument that can be made in favour of the dilution not mattering: Drs-RNAi has a mild but opposite effect at 96h. However it should be noted that RNAi vs. protein overexpression leads to significantly different metabolic burdens on the cell, and thus RNAi is not a good control for this dilution effect. The authors should use a control overexpression construct such as UAS-GFP or UAS-RFP throughout.

3) There is no logic presented within the paper that this effect should be Drs-specific. Many other AMPs and immune effectors were similarly upregulated in their transcriptome (e.g. AttD, Def). If AttD and Def are similarly induced in this model, it is assumed their expression is similarly regulated, and that overexpression of AttD or Def could similarly rescue RasV12 SGs, and RNAi of AttD or Def could similarly exacerbate RasV12 SGs. I would recommend that, alongside a UAS-GFP control, the authors also test at least some subset of NF-κB effectors alongside their Drs model to elucidate if this is specific to Drs or a general effect. Given findings in other papers on AMP-tumor interactions where overexpressing Drs, Dpt, or Def all had similar effect (Araki et al., 2019), it seems likely that many AMPs should result in the same phenotype, implying there is no molecular pathway where specifically Drs inhibits JNK.

Reviewer #2:

Krautz et al. presents an interesting finding on regulation of stress signaling during hypertrophy via tissue-autonomous immune signaling. Data shows that Drosomycin, downstream of the NFkB factor Dorsal, interferes with overgrowth of the salivary gland caused by RasV12 expression. The authors investigate the underlying mechanism of this finding using different approaches, presenting copious data which are generally supportive their model.

However, in some cases the findings lack direct demonstration of key aspects of their model. The conclusions, highlighted in the model shown in Figure 8, claim that Drs inhibits JNK signaling which is shown with some key experiments. Yet, this model also claims that JNK signaling drives tissue disintegration effects of RasV12 through induction of MMP2, yet MMP2 is not upregulated until a very late time point and only to a modest degree. Moreover, they don't show that MMP2 or Hid (another JNK target) are required for these events. Also, in some cases the data is not as robust as would be expected in the field; activation of JNK is shown with compelling reporter data and pBsk staining, but the gene expression for JNK targets is underwhelming, for example Puc. Perhaps the issue is more with the model figure over-extending beyond the actual data presented.

In addition, the authors claims have relied mostly on qualitative nature of the data, whereas quantification could have strengthened their claim. For example, authors conclude that endoreplication leads to increase in nuclear volume but have not shown any quantitative increase in nuclear DNA content. The DAPI data shows most of the nuclei as disintegrated which will amount to higher nuclear volume but is less informative about DNA content. Further, authors have not stated whether the disintegrated nuclei were excluded from the analysis or not. Similarly, RasV12 stimulated MMPs-dependent tissue disintegration and Hid-dependent nuclear disintegration are not directly demonstrated.

Apart from the above-stated experimental concerns, issue with data presentation also significantly dampens the enthusiasm for this study. The data-presentation is incredibly cumbersome; throughout, the text does not present clearly the rationale of their experimental design, some key figure panels are not even mentioned in the text, other key data is buried in the supplemental figures, while other data is replicated in separate figure panels without clear reasoning, and importantly most of the legends/labeling on the graphs are so small, they are illegible. This last issue makes this work nearly impossible to critically evaluate. Also, I would encourage the authors to use gene names and symbols on their figures that are understood by a wide readership. For example, Bsk might be alternatively JNK in S5.

Reviewer #3:

This article is interesting because it shows that some immune genes, mostly antimicrobial peptide (AMP) genes, are induced in a highly artificial system in which the overgrowth of salivary glands is induced by ectopic expression of the Ras[V12] gain-of-function allele. The expression of immune genes would rely solely on Dorsal and not the Toll pathway (see below). The most important point is that the overexpression of Drosomycin is sufficient to reduce the induction of JNK-pathway regulated genes that occurs in the distal part of salivary glands, including pro-apoptotic genes, thus introducing the concept of a direct or indirect intracellular signaling function for an antimicrobial peptide.

In the longer term, it will be interesting to determine whether Dorsal plays a role in wild-type larval salivary glands and why it is differentially repressed in the DP.

Substantive concerns

1) The authors use RNAi lines to determine whether the Toll pathway is required for the induction of immune genes, which are not even validated. Since it is close to impossible to demonstrate the generation of a null phenotype by RNAi, the authors should imperatively use at least one null mutant of the Toll pathway, e.g., MyD88 and possibly spz.

2) The authors should look at and quantify Dorsal nuclear localization in the salivary glands and not solely at its expression. Upon looking at Figure 2B, one gets the impression it is nuclear only in the very proximal region of the organ.

3) Multiple AMPs are induced in the PP and it is surprising that the one AMP they have overexpressed is having such a biological function. A control is missing, ideally a Drosomycin mutant in which disulfide bridges cannot form; at least a nonrelevant gene should be used as a control to exclude nonspecific effects linked to overexpression in the Ras[V12] context. Also, the Drosomycin RNAi line does not abolish the expression of its target gene and does not have such a "drastic" effect: several genes are only mildly more strongly expressed. It would be interesting to use a Drosomycin null mutant. Perhaps less important for this work but of interest, it would have been informative to test mutants affecting AMP families that have recently been published.

4) It is not clear whether Drosomycin is secreted in the lumen or basally in the hemolymph. To really conclude that this is a tissue-autonomous response, inasmuch the authors do not state whether Drosomycin is expressed in other tissues, it would be useful to overexpress Drosomycin in the fat body and test its effects on the DP of the salivary glands.

https://doi.org/10.7554/eLife.64919.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

This work proposes that tissue-autonomous Drosomycin (Drs) negatively regulates JNK signalling in a hypertrophic salivary gland (SG) model. The flow of experiment ideas is thorough, but lacking critical control treatments. There appears to be a spatial bias in the BxGal4 expression model used. Additionally, there is no demonstration that this is Drs-specific, as the authors never test for the effect of Gal4 dilution by involving additional UAS constructs such as other AMPs or a control like UAS-GFP. As the paper relies entirely on a BxGal4>UAS model, alternate Gal4 and UAS control treatments are essential to avoid spatial bias and the effect of multiple UAS diluting RasV12 production and consequent hypertrophy.

List of substantive concerns:

1) The authors use a BxGal4 model to drive SG expression throughout this manuscript. The assumption seems to be that BxGal4 is a pan-SG driver, and that differences in the proximal and distal parts of the SG arise due to differences in Drs expression in these regions. However it is revealed in Figure 2B that BxGal4 does not efficiently drive dorsal-RNAi in the proximal part (PP) of the SG, as this tissue retains abundant dorsal expression.

In order to address this very valid aspect systematically, we devised a novel algorithm to aggregate experimentally acquired fluorescence intensities across the vertical axes for every position along the longitudinal axis of salivary glands (Figure 1—figure supplement 1A-B; Author response image 1A). This allowed to quantify Gal4-driver expression (i.e., BxGal4>UAS-mCD8::RFP), fluorescence signals from stainings (i.e., Phalloidin) or reporter lines (i.e., DrsGFP) along the salivary gland beyond a bipartite proximal-distal separation. To avoid artifacts due to differential thickness, tissue curvature and thus internal reflection we normalized BxGal4>UAS-mCD8::RFP signal intensities with Phalloidin signal from the corresponding positions and averaged for every genotype-time combination (e.g., RasV12 at 96 h) over more than 10 samples (i.e., 11-15).

Author response image 1
RasV12-glands show intrinsic anterior-posterior separation.

(A) Right: Schematic representation of deployed algorithm to measure fluorescene signals along all vertical axis perpendicular to the glands longitudinal axis. Left: mCD8::RFP;;RasV12- and mCD8::RFP -control glands stained with Phalloidin (green) to assess BxMS1096 driven, Phalloidin-normalized RFP-expression at 96 h and 120 h AED along the longitudinal gland axis (middle). Signals averaged across n=15 glands showed 86% of the signal intensity in the proximal gland part (PP) compared to the distal (DP) (right). DrsGFP-reporter intensities along the longitudinal gland axis in RasV12- and w1118-control glands at 96 h and 120 h AED (bottom). (B) Upper: Schematic representation of the correlation assay. DAPI-signal defines nuclei location and volume used to subsequently accumulate fluorescence signals from nuclear DrsGFP-reporter and dl-staining. Lower: Scatterplot of associated dl- and Drs-signals per nucleus in PP or DP across 5 glands at 96 h AED. Pearson correlation coefficient shown for all 5 samples. Lines with confidence intervals indicate linear regression for Drs- as a function of dl-signal per sample and gland compartment. (C) AttD gene expression as determined by qPCR in separated PP and DP of RasV12- and w1118-control glands at 96 h and 120 h AED (log2-transformed, fold-change over Rpl32). Lower/upper hinges of boxplots indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Student's t-tests determine statistical significance (** p<0.01, n.s. p≥0.05). (D) RasV12-, lglRNAi;RasV12- and w1118-control glands carrying the 10xStat92E-GFP reporter (green) at 96 h and 144 h AED (w1118-larvae pupate at 120 h AED). (E) DrsGFP-reporter signal in RasV12-glands at 144 h AED. (F) Expression of AMP genes in RasV12, lglRNAi;RasV12 or lglRNAi;RasV12-PP- in comparison to w1118-samples. Missing 1009 bars indicate absence of expression values in the RNAseq data. Insets: (D-E) DAPI. Scalebars: (D-E) 100 µm.

