The cell adhesion molecule Echinoid functions as a tumor suppressor and upstream regulator of the Hippo signaling pathway

Summary

The Hippo (Hpo) signaling pathway controls tissue growth and organ size in species ranging from Drosophila to mammals and is deregulated in a wide range of human cancers. The core pathway consists of the Hpo/Warts (Wts) kinase cassette that phosphorylates and inactivates the transcriptional coactivator Yorkie (Yki). Here, we report that Echinoid (Ed), an immunoglobulin domain-containing cell adhesion molecule, acts as an upstream regulator of the Hpo pathway. Loss of Ed compromises Yki phosphorylation, resulting in elevated Yki activity that increases Hpo target gene expression and drives tissue overgrowth. Ed physically interacts with and stabilizes the Hpo-binding partner Salvador (Sav) at adherens junctions. Ed/Sav interaction is promoted by cell-cell contact and requires dimerization of Ed cytoplasmic domain. Overexpression of Sav or dimerized Ed cytoplasmic domain suppressed loss-of-Ed phenotypes. We propose that Ed may link cell-cell contact to Hpo signaling through binding and stabilizing Sav, thus modulating the Hpo kinase activity.

Introduction

How multi-cellular organisms control their growth to reach proper organ size during development is a fascinating question. Recent studies, initially from Drosophila, have identified an evolutionarily conserved pathway, the so-called Hippo (Hpo) tumor suppressor pathway, as a crucial mechanism that controls tissue growth and organ size by simultaneously inhibiting cell growth and proliferation and promoting cell death (Halder and Johnson, 2011; Pan, 2010; Zhang et al., 2009a). The Hpo signaling pathway has also been implicated in cell contact-dependent growth inhibition (Zhao et al., 2007), and deregulation of the Hpo pathway has been linked to a wide range of human cancers (Pan, 2010).

Central to the Hpo pathway is a kinase cassette consisting of four tumor suppressor proteins, the Ste20-like kinase Hpo (Harvey et al., 2003; Jia et al., 2003; Pantalacci et al., 2003; Udan et al., 2003a; Wu et al., 2003), the WW domain-containing protein Salvador (Sav) (Kango-Singh et al., 2002; Tapon et al., 2002), the NDR family kinase Warts (Wts) (Justice et al., 1995; Xu et al., 1995) and the Mob family protein Mats (Lai et al., 2005). The kinase activities of Hpo and Wts are facilitated by their regulatory proteins Sav and Mats, respectively. Activated Hpo/Sav complex phosphorylates and activates the Wts/Mats complex (Jia et al., 2003; Wei et al., 2007), which in turn phosphorylates and inactivates the transcriptional coactivator Yorkie (Yki) (Huang et al., 2005). Phosphorylation of Yki restricts its nuclear localization through recruiting 14-3-3 (Dong et al., 2007; Oh and Irvine, 2008; Ren et al., 2010; Zhang et al., 2008). When the activity of the Hpo/Wts kinase cassette is compromised, Yki forms complexes with transcription factors including Scalloped (Sd) and translocates to the nucleus to activate Hpo pathway target genes, including cyclin E, diap1, and the microRNA bantam that regulate cell growth, proliferation and survival (Goulev et al., 2008; Wu et al., 2008; Zhang et al., 2008).

The Hpo/Wts kinase cassette is regulated at several levels by multiple upstream modulators, including the atypical cadherins Dachsous (Ds) and Fat (Ft) (Bennett and Harvey, 2006; Cho et al., 2006; Matakatsu and Blair, 2006; Silva et al., 2006; Willecke et al., 2006), the FERM-domain proteins Expanded (Ex) and Merlin (Mer) (Cho et al., 2006; Hamaratoglu et al., 2006; Maitra et al., 2006; Pellock et al., 2007), the WW- and C2 domain-containing protein Kibra (Baumgartner et al., 2010; Genevet et al., 2010; Yu et al., 2010), and cell polarity determinants such as Crumbs (Crb) and lethal (2) giant larvae (Lgl) (Chen et al., 2010; Grzeschik et al., 2010; Ling et al., 2010; Robinson et al., 2010). Ds functions as a ligand for Ft and binding of Ds to Ft stabilizes Wts by inhibiting Dachs-mediated degradation of Wts (Cho et al., 2006). The apical membrane-localized Ex/Mer/Kibra complex interacts with Hpo/Sav and facilitates Hpo activation likely by retaining Hpo at the apical surface (Baumgartner et al., 2010; Genevet et al., 2010; Yu et al., 2010). Reduced protein levels or mislocalization of Ex at the apical surface have been observed in imaginal disc cells overexpressing the Crb intracellular domain or lacking Crb function, which may contribute to the deregulation of Hpo signaling in these cells (Chen et al., 2010; Grzeschik et al., 2010; Ling et al., 2010; Robinson et al., 2010).

