Apontic directly activates hedgehog and cyclin E for proper organ growth and patterning

Hedgehog (Hh) signaling pathway and Cyclin E are key players in cell proliferation and organ development. Hyperactivation of hh and cyclin E has been linked to several types of cancer. However, coordination of the expression of hh and cyclin E was not well understood. Here we show that an evolutionarily conserved transcription factor Apontic (Apt) directly activates hh and cyclin E through its binding site in the promoter regions of hh and cyclin E. This Apt-dependent proper expression of hh and cyclin E is required for cell proliferation and development of the Drosophila wing. Furthermore, Fibrinogen silencer-binding protein (FSBP), a mammalian homolog of Apt, also positively regulates Sonic hh (Shh), Desert hh (Dhh), Cyclin E1 (CCNE1) and Cyclin E2 (CCNE2) in cultured human cells, suggesting evolutionary conservation of the mechanism. Apt-mediated expression of hh and cyclin E can direct proliferation of Hh-expressing cells and simultaneous growth, patterning and differentiation of Hh-recipient cells. The discovery of the simultaneous expression of Hh and principal cell-cycle regulator Cyclin E by Apt implicates insight into the mechanism by which deregulated hh and cyclin E promotes tumor formation.

the wing disc 5 . This contradiction suggests that other factors are involved in regulating the expression of cyclin E. Therefore, it is fruitful to investigate the regulation of cyclin E in the wing disc and the relationship between Cyclin E and Hh pathway.
Apt has been identified as a transcription factor involved in development of tracheae, head, heart and nervous system [22][23][24][25] . Apt can suppress metastasis 26 and is required in the nervous system for normal sensitivity to ethanol sedation 27 . Moreover, Apt participates in JAK/STAT signaling pathway to limit border cells migration 28 . The human homolog of Apt, FSBP, is a cancer-related factor that is expressed in many tissues 29,30 . However, the role of Apt in the organ growth and patterning is unknown.
In this study, we unveiled a fundamental role of Apt in growth and patterning of the wing disc through coordinated expression of morphogen hh and cell cycle regulator cyclin E. Both loss of function and overexpression of apt resulted in defective wings. Further studies demonstrated that loss of apt function attenuated the expression of hh and cyclin E, while apt overexpression upregulated hh and cyclin E. Mutating the inherent Apt binding sites in the promoter region of hh and cyclin E compromised the expression of hh and cyclin E. Collectively, Apt directly activates the expression of hh and cyclin E to allow proper wing development. In addition, we found that Apt-dependent expression of hh and cyclin E is evolutionarily conserved in human cells.

