Loss of flfl Triggers JNK-Dependent Cell Death in Drosophila

falafel (flfl) encodes a Drosophila homolog of human SMEK whose in vivo functions remain elusive. In this study, we performed gain-of-function and loss-of-function analysis in Drosophila and identified flfl as a negative regulator of JNK pathway-mediated cell death. While ectopic expression of flfl suppresses TNF-triggered JNK-dependent cell death, loss of flfl promotes JNK activation and cell death in the developing eye and wing. These data report for the first time an essential physiological function of flfl in maintaining tissue homeostasis and organ development. As the JNK signaling pathway has been evolutionary conserved from fly to human, a similar role of PP4R3 in JNK-mediated physiological process is speculated.


Introduction
falafel (flfl) is a Drosophila protein phosphatase 4 (PP4) regulatory subunit 3 (PP4R3) [1], which specifically mediates Miranda (Mira) localization and determinants cell fate during both interphase and mitosis [2]. flfl binds to CENP-C with its EVH1 domain [3] that is crucial for PP4 catalytic activity to centromeres at chromosomes during mitosis. Previous study proposed that PP4 functions through the modular activity of its component subunits [3]. Although in vitro studies have reported that PP4 is involved in a variety of molecular and cellular processes including regulation of c-Jun N-terminal kinase (JNK) pathway [4], NF-kB pathway [5], hematopoietic progenitor kinase 1 [6], apoptosis [7], and cell division [8], flfl's in vivo functions remain poorly understood. The human homolog of flfl is SMEK, which recruits PP4c to promote neuronal differentiation by dephosphorylating Par3 [9]. However, other in vivo functions of SMEK remain largely elusive.
To identify additional factors that regulate Egr-triggered JNK-mediated cell death, we performed a genetic screen for dominant modifiers of the GMR > Egr small eye phenotype. From the screen, we found that expression of flfl suppresses Egr-triggered cell death. On the other hand, knocking down flfl induced JNK activation and JNK pathway-dependent cell death, suggesting a physiological function of flfl in animal development. To our knowledge, this is the first report that flfl negatively regulate TNF-JNK signaling-induced cell death in vivo.  Center, and UAS-bsk-IR (5680R-2) was from Fly Stocks of National Institute of Genetics (NIG). puc E69 [18], GMR-Gal4, en-Gal4, pnr-Gal4, UAS-GFP [19,20], UAS-Egr [16], UAS-Egr w [6], UAS-Hep, and UAS-Bsk DN [21] were previously described.

AO Staining.
Eye discs from 3rd instar larvae were dissected in 1% PBS buffer. AO staining procedure was based on previous assay [22]. Florescent image of eye discs labeled with AO was collected with Olympus Microscope BX51. 10 discs of each genotype were collected for statistics analysis.

Light Image.
3-day-old flies of each genotypes were collected and immediately frozen at −80 ∘ C. For the image, flies were mounted on 1% agarose plates. Light images of eye and thorax were documented with OLYMPUS stereo microscope SZX16.

X-Gal
Staining. X-Gal staining was performed as previously described with minor modification [23,24]. Wing imaginal discs from 3rd instar larvae were dissected in 1% PBS buffer and fixed with 1% glutaraldehyde for 15 minutes at room temperature and incubated with -galactosidase at 37 ∘ C for 24 hours.

flfl Suppresses Egr-Induced Cell Death in Eye Development.
As previous study showed, ectopic expression of Egr under the control of GMR-Gal4 induced a small eye phenotype [17]. This phenotype is mostly suppressed by coexpressing a dominant negative allele of Bsk (Bsk DN ) encoding the Drosophila JNK ortholog [21], which indicates Egr-induced cell death is mainly mediated by JNK signaling [25]. To identify additional components of the Egr-JNK pathway or factors interacting with the pathway, we performed a genetic screen for dominant modifiers of the GMR > Egr small eye phenotype and identified Nopo, Ben, Wnd, and Wg signaling as essential regulator of Egr-JNK pathway induced cell death [21,26]. From the screen, we also found that the GMR > Egr small eye phenotype (Figure 1(c)) was significantly suppressed by EY03585 (Figure 1(e)), a P-element inserted in the first intron of flfl. This P-element carries the UAS sequence located about 1 kb upstream of the coding region and is able to drive the expression of flfl by the GMR-Gal4 driver. However, expression of flfl by itself had no effect on the eye size ( Figure 1(b)), compared to the GMR-Gal4 control ( Figure 1(a)). As a negative control, coexpressing GFP did not suppress GMR > Egr-triggered small eye phenotype ( Figure 1(d)). Thus, the data indicate that flfl is able to suppress Egr-induced cell death in the eye.

Loss of flfl Enhances Egr-Induced Cell Death in Eye Development.
As flfl gain of function suppressed Egr-induced cell death, we wonder whether loss of flfl could enhance Egrtriggered cell death. To this end, we knocked down flfl in the eye by expressing flfl RNAi with GMR-Gal4 and observed a rough eye phenotype (Figure 2(d)), compared to the control (Figure 2(a)). Consistent with previous reports, expression of a weaker UAS-Egr allele (UAS-Egr w ) driven by GMR-Gal4 resulted in a rough eye phenotype (Figure 2(b)). This phenotype is severely enhanced by knocking down flfl as there was almost no eye tissue left (Figure 2(e)). As a negative control, expressing a RNAi sequence specifically targeting green fluorescent protein (GFP) has no effect on GMR > Egr w -triggered rough eye phenotype (Figure 2(c)). These results show that flfl loss of function rigorously enhances Egrtriggered eye phenotype. It was previously reported that ectopic Egr-induced eye phenotype is caused by cell death [16]. To examine cell death in vivo, we performed acridine orange (AO) staining that specifically labels dying cell. As reported previously [12], ectopic expression of a weak UAS-Egr transgene (UAS-Egr w ) driven by GMR-Gal4 induced mild cell death in eye discs posterior to the morphogenetic furrow (MF), as revealed by AO staining (Figure 2(g)). Egr-triggered cell death was rigorously enhanced by expressing flfl RNAi (Figure 2(j)) but remained unaffected by expressing GFP RNAi (Figure 2(h)). Consistent with its rough eye phenotype, knocking down flfl provoked weak cell death (Figure 2(i)). These data suggest that loss of flfl enhances Egr-induced cell death in eye development.

Loss of flfl Enhances JNK-Mediated Cell Death in Thorax
Development. To investigate whether flfl suppresses JNKmediated cell death in other tissues, we activated JNK signaling in the notum with pannier-Gal4 (pnr-Gal4). Expression of Hep, the Drosophila homolog of JNK, driven by pnr-Gal4 induced cell death and produced a small scutellum in adult fly (Figure 3(d)) [21]. Knocking down flfl by pnr-Gal4 slightly decreased scutellum size (Figure 3(c)) and dramatically enhanced Hep-induced cell death by producing a no scutellum phenotype as well as a split thorax in adult flies (Figure 3(f)). As a negative control, expression of a GFP RNAi did not produce any effect on scutellum size (Figures 3(b) and 3(e)). Together, the results indicated that flfl negatively regulates JNK-mediated cell death in thorax development.
During Drosophila imaginal discs development, slowproliferating cells are eliminated by a process called "cell competition" [27], which regulates tissue's homeostasis and organs' fitness and final cell number. JNK pathway was shown to play a crucial role in cell competition by eliciting cell death in "loser cells" [28,29]. Since our data suggest that flfl impedes JNK-mediated cell death in a nontissue specific manner, flfl is likely a negative regulator of JNK-dependent cell competition and tissue homeostasis.