While Phalloidin-normalized BxGal4-expression increases between 96 and 120 h, no significant bias can be detected for either gland compartment (i.e., proximal or distal) in control glands (BxGal4,UAS-mCD8::RFP x w1118) apart from a minor tendency for increased BxGal4-driven expression in the proximal part at 120 h AED (Figure 1—figure supplement 1A-B; Author response image 1A). Upon co-expression of RasV12, the RFP-signal at 120 h compared to control glands increases further across the entire gland length. The proximal tendency from control glands is only reversed in RasV12-glands at 120 h with a drop in BxGal4-expression along the proximal part (Figure 1—figure supplement 1A-B; Author response image A). This shift coincides with the differential replication of genomic DNA (i.e., endoreplications) in distal vs. proximal nuclei. However, while the differing number of endoreplications would suggest a substantially differing expression, on average the BxGal4driven expression in the proximal part of RasV12-glands still reaches 86% of the average expression in the distal part (Figure 1—figure supplement 1A-B; Author response image 1A). In addition, the expression strength in the proximal part still increases upon co-expression of RasV12 compared to control glands (Bx, RFP x w1118: 1.58; Bx, RFP x RasV12: 2.30).

In parallel, quantifying DrsGFP-fluorescence across the length of the gland showed that Drs-expression and BxGal4-expression are neither correlated, nor anti-correlated over time (i.e., 96 and 120 h) and space (i.e., proximal and distal; Figure 1—figure supplement 1A-B; Author response image 1A). Thus, while RasV12-expression is qualitatively necessary for Drs-expression, already the comparably low BxGal4-driven RasV12-expression at 96 h across the gland epithelium is sufficient for Drs to become fully expressed (i.e., same level as at 120 h AED). This makes Drs-expression qualitatively, but not quantitatively dependent on BxGal4-driven RasV12-expression. In this regard, strong Drs-expression also precedes the accumulation of programmed cell death (PCD) in the gland as indicated by CC3 and nuclear disintegration (Figure 6C-F), a prerequisite for the model we propose, in which Drs actively inhibits induction of PCD.

In contrast, the intrinsic expression pattern of Dorsal, that appears independent of RasV12-expression correlates not only qualitatively, but crucially quantitatively with the presence of Drs (Figure 2B-C). Under physiological conditions and thus independent of RasV12, Dorsal expression is higher in the proximal part at 96 and 120 h AED than in the distal and eventually lost almost entirely in the distal part at 120 h (Figure 2B-C). This expression pattern persists upon RasV12-coexpression. However, upon dl-knock-down without RasV12-expression, Dorsal is indeed greatly reduced in all gland parts across the two developmental time points. We included separating lines in Figure 2B. to emphasize this aspect further. At 96 h, when Bx driven expression is similar in proximal and distal, residual dl-expression in the proximal part upon dlRNAi-expression indicates an insufficiency of the knock-down construct to abolish dl-expression (Figure 2B-C). This is in line with qPCR-data we acquired for Drs-expression in separated proximal and distal gland parts upon co-expression of the dlRNAi-line with RasV12 (Figure 2—figure supplement 3A; Author response image 3A). However, in spite of a reduced Bx-driven expression in the proximal part of RasV12-glands at 120 h and an insufficient dlRNAi-construct, dl-knock-down not only decreases Dorsal expression, but crucially also Drs-expression in both gland parts, as well as both time points (Figure 2B) emphasizing the dependency of Drs on dl rather than changes in Bxdriver expression further.

Author response image 2
Drs asserts specific function on JNK-signaling not mimicked by other detected AMPs.

(A) SG size as measured by outlining images of the 120 h-old, experimental and control glands with the indicated genotypes. (B) Quantified CC3-staining in RasV12-glands with coexpressed Drs or mCD8::RFP as a UAS-dilution control, normalized for gland size. (C). Quantifications of activated jnk assessed by antiphosphoJNK-antibody staining and normalized for gland size in RasV12-glands coexpressing Drs or mCD8::RFP. (D) Prototypical examples of Dronc- and JNK-activation in RasV12-glands with coexpressed Drs or mCD8::RFP. (E) Expression values for Drs and JNK target genes as determined by qPCR for RasV12-glands coexpressing either Drs or mCD8::RFP as a UAS-dilution control at 96 h and 120 h AED (log2-transformed; normalized to Rpl32 expression). (F) CecA1-mRNA detected by in-situ hybridization in RasV12- and w1118-control glands at 96 h and 120 h AED upon coexpression of either a dlRNAi-knock-down or a CecA1-overexpression construct. (G) Effect size of Drs-knock-down on Drs-expression in RasV12-glands. Mean RasV12-expression was set to “1” separately at 96 h and 120 h AED. (H) ISH with Drs-probe in RasV12-glands with or without coexpressed DrsRNAi-construct. Scalebars: (D/F/H) 100 µm. Insets: (D) DAPI. Boxplots: (A-C,E,G) lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student's t-tests (*** p<0.001, ** p<0.01, * p<0.05, n.s. p≥0.05).

Author response image 3
Apart from dorsal Drs is putatively regulated by Mef2.

(A) Effect sizes of dlRNAi-knock-down on Drs-expression in PP- and DP-compartments of and compared to RasV12-glands at 96 and 120 h AED. Mean of Drs-expression in RasV12-glands was set to “1”. Boxplots: lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. (B) Transcription factor binding site analysis in up- and down-regulated genesets of RasV12-, lglRNAi;RasV12- and lglRNAi;RasV12-PP- samples. Normalized enrichment scores represent statistical overrepresentation (NES>2) of transcription factor binding motifs in the 6 screened genesets. Cis-BP-derived motifs were sorted into the displayed transcription factor groups. NFκB- and jra- as well as Myb-, eg- and br-motif groups were included as positive and negative examples. (C) DrsGFP-reporter based assay including screened RNAi-lines targeting transcription factors identified in binding site analysis. 3 different phenotypes were scored (right) per genotype and their mean and standard deviation plotted. Dunn’s post-hoc test conducted on “None”phenotype to validate significant differences (*** p<0.001, n.s. p≥0.05). To avoid false positives, melanisation was monitored in pupae (insets) to confirm RasV12-expression (Hauling et al., 2014). (D) Wider Drs-locus on “3L”- and “Dif/dl”-locus on “2L”-chromosome shown including outlines for gene loci encoded on plus and minus strands. Log-transformed motif scores are represented as a function of the motif’s location along the respective gene loci. (E) Drs in-situ hybridization in 120 h-old RasV12- and control-glands hetero- or homozygous mutant for Myd88 (Myd88KG03447). Scalebar corresponds to 100 µm.

To confirm a correlation between the intrinsic, tissue-autonomous and largely RasV12-independent dl-expression on the one and Drs-expression patterns in these hypertrophic glands on the other side, we quantified nuclear dl- parallel to DrsGFP-signal in 120 h-old RasV12-glands. In fact, Drs and dl show a strong linear correlation across the entire scale (i.e., log10-transformed) in both the proximal as well as the distal gland part (Figure 2—figure supplement 2B; Author response image 1B).

In Figure 4A there is a weakening of Drs signal in situ in the BxGal4>Drs at the PP, indicating BxGal4 is not efficiently driving Drs in the PP.

While ISH can be performed in a controlled, quantitative manner by using fluorescence probes and confocal microscopy, the ISH results presented throughout the manuscript were done by chromogenic staining and results were captured by brightfield microscopy. Thus, all results can only be seen in a comparative, semi-quantitative manner with the aim to characterize qualitative expression patterns. In addition, increased curvature and lumen width, as well as reduced tissue thickness and cell size of the proximal gland part complicate using ISH-quantifications as proxy for expression strengths further.

While we therefore refer to the presented quantifications along the longitudinal axis (Figure 1—figure supplement 1A-B; Author response image 1A), we certainly acknowledge the point raised by the reviewer of a reduced expression in the proximal part of RasV12-glands at 120 h, but present evidence that this reduced expression is not causal for the differences seen between proximal and distal gland compartments such as Drs-expression. To credit the reviewer’s comments and for full transparency, Figure 1—figure supplement 1A was included in the manuscript and its consequences for the model discussed in the manuscript’s main text.

Finally, the PCA analysis of RNAseq data shows totally independent clustering of the PP transcriptome from whole SGs; no BxGal4>w1118 PP control is given for comparison.

PC1 explains 94.26% of the variability in the data and all samples derived from overexpressing RasV12 (i.e., RasV12, lglRNAi;RasV12, lglRNAi;RasV12-PP) are falling almost completely in line with one another in PC1 (Figure 3A). Thus, while all 9 RasV12-samples separate in PC1 from the control gland samples with the majority of variability in the expression data being explained by PC1, the lglRNAi;RasV12-PP samples separate mainly in PC2, which only explains 3.05% of the intrinsic variability of the data.

