Tumor-derived ligands trigger tumor growth and host wasting via differential MEK activation

Abstract

Interactions between tumors and host tissues play essential roles in tumor-induced systemic wasting and cancer cachexia, including muscle wasting and lipid loss. However, the pathogenic molecular mechanisms of wasting are still poorly understood. Using a fly model of tumor-induced organ wasting, we observed aberrant MEK activation in both tumors and host tissues of flies bearing gut-yki^3SA tumors. We found that host MEK activation results in muscle wasting and lipid loss, while tumor MEK activation is required for tumor growth. Strikingly, host MEK suppression alone is sufficient to abolish the wasting phenotypes without affecting tumor growth. We further uncovered that yki^3SA tumors produce the vein (vn) ligand to trigger autonomous Egfr/MEK-induced tumor growth, and produce the PDGF- and VEGF-related factor 1 (Pvf1) ligand to non-autonomously activate host Pvr/MEK signaling and wasting. Altogether, our results demonstrate the essential roles and molecular mechanisms of differential MEK activation in tumor-induced host wasting.

Graphical Abstract

eTOC Blurb

Mechanisms of tumor-induced host wasting are largely unknown. Song et al. provide evidence that tumor-derived ligands trigger tumor growth and host wasting via differential MEK activation. They demonstrate that yki^3SA-gut tumors produce vn and Pvf1 ligands to trigger MEK-associated tumor growth and host wasting, respectively.

Introduction

Many patients with advanced cancer exhibit a systemic wasting syndrome, referred to as “cancer cachexia”, with major features of progressive loss of muscle and adipose tissues. Cachexia is associated with poor chemotherapy response, reduced life quality and increased mortality (Fearon et al., 2013). Unlike malnutrition conditions, cachexia can rarely be reversed by nutritional supplementation and is frequently accompanied with hyperglycemia (Chevalier and Farsijani, 2014). A number of findings in cultured cells have indicated that, in addition to systemic inflammatory responses , tumors produce secreted factors (e.g., Interleukines and activins) that directly target myotubes and adipocytes to cause myotube wasting and lipid loss, respectively (Miyamoto et al., 2016). Antibody neutralization of tumor-derived cachectic ligands (PTHrP) also significantly improves host wasting (Kir et al., 2014). Despite these advances, genetic animal models to comprehensively assess tumor-secreted ligands, the signaling pathways they regulate and their effects in various tissues are far less established.

The adult Drosophila midgut has emerged as a model system to study tumorigenesis. Many mutations involved in human cancer have been found to result in overproliferation of fly intestinal stem cells (ISCs) and tumor formation (Patel and Edgar, 2014). The fly midgut has also been established as a conserved genetic model to study tumor-induced host wasting. We found that induction of an active oncogene yorkie (yki^3SA), the homolog of human Yap1, that causes gut tumor formation is associated with organ wasting phenotypes, including muscle dysfunction, lipid loss, and hyperglycemia. Mechanisms included that yki^3SA-gut tumors produce the IGF-antagonizing peptide ImpL2 that suppresses systemic insulin signaling and anabolism and contributes to host wasting (Kwon et al., 2015). In addition, ImpL2 regulation of host wasting has been also observed in transplanted tumors that are generated from fly imaginal discs (Figueroa-Clarevega and Bilder, 2015). Together, these findings emphasize that communication between tumor and host organs is a general phenomenon, and that Drosophila can be used to dissect the molecular mechanisms involved in tumor-host interaction.

The MEK/ERK cascade is a highly conserved Mitogen-Activated Protein Kinase (MAPK) pathway involved in various biological regulations in both fly and mammals (Friedman and Perrimon, 2006). MEK signaling, in addition to controlling cell proliferation, promotes muscle atrophy via modulation of ubiquitin-dependent protein degradation and enhances lipid mobilization via modulating GPCR/cAMP cascade (Hong et al., 2018; Zheng et al., 2010). These results indicate that MEK signaling is associated with loss of muscle and adipose tissues, the main features of cancer cachexia, and suggest that MEK activation may be involved in tumor-induced host wasting. However, administration of MEK inhibitors in tumor-bearing mice and patients are associated with inconsistent results (Au et al., 2016; Prado et al., 2012; Quan-Jun et al., 2017).

In this study, we revealed aberrant MEK activation in both tumors and host tissues (muscle and fat body) of flies bearing yki^3SA-gut tumors. In the context of gut-tumor growth, pharmaceutical inhibition of MEK signaling in the host tissues only is sufficient to alleviate host wasting, including muscle wasting, lipid loss, hyperglycemia, and elevated mortality, in tumor-bearing flies. Integrating RNA-seq and RNAi screening, we demonstrate that yki^3SA-gut tumors produce Pvf1 ligand to activate MEK signaling and enhance catabolism in host tissues. yki^3SA-gut tumors also produce vn ligand to autonomously promote MEK signaling and self-growth.

Results

MEK activation in yki^3SA-tumor-bearing flies contributes to wasting of host tissues

To analyze the role of MEK/ERK signaling in tumor-host interaction, we first examined whether MEK signaling is activated in host tissues during tumor-induced wasting. We expressed an activated form of yorkie (yki^3SA, yki^{S111A-S168A-S250A} triple mutant) in adult ISCs using the temperature-sensitive GAL4 driver to trigger development of GFP-labeled gut tumors (esg-GAL4, tub-GAL80^TS,UAS-GFP/+; UAS-yki^3SA/+, referred to as yki^3SA tumors) and host wasting (Fig. 1A and Fig. S1A-G). Interestingly, the canonical readout of MEK signaling, pdERK (encoded by rl), in the muscle and fat body remained unchanged at day 2 of tumor induction (referred to as the “proliferation” state) but was significantly increased at day 4 (yki^3SA tumors expand to the whole midgut - referred to as the “tumorigenesis” state), day 6 (tumor-bearing flies exhibit swollen abdomen and moderate TAG decrease and muscle dysfunction - referred as the “ascites” state), and day 8 (tumor-bearing flies exhibit translucent abdomen, severe TAG decrease and carbohydrate increase and muscle dysfunction - referred as the “wasting/bloating” state) (Fig. 1A-C and Fig. S1A-G). These observations indicate that MEK signaling is activated in host tissues during yki^3SA tumor-induced wasting.

Figure 1. MEK activation in host tissues of yki^3SA-tumor-bearing flies.