Results
Apt is expressed in the wing disc and is required for wing development. As the first attempt to investigate the function of apt during wing development, we analyzed apt expression pattern in the wing disc by immunostaining using anti-Apt antibody. In the wing disc, Apt was detected in PE cells as revealed by co-localization with a PE marker Ubx (Fig. 1A). Apt was also detected in DP cells (Fig. 1B). These data clearly demonstrate that Apt is expressed in both the PE and DP of the wing disc, suggesting its possible role in wing development.
To analyze the role of Apt during wing development, we would examine the developing wing of homozygous apt null mutant. However, apt null homozygotes die as embryos 22 . Therefore, we firstly examined the phenotype of apt knockdown using an MS1096-GAL4 driver. RNAi-mediated knockdown of apt resulted in a small wing, and also reduced the width between vein 3 and vein 4 ( Fig. 1C-E). Overexpression of a dominant-negative form of Smoothened (Smo −PKA ) caused a "fused wing" phenotype [31][32][33] . Knockdown of apt enhanced the "fused wing" phenotype ( Fig. 1F, arrowhead indicates the "fused wing" phenotype and arrow indicates enhancement of the "fused wing" phenotype). Furthermore, we induced apt loss of function mutant clones in the wing disc using the FLP/FRT system 34 . The formation of these clones resulted in a small wing with a blistered phenotype (Fig. 1H) compared with the control wing (Fig. 1G). To investigate the effect of apt overexpression, we employed the MS1096-Gal4 driver. Abnormal wings were induced by overexpression of apt (Fig. 1I). The wing was diminished and blistered, and the pattern of veins was disrupted and extra abnormal bristles were induced in the wing margin. In addition, when apt was overexpressed by a stronger driver (sd-Gal4), both wings and halters were lost (Fig. 1J). Taken together, the loss-of-function and overexpression analyses indicate that Apt is indispensable for wing development.
Apt activates the expression of hh in the wing disc. Given that the space between vein 3 and vein 4 is a characteristic monitor of Hh activity 21,35 , the observed narrowing the space between vein 3 and vein 4 upon knockdown of apt ( Fig. 1C-E) implies that Apt can modulate expression of hh in the wing disc. The enhanced dominant-negative phenotype of Smo −PKA by knockdown of apt (Fig. 1F) supports the notion. To examine the relationship between apt and hh, we first compared the expression of apt and hh, and found that Apt and hh-lacZ were co-expressed in PE cells ( Fig. 2A-C) and P compartment cells of the DP (Fig. 2D-F) in the early third instar larval disc. Furthermore, apt exhibited genetic interaction with hh. Ninety-seven percent of hh bar3 mutant (n = 40) showed slightly reduced area between L3 and L4 (Fig. 2H) and the remaining three percent showed wing blistering phenotype ( Supplementary Fig. S5B). While heterozygotes of apt null allele showed normal wings (Fig. 2G), the same heterozygotes under the hh bar3 background exhibited more severe phenotypes of reduced L3-L4 area and smaller wing with blister ( Fig. 2I), which reproduced the apt loss of function phenotype (Fig. 1H). Transheterozygotes of two sets of hh alleles (hh bar3 /hh 2 and hh Mir /hh 2 ) showed a smaller wing with an extra crossvein ( Supplementary Fig. S1A-E), demonstrating that it is a loss of function phenotype of hh. While wings of animal heterozygous for hh 2 or apt null mutant were normal, trans-heterozygotes of the apt null allele and hh 2 showed the same wing phenotype ( Supplementary Fig. S1F-I). These results suggest that Apt regulates the expression of hh. To address the issue directly, we analyzed the expression of hh under loss-of-function and overexpression of Apt. The expression of hh-lacZ was significantly reduced in the apt mutant clones in the PE ( Fig. 3A-C) and the DP (Fig. 3D-F). In addition, the expression of hh-lacZ also decreased in the apt-knocked down region (Fig. 3G-I). By contrast, overexpression of Apt increased the expression of hh-lacZ ( Fig. 3J-L). Moreover, Apt regulates the mRNA levels of hh and its target dpp ( Supplementary Fig. S2). These results demonstrate that Apt activates the expression of hh.
Apt directly controls hh in the wing disc. To address how Apt activates the expression of hh, we focused on a 15-kb region of the hh locus known to reproduce the normal hh expression pattern in the wing disc 36 . We identified one potential Apt binding sequence 25 within the region (Fig. 4A). Chromatin immunoprecipitation (ChIP) assays using early third instar wing discs detected Apt protein on the predicted wild-type Apt binding site but not in the regions upstream and downstream of the binding site ( Fig. 4A and D). We next assessed the function of the Apt-binding site in hh using a CRISPR-Cas9 system 37 . Since the designed gRNA contained the Apt-binding site, four Apt-binding site deletion mutants and two insertion mutants were generated (  (Fig. 4L). Taken together, these data demonstrate that Apt directly activates transcription of hh in the wing disc for proper wing development.
Apt activates the cyclin E expression in the wing disc. We have reported that Apt induces the cyclin E expression in the eye disc 38 . Therefore, we examined whether Apt consistently regulates cyclin E in the wing disc. To do this, we performed a double-staining experiment using Apt antibody and Cyclin E antibody. In the wild-type wing disc, Apt and Cyclin E were co-expressed ( Fig. 5A-C). Furthermore, the expression of Cyclin E was significantly reduced in the apt mutant clones (Fig. 5D-F). Compared with control disc (Fig. 5G), apt knockdown decreased Cyclin E level (Fig. 5H), while apt overexpression increased Cyclin E (Fig. 5I). In addition, the Apt directly controls cyclin E in the wing disc. Since Apt directly activates the expression of cyclin E in the eye disc, we anticipated a direct role of Apt in the expression of cyclin E also in the wing disc. This expectation was verified by transgenic reporter assays. The reporter gene 38 carries the endogenous promoter and the cyclin E regulatory element containing a wild-type Apt-binding site (cycEPlacZ) or a mutated site (cycEMPlacZ). cycE-PlacZ with the wild type binding site recapitulated the cyclin E expression in the wing disc ( Fig. 5J-L). However, base substitutions in the Apt-binding site in cycEMPlacZ abolished the lacZ expression ( Fig. 5M-O). These results indicate that Apt directly activates cyclin E through its binding site in the regulatory region of cyclin E.
Apt is a growth sensor to control organ growth and patterning. Because both Hh and Cyclin E are involved in cell death and cancer [39][40][41] , we asked whether the overexpression phenotypes are caused by apoptosis. To test this, we investigated apoptosis in wing discs by staining with anti-Caspase-3 antibody. In the third instar wing disc, apt mutant clones showed few apoptotic cells (Fig. 6A). However, in the wing disc from an Apt-overexpressed larva, the number of apoptotic cells significantly increased compared with a control disc ( Fig. 6B and C). This presumably explains why wing size was reduced upon strong overexpression of Apt ( Fig. 1K and L). Besides, we examined the growth of wing discs upon apt knockdown in the dorsal region using an ap-GAL4 driver. Compared with a control disc (Fig. 6D), apt knockdown region exhibited growth disadvantage ( Fig. 6E and H). Overexpression of Apt using the ap-GAL4 driver resulted in severely reduced dorsal region (Fig. 6F). When we inhibited cell death by simultaneous overexpression of a Caspase inhibitor P35, we observed outgrowth of cell layers from the disc in the apt and p35 overexpressed region ( Fig. 6G and H). Homozygotes of hh mutations for the Apt-binding site exhibited the small wing but not the blistered phenotype. However, hh and cyclin E double mutant recapitulates the smaller and blistered wing. While CycE 2 /+ flies showed normal wings, three percent of hh bar3 /hh bar3 and eighteen percent of CycE 2 /+; hh bar3 /hh bar3 flies showed the smaller and blistered phenotypes ( Supplementary Fig. S5A-C). We also observed genetic interaction between   Supplementary Fig. S5D-F). Collectively, these data suggest that Apt controls wing development by inducing appropriate amounts of Hh and Cyclin E.

FSBP positively regulates Shh and cyclin E in human cells. FSBP, the mammalian homologue of
Drosophila Apt, is a cancer related factor. To examine whether FSBP regulates Shh and cyclin E, we used human 293T cells to knockdown or overexpress FSBP and analysed the mRNA levels of Shh, its signaling pathway genes and cyclin E. After transfection of FSBP siRNA, the mRNA level of FSBP decreased nearly 60 percent compared with mock. Under the condition, we observed marked decrease in the mRNA levels of Shh and Shh signaling pathway genes such as Ptch, Gli and Hhip (Fig. 7A). The levels of cyclin E (CCNE1 and CCNE2) mRNA showed less prominent but statistically significant decrease. When we overexpressed FSBP, mRNA level of FSBP increased nearly 7.5 folds, and that of Shh increased dramatically 9 folds. The mRNA levels of Shh targets and cyclin E were also increased upon overexpression of FSBP (Fig. 7B). Interestingly, FSBP also regulates the expression of Dhh, but not Ihh. Taking together, these data suggest that the regulation of hh/Shh and cyclin E by Apt/FSBP is conserved from Drosophila to humans.

Discussion
Morphogen Hh and cell cycle regulator Cyclin E control growth and patterning in vertebrate and invertebrate. Here, we unravel a fundamental role of transcription factor Apt/FSPB as a conserved regulator of hh/Shh and cyclin E/CCNE. During Drosophila wing development, Apt directly activates the expression of hh and cyclin E to control wing growth and patterning. Both loss-of-function and overexpression assays clearly demonstrated that Apt is vital for wing development. Further studies showed that loss of apt function attenuates, while overexpression of apt activates the expression of hh and cyclin E. Moreover, we found that the homolog of Apt, FSBP, can positively regulate Shh and its pathway genes, and CCNE in human cells.
Hyperactivation of Hh pathway and cyclin E has been implicated in many tumors 16,39 . In contrast, during development, cell proliferation must be precisely regulated and coordinated with the processes of cell patterning and differentiation, which are also regulated by Hh and Cyclin E 41,42 . This delicate balance is probably maintained by Apt-mediated proper expression of Hh and Cyclin E. Indeed, overexpression of Apt in the presence of apoptosis inhibitor P35 generated tumor-like outgrowth of cell layers in the wing disc. Apt-dependent expression of hh and cyclin E can direct proliferation of Hh-expressing cells and simultaneous growth, patterning and fate specification of Hh-recipient cells. Although mechanisms are quite different, this provides similar effects as an asymmetric division of a stem cell.
To assess the importance of the Apt-binding site in the promoter region of hh, we first tried a transgenic reporter assay. However, the regulatory region of hh encompassing the upstream region and the 1st intron (~15 kb) 36 is too large to make a reporter construct for conventional P-element mediated transgenesis. Therefore, we employed the CRISPR-Cas9 system 37 to mutagenize the endogenous Apt-binding site in the hh promoter. All 6 independent mutants exhibited the same phenotypes (reduced expression of hh, reduced wing size and the space between L3 and L4), suggesting that the observed phenotypes are not due to off-target effect of Cas9. Nevertheless, we inspected the possibility of off-target effect. Since our gRNA carries the binding sequence for Apt, a binding site of Apt in other than the hh promoter could be the most likely candidate for off-target. However, all the 6 mutants showed the wild type sequence around the Apt-binding site in the cyclin E promoter (Supplementary Fig. S6). Furthermore, we observed clear genetic interactions between hh ΔaptDB1 and other hh mutants. Taken together, these data strongly suggest that the observed phenotypes are not due to off-target effect.
Although the expression of Apt and Hh overlapped in the P compartment of wing disc, how Apt specifically induces hh in the P compartment is still not clear. Since Ci R has been known to repress the expression of hh in anterior cells 14 , Ci R may interrupt the activation of hh by Apt in anterior cells. While our data strongly support that Apt is a transcription factor of hh, mutating the Apt-binding site on hh promoter alone induced the weak phenotype. However, the binding site mutation showed strongly enhanced "fused wing" phenotype in the background of overexpression of Smo domain-negative form (Smo −PKA ). These observations suggest that besides Apt, other factor(s) might also regulate hh transcription during development. Therefore, both knockdown and overexpression of Apt only moderately affected the expression of hh. Hh, as an important morphogen, plays multifaceted roles in segmentation and wing patterning. Previous findings paid more attention on the protein modification of Hh but the regulatory mechanism underlying hh transcription was not well understood. Here we identified Apt as the first regulatory factor that directly activates hh transcription. third instar larvae from yw or hh ΔaptDB1 . Error bars, SEM from three independent experiments. Student's t tests, ***p < 0.001. (F-L) Deletion of the Apt-binding site in the hh promoter affects wing development. The wing size and the intervein region between L3 and L4 (control value was set as 100%) were decreased in hh ΔaptDB1 . Error bars, SEM. Student's t tests, ***p < 0.001. hh 2 /+ (G) or hh ΔaptDB1 /+ (H) adult wings show normal phenotype. All adult wings of hh ΔaptDB1 /hh 2     Generation of CRISPR constructs. To induce mutations in the Apt-binding site in the hh promoter region, we used a Cas9-gRNA system. We designed gRNA in the hh promoter region carrying the binding sequence of Apt (Fig. 3A). The corresponding sequence was introduced into the pBFv-U6.2 vector and the gRNA transgenic flies were generated as described 37 . gRNA females were crossed to Cas9 males to obtain the founder animals. Male founders were crossed to female balancer. Offspring male flies were balanced and stocked. Genomic DNA was extracted from each offspring male and used for molecular characterization. PCR primers were designed to construct gRNA expression vectors and to amplify the promoter region of cyclin E (Supplementary Table S1).