The relative contribution of the upstream Hpo pathway regulators varies depending on tissue types. For example, loss of Ft in wing imaginal discs resulted in strong activation of Hpo responsive genes and marked tissue overgrowth but did not perturb Hpo signaling in Drosophila ovarian follicle cells (Meignin et al., 2007; Polesello and Tapon, 2007). To the contrary, Kibra is a critical regulator of Hpo signaling in Drosophila ovarian follicle cells but its loss of function mutation only weakly activated Hpo pathway genes and caused minor overgrowth of imaginal tissues (Genevet et al., 2010; Yu et al., 2010). Thus, the wiring of the Hpo signaling network may depend on the developmental context.

To identify additional Hpo pathway regulators, we carried out a genetic modifier screen and identified Echinoid (Ed), an immunoglobulin domain-containing cell adhesion molecule (Bai et al., 2001; Wei et al., 2005), as a upstream regulator of the Hpo pathway. We found that loss of Ed in imaginal discs attenuates Yki phosphorylation at S168 and promotes Yki nuclear localization, resulting in elevated Yki activity that drives tissue overgrowth. We also found that Ed is required for proper Hpo signaling in ovarian follicle cells. We demonstrated that Ed forms a complex with Sav to regulate its stability in S2 cells and that Ed is required for the enrichment of Sav at adherens junctions in vivo. In addition, we found that Ed/Sav interaction is promoted by cell-cell contact and requires dimerization/oligomerization of Ed cytoplasmic domain.

Results

Loss of Ed promotes tissue overgrowth

To identify new components of the Hpo signaling pathway, we carried out a genetic modifier screen in which flies carrying GMR-Gal4 and UAS-Yki (referred to as GMR>Yki) were crossed to a collection of transgenic RNAi lines from Vienna Drosophila RNAi center (VDRC) and Bloomington stock center, and looked for enhancers or suppressors of the overgrown eye phenotype caused by Yki overexpression. From this screen, we found that an RNAi line that targets Ed (UAS-Ed-RNAi³⁰⁸⁷; Fig. S1A) enhanced GMR>Yki mediated overgrowth although expression of UAS-Ed-RNAi³⁰⁸⁷ using GMR-Gal4 in otherwise wild type eyes did not cause any discernible overgrowth (Fig 1A–C′; data not shown). However, expression of UAS-Ed-RNAi³⁰⁸⁷ using the eyeless-Gal4 (ey-Gal4) driver, which drives the expression of UAS transgene in eye discs starting at a much earlier stage than GMR-Gal4 (Halder et al., 1998), resulted in eye overgrowth (Fig. 1D–D′).

Figure 1. Inactivation of Ed induces tissue overgrowth and enhances Yki gain-of-function phenotype.

(A–D′) Side (A–D) and dorsal (A′ –D′) views of a control adult fly eye (A–A′) or eyes expressing UAS-Yki (B–B′) or UAS-Yki + UAS-Ed-RNAi³⁰⁸⁷ (C–C′) with GMR-Gal4, or expressing UAS-Ed-RNAi³⁰⁸⁷ with ey-Gal4 (D–D′).

(E–F′) Side (E–F) and dorsal (E′–F′) views of mosaic eyes containing control clones (E–E′) or ed clones (F–F′) induced by ey-FLP.

(G–J) A control adult fly wing (G) and wings expressing UAS-Ed-RNAi⁹³⁷ (H), UAS-Sav-RNAi (I), or UAS-Ed-RNAi⁹³⁷ + UAS-Ed (J) with hh-Gal4 at 29 °C. Red dashed lines indicate the A/P compartment boundary.