In addition, principal components analysis subsumes qualitative differences (i.e., genesets expressed) into quantitative differences (i.e., gene expression intensities), indicating that the separation in PC2 remains largely due to qualitative differences in significantly differentially expressed genes that do not occur in the DP. This is clearly underlined in the barplot (UpSetR-plot) in Figure 2A, which shows a set of 309 genes solely, significantly differentially expressed in the PP and further exemplified in Figure 2C. for “immune response” genes. Many of these genes show a strong qualitative difference (i.e., unilateral upregulation in the PP compared to whole-tissue samples). The whole-tissue samples include not only the distal, but also the proximal part, which dampens the detectable difference in gene expression strength between both parts. Thus, we performed additional qPCRs for AttD as the highest upregulated gene in the PP on dissected, separated gland parts (DP vs. PP), which revealed a complete absence of AttD-expression in the distal part at 96 and 120 h (Figure 3—figure supplement 1D; Author response image 1C).

In summary, the qualitative differences between the PP and the DP as reflected in the expression of different sets of genes (Figure 2A-C) gives rise to the separation of the PPsamples along PC2 which in turn however only explains a minor fraction of the data’s inherent variability.

This spatial separation is further corroborated by the expression patterns shown for amongst others dl (Figure 2B), DrsGFP (Figure 1E; Figure 1—figure supplement 1E; Figure 5A) and TREGFP1b (Figure 3F; Figure 3—figure supplement 1F; Figure 5F; Figure 5—figure supplement 2E). However, a reporter indicating JAK-STAT-activation (related GOterms were found to be enriched in the RNAseq-datasets; data not shown), the 10xSTAT92EGFP-reporter, highlights this separation best (Figure 3—figure supplement 1F; Author response image 1D). Upon RasV12-expression, the construct reports JAK-STAT-activation solely in the PP between 96 h to 144 h (Figure 3—figure supplement 1F; Author response image 1D). The presented evidence for an intrinsic spatial separation into a proximal and a distal gland compartment is in the following also central for the proposed model, in which a dl-dependent reduction solely in the DP reverts inhibition of JNK-signaling.

Thus, throughout the manuscript, the PP is very likely a region with significantly reduced RasV12 expression, readily explaining most differences between the sick and apoptotic DP and the (relatively) healthy PP. For instance, this would readily explain differences in hemocyte attachment between the DP and PP. This may also imply a Drs response is seen in the PP solely because this relatively healthy tissue remains immune competent for longer than the DP.

If PCD and tissue degradation is the cause for the reduced Drs expression or expression in general, the assumption would be a maintained Drs-expression upon inhibition of PCD in the distal part of RasV12-glands at 120 h. Upon inhibition of the JNK-pathway by co-expression of a jnkDN-construct (bskDN), JNK-activation (Figure 5—figure supplement 1B) as well as hemocyte attachment (Figure 4D) are reversed almost to the level of control glands in line with the block of PCD. However, in spite of inhibited PCD, Drs-expression does not recover in the DP, ultimately rendering Drs-expression independent of the quantitative expression levels from the BxGal4-driver (Figure 4A; Figure 5—figure supplement 1A). Drs is only dependent on the driver in as much as it qualitatively expresses RasV12 in this model. In parallel, jnkDN;;RasV12-glands undergo the same increase in endoreplications, thus mimicking the changes in Bx-driven expression present in RasV12-glands (Figure 6C).

Moreover, Figure 5—figure supplement 2C. shows an upregulation of many JNK target genes upon co-expression of an RNAi-knock-down construct against Drs at 96 h, when the DP of RasV12glands still expresses Drs and is thus susceptible to a Drs-knock-down. Despite the use of two UAS-constructs, an increased expression of these JNK target genes indicates that a Drs-reduction has a specific, active effect on the expression of the JNK target genes.

In summary, the state of a gland part (i.e., healthy or not) is not determined by the strength of Bx-driven expression, but instead by the presence or absence of Drs, whose expression is in turn regulated by the distribution of dl.

Finally, Drs expression in the SG increases between the L3 and prepupa stages of development (FlyAtlas). It is unclear how BxGal4>RasV12 might affect SG developmental patterning, and indeed the model relies on rapid overgrowth of the tissue. Moreover there is an expected Drs response in immune-responsive tissues to sterile injury or stress. It is likely that SG remodelling during RasV12-induced stress leads to corresponding programmed cell death and similarly induction of NF-κB. This interpretation is congruous with the findings of Araki et al. (2018) and Parvy et al. (2019) that suggest NF-κB effectors cooperate in clearing tumorous tissue and responds to cell death signals.

The model presented in this manuscript and the reviewers alternative fundamentally differ in their interpretation of the causality between Drs-activation and the induction of PCD. (1) Either Drs reduced cell death by inhibiting JNK-signaling or (2) PCD leads to the erasure of Drs-expressing cells and thus eventually to Drs-expression itself. While we acknowledge the 14%reduction in the PP compared to the DP of BxGal4>RasV12-glands at 120 h, the alternative proposition can however not be sufficiently aligned to all the above presented evidence at once: (1) no correlation between strength of Bx-driven and Drs-expression (Figure 1—figure supplement 1A-B; Author response image 1A), (2) Drs-expression precedes occurrence of PCD (Figure 1—figure supplement 1A-B; Author response image1A), (3) Drs-expression is determined by dl (i.e., dl-reduction leads to Drs-reduction; Figure 2B; Figure 2—figure supplement 2B; Figure 2—figure supplement 3A; Author response image 1B; Author response image 3A), (4) tissue-autonomous proximal-distal-separation of the gland epithelium (Figure 1E; Figure 1—figure supplement 1G,I; Figure 2B; Figure 3F; Figure 3—figure supplement 1F; Figure 5F; Author response image 1D) and (5) no recovery of Drs-expression by blocking PCD via JNK-inhibition (Figure 4A; Figure 5—figure supplement 1A).

Most importantly, Drs is not a mere bystander parallel to the tissue’s disintegration, but has functional relevance in preventing it as is shown by Drs-overexpression in RasV12-glands. Not only are markers for PCD reduced (Figure 6D. – nuclear disintegration; Figure 6E-F. – CC3-activation; Figure 5E-F. – JNK-activation) and thus tissue-viability maintained, but the glands continue to overgrow (Figure 4C; Figure 4—figure supplement 1A; Author response image 2A). In this regard, the here presented model of a tissue-autonomous immune response aiding hypertrophic growth is also opposed to the models published by Araki et al. and Parvy et al. (Araki M et al. Dis Model Mech. 2019 Jun18;12(6):dmm037721. doi:10.1242/dmm.037721; Parvy JP et al. eLife. 2019 Jul 30;8:e45061. doi:10.7554/eLife.45061). Clearance of the distal parts of RasV12-glands at 120 h not only occurs in the absence of Drs, but in its presence such as by Drs-overexpression, no gland tissue needs to be cleared, since Drs prevents its disintegration in the first place.

Kenmoku et al. indeed showed the induction of Drosomycin upon sterile injury independent on the position of the actual injury in the larval epidermis (i.e., pinching; Kenmoku H et al. Dis Model Mech. 2017 Mar1;10(3):271-281. doi:10.1242/dmm.027102). However, this type of sterile injury leads exclusively to an upregulation of Drs in the fat body as part of a systemic immune response. While we use the same transgenic reporter construct as Kenmoku H et al. we exclusively identify expression of Drs in the larval salivary glands coinciding with RasV12-overexpression, indicating the absence of artefacts due to mechanically induced stress (Ferrandon D et al. EMBO J. 1998 Aug10;17(5):1217-27. doi:10.1093/emboj/17.5.1217; Jung AC et al. Biotechniques. 2001 Mar;30(3):594-8, 600-1. doi: 10.2144/01303rr04). In parallel, neither are larval salivary glands established immune responsive epithelia, nor did other Drosophila overgrowth models show expression of immune effectors in a tissue-autonomous manner. Apart from the qualitative differences in tissue expression, we also identify strong quantitative differences in the Drosomycin expression levels with respect to the reported levels upon sterile injury. While Kenmoku report a fold change of maximal 10 in Drs expression upon injury over control levels (i.e., in fat bodies), Drs expression in RasV12-glands over control glands is two orders of magnitude higher (i.e., log2(FC) > = 10, FC > = 1024).

Apart from our findings presented here, Pierce SB et al., 2004 showed a similar proximal, distal separation of dMyc and dMnt expression along the gland epithelium in larval salivary glands (Pierce SB et al. Development. 2004 May;131(10):2317-27. doi:10.1242/dev.01108). In addition, our findings on the differential expression and abundance of dl in the absence of RasV12 (Figure 2B) and the differing nuclei volume in PP and DP consistent with previous reports indicate that the SG pattern identified in RasV12 glands is not altered, but that the RasV12-related phenotypes are indeed based on an intrinsic separation between PP and DP that is pre-established in wandering stage larvae (Berendes HD, Ashburner M 1978, The Salivary Glands. The Genetics and Biology of Drosophila, vol. 2b, pp. 453-498. London: Academic Press; Hammond MP, Laird CD. Chromosoma. 1985;91(3-4):279-86. doi:10.1007/ BF00328223).

2) Throughout the manuscript, the authors combine additional UAS constructs without controlling for the dilution of their Gal4 drive. This affects almost every figure in the latter half of the manuscript, and the effect of this dilution is apparent in Figure S5C, where Drs levels in BxGal4>Drs alone are 1-2 fold higher than BxGal4>Drs,RasV12 double drive.