(A-B) Changes in the bloating appearance (A, bright field), gut tumors (A, fluorescent, green), and pdERK in the fat body and muscles (B), of yki^3SA-tumor-bearing (esg- GAL4, UAS-GFP, tub-GAL80^TS /+; UAS-yki^3SA/+) (yki^3SA) and control (esg-GAL4, UAS-GFP, tub-GAL80^TS/+) (ctrl) flies. (C) pdERK and Bodipy staining (lipid) in the fat bodies at day 6. (D-G) Muscle pdERK (D), protein degradation rates (E, n=3, 30 flies/group), degenerative phenotypes (F, arrow indicates impaired myofibril integrity), and climbing speeds (G, centimeter/sec, normalized to control) (n=60) of dMef2-GAL4, tub-GAL80^TS > UAS flies after 8 days at 29°C. (H-J) pdERK (H) and in vitro lipolysis rates in the fat body (I, mg released glycerol/mg protein/hour, normalized to control) (n=3, 30 abdomens/group) and lipid storages (J, mg TAG/mg protein, normalized to control) (n=3, 30 flies/group) of CG-GAL4, tub-GAL80^TS > UAS flies after 8 days at 29°C . (K) Nile Red staining (lipid) in adult fat body clones after transgene induction for 3 days. Data are presented as means ± SEM. * p < 0.05.

Further, we manipulated MEK signaling specifically in wild-type muscle or fat body. Consistent with yki^3SA-tumor-bearing flies (Fig. S1E-G), specific activation of MEK signaling via expression of an activated form of Raf (Raf^F179) or simply wild-type dERK in the wild-type muscle resulted in muscle-wasting phenotypes, including enhanced muscular protein degradation, impairment of myofiber integrity (gaps between myofibers and mitochondria as indicated), and climbing defects (Fig. 1D-G). MEK suppression via dERK knockdown in wild-type muscles decreased protein degradation and improved fly climbing ability (Fig. 1D-G). MEK signaling manipulation in wild-type fat bodies significantly affected lipolysis rate and lipid storage (Fig. 1H-K and Fig. S1H), phenocopying lipid dysregulation in yki^3SA-tumor-bearing flies (Fig. S1A and S1I). Thus, our results demonstrate that MEK activation in host tissues results in muscle wasting and lipid loss.

Autonomous MEK activation is essential for yki^3SA-tumor growth

We next tested whether host MEK inhibition is sufficient to alleviate tumor-induced wasting. As systemic disruption of MEK signaling causes developmental lethality, we fed flies trametinib (Tram), an efficient MEK inhibitor (Slack et al., 2015). We initially fed flies normal food containing 10 μM or 100 μM Tram at tumor initiation (day 0). Both doses resulted in the strong MEK suppression in host tissues and diminished the bloated/wasting phenotypes of yki^3SA-tumor-bearing flies at day 8 (Fig. 2A-B). However, we observed a strong reduction of yki^3SA tumor growth in the midgut (Fig. 2A), suggesting that MEK activation is crucial for yki^3SA-tumor growth. Consistently, we observed a robust increase of pdERK in yki^3SA-tumor gut cells (Fig. 2C). Specific dERK knockdown in yki^3SA-tumors was sufficient to terminate tumor growth (Fig. 2D and Fig. S1J-K). However, dERK gain-of-function alone in ISCs only slightly increased ISCs proliferation but failed to cause bloating/wasting (Fig. S1L-N). Altogether, our results indicate that autonomous MEK activation in ISCs is required for yki^3SA-tumor growth but is not sufficient to induce host wasting.

Figure 2. Pharmaceutical MEK inhibition in host tissues abolishes yki^3SA tumors-induced wasting.

(A-B) Bloating phenotypes (A, up), gut tumors (A, down), and pdERK in the fat body and muscles (B) were measured after Tram treatment simultaneously with tumor induction for 8 days. (C) pERK in yki^3SA-tumor midgut at day 8. (D) Bloating (up) and gut tumors (down) of flies bearing yki^3SA-gut tumors with dERK-RNAi (esg-GAL4, UAS-GFP, tub-GAL80^TS/+; UAS-yki^3SA/UAS-dERK-RNAi). (E) Schematic host MEK suppression by Tram in yki^3SA+dERK^SEM tumor-bearing flies (esg-GAL4, UAS-GFP, tub-GAL80^TS/UAS-dERK^SEM; UAS-yki^3SA/+). (F-J) Wasting phenotypes of flies treated with Tram simultaneously with yki^3SA+dERK^SEM tumor induction for 8 days: bloating (F, up), gut tumors (F, down), pdERK (G), muscle degeneration (H), lipid and trehalose storage (I, n=3, 30 flies/group), climbing speed (I, n=60), and lifespan (J, n=120. Genotype of ctrl is esg-GAL4, UAS-GFP, tub-GAL80^TS/+). Data are presented as means ± SEM. * p < 0.05.

Host MEK inhibition is sufficient to suppress yki^3SA tumor-induced wasting

To further investigate the role of MEK signaling in host tissues in the context of yki^3SA-tumor growth, we fed yki^3SA-tumor-bearing flies Tram with a lower dosage after tumor formation from day 4 (Fig. S2A-B). 1, 10, and 100 μM Tram significantly decreased pdERK in both muscles and fat body of yki^3SA-tumor-bearing flies in a dose-dependent manner (Fig. S2D). 100 μM Tram still terminated, while 1 and 10 μM Tram hardly affected, yki^3SA-tumor growth at day 8 (Fig. S2C). Strikingly, 1 and 10 μM Tram potently alleviated wasting phenotypes, including abdomen bloating, muscle degeneration, lipid loss, and hyperglycemia in the presence of yki^3SA tumors (Fig. S2C-K). To validate the direct effect of Tram on host tissues, we treated isolated adult fat bodies from yki^3SA-tumor-bearing flies with Tram in vitro and confirmed that Tram robustly suppressed yki^3SA-tumor-induced lipolysis (Fig. S1I). Similar to wasting-associated mortality of cancer patients (Fearon et al., 2013), yki^3SA-tumor-bearing flies also exhibited a shortened lifespan, while it was surprisingly extended by 1 and 10 μM Tram (Fig. S2L-M). Note that, feeding 1 μM Tram did not affect yki^3SA-tumor growth even at day 30 when most yki^3SA-tumor-bearing flies died (Fig. S2N).

To further exclude the drug effect on yki^3SA-gut tumors and evaluate the impact of MEK signaling in host tissues only, we generated drug-resistant yki^3SA tumors by specifically overexpressing an active form of dERK (encoded by rl^SEM) (esg-GAL4, tub-GAL80^TS, UAS-GFP/UAS-rl^SEM; UAS-yki^3SA/+, referred to as “yki^3SA+dERK^SEM tumor”). Feeding 10 μM Tram simultaneously at yki^3SA+dERK^SEM tumor initiation from day 0 could no longer affect gut-tumor growth at day 8 (Fig. 2E-F), while dramatically decreasing host MEK activation (Fig. 2G). Strikingly, bloating/wasting phenotypes, including muscle defect, lipid loss, hyperglycemia, and mortality, were significantly alleviated (Fig. 2F-J). Note that Tram administration in control flies (esg-GAL4, tub-GAL80^TS, UAS-GFP/+) rarely affected wasting effects (Fig. S3A-B). We also confirmed the anti-wasting effects of another MEK inhibitor, PD0325901 (PD), in flies bearing yki^3SA+dERK^SEM tumors (Fig. S3C-E, albeit at a higher dose). Collectively, our results demonstrate that host MEK suppression is sufficient to improve yki^3SA-tumor-induced wasting.

