Rap1, AF6 and Pak4 cooperate with Par3 to promote ZA assembly and remodeling in the fly photoreceptor

The epithelial Zonula adherens (ZA) is a main adhesion compartment that enables organogenesis by allowing epithelial cells to assemble into sheets. How ZA assembly is regulated during epithelial cell morphogenesis is not fully understood. We show that during ZA morphogenesis, the function of the small GTPase Rap1 and the F-actin binding protein AF6/Cno are both linked to that of the P21-activated kinase Pak4/Mbt. We find that Rap1 and Mbt regulate each other’s localization at the ZA and cooperate in promoting ECadherin stabilization. During this process Cno regulates the recruitment of Baz at the ZA, a process that is also regulated by Arm phosphorylation by Mbt. Altogether, we propose that Rap1, Cno and Mbt regulate ZA morphogenesis by coordinating ECadherin stabilization and Baz recruitment and retention. In addition, our work uncovers a new link between two main oncogenes, Rap1 and Pak4/Mbt, in a model developing epithelial cell.


Introduction
The epithelial ZA consists of a lateral circumferential belt of Adherens Junction (AJ) material that allows for epithelial cells to assemble into sheets.
Loss of epithelial adhesion is a hallmark of cancer and there is therefore a strong interest in better understanding how ZA morphogenesis, remodeling and maintenance are regulated. The adhesion molecule ECadherin/Shotgun (ECad) and its effector βcatenin/Armadillo (Arm) are main AJ components of the ZA in animal epithelial cells. Work in Drosophila and vertebrate cells points to multiple pathways that regulate AJ material morphogenesis during epithelial morphogenesis, including membrane delivery, endocytosis, local accumulation and stabilization at the plasma membrane (Bryant and Stow, 2004;Tepass, 2012). However, we still lack an integrated view of how these pathways and the corresponding molecular players come together to regulate ZA morphogenesis and remodeling during development.
The fly retina has long been used as a model system to study the genetic and molecular basis of ZA morphogenesis and remodeling during organogenesis.
During pupal development, photoreceptors build a new ZA to accommodate the nascent apical light-gathering structure, called the rhabdomere (Ready, 2002). During this process the Par complex, which consists of Cdc42-Par6-aPKC and Par3/Bazooka (Baz), regulates the specification of the photoreceptor cortex and plasma membrane into a sub-apical domain (stalk membrane) and ZA (Hong et al., 2003;Nam and Choi, 2003;Walther et al., 2016;Walther and Pichaud, 2010). During ZA morphogenesis, phosphorylation of Baz by aPKC at the conserved S980 leads to the apical exclusion of P-S980-Baz, a step of molecular sorting that also depends on Crumbs (Crb) capturing Par6-aPKC and Stardust (Sdt) at the stalk membrane (Krahn et al., 2010;Morais-de-Sa et al., 2010;Walther and Pichaud, 2010).
Confined to the apical-lateral border of the cell, P-S980-Baz is thought to promote ZA assembly by binding to Arm and Echinoid (Ed), two main AJ components (Wei et al., 2005).
Next to Baz, the Cdc42 effector type-2 p21-activated kinase Mushroom bodies tiny (Mbt/Pak4) has also been shown to regulate ZA morphogenesis in several epithelial cell types by promoting Baz retention at the ZA and regulating the accumulation of the ECad-Arm complex via phosphorylating βCat/Arm (Jin et al., 2015;Law and Sargent, 2014;Menzel et al., 2008Schneeberger, 2003Walther et al., 2016). In vertebrate cells, Pak4 has also been linked to the Par complex via Par6b phosphorylation, indicating a potential cross talk between Par6b-aPKCγ/ζ and Pak4, downstream of Cdc42 (Jin et al., 2015;Wallace et al., 2010). While in fly photoreceptors both Mbt and Baz are main regulators of AJ accumulation at the developing ZA, they seem to operate as part of parallel convergent pathways. This is demonstrated by the fact that while Arm accumulation is reduced in mbt null mutant photoreceptors, AJ material is no longer present at the plasma membrane in cells mutant for both mbt and baz (Walther et al., 2016). How exactly Baz contributes to regulating AJ material accumulation at the ZA is not fully understood. Similarly, phosphorylation of βCat/Arm by Pak4/Mbt (Law and Sargent, 2014;Menzel et al., 2008) cannot fully account for mbt function during ZA morphogenesis, as re-introducing a phosphomimetic form of Arm does not rescue the loss of mbt function (Walther et al., 2016). Therefore other factors must regulate AJ morphogenesis during ZA maturation, either in concert with Mbt and Baz, or as part of the Mbt and Baz pathways.
An interesting candidate in contributing to AJ accumulation at the epithelial ZA is Rap1. This member of the Ras subfamily of small GTPases has been shown to localize at the AJ in various fly epithelia, and to be an essential AJ regulator (Boettner et al., 2003;Boettner and Van Aelst, 2007;Choi et al., 2013;Knox and Brown, 2002;O'Keefe et al., 2009;Spahn et al., 2012;Wang et al., 2013). In the early embryo, Rap1 and its effector F-actin binding protein AF6/Cno (Boettner et al., 2003;Mandai et al., 2013;Sawyer et al., 2009) (Hogan et al., 2004). Here we sought to examine the relationship between Rap1, its GEF Dizzy/PDZ-GEF and the protein network that drives epithelial apical cortex and plasma membrane specification using the pupal photoreceptor as a model system.

