The dark side of fly TNF

Abstract

The fruit fly Drosophila is an important model for biological research; however, due to its relatively short lifespan its relevance in cancer research is often questioned. Nevertheless, among many other intriguing Drosophila models, scribble group mutants provided early evidence for the existence of tumor suppressor genes, and their importance in mammalian systems is beginning to emerge. In this extra view, I discuss recent advances in our understanding of the phenotypes of scrib-group mutants, in which the activation of JNK signaling plays a crucial role. Several mechanisms can account for the activation of JNK within scrib-group mutant cells, including a mechanical stress triggered by the loss of polarity, cell competition, intrinsic tumor suppression by autonomous production of Eiger and an inflammatory response mediated by Eiger-producing hemocytes. Eiger, the sole Drosophila homolog of tumor necrosis factor, is emerging as a “danger signal” initiated upon the presence of external pathogens, damaged tissues and the appearance of pre-malignant cells. Remarkably, in the presence of the Ras oncoprotein, Eiger can act as a tumor promoter by stimulating invasive migration and delaying the onset of metamorphosis.

Eiger, the TNF Homolog in Drosophila

The Drosophila genome encodes a single member of the TNF superfamily, Eiger (eda-like cell death trigger, egr).^1,2 The initial reports characterizing Eiger/dTNF indicated that its misexpression in imaginal disc epithelial cells results in JNK-dependent cell death. Nevertheless, egr mutants develop normally to become viable, fertile adults and no role for developmental cell death has been ascribed to egr. A physiological role of egr was first unveiled by a report showing that egr mutants are differentially sensitive to bacterial infection.³ This suggested that egr is involved in the fly immune response rather than in normal development. In this way, egr plays a similar role to its mammalian counterpart as a key pro-inflammatory cytokine of the innate immune system. More recent studies have indicated that while egr mutants are sensitive to extracellular bacterial pathogens, they are equally sensitive or even more tolerant than wild-type flies to infections by intracellular pathogens.⁴ This latter case suggests that egr could mediate a ‘septic shock’ response and participate in the pathology of the infection. Interestingly, the expression of egr in the fat body is sufficient to mediate these immune phenotypes.⁵

Drosophila Polarity Tumor Suppressor of the Scribble Group

Because cancer is an aging-associated disease in humans, it has been speculated that short-lived organisms such as flies are excluded from suffering from cancer.⁶ Therefore, the complex tumor suppressor mechanisms present in mammals may have appeared rather recently in the evolution of highly derived organisms. Based on similarities from their molecular machinery, it is speculated that such tumor suppressor mechanisms may have evolved from the stress-response pathways. An interesting example is p53, which plays key roles both as a stress-response mediator and tumor suppressor in mammals but in Drosophila only a role in the stress response has been described.

Nevertheless, we have previously speculated that most organisms composed of multicellular communities have to deal with cancer in the form of ‘rebel’ cells that fail to restrict their proliferation and growth when needed for the benefit of the multicellular community, the organism.⁷ In this way, the less derived but genetically simpler invertebrate models could provide insights into the fundamental biology of cancer.

In fact, spontaneous mutations reported in the 1930s, such as lethal giant larvae (lgl), presented founding evidence for the existence of tumor suppressor genes.⁸ In such mutants, the mitotic tissues, such as the imaginal discs, display tumor-like dramatic overgrowth. For a recent review on the molecular and biochemical nature of lgl and its partners of the “scribble group”, see reference ⁹. Importantly, in these mutants this overgrowth is not due to faster proliferation. Instead, a key component of this phenotype is a developmental arrest in the larval stage of the fly life cycle. During the normal fly life cycle, wild-type animals proceed to the pupa (metamorphosis) stage where tissues eventually stop proliferating and growing to start differentiation and apoptotic programs that shape the organs of the adult fly. In lgl mutants, the arrest in the larva stage allows an extended window of time of growth for the tumor-like mitotic tissues.

A key question is then whether these tumor-like outgrowths from already proliferating tissues could successfully model aspects of human cancer within a developing organism. The roles of lgl and its partner's scribble (scrib) and disc large (dlg) are only beginning to be characterized in murine models for cancer, and tumor genome analyses should indicate whether they are bona fide tumor suppressors in human cancer. Importantly, recent work indicates that scrib acts as a tumor suppressor gene in the murine breast epithelium.¹⁰

Whole animal phenotypes versus genetic mosaics.

An important characteristic of scribble-group mutants is that while fully mutant animals develop tumors, the outcome is different in genetic mosaics where the mutant cells are surrounded by wild-type ones. In this case, the mutant cells undergo JNK-dependent cell death and are eliminated from the tissues.^11,12 This is intriguing because the mosaic situation could perhaps better model the appearance of tumors within normal tissues. The mechanisms by which these mutant cells are eliminated by apoptosis have been intensely investigated and the recent advances are discussed below.

Cancer is thought to be the result of multiple genetic and environmental insults to a tissue, and although the precise number of ‘driver’ mutations in human tumors is still undefined, it is widely accepted that most tumors are driven by more than one genetic insult. In line with this concept, Brumby and Richardson performed a seminal genetic screen to identify genes that, when overexpressed, could prevent the death of scrib clonal patches. Importantly, the screen identified proto-oncogenes such as Raf and Notch.¹² In a reciprocal genetic screen, Pagliarini and Xu identified scrib, lgl, dlg and other polarity genes as cooperators of oncogenic Ras (Ras^V12) needed to drive overgrowth, invasion and metastasis in the context of genetic mosaics within the eye imaginal disc.¹³

Because Raf is a major signaling kinase downstream of Ras, the oncogenic cooperation of Ras/Raf and scribble became a paradigmatic model of oncogenic cooperation in Drosophila. Unexpectedly, a recent study indicates that such cooperation is not necessarily cell-autonomous: Ras^V12 clones could progress to invasive tumors when juxtaposed to scrib clones.¹⁴ A key event downstream of scrib loss is JNK activation. Importantly, while as mentioned above JNK activation promotes the death of scrib clones, JNK drives tumor progression in the context of Ras^V12/scrib cells.^11,15 JNK activation propagates across the imaginal disc epithelium from a wound and presumably, from a scrib patch of cells as well.¹⁴ Therefore, JNK is not only activated within the scrib cells themselves but within their neighbors as well. In this way, scrib cells could remotely turn on JNK within the Ras^V12 cells. Taken all together, these studies suggest that the concomitant activation of the Ras/RAF/ERK and JNK MAPK pathways is key for the progression of these tumors.

The Link of Scribble Loss and JNK Activation

Several mechanisms have been proposed for causing the key JNK activation within scrib group mutants (Fig. 1), including: (1) cell competition, (2) loss of polarity-induced mechanical stress, (3) intrinsic tumor suppression and (4) an innate immune response to tumors, which could be termed “extrinsic tumor suppression.”

Figure 1.

All roads lead to JNK. Diagram with the proposed mechanisms for the activation of JNK after the loss of scrib group genes. JNK pathway activation normally results in the death of the mutant cells; however, in the presence of oncogenic Ras JNK can promote tumor progression. See the text for details.

(1) Cell competition (CC) is a process in which faster-growing cells actively eliminate slower-growing cells in proliferating epithelia. In this way, viable but suboptimal cells are removed. This process can be observed experimentally by creating genetic mosaics with mutants of minute genes, which encode for components of the protein synthesis machinery, and also by juxtaposing populations of cells expressing different levels of the Myc protein.¹⁶

JNK plays a role in cell competition¹⁷ although the absolute importance of such role is disputed.¹⁸ Uhlirova and Bohmann first proposed that cell competition between scrib cells and their wild type neighbors is the cause of JNK activation.¹⁵ Unlike in the case of genetic mosaics, animals fully deficient for any scribble group gene develop tumors. Furthermore, scrib clones survive in the eye imaginal disc when the wild-type cells are genetically ablated.¹² These results suggest the importance of these wild-type neighboring cells in the fate of scrib group mutants in a similar way to Minute- or Myc-driven cell competition.

Furthermore, a recent study indicated that lgl clones display reduced Myc levels relative to the surrounding wild-type neighbors. Most importantly, the death of lgl clones was rescued both by Myc over-expression and by reducing the growth rate of the neighboring cells in the genetic mosaics.¹⁹

(2) Polarity stress (PS): JNK is also known as “Stress Activated Protein Kinase” (SAPK) and is a general mediator of the stress response. A whole-genome RNAi screen for genes that when knocked down result in JNK activation identified dlg and other components of the apical-basal polarity machine. Interestingly, this study utilized a cultured Drosophila epithelial cell line and a highly sensitive reporter for JNK activity.²⁰ This suggests that dlg loss (and other related genetic alterations) results in JNK activation due to a stress response triggered by the loss of polarity. The fact that this was observed in a homogenous cell population in vitro precludes the influence of wild-type neighbors or other non-cell autonomous effects. A similar mechanism has been suggested for patches of cells deficient for dpp signal transduction, in this case the cells are extruded from the epithelium and JNK may be activated due to a mechanical stress.²¹

(3) Intrinsic Tumor Suppression (ITS). Recent work indicated that the death of scrib and dlg clones depends on Eiger/dTNF, in a process defined as Intrinsic Tumor Suppression (ITS). This mechanism requires an autocrine (cell autonomous) production of Eiger by the mutant cells, which is then internalized to signal from endocytic compartments. Nevertheless, while the production of Eiger seems cell-autonomous, the presence of wild-type neighbors is required to enhance the endocytic rate of the mutant cells.²² As such, the ITS mechanism for the removal of scrib cells is not completely cell-autonomous and thus the term “Intrinsic” could be misleading.

Froldi et al. reported that ITS does not play a major role in the removal of lgl clones within the pouch region of the wing-disc epithelium: when knocking down eiger by RNAi within the lgl clones, JNK activation was still observed and no significant change in lgl clone survival was detected.¹⁹ Therefore, although there could be gene-specificities for different scrib group components (e.g., lgl versus scrib and dlg) and tissue specificities for ITS versus CC, the generality of ITS seems disputed. It is worth noting that Igaki et al. reported a drastic improvement in survival of scrib cells against an egr mutant background. However, the improvement was rather mild when knocking down egr exclusively within the scrib clones by RNAi.²² One plausible explanation for this discrepancy is that the RNAi knockdown could result in weaker phenotypes than a null allele. An alternative possibility is that Eiger is not produced primarily by the mutant cells, and in this way acts non cell-autonomously/extrinsically. Our recent data described bellow supports this hypothesis.

(4) Extrinsic Tumor Suppression. We confirmed the importance of the egr gene in the removal of scrib, dlg and lgl-deficient cells.²³ Immunostaining experiments detected the Egr protein in a punctuated, intracellular vesicle-like pattern (Fig. 2, red arrow; ref. 23). This is consistent with Igaki et al.²² However, the highest Egr protein levels were not within the scrib cells themselves; instead, they were at the periphery of associated hemocytes (Fig. 2, white arrow; reviewed in ref. 23). Haemocytes are the immune surveillance cells in Drosophila, known to target external pathogens.²⁴ This finding was intriguing because hemocytes associate to Ras^V12/scrib tumors and impact negatively the growth of tumors within scrib fully mutant larvae.²⁵

Figure 2.

Eiger/dTNF is primarily expressed by tumor-associated hemocytes. The confocal image displays a Ras^V12/scrib imaginal disc tumor. The tumor cells are labelled with GFP (green), Eiger immunostaining is in red and nuclei are shown in blue. The Eiger staining is shown in grey in the lower part. The white arrow points to a tumor-associated cell that expresses high levels of Eiger. These tumor-associated cells were identified as plasmatocytes, a sub-type of haemocytes.²³ The red arrow points to a tumor cell displaying Eiger staining in the form of intracellular puncta, identified previously as endocytic compartments.²²

The functional relevance of egr expression by hemocytes was tested: the survival of lgl clones, both in the wing and eye-antenna discs was dramatically improved by knocking down egr through RNAi, specifically in hemocytes. Therefore, Eiger expression by hemocytes is necessary for the efficient removal of lgl cells. Reciprocally, hemolymph transfusion experiments using larvae with different genotypes indicated that wild-type—but not egr hemolymph—could rescue the expression of the Egr/JNK pathway target dMMP1 in scrib, egr double mutant cells.²³ This indicates that Eiger expression by the hemocytes is sufficient for Egr/JNK pathway activation within the scrib mutant cells.

This study suggests the importance of Eiger expression by the hemocytes associated with the mutant cells and could explain the discrepancy between the experiments using a fully mutant egr background (which depletes every cell within the organism of the Eiger protein) and RNAi targeting of egr exclusively within the scrib group mutant cells. It is worth noting that all possible mechanisms described to account for JNK activation within scrib group mutant cells (PS, CC, ITS and ETS) could potentially act synergistically or compensate for each other in different situations to ensure an efficient removal of the mutant cells.

TNF and the Lethal Giant Larva Phenotype

Unlike the case of genetic mosaics, experiments with fully mutant scrib group animals can provide evidence for mechanisms that are independent of the presence of nearby normal epithelial cell. For example, cell competition by definition requires the presence of wild type cells. As mentioned above, the larval arrest is a major characteristic of scrib group mutants, which die as giant larvae with their mitotic tissues transformed into tumors. Remarkably, this larval arrest is fully dependent on the egr locus: double mutant egr; scrib animals progress to the pupa stage. Because they lack additional time for overgrowth, the imaginal disc tumors of egr; scrib animals do not reach the dramatic size typical of scrib-group mutants.²³

Furthermore, there are pockets of apoptotic cell death within the discs of scrib animals, but not in the discs of egr; scrib double mutant animals.²³ This facet of the egr-dependent response to scrib group mutants does not require the presence of wild type cells and therefore distinguishes it from CC and ITS. Importantly, the characteristic larval arrest that allows tumor development is fully dependent on egr. Because this response does not discriminate between patches of mutant cells or fully mutant tissues, it can be speculated that the egr-dependent larva arrest constitutes a failed attempt from the host to completely eliminate the mutant cells. Early work in fruit flies identified a developmental checkpoint that delays the onset of metamorphosis in the case of imaginal disc damage.²⁶ This ensures that tissue regeneration has the time needed to heal the wounds. For example, Halme et al. recently blocked the irradiation-induced developmental delay using the metamorphosis-inducing hormone Ecdysone. This resulted in either lethality or animals with aberrant adult appendages, depending on the initial degree of tissue damage.²⁷

The similarities between wounds and tumors have been long appreciated, and it may be the case that similar mechanisms act for the developmental delay upon both irradiation-induced tissue damage and pre-malignant lesions.

TNF as an ancestral danger signal.

Egr overexpression results in epithelial cell death and intriguingly egr is upregulated by p53 in response to ionizing radiation. Surprisingly, the p53-induced cell death is independent of egr.²⁸ The physiological role of Eiger in this context was unknown until recently. When produced by dying irradiated epidermal cells, Eiger acts on sensory neurons expressing its receptor, Wengen, to mediate the heightened nociceptive (pain) sensitivity response. This induces an evolutionarily conserved behavior that protects the damaged tissue until it heals. Babcock et al. propose that TNF represents an ancestral danger signal to deal with tissue damage.²⁹ Indeed, taken together, the recent research on egr positions this cytokine as a central signaling node, coordinating responses to external pathogens, tissue damage and developing tumors. Such response involves monitoring of peripheral tissues through both the nervous and immune systems.

The Dark Side of TNF

Metazoa evolved mechanisms to eliminate external pathogens such as invading bacteria, internal pathogens such as mutant and pre-malignant cells and to coordinate animal behavior upon tissue damage. However, there can be cases when the host's own weapons of defense ‘misfire’ to cause detrimental effects. In the case of some intracellular bacterial pathogens, egr can make the host intolerant to the infection; in this way it contributes to the pathology and the result is accelerated animal death.⁴ Additionally, an exaggerated nociceptive response could lead to chronic pain, which is a major health care issue worldwide.

Perhaps relevant to cancer biology, when fly larvae are challenged with patches of pre-malignant of cells, an egr-dependent immune response results in the death of such mutant cells. If needed, this egr response could delay the onset of metamorphosis until the mutant cells are completely removed. However, when the removal of the aberrant cells cannot be completely achieved, the developmental arrest is permanent until the animals die as giant larvae. This is the case with animals fully mutant for genes of the scrib group or when discrete patches of scrib group mutant cells become resistant to Egr/JNK-mediated cell death through the expression of the Ras^V12 oncoprotein.

It could be interpreted as mutant cells within mitotic tissues “usurping” the arrest in the proliferative larval stage of development—which likely evolved as a wound healing and tumor suppressor mechanism—as an opportunity to develop into tumors. Furthermore, the activation of JNK in Ras^V12/scrib cells does not just fail to stimulate cell death but drives invasive migration instead. This could be an unwanted “side effect” of Egr, which relies on JNK as its signal transducer, while JNK is a general factor with multiple roles in homeostasis and development. For example, JNK participates in the invasive migration of cells during imaginal disc eversion, a developmental process with a clear resemblance to malignant cell invasion.³⁰ Through expression of the Ras oncoprotein, pre-malignant cells can thus highjack Egr signaling to re-direct it from a pro-death into a pro-growth, pro-invasion stimulus.

Implications for Cancer

The parallels between the fly and mammalian TNF are emerging in different fields of research. Murine TNF has long been known to act both as a tumor promoter and as a tumor suppressor, and this seems to be true in Drosophila as well. Further parallels can be traced from these apparently disparate models; importantly, the powerful genetic toolkit of Drosophila could identify key determinants for the context-dependent roles of TNF; for example, we identified the Ras oncoprotein as a switch turning TNF into a tumor promoter in the fly. This is consistent with previous reports from murine models of cancer: in a Ras-driven genetic cancer model, induction of pancreatic inflammation accelerates tumorigenesis.³¹ In the papilloma model based on DMBA-TPA treatment, activating Ras mutations are a highly frequent event.³² Intriguingly, in such a model, TNF plays a key role as a tumor-promoting cytokine.³³ Together, the findings reviewed here indicate that despite apparent caveats, short-lived organisms like Drosophila could model important aspects of cancer. Remarkably, mechanisms that evolved for developmental proofreading to ensure developmental robustness could also act as tumor promoters.