To further probe the role of Ed in growth control, we expressed several UAS- Ed-RNAi lines from VDRC in the posterior compartment of wing imaginal discs using hh-Gal4 (referred to as hh>Ed-RNAi). As shown in Fig. 1H and Fig. S1, inactivation of Ed by hh>Ed-RNAi produced adult wings with enlarged posterior compartments. RNAi-mediated knockdown of ed expression was confirmed by immunostaining with an Ed antibody (Fig. 2A–A′). The overgrowth phenotype induced by hh>Ed-RNAi is unlikely due to an off-target effect because expression of three different transgenic RNAi lines (VDRC# 937, 3087, and 104279) targeting two non-overlapping regions of the ed coding sequence produced a similar overgrowth phenotype in adult wings (Fig. S1A, C–E). We also generated a UAS-Ed-RNAi line that targets a C-terminal region of Ed (UAS-Ed-RNAi^C; Fig. S1A) and found that expression of this line by hh-Gal4 also produced an enlarged posterior compartment (Fig. S1F). Furthermore, overexpression of a UAS-Ed transgene suppressed wing overgrowth caused by hh>Ed-RNAi (Fig. 1J).

Figure 2. Inactivation of Ed increases the expression of multiple Hpo pathway target genes.

(A–A′) A wing disc expressing UAS-GFP + UAS-Ed-RNAi¹⁰⁴²⁷⁹ with hh-Gal4 and immunostained with anti-GFP and anti-Ed antibodies. GFP marked the posterior compartment (arrows) where hh-Gal4 is expressed.

(B–E′) Wild type wing discs (B, C, D, and E) or wing discs expressing UAS-Ed-RNAi¹⁰⁴²⁷⁹ (B′, C′, D′, and E′) or UAS-Ed-RNAi¹⁰⁴²⁷⁹ + UAS-Ed (C″) with hh-Gal4 and immunostained to show the expression of diap1-GFP (B–B′), ex-lacZ (C–C″), Wg (green in D–D′), Ci (red in D–D′), and bantam sensor (BS; E–E′). Arrows in B–C″ and E–E′ indicate posterior compartments. Arrows and asterisks in D–D′ indicate Wg expression at the periphery of the wing pouch region and along the D/V compartment boundary, respectively. Ci expression (red in D–D′) marks the A-compartment.

(F–F′) A large magnification view of an eye disc containing ed mutant clones and immunostained to show the expression of LacZ (green) and DIAP1 (red). ed mutant clones are marked by the lack of LacZ expression and exhibit elevated DIAP1 expression (arrows).

(G–G′) A wing disc carrying ed mutant clones was immunostained to show the expression of LacZ (green) and Wg (red). ed mutant cells are marked by the lack of LacZ expression. ed mutant cells exhibit elevated Wg expression at the periphery of the wing pouch region (arrows) but normal Wg expression along the D/V boundary (asterisks).

We then determined whether ed mutation could render tissue overgrowth. We generated mosaic eyes that contain either wild type control clones or ed mutant clones using the ey-FLP system (Newsome et al., 2000). As shown in Fig. 1E–F′, mosaic eyes containing ed mutant clones were overgrown compared with the control mosaic eyes. The overgrowth phenotype caused by ed mutation is less severe compared with those caused by hpo mutations (Jia et al., 2003) but is comparable to those caused by kibra mutations (Baumgartner et al., 2010; Genevet et al., 2010; Yu et al., 2010).

Loss of Ed upregulates Hpo pathway target genes

The enlarged wing phenotype induced by hh>Ed-RNAi resembles those caused by loss of Hpo pathway components including Sav and Wts in posterior compartments (Fig. 1I) (Genevet et al., 2010). We therefore determined whether loss of Ed led to deregulation of Hpo pathway target genes. As shown in Fig. 2, expression of UAS-Ed-RNAi by hh-Gal4 led to the upregulation of diap1-GFP and ex-lacZ that faithfully report the expression of diap1and ex, respectively (Fig. 2B–C′; Hamaratoglu et al., 2006; Zhang et al., 2008). Consistent with this, DIAP1 was upregulated in ed mutant clones induced in eye discs (Fig. 2F–F′). In addition, overexpression of a UAS-Ed transgene suppressed elevated ex-lacZ expression caused by hh>Ed-RNAi (Fig. 2C″). Of note, we cannot examine ex-lacZ expression in ed mutant clones because ex-lacZ and ed are located on the same chromosomal arm.