Previously, we co-expressed a cytoplasmic GFP alongside RasV12, which did not diminish any of the reported effects and phenotypes induced by RasV12 alone (Figure 2 and Figure 3 in Hauling T et al. Biol Open. 2014 Apr15;3(4):250-260. doi:10.1242/bio.20146494). However, to address this aspect more systematically, we used a mCD8-tagged RFP-line to balance the number of UAS constructs with UAS-RasV12 and UAS-Drs. Crucial experiments including the activation of JNK (Figure 5—figure supplement 3B-C; Author response image 2C-D) and CC3 (Figure 6—figure supplement 1C-D; Author response image 2B,D), as well as the expression of JNK-target genes (Figure 5—figure supplement 3A; Author response image 2E) were repeated, while accounting for the number of UAS-constructs. Consistent with our previous results, Drs,RasV12-glands exhibit significantly reduced CC3- and pJNK-signal compared to RasV12-samples, while RFP;;RasV12-glands do not differ significantly from RasV12-glands (Figure 5—figure supplement 3B-C; Figure 6—figure supplement 1C-D; Author response image 2B-D). Most importantly, while Drs,RasV12-glands show a drastic increase in gland size emphasizing the hypertrophic overgrowth, co-expression of RFP alongside RasV12 has no significant influence on the size of the glands (Figure 4—figure supplement 1A; Author response image 2A). This indicates a physiological influence of Drs that is not replicated by simply co-expressing a second protein such as RFP.

Quantifications of JNK-target gene expressions corroborated the difference between Drs- and RFP-co-expression further. At 96 h, apart from Mmp2 all monitored target genes show no significant differences in expression between RFP;;RasV12- and RasV12-glands, while for the same genes Drs,RasV12-glands exhibit significantly reduced expression compared with RasV12-glands (Figure 5—figure supplement 3A; Author response image 2E). At 120 h, Mmp1, Mmp2 and upd1 show the same pattern with these genes being significantly downregulated in Drs,RasV12-, but not in RFP;;RasV12-glands compared to RasV12. While the absolute reduction in gene expression is still stronger in Drs,RasV12-glands for puc, hid and rpr, their RFP;;RasV12-counterparts show a significant decrease, too (Figure 5—figure supplement 3A; Author response image 2E).

While we acknowledge the partial reduction of these genes in RFP;;RasV12-compared to RasV12-glands, several confounding factors complicate the comparability of results from co-expressed constructs used as controls: (1) insertion site of the UAS-constructs, that have an influence on their expressability, (2) metabolic burden due to the size and cellular location of the final protein, in line with the reviewers comment (see below), (3) activity and function of the protein and (3) whether the protein is secreted or not.

Since Drs is a secreted peptide, we opted for mCD8::RFP as control to mimic the entrance into the secretory pathway, too. However, while mCD8::RFP-constructs were shown to be inert in other cell systems, their influence on salivary gland physiology with e.g., its particular high secretory activity has not been documented.

Moreover, sole expression of RasV12 and co-expression of RasV12 with Drs show very similar Drs-expression levels, both at 96 h and 120 h AED. In addition, as indicated by the reviewer, the co-expression of DrsRNAi with RasV12 results in a strong reduction of Drs expression (i.e., only 8% at 96 h and 4% at 120 h of Drs-expression remains after DrsRNAi-co-expression; Figure 5—figure supplement 2A; Author response image 2G), and thus shows the opposite effect as in Drs,RasV12-glands (Figure 5—figure supplement 2CD; Figure 6A). In this regard, the most parsimonious explanation for the differences in expression are the actual constructs driven by BxGal4 together with RasV12 (i.e., alone or in combination) and their differing downstream effects. In fact, the RNAseq data for RasV12 (Figure 1D; Figure 3A-C) is consistent with this explanation highlighting the comprehensive reprogramming, the glands undergo. Moreover, the co-expression of another RNAi-construct targeting lgl increases the severity of the phenotypes (i.e., nuclear and tissue disintegration) further (Figure 1—figure supplement 1C-H).

We also followed the reviewer’s suggestion to overexpress one of the other AMPs upregulated in RasV12-glands, CecA1 (4th highest upregulated; Figure 3C). ISH with a CecA1specific probe revealed an expression of CecA1 exclusively in the PP of RasV12-glands at 96 and 120 h AED in line with the RNAseq-results (Figure 3C; Figure 3—figure supplement 1E; Author response image 2F). By using the same probe, overexpression throughout the whole gland epithelium at both time points with the help of the CecA1-overexpression construct could also be confirmed. However, CecA1;RasV12glands do not exhibit the same abundant CecA1-expression as in CecA1-glands, but instead CecA1-expression is confined to the PP. While the implications of a RasV12-dependent inhibition of CecA1-transcript accumulation in the distal part are beyond the scope of this manuscript, it is obvious that CecA1-coexpression in methodological terms is not simply controlling for Drs-co-expression. In fact, this result suggests further layers of direct interactions between RasV12-driven hypertrophy and the tissue-autonomous immune response beyond immediate expectations.

If RasV12 is diluted in combined UAS lines, this would promote a delayed phenotype, resulting in an apparent rescue effect if you simply compare equal time points. But it would be expected that continued monitoring to 144h would reveal the same apoptotic effects as the lone BxGal4>RasV12 tissue at 120h, similar to how BxGal4>RasV12 at 96h displays less apoptosis than 120h.

We have included a prototypical example of a RasV12-gland at 144 h AED. At this time point, Drs is not only still expressed, but remains confined to the PP reminiscent of its expression at 120 h (Figure 1—figure supplement 1I; Author response image 1E).

In addition, a model of gradual induction of PCD and its delay in Drs,RasV12-glands or other genotypes due to dilution effects implies the reduction of expression of all genes in the gland including Drs due to apoptotic induction. However, given the necessity for JNK signaling during the induction of PCD in RasV12-expressing glands, the inhibition of JNK signaling throughout the entire time of RasV12-expression rescues the glands from undergoing PCD (Figure 6A,D,E-F). However, the glands continue to show a difference in Drs expression between proximal and distal part with a continuous loss of Drs expression in the latter at 120 h AED (Figure 4A; Figure 5—figure supplement 1A).

There is one argument that can be made in favour of the dilution not mattering: Drs-RNAi has a mild but opposite effect at 96h. However it should be noted that RNAi vs. protein overexpression leads to significantly different metabolic burdens on the cell, and thus RNAi is not a good control for this dilution effect. The authors should use a control overexpression construct such as UAS-GFP or UAS-RFP throughout.

We hope that the above outlined additional experiments based on the co-expression of especially mCD8::RFP as well as CecA1 combined with the accompanied clarifications emphasize the unique role Drs has in this hypertrophic model and that its significance is not merely explained by dilution effects. In this regard, we also like to emphasize the significant differences upon DrsRNAi- and Drs-co-expression with RasV12 and the log2-scale used to represent associated qPCR expression values, which can influence the impression of expression differences.

3) There is no logic presented within the paper that this effect should be Drs-specific. Many other AMPs and immune effectors were similarly upregulated in their transcriptome (e.g. AttD, Def). If AttD and Def are similarly induced in this model, it is assumed their expression is similarly regulated, and that overexpression of AttD or Def could similarly rescue RasV12 SGs, and RNAi of AttD or Def could similarly exacerbate RasV12 SGs. I would recommend that, alongside a UAS-GFP control, the authors also test at least some subset of NF-κB effectors alongside their Drs model to elucidate if this is specific to Drs or a general effect. Given findings in other papers on AMP-tumor interactions where overexpressing Drs, Dpt, or Def all had similar effect (Araki et al., 2019), it seems likely that many AMPs should result in the same phenotype, implying there is no molecular pathway where specifically Drs inhibits JNK.

Based on the acquired RNAseq-datasets, Drosomycin is the most highly upregulated gene in whole RasV12-glands (Figure 1D). More importantly, Drosomycin is also the only AMP that is consistently highly upregulated in all three genotypes whose transcriptomes were sequenced (Figure 3—figure supplement 1C; Author response image 1F). Whole tissue samples of RasV12- and lglRNAi;RasV12-glands showed high expression for only one more AMP, AttD (Figure 3—figure supplement 1C; Author response image 1F). However, precise transcript quantifications via qPCR on separated PPs and DPs revealed that this expression results from the upregulation of AttD solely in the PP (Figure 3—figure supplement 1D; Author response image 1C). Crucially, this restricted upregulation not only occurs at 120 h AED as for Drs, but also at 96 h AED. Thus, Drs is the only AMP that is expressed at 96 h AED in the distal part of RasV12-glands and whose downregulation coincides with an increase in JNK-activation and initiation of PCD from 96 h to 120 h AED. In this regard, Drs emerged as the only likely candidate to influence the shift in the DP, an aspect we definitely have missed to clarify in the manuscript before, but have added now.

However, our intention was to merely indicate that Drs expression is capable to interfere with JNK-signaling and sufficient to inhibit it in the presented hypertrophy model, but we did not comment on a unique function of Drs in JNK-inhibition. In this way, we certainly acknowledge the possibility of additional AMPs having similar or different effects on stress signaling in this or other models of overgrowth. However, the mechanism of Drs directly interfering with JNK-signaling rather than with the accumulation of initial stress is a novel concept shedding new light on the necessity of tissue-autonomous immune responses on tissue integrity.

As a proof-of-principle, we overexpressed CecA1 alone or in combination with RasV12 leading to a block of CecA1-transcript accumulation in the DP of RasV12-glands (Figure 3—figure supplement 1E; Author response image 2F). While the underlying mechanism resulting in this block awaits further investigations, it highlights the role of Drs as the only AMP expressed in the DP until 96 h AED further.

Reviewer #2:

Krautz et al. presents an interesting finding on regulation of stress signaling during hypertrophy via tissue-autonomous immune signaling. Data shows that Drosomycin, downstream of the NFkB factor Dorsal, interferes with overgrowth of the salivary gland caused by RasV12 expression. The authors investigate the underlying mechanism of this finding using different approaches, presenting copious data which are generally supportive their model.

However, in some cases the findings lack direct demonstration of key aspects of their model. The conclusions, highlighted in the model shown in Figure 8, claim that Drs inhibits JNK signaling which is shown with some key experiments. Yet, this model also claims that JNK signaling drives tissue disintegration effects of RasV12 through induction of MMP2, yet MMP2 is not upregulated until a very late time point and only to a modest degree.

According to the acquired qPCR-data Mmp2 is for the first time significantly differentially expressed at 120 h AED (Figure 4D; p=0.0263; FC=2.60; log2(FC)=1.38). However, while the upregulation is not significantly changed compared to w1118-control glands, MMP2 is on average already differentially expressed in RasV12-glands at 96 h (Figure 4D; p=0.2928; FC=2.15; log2(FC)=1.10). Given the time necessary for JNK-dependent transcriptional activation around 96 h AED, generating and secreting the Mmp2-protein, as well as enzymatically disrupting the basement membrane and consequentially recruiting hemocytes, the presented data is consistent with the model proposed in the manuscript. While a non-significant, but differential Mmp2expression already at 96 h AED indicates an incline in expression over time, the detection of Mmp2-dependent disruption of the basement membrane at 120 h characterizes the functional consequences in a time-appropriate manner (Figure 4D).

Moreover, overexpression of Mmp1 or Mmp2 independently of one another and of RasV12 indicates that Mmp2 is the only matrix metalloproteinase capable of opening the BM surrounding the glands irrespective of strong differences in the expression of Mmp1 and Mmp2 in RasV12-glands (Figure 4E; Figure 4—figure supplement 1B).

Moreover, they don't show that MMP2 or Hid (another JNK target) are required for these events.

While the results in the manuscript covered presence (i.e., expression, Figure 4D) and functional capability (i.e., sufficiency, Figure 4E; Figure 4—figure supplement 1B) of Mmp2 to degrade the gland’s basement membrane, we indeed failed to show the necessity of Mmp2 for the recruitment of hemocytes as a function of basal membrane degradation in RasV12-glands in line with the reviewer’s comment.

For this purpose, we crossed a Mmp2-mutant allele (Mmp2k00604) with RasV12 and BxGal4 to obtain a hetero- or homozygous knock-out of the Mmp2-locus (Figure 4F-G; Figure 4—figure supplement 1C; Author response image 4A-B,D). Both, in a homozygous, but also already in a heterozygous mutant state, hemocyte attachment is significantly impaired compared to RasV12-glands (Figure 4F-G; Figure 4—figure supplement 1C; Author response image 4A-B,D). To capture the true extent of this reduction (i.e., hemocyte attachment measured as ln(hem-area)/ln(gland size)), we also plotted non-log-transformed effect sizes of hemocyte attachment in Mmp2-/+;RasV12- and Mmp2-/-;RasV12- compared to RasV12-glands (Figure 4G; Author response image 4B). In fact, in both mutant backgrounds, hemocyte attachment has dropped to 33% of the average quantified in RasV12-glands. In addition, the remaining, attached hemocytes are far less spread, rounded and do not form filo- or lamellipodia as their counterparts on the surface of RasV12-glands do (insets in Figure 4—figure supplement 1C; Author response image 4D). Both, the significant reduction of attached hemocytes and the morphology of the residual, surface-bound hemocytes indicates the crucial involvement of JNK-dependent Mmp2-expression in recruitment and activation of hemocytes as a proxy for the systemic immune response towards RasV12-glands.

Author response image 4
Mmp2 expression necessary for hemocyte attachment to the surface of RasV12-glands.

(A) Hemocyte attachment at 120 h AED in RasV12-glands hetero- and homozygous mutant for Mmp2 (Mmp2k00604). (B) Effect size of hetero- and homozygous mutant Mmp2-allele (Mmp2k00604) on hemocyte attachment to and compared to RasV12-glands. Mean of hemocyte attachment in RasV12-glands at 120 h was set to “1”. E. and F. represent the same data points. (C) Quantification of CC3 to measure Dronc-activation upon hidknock-down in RasV12-glands. (D) w1118-control and RasV12-glands hetero- or homozygous for Mmp2 (Mmp2k00604) stained for attached hemocytes with anti-Hemese-antibody. Lower image row shows magnifications of the insets in the upper row. Filopodia (filled arrowheads) and lamellipodia (open arrowheads) indicate hemocyte activation on the surface of RasV12-, but not glands mutant for Mmp2. (E) H2DCF-staining of live glands as general means to evaluate the presence of ROS at 96 h and 120 h AED. (F) Dronc-activation upon hid-knock-down in RasV12glands indicated by CC3-staining normalized by SG size per gland. Insets: (D,F) DAPI, (E) brightfield. Scalebars: (D. Upper,E-F) 100 µm, (D. Lower) 50 µm. Boxplots (A-C): lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median. Significance evaluated by Student's t-tests (*** p<0.001, ** p<0.01, * p<0.05).

These findings are in line with previous reports on the role of Mmp2 in cell dissociation, the integrity of basement membranes and hemocyte attraction (Jia Khan Q et al. Sci Rep. 2014;4:7535. doi:10.1038/srep07535; Kim MJ, Choe KM. PLoS Genet. 2014;10(10):e1004683. doi:10.1371/ journal.pgen.1004683). Moreover, recent, systematic work from the Bergmann lab has not only emphasized the role of Mmp2 in hemocyte recruitment to undead cells and a clonally induced, neoplastic tumor model, but also showed the importance of reactive oxygen species (ROS) in this process (Diwanji N, Bergmann A. Nat Commun. 2020;11(1):3631. doi:10.1038/s41467-020-17399-8). In order to evaluate, whether ROS are also present in the here presented hypertrophic model and thus could aid in the process of basement membrane degradation, we stained live tissues with H2DCF-dye to detect a wide spectrum of ROS in salivary glands (Figure 4—figure supplement 1D; Author response image 4E). In fact, while RasV12-glands show a strong ROS-signal in the DP, co-expression of either Drs or jnkDN with RasV12 appears to prevent ROS-generation. This indicates that apart from inhibiting Mmp2-expression by blocking JNKsignaling, Drs might also block hemocyte recruitment to the gland surface by inhibiting ROS production through its block of JNK-activation and could explain the complete absence of attached hemocytes on jnkDN;;RasV12- and Drs,RasV12-glands in contrast to Mmp2-/-;RasV12- and Mmp2-/+;RasV12-glands (Figure 4D,F-G; Figure 4—figure supplement 1C; Author response image 4A-B,D). However, an association between JNK-signaling and ROS-production in RasV12-glands needs further investigations.

In parallel, to provide further evidence on the significance of hid in JNK-dependent salivary gland degradation, we coexpressed a hid-knock-down construct with RasV12 and stained for CC3 as an indicator for the induction of PCD (Figure 6—figure supplement 1E-F; Author response image 4C,F). Upon reduction of hid expression, the downstream activation of Dronc is significantly reduced compared to RasV12-glands. In fact, coexpression of hidRNAi reduces CC3 to the same level as upon coexpression of Drs with RasV12 indicating that hid plays a crucial quantitative role in the PCD blocked by Drs (Figure 6F; Figure 6—figure supplement 1E-F; Author response image 4C,F).

Also, in some cases the data is not as robust as would be expected in the field; activation of JNK is shown with compelling reporter data and pBsk staining, but the gene expression for JNK targets is underwhelming, for example Puc.

Due to the high expression values identified for JNK target genes in RasV12-expressing glands, we consistently plotted all expression data as log2-transformed values, which for instance implies puc is significantly increased in RasV12-glands at 96 h over control glands. The presented qPCR values for puc are not only in line with the order of magnitude of previously reported puc expression, but comparable to puckered expression values after immune stimulation or initiation of regeneration (Hu Y et al. G3 (Bethesda). 2013 Sep 4;3(9):160716. doi: 10.1534/g3.113.007021; Chakrabarti S et al. PLoS Genet. 2016 May27;12(5):e1006089. doi:10.1371/journal.pgen.1006089. eCollection; Khan SJ et al. PLoS Genet. 2017 Jul28;13(7):e1006937. doi:10.1371/journal.pgen.1006937).

For instance, we obtained the sequencing data for scrib- and dlg-tumor models published in Bunker et al. and compared the fold-change for puc-expression with the values from the salivary gland models presented here (Author response image 5A; Bunker BD et al. eLife. 2015 Feb26;4. doi:10.7554/ eLife.03189). Apart from an expected reduction in proximal gland parts (i.e., lglRNAi;RasV12-PP), our RNAseq data shows comparable puc-expression in RasV12- and lglRNAi;RasV12-glands. Moreover, qPCR-data for puc (presented as non-log-transformed data for comparison) in RasV12-glands at 96 h and 120 h, indicates that the actual puc-expression in hypertrophic glands is even higher (i.e., FC=3-4; Author response image 5B). These values are above the induction level reported for puc during JNK-dependent wing disc pouch region regeneration and similar to puc-levels in gut, hemocytes and the fat body upon septic injury (Khan SJ et al. PLoS Genet. 2017 Jul28;13(7):e1006937. doi: 10.1371/journal.pgen.1006937).

Author response image 5
Puc expression levels in RasV12-glands comparable to other tumor models.

(A) Differential expression values for puc as quantified from the here presented RNAseq datasets and Bunker, B.D. et al., 2015. Comparison of sequenced genotypes are presented in the respective bars. (B) Fold-change expression values for puc without log-transformation in RasV12-glands with Drs-overexpression (Drs) and -knock-down (DrsRNAi). Data reproduced from Figure 5—figure supplement 2C-D. (C) Effect sizes of puc and hid upon DrsRNAi-co-expression in RasV12-glands with respective mean expression in RasV12-glands set to “1” for each gene and time point. (D) Differential expression values for egr, grnd and wgn across the three sample populations screened with RNAseq (i.e., RasV12, lglRNAi;RasV12 or lglRNAi;RasV12-PP each compared to w1118-glands). Q-values are presented in the respective bars. (E) DrsGFP-signal normalized to SG size per gland in RasV12- and w1118control glands each separated by sex into female and male specimens. (F) Effect size for DrsGFP-signal in male RasV12-glands compared to their female equivalents with mean of female RasV12-glands set to “1”. Boxplots (BC,E-F): lower/upper hinges indicate 1st/3rd quartiles, whisker lengths equal 1.5*IQR, red circle and bar represent mean and median.

We have performed similar comparisons with reported and published JNK-target genes beyond puc (data not shown), which are similarly in line with the differential expression values identified in RasV12-glands.

Perhaps the issue is more with the model figure over-extending beyond the actual data presented.

We hope that the additional evidence for Mmp2 and hid being involved in the activation of a systemic immune response and degradation of the gland tissue, reformatting of the figures and a softened choice of words cover the aspects rightfully criticized by the reviewer.

In addition, the authors claims have relied mostly on qualitative nature of the data, whereas quantification could have strengthened their claim. For example, authors conclude that endoreplication leads to increase in nuclear volume but have not shown any quantitative increase in nuclear DNA content. The DAPI data shows most of the nuclei as disintegrated which will amount to higher nuclear volume but is less informative about DNA content. Further, authors have not stated whether the disintegrated nuclei were excluded from the analysis or not. Similarly, RasV12 stimulated MMPs-dependent tissue disintegration and Hid-dependent nuclear disintegration are not directly demonstrated.

We agree entirely with the reviewer’s notion on comprehensively quantifying phenotypes as a necessity for repeatable and transparent science. While several phenotypes throughout the paper were indeed demonstrated only in a qualitative, proof-of-principle manner (i.e., Mmp1/2overexpression; spatio-temporal dl-distribution), conclusions drawn from these instances were kept to their essential minimum (i.e., Mmp2, but not Mmp1 capable of disrupting the basement membrane; overlap of dl with Drs expression patterns). In parallel, phenotypes which the model relies upon were not only quantified, but validated by independent repetitions of the experiments (data not shown), including nuclear volume (Figure 1C; Figure 1—figure supplement 1D-E; Figure 6C), hemocyte attachment (Figure 1F; Figure 1—figure supplement 1H; Figure 4D), Drs-expression (semi-quantitatively as part of the regulator screen; Figure 2A; Figure2—figure supplement 2A), Drs- and JNK-target gene expression via RNAseq and qPCR (Figure 3C,E; Figure 3—figure supplement 2A; Figure 4E; Figure 5B; Figure 5—figure supplement 1B; Figure 5—figure supplement 2C-D; Figure 6A), SG-size (Figure 4C), phosphorylation/activation of JNK (Figure 5C; Figure 5—figure supplement 2F), JNK-reporter activation (Figure 5D; Figure 5—figure supplement 2E; Figure 7B,D) and Dronc-activation (Figure 6F; Figure 6—figure supplement 1A; Figure 7C; Figure 7—figure supplement 1B).

In the course of revising the manuscript, we included additional quantifications, i.e., BxGal4-driver strength (Figure 1—figure supplement 1A-B; Author response image 1A), dl-Drs-correlation (Figure 2—figure supplement 2B; Author response image 1B), dlRNAi- and DrsRNAi-effect sizes (Figure 2—figure supplement 3A; Author response image 3A; Figure 5—figure supplement 2A; Author response image 2G), Drs-expression (semi-quantitatively as part of the regulator screen, Figure 2—figure supplement 3C; Author response image 3C), AttD-expression (Figure 3—figure supplement 1D; Author response image 1C), SG-size (Figure 4—figure supplement 1A; Author response image 2A), JNK activation (Figure 5—figure supplement 3B; Author response image2C), activated Dronc (Figure 6—figure supplement 1B-C; Author response image 2B; Author response image4C), Drs- and JNK-target expression via RNAseq and qPCR (Figure 3—figure supplement 1C; Author response image 1F; Figure 5—figure supplement 3A; Author response image 2E) and hemocyte attachment (Figure 4F-G).

To characterize the impact of continuous Ras-MAPK-signaling, we assayed nuclear volume as an indicator of pre-/absence of additional rounds of endoreplication. In line with the reviewers comment, DAPI has been described as a less efficient fluorochrome for measuring DNA content in the past due to its heterogeneity in nuclear distribution (Santisteban MS et al. J Histochem Cytochem. 1992;40(11):1789-1797. doi:10.1177/40.11.1431064). However, adjustments in the mounting and staining protocol as well as during image capture allow for DAPI to be used in a quantitative manner (Hamada S, Fujita S. Histochemistry. 1983;79, 219–226. doi:10.1007/BF00489783). By using Fluoromount G and reduced laser intensity we followed these recommendations to avoid photobleaching and adapt DAPI staining to quantify nuclear volumes. Moreover, DAPI has been used previously to quantify the occurrence of additional endoreplications in regenerating Drosophila midguts (Xiang J et al. Nat Commun. 2017 May9;8:15125. doi:10.1038/ncomms15125). However, having been aware of possible inconsistencies, we used a threshold-based method rather than aggregating fluorescence intensities. The former is more robust towards variability in DAPI intercalation enabling us to precisely locate and outline nuclei in confocal z-stacks and to quantify nuclear volume as a proxy rather than DNA content itself (Figure 1C; Figure 1—figure supplement 1D,E; Figure 6C).

In parallel, we observed the occurrence of nuclear disintegration in the DPs of RasV12-glands, which poses a two-fold problem for quantifications: (1) separating enlarged, but disintegrating from enlarged, intact nuclei in an unbiased manner and (2) separating nuclei in general due to the propensity of the DAPI-signal of disintegrating nuclei to fuse across cells and nuclei. Attempts to employ machine learning (i.e., support vector machines, k-means, Gaussian Mixture Models) to cluster disintegrating from enlarged nuclei as a prerequisite to quantify their volumes separately were not satisfactory and need more in-depth development. However, we recognized a clear separation between genotypes regarding the occurrence of nuclear disintegration, whereby p35;RasV12-, jnkDN;RasV12- and Drs;RasV12-glands do not show any disintegrating nuclei, only RasV12-glands (Figure 6D). Thus, we chose to show quantifications of nuclear volume as a proxy for the amount of endoreplications alongside images of DAPI stained nuclei to visualize the sole occurrence of nuclear disintegrations in RasV12-glands (Figure 6C-D). We believe that the conclusion we drew from these two datasets is covered by the presented evidence, namely (1) by inhibiting PCD (i.e., p35;RasV12) or inhibiting JNK signaling (i.e., jnkDN;RasV12, Drs;RasV12) nuclear disintegration is prevented, whereas (2) the same interventions do not block the induction of further endoreplications.

As indicated in our summary model, Mmp2 is involved in tissue disintegration in as much as its secretion promotes disruption of the basement membrane, the attraction of hemocytes and thus the activation of a systemic immune response towards the hypertrophic glands (Figure 4E-G; Figure 4—figure supplement 1B-C; Author response image A-B,D; Figure 8). We agree with the reviewer that this however needed clarification throughout the manuscript.

Apart from the above-stated experimental concerns, issue with data presentation also significantly dampens the enthusiasm for this study. The data-presentation is incredibly cumbersome; throughout, the text does not present clearly the rationale of their experimental design, some key figure panels are not even mentioned in the text, other key data is buried in the supplemental figures, while other data is replicated in separate figure panels without clear reasoning, and importantly most of the legends/labeling on the graphs are so small, they are illegible. This last issue makes this work nearly impossible to critically evaluate. Also, I would encourage the authors to use gene names and symbols on their figures that are understood by a wide readership. For example, Bsk might be alternatively JNK in S5.

We have changed formatting of the figures including increasing font sizes, checked the presence of all figure panel references in the text, replaced gene names where possible (e.g., “jnk” instead of “bsk” as suggested by the reviewer). Supplemental figures that include data presented in the main figures show this data alongside additional genotypes in an extended format. By presenting a complete picture of all genotypes in the supplemental figures, we hope to ensure comparability amongst all datapoints.

Most importantly, we revised the manuscript’s text in order to emphasize the reason for conducted experiments as well as for the experimental set-ups.

Reviewer #3:

This article is interesting because it shows that some immune genes, mostly antimicrobial peptide (AMP) genes, are induced in a highly artificial system in which the overgrowth of salivary glands is induced by ectopic expression of the Ras[V12] gain-of-function allele. The expression of immune genes would rely solely on Dorsal and not the Toll pathway (see below). The most important point is that the overexpression of Drosomycin is sufficient to reduce the induction of JNK-pathway regulated genes that occurs in the distal part of salivary glands, including pro-apoptotic genes, thus introducing the concept of a direct or indirect intracellular signaling function for an antimicrobial peptide.

In the longer term, it will be interesting to determine whether Dorsal plays a role in wild-type larval salivary glands and why it is differentially repressed in the DP.

Substantive concerns

1) The authors use RNAi lines to determine whether the Toll pathway is required for the induction of immune genes, which are not even validated. Since it is close to impossible to demonstrate the generation of a null phenotype by RNAi, the authors should imperatively use at least one null mutant of the Toll pathway, e.g., MyD88 and possibly spz.

We have almost exclusively used RNAi-lines for the screen of Toll-/imd-signal pathway components that were previously published and found to impact the expression of the targeted gene, alter expression and have a phenotype. The corresponding references are now listed in the Supplementary file 3. However, we agree entirely with the reviewer’s comment on the necessity to validate the involvement of the Toll-pathway independently. Thus, we crossed a mutant Myd88-allele in a hetero- or homozygous state into the RasV12-background (Figure 2—figure supplement 2E; Author response image 3E). However, Drs-expression is not diminished at 120 h AED in the PP of Myd88/+;RasV12- or Myd88-/-;RasV12- compared to RasV12-glands, whereas a dl-knock-out reduced Drs-expression significantly, indicating a dependence on dl, but not upstream Toll-signaling modules (Figure 2C; Figure 2—figure supplement 2D).

2) The authors should look at and quantify Dorsal nuclear localization in the salivary glands and not solely at its expression. Upon looking at Figure 2B, one gets the impression it is nuclear only in the very proximal region of the organ.

We agree with the reviewer’s comment on the value of quantifications based on antibody staining parallel to evaluating the expression. For this purpose, we captured z-stacks of RasV12glands containing the DrsGFP-reporter and stained for the presence of dorsal-protein. We used DAPI to define location and dimensions of the glands nuclei and exploited the nuclear signal of DrsGFP (i.e., proxy for cellular Drs-expression) to correlate dl and Drs expression (Figure 2—figure supplement 2B; Author response image 1B). Across the entire scale of quantified intensities (i.e., log10-transformed; range covers entirety from non-detectable to high intensities throughout the nucleus), dl- and Drs-signal correlate strongly with one another in both the PP and DP as shown for 5 captured glands (i.e., Pearson correlation coefficient included for all 5 samples). More importantly, dlstaining is present in both PP and DP and occurs to stimulate Drs-expression in the entire gland epithelium.

3) Multiple AMPs are induced in the PP and it is surprising that the one AMP they have overexpressed is having such a biological function. A control is missing, ideally a Drosomycin mutant in which disulfide bridges cannot form; at least a nonrelevant gene should be used as a control to exclude nonspecific effects linked to overexpression in the Ras[V12] context.

Among the induced AMPs in the acquired transcriptomes, only Drs and AttD showed considerable expression in the two sample populations derived from whole glands (i.e., RasV12, lglRNAi;RasV12; Figure 3C; Figure 3—figure supplement 1C; Author response image 1F). Given the strong AttD-induction in the PP, its expression in the whole gland might not reflect the actual distribution (Figure 3C). Thus, we performed qPCRs on separated gland parts with AttD-primers, which revealed that AttD is exclusively expressed in the PP at 96 and 120 h AED (Figure 3—figure supplement 1D; Author response image 1C). According to our data, Drs is thus the only AMP that is expressed in the DP of 96 h-old RasV12-glands. While this doesn’t exclude the possibility of other AMPs to function in a fashion similar to Drs, the latter is the only one present in the DP to alter RasV12-induced phenotypes until 96 h AED.

In parallel, we also overexpressed CecA1, the 4th highest expressed AMP in RasV12- and lglRNAi;RasV12-glands and monitored CecA1-mRNA occurrence via ISH (Figure 3C; Figure 3—figure supplement 1C; Author response image 1F). While CecA1 is solely expressed in the PP of 120 h-old RasV12-glands, the employed CecA1-overepression construct driven by BxMS1096 validates the possibility for CecA1 to be expressed and detected throughout the entire salivary gland (Figure 3—figure supplement 1E; Author response image 2F). However, CecA1;RasV12-glands only exhibited CecA1-staining in the PP, indicating a RasV12induced block of post-transcriptional mRNA-accumulation. While the precise mechanism for this block remains to be elucidated, this result is opposed to the Drs,RasV12-coexpression, where co-expressed Drs could be detected throughout the gland (Figure 4A). This highlights the unique presence and function of Drs in the hypertrophic RasV12-model even further.

Nevertheless, we agree strongly with the reviewer’s comment on the relevance of a control to rule out non-specific effects derived from the overexpression of the UAS-Drs construct. To mimic Drosomycin’s entry into the secretory pathway, we combined a mCD8::RFP-construct with RasV12 and repeated crucial experiments (i.e., CC3- and pJNKstaining, qPCR for Drs and JNK-target genes, measurement of gland size; Figure 5—figure supplement 3A-C; Figure 6—figure supplement 1C-D; Author response image 2A-D). Co-expression of Drs with RasV12 significantly reduces the activation of Dronc and JNK compared with RasV12 alone, while co-expression of mCD8::RFP does not (Figure 5—figure supplement 3B; Figure 6—figure supplement 1C; Author response image 2B-D).

The difference between Drs,RasV12- and RFP;;RasV12-glands is further confirmed by the performed qPCRs. Except for the JNK-target Mmp2, none of the screened genes displays significant expression differences in RFP;; RasV12- compared to RasV12-glands at 96 h AED (Figure 5—figure supplement 3A; Author response image 2E). At the same time point, the same genes are indeed significantly downregulated in Drs,RasV12- in contrast to RasV12-glands. At 120 h, co-expression of Drs with RasV12, but not mCD8::RFP, still reduces expression of Mmp1, Mmp2 and upd1 significantly (Figure 5—figure supplement 3A; Author response image 2E). In addition, Drs,RasV12-glands show a stronger absolute reduction of puc-, hid- and rpr-expression than RFP;;RasV12-glands compared to RasV12-glands alone, in spite of a significant reduction of these genes in RFP;;RasV12-glands.

Most importantly, while co-expression of mCD8::RFP has a moderate, but significant impact on the expression of few of the JNK target genes at 120 h AED, RFP;;RasV12glands do not differ significantly in size from their RasV12-counterparts (Figure 4—figure supplement 1A; Author response image 2A). Drs,RasV12-glands in turn are strongly overgrown indicating prolonged hypertrophic growth in the absence of JNK-activation and PCD. Crucially, the set of RFP;;RasV12-control experiments indicates that Drs has a specific role in altering the physiology of RasV12-glands that is not mimicked by merely overexpressing a mCD8::RFP-construct.

While not significant, RFP;;RasV12-glands are indeed slightly smaller than RasV12-glands. Since mCD8::RFP-marker constructs were to our knowledge not rigorously tested in polyploid tissues like salivary glands, it remains open as to whether mCD8::RFP itself has specific effects separate of Drs that would explain the minor reduction in tissue size. In addition, we have previously described RasV12-induced phenotypes, while co-expressing a cytoplasmic GFP, which did not diminish the reported effects either (Figure 2 and Figure 3 in Hauling T et al. Biol Open. 2014 Apr15;3(4):250-260. doi:10.1242/bio.20146494).

Also, the Drosomycin RNAi line does not abolish the expression of its target gene and does not have such a "drastic" effect: several genes are only mildly more strongly expressed. It would be interesting to use a Drosomycin null mutant. Perhaps less important for this work but of interest, it would have been informative to test mutants affecting AMP families that have recently been published.

While we absolutely agree with the reviewer’s suggestion to use a null mutant to completely abolish Drs expression in further investigations, the DrsRNAi-line used throughout the manuscript exhibits a comprehensive reduction of Drs-expression levels. On average only 8% of the Drs-expression in RasV12-glands remains upon DrsRNAi-co-expression at 96 h AED and only 4% at 120 h AED, in line with the virtual absence of detected Drs-mRNA in ISH (Figure 5—figure supplement 2A-B; Author response image 2G). In addition, as exemplified for hid and puc, by knocking Drs-expression down, puc-expression increases by almost 50% at 96 h AED and hid-expression increases by almost 100% at the same time point (Author response image 5C). As expected this increase of expression is diminished at 120 h AED (i.e., compared to RasV12), due to the absence of Drs-expression in the distal part of RasV12-glands, making a knock-down of Drs obsolete. Given the effect size of the used DrsRNAi-construct and its impact on JNK-target genes at 96 h AED, we believe that even without additional Drs-mutant experiments, we were able to show an effect opposite of Drs-overexpression by DrsRNAi-co-expression to strengthen the validity of our model (Figure 5—figure supplement 2C-D; Author response image 5C).

The reviewer’s suggestion to use the here presented hypertrophic model to test the functionality of additional AMPs as done in 2 recently published landmark papers paves intriguing possibilities for follw-up studies (Araki et al., 2019; Parvy et al., 2019). Since Drs was the only AMP expressed in the distal part until 96 h AED, downregulated thereafter and thus anti-correlated with the activation of JNK-signaling, we reasoned that Drs remains the only possible candidate to mount a tissue-autonomous immune response capable of explaining the shift in phenotypes from 96 h to 120 h AED. While this does not rule out similar activities of other AMPs in this or alternative models, the focus of the presented manuscript rested upon Drs as the naturally occurring AMP in this model and its role in regulating hypertrophic growth.

4) It is not clear whether Drosomycin is secreted in the lumen or basally in the hemolymph. To really conclude that this is a tissue-autonomous response, inasmuch the authors do not state whether Drosomycin is expressed in other tissues, it would be useful to overexpress Drosomycin in the fat body and test its effects on the DP of the salivary glands.

In order to detect Drs-expression in an unambiguous manner, we employed the previously published Drs-reporter including 2450 bp of the regulatory sequence upstream of the translation start site (Ferrandon D et al. EMBO J. 1998 Aug10;17(5):121727. doi:10.1093/emboj/ 17.5.1217). Due to its systemic presence, any non-tissue autonomous induction of Drosomycin in BxGal4>UAS-RasV12- or BxGal4>UAS-lglRNAi;UAS-RasV12-larvae would thus become obvious when monitored over time, as shown in other infection and wounding paradigms (Kenmoku H et al. Dis Model Mech. 2017 Mar1;10(3):271-281. doi:10.1242/dmm.027102; Kounatidis I et al. G3 (Bethesda). 2018;8(5):1637-1647. doi:10.1534/g3.118.200182). However, throughout larval and pupal development, a DrsGFP-signal was solely detected in the salivary glands with lymph glands, fat body, hemocytes or hematopoietic pockets all Drsnegative (Figure 1E; Figure 1—figure supplement 1E).

While we cannot unequivocally rule out interference from other tissues, the most parsimonious explanation based on the presented evidence is a tissue-autonomous expression and activity of Drs in the hypertrophic glands: (1) Expression and presence of Drs in a spatiotemporal pattern that precisely anti-correlates with the induction of JNK-signaling. (2) Rescue of tissue degradation in the distal part by Drs-overexpression across the gland’s epithelium at 120 h AED. (3) Further increase of JNK-activation upon tissue-wide knock-down of Drs at 96 h AED. (4) Physiological expression of dl as a crucial upstream activator of Drs in a spatiotemporal pattern completely overlapping with Drs. (5) Absence of any non-tissue-autonomous expression of Drs, especially no other dedicated immune organ.

We very much appreciate the reviewer’s suggestion to test further inter-organ communication between the hypertrophic glands and amongst others the fat body and agree that further evaluations would be of interest. However, in the light of substantial evidence for a tissue-autonomous activity of Drs, we deem establishing, optimizing and validating a dualexpression system (i.e., 2 two-component expression systems, QF-QUAS or LexA-lexAop alongside Gal4-UAS) for parallel expression of RasV12 in the salivary glands and Drs in the fat body exceptional for an evaluating experiment.

https://doi.org/10.7554/eLife.64919.sa2

Article and author information

Author details

  1. Robert Krautz

    Department of Molecular Biosciences, The Wenner-Gren Institute (MBW), Stockholm University, Stockholm, Sweden
    Present address
    The Bioinformatics Centre, Department of Biology, University of Copenhagen, Copenhagen, Europe
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing
    Contributed equally with
    Dilan Khalili
    For correspondence
    rkrautz@binf.ku.dk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0457-1348
  2. Dilan Khalili

    Department of Molecular Biosciences, The Wenner-Gren Institute (MBW), Stockholm University, Stockholm, Sweden
    Contribution
    Data curation, Validation, Investigation, Visualization, Writing - review and editing
    Contributed equally with
    Robert Krautz
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9785-9641
  3. Ulrich Theopold

    Department of Molecular Biosciences, The Wenner-Gren Institute (MBW), Stockholm University, Stockholm, Sweden
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing - review and editing
    For correspondence
    uli.theopold@su.se
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1009-8254

Funding

Vetenskapsrådet (VR-2010-5988)

  • Ulrich Theopold

Vetenskapsrådet (VR 2016-04077)

  • Ulrich Theopold

Swedish Cancer Foundation (CAN 2010/553)

  • Ulrich Theopold

Swedish Cancer Foundation (CAN 2013/546)

  • Ulrich Theopold

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank S Höglund and the Imaging facility at Stockholm University for microscopy support, the Bloomington Drosophila Stock Center and the Vienna Drosophila Resource Center for fly stocks, the Drosophila Genomics Resource Center for the Drs and CecA1 cDNAs, the Developmental Studies Hybridoma Bank for the anti-dl antibody and D Bohmann, S Certel, Y Engström, W-J Lee, B Lemaitre as well as M Suzanne for providing us with flies. We are especially grateful to R Karlsson, I Söll and G Hauptmann for decisive feedback on experimental procedures and J van den Ameele and the three reviewers for their critical review of the manuscript. This work was supported by the Swedish Research Council (VR-2010–5988 and VR 2016–04077) and the Swedish Cancer Foundation (CAN 2010/553 and CAN 2013/546).

Senior Editor

  1. Utpal Banerjee, University of California, Los Angeles, United States

Reviewing Editor

  1. Bruno Lemaitre, École Polytechnique Fédérale de Lausanne, Switzerland

Reviewer

  1. Mark Austin Hanson, École Polytechnique Fédérale de Lausanne, Switzerland

Publication history

  1. Received: November 15, 2020
  2. Accepted: December 29, 2020
  3. Accepted Manuscript published: December 30, 2020 (version 1)
  4. Version of Record published: February 12, 2021 (version 2)

Copyright

© 2020, Krautz et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,653
    Page views
  • 263
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Robert Krautz
  2. Dilan Khalili
  3. Ulrich Theopold
(2020)
Tissue-autonomous immune response regulates stress signaling during hypertrophy
eLife 9:e64919.
https://doi.org/10.7554/eLife.64919

Further reading

    1. Cancer Biology
    2. Computational and Systems Biology
    Xiangkun Wu, Hong Yan ... Li Liang
    Research Article

    Colorectal cancer (CRC) remains a challenging and deadly disease with high tumor microenvironment (TME) heterogeneity. Using an integrative multi-omics analysis and artificial intelligence-enabled spatial analysis of whole-slide images, we performed a comprehensive characterization of TME in colorectal cancer (CCCRC). CRC samples were classified into four CCCRC subtypes with distinct TME features, namely, C1 as the proliferative subtype with low immunogenicity; C2 as the immunosuppressed subtype with the terminally exhausted immune characteristics; C3 as the immune-excluded subtype with the distinct upregulation of stromal components and a lack of T cell infiltration in the tumor core; and C4 as the immunomodulatory subtype with the remarkable upregulation of anti-tumor immune components. The four CCCRC subtypes had distinct histopathologic and molecular characteristics, therapeutic efficacy, and prognosis. We found that the C1 subtype may be suitable for chemotherapy and cetuximab, the C2 subtype may benefit from a combination of chemotherapy and bevacizumab, the C3 subtype has increased sensitivity to the WNT pathway inhibitor WIKI4, and the C4 subtype is a potential candidate for immune checkpoint blockade treatment. Importantly, we established a simple gene classifier for accurate identification of each CCCRC subtype. Collectively our integrative analysis ultimately established a holistic framework to thoroughly dissect the TME of CRC, and the CCCRC classification system with high biological interpretability may contribute to biomarker discovery and future clinical trial design.

    1. Cancer Biology
    Jiangfei Chen, Kunal Baxi ... Myron S Ignatius
    Research Article

    In embryonal rhabdomyosarcoma (ERMS) and generally in sarcomas, the role of wild-type and loss or gain-of-function TP53 mutations remains largely undefined. Eliminating mutant or restoring wild-type p53 is challenging; nevertheless, understanding p53 variant effects on tumorigenesis remains central to realizing better treatment outcomes. In ERMS, >70% of patients retain wild-type TP53, yet mutations when present are associated with worse prognosis. Employing a kRASG12D-driven ERMS tumor model and tp53 null (tp53-/-) zebrafish, we define wild-type and patient-specific TP53 mutant effects on tumorigenesis. We demonstrate that tp53 is a major suppressor of tumorigenesis, where tp53 loss expands tumor initiation from <35% to >97% of animals. Characterizing three patient-specific alleles reveals that TP53C176F partially retains wild-type p53 apoptotic activity that can be exploited, whereas TP53P153D and TP53Y220C encode two structurally related proteins with gain-of-function effects that predispose to head musculature ERMS. TP53P153D unexpectedly also predisposes to hedgehog expressing medulloblastomas in the kRASG12D-driven ERMS-model.