Dizzy and Rap1 regulate pupal photoreceptor ZA morphogenesis
In the fly retina, Rap1 has been previously shown to regulate AJ remodeling between newly specified photoreceptors, and between retinal accessory cells (cone and pigment cells) (O'Keefe et al., 2009). To test whether dizzy and Rap1 are required during ZA morphogenesis in the pupal photoreceptor, we made use of the rap1-Rap1::GFP and dizzy-Dizzy::GFP transgenes, which allow for expression of these proteins under their endogenous promoter. We found that Rap1::GFP accumulates predominantly at the developing pupal photoreceptor ZA ( Figure 1A), and can also be detected at lower levels at the apical photoreceptor membrane, which includes the stalk membrane. were also significantly reduced in Rap1IR photoreceptors. Furthermore, Cno could no longer be detected at the ZA ( Figure 1C'') and Mbt levels were significantly decreased when compared to wild type ( Figure 1D'' and 1G'). In some cases, ZA domains were present that did not contain Mbt, resulting in a significant alteration of the ratio between Mbt and Arm. In wild type, ZA levels of Mbt/Arm are correlated and follow a normal distribution. In Rap1IR ZA this correlation was disrupted, with significantly more junctions presenting either high or low Mbt/Arm ratios that fall outside of a normal Gaussian distribution ( Figure 1H). In these shortened ZA, levels of Arm and Baz were comparable to wild type ( Figure 1G, 1G'') and the ratios of Baz/Arm in Rap1IR cells remained similar to wild type ( Figure 1H'). Apical levels of F-actin ( Figure   1C'''), aPKC ( Figure 1D'''), and Crb ( Figure 1E'''), were not affected in Rap1IR photoreceptors when compared to wild type. These data indicate that Rap1 is required for the accumulation of AJ material at the developing ZA. This function appears most critical when considering the ZA levels of Cno and Mbt, Next, to examine the function of dizzy during ZA morphogenesis, we made use of the strong dizzy Δ12 allele. We found that reducing dizzy expression leads to a phenotype similar to that seen in Rap1IR photoreceptors, including a shortening of the ZA along the apical-basal axis ( Figure

Rap1 promotes AJ stabilization during ZA remodeling
We have previously shown that in pupal photoreceptors, loss of mbt function leads to an increase in the mobile fraction of ECad at the ZA when compared to wild type over 250 secondes (Walther et al., 2016). Our analysis of Rap1IR indicates that Mbt accumulation is reduced in the corresponding ZA ( Figure   1D'' and 1G'), which should therefore be accompanied by an increased in ECad mobility. To assess whether this is the case, we made use of FRAP and compared the recovery after photo-bleaching of a ubi-ECad::GFP transgene in wild type and Rap1IR photoreceptors. In wild type cells, over approximately 250 sec, we estimated that 25% of ECad::GFP is mobile, which is consistent with previous estimations from our lab (Walther et al., 2016) (not shown).
However, while ECad::GFP shows a stronger recovery over this relatively short time scale in Rap1IR when compared to wild type, the GFP signal failed to plateau (not shown), preventing us from extrapolating the mobile fraction.
We therefore performed FRAP over a longer time scale (1000 sec). Over this long time scale, we found approximately 35% of ECad::GFP is mobile in wild type ZA, while ~70% is mobile in Rap1IR photoreceptors (Figure 2A-C).
These data indicate that Rap1 promotes ECad stabilization at the ZA, and are compatible with Mbt mediating part of Rap1 function during this process.

Cno couples Arm and Baz at the ZA and is required for the apical accumulation of aPKC and Crb
Next to regulating Mbt accumulation at the ZA, one likely mechanism whereby Rap1 might promote ECad stabilization is through the F-actin linker Cno (Kooistra et al., 2007). In the pupal photoreceptor, Cno localizes at the ZA and this localization is also strongly decreased in Rap1IR photoreceptors ( Figure   1C''). Decreasing cno expression using the strong cno R2 allele leads to delamination of the mutant photoreceptors through the floor of the retina ( Figure 3A-B), a phenotype resembling that obtained when strongly reducing These data further indicate that during ZA morphogenesis, the function of Rap1 and mbt are interlinked.
To complement these experiments, we next asked whether decreasing Rap1 expression could modify the mbt phenotype. Combining the null allele mbt P1 to Rap1IR led to a very strong additive effect as nearly all photoreceptors delaminated from the retina, a phenotype due to strong defects in recruiting cone and pigment cells around the photoreceptor clusters ( Figure 5D).
Nevertheless, a majority of the delaminated, photoreceptors still presented Arm domain linking flanking photoreceptors ( Figure 5D'''' and 5E). Altogether, these genetic experiments argue in favor of Rap1 functioning together with mbt during Baz-dependent ZA morphogenesis.

Discussion
In the pupal photoreceptor, ZA morphogenesis is orchestrated by a conserved protein network that includes the Par complex, Crb and its binding partners Sdt and PATJ, together with the lateral kinase Par1 (Berger et al., 2007;Hong et al., 2003;Izaddoost et al., 2002;Nam and Choi, 2003;Pellikka et al., 2002;Richard et al., 2006;Walther et al., 2016;Walther and Pichaud, 2010 (Hogan et al., 2004). Our work raises the possibility that Pak4 is one of the downstream effector of Cdc42 during this process.
Rap1 and cno have been shown to regulate apical-basal polarity in the cellularizing embryo, a system that allows for examination of the net contribution of the Par complex and AJ material toward epithelial cell polarization (Choi et al., 2013)

Analysis of gene function
Clonal analysis of mutant alleles in the retina was performed using the standard FLP-FRT technique (Xu and Rubin, 1993) with appropriate FRT, ubi-GFP chromosomes used to generate negatively marked mutant tissue in combination with eyFLP (Newsome et al., 2000). Retina expressing RNAi in clones were generated using the coinFLP system (Bosch et al., 2015).
Clones of retinal tissue expressing RNAi against Rap1 were generated both with and without UAS-dicer, while clones of retinal tissue expressing RNAi against cno were generated without UAS-dicer only. In order to mitigate the strong Rap1 loss of function phenotype, Rap1IR animal were raised at 20 degrees and shifted to appropriate temperature (25 or 29 degrees) at puparium formation.

Antibodies and immunological methods
Whole mount retinas at 40% after puparium formation (APF) were prepared as previously described (Walther and Pichaud, 2006). VectaShield™ with or without DAPI as appropriate and imaging was performed using a Leica SP5 confocal. Images were edited using ImageJ and Adobe Photoshop 7.0.

Data analysis
For length and pixel intensity measurements, a threshold was applied to define the ZA domain, and a line was drawn along the apical-basal axis of the cell, running in the middle of the ZA to measure the length of the Arm, Baz, Mbt domains. Mean pixel intensity was measured using the wand (tracing) tool in Fiji (Schindelin et al., 2012). In all cases, at least four independent mosaic retinas were used for each genotype. To compare the Mbt/Arm or
At 40% APF the pupal cuticle was removed to expose the retina and the animal was mounted in Voltalef oil. Live imaging was performed on a Leica SP5 confocal using a 63x 1.4 NA oil immersion objective at the following settings: pixel resolution 512 x 512, speed 400 Hz, 10% 488 nm laser power at 20% argon laser intensity and 5x zoom. FRAP analysis of ubi-ECad::GFP was performed by marking the basal tip of the AJ with a 5 pixel-diameter circle ROI followed by photo-bleaching with a single pulse using 90 % 488 nm laser power at 20 % argon laser intensity. AJ recovery was recorded every 1.293 seconds with the previously mentioned settings for approximately 1000 sec.
FRAP data were drift corrected in Fiji (Schindelin et al., 2012) using the StackReg plugin. Three different z axis profiles were analysed: (1) from the photo-bleached area; (2) from an equivalent area of a neighbouring non-photo-bleached AJ; and (3) from an equivalent area of background. The data were normalized using easyFRAP. ECad::GFP data were fitted to a twophase association curve in GraphPad Prism. The p values were calculated with an unpaired two-tailed Student's t test with Welch's correction.

Scanning Electron Microscopy
Flies were fixed in 2% paraformaldehyde, 2% glutaraldehyde and 0.1 M cacodylate for 2 hours and then dehydrated in ethanol, as previously described (Richardson and Pichaud, 2010). The samples were then criticalpoint dried and mounted on aluminum stubs before gold coating. Imaging was carried out on a JEOL Variable Pressure scanning electron microscope (SEM).