Wg expression at the periphery of wing pouches is under the control of the Hpo signaling activity (Cho et al., 2006). We found that wing discs expressing hh>Ed-RNAi exhibited elevated Wg expression in these regions of the posterior compartment (indicated by the arrows in Fig. 2D′), consistent with the notion that loss of Ed deregulates the Hpo signaling activity. In contrast, Wg expression along the dorsoventral (D/V) compartment boundary, which is under the control of Notch (N) signaling activity, was not affected by Ed RNAi (indicated by the asterisk in Fig. 2D′). Similarly, ed mutant clones exhibited elevated Wg expression at the periphery of wing pouches (arrows in Fig. 2G–G′). In contrast, Wg expression along the D/V compartment boundary remained unchanged in ed mutant clones (asterisks in Fig. 2G–G′).

We also examined the expression of a bantam sensor, bantam-GFP, which inversely reports the expression level of the bantam microRNA (Thompson and Cohen, 2006). As shown in Fig. 2E′, wing discs expressing hh>Ed-RNAi exhibited reduced levels of bantam-GFP expression in posterior compartment cells, suggesting that loss of Ed upregulates bantam. Taken together, these observations demonstrate that loss of Ed increases the expression of multiple Hpo pathway target genes and further suggest that tissue overgrowth caused by Ed inactivation is due to deregulation of Hpo signaling activity.

Ed regulates Hpo signaling in ovarian follicle cells

The Hpo pathway is required for the proper maturation of posterior follicle cells and oocyte polarity formation during oogenesis (Meignin et al., 2007; Polesello and Tapon, 2007; Yu et al., 2008). In large posterior follicle cell clones mutant for Hpo pathway components including hpo, wts, sav, and ex, the N pathway is deregulated, leading to a failure of posterior follicle cells to undergo a switch from mitotic cycle to endocycle. As a consequence, mutant follicle cells undergo extra cell division and accumulate in multiple layers. We found that the N pathway target gene cut was ectopically expressed and multi-layers formed in posteriorly situated ed mutant follicle cell clones (Fig. 3B–B‴). In addition, posterior ed mutant follicle cells exhibited elevated expression of the Hpo pathway reporter diap1-GFP (Fig. 3D–D‴). Posterior follicle cell clones mutant for hpo pathway components cause oocyte polarity defects such as mislocalization of oocyte nuclei (Meignin et al., 2007; Polesello and Tapon, 2007; Yu et al., 2008). We found that posterior follicle cell clones mutant for ed also resulted in mislocalization of oocyte nuclei and Gurken (Grk) protein (Fig. 1F–F′ and H–H′). Thus, loss of Ed phenocopies loss of Hpo signaling in Drosophila ovaries.

Figure 3. Loss of Ed phenocopies loss of Hpo signaling in ovary.

(A–F′) Stage 8 wild-type (A–A‴, C–C‴) and mosaic egg chambers carrying ed mutant clones (B–B‴; D–D‴) were immunostained with antibodies against Cut (green in A and B), GFP (green in C and D) and LacZ (red), and the nuclear dye, DRAQ5 (blue). In all panels, the posterior is to the right. The dashed lines outline the mutant follicle cell clones that are marked by the lack of LacZ expression, and the solid lines indicate the posterior follicle cells. Cut expression is downregulated in follicle cells at stage 8 and its expression remains in the anteriorly and posteriorly localized polar cells (A). Cut expression is prolonged in posterior ed mutant follicle cell clones (arrow in B). There is a singer layer of follicle cells surrounding the wild type oocyte (A″ and C″) but multiple layers of follicle cells are found in posterior ed mutant follicle cell clones (B″ and D″). The expression of Hpo pathway target gene diap1-GFP is weak in posterior follicle cells (C) but is upregulated in posterior ed mutant follicle cells (arrow in D). The oocyte nucleus is localized at the anterior-dorsal corner (arrows in E–E′) but is mislocalized at the posterior pole of the oocyte with large ed follicle cell clones (arrows in F–F′). (G–H′) Stage 9 wild-type (G–G′) and mosaic egg chambers carrying ed mutant clones (H–H′) were immunostained with antibodies against Grk (red) and LacZ (green), and the nuclear dye, DRAQ5 (blue). Grk is localized at the anterior-dorsal corner of wild type oocytes (arrows in G–G′) but is mislocalized at the posterior pole of oocytes associated with large posterior ed follicle cell clones marked by the lack of LacZ staining (arrows in H–H′).