Loss of flfl Induces JNK Pathway Activation and Cell
Death in Wing Development. To investigate the physiological functions of flfl in wing development, we specifically knocked down flfl in the posterior compartment of wing discs by engrailed-Gal4 (en-Gal4) and checked cell death with AO staining. We found that loss of flfl triggered extensive cell death in the posterior compartment of wing discs (Figure 4(c)), compared with the en-Gal4 control (Figure 4(a)) and en > GFP-IR (Figure 4(b)). These results suggest that flfl is physiologically required for cell survival in Drosophila wing development.
To examine whether JNK signaling plays a role in loss of flfl induced cell death, we checked the expression of puc, a transcriptional target of JNK pathway [30]. puc E69 is a puc mutant allele with a LacZ bearing P-element inserted into the puc locus and serves as a puc-LacZ reporter [31] whose expression could be easily visualized by X-Gal staining. We found that knocking down flfl in the posterior compartment of wing discs resulted in upregulated puc-LacZ expression (Figure 4(f)), compared with the en-Gal4 control The JNK pathway is evolutionary conserved from fly to human. Compared with the compact Drosophila genome, there are three homologs of flfl, SMEK1, SMEK2, and SMEK3P, and dozens of Puc homologs named dual specificity phosphatase (DUSP) in human. Previous study has reported that JNK signaling is essential for cell migration and tumor invasion [32]. Based on the above data, we speculate that SMEK is downregulated and DUSP is upregulated in metastatic tumor. Consistent with the hypothesis, we found from the Oncomine database (https://www.oncomine.org/) that SMEK1 expression is indeed downregulated whereas DUSP1 is upregulated in invasive breast carcinoma stroma compared to normal tissue (Figures 4(g) and 4(h)) [33]. These data imply that the role of flfl in modulating JNK pathway is likely conserved by SMEK1 from Drosophila to human.
Although our data mining and previous study found that JNK activity is elevated in several cancer cell lines, its role in tumor development is context-dependent [8]. JNK pathway was implicated as both procancer and anticancer signaling in cancer development for its regulation on cell proliferation and cell death, respectively [6]. In certain mouse models of cancer, JNK deficiency enhances tumor formation and metastasis [20,34]. In Drosophila, clones with ectopic oncogene Src expression induce no-autonomous tumor growth [35], while Src expression also induces cell death through JNK pathway [22]. Cells in Src clone could escape from cell death if JNK pathway is blocked [35]. Intriguingly, another important oncogene Ras can also switch JNK pathway from anti-to protumor signaling [6]. Thus, upon the presence of different regulating factor(s), JNK pathway modulates cell death, tumor genesis, and progression in a cell context-dependent manner.

Loss of flfl Induced Cell Death Is JNK Pathway-Dependent.
Knocking down flfl by GMR-Gal4 induced cell death in eye discs (Figure 2(i)) and produced a rough eye phenotype in adults (Figure 2(d)). These results were confirmed by another independent line of flfl RNAi (Figures 5(b) and 5(b )). To understand whether loss of flfl induced cell death is JNK pathway dependent, we blocked JNK signaling by expressing a bsk RNAi or a dominant negative allele of Bsk (Bsk DN ). We found that loss of flfl triggered rough eye phenotype ( Figure 5(b)) and increased cell death in eye discs ( Figure 5(b )) were significantly suppressed by compromised JNK activity (Figures 5(c)-5(e)). As a control, GFP RNAi and

Conclusions
In this study we have identified flfl as a negative regulator of TNF-trigger JNK-mediated cell death in Drosophila. While ectopic expression of flfl impedes JNK signaling-induced cell death, loss of flfl induces JNK pathway activation and cell death in Drosophila eye and wing discs and produced morphological defects in the adult eye. These data suggest an important physiological function of flfl in maintaining tissue homeostasis in Drosophila organ development. flfl's ability to inhibit JNK signaling is likely retained by its human homolog SMEK1. Consistently, while activated JNK pathway promotes dermal fibroblasts cell migration in wound healing [36], ectopic expression of SMEK1 significantly decreased the migration ability of carcinoma cells [37]. In addition, we found from Oncomine database that SMEK1 is downregulated whereas JNK signaling target gene DUSP1 is upregulated in human invasive carcinoma [33].