MEK activation contributes to LLC-induced wasting of adipocytes and myotubes

We next tested whether the similar effects of MEK activation could be observed in mammalian wasting models. Conditioned medium from Mouse LLC (Lewis lung carcinoma) cancer cells strongly induces lipid loss in adipocytes and myotube atrophy (Rohm et al., 2016; Zhang et al., 2017). We found that LLC-conditioned medium activates MEK signaling in both cultured C2C12 myotubes and 3T3-L1 adipocytes after 15-min treatment (Fig. S4A and S4F). Interestingly, LLC-conditioned medium robustly resulted in, while adding Tram significantly alleviated, MEK-associated induction of ubiquitination-related genes (UbC and USP19) and protein degradation, decrease of MHC level, and atrophy phenotype in C2C12 myotubes (Liu et al., 2011; Zheng et al., 2010) (Fig. S4A-E). Similarly, LLC-conditioned medium potently enhanced, while adding Tram significantly hampered, lipolysis rate and TAG decline in differentiated 3T3-L1 adipocytes (Fig. S4F-I). Thus, our results demonstrate that MEK activation also causes tumor-induced wasting in mammalian myotubes and adipocytes.

Using RNA-seq and RNAi screening to identify yki^3SA-tumor-derived cachectic ligands

We hypothesized that MEK non-autonomous regulation of host wasting is caused by secreted proteins from yki^3SA-gut tumors. To identify such factors, we characterized the transcriptomic changes in yki^3SA-tumor midguts using RNA-seq. Following statistical analysis, 2211 genes were found to be significantly changed (fold change > 2), encompassing 1659 up- and 552 down-regulated genes in yki^3SA midguts (Fig. 3A and Table. S1). Among the 794 genes encoding secreted proteins annotated by GLAD Gene Ontology (Hu et al., 2015), 92 were significantly up-regulated and 36 were down-regulated (Table. S1). Genes encoding different trypsins were the most differentially regulated ones in yki^3SA-tumor guts (Fig. S3F, up). As flies bearing yki^3SA tumors did not exhibit digestion/absorption problem (Kwon et al., 2015), we speculate that yki^3SA tumors only change the composition of trypsin production but do not affect trypsin-associated nutrient absorption in the gut. Interestingly, the transcriptional levels of gut hormones, which are produced in enteroendocrine cells (EEs) (Reiher et al., 2011; Song et al., 2017a; Song et al., 2014), were decreased in yki^3SA midguts (Fig. S3F, down). Immunostainings also revealed that EEs (Pros+ cells) were largely missing in yki^3SA midguts (Fig. S3G). We next asked whether yki^3SA-gut tumors cause host wasting via loss of EEs and gut hormones, and overexpressed an active form of Notch (N^act) in ISCs to genetically suppress EE generation in the midgut (Takashima et al., 2011). Consistently, ISC overexpression of N^act potently eliminated EEs in the midgut (Fig. S3G), but it failed to affect bloating/wasting effects (Fig. S3H). N^act overexpression slightly reduced survival rates of flies (Fig. S3I), but the extent of decline was very marginal as compared to yki^3SA-tumor-bearing flies.

Figure 3. yki^3SA-gut tumors cause host MEK activation and wasting via Pvf1 production.

(A) Scatter plots comparing gene expression in ctrl and yki^3SA midguts. (B) Heat map showing up-regulated genes that encode secreted proteins. (C) Bloating and tumor phenotypes of flies bearing gut-yki^3SA tumors with RNAi against ligand-encoding genes. (D) RTK-ligand expressions in yki^3SA-tumor midguts at day 8 (n=3, 30 midguts/group). (E-L) Wasting phenotypes of flies bearing yki^3SA-tumor with indicated transgenes (esg-GAL4, UAS-GFP, tub-GAL80^TS/+; UAS-yki^3SA/UAS-X): bloating (E-F, up), gut tumors (E-F, down), Pvf1 expression (G, antibody staining in control midgut; and H, n=3, 30 midguts/group), dERK (I), muscle degeneration (J), lipid and trehalose levels (K, n=3, 30 flies/group), climbing speeds (K, n=60), and lifespans (L, n=120). (M-N) Bloating and gut phenotypes (M) and wasting effects, including TAG and trehalose (n=3, 30 flies/group) levels and climbing speed (n=60) (N), in Pvf1-null flies bearing yki^3SA-gut tumor (Pvf1⁻/Pvf1⁻; esg-GAL4, UAS-GFP, tub-GAL80^TS/+; UAS-yki^3SA/+). Data are presented as means ± SEM. * p < 0.05.

We next hypothesized that yki^3SA-gut tumors cause host wasting via increasing the production of a cachectic ligand(s) such as ImpL2. We thus performed an RNAi screening of most up-regulated ligand-encoding genes in yki^3SA tumors (Fig. 3B). Interestingly, knockdown of 39 ligand-encoding genes (60 RNAi lines) in yki^3SA-gut tumors still exhibited both tumor growth and abdomen bloating. However, knockdown of 17 ligand-encoding genes (24 RNAi lines) in yki^3SA-gut tumors, referred to as Group “Tumor without Bloating”, diminished bloating without affecting tumor growth, suggesting that these ligands regulate tumor-induced wasting. Knockdown of 8 genes (11 RNAi lines), referred as Group “Non-tumor”, exhibited no tumor growth, suggesting that these ligands are essential for yki-induced tumorigenesis (Fig. 3C and Table. S2).

yki^3SA tumors produce vn to autonomously promote Egfr/MEK-induced tumor growth

Among the established MEK-activating ligands that are induced in yki^3SA-tumor guts (Fig. 3D and Fig. S3J), only Pvf1 knockdown in yki^3SA tumors potently abolished the bloating phenotypes without perturbing yki^3SA-tumor growth, while only vn knockdown terminated yki^3SA-tumor growth (Fig. 3C and 3E). Consistently, blockade of vn/Egfr signaling via knockdown of Egfr or its downstream Ras85D, but not blockade of Pvf1/Pvr signaling via overexpressing a dominant negative Pvr (Pvr.DN), terminated yki^3SA-tumor growth (Fig. 3F). These results indicate that yki^3SA-tumor cells produce vn to activate autonomous Egfr/MEK signaling and promote growth.

yki^3SA tumors produce Pvf1 to activate host Pvr/MEK signaling and result in host wasting

We next confirmed Pvf1 expression in the ISCs (Fig. 3G) and found that Pvf1 knockdown in yki^3SA tumors suppressed pdERK levels significantly in the fat body and moderately in muscles and subsequently improved systemic wasting, including TAG decline, trehalose elevation, and muscle defects, in tumor-bearing flies (Fig. 3H-K). We also generated yki-tumors in Pvf1-null mutant flies (Wu et al., 2009) and confirmed that systemic removal of Pvf1 also significantly alleviated the wasting effects in host tissues without perturbing yki^3SA-tumor growth (Fig. 3M-N). Despite restoration of host wasting, Pvf1 knockdown in yki^3SA tumors failed to extend longevity of tumor-bearing flies (Fig. 3L), indicating that tumor-associated mortality involves other processes than host energy regulation. Since vn knockdown diminished tumor growth, systemic wasting effects were not observed in flies bearing yki^3SA tumors with vn RNAi (Fig. 3K-L).

We further validated the autonomous effects of Pvr/MEK signaling by overexpressing an active form of Pvr (Pvr. λ) in the host tissues. Consistently, Pvr gain-of-function in the fat body resulted in MEK activation, lipid loss, and trehalose elevation (Fig. 4A-C), while in muscle potently caused climbing defects (Fig. 4D-E). To mimic Pvf1 induction in yki^3SA-tumor-bearing midgut, we further overexpressed Pvf1 in wild-type enterocytes and found MEK activation in both muscle and fat body and wasting effects, even though no obvious abdomen bloating was observed (Fig. 4F-G). Note that, Tram administration in flies bearing yki^3SA+dERK^SEM tumors failed to affect Pvf1 levels in the midgut (Fig. S3J-K), suggesting that Tram suppresses host MEK signaling independent of midgut Pvf1. Collectively, our results demonstrate that yki^3SA-gut tumors produce Pvf1 to non-autonomously enhance Pvr/MEK signaling in host tissues and cause wasting.

Figure 4. Pvf1/Pvr axis regulates MEK activation and host wasting.

(A-C) TAG and trehalose levels (A, C) (n=3, 30 flies/group) and pdERK (B) in wild-type flies with fat body Pvr activation at 29°C for 4 da ys. (D-E) Climbing speed (D, n=60) and pdERK (E) in wild-type flies with muscle Pvr activation at day 8. (F-G) Midgut Pvf1 mRNA (F, left, n=3, 30 midguts/group), wasting effects (F, right), including TAG and trehalose (n=3, 30 flies/group) levels and climbing speed (n=60), and pdERK (G) in flies with enterocyte Pvf1 overexpression at 29°C for 8 days. (H) Ligand-associated dual-MEK regulation of tumor growth and host wasting. (I) pdERK in flies bearing yki^3SA-tumors with ImpL2 RNAi (esg-GAL4, UAS-GFP, tub-GAL80^TS/UAS-ImpL2-RNAi-NIG15009R-3; UAS-yki^3SA/+). (J-K) Bloating and gut tumors (J) and wasting effects (K), including bloating rates (n=3, 60 flies/group), TAG and trehalose levels (n=3, 30 flies/group), as well as climbing speeds (n=60), of flies bearing yki^3SA tumors with RNAi at day 10. UAS-Pvf1-RNAi-NIG7103R-1 and UAS-ImpL2-RNAi-VDRC30931 were used. Data are presented as means ± SEM. * p < 0.05.

Pvf1 and ImpL2 are independent regulators of yki^3SA-tumor-associated host wasting

We previously showed that yki^3SA-tumor-derived ImpL2 contributes to host wasting (Kwon et al., 2015). Thus, we wondered whether ImpL2 and Pvf1 are independent regulators. We first examined whether tumor-derived ImpL2 affects host MEK signaling and found that ImpL2 knockdown in yki^3SA tumors failed to suppress host pdERK (Fig. 4I). In addition, MEK suppression does not impinge on tumor-derived ImpL2 to affect wasting, as administration of 10 μM Tram in flies bearing yki^3SA+dERK^SEM tumors failed to affect tumor ImpL2 production, brain ILPs levels, as well as feeding behavior (Fig. S3K-M). Similarly, Pvf1 removal in yki^3SA-gut tumors also barely affected tumor ImpL2 production (Fig. S3N). To examine a potential crosstalk between Pvf1 and ImpL2, we knocked down both Pvf1 and ImpL2 in yki^3SA tumors. Surprisingly, as compared to knockdown of either Pvf1 or ImpL2, double knockdown of these two ligands in yki^3SA tumors further alleviated host wasting, including bloating rates, lipid loss, as well as hyperglycemia, without affecting tumor growth (Fig. 4J-K). Taken together, our results demonstrate that yki^3SA-tumor-derived Pvf1 and ImpL2 are independent regulators of host wasting.

Discussion

MEK/ERK signaling induces muscle wasting and lipid loss

Lipid loss and protein degradation-associated muscle wasting are two major features of tumor-induced host wasting in mammals (Fearon et al., 2013). We observe similar outcomes in yki^3SA-tumor-bearing flies and show that the MEK/ERK pathway acts as a critical pathogenic factor. Although the molecular mechanisms of MEK action in fly are unknown, our results are reminiscent of some mammalian studies. For example, MEK signaling was found to regulate ubiquitination-associated proteolysis in mouse myotubes and promote lipid mobilization in mammalian adipocytes (Hong et al., 2018; Zheng et al., 2010). We further confirmed similar MEK regulation in mammalian wasting models. Therefore, our study reveals that host MEK signaling plays a conserved role in tumor-induced wasting, including muscle wasting and lipid loss.

Ligand-mediated dual-MEK regulation of tumor-host interaction

The mechanism underlying the regulation of MEK signaling in tumor-induced host wasting is poorly understood. Here we identified the secreted protein Pvf1 as a tumor-derived factor activating host MEK signaling. Pvr, the Pvf1 receptor, is expressed in both fat body and muscle (Kwon et al., 2015; Zheng et al., 2017). Activation of Pvf/Pvr signaling by overexpressing an active form of Pvr in the muscle and fat body is sufficient to cause muscle wasting and lipid loss, respectively. Reducing Pvf1 production in yki^3SA-gut tumors decreased host MEK signaling and robustly alleviated wasting effects, while Pvf1 overexpression in wild-type midgut enterocytes somehow mimicked tumor-induced-wasting effects, confirming that yki^3SA-gut tumor-derived Pvf1 non- autonomously promotes host MEK signaling and wasting. Note that, as compared to pharmaceutical MEK inhibition in host tissues, Pvf1 removal from yki^3SA-gut tumors fails to prolong survival, suggesting that systemic regulation of survival is more complicated than energy wasting in the context of tumor growth. We speculate that Pvf1/Pvr/MEK signaling is missing from some host tissues and that additional MEK-activating ligands produced either directly from yki^3SA tumors or indirectly in host tissues contribute to mortality.

Autonomous MEK activation in yki^3SA-gut tumors is required for tumor growth. We demonstrate that only vn, a MEK-activating ligand essential for ISC proliferation (Xu et al., 2011), is critical for yki^3SA-gut tumor growth by activating Egfr/MEK signaling. Altogether, we propose a dual-MEK regulatory model whereby yki^3SA-gut tumors produce vn to autonomously promote self-growth, and Pvf1 to non-autonomously trigger Pvr/MEK signaling in host tissues and cause wasting (Fig. 4H).

Organ wasting involves energy balance that is regulated by multiple tumor-derived ligands.

We have previously shown that yki^3SA-gut tumors produce ImpL2 to suppress systemic insulin signaling and anabolism, which contribute to host wasting. However, ImpL2 removal from yki^3SA-gut tumors fails to completely abolish host wasting (Kwon et al., 2015), suggesting the existence of other cachectic factor(s). Here, we demonstrate that yki^3SA-gut tumors also produce Pvf1 that remotely promotes MEK signaling and catabolism in host tissues and causes wasting. Further, removal of both ImpL2 and Pvf1 in yki^3SA-gut tumors exhibits additive improvement of host wasting, as compared to removal of either Impl2 or Pvf1. Thus, yki^3SA-gut tumors produce, at least, Pvf1 and ImpL2 to orchestrate organ wasting, involving both MEK and insulin signaling-associated catabolism and anabolism, respectively.

Interestingly, induction of these cachectic ligands is highly tumor-context dependent. Compared to yki^3SA, overexpression of a mild active form, yki^S168A (Oh and Irvine, 2009), in the ISCs only results in weak tumor growth, mild induction of Pvf1 and ImpL2, and slight host wasting (data not shown). In Ras^V12scrib⁻ imaginal disc tumors (Figueroa-Clarevega and Bilder, 2015), only ImpL2, but not Pvf1, is highly induced (data not shown). Our study also suggests other potential cachectic factors in yki^3SA tumors, as removal of other ligands than ImpL2 and Pvf1 diminished the bloating as well (Fig. 3C). As tumor-induced wasting is a complicated physiological process, it is not surprising to observe differential involvement of multiple ligands/regulators. Further studies will be needed to characterize the functions of tumor-derived ligands and the intricate cross-talks between them.

Relevance to MEK cascade and cancer cachexia.

In addition to flies and mouse cells, we also observed that Tram remarkably abolished MEK activation in host tissues and improved wasting in LLC-tumor-bearing mice (data not shown). However, we cannot conclude the beneficial effects from host MEK inhibition, as Tram also reduced tumor sizes by 50% (data now shown). Thus, we speculate that the inconsistent results of MEK inhibition (Au et al., 2016; Quan-Jun et al., 2017) might be caused by different tumor responses to MEK inhibitors. To address the impact of MEK inhibition in host tissues but not in tumors, we generated drug-resistant yki^3SA+dERK^SEM gut tumors in flies. Strikingly, MEK inhibitors, while no longer affecting tumor growth or tumor-derived ligand production, remarkably suppressed MEK signaling in host tissues (fat body, muscle, and maybe other secondary responsive organs) and abolished energy wasting. Thus, our results indicate that host MEK suppression is sufficient to reduce systemic wasting. Meanwhile, it will be worthy to obtain drug-resistant LLC cancer cells, via either random mutagenesis or genetic manipulation of MEK/ERK pathway, and evaluate the effects of MEK inhibitors in mice bearing drug-resistant LLC tumors in the future.

Finally, our results provide genetic evidence of tumor-secreted proteins, rather than systemic inflammatory responses, in tumor-induced wasting. VEGF exists in mammals as a homolog of Drosophila Pvf1 and activates Ras/MEK signaling (Holmes and Zachary, 2005). Similar to cachectic fly Pvf1, VEGF has been reported to be associated with lipid mobilization and muscle atrophy in mice (Gao et al., 2015; Sun et al., 2012). Administration of antibodies against VEGF or VEGFR-2 elicits beneficial effects on multiple organs and prolongs survival of mice bearing high-VEGF tumors (Xue et al., 2008). However, other tumor-derived ligands, including LIF and ILs, are also important MEK-activating factors and are associated with systemic catabolism (Miyamoto et al., 2016; Seto et al.,2015). We therefore propose that cachectic tumors might produce multiple MEK-activating ligands to trigger organ wasting in mice and patients.

STAR Methods

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Wei Song (songw@whu.edu.cn).

Experimental Model and Subject Details

Fly strains

Driver lines for the midgut ISC (esg-GAL4, tub-GAL80^TS, UAS-GFP) (Kwon et al., 2015), fat body (CG-GAL4, tub-GAL80^TS), (tub-GAL80^TS; R4-Gal4), and (tub-GAL80^TS; Lpp-Gal4) , muscle (tub-GAL80^TS; dMef2-GAL4) and (Mhc-Gal4) (Song et al., 2017b), and midgut enterocytes (tub-GAL80^TS; Myo1A-GAL4, UAS-GFP) (Song et al., 2014) have been described previously. UAS-yki^3SA (#28817), UAS-Raf^F179 (#2033), UAS-rl (referred to as “UAS-dERK”, #36270), UAS-rl^SEM (referred to as “UAS-dERK^SEM”, #59006), UAS-N^intra (#52008), UAS-Pvf1 (#58426), UAS-Pvr.λ (#58428 and 58496), UAS-Pvr.DN (#58430) were obtained from the Bloomington Drosophila Stock Center. Pvf1-null mutant line (Vegf17E^1624ex3) was a kind gift from Michael J Galko. eyFLP1; Act5C>y+>GAL4, UAS-GFP; [FRT]82B, tub-GAL80 and UAS-Ras^V12, [FRT]82B, scrib¹ lines used to generated Ras^V12scrib¹ tumors were obtained from Tian Xu and Xianjue Ma.

Established RNAi lines were obtained from Bloomington Drosophila Stock Center, TRiP at Harvard Medical School, VDRC Stock Center, and the NIG-FLY stock center, including: UAS-dERK (rl)–RNAi (TRiP HMS00173), UAS-ImpL2-RNAi (NIG 15009R-3 and VDRC 30931) (Kwon et al., 2015), UAS-Egfr-RNAi (TRiP JF01368) and UAS-Ras85D-RNAi (TRiP HMS01294). Multiple RNAi lines against Pvf1 (VDRC 102699, NIG 7103R-1 and R-2) or vn (TRiP HMC04390, VDRC 109437, NIG 10491R-2) exhibited similar phenotypes, thus only UAS-Pvf1-RNAi-VDRC 102699 and NIG 7103R-1 and UAS-vn-RNAi-NIG 10491R-2 were shown in Fig. 3 and 4. Negative controls, w¹¹¹⁸ and UAS-w-RNAi, exhibited similar phenotypes and only UAS-w-RNAi is shown in the figures. Others RNAi lines used for in vivo RNAi screening are shown in Table S2.

Cell lines

Mouse pre-adipocyte 3T3-L1 and mouse myoblasts C2C12 were purchased from American Type Culture Collection (ATCC). C2C12 myoblasts (< 15 passages) were cultured in growth medium (DMEM with 10% FBS and antibiotics) and differentiated into myotubes in differentiation medium, DMEM containing 2% horse serum (ThermoFisher, 16050130), for 5 days at 37 °C in 5% CO ₂. 3T3-L1 pre-adipocytes were cultured in growth medium. For differentiation, after 2 days at full confluency (day 0), 3T3-L1 pre-adipocytes were switched into DMEM containing 10% FBS, 0.5 mM IBMX, 0.25 μM dexamethasone, 1 μg/mL insulin, and 2 μM rosiglitazone (Sigma) for 2 days (day 2), and switched to DMEM with 10% FBS, 1 μg/mL insulin, and 2 μM rosiglitazone for another 2 days (day 4). Differentiated 3T3-L1 adipocytes were then maintained in growth medium for at least 4 days. LLC mouse tumor cells were cultured in growth medium. For LLC-conditioned medium preparation, after reaching 80% confluency, LLC cells were washed with PBS and cultured in growth medium (for adipocytes) or differentiation medium (for myotube) for the next 24h. Then conditioned mediums were centrifuged at 1,200 g for 10 min, filtered with a 0.2 μM syringe filtered, and stored at −80 °C or used immediately after 3:1 dilu tion with fresh medium.

Method details

Gut tumor induction

Crosses were set up with esg-GAL4, tub-GAL80^TS, UAS-GFP and UAS-yki^3SA or UAS-dERK^SEM; UAS-yki^3SA at 18°C to inactivate GAL4. 4-day-old adult progenies were placed at 29°C to induce the transge nes. For the in vivo RNAi screening of ligands in yki^3SA-tumor guts, different RNAi lines were crossed to esg-GAL4, tub-GAL80^TS, UAS-GFP/CyO; UAS-yki^3SA/TM6B at 18 °C . Virgin progenies were maintained at 18 °C for at least 8 d ays and then switched to 29 °C to induce transgenes. Lipid and sugar levels, cl imbing speed, and muscle morphology, gut morphology, and bloating phenotypes were all measured after switching to 29 °C for 8 days. For drug treatment, flies were transferred onto food containing inhibitors at day 0 (simultaneously with tumor induction) or day 4 (after tumor formation). MEK/ERK inhibitors PD0325901 (S1036) and Trametinib (S2673) were purchased from Selleckchem.

Protein degradation

To measure protein degradation rates in adult fly muscles, 20 flies were cultured on normal food at 29°C to induce tumors for 4 days and then transferred onto normal food with 5 μCi [³H]-tyrosine (PerkinElmer, NET127250UC) for 2 days to label fly proteins. 10 flies were washed with PBS and the thorax parts were dissected and lysed with 500 μL RIPA buffer at 4 °C. After centrifugation at 12,000 rpm for 10 min at 4°C, 400 μL supernatant was used for measurement of labeled proteins using 5 mL Ultima Gold liquid scintillation cocktail (PerkinElmer) and scintillation counter. Radioactivity values were referred to as “basal” level. The other 10 flies were then transferred to normal food without [³H]-tyrosine for another 2 days. Similarly, radioactivity in the thoraces was measured and referred to as “wasted” level. The protein degradation rate was defined by subtracting “wasted” level from “basal” level and then normalized to protein levels. To measure protein degradation in mouse myotubes, 3d-differentiated C2C12 myotube were incubated with differentiation medium (2% horse serum) containing 1 μCi/mL [³H]-tyrosine for 48h to label cellular proteins. The cells were washed with PBS and transferred to differentiation medium with 2 mM tyrosine for 2h to exclude short-lived proteins. Cells were then cultured with control or LLC-conditioned differentiation medium (3:1 diluted with normal differentiation medium) containing 2 mM tyrosine for 12h to measure release of [³H]-tyrosine. 500 μL culture medium was mixed with 100 μL 10 mg/mL BSA and precipitated with 600 μL 20% (wt/vol) TCA overnight at 4°C. After centrifu gation at 12,000 rpm for 5 min, the TCA-soluble radioactivity from supernatant, which reflect the protein degradation rate, was measured using scintillation counter.

Lipid and carbohydrate measurements in flies

We measured fly TAG and carbohydrates as described previously (Kwon et al., 2015; Song et al., 2014). Briefly, 10 flies from each group were homogenized using Multi-sample Tissuelyser-24 (Shanghai Jingxin Technology) with 1 mL PBS containing 0.2% Triton X-100 and heated at 70 °C fo r 5 min. The supernatant was collected after centrifugation at 14,000 rpm for 10 min at 4 °C 10 μl of supernatant was used for protein quantification using Bradford Reagent (Sigma, B6916-500ML). Whole body trehalose levels were measured from 10 μl of supernatant treated with 0.2 μl trehalase (Megazyme, E-TREH) at 37 °C for 30 min using glucose assay reagent (Megazyme, K-GLUC) following the manufacturer’s protocol. We subtracted the amount of free glucose from the measurement and then normalized the subtracted values to protein levels in the supernatant. To measure whole body triglycerides, we processed 10 μl of supernatant using a Serum Triglyceride Determination kit (Sigma, TR0100). We subtracted the amount of free glycerol from the measurement and then normalized the subtracted values to protein levels. To measure circulating trehalose concentrations, hemolymph was extracted from 20 decapitated adults by centrifugation at 1,500 g for 10 min. 0.5 μL collected hemolymph was diluted in 40 μl of PBS, heated at 70°C for 5 min, and centrifuged at 14,000 rpm at 4°C for 10 min. 10 μl supernatant was treated with 0.2 μl trehalase (Megazyme, E-TREH) at 37°C for 30 min and then used to measure circulating trehalose levels with glucose assay reagent (Megazyme, K-GLUC). We subtracted the amount of free glucose in the supernatant from the measurement.

Lipolysis measurements

For lipolysis measurement in flies, 10 adult abdomens containing fat bodies with midguts removed were dissected and washed with 1 mL M3 Insect Medium (Sigma, S8398) at room temperature. Supernatant was aspirated after a brief centrifugation and 10 abdomens were incubated with 100 μL M3 medium containing 4% fatty-acid-free BSA (w/v, Sigma, A7030-1KG) with or without 1 mM Trametinib for 1h at room temperature. Released glycerol in M3 medium was determined using Free Glycerol Reagent (Sigma, F6428-40ML). Abdomens were later lysed in RIPA buffer and used for protein quantification with Bradford Reagent (Sigma, B6916-500ML). Final lipolysis rate was calculated by normalizing released glycerol level to abdomen protein amount. For lipolysis measurement in adipocytes, after treated with control or conditioned medium 24h, 3T3-L1 adipocytes were washed with PBS and incubated with serum- and phenol-free DMEM containing 4% fatty-acid-free BSA with or without trametinib for 2h. The glycerol amount in the medium was immediately measured using a Free Glycerol Reagent kit (F6428, Sigma). After treated with conditioned medium and drugs for 24h, 3T3-L1 adipocytes were washed and harvested in TNET buffer (50 mM Tris-HCl (pH7.4), 150 mM NaCl, 2 mM EDTA, and 1% Triton X-100) with 0.5% cholate. Dissolved TAG was measured using a Serum Triglyceride Determination kit (TR0100, Sigma) and normalized to protein level.

Climbing activity

Ten flies were placed in an empty vial and then tapped down to the bottom. They were allowed to climb for 3 seconds. Climbing was recorded and climbing height and speed were calculated from the video. A minimum of 60 flies and 10 separate trials were run per condition.

Immunostainings

Brains, midguts, and abdomens containing fat bodies were dissected in PBS and fixed for 15 min in PBS containing 4% paraformaldehyde. After fixation, the samples were washed with PBS containing 0.2% Triton X-100 (PBST) and blocked with 1% BSA in PBST for 30 min. After incubation with primary antibodies overnight at 4°C: p-ERK (1:100, Cell Signaling, 4370), Prospero (1:100, DSHB, MR1A), ILP2 (1:1000, a kind gift from Hugo Stocker), or Pvf1 (1:50, a kind gift from Ben-Zion Shilo). Tissues were washed and then incubated with secondary antibody and DAPI for 1h, washed, and mounted in Vectashield (Vector). C2C12 myoblasts were cultured and differentiated on cover slides. Treated C2C12 myotubes were washed and fixed for 15 min in PBS containing 4% formaldehyde. After fixation, the samples were washed with PBST, blocked with 1% BSA in PBST, and incubated with primary antibody against MHC (1:50, DSHB, MF20) overnight at 4°C. Cells were then incub ated with secondary antibody and DAPI for 1h, washed and mounted in Vectashield (Vector). Treated 3T3-L1 mature adipocytes were incubated with Bodipy 493/503 (1 μg/mL, Life Technologies, D3922) for 20 min, washed, and imaged. Regular microscopy was performed on a Zeiss Axioskop 2motplus or a Nikon SMZ18 and confocal images were obtained using a Leica system.

Western blot

10 adult thoraces, 10 adult abdomens without midgut, and treated C2C12 myotubes and 3T3-L1 mature adipocytes were lysed in RIPA buffer containing inhibitors of proteases and phophatases. Extracts were immunoblotted with indicated antibodies: rabbit anti-ERK (1:1000, Cell Signaling, 4695), rabbit anti-phospho-ERK (1:1000, Cell Signaling, 4370), ±-tubulin (1:5000, Sigma, T5168).

RNA-seq analysis of adult midgut

15 midguts of ctrl or yki^3SA flies incubated for 8 days at 29Ό were dissected for total RNA extraction. After assessing RNA quality with Agilent Bioanalyzer (RIN > 7), total RNAs were sent to Columbia Genome Center for RNA-seq analysis following the standard protocol. Briefly, mRNAs were enriched by poly-A pull-down. Sequencing libraries were prepared with Illumina Truseq RNA preparation kit and were sequenced using Illumina HiSeq 2000. Samples were multiplexed in each lane, which yields target number of single-end 100-bp reads for each sample, as a fraction of 180 million reads for the whole lane. The RNA-seq data were deposited in the Gene Expression Omnibus (accession number GSE113728). After trimming, sequence reads were mapped to the Drosophila genome (FlyBase genome annotation version r6.15) using Tophat. With the uniquely mapped reads, gene expression was quantified using Cufflinks (FPKM values) and HTseq (read counts per gene). Differentially expressed genes were analyzed based on both adjusted P value using DSeq2 as well as fold change cut-off. Prior to fold change calculation, we set to a value of “1” for any FPKM value between 0 and 1 to reduce the possibility that we get large ratio values for genes with negligible levels of detected transcript in both experimental and control samples (e.g. FPKM 0.1 vs. 0.0001), as those ratios are unlikely to have biological relevance. A cut-off of 2-fold change consistently observed among replicates and the adjusted P value of 0.05 or lower from DSeq2 analysis were used as criteria to define the set of 552 down-regulated and 1659 up-regulated genes. Heatmap was generated using MEV_4_7 based on FPKM changes.

qPCR

10 adult midguts or heads of each genotype and C2C12 myotubes were lysed with Trizol for RNA extraction and cDNA transcribed using the iScript cDNA Synthesis Kit (Bio-rad). qPCR was then performed using iQ SYBR Green Supermix on a CFX96 Real-Time System/C1000 Thermal Cycler (Bio-rad). Drosophila and mouse gene expression were normalized to RpL32 and β-actin, respectively. qPCR primers are listed in STAR methods.

Electron Microscopy

Adult thoraces were processed and analyzed in cross-section following standard protocols at Electron Microscopy Facility in Harvard Medical School. Briefly, thoraces were fixed in 0.1 M sodium cacodylate buffer (pH 7.4) containing 2.5% glutaraldehyde, 2% paraformaldehyde overnight. The fixed samples were washed in 0.1M cacodylate buffer, fixed again with 1% osmiumtetroxide (OsO4) and 1.5% potassium ferrocyanide (KFeCN6) for 1 hour, and washed 3 times in water. Samples were incubated in 1% aqueous uranyl acetate for 1 hour and followed by 2 washes in water and subsequent dehydration in grades of alcohol. The samples were then put in propyleneoxide for 1 hour and embedded in TAAB Epon (Marivac Canada Inc.). Ultrathin sections (about 60nm) were cut on a Reichert Ultracut-S microtome, moved to copper grids, and then stained with lead citrate. Sections were examined in a JEOL 1200EX Transmission electron microscope, and images recorded with an AMT 2k CCD camera.

Statistical Analyses.

Data are presented as the mean ± SEM. Unpaired Student’s t test and one-way ANOVA followed by post-hoc test were performed to assess the differences. p < 0.05 was considered statistically significant.

Supplementary Material

1

Supplemental Table 1. Differential gene expression in yki3SA-tumor midguts. Related to Figure 3.

2

Supplemental Table 2. In vivo RNAi screening of ligands in yki^3SA-tumor midguts. Related to Figure 3.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-pERK	Cell Signaling Technology	Cat# 4370; RRID: AB_2315112
Rabbit anti-ERK	Cell Signaling Technology	Cat# 4695; RRID: AB_390779
Mouse anti-α-Tubulin	Sigma-Aldrich	Cat# T5168; RRID: AB_477579
Mouse anti-Prospero	Developmental Studies Hybridoma Bank	Cat# Prospero (MR1A); RRID: AB_528440
Mouse anti-MHC	Developmental Studies Hybridoma Bank	Cat# MF 20; RRID: AB_2147781
Rabbit anti-ILP2	(Broughton et al.,2008)	N/A
Rat anti-Pvf1	(Rosin et al., 2004)	N/A
Bacterial and Virus Strains
N/A
Biological Samples
N/A
Chemicals, Peptides, and Recombinant Proteins
TRizol reagent	Thermo Fisher Scientific	15596018
iScript Reverse Transcription Supermix	Bio-Rad	1708896
iQ SYBR Green Supermix	Bio-Rad	1708880
Protease and phosphatase inhibitor cocktail	Pierce	78440
Glycerol standard	Sigma-Aldrich	G7793-5ML
D-(+)-Glucose	Sigma-Aldrich	G7021
Trehalase	Megazyme	K-TREH
PD0325901	Selleckchem	S1036
Trametinib	Selleckchem	S2673
DMEM, High glucose	Thermo Fisher Scientific	11965092
DMEM, High Glucose, HEPES, no Phenol Red	Thermo Fisher Scientific	21063029
Fetal Bovine Serum	Thermo Fisher Scientific	10437028
Horse Serum	Thermo Fisher Scientific	16050130
Bovine Serum Albumin	Sigma-Aldrich	A7030-1KG
IBMX	Sigma-Aldrich	I5879
Dexamethasone	Sigma-Aldrich	D1756
Rosiglitazone	Sigma-Aldrich	R2408
Insulin	Sigma-Aldrich	I2643
(Hydroxypropyl)methyl cellulose	Sigma-Aldrich	H7509
Paraformaldehyde 16% Solution	Fisher Scientific	50-980-487
[3H]-tyrosine	PerkinElmer	NET127250UC
M3 Insect Medium	Sigma-Aldrich	S8398
Bodipy 493/503	Thermo Fisher Scientific	D3922
Critical Commercial Assays
Free glycerol reagent	Sigma-Aldrich	F6428-40ML
Triglyceride reagent	Sigma-Aldrich	T2449-10ML
Bradford Reagent	Sigma-Aldrich	B6916-500ML
D-Glucose Assay Kit	Megazyme	K-GLUC
Deposited Data
Data files for RNA sequencing	This paper	GSE113728
Experimental Models: Cell Lines
Mouse: C2C12	ATCC
Mouse: 3T3-L1	ATCC
Mouse: LLC	(Kir et al., 2014)	N/A
Experimental Models: Organisms/Strains
Esg-Gal4, UAS-GFP, tub-Gal80ts	(Kwon et al., 2015)	N/A
Cg-Gal4	(Song et al., 2017a)
R4-Gal4	(Song et al., 2017a)
Lpp-Gal4	(Song et al., 2017a)
dMef2-Gal4	(Song et al., 2017b)
Mhc-Gal4	(Song et al., 2017b)
Myo1A-Gal4	(Song et al., 2014)	N/A
UAS-yki3SA	BDSC	28817
UAS-Raf F179	BDSC	2033
UAS-rl	BDSC	36270
UAS-rlSEM	BDSC	59006
UAS-Nintra	BDSC	52008
UAS-Pvfl	BDSC	58426
UAS-Pvr.λ.mp1 (Pvr.λ1)	BDSC	58428
UAS-Pvr.λ.mp10 (Pvr.λ10)	BDSC	58496
UAS-Pvr.DN	BDSC	58430
Pvf1-null (Vegf17El624ex3)	(Wu et al., 2009)	N/A
UAS-dERK (rl)-RNAi	TRiP	HMS00173
UAS-Egfr-RNAi	TRiP	JF01368
UAS-Ras85D-RNAi	TRiP	HMS01294
UAS-ImpL2-RNAi	NIG	15009R-3
UAS-ImpL2-RNAi	VDRC	30931
UAS-vn-RNAi	NIG	10491R-2
UAS-Pvf1-RNAi	VDRC	102699
UAS-Pvf1-RNAi	NIG	7103R-1
Other RNAi lines	This paper	See Table S2
Oligonucleotides
Primers for Drosophila grk qPCR F: GTCGCCGTCACAGATTGTTG R: GATTGAGCAACTCAACCGCG	This paper	N/A
Primers for Drosophila Krn qPCR F: CCGCTTTAATCGGCGCTTAC R: ATCGGGAAGGTGACATTCGG	This paper	N/A
Primers for Drosophila spi qPCR F: TGCGGTGAAGATAGCCGATC R: TTCGCATCGCTGTCCCATAA	This paper	N/A
Primers for Drosophila vn qPCR F: GAACGCAGAGGTCACGAAGA R: GAGCGCACTATTAGCTCGGA	This paper	N/A
Primers for Drosophila Pvf1 qPCR F: CTGTCCGTGTCCGCTGAG R: CTCGCCGGACACATCGTAG	This paper	N/A
Primers for Drosophila Pvf2 qPCR F: GGTGGTCCACATCACGAGAG R: CGACTTTGTCGCTGCATCTG	This paper	N/A
Primers for Drosophila Pvf3 qPCR F: TCGTGAAGAGCAGTAAGCATCG R: AGGTGCAACTCAGTATGGTGG	This paper	N/A
Primers for Drosophila ImpL2 qPCR F: AAGAGCCGTGGACCTGGTA R: TTGGTGAACTTGAGCCAGTCG	This paper	N/A
Primers for Drosophila Ilp2 qPCR F: ATGAGCAAGCCTTTGTCCTTC R: ACCTCGTTGAGCTTTTCACTG	This paper	N/A
Primers for Drosophila Ilp3 qPCR F: ATGGGCATCGAGATGAGGTG R: CGTTGAAGCCATACACACAGAG	This paper	N/A
Primers for Drosophila Ilp5 qPCR F: CGCTCCGTGATCCCAGTTC R: AGGCAACCCTCAGCATGTC	This paper	N/A
Primers for Drosophila RpL32 qPCR F: GCTAAGCTGTCGCACAAATG R: GTTCGATCCGTAACCGATGT	This paper	N/A
Primers for Mouse MHC2B qPCR F: AGT CCCAGGTCAACAAGCT G R: TTTCTCCTGTCACCTCTCAACA	This paper	N/A
Primers for Mouse UbC qPCR F: CCAGTGTTACCACCAAGAAGG R: ACACCCAAGAACAAGCACAA	This paper	N/A
Primers for Mouse USP19 qPCR F: GTGGCCCTCTCTCCTGAAAC R: GGCAATGGGAGACAGCTCTT	This paper	N/A
Primers for Mouse β-actin qPCR F: CGGTTCCGATGCCCTGAGGCTCTT R: CGTCACACTTCATGATGGAATTGA	This paper	N/A
Recombinant DNA
N/A
Software and Algorithms
Photoshop	Adobe	N/A
ImageJ	NIH	N/A
Excel	Microsoft	N/A
MeV_4_7	MeV team	N/A
GLAD	(Hu et al., 2015)	N/A
Other
SMZ18	Nikon	N/A
Axioskop	Zeiss	N/A
TCS SP2 AOBS Confocal	Leica	N/A
CFX96 Real-Time System	Bio-Rad	N/A
Multi-sample tissuelyser-24	Shanghai Jingxin Technology	N/A

Highlights.

• yki^3SA tumors activate systemic MEK signaling to cause tumor growth and host wasting.

• Pharmaceutical MEK inhibition in host tissues alone alleviates wasting.

• yki^3SA tumors produce vn to autonomously promote MEK signaling and tumor growth.

• yki^3SA tumors produce Pvf1 to non-autonomously trigger host MEK signaling and wasting.