Ed regulates Yki phosphorylation and nuclear localization

The expression of Hpo pathway target genes is controlled by the transcriptional coactivator Yki, whose activity is primarily regulated by phosphorylation-mediated nuclear exclusion. Phosphorylation of Yki at S168 recruits 14-3-3 that restricts Yki nuclear localization, thereby limiting Yki activity in the nucleus (Dong et al., 2007; Oh and Irvine, 2008; Ren et al., 2010; Zhang et al., 2008). The increased expression of multiple Hpo pathway genes in hh>Ed-RNAi wing discs suggests that Yki activity is upregulated by loss of Ed. We therefore examined the Yki phosphorylation status using a phospho-specific antibody that recognizes phosphorylated S168 (pS168) (Dong et al., 2007). We found that expression of UAS-Ed-RNAi using the wing specific Gal4 driver MS1096 led to a marked reduction in the level of S168 phosphorylation (Fig. 4A). Consistent with diminished S168 phosphorylation, loss of Ed led to increased nuclear localization of Yki in hh>Ed-RNAi wing discs (Fig. 4B–B″). These results suggest that Ed inactivation may compromise the Hpo/Wts kinase activity, leading to reduced Yki phosphorylation and increased Yki nuclear localization and activity.

Figure 4. Ed acts upstream of Sav to regulate Yki phosphorylation and nuclear localization.

(A) Western blot analysis of Yki and phospho-Yki for wild type wing discs or wing discs expressing UAS-Ed-RNAi¹⁰⁴²⁷⁹ with the MS1096 Gal4 driver. The bottom panel is western blot for α-tubulin as a loading control.

(B–B″) A wing disc expressing ex-lacZ and UAS-Ed-RNAi¹⁰⁴²⁷⁹ with hh-Gal4 was immunostained with Yki (green) and LacZ (red) antibodies. The A/P boundary is demarcated by white lines based on Ci staining (not shown). Of note, Yki colocalized with LacZ in the nucleus of posterior compartment cells (arrows).

(C–E) Adult fly wings expressing UAS-Ed-RNAi⁹⁷³ (C), or UAS-Ed-RNAi⁹⁷³ + UAS-Yki-RNAi (D), or UAS-Ed-RNAi⁹⁷³ + UAS-Sav (E) with hh-Gal4 at 29 °C. Red dash lines demarcate the A/P compartment boundary.

(F–I) A wild type wing disc (F) or wing discs expressing UAS-Ed-RNAi¹⁰⁴²⁷⁹ (G), UAS-Ds-RNAi (H), UAS-Ed-RNAi¹⁰⁴²⁷⁹ + UAS-Ds-RNAi (I) with MS1096. The discs also expressed UAS- dicer2 to enhance the RNAi efficiency.

If upregulation of Yki activity is responsible for the phenotypes induced by Ed inactivation, one would expect that reduction of Yki activity should attenuate the phenotypes caused by loss of Ed. Indeed, Yki RNAi suppressed the overgrowth phenotype induced by loss of Ed (Fig. 4D). In addition, Yki RNAi suppressed the ed mutant phenotypes in Drosophila ovaries (Fig. S2B–B″).

Ed acts in parallel with Ds/Ft

We next investigated how loss of Ed compromised the activity of the Hpo/Wts kinase cassette. Ed is a multi-functional protein that has been implicated in restricting the activity of EGFR (Bai et al., 2001; Spencer and Cagan, 2003). However, