RESEARCH ARTICLE

Spz/Toll-6 signal guides organotropic metastasis in DrosophilaKetu Mishra-Gorur1,*,‡, Daming Li1,*, Xianjue Ma1,2,*, Yanki Yarman3, Lei Xue1,4 and Tian Xu1,2,‡

ABSTRACTTargeted cell migration plays important roles in developmentalbiology and disease processes, including in metastasis. Drosophilatumors exhibit traits characteristic of human cancers, providing apowerful model to study developmental and cancer biology. We nowfind that cells derived from Drosophila eye-disc tumors also displayorgan-specific metastasis, invading receptive organs but not wingdisc. Toll receptors are known to affect innate immunity and the tumorinflammatory microenvironment by modulating the NF-κB pathway.Our RNA interference (RNAi) screen and genetic analyses show thatToll-6 is required for migration and invasion of the tumor cells. Further,receptive organs express Toll ligands [Spatzle (Spz) familymolecules], and ectopic Spz expression renders the wing discreceptive to metastasis. Finally, Toll-6 promotes metastasis byactivating JNK signaling, a key regulator of cell migration. Hence,we report Toll-6 and Spz as a new pair of guidance moleculesmediating organ-specific metastatic behavior and highlight a novelsignaling mechanism for Toll-family receptors.

KEY WORDS: JNK, Tumor, Cancer, Toll-like receptors, Spatzle,Cell migration

INTRODUCTIONTargeted cell migration and invasion play important roles in avariety of biological and disease processes. This phenomenon isexemplified by organotropic metastasis in cancer progression.Tumor metastasis is the leading cause of death for cancer patients(Lambert et al., 2017). A fundamental feature of metastasis is theability of distinctive tumor types to colonize different organ sites,which depends on the inherent properties of the tumor cells and theirinteraction with the host tissue. However, the mechanism oforganotropic metastasis is not well delineated. Deciphering theunderlying molecular mechanism(s) of interactions between tumorcells and secondary host tissue is essential for our understanding ofmetastasis.Almost a century ago, two theories were stated to explain

organotropic tumor metastasis. James Ewing described the‘anatomical mechanical theory’ to account for cancer metastasis

(Ribatti et al., 2006). He theorized that the patterns of blood flowfrom the primary tumor can predict the first metastasized organs. Onthe other hand, Stephen Paget’s ‘seed and soil’ theory hypothesizedthat tumor cells migrate to tissues that support their growth (Fidler,2003). In other words, he suggested that the site of metastasisdepended on the affinity of the tumor for the microenvironment. Amodern version of the seed and soil theory focuses on a ‘homingmechanism’, which suggests that tumor cells are drawn to specificorgan sites because of complex signaling crosstalk between thetumor cells and the cells of the organ (Fidler, 2002). Consistent withthis idea, recent studies in mammalian systems indicate thatchemokines and their receptors are involved in organ-specifictumor metastasis (Takeuchi et al., 2007; Zlotnik, 2006).

The strength of forward genetic screens and mosaic analysis inDrosophila has empowered investigators to utilize this modelorganism for identifying genes of cancer relevance, defining cancersignaling pathways and deciphering cancer biology. We havepreviously performed a genetic screen and developed a Drosophilamodel for tumor metastasis (Pagliarini and Xu, 2003). Somatic cellsexpressing oncogenic Ras protein (RasV12) and simultaneouslycarrying a loss-of-function mutation in any of the cell polaritygenes, including scribbled (scrib), lethal giant larvae (lgl) or discslarge (dlg), develop into malignant tumors (known as RasV12/cell-polarity-defect tumors). This fly tumor model displays the mainhallmarks of human metastatic cancers, including uncontrolledgrowth, basement membrane (BM) degradation, loss of E-cadherin,migration, invasion, and secondary-tumor formation at distantorgans. Further studies have revealed that JNK (c-Jun N-terminalkinase) signaling is activated in the tumor cells and is required fortumor-cell migration and invasion (Ma et al., 2013, 2014; Igakiet al., 2006). We also learned that JNK signaling can be activatednon-autonomously or under stress conditions and can propagate tocollaborate with RasV12-expressing cells to induce tumorigenesis(Wu et al., 2010).

In this study, we have conducted a pathological and chronologicalexamination of tumor progression and invasion in the fly tumormodel. We have discovered that tumor cells metastasize to distalorgans in a tissue-specific pattern. To identify genes responsible fororganotropic metastasis, we successfully established fly tumor celllines from the imaginal disc of a single larva and performed agenome-wide RNA interference (RNAi) screen. We identified Toll-6, a Toll-receptor family member, as a crucial gene for tumor cellmigration. We also show that Toll-6 is required in tumor cells fororgan-specific metastasis in vivo by inducing JNK signalingactivation. Finally, we have discovered that the expression of theSpätzle (Spz) ligands in targeted organs serves as a cue for theguided migration and invasion. The Spz/Toll-6 system provides anovel molecular mechanism for organotropic metastasis.

RESULTSOrgan-specific metastasis of epithelial tumors in DrosophilaWe have previously shown that, in Drosophila eye-antennal discs,oncogenic Ras (RasV12) or Raf (Rafgof ) can cooperate with mutantsReceived 18 March 2019; Accepted 20 August 2019

1Howard Hughes Medical Institute, Department of Genetics, Yale University Schoolof Medicine, New Haven, CT 06519, USA. 2School of Life Sciences, WestlakeUniversity, 18 Shilongshan Road, Hangzhou 310024, China. 3Department ofNeurosurgery, Yale University School of Medicine, New Haven, CT 06519, USA.4Shanghai Key Laboratory for Signaling and Diseases, School of Life Science andTechnology, Tongji University, Shanghai 200092, China.*These authors contributed equally to this work

‡Authors for correspondence (ketu.mishra@yale.edu; tian.xu@yale.edu)

T.X., 0000-0002-2160-0027

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2019. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

that disrupt cell polarity, including scrib, dlg or lgl, to induce tumorinvasion and metastasis (Pagliarini and Xu, 2003). These fly tumorsdisplay uncontrolled proliferation, BM degradation, migration,invasion and secondary-tumor formation, which are characteristicof human tumors (Igaki et al., 2006). To further utilize Drosophilaas a model system to explore the underlying in vivo mechanism oftumor metastasis, we performed a detailed study of the pathology ofepithelial tumor dissemination and investigated the spatial andtemporal pattern of tumor cell migration and invasion using macro-and microscopic methods. Previously, we have shown that thesetumors metastasize at around day 7 after egg laying and progressivelyinvade other tissues until larval death at approximately day 15(Pagliarini and Xu, 2003). We discovered that the appearance ofsecondary tumors gradually radiates away from the primary tumors inthe eye-antennal disc, suggesting a more directed migration ratherthan the hemolymph being the main route for metastasis. Hence, weexamined early-stage larvae and progressively followed theirmetastasis pattern from day 7 to day 12. We found that tumor cellsinitially migrate anteriorly towards the mouth hooks and posteriorlytowards the ventral nerve cord (VNC), both of which are physicallyattached to the eye-antennal disc, the host tissue of the primary tumors(Fig. 1A,B; Figs S1A,B,J and S2A,B). Strikingly, we also observedan organotropic tumor metastatic behavior. The tumor cells neverinvade or migrate onto the wing disc (Fig. 1K; Figs S1G and S2I)while readily metastasizing onto other tissues, including the VNC,mouth hooks, salivary glands (SGs), leg and haltere discs, gut, and fatbody (FB) (Fig. 1A-J,L; Figs S1A-S‴ and S2C-H).To dissect the molecular events underlying this organotropic

tumor metastatic behavior observed in vivo, we developed anin vitro culture system. We established cell lines of various

genotypes from Drosophila epithelial tumors derived from eitherthe eye- or wing-disc and optimized their culture conditions(Fig. 2A-D, Fig. S3A-F). The tumor cell lines were derived fromsingle larvae, thus minimizing genotypic variability (see Materialsand Methods). Karyotype analysis revealed that, in contrast toSchneider 2 (S2) and Kc cells, which showed 100% penetrance ofchromosomal aneuploidy (Fig. 2E,F), the karyotype of tumor celllines remains largely normal at passages as late as P60 (Fig. 2G-L,Fig. S4), but extensive culture results in partial chromosomalabnormalities (Fig. 2M, Fig. S4). These tumor cell cultures with auniform genetic background were then used to establish clonalRasV12/scrib−/− mutant cell lines by fluorescence-activated cellsorting (FACS) and dilution methods (see Materials and Methods).Twelve percent of the single cells from FACS sorting (48/400) weresuccessfully cloned (Fig. S3G,H) and tested for ease of applicationto various genetic and biological manipulations. Additionally, thecells are easily transfectable and phenotypically display invasivebehavior both in vitro and in vivo (Figs S5, S6, S7), highlighting theversatile potential of these fly tumor cell lines.

Spz/Toll-6 axis regulates organotropic metastatic behaviorWe then performed a genome-wide RNAi screen using theseRasV12/scrib−/− tumor cells to identify the potential receptorresponsible for organ-specific tumor cell migration (Echeverri andPerrimon, 2006). In the scratch assay (Hall, 2005), the fly tumorcells were found to be highly migratory, capable of covering theentire scratch areawithin 22 h (Fig. 2N). Multiple core componentsof the JNK pathway were identified in this screen, including eiger(egr), dTAK1, hemipterous (hep) and basket (bsk) (Fig. 2P-S), inaccordance with their roles in promoting tumor metastasis (Igaki

Fig. 1. Organotropic metastasisbehavior in Drosophila neuro-epithelial tumors. (A-K) GFP-labeledRasV12/lgl−/− clones were generated ineye discs. The ensuing clones developinto tumors and metastasize onto theventral nerve cord (VNC; A), mouth hooks(B), salivary glands (SGs; C), first leg disc(D), second leg disc (E), third leg disc andhaltere disc (F), gut (G), fat body (FB; H),trachea (I) and skin (J). Note that tumorcells do not metastasize onto the wingdisc (K). Numbers at bottom left of eachpanel indicate number of larvaedisplaying metastasis to a particularorgan out of the total number of larvaeexamined. (L) A schematicrepresentation of the organotropicmetastasis pattern in Drosophila. Scalebars: 50 μm.

2

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

et al., 2006). One of the remaining candidates is Toll-6, a memberof the Toll-family receptors (Tauszig et al., 2000), as knockdownof Toll-6 dramatically blocked tumor cell migration (Fig. 2T,T′).Toll-6 expression is detected at both the RNA and protein level

in RasV12/scrib−/− as well as RasV12/lgl−/− tumors (Fig. 3A,Fig. S8A-E′). Consistent with our in vitro data, while having aminor effect on tumor growth, expression of the Toll-6 dsRNA(RNAi) in the tumor-bearing flies dramatically blocks metastasis

Fig. 2. Establishment and characterization of fly tumor cell lines and RNAi screen. (A,B) RasV12/scrib−/− tumor cell lines at low (A) and high (B)magnification. (C,D) Fluorescent (C) and bright-field (D) image of RasV12/scrib−/− tumor cells. (E-M) Karyotyping of various cultured cell lines. All (100%) S2 (E)and Kc (F) cells exhibit abnormal chromosomes (Abnl). RasV12/scrib−/− female cells and male cells possess a normal karyotype (Nml) at early passages (G-L)and abnormal karyotype after a long period in culture (P141: M, tetraploidy except for third chromosome). We noticed that the female-derived cells are more proneto losing one of the X chromosomes (data not shown). (P: passage number; n: number of cells examined; ‘F’: females; ‘M’: males). (N-T′) RasV12/scrib−/− cellswere used in an in vitro scratch assay to perform a genome-wide screen. Negative control: lacZ RNAi (N,N′); positive control: Thread (Diap1) RNAi (O,O′).Knockdown of Eiger (P), dTAK1 (Q) and Hep (R) partially blocked migration of cells, whereas knockdown of Bsk (S) and Toll-6 (T,T′) dramatically blocked tumorcell migration. Yellow boxes indicate area of higher magnification corresponding to dashed boxes.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

and tumor-induced BM degradation (Fig. 3B-E′). Furthermore, likethe RasV12/scrib−/− tumors, the RasV12/lgl−/− tumors also grow andencounter other organs. However, unlike the invasive tumors,

Toll-6-IR (Toll-6 RNAi) tumors do not invade these organs,which can be easily and cleanly separated away from the tumortissue (Fig. S8F-F″).

Fig. 3. See next page for legend.

4

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

Spz serves as the ligand for Toll receptors and activatesdownstream signaling events (Lewis et al., 2013; Parthier et al.,2014; Gangloff et al., 2008). Since Toll-6 is required in the tumorcells for metastasis, we asked whether Spz and related molecules areexpressed on the target organs and serve as an attractant for thetumor cells. Several larval tissues are known to express spz mRNA,including the FB, SG and hemocytes (Irving et al., 2005; Meyeret al., 2014; Shia et al., 2009). In addition to Spz, there are five Spz-related genes in the Drosophila genome (Parker et al., 2001) thatcould encode potential ligands for the Toll-family receptors. Weperformed reverse-transcription PCR (RT-PCR) and found thatseveral Spz-related genes are expressed in metastasis-receptiveorgans, including VNC, mouth hook, SG, leg disc, FB and gut, butnot in the non-receptive wing discs (Fig. S9A).Next, to test whether Spz could serve as an attractant for Toll-6-

mediated metastasis, we developed an in vitro culture system,similar to the one used for studying neuronal migration (Jia et al.,2005). Briefly, freshly dissected individual organs were placedadjacent to tumor cell lines in culture medium, and the migrationprocess of tumor cells towards the specific organ was examined(Fig. 3F). Using this assay, we can recapitulate the organ-specificmetastasis pattern observed in vivo. Within 24 h, we can clearlydetect migration of the tumor cells towards the VNC (Fig. 3H),which is distinctly different from the random migration of the tumorcells in culture (Fig. 3G). Similar migration and invasion patternsare observed for all the organs that are invaded by tumor cells invivo, including leg disc, haltere disc, FB, trachea and SG (Fig. 3J,J′and data not shown). In sharp contrast to the migration of cellstowards receptive organs like the VNC, when the wing disc wasexamined in this assay we found no directed migration of tumorcells towards the disc (Fig. 3I). In fact, when both leg and wing discsare placed together in the same assay, tumor cells preferentiallymigrate towards the leg disc (Fig. 3J,J′). This suggests that thedissemination of the RasV12/cell-polarity-defect tumors inDrosophila is an active migratory process, which exhibits organ-specific metastatic behavior (Fig. 1L). To further investigatewhether Spz serves as a guide for Toll-6-mediated organotropicmetastasis, we overexpressed an activated form of Spz (SpzACT) inthe wing disc under the apterous (ap) promoter (Ligoxygakis et al.,2002) and used it in our in vitro culture system. In contrast to wild-type wing disc, tumor cells migrated towards the SpzACT-expressingwing disc and invaded it within 24 h (Fig. 3K). When the SpzACT-

expressing wing disc was placed on a lawn of semi-confluent tumorcells, the cells immediately adjacent to the receptive tissue neverdisplayed random migration. Instead, they migrated onto the discand accumulated on the tissue (Fig. S9B′, Fig. S10). Interestingly,in the case of the control wing disc, the cells migrated ontothe attached tissue (nerves, tracheal fibers) but not onto the discsurface (Fig. S9B). This observation supports the role of Spz as achemoattractant. To further test whether Toll-6 was serving as thecorresponding receptor in the tumor cells responding to SpzACT,we used RNAi to downregulate Toll-6 expression in the cellsand assessed their migration towards SpzACT-expressing wingdiscs. Compared with cells treated with lacZ dsRNA for RNAi(Fig. 3L-N), tumor cells treated with Toll-6 dsRNA displayed adramatically reduced migration towards both the VNC as well asSpzACT-expressing wing disc (Fig. 3O-Q; for quantification of cellmigration see Fig. S9C,D). Together, these data indicate that theSpz/Toll-6 axis serves as a molecular apparatus mediating organ-specific metastasis inDrosophila and strongly suggest the existenceof underlying molecular machinery guiding the targeted migrationand invasion of tumor cells.

Toll-6 activates JNK signalingGiven that JNK signaling is required for cell migration and itsinhibition blocks metastasis (Igaki et al., 2006, 2009), we reasonedthat Toll-6 could mediate metastasis by regulating JNK activation.Indeed, knockdown of Toll-6 significantly reduces JNK activationas revealed by phospho-JNK (p-JNK) immunostaining and westernblotting (Fig. 4A,B; Fig. S11A). In addition, ectopic expression ofactivated Toll-6 (Toll-6ACT) in the wing disc results in JNKactivation (Fig. 4C-D″). Hence, to test whether Spz could triggerToll-6-mediated JNK activation, we ectopically expressed SpzACT

and wild-type Toll-6 (Toll-6WT) in wing discs under ptc-Gal4.Although expression of SpzACT or Toll-6WT alone induces no ormild JNK activation, respectively (Fig. 4E-F″), co-expression ofboth strongly induces JNK activation (Fig. 4G-G″), suggesting thatSpz can serve as a ligand for Toll-6 to activate JNK signaling.Furthermore, in accordancewith the observation that JNK activationis associated with endocytic vesicles (Igaki et al., 2009), doublestaining reveals that Toll-6ACT colocalizes with p-JNK (Fig. S11B-D).Interestingly, a closer examination also revealed that co-expression ofSpzACT and Toll-6WT synergistically induces cell invasion behavior,as GFP-positive cells collectively migrate away from the anterior-posterior boundary and towards the posterior part of the wing disc(Fig. S12A-C″), phenocopying JNK activation, as has beenreported previously (Ma et al., 2014, 2017). Finally, it has beenreported that, although there is a certain redundancy in the ligandsthat activate Toll-6, Spz5 is the main ligand for Toll-6 (McIlroyet al., 2013). Since we detected that SpzACT can activate JNK viaToll-6, we then also tested the ability of Spz5 to activate Toll-6,and hence JNK, in vivo. We find that, although Spz5overexpression is sufficient to induce a slight JNK activation(Fig. S12D″) (Foldi et al., 2017), it can significantly increasewild-type Toll-6-mediated JNK activation and cell migration(Fig. S12A-E″). Taken together, these data demonstrate that Toll-6activates JNK signaling in vivo.

Next, to further delineate the action of Toll-6 in the JNK pathway,we performed genetic analysis between Toll-6 and the knowncomponents of the JNK signaling pathway. We overexpressedactivated Toll-6 in the developing eye using the GMR-Gal4 driver(GMR>Toll-6ACT), which produces a small-eye phenotype (Fig. 4H,I). This resembles the effect of expressing a constitutively activeform of Hep (HepCA), a JNK kinase (Fig. 4P), consistent with the

Fig. 3. Toll-6 mediates migration of tumor cells in vivo and in vitro.(A) RT-qPCR to determine expression of Toll-6 in RasV12/scrib−/−- andRasV12/lgl−/−-induced tumors. RP49 is used as an internal control. (B-E)RasV12/lgl−/−-induced tumor invasion (B,B′) and BM degradation (C,C′) areblocked by Toll-6 RNAi (D-E′); C,C′,E,E′ show margin of tumor stainedfor LaminA/C (red). Arrows indicate BM degradation. (F) Schematicrepresentation of a test plate for in vitro migration assay: a confluent lawn ofRasV12/scrib−/− cells in a 35 mm tissue-culture dish is scratched, twowild-type(WT) wing discs or wing discs expressing SpzACT are placed in the same dish.(G-K) Within 24 h, tumor cells migrate towards the VNC (H) or wing discexpressing SpzACT (K), but not WT wing disc (I). All tissues (controls and test)were placed in the same dish to minimize variability and each experimentrepeated at least three times; a representative experiment is shown. Whenleg disc was placed adjacent to a wing disc, tumor cells selectively migrateto the leg disc (J,J′). Yellow panel: higher magnification of boxed area.(L-Q) Knockdown of Toll-6 but not lacZ in RasV12/lgl−/− tumor cells blockedin vitro organotropic metastasis towards VNC (compare M and P) or SpzACT-expressed wing discs (compare N and Q). (G-Q) Arrowhead (pink) indicatesedge of wound/scratch. Arrow in G indicates the direction of migration. Cellsthat migrated beyond 200 µm (indicated by the length of the arrow) werecounted and graphed (see Fig. S8). Scale bars: B,B′,D,D′: 200 µm; C,C′,E,E′:50 µm; J′: 200 µm; G-Q: arrow equals 200 µm.

5

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

notion that Toll-6 activates JNK signaling. The Toll-6ACT eyephenotype is suppressed by knocking down Hep (Fig. 4K), byexpression of a dominant-negative form of bsk (Drosophila JNK;Fig. 4L) or by expression of JNK inhibitor Puckered (Puc; Fig. 4M).However, it is not affected by inactivation of an upstreamcomponent of the pathway, dTAK1 (Fig. 4J). Consistently,knockdown of Toll-6 suppresses the small-eye phenotype inducedby ectopic expression of Egr or dTAK1, but not that of HepCA

(Fig. 4N-S). Collectively, these genetic data suggest that Toll-6 actsin the JNK signaling pathway upstream of Hep and Bsk.

DISCUSSIONA Drosophila model for organotropic metastasisMetastasis is the leading cause of mortality in cancer patients.Befittingly, both the words ‘cancer’ (Latin: ‘crab’) and ‘metastasis’(Greek: ‘displacement’) refer to cell movement: the crab-like

Fig. 4. Toll-6 activates JNK signaling. (A-B′′) Knockdown of Toll-6 reduces RasV12/lgl−/− tumor-induced JNK activation, as shown by phosphorylatedJNK (p-JNK) staining. (C-G″) p-JNK staining of wing discs under patched promoter. Compared with control (C-C″), ectopic expression of activatedToll-6 (Toll-6ACT) activates JNK (D-D″). (E-G) Co-expression of SpzACT strongly increased wild-type Toll-6-induced JNK activation (F-G″), whereas expression ofSpzACT alone is not sufficient to activate JNK (E-E″). (H-S) Light micrographs of Drosophila adult eyes are shown. Compared with the control (H), ectopicexpression of activated Toll-6 under the GMR promoter induces a small-eye phenotype (I), which is suppressed by expression of a Hep dsRNA (K), a dominant-negative form of Bsk (L), or Puc (M), but not by that of a dominant-negative form of dTAK1 (J). Knockdown of Toll-6 suppresses ectopic Egr- or dTAK1-expression-induced small-eye phenotypes (N,O,Q,R) but not that of HepCA expression (P,S). (T) A working model of Spz/Toll-6 in guiding organotropic metastasis inDrosophila.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

invasion of cancer into healthy tissue and the migration of cancercells to secondary sites (Murphy, 2001). Since Paget’s initialobservation more than 100 years ago, pathologists have recognizedthat the movement of cancer cells is not random and that differenttypes of cancer have different destinations or organ-specificmetastasis (Murphy, 2001). For example, colon carcinomasusually metastasize to liver and lung but rarely to bone, skin orbrain and almost never to kidneys, intestine or muscle (Obenauf andMassagué, 2015). In contrast, other tumors, such as breastcarcinomas, frequently form metastases in most of these organs,while prostate cancer metastasis occurs most prominently in bone(Ribatti et al., 2006). Here, we report a similar phenomenon in thespread of malignant tumors in Drosophila. The RasV12/cell-polarity-defect tumors exhibit organ-specific metastasis. Tumorcells originating from the eye-antennal imaginal disc in the larvaemetastasize to almost all organs (mouth hook, VNC, SG, leg discs,haltere disc, gut, FB) except the wing disc.In mouse cancer models and human patients, it is known that

tumor cells are disseminated through vascular networks such as theblood and lymphatic vessels (Karaman and Detmar, 2014). Thetracheal system in Drosophila is a tubular network for supplyingoxygen, functioning as the equivalent of human lungs and bloodvessels. Interestingly, in flies, we also observed that tumor cells fromprimary tumors can metastasize to the trachea, suggesting that thetrachea might function as an essential media to facilitate tumormetastasis. Consistent with this hypothesis, a recent study by RossCagan and Benjamin Levine showed that trachea-derived tumor cellscan migrate significant distances (Levine and Cagan, 2016).Additionally, RasV12/cell-polarity-defect tumors have been shownto express tracheal markers and migrate along the trachea (Grifoniet al., 2015). It will be interesting to further study the role of thetrachea in organotropic metastasis, and possibly investigate whethera similar conserved mechanism exists in mammalian tumor cellmigration.

Spz/Toll-6: a new nexus for guiding targeted migration oftumor cellsTargeted cell migration plays a key role in normal development.Studies of neural development have identified multiple receptorsand their ligands that regulate guided neuronal migration (e.g. Robo/Slit, Trk receptors/neurotrophins) (Andrews et al., 2007; Puehringeret al., 2013). For organotropic metastasis, Paget proposed the ‘seedand soil’ theory, which states that metastases develop only when the‘seed’ (tumor cells) and the ‘soil’ (target organs) are compatible(Paget, 1989; Fidler et al., 2002). It is only recently that studies havebegun to reveal some of the identities of the ‘seed and soil’molecules that regulate the spread of tumor cells. Experiments withbreast cancer cell lines showed that inhibition of the chemokinereceptor CXCR4 by a neutralizing antibody abrogates theirmetastasis to the lung, which expresses the correspondingchemokine ligand, CXCL12 (Muller et al., 2001). However, theCXCR4/CXCL12 interaction is unlikely involved in the metastasisof breast cancers to the liver (Wang et al., 2006). The diverse typesof cancers and their complex metastatic patterns indicate that other‘seed and soil’ molecules must be involved in organ-specificmetastasis.Here, we report that Toll-6, a Toll-family neurotrophin receptor,

plays an essential role in tumor cell metastasis in Drosophila. Ourstudy shows that Toll-6 is expressed in the RasV12/cell-polarity-defect tumors and that downregulation of Toll-6 in the tumor cellsblocks their migration and invasion. Interestingly, we also showedthat Toll-6-expressing tumor cells migrate towards and invade the

organs that express Spz or Spz-related molecules. Furthermore, theToll-6-expressing tumor cells do not metastasize to thewing disc, anorgan with no detectable Spz or Spz-related gene expression in oursystem, strongly arguing that Spz or a Spz-related molecule could beserving as a cue for guiding Toll-6-expressing tumor cells.Interestingly, a recent study by Alpar et al. also shows that thewing disc produces extremely low levels of Spz in a tightlyregulated spatially restricted pattern with no resultant signalingevents (Alpar et al., 2018). Indeed, artificial overexpression ofSpzACT in the wing disc converts it to a tissue receptive for themigration and invasion by Toll-6-expressing tumor cells (Fig. S9B′).Together, these data indicate that Spz/Toll-6 serve as the ‘seed andsoil’ molecules for organotropic metastasis in the fly. In support ofour conclusion, clinical studies have revealed a correlation betweenincreased expression levels of Toll-like receptors (TLRs) andmalignancy of multiple cancers (O’Neill, 2008; Zhang et al.,2009). TLRs have also been shown to affect malignancy by alteringthe tumor inflammatory microenvironment (Fidler et al., 2007; Kimand Karin, 2011; Kim et al., 2009; Rakoff-Nahoum and Medzhitov,2009). Given that the Toll-family receptors are evolutionarilyconserved, consisting of nine members in Drosophila and ten TLRsin mammals (Yagi et al., 2010; Imler and Zheng, 2004), our studyraises the possibility that mammalian TLRs could play a similar rolein mediating organotropic metastasis and cell migration.

In Drosophila, Toll regulates dorsoventral patterning in embryosand anti-fungal defense in adults (Anderson et al., 1985; Lemaitreet al., 1996). Toll has also been found to have an inhibitory role inthe formation of neuromuscular junctions (Halfon et al., 1995; Roseet al., 1997) and, recently, 18 wheeler, a Toll-like receptor protein,was found to play a role in border cell migration (Kleve et al., 2006).Studies in both flies and mammals show that Toll-family receptorsmediate NF-κB signaling activation (Rakoff-Nahoum andMedzhitov, 2009; Brikos and O’Neill, 2008; Valanne et al.,2011). A study using biochemical inhibitors suggests that Tollmolecules could affect random migration of neutrophils byactivating ERKs (Aomatsu et al., 2008). Our study shows thatToll-6 activates JNK signaling, which is in accordance with therecent study by Foldi et al. that shows the role of Toll-6 and JNK incell death (Foldi et al., 2017). However, JNK signaling is also a keyregulator for cell migration during development (Davis, 2000), andwe have previously shown that JNK signaling is activated in RasV12/cell-polarity-defect tumors and is essential for metastasis (Igakiet al., 2006). Here, we report that, first, Toll-6 knockdown in RasV12/cell-polarity-defect tumors completely blocks metastasis byeffectively reducing JNK signaling; second, the role of JNKactivation by Toll-6 is highlighted in the in vitro assay, as Toll-6knockdown reduced migration of the tumor cells even when discswere not placed (Fig. 3, compare panels K and N; Fig. S9D),suggesting that JNK activation by Toll-6 might be important forgeneral cell migratory behavior. Finally, ectopic expression of Toll-6ACT in the wing disc results in JNK activation (Fig. 4D-D″). Thesedata indicate that Toll-6 regulates metastasis by activating JNKsignaling. Recently, Grindelwald (Grnd) has been identified as anovel tumor necrosis factor receptor (TNFR) mediating RasV12/scrib−/−-induced tumor growth and invasion (Andersen et al., 2015).Our data here suggest that Toll-6 could be providing a second signalfor the activation of JNK. Indeed, our epistasis data support the modelthat Toll-6 genetically interacts with the JNK pathway (Fig. 4H-S). Itis possible that inputs from both Toll-6 and Grnd signals could resultin or are required for a high level of JNK activation.

In addition to Spz, there are five Spz-related genes in theDrosophila genome, which could serve as ligands for Toll-6. Toll-6

7

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

has been recently reported to function as a neurotrophin receptor inregulating motor neuron targeting and survival, and it alsophysically binds to Spz5 (McIlroy et al., 2013; Foldi et al., 2017).Consistent with this, we found that co-expression of Toll-6 and Spz5synergistically promote collective cell invasion in the developingwing, phenocopying Toll-6 and SpzACT co-expression (Fig. S12).Interestingly, although both SpzACT and Spz5 can activate Toll-6-mediated JNK signaling, we found that, unlike Spz5, expression ofSpzACT alone is not sufficient to activate JNK (Fig. S12A-A″, D-D″).Given our results, we infer that Spz and Spz5 might display someredundancy. Indeed, a previous study by Foldi et al. has demonstratedthe redundancy of Spz2 (DNT1) and Spz5 (DNT2) in binding toToll-6 (Foldi et al., 2017). Furthermore, a redundancy betweenSpz, Spz2 and Spz5 has also been demonstrated by Sutcliffe et al.(2013), suggesting that Spz proteins may function as promiscuousligands in some circumstances (e.g. upon overexpression) and canbind multiple Toll receptors. We believe that, under overexpressionconditions, Spz is capable of activating Toll-6 (albeit at lower levelsthan is Spz5, as shown in Fig. S12). Therefore, as has been reportedbefore (Sutcliffe et al., 2013), we also find that different Spz proteinsmight act redundantly to induce Toll-6-mediated JNK activation andcell migration. In conclusion, our genetic and biochemical data showthat Toll-6 and Spz compose a new pair of guidance molecules fordirecting cell migration and that their interaction mediatesorganotropic metastasis by activating JNK signaling.

MATERIALS AND METHODSFly strains and generation of clonesFluorescently labeled clones were produced in larval imaginal discs usingthe following strains: y, w, eyFLP1; Act>y+>Gal4, UAS–GFP; FRT82B,Tub-Gal80 (82B tester) and y, w, eyFLP1; Tub-Gal80, FRT40A;Act>y+>GAL4, UAS-GFP (40A tester). Additional strains used were asfollows: GMR-GAL4, ptc-GAL4, ap-GAL4, sev-GAL4, UAS-GFP andpucE69 ( puc-lacZ) were obtained from Bloomington Drosophila StockCenter.UAS-spzACTwas gift from J.-M. Reichart (Ligoxygakis et al., 2002).Five independent UAS-Toll-6-IR transgenic lines generated from threedifferent constructs were obtained fromViennaDrosophilaResource Center(VDRC). UAS-spz5HA was obtained from FlyORF. UAS-egr (Igaki et al.,2002), UAS-HepCA, UAS-dTAK1, UAS-hep-IR, UAS-bskDN, UAS-dTAK1DN

and UAS-puc (Ma et al., 2012) were previously described. UAS-TollWT andUAS-Toll-6ACT-Flag transgenic flies were generated by standard P-element-mediated transformation (Bestgene, Inc.). More than five independent lineswere produced and examined for each transgene. Two RNAi lines (v27102and v27103) were recombined and used to perform experiments unlessindicated. Gene expression was verified by immunostaining.

Genotypes for fly-related figuresThe genotypes of flies for results shown in the figures are as shown below.

Fig. 1: (A-K) y, w, ey-Flp/+; tub-Gal80, FRT40A/lgl4, FRT40A, UAS-RasV12; Act>y+>Gal4, UAS-GFP/+.

Fig. 3: (B,C) y, w, ey-Flp/+; tub-Gal80, FRT40A/lgl4, FRT40A, UAS-RasV12; Act>y+>Gal4, UAS-GFP/+; (D,E) y, w, ey-Flp/+; tub-Gal80,FRT40A/lgl4, FRT40A, UAS-RasV12; Act>y+>Gal4, UAS-GFP/UAS-Toll-6-IRV27102, V27103.

Fig. 4: (A-A″) y, w, ey-Flp/+; tub-Gal80, FRT40A/lgl4, FRT40A, UAS-RasV12; Act>y+>Gal4, UAS-GFP/+; (B-B″) y, w, ey-Flp/+; tub-Gal80,FRT40A/lgl4, FRT40A, UAS-RasV12; Act>y+>Gal4, UAS-GFP/UAS-Toll-6-IRV27102, V27103; (C) ptc-Gal4, UAS-GFP/+; (D) ptc-Gal4, UAS-GFP/+;UAS-Toll-6Act/+; (E) ptc-Gal4, UAS-GFP/+; UAS-spzAct/+; (F) ptc-Gal4,UAS-GFP/UAS-Toll-6WT; (G) ptc-Gal4, UAS-GFP/UAS-spzWT; UAS-Toll-6Act/+; (H) GMR-Gal4/+; (I) GMR-Gal4, UAS-Toll-6Act/+; (J) GMR-Gal4,UAS-Toll-6Act/UAS-dTAK1DN; (K) GMR-Gal4, UAS-Toll-6Act/UAS-hep-IR;(L)GMR-Gal4, UAS-Toll-6Act/UAS-bskDN; (M) GMR-Gal4, UAS-Toll-6Act/UAS-Puc; (N)UAS-Egr/+; GMR-Gal4/+; (O) sev-Gal4, UAS-dTAK1/+; (P)GMR-Gal4, UAS-HepCA/+; (Q) UAS-Egr/+; GMR-Gal4/UAS-Toll-6-

IRV27102, V27103; (R) sev-Gal4, UAS-dTAK1/+; UAS-Toll-6-IRV27102, V27103/+;(S) GMR-Gal4, UAS-HepCA/UAS-Toll-6-IRV27102, V27103.

Fig. S1: (A-I) y, w, ey-Flp1; Act>y+>Gal4, UAS-GFP/+; FRT82B, tub-Gal80/UAS-RafGOF, FRT82B, scrib1; (J-O) FRT19A, dlgm52/tub-Gal80,FRT19A; ey-Flp5, act>y+>Gal4, UAS-GFP/UAS-RasV12; (P-S‴) y, w, ey-Flp/+; tub-Gal80, FRT40A/lgl4, FRT40A, UAS-RasV12; Act>y+>Gal4,UAS-GFP/+.

Fig. S2: y, w, ey-Flp1; Act>y+>Gal4, UAS-GFP/UAS-RasV12; FRT82B,tub-Gal80/FRT82B, scrib1.

Fig. S3: (E) y, w, hs-Flp1; Act>y+>Gal4, UAS-GFP/UAS-RasV12;FRT82B, tub-Gal80/FRT82B, scrib1.

Fig. S8: (A-A″) ptc-Gal4, UAS-GFP/+; (B-B″) ptc-Gal4, UAS-GFP/UAS-spzWT; (C-C″) ptc-Gal4, UAS-GFP/UAS-spzWT; UAS-Toll-6-IRV27102, V27103;(D-E′) y, w, ey-Flp/+; tub-Gal80, FRT40A/lgl4, FRT40A, UAS-RasV12;Act>y+>Gal4, UAS-GFP/+; (F) y, w, ey-Flp/+; tub-Gal80, FRT40A/lgl4,FRT40A, UAS-RasV12; Act>y+>Gal4, UAS-GFP/UAS-Toll-6-IRV27102, V27103.

Fig. S11: (B) ptc-Gal4, UAS-GFP/UAS-Toll-6WT; (C,D) ptc-Gal4, UAS-GFP/+; UAS-Toll-6Act/+.

Fig. S12: (A-A″) ptc-Gal4, UAS-GFP/+; UAS-spzAct/pucE69; (B-B″) ptc-Gal4, UAS-GFP/UAS-Toll-6WT; pucE69/+; (C-C″) ptc-Gal4, UAS-GFP/UAS-Toll-6WT; UAS-spzAct/pucE69; (D-D″) ptc-Gal4, UAS-GFP/+; UAS-spz5/pucE69; (E-E″) ptc-Gal4, UAS-GFP/UAS-Toll-6WT; UAS-spz5/pucE69.

Establishing Drosophila tumor cell linesA single or 30 third-instar larvae were washed five times in a 10 cm Petridish with sterile 1× PBS. The larvae were transferred into an Eppendorftube, treated with four washes of 70% ethanol for 2 min followed by a briefvortexing, followed by rinsing twice with sterile 1× PBS. These wash stepswere critical for preventing contamination in primary cultures. Using pre-sterilized forceps, the larvae were dissected in basic M3 FBS-free medium(Sigma-Aldrich) supplemented with 100 U/ml of penicillin, 100 µg/ml ofstreptomycin and 400 µg/ml of G418 (Life Technologies). The eye or wingdiscs with tumors were cut with a dissecting needle into several piecesand kept in an Eppendorf tube containing 1 ml of basic M3 mediumwith 400 µg/ml of G418 (100 µl medium for single-animal experiments).The tissues were washed once with 1× PBS, changed to 1 ml 1× trypsin-EDTA, and incubated at 37°C for 10 min (250 µl for single-animalexperiments). After incubation, cells were drawn up and down by pipettingto dissociate cell clumps. Cells were spun at 1000 rpm (900 g) for 5min. Thesupernatant was removed, and the cells were resuspended in 1 ml cM3/FEmedium (cM3 plus 2.5% fly extract) with 400 µg/ml of G418, beforetransfer to a 12.5 cm2 T-flask with 2 ml of medium (for single-animalexperiments, 100 µl medium and a 96-well plate, respectively; forexperiments using cM3/Ins or cM3/FE/Ins media, 0.125 IU/ml insulin wasalso added). Following overnight incubation of the cells at 25°C, mediumwasreplacedwith freshmedium at the same volume. Cell lines were then passagedat the ratio of 1:2 to 1:3 at the early passage (passages 1-3) every 3-4 days, andlater the cells were passaged at the ratio of 1:3 to 1:6 every 3 days. For single-animal experiments, the cells were passaged every 3-4 days in the samemedium sequentially from 96- to 48- to 24- to 12-well plates, and finally to12.5 cm2 T-flasks. For amplifying cells for large-scale experiments, cM3/Extra FBS medium was used (cM3 plus 7.5% extra FBS, Invitrogen). Thesecell lines can be frozen in culture medium plus 10% DMSO in liquid N2,thawed and cultured again. Fly extract was prepared as previously reported(Currie et al., 1988), except that we homogenized 1 g of female flies in 10 mlof basic M3 serum-free medium.

Cloning of RasV12/scrib−/− cellsThe cells were trypsinized and large clumps broken by repeated pipettingfollowed by filtering through a 25-μm mesh (Falcon). The cells wereallowed to sit for 10 min to allow any clumps to settle and the supernatantcontaining mostly single cells was resuspended in 1× M3 containing 1%FCS. Single cells were collected by sterile FACS at the Yale FACS facility.The cells were then resuspended in cM3 medium and plated into 96-wellplates to obtain a single cell/well. Of the 400 cells that were plated, 48 weresuccessfully cloned. To select clones with a more spread-out monolayermorphology, 28 were examined and four were chosen.

8

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

KaryotypingKaryotype analysis followed the protocol of Cherbas and Cherbas (2007),except that we increased the vinblastine sulfate to 2 µg/ml and the cells weretreated with this drug for 3-4 h.

Transplantation assayThe cells were trypsinized, washed two times with sterile 1× PBS, andresuspended in 1× PBS with a concentration of approximately 5×107 cells/ml. Transplantation experiments were performed as previously described(Pagliarini and Xu, 2003).

RNAi treatment and screenRNAi treatment conditions were as previously described (Mihaylov et al.,2002; Baeg et al., 2005). For the RNAi screen, RasV12/scrib−/− cells (clone1) were amplified in cM3/Extra FCS medium. The screen was performed in96-well plates using the Drosophila dsRNA library (Ambion, Inc.). Cellswere trypsinized and resuspended in 1× M3 medium with no additives andan equal number of cells were added into each well containing dsRNA. Theplates were gently rocked for 1 h at room temperature. A total of 225 μl ofcM3/Extra FBS was then added per well. The plates were incubated at 25°Cfor 3 days. On day 3, a pipet tip was used to generate a scratch wound in thelawn of cells. The plates were incubated at 25°C for 22 h and the migrationof cells was monitored and imaged using a Zeiss microscope. The RNAiscreen was scored based on the extent of coverage of the scratch area, andonly genes that showed greater than 90% inhibition of migration were scoredas positive hits. The screen results were segregated according to proteinfunction.

In vitro Transwell invasion assayThe assay was carried out using BioCoat Matrigel Invasion Chambers (BDScientific). In total, 5×105 cells were resuspended in serum-freeM3mediumand grown in the Matrigel chamber. The Transwells contained cM3/FE/Insmedium as a chemoattractant. At 40 h post-incubation, the cells were stainedusing the HEMA 3 stain kit (Fisher Scientific).

In vitro migration assayFor the VNC migration assay, tumor cells were plated at near confluency in35 mm dishes in 1×M3 supplemented with 20% FCS and allowed to adhereovernight. Larval tissues were dissected from L3 larvae and stored in 1×M3complete medium on ice until use. A scratch was made in the cell monolayerand the medium aspirated from the plate. Larval tissues were then placed inthe scratch area, held in placewith a coverslip and fresh medium added to thecells. Control and test organs were placed in the same dish to minimizevariability. Each organ (control and test) was placed in duplicate. The platewas labeled to indicate which half had the controls and test discs. A grid wasdrawn at the bottom of the plate to serve as a guide for wounding themonolayer as well as to direct the placement of the discs/VNC. Drosophilaimaginal discs and organs can be maintained in vitro overnight and stillmaintain their characteristics, as has been previously reported (Schubigerand Truman, 2000); hence, the migration of cells was monitored after 24 hand imaged using a Zeiss microscope (5×, 10× and 20× objectives). Toll-6and lacZ dsRNAs were synthesized using the MEGAscript RNAi kit(Ambion, Inc.) and 25 µg of RNAi used in each 35 mm tissue culture dish.Cells were treated with RNAi in 1× M3 medium for 1 h and thensupplemented with 20% FCS. After 48 h, a scratch was created, and larvaltissues placed in the wound area. Migration of cells was assessed after 24 h,as described above. Cells that migrated beyond 200 µm from the edge of thescratch were counted and plotted for quantification. The experiments wererepeated at least three times. The 10× or 20× images were used to count cellsthat traversed beyond the 200 μm using Adobe Photoshop. Duplicate valuesfrom each experiment were averaged and the final three values used forstatistical analysis with GraphPad.

Plasmid constructionToll-6 was cloned into the EcoR1/Xba site of the pUAST vector while theactivated form of Toll-6 (Toll-6ACT) was cloned into the EcoRV site ofpUAST (Toll-6ACT was a kind gift from Dr J. Imler, Université deStrasbourg, CNRS, Strasbourg, France). Toll-6ACT is a recombinant chimera

in which the constitutively active extracellular domain of Toll is fused to thetransmembrane and intracytoplasmic domains of Toll-6 (Tauszig et al.,2000). Toll-6 cDNAwas obtained from theDrosophila Genomics ResourceCenter (DGRC). The restriction sites were introduced into the fragments byPCR amplification, sequenced and inserted into a pUAST vector. RNA fromdifferent larval organs was isolated using the Trizol reagent (Invitrogen) andRT-PCR was performed using routine protocols.

Real-time and quantitative real-time PCRThe reverse transcription of total RNA (1 µg) of Drosophila larval tissuesinto cDNA was performed using the Bio-Rad iScript cDNA Synthesis Kitfollowing the standard protocol of the manufacturer (Bio-Rad). RT-PCR forSpz-family ligands was performed using primers (Table 1) and conditionspreviously described (Parker et al., 2001). RP49 (reference gene) PCR wasperformed as previously described (Mishra-Gorur et al., 2002). (Larvaltissues were dissected with extreme care to eliminate contamination fromattached nerves, tracheal fibers or other tissues.)

Quantitative real-time PCR (qPCR) primers were designed with anamplicon size range of 110-160 bp with CG clamp at the 3′ end usingPrimer3 software (Untergasser et al., 2012; Koressaar and Remm, 2007) andtested for specificity by blasting the sequences against the Drosophilagenome on Ensembl BLAST. Drosophila RP49 was used as a referencegene. The qPCRwas performed in a total of 15 µl, containing 2 µl of cDNA,7.5 µl of Roche FastStart Universal SYBR Green master mix, and 300 nMof the forward and reverse primers. The reaction was carried out on a Bio-Rad CFX-384 real-time PCR system under the following conditions:activation of FastStart Taq DNA polymerase at 95°C for 10 min followed by40 cycles of 15 s denaturation at 95°C and 1 min annealing at 60°C. Afterthis run, melting-curve analysis was performed to identify primer dimers.Genomic DNA contamination and cross-contamination were checked byincluding a no reverse transcriptase (no-RT) control for each sample and ano-template control for every different gene analyzed during the preparationof samples. Each sample is run in triplicates for each gene to be assayed(Schmittgen and Livak, 2008).

Antibodies, histology and imagingA rabbit polyclonal antibody was generated against the intracellular domainof Toll-6. Two C-terminal peptides in Toll 6 (GC-SLNDDEDEDHDQQ-KNLWA and GC-TLEHQHHHNHQANRRSQH) were used for injectinginto rabbits (Cocalico, Inc.). The antibody was affinity purified and used at a1:50 dilution for antibody staining. The specificity of the anti-Toll-6antibody was confirmed by immunostaining of either S2 cells or wingimaginal discs expressing Toll-6. Larval tissues were stained with standardimmunohistochemical procedures using rabbit anti-MMP-1 antibody (1:40;DSHB) or rabbit anti-phospho-JNK polyclonal antibody (1:50; Calbiochem),mouse anti-beta-gal antibody (1:200; Sigma), secondary Alexa-Fluor-555- orAlexa-Fluor-488-conjugated antibodies (1:500 dilution for 90 min;

Table 1. Primers used in this study

Gene Primer sequence (5′-3′) Size (bp)

RP49 F: AGTATCTGATGCCCAACATCGR: TTCCGACCAGGTTACAAGAAC

300

Toll-6 F: AATATTGTGGAGTGCTCGGGR: GCGTTTAAGGCCACTGAAAG

204

spatzle F: GGATATCGCTGTTCAAAGTCCR: GAAGACAAAGGTATCCGATGG

119

spatzle 2 F: AAAAGAGGCGTGAAGACGAAGR: GGGATTACTGCATTTTTCCAATC

161

spatzle 3 F: AAGGCCAAGAGAATGATAAGTCGR: TGAGTTTCGGATAGGAGAAGAGC

123

spatzle 4 F: GGTTATCGGGTATTTGCTGATCCR: CACCACTTCATCCTTCACGGG

110

spatzle 5 F: CCGATAACTACCCCACGTATCTCR: ACAGGTTCGAGCAGGTGTTC

136

spatzle 6 F: GGTGGAGTATTTTGCACCTTATACAR: TACTCTACCTTGGCGGGACA

146

9

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

Invitrogen). Tissues were mounted in a drop of Vectashield-DAPI (VectorLaboratories, Burlingame, CA, USA). A Zeiss LSM510 Meta ConfocalMicroscope (Zeiss, Jena, Germany) was used for analysis. In all experiments,identical confocal settings were used for imaging when samples were beingcompared.

Western blotTumors were dissected and homogenized in Tris lysis buffer (50 mM Tris,pH 7.4; 150 mMNaCl; 1 mM EDTA; 1%NP-40). The samples were run on10% SDS-PAGE gels (Bio-Rad) and western blots were performedaccording to standard methods. In brief, lysates were separated on 10%SDS-PAGE gels and transferred to nitrocellulose membranes. Afterblocking with 5% milk, the membranes were incubated first withpolyclonal antibodies against phosphor-JNK (Calbiochem) or tubulin(DSHB), and then with a secondary HRP-conjugated anti-rabbit IgG(Jackson ImmunoResearch). Signals were detected with an ECL-plus kit(PerkinElmer).

AcknowledgementsWe thank J. Imler for the Toll-6ACT plasmid, X. J. Gao for cloning the UAS-Toll-6 andUAS-Toll-6ACT constructs, DGRC for the Toll-6 cDNA, VDRC for fly lines, and DSHBfor antibodies. We thank O. Henegariu and A. G. Ercan-Sencicek for lending theirexpertise for the qPCR experiments and analysis.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: K.M.-G.,T.X.;Methodology:K.M.-G.,D.L.;Validation:K.M.-G.,D.L.,X.M., L.X.; Investigation: K.M.-G., D.L., X.M., Y.Y., L.X.; Data curation: K.M.-G., D.L.,X.M.; Writing - original draft: K.M.-G., T.X.; Writing - review & editing: K.M.-G., X.M.,T.X.; Visualization: K.M.-G., D.L., X.M.; Supervision: T.X.; Funding acquisition: T.X.

FundingThis work is partly supported by aNational Institutes of Health (NIH)/National CancerInstitute (NCI) grant to T.X. (5R01CA069408-12), and grants from Shanghai PujiangProgram (09PJ1409400), the Innovation Program of Shanghai Municipal EducationCommission (10ZZ27), and Shanghai Committee of Science and Technology(09DZ2260100) to L.X.

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.039727.supplemental

ReferencesAlpar, L., Bergantin os, C. and Johnston, L. A. (2018). Spatially restrictedregulation of spatzle/toll signaling during cell competition. Dev. Cell 46,706-719.e5. doi:10.1016/j.devcel.2018.08.001

Andersen, D. S., Colombani, J., Palmerini, V., Chakrabandhu, K., Boone, E.,Rothlisberger, M., Toggweiler, J., Basler, K., Mapelli, M., Hueber, A.-O. et al.(2015). The Drosophila TNF receptor Grindelwald couples loss of cell polarity andneoplastic growth. Nature 522, 482-486. doi:10.1038/nature14298

Anderson, K. V., Bokla, L. and Nusslein-Volhard, C. (1985). Establishment ofdorsal-ventral polarity in the Drosophila embryo: the induction of polarity by theToll gene product. Cell 42, 791-798. doi:10.1016/0092-8674(85)90275-2

Andrews, W. D., Barber, M. and Parnavelas, J. G. (2007). Slit-Robo interactionsduring cortical development. J. Anat. 211, 188-198. doi:10.1111/j.1469-7580.2007.00750.x

Aomatsu, K., Kato, T., Fujita, H., Hato, F., Oshitani, N., Kamata, N., Tamura, T.,Arakawa, T. andKitagawa, S. (2008). Toll-like receptor agonists stimulate humanneutrophil migration via activation of mitogen-activated protein kinases.Immunology 123, 171-180. doi:10.1111/j.1365-2567.2007.02684.x

Baeg, G.-H., Zhou, R. and Perrimon, N. (2005). Genome-wide RNAi analysis ofJAK/STAT signaling components in Drosophila. Genes Dev. 19, 1861-1870.doi:10.1101/gad.1320705

Brikos, C. and O’Neill, L. A. J. (2008). Signalling of Toll-Like Receptors. In: Toll-Like Receptors (TLRs) and Innate Immunity. Bauer S., Hartmann G. (eds).Handbook of Experimental Pharmacology, vol. 183, pp. 21-50. Springer, Berlin,Heidelberg. doi:10.1007/978-3-540-72167-3_2

Cherbas, L. and Cherbas, P. (2007). Drosophila cell culture and transformation.CSH Protoc. 2007, pdb top6. doi:10.1101/pdb.top6

Currie, D. A., Milner, M. J. and Evans, C. W. (1988). The growth and differentiationin vitro of leg and wing imaginal disc cells from Drosophila melanogaster.Development 102, 805-814.

Davis, R. J. (2000). Signal transduction by the JNK group of MAP kinases.Cell 103,239-252. doi:10.1016/S0092-8674(00)00116-1

Echeverri, C. J. and Perrimon, N. (2006). High-throughput RNAi screening incultured cells: a user’s guide. Nat. Rev. Genet. 7, 373-384. doi:10.1038/nrg1836

Fidler, I. J. (2002). The organ microenvironment and cancer metastasis.Differentiation 70, 498-505. doi:10.1046/j.1432-0436.2002.700904.x

Fidler, I. J. (2003). The pathogenesis of cancer metastasis: the ‘seed and soil’hypothesis revisited. Nat. Rev. Cancer 3, 453-458. doi:10.1038/nrc1098

Fidler, I. J., Kim, S.-J. and Langley, R. R. (2007). The role of the organmicroenvironment in the biology and therapy of cancer metastasis. J. Cell.Biochem. 101, 927-936. doi:10.1002/jcb.21148

Fidler, I. J., Yano, S., Zhang, R.-D., Fujimaki, T. and Bucana, C. D. (2002). Theseed and soil hypothesis: vascularisation and brain metastases. Lancet Oncol. 3,53-57. doi:10.1016/S1470-2045(01)00622-2

Foldi, I., Anthoney, N., Harrison, N., Gangloff, M., Verstak, B., Nallasivan, M. P.,Alahmed, S., Zhu, B., Phizacklea, M., Losada-Perez, M. et al. (2017). Three-tierregulation of cell number plasticity by neurotrophins and Tolls in Drosophila. J. CellBiol. 216, 1421-1438. doi:10.1083/jcb.201607098

Gangloff, M., Murali, A., Xiong, J., Arnot, C. J., Weber, A. N., Sandercock, A. M.,Robinson, C. V., Sarisky, R., Holzenburg, A., Kao, C. et al. (2008). Structuralinsight into the mechanism of activation of the Toll receptor by the dimeric ligandSpatzle. J. Biol. Chem. 283, 14629-14635. doi:10.1074/jbc.M800112200

Grifoni, D., Sollazzo, M., Fontana, E., Froldi, F. and Pession, A. (2015). Multiplestrategies of oxygen supply in Drosophila malignancies identify tracheogenesis asa novel cancer hallmark. Sci. Rep. 5, 9061. doi:10.1038/srep09061

Halfon, M. S., Hashimoto, C. and Keshishian, H. (1995). The Drosophila toll genefunctions zygotically and is necessary for proper motoneuron and muscledevelopment. Dev. Biol. 169, 151-167. doi:10.1006/dbio.1995.1134

Hall, A. (2005). Rho GTPases and the control of cell behaviour. Biochem. Soc.Trans. 33, 891-895. doi:10.1042/BST0330891

Igaki, T., Kanda, H., Yamamoto-Goto, Y., Kanuka, H., Kuranaga, E., Aigaki, T.andMiura, M. (2002). Eiger, a TNF superfamily ligand that triggers the DrosophilaJNK pathway. EMBO J. 21, 3009-3018. doi:10.1093/emboj/cdf306

Igaki, T., Pagliarini, R. A. and Xu, T. (2006). Loss of cell polarity drives tumor growthand invasion through JNK activation in Drosophila. Curr. Biol. 16, 1139-1146.doi:10.1016/j.cub.2006.04.042

Igaki, T., Pastor-Pareja, J. C., Aonuma, H., Miura, M. and Xu, T. (2009). Intrinsictumor suppression and epithelial maintenance by endocytic activation of Eiger/TNF signaling in Drosophila. Dev. Cell 16, 458-465. doi:10.1016/j.devcel.2009.01.002

Imler, J.-L. and Zheng, L. (2004). Biology of Toll receptors: lessons from insectsand mammals. J. Leukoc. Biol. 75, 18-26. doi:10.1189/jlb.0403160

Irving, P., Ubeda, J.-M., Doucet, D., Troxler, L., Lagueux, M., Zachary, D.,Hoffmann, J. A., Hetru, C. and Meister, M. (2005). New insights into Drosophilalarval haemocyte functions through genome-wide analysis. Cell. Microbiol. 7,335-350. doi:10.1111/j.1462-5822.2004.00462.x

Jia, L., Cheng, L. and Raper, J. (2005). Slit/Robo signaling is necessary to confineearly neural crest cells to the ventral migratory pathway in the trunk.Dev. Biol. 282,411-421. doi:10.1016/j.ydbio.2005.03.021

Karaman, S. and Detmar, M. (2014). Mechanisms of lymphatic metastasis. J. Clin.Invest. 124, 922-928. doi:10.1172/JCI71606

Kim, S. and Karin, M. (2011). Role of TLR2-dependent inflammation in metastaticprogression. Ann. N. Y. Acad. Sci. 1217, 191-206. doi:10.1111/j.1749-6632.2010.05882.x

Kim, S., Takahashi, H., Lin, W.-W., Descargues, P., Grivennikov, S., Kim, Y.,Luo, J.-L. and Karin, M. (2009). Carcinoma-produced factors activate myeloidcells through TLR2 to stimulate metastasis. Nature 457, 102-106. doi:10.1038/nature07623

Kleve, C. D., Siler, D. A., Syed, S. K. and Eldon, E. D. (2006). Expression of 18-wheeler in the follicle cell epithelium affects cell migration and egg morphology inDrosophila. Dev. Dyn. 235, 1953-1961. doi:10.1002/dvdy.20820

Koressaar, T. and Remm, M. (2007). Enhancements and modifications of primerdesign program Primer3. Bioinformatics 23, 1289-1291. doi:10.1093/bioinformatics/btm091

Lambert, A. W., Pattabiraman, D. R. and Weinberg, R. A. (2017). Emergingbiological principles of metastasis. Cell 168, 670-691. doi:10.1016/j.cell.2016.11.037

Lemaitre, B., Nicolas, E., Michaut, L., Reichhart, J. M. and Hoffmann, J. A.(1996). The dorsoventral regulatory gene cassette spatzle/Toll/cactus controls thepotent antifungal response in Drosophila adults. Cell 86, 973-983. doi:10.1016/S0092-8674(00)80172-5

Levine, B. D. and Cagan, R. L. (2016). Drosophila lung cancer models identifytrametinib plus statin as candidate therapeutic. Cell Rep 14, 1477-1487. doi:10.1016/j.celrep.2015.12.105

Lewis, M., Arnot, C. J., Beeston, H., McCoy, A., Ashcroft, A. E., Gay, N. J. andGangloff, M. (2013). Cytokine Spatzle binds to the Drosophila immunoreceptorToll with a neurotrophin-like specificity and couples receptor activation. Proc. Natl.Acad. Sci. USA 110, 20461-20466. doi:10.1073/pnas.1317002110

10

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms

Ligoxygakis, P., Pelte, N., Hoffmann, J. A. and Reichhart, J. M. (2002). Activationof Drosophila Toll during fungal infection by a blood serine protease. Science 297,114-116. doi:10.1126/science.1072391

