Chi and dLMO function antagonistically on Notch signaling through directly regulation of fng transcription

Gene apterous (ap), chip (chi) and beadex (bx) play important roles in the dorsal-ventral compartmentalization in Drosophila wing discs. Meanwhile, Notch signaling is essential to the same process. It has been reported that Ap and Chi function as a tetramer to regulate Notch signaling. At the same time, dLMO (the protein product of gene bx) regulates the activity of Ap by competing its binding with Chi. However, the detailed functions of Chi and dLMO on Notch signaling and the relevant mechanisms remain largely unknown. Here, we report the detailed functions of Chi and dLMO on Notch signaling. Different Chi protein levels in adjacent cells could activate Notch signaling mainly in the cells with higher level of Chi. dLMO could induce antagonistical phenotypes on Notch signaling compared to that induced by Chi. These processes depend on their direct regulation of fringe (fng) transcription.

opposite phenotypes on Notch signaling relative to Chi. Meanwhile, the functions of chi overexpression and bx RNAi on Notch signaling is limited to the regions far away from A/P boundary and this might be explained by that the interaction between Chi and dLMO is regulated by Dpp signaling. In addition, the function of Chi and dLMO on Notch signaling is dependent on their direct regulation of fng transcription. Our findings uncovered the detailed functions and the mechanism of how Chi and dLMO regulate Notch signaling in the Ap-independent manner. Meanwhile, the finding that Dpp could regulate the interaction between Chi and dLMO implied a potential crosstalk between Dpp and Notch signaling.

Results
Different Chi protein levels in adjacent cells are essential for Notch signaling activity. chi RNAi was found to induce serrated wing in a loss-of-function screen (Fig. S1a-S1c). To further study the function of Chi on Notch signaling, we generated a Chi antibody that was proved to work well in both immunofluorescence (IF) and immunoprecipitation (IP) assays ( Fig. S2a-S2b). chi e5.5 is a widely used chi mutation. To exclude the potential functions of 141 correct amino acids residual in chi e5. 5 , we generated a new chi mutant with 26 amino acids residual named chi 26 using Cas9/sgRNA system 28 (Fig. S2c-S2e'). Previous study suggested that the functions of Chi on Notch signaling are different at different larvae stages 27 . To figure out the details, we then did the Chi staining in the early and late third instar stage, respectively. The expression level of chi is higher in the D compartment than that in the V compartment at the early third instar larvae stage. The Notch signaling classical target gene cut expressed in the junctional area between the different Chi level regions ( Fig. 1a-a'''). However, in the late third instar, Chi is equally expressed throughout the wing discs (Fig. S3a). We reasoned that the different Chi levels might contribute to the Notch signaling regulation. So we first employed chi RNAi and chi e5.5 to induce different Chi levels in wing discs. Since the RNAi efficiency was good ( Fig. 1b-b'), Chi protein level should be higher at the outside of chi RNAi and chi e5.5 clones than that inside. The Notch signaling classical target genes in wing discs, cut and wg, were employed to monitor the Notch signaling activity. Consistent with previous reports, the Notch signaling activity only changed in the clones at the D compartment ( The statistical analysis showed that around 30% Cut upregulating cells located inside the clones, while around 70% located outside, indicating Notch signaling tends to be activated in the cells with higher Chi protein level (Fig. 1i). We then employed chi 26 to confirm all the aforementioned observations. All the results gotten from chi RNAi and chi e5.5 were reproduced in chi 26 clones ( Fig. 1g-h" and Fig. S3e-S3e'). In the Notch signal upregulating cells along the chi 26 clones, the majority of them located outside the clones (Fig. 1g-h"). Similar to the results from chi RNAi, the statistical analysis showed that around 30% Cut upregulating cells located inside the clones, while around 70% located outside (Fig. 1i).
We further reasoned that if the different Chi levels are critical for Notch signaling activation, the loss of Chi crossing over the D/V boundary should induce downregulation of Notch signaling along the D/V boundary. To knockdown the endogenous Chi at different regions of wing disc, we then employed ciGal4, hhGal4 or dppGal4 to drive chi RNAi overexpression specifically in the A, P compartment or A/P boundary regions, respectively. In these RNAi overexpression regions, Chi protein should be downregulated and the pattern of different Chi expression levels between D/V compartments should be lost. For example, when chi RNAi was overexpressed in the P compartment, the endogenous Chi was downregulated and the Chi protein should be at a very low level. The pattern of different Chi levels between D/V compartments in P compartment region should then disappear. IF staining showed that Notch target genes along the D/V boundary totally disappeared in the corresponding chi RNAi regions, indicating Notch signaling activation depends on the different Chi protein levels. In addition, since chi RNAi induced a low chi expression region compared to other regions, we also noticed that Notch signaling target genes were activated along the boundary of RNAi regions (Fig. S3f). ciGal4 and dppGal4 induced similar phenotypes as hhGal4, further confirming that the Chi protein level is essential for the Notch signaling activation ( Fig. S3g-S3i). To further confirm this conclusion, we induced chi RNAi overexpression in the whole disc with actinGal4 and noticed a total loss of Cut and Wg expressions ( Fig. 1j and Fig. S3j-S3k').

Chi overexpression induces Notch signaling activation mainly inside clones.
To further investigate the effect of Chi levels on Notch signaling, we employed AGGal4 to overexpress chi in the clones ( Fig. 2a-a'). Compared to the endogenous Chi protein outside the clones, the overexpressed Chi protein level was higher inside the clones. In the experiment, we found that Notch signaling was also upregulated along the clone boundary ( Fig. 2b-c"), only that around 70% of the Notch activated cells located inside clones (Fig. 2d). Together with loss-of-Chi results, we concluded that Notch signaling could be activated at the boundary of different Chi protein regions and tends to be activated in the cells with higher Chi protein level (Fig. 2e). In addition, chi overexpression induced by hhGal4 caused loss of Notch target genes along D/V boundary in the P compartment ( Fig. S4a-S4b). These phenotypes further verified that different Chi levels are essential for Notch signaling activation.
However, we noticed that phenotypes above could not be induced in the A/P boundary ( Fig. 2b-c"). To further investigate this phenotype, we employed dppGal4 to induce chi overexpression. This chi overexpression caused only slight loss of Notch target genes ( Fig. S4c-S4d). These phenotypes indicated that the regulation on Notch signaling by Chi levels might be regulated by signals locate in the A/P boundary, such as BMP signaling.
Chi is not involved in the Notch signaling transcription complex directly. Since Chi is a transcription cofactor, we set out to distinguish the genetic relationship between Chi and NICD. In the experiments, we found that NICD level was upregulated outside chi 26  Chi and dLMO induce opposite functions on Notch signaling. In order to demonstrate the regulation mechanism of Chi on Notch signaling, we set out to seek the potential factor(s) functioning with Chi. Since the phenotypes induced by Chi are limited to the D compartment, we reasoned two candidates, Ap and dLMO, which are reported to share different expression levels between D and V compartment 16,18,22 . We first confirmed the functions of these two proteins on Notch signaling. The RNAi or overexpression of ap induced slight changes in our experiment condition (Fig. S5a-S5f). However, bx RNAi induced cut and wg upregulation along the clone boundary in the regions away from A/P boundary and a large proportion of the Notch signaling activated cells are located inside clones (Fig. 3a-b'). These phenotypes induced by loss-of-dLMO are similar to that induced by gain-of-Chi. Meanwhile, the bx overexpression induced cut and wg upregulation along the clone boundary excepted that a large proportion of these Notch signaling activated cells located outside clones, similar to those induced by loss-of-Chi (Fig. 3c-d'). Previous study suggested that the expression of bx could downregulate fng and allows the expression of delta in the D compartment 27 . We then employed delta-lacZ and fng-lacZ. Along the boundary of the chi 26 and chi RNAi clones, delta-lacZ was upregulated, resembling those induced by the bx overexpression ( Fig. 3e-g'). In addition, the chi overexpression induced delta-lacZ downregulation in the whole clones, similar to that induced by overexpression of bx RNAi (Fig. 3h-i'). For fng-lacZ, in chi 26 and chi RNAi clones, lacZ was downregulated dramatically, similar to that induced by bx overexpression (Fig. 3j-l'). Taken together, Chi and dLMO induce exactly opposite functions on Notch signaling.
Chi and dLMO form a complex to regulate fng transcription. Since fng-lacZ is downregulated inside the clones of loss-of-Chi and gain-of-dLMO, we reasoned that Chi and dLMO might regulate fng transcription. To test this possibility, we first performed Co-IP to confirm whether there was an interaction between them as reported 26 . Since the functions of Chi and dLMO might be regulated by signals located in the A/P boundary, we performed this experiment with or without Dpp treatment, a morphogen that is expressed by the cells at the A/P boundary 29,30 . The results showed that Chi and dLMO could interact with each other and this interaction could be weakened by Dpp signal (Fig. 4a,b).
The next question we wanted to figure out was whether Chi and dLMO regulate the transcription of fng directly. We performed chromatin immunoprecipitation (ChIP) assay with Chi antibody and followed by qPCR. We designed 29 pairs of qPCR primers covering about 2,200 bases upstream of the transcription start site (TSS) and 2,000 bases downstream of TSS. Three peaks were observed in the result, indicating there are three binding sites between Chi and fng locus. The highest peak appeared at about 700 bases upstream of TSS. The other two peaks appeared at about 1,500 bases upstream of TSS and 800 bases after TSS (Fig. 4c). These data suggested that Chi directly promotes fng transcription in wing discs.
We also performed ChIP experiment for dLMO. Due to lack of endogenous dLMO antibody, we overexpressed HA-dLMO with ms1096Gal4 and then performed ChIP with HA antibody. The same peaks as those in Chi ChIP were also observed in the result, only that the lowest peak in dLMO ChIP assay was the highest peak in the Chi ChIP assay (Fig. 4d). All these data together with previous biochemical results indicate that Chi and dLMO are in a complex to regulate fng transcription directly. If this conclusion is right, the fng downregulation induced by bx overexpression should be revived by chi overexpression. We then overexpressed chi and bx together and noticed the fng-lacZ downregulation was partially revived (Fig. 4e-f '). In addition, the functions of bx overexpression on Notch target genes were partially revived as well (Fig. 4g-h'), indicating the roles of Chi and dLMO on Notch signaling were dependent on their regulation on fng.

Discussion
The Notch signaling plays a very important role in metazoan development. Since Notch was named a century ago, more and more detailed mechanisms are uncovered 5,31 . However, some important detailed information is still weakly defined. Such as, it is already known that Ap, Chi and dLMO are involved in the regulation of Notch signaling, Ap and Chi function together as a tetramer to regulate Notch signaling and Bx represses Ap's function by competing binding to Chi 22,26 . However, the detailed function of Chi and dLMO on Notch signaling remains unknown and the mechanism is also obscure.
Here we demonstrated the detailed functions and the mechanisms of Chi and dLMO in the Notch signaling. Different chi and bx expression levels in the adjacent cells are essential for Notch signaling activation. The roles of Chi and dLMO on Notch signaling depend on their regulation of the fng transcription. Both of them could bind to the same fragments in fng locus: 1.5k bases and 700 bases in the upstream of TSS and 800 bases in the downstream of TSS. In addition, the downregulation of fng expression induced by bx overexpression could be revived by chi overexpression. In our study, we observed that chi overexpression and bx downregulation could not induce the changes of Notch signaling in regions near A/P boundary. We noticed that fng is also not expressed in the A/P boundary region at the late stage of wing disc. In addition, we found that the binding between Chi and dLMO is weakened in the presence of Dpp. All these data indicated that Chi and dLMO function together to directly regulate the transcription of fng, this might be regulated by Dpp signals. It was reported that Notch2 signaling in the outer ciliary epithelium is required for maintaining bone morphogenetic protein (BMP) signaling in mouse eye 32 . Our results here offered a possibility that the crosstalk between BMP signaling pathway and Notch signaling pathway might partially depend on the interaction of Chi and dLMO.
The overexpression and downregulation of both chi and bx induce similar phenotypes on Notch signaling with just a slight difference in the locations of activated cells, further suggesting the proper amount of Chi and dLMO is essential in the wing discs development. Previous studies showed that both of dLMO and Ap bind with Chi and function in the Notch signaling 22,24 . Our ChIP assay results here showed that, the highest peak of Chi binding fragments on the fng locus appeared at about 700 bases in the upstream of TSS and this peak is the lowest on the ChIP experiment of dLMO. These results indicate that there might be another transcription factor in the complex with Chi to regulate fng transcription. However, Ap is excluded from the candidates, since in our experiment condition, it did not play an important role and showed only slight effects on Notch signaling. It will be very important to investigate this unknown factor(s) in the future study.
Although both loss-of-Chi/dLMO and gain-of-Chi/dLMO induced fng-lacZ downregulation, their roles on delta-lacZ are opposite and their functions on Notch signaling are not exactly the same. It is normal to reason that there are other mechanisms of how Chi and dLMO are involved in Notch signaling. One of the interesting things we have noticed in our experiments is that delta-lacZ upregulation induced in our experiments is along the clone boundary while downregulation is inside the whole clones. There is a feedback loop in Notch signaling, it might explain why delta-lacZ is upregulated along the clone boundary. However, it is hard to imagine that cells in the middle of the clones could respond to the activated Notch signaling along the clone boundary. These data implied that Chi and dLMO might also regulate delta transcription.  (f-f '). In all IF results, clones are marked with GFP (green) and circled with dashed line. (g-h') Immunostaining of Cut (g-g'; red) and Wg (h-h'; red) in the discs of indicated genotype flies. Compared to the overexpression of HA-bx alone (Figure 3 c-d'), Cut and Wg activation could be partially rescued by the chi overexpression. Clones are marked with GFP (green). (g',h') are the enlarged pictures of (g,h). In summary, we uncovered the detailed functions of Chi and dLMO on Notch signaling. Notch signaling tends to be activated along the adjacent cells with different Chi levels in the D compartment of wing discs. The chi overexpression and bx RNAi could not function on Notch signaling in the A/P boundary. This might be explained by that Dpp could regulate the interaction between Chi and dLMO. In addition, the function of Chi and dLMO on Notch signaling is depended on the regulation of fng transcription. The antagonistical regulation of Chi and dLMO on Notch signaling through fng expression is important for wing development.
ChIP-qPCR. ChIP assay was performed as previously described with some modifications (for details, please see Supplemental Information) 36,37 . Primer pairs and detailed protocols used in this study are listed in the Supplemental Information.
Immunostaining of wing discs. The larvae were heatshock at 37 °C for 30 minutes after birth for 72 hours.
The wing discs were dissected for immunostaining with standard protocol as previously described. The antibodies information are listed in the Supplemental Information. Confocal imagings were collected using a Leica TCS system and processed by Adobe Photoshop and Image J.
Cell culture, transfection and Western blotting. S2 cells were cultured in Drosophila Schneider's Medium with 10% fetal bovine serum, 100 U/ml of penicillin, and 100 mg/ml of Streptomycin (Invitrogen). Plasmids were transfected with LipofectAMINE (Invitrogen) according to manufacturer's instructions. Immunoprecipitation and Western blot were carried out according to standard protocols as previously described (Jin et al., 2012). Antibodies used in this study were as follows: mouse anti-flg (1:5000; Sigma), mouse anti-HA (1:5000; Sigma), rabbit anti-Chi (produced by immunizing rabbits with the whole protein). S2 cells were treated with the Recombinant Drosophila Decapentaplegic (R&D systems) according to the instruction for 6 hours before harvesting. siRNA target of chi is the first 500 bases downstream of TSS. siChi were generated according to the instruction of in vitro transcription T7 kit (Takara).