Chromatin immunoprecipitation (ChIP).
ChIP assays of wing discs were performed as previously described 43 . Briefly, 100 early third instar wing discs were dissected in PBS and fixed by 1% formaldehyde at room temperature for 20 minutes. Sonicated chromatin was immunoprecipitated using 10 μl anti-Apt antibody. Quantitative PCR using 4 μl of the purified DNA. Scientific REPORTS | 7: 12470 | DOI:10.1038/s41598-017-12766-w RT-qPCRanalysis. Total RNAs were prepared from the dissected tissues using an RNAprep Pure Tissue kit TIANGEN #DP431). cDNAs were synthesized using a Prime Script TM II1 st strand cDNA synthesis kit (TaKaRa #6210A). The real-time qPCR was conducted with Bio-Rad CFX96 real-time system using a SuperRealPreMix Plus (SYBR Green) Kit (TIANGEN #FP205) in a 20 ul reaction containing 2 pmol of relevant primers. The amount of mRNA was normalized to that of control tubulin mRNA. PCR primers used are shown in Supplementary Table  (Supplementary Table S1).
Cell culture and transfection. 293 T cells were cultured in DMEM (Gibco) containing 10% fetal bovine serum and 100 U/ml of penicillin/streptomycin. For RNA interference experiment, FSBP siRNA was designed by ourselves, the sequences were: siFSBP-F: 5′-GCCUGGUAAGAGACAGGAAdTdT-3′, siFSBP-R: 5′-UUCCUGUCUCUUACCAGGCdTdT-3′. After cells were cultured for 24 h in 12-well plate, the culture medium was changed to serum-free medium. Mock is no siRNA treatment. siRNA duplexes were transfected at a final concentration of 20 nM using lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Cells are harvested for real-time PCR after cultured for 48-96 h with serum containing medium. For overexpression experiment, transfection was carried out using PEI (polyethylenimine) transfection method. 293 T cells were transfected in 60 mm plates with 5 µg plasmid pcDNA3.1-FSBP, which was constructed by our lab, and pcDNA3.1-HisA/V5 plasmid was transfected as control. Forty-eight to ninety-six hours after transfection, cells are harvested for real-time PCR analyses with standard protocols. The primers were used as showing in the Supplementary Table S1.
Microscopy and image treatment. Images were acquired in Olympus FV1200 confocal microscope and Olympus cellSens, treated with Adobe Photoshop CS6 image programs. Wing size and space between vein 3 and vein 4 were measured on each picture using the ImageJ computer program.

Statistical analysis.
Results are given as means SEM; each experiment included at least three independent samples and was repeated at least three times. Group comparisons were made by two-tailed unpaired Student's t-tests. *P < 0.05; **P < 0.01, and ***P < 0.001.