NPM-hMLF1 fusion protein suppresses defects of a Drosophila FTLD model expressing the human FUS gene

Fused in sarcoma (FUS) was identified as a component of typical inclusions in frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS). In FTLD, both nuclear and cytoplasmic inclusions with wild-type FUS exist, while cytoplasmic inclusions with a mutant-form of FUS occur in many ALS cases. These observations imply that FUS plays a role across these two diseases. In this study, we examined the effect of several proteins including molecular chaperons on the aberrant eye morphology phenotype induced by overexpression of wild-type human FUS (hFUS) in Drosophila eye imaginal discs. By screening, we found that the co-expression of nucleophosmin–human myeloid leukemia factor 1 (NPM-hMLF1) fusion protein could suppress the aberrant eye morphology phenotype induced by hFUS. The driving of hFUS expression at 28 °C down-regulated levels of hFUS and endogenous cabeza, a Drosophila homolog of hFUS. The down-regulation was mediated by proteasome dependent degradation. Co-expression of NPM-hMLF1 suppressed this down-regulation. In addition, co-expression of NPM-hMLF1 partially rescued pharate adult lethal phenotype induced by hFUS in motor neurons. These findings with a Drosophila model that mimics FTLD provide clues for the development of novel FTLD therapies.


Results
Ectopic expression of hFUS in Drosophila eye imaginal discs induces aberrant eye morphology in adults. It was reported that ectopic expression of hFUS in eye imaginal discs induces aberrant eye morphology [31][32][33]37 . To confirm the effect of hFUS expression on eye morphology, we crossed w; UAS-hFUS /CyO Act-GFP with GMR-GAL4 driver to express hFUS in the region posterior to the morphogenetic furrow (MF) and then inspected the adult compound eyes. Expression of hFUS in eye imaginal discs exhibited aberrant eye morphology at 25 °C ( Fig. 1d-f). Scanning electron microscope and stereomicroscope images of the compound eye revealed the many malformed bristles that are smaller than the normal bristles as observed previously [31][32][33]37 . Fusion of ommatidia was also observed in some regions of the compound eye compared with control flies expressing GFP (Fig. 1a-c). Severer eye phenotype was observed at 28 °C ( Fig. 1g-i). The number of malformed small bristles increased and fusion of ommatidia was observed in a wider area in the compound eye ( Fig. 1g-i). Loss of the pigment phenotype was also evident at 28 °C (Fig. 1i). Here, we thus confirmed the eye phenotype of the flies expressing hFUS.

Co-expression of NPM-hMLF1 suppresses the aberrant eye morphology phenotype induced by hFUS.
To identify genes that modify the aberrant eye morphology phenotype induced by expression of hFUS, we crossed the hFUS-overexpressing flies with flies overexpressing the following genes: Drosophila mammalian relative DnaJ (dMRJ), Drosophila myeloid leukemia factor (dMLF) and hMLF1. Overexpression of these genes is reported to suppress the eye degeneration phenotype of polyQ disease model flies 40,41 . However, co-expression of these three genes exhibited no apparent suppression of the aberrant eye morphology phenotype induced by hFUS (Fig. 2a). It is reported that dMLF mainly localizes in nucleus, although its localization changes depending on the tissue, the time of development, the cell cycle and its isoforms 46 . In contrast, it is shown that hMLF1 mainly localizes in cytoplasm in mammalian cultured cells, but it also partly localizes in nucleus 42,43,47,48 . Subcellular localization of dMRJ was not fully examined, but overexpressed dMRJ was co-localized with cytoplasmic polyQ aggregates in polyQ disease model flies 40 . We considered that the difference in subcellular localization may explain the apparent lack of expression from these proteins on the aberrant eye morphology phenotype induced by expression of hFUS. We therefore examined the effect of overexpression of NPM-hMLF1 as it is known that this fusion protein mainly localizes in the nucleus 43 . No aberrant eye morphology induced by NPM-hMLF1 was observed compared with control flies expressing GFP (Fig. 2b-g). Strong suppression of the aberrant eye morphology phenotype induced by hFUS was observed by co-expression of NPM-hMLF1 ( Fig. 2h-m). In the following studies, we therefore focused on the detailed mechanism for this suppression.
(c,f,I,l) Enlarged images of (b,e,h,k) respectively. The eye phenotype of at least five newly eclosed adult male flies of each line was examined and the experiments were done in triplicate. No significant variation in eye phenotype was observed among the five individuals. Anterior is to the left and dorsal to the top. The flies were developed at 28 °C. The bars indicate 50 μm (b,d,e,g,h,j,k,m) and 14.2 μm (c,f,I,l), respectively. White dotted lines define the area of aberrant eye morphology. single major band corresponding to 53 kDa was detected on the immunoblot using the anti-hMLF1 IgG, which recognized the C-terminal region of hMLF1 protein, but not with the extracts of the control flies expressing GFP (Fig. 3a). The reported molecular mass of NPM-hMLF1 is 56 kDa, which is generated by fusion of NPM (38 kDa) with hMLF1 (31 kDa) 43 . Thus, we confirmed the expression of NPM-hMLF1. The data also confirmed the specificity of anti-hMLF1 IgG.
Then, to examine subcellular localization of the NPM-hMLF1 fusion protein, we conducted immunostaining of eye imaginal discs with the anti-hMLF1 IgG. Strong NPM-hMLF1 signals were detected in the posterior region to MF in the eye imaginal discs (Fig. 3h). No such signal was detected in eye imaginal discs of the control flies expressing lacZ (Fig. 3d). Merged images showed that NPM-hMLF1 signals mainly co-localized with 4′ 6-diamidono-2-phenylindole (DAPI), especially in the DAPI faint region (Fig. 3j). The data suggest that NPM-hMLF1 mainly localizes in nucleoplasm of eye imaginal disc cells. However, since in some cells NPM-hMLF1 signals partially co-localized with cytoplasmic actin signals (Fig. 3k), we cannot exclude the possibility that some fractions of NPM-hMLF1 proteins also localize in cytoplasm. hFUS appears to co-localize with NPM-hMLF1 mainly in nucleoplasm of eye imaginal disc cells. To confirm specificity of anti-hFUS IgG, we performed Western immunoblotting analysis with extracts of adult heads carrying GMR-GAL4/Y; UAS-hFUS/+. Two major bands corresponding to 83 kDa and 72 kDa were detected on the immunoblot using the anti-hFUS IgG, but not with the extracts of the control flies expressing GFP (Fig. 4a). The molecular mass of these bands is within a similar range predicted from the amino acid sequences of hFUS (75 kDa). The two hFUS-immunopositive bands with similar mass were reported previously 35 . Here, we confirmed the specificity of anti-hFUS IgG.
To examine subcellular localization of hFUS protein in eye imaginal discs, we performed immunostaining with anti-hFUS IgG. Strong hFUS signals were detected in the posterior region to MF in the eye imaginal discs  ( Fig. 4h). No such signal was detected in eye imaginal discs of the control flies expressing lacZ (Fig. 4d). Merged images showed that hFUS signals co-localized with DAPI, especially in the DAPI faint region (Fig. 4I'). The data indicate that hFUS likely localizes in nucleoplasm.
In addition, we also performed double immunostaining of eye imaginal discs from flies co-expressing hFUS and NPM-hMLF1 with the anti-hFUS IgG and anti-hMLF1 IgG. Merged images indicated that hFUS signals co-localize with NPM-hMLF1, mainly in the DAPI faint region (Fig. 4r,q'), suggesting that hFUS co-localizes with NPM-hMLF1 mainly in the nucleoplasm of eye imaginal disc cells. However, since some merged signals of hFUS and NPM-hMLF1 also localized outside of DAPI -stained region ( Fig. 4q'), we cannot exclude the possibility that some fractions of hFUS and NPM-hMLF1 proteins also localize in cytoplasm.

The driving of hFUS expression at 28 °C down-regulates levels of hFUS and endogenous caz
proteins. Previous studies reported that ectopic expression of hFUS reduced endogenous caz 35,37 . Therefore, we re-examined the level of caz together with hFUS by Western immunoblotting analysis with extracts of adult heads carrying GMR-GAL4/Y; UAS-hFUS/+ developed at either 25 °C or 28 °C. At 25 °C, expression of hFUS was confirmed as expected and the level of endogenous caz was decreased to 61 percent compared to the control flies expressing GFP (Fig. 5a-c). Normally higher expression of hFUS is expected at 28 °C due to higher activity of the transcription factor GAL4 at 28 °C than at 25 °C 49 . However, surprisingly, not only endogenous caz level, but also hFUS level decreased to 11 percent and 37 percent at 28 °C compared with those at 25 °C, respectively ( Fig. 5a-c). The data suggest a possible negative feedback mechanism that may be mediated by protein degradation at 28 °C.

Down-regulation of both hFUS and caz are likely mediated by proteasome dependent degradation.
To examine the possibility that the degradation of both hFUS and caz proteins are mediated by proteasome dependent degradation, we added proteasome inhibitor (MG132) in fly food and flies were reared at 28 °C. Interestingly, dosage-dependent feeding of MG132 from 5 μM to 10 μM inhibited both hFUS and caz protein degradation (Fig. 5d-f). The data suggest that the down-regulation of both hFUS and caz are mediated by proteasome-dependent degradation.
Co-expression of NPM-hMLF1 recovered the hFUS protein level but not caz protein level. To examine the effect of co-expression of NPM-hMLF1 on levels of hFUS and caz proteins at 28 °C, we performed Western immunoblotting analysis with extracts of adult heads carrying GMR-GAL4/Y; UAS-hFUS/UAS-GFP and GMR-GAL4/Y; UAS-hFUS/UAS-NPM-hMLF1. Expression of hFUS decreased caz protein level by 51 percent (Fig. 6a,c), which corelates with the eye phenotype ( Fig. 2h-j). Co-expression of NPM-hMLF1 increased hFUS protein level approximately two fold, but the decreased caz protein level was not recovered (Fig. 6a-c). The eye phenotype was suppressed under these conditions ( Fig. 2k-m). These results indicate that co-expression of NPM-hMLF1 could suppress hFUS degradation but not caz degradation. The data implicate that decreased caz level is critical to induce the eye phenotype ( Fig. 2h-j), and that the caz function may be partially complemented by an increased level of hFUS ( Fig. 2k-m) 30 .

Co-expression of NPM-hMLF1 slightly increased the solubility of hFUS protein.
In transgenic flies carrying mutant or wild-type hFUS, there is no obvious evidence showing apparent FUS-immunopositive inclusions [31][32][33][34][35][36][37] . To further confirm these previous observations obtained by cytological and biochemical analyses, we examined a possible formation of FUS-immunopositive aggregates. We performed solubility tests on hFUS protein with extracts of adult heads carrying GMR-GAL4/Y; UAS-hFUS/UAS-GFP developed at 28 °C as summarized in Fig. 7a. This solubility test has been used to evaluate protein aggregates containing human TDP-43 (hTDP-43) protein in Drosophila cells 50 . Nearly 28 (mean; 28.03 ± 2.78) percent of the Input hFUS proteins in extracts of adult heads were recovered in low salt soluble fraction (LS) (Fig. 7b,c). Nearly 65 (mean; 65.08 ± 1.21) percent of the Input hFUS proteins were recovered in low salt insoluble and 1 percent Sarkosyl soluble fraction (SARK) (Fig. 7b,c). Nearly 7 (mean; 6.88 ± 1.58) percent of the Input hFUS proteins were recovered in 1 percent Sarkosyl insoluble and 7 M Urea soluble fraction (Urea) (Fig. 7b,c). No hFUS proteins were recovered in 7 M Urea insoluble fraction. Here, we thus confirmed the biochemical property of hFUS protein in Drosophila cells.
To examine the effects of co-expression of NPM-hMLF1 on the solubility of hFUS protein, we carried out solubility tests with extracts of adult heads of flies co-expressing hFUS and NPM-hMLF1 developed at 28 °C. Nearly 28 (mean; 28.43 ± 1.54) percent of the Input hFUS proteins were recovered in LS fraction (Fig. 7b,c). Nearly 68 (mean; 68.41 ± 1.97) percent of the Input hFUS proteins were recovered in SARK fraction (Fig. 7b,c). Nearly 3 (mean; 3.15 ± 0.42) percent of the Input hFUS proteins were recovered in Urea fraction (Fig. 7b,c). Here, we thus found that co-expression of NPM-hMLF1 slightly increased the solubility of hFUS protein.  Co-expression of NPM-hMLF1 partially suppressed pharate adult lethal phenotype induced by hFUS (Fig. 8). No eclosion was observed with flies carrying UAS-hFUS/UAS-GFP; D42-GAL4/+. Nearly 100 percent eclosion was observed with control flies expressing GFP under control of D42 GAL4. The eclosion rate increased to approximately 10 percent by co-expression of NPM-hMLF1. The data indicate that co-expression of NPM-hMLF1 partially rescued the defect induced by hFUS in motor neurons.

Discussion
In the present study, we observed a strong eye phenotype at 28 °C. Unexpectedly, the hFUS level decreased at 28 °C compared to 25 °C and this hFUS reduction was inhibited by feeding a proteasome inhibitor to flies. The data suggest that the strong eye phenotype at 28 °C is likely due to proteasome-dependent degradation of hFUS protein. Interestingly, co-expression of NPM-hMLF1 suppressed the eye phenotype with the increase in hFUS protein level, but not caz. In addition, we also revealed that co-expression of NPM-hMLF1 partially rescued pharate adult lethal phenotype induced by motor neuron-specific expression of hFUS, implying that NPM-hMLF1 may modify FUS-related FTLD pathology.
Negative feedback mechanisms associated with hFUS were reported 51,52 . For instance, it was reported that hFUS binds to the exon 7 region of its own gene to inhibit splicing of hFUS mRNA that results in reduction of hFUS expression 51 . The other study revealed a different negative feedback mechanism that is mediated by repressive miRNAs, such as miR-141 and miR-200a that hybridize to 3′ UTR of hFUS mRNA 52 . In addition to these, here we suggest a novel proteasome-dependent negative feedback mechanism of hFUS. It was reported that hFUS can be transported from cytoplasm to the nucleus that is mediated by transportin1 17 . Previous studies indicate that tight binding of transportin1 with hFUS in the nucleus leads to re-export of this complex to cytoplasm in FTLD pathology 53,54 . By this mechanism, this complex accumulates in cytoplasm, resulting in the depletion of hFUS in the nucleus 53,54 . Based on these reports, we can propose a negative feedback model on hFUS protein stability indicating that overexpression of hFUS may activate both this transportation system and proteasome-dependent depletion of hFUS. The accumulated hFUS in cytoplasm may also be degraded by proteasome and consequently hFUS in the nucleus can be depleted. Further analysis is necessary to clarify a detailed mechanism.
In the present study, it was shown that the co-expression of three genes (dMRJ, dMLF and hMLF1) exerts no effect on the eye phenotype induced by hFUS. Interestingly, both hFUS and NPM-hMLF1 were found to be co-localized in nucleoplasm of Drosophila eye imaginal disc cells in the flies co-expressing hFUS and NPM-hMLF1. Therefore, co-localization of NPM-hMLF1 with hFUS in nucleoplasm is likely critical for the suppression of the aberrant eye morphology phenotype induced by hFUS. It was reported that dMLF directly binds to RUNX transcription factor Lozenge and regulates the stability of Lozenge in a proteasome-dependent manner [55][56][57] . It should also be noted that NPM co-immunoprecipitates with hFUS in human hepatocellular carcinoma, HepG2 cells 58 . Considering these findings, NPM-hMLF1 possibly binds to hFUS to protect from degradation.
Pathological features in FTLD/ALS are represented by formation of neuronal and glial inclusions. TDP-43 is another DNA/RNA binding protein that is contained in the inclusion observed in FTLD patients similar to FUS 59 . Biochemical studies revealed that inclusions containing hTDP-43 in FTLD patients tissue are recovered in Urea soluble fractions 60 . Moreover, previous studies showed the evidence of Urea soluble hTDP-43 protein in transgenic flies expressing wild-type hTDP-43 50 . In contrast, inclusions containing hFUS are recovered in sodium dodecyl sulfate (SDS) fractions in tissues of human FTLD patients 10 . Here, we observed a higher level of Sarkosyl soluble hFUS protein than that of Urea soluble hFUS protein in transgenic flies expressing hFUS, suggesting that hFUS is less insoluble than hTDP43 in Drosophila cells as observed with human FTLD patients. In addition, it was also reported that increasing the solubility of hTDP-43 protein by co-expression of futsch closely correlates with extension of life span and improvement of locomotive defects in flies expressing hTDP-43 50 . Similarly to this report, we found that co-expression of NPM-hMLF1 slightly increased the solubility of hFUS protein. Furthermore, eye phenotype and pharate adult lethal phenotype induced by hFUS were suppressed by co-expression of NPM-hMLF1. These data indicate that refolding of hFUS protein aggregates by NPM-hMLF1 can be applied to novel therapy in human FTLD pathology.
Tukey test, was used to compare measures among three groups. *Indicates p < 0.05. Error bars indicate SEM. (d) Western immunoblotting analysis. Proteins were extracted from adult heads carrying GMR-GAL4 > UAS-GFP (GMR-GAL4/Y; UAS-GFP/+) and GMR-GAL4 > UAS-hFUS (GMR-GAL4/Y; UAS-hFUS/+). The blots were probed with anti-hFUS IgG or anti-caz IgG. The bands with an apparent molecular mass of 83 kDa and 72 kDa correspond to hFUS protein (upper panel). The bands with molecular mass of 45 kDa and 39 kDa correspond to two caz isoforms (lower panel) 29,30 . The flies were fed with fly food containing 0 μM, 5 μM or 10 μM of MG132 at 28 °C. α-tubulin was used as a loading control. (e) Quantification of relative hFUS protein level in two independent experiments (n = 2). hFUS protein levels were normalized with α-tubulin and shown relative to the one at GMR-GAL4 > UAS-hFUS (0 μM). Error bars indicate SEM. (f) Quantification of relative caz protein level in two independent experiments (n = 2). caz protein levels were normalized with α-tubulin and shown relative to the one at GMR-GAL4 > UAS-GFP (0 μM). Error bars indicate SEM. Establishment of transgenic flies. This method was adopted from a protocol described previously [63][64][65] .
pUAST-NPM-hMLF1 were constructed by inserting the NPM-MLF1 cDNA into the pUAST and then injected into embryos to obtain a stable transformant line carrying UAS-NPM-hMLF1 42,43 . P-element-mediated germline transformation was performed. F1 transformants were selected based on white eye color rescue. To generate pUAST-hFUS vectors, we used the Gateway ® Vector Conversion System (Thermo Fisher Scientific). The human FUS cDNA was subcloned into the pENTR ™ /D-TOPO ® vector (Thermo Fisher Scientific). To generate the Gateway destination vector pUAST-DEST, we inserted the Gateway cassette A sequence (Thermo Fisher  29,30 . The flies were developed at 28 °C. α-tubulin was used as a loading control. (b) Quantification of relative hFUS protein level in three independent experiments (n = 3). hFUS protein levels were normalized with α-tubulin and shown relative to the one at GMR-GAL4 > UAS-hFUS/UAS-GFP. Student t-test was used to compare measures between two groups. *Indicates p < 0.05. Error bars indicate SEM. (c) Quantification of relative caz protein level that were normalized with α-tubulin and shown relative to the one at GMR-GAL4 > UAS-GFP. Three independent experiments were carried out (n = 3). One-way ANOVA, followed by Tukey test, was used to compare measures among three groups. *Indicates p < 0.05. Error bars indicate SEM. Scientific) into the pUAST vector. The pUAST-hFUS vector was generated using Gateway recombination reactions (Thermo Fisher Scientific). To establish transgenic flies harboring UAS-hFUS, the pUAST-hFUS vector was injected into fly embryos using standard methods, by BestGene Inc.  Western immunoblotting analysis. Protein extracts from one hundred adult heads were prepared. Briefly, the adult heads were homogenized in a sample buffer containing 50 mM Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 0.1% bromophenol blue and 1.2% β-mercaptoethanol. The homogenates were incubated at 95 °C for 3 minutes, and then centrifuged. The supernatants were separated by electrophoresis on SDS-polyacrylamide gels containing 8% or 12% acrylamide and then the polypeptides were transferred to polyvinylidene difluoride membranes (Bio-Rad). The blotted membranes were subsequently blocked with TBS/0.3% Tween 20 containing 0.1% skim milk for 1 hour at 25 °C, and incubated with primary antibody for 16 hours at 4 °C. The following primary antibodies were used: mouse anti-FUS IgG at a 1:200 dilution (Santa-Cruz), rabbit anti-caz IgG at a 1:800 dilution and rabbit anti-hMLF1 IgG at a 1:200 dilution (Santa-Cruz) 30 . After extensive washing, the membranes were incubated with HRP-conjugated anti-mouse IgG (GE healthcare) or HRP-conjugated anti-rabbit IgG (GE healthcare) at a 1:10,000 dilution for 2 hours at 25 °C. The bound antibody was detected using ECL-select Western blotting detection reagents (GE healthcare) and images were analyzed with AE-9300H Ez-Capture MG (ATTO). To ensure equal protein loading in each lane, the membranes were also probed with an anti-α-tubulin IgG after stripping the complex of primary antibody and HRP-conjugated anti-mouse IgG or HRP-conjugated anti-rabbit IgG. For the detection of α-tubulin, mouse anti-α-tubulin monoclonal IgG at a 1:5,000 dilution (Sigma-Aldrich) and HRP-conjugated anti-mouse IgG at a 1:10,000 dilution (GE healthcare) were used as the primary and secondary antibodies, respectively. Solubility test. This method was adopted from a protocol described previously 50 . Briefly, one hundred adult heads were homogenized in 300 μl of low-salt buffer (10 mM Tris-HCl, 5 mM EDTA, 10% sucrose, pH 7.5).
Homogenates were centrifuged at 2,000 × g for 10 seconds. After discarding cellular debris and cuticle, 70 μl of the homogenate was set aside as Input fraction. The remaining 230 μl of homogenate was centrifuged at 25,000 × g for 30 minutes at 4 °C. The supernatant from the solution was set aside as LS fraction. The pellet was then extracted with Sarkosyl-containing buffer (10 mM Tris, 5 mM EDTA, 1% Sarkosyl, 10% sucrose, pH 7.5) and centrifuged at 180,000 × g for 20 minutes at 4 °C. The supernatant from the solution was set aside and designated as SARK fraction. The remaining Sarkosyl-insoluble pellet was then solubilized in Urea-containing buffer (30 mM Tris, 7 M Urea, 2 M Thiourea, 4% CHAPS, pH 8.5). The recovered fraction was named the Urea fraction. Protein concentrations were determined and 20 μg of each fraction was loaded on 8 or 12% SDS-PAGE. All buffers contained 2 × Complete Protease Inhibitor Cocktail (Roche) supplemented with 0.5 mM phenylmethylsulfonylfluoride to inhibit proteolysis and were filtered through 0.45 μm filter.
Immunostaining. For immunohistochemistry, eye imaginal discs from larvae were dissected and fixed in 4% paraformaldehyde/PBS for 30 minutes at 25 °C. These tissue samples were washed with PBS containing 0.3% Triton X-100; fixed samples were then incubated with Alexa 488-conjugated phalloidin (1unit/200 μl) (Thermo Fisher Scientific) in PBS containing 0.3% Triton X-100 for 20 minutes at 25 °C. The samples were then blocked with PBS containing 0.15% Triton X-100 and 10% normal goat serum (blocking buffer) for 30 minutes at 25 °C and incubated with primary antibodies in PBS containing 0.15% Triton X-100 and 10% normal goat serum for 16 hours at 4 °C. The following antibodies were used; mouse anti-FUS IgG at a 1:50 dilution (Santa-Cruz) and rabbit anti-hMLF1 IgG at a 1:50 dilution (Santa-Cruz). After washing with PBS containing 0.3% Triton X-100, the samples were stained with DAPI (0.5 μg/ml) (Thermo Fisher Scientific) in PBS containing 0.1% Triton X-100. After extensive washing with PBS containing 0.3% Triton X-100, samples were incubated with secondary antibodies labeled with Alexa 488 or Alexa 594 at a 1:400 dilution (Invitrogen) for 3 hours at 25 °C. After extensive washing with PBS containing 0.3% Triton X-100, samples were mounted in the Vectashield (Vector laboratories), and inspected using a confocal laser-scanning microscope (Olympus Fluoview FV10i). Proteasome inhibition. This method was adopted from a protocol described previously 66 . Flies were reared at 28 °C in a single plastic vial containing 5 mL of Drosophila instant medium (Formula 4-24, Blue, Carolina Biological Supply Company) with or without indicated amount of MG132 (Sigma-Aldrich). Briefly, one hundred newly eclosed adult flies were collected. In parallel, individual flies were used for Western immunoblotting analysis.
Scanning electron microscopy and stereomicroscopy. Adult flies were anesthetized, mounted and inspected under a scanning electron microscope (Keyence VE-7800) in high vacuum mode and stereomicroscope (Olympus SZX-ILLB100). In every experiment, the eye phenotypes of at least five newly eclosed adult male flies of each line were simultaneously examined by scanning electron microscope and stereomicroscope. Three independent experiments for each line were carried out and no significant variation in the compound eye phenotype was observed among individuals. Viability assay. This method was adopted from a protocol described previously 67,68 . Briefly, at least thirty first instar larvae were collected and transferred to fresh food. After eclosion, the rate of eclosed adult flies, dead pharate adults and dead larvae were calculated. Data analysis. GraphPad Prism version 6.02 was used for statistics analysis. Student t-test was used to statistically compare between two independent groups. One-way ANOVA followed by Tukey test was used to statistically compare measures among three groups. The significance between the variables was described based on the p-value: NS indicates p > 0.05, *Indicates p < 0.05. Values are described as means and error bars indicate standard error means (SEM). Data availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.