Ma, X., Huang, J., Yang, L., Yang, Y., Li, W. and Xue, L. (2012). NOPOmodulatesEgr-induced JNK-independent cell death in Drosophila. Cell Res. 22, 425-431.doi:10.1038/cr.2011.135

Ma, X., Yang, L., Yang, Y., Li, M., Li, W. and Xue, L. (2013). dUev1a modulatesTNF-JNK mediated tumor progression and cell death in Drosophila. Dev. Biol.380, 211-221. doi:10.1016/j.ydbio.2013.05.013

Ma, X., Li, W., Yu, H., Yang, Y., Li, M., Xue, L. and Xu, T. (2014). Bendlessmodulates JNK-mediated cell death andmigration in Drosophila.Cell Death Differ.21, 407-415. doi:10.1038/cdd.2013.154

Ma, X., Huang, J., Tian, Y., Chen, Y., Yang, Y., Zhang, X., Zhang, F. and Xue, L.(2017). Myc suppresses tumor invasion and cell migration by inhibiting JNKsignaling. Oncogene 36, 3159-3167. doi:10.1038/onc.2016.463

Mcilroy, G., Foldi, I., Aurikko, J., Wentzell, J. S., Lim, M. A., Fenton, J. C., Gay,N. J. and Hidalgo, A. (2013). Toll-6 and Toll-7 function as neurotrophin receptorsin the Drosophila melanogaster CNS. Nat. Neurosci. 16, 1248-1256. doi:10.1038/nn.3474

Meyer, S. N., Amoyel, M., Bergantin os, C., De La Cova, C., Schertel, C., Basler,K. and Johnston, L. A. (2014). An ancient defense system eliminates unfit cellsfrom developing tissues during cell competition. Science 346, 1258236. doi:10.1126/science.1258236

Mihaylov, I. S., Kondo, T., Jones, L., Ryzhikov, S., Tanaka, J., Zheng, J., Higa,L. A., Minamino, N., Cooley, L. and Zhang, H. (2002). Control of DNA replicationand chromosome ploidy by geminin and cyclin A. Mol. Cell. Biol. 22, 1868-1880.doi:10.1128/MCB.22.6.1868-1880.2002

Mishra-Gorur, K., Rand, M. D., Perez-Villamil, B. and Artavanis-Tsakonas, S.(2002). Down-regulation of Delta by proteolytic processing. J. Cell Biol. 159,313-324. doi:10.1083/jcb.200203117

Muller, A., Homey, B., Soto, H., Ge, N., Catron, D., Buchanan, M. E.,McClanahan, T., Murphy, E., Yuan, W., Wagner, S. N. et al. (2001).Involvement of chemokine receptors in breast cancer metastasis. Nature 410,50-56. doi:10.1038/35065016

Murphy, P. M. (2001). Chemokines and the molecular basis of cancer metastasis.N. Engl. J. Med. 345, 833-835. doi:10.1056/NEJM200109133451113

Obenauf, A. C. and Massague, J. (2015). Surviving at a distance: organ specificmetastasis. Trends Cancer 1, 76-91. doi:10.1016/j.trecan.2015.07.009

O’Neill, L. A. J. (2008). Toll-like receptors in cancer.Oncogene 27, 158-160. doi:10.1038/sj.onc.1210903

Paget, S. (1989). The distribution of secondary growths in cancer of the breast.1889. Cancer Metastasis Rev. 8, 98-101.

Pagliarini, R. A. and Xu, T. (2003). A genetic screen in Drosophila for metastaticbehavior. Science 302, 1227-1231. doi:10.1126/science.1088474

Parker, J. S., Mizuguchi, K. and Gay, N. J. (2001). A family of proteins related toSpatzle, the toll receptor ligand, are encoded in the Drosophila genome. Proteins45, 71-80. doi:10.1002/prot.1125

Parthier, C., Stelter, M., Ursel, C., Fandrich, U., Lilie, H., Breithaupt, C. andStubbs, M. T. (2014). Structure of the Toll-Spatzle complex, a molecular hub inDrosophila development and innate immunity. Proc. Natl. Acad. Sci. USA 111,6281-6286. doi:10.1073/pnas.1320678111

Puehringer, D., Orel, N., Luningschror, P., Subramanian, N., Herrmann, T.,Chao, M. V. and Sendtner, M. (2013). EGF transactivation of Trk receptorsregulates the migration of newborn cortical neurons. Nat. Neurosci. 16, 407-415.doi:10.1038/nn.3333

Rakoff-Nahoum, S. and Medzhitov, R. (2009). Toll-like receptors and cancer. Nat.Rev. Cancer 9, 57-63. doi:10.1038/nrc2541

Ribatti, D., Mangialardi, G. and Vacca, A. (2006). Stephen Paget and the ‘seedand soil’ theory of metastatic dissemination. Clin. Exp. Med. 6, 145-149. doi:10.1007/s10238-006-0117-4

Rose, D., Zhu, X., Kose, H., Hoang, B., Cho, J. and Chiba, A. (1997). Toll, amuscle cell surface molecule, locally inhibits synaptic initiation of the RP3motoneuron growth cone in Drosophila. Development 124, 1561-1571.

Schmittgen, T. D. and Livak, K. J. (2008). Analyzing real-time PCR data by thecomparative C(T) method. Nat. Protoc. 3, 1101-1108. doi:10.1038/nprot.2008.73

Schubiger, M. and Truman, J. W. (2000). The RXRortholog USP suppresses earlymetamorphic processes in Drosophila in the absence of ecdysteroids.Development 127, 1151-1159.

Shia, A. K., Glittenberg, M., Thompson, G., Weber, A. N., Reichhart, J.-M. andLigoxygakis, P. (2009). Toll-dependent antimicrobial responses in Drosophilalarval fat body require Spatzle secreted by haemocytes. J. Cell Sci. 122,4505-4515. doi:10.1242/jcs.049155

Sutcliffe, B., Forero, M. G., Zhu, B., Robinson, I. M. and Hidalgo, A. (2013).Neuron-type specific functions of DNT1, DNT2 and Spz at the Drosophilaneuromuscular junction. PLoS ONE 8, e75902. doi:10.1371/journal.pone.0075902

Takeuchi, H., Ikeuchi, S., Kitagawa, Y., Shimada, A., Oishi, T., Isobe, Y.,Kubochi, K., Kitajima, M. and Matsumoto, S. (2007). Pretreatment plasmafibrinogen level correlates with tumor progression and metastasis in patients withsquamous cell carcinoma of the esophagus. J. Gastroenterol. Hepatol. 22,2222-2227. doi:10.1111/j.1440-1746.2006.04736.x

Tauszig, S., Jouanguy, E., Hoffmann, J. A. and Imler, J.-L. (2000). Toll-relatedreceptors and the control of antimicrobial peptide expression in Drosophila. Proc.Natl. Acad. Sci. USA 97, 10520-10525. doi:10.1073/pnas.180130797

Untergasser, A., Cutcutache, I., Koressaar, T., Ye, J., Faircloth, B. C., Remm, M.and Rozen, S. G. (2012). Primer3–new capabilities and interfaces. Nucleic AcidsRes. 40, e115. doi:10.1093/nar/gks596

Valanne, S., Wang, J.-H. and Ramet, M. (2011). The Drosophila Toll signalingpathway. J. Immunol. 186, 649-656. doi:10.4049/jimmunol.1002302

Wang, J., Loberg, R. and Taichman, R. S. (2006). The pivotal role of CXCL12(SDF-1)/CXCR4 axis in bone metastasis. Cancer Metastasis Rev. 25, 573-587.doi:10.1007/s10555-006-9019-x

Wu, M., Pastor-Pareja, J. C. and Xu, T. (2010). Interaction between Ras(V12) andscribbled clones induces tumour growth and invasion. Nature 463, 545-548.doi:10.1038/nature08702

Yagi, Y., Nishida, Y. and Ip, Y. T. (2010). Functional analysis of Toll-related genes inDrosophila. Dev. Growth Differ. 52, 771-783. doi:10.1111/j.1440-169X.2010.01213.x

Zhang, Y.-B., He, F.-L., Fang, M., Hua, T.-F., Hu, B.-D., Zhang, Z.-H., Cao, Q. andLiu, R.-Y. (2009). Increased expression of Toll-like receptors 4 and 9 in humanlung cancer. Mol. Biol. Rep. 36, 1475-1481. doi:10.1007/s11033-008-9338-9

Zlotnik, A. (2006). Chemokines and cancer. Int. J. Cancer 119, 2026-2029. doi:10.1002/ijc.22024

11

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm039727. doi:10.1242/dmm.039727

Disea

seModels&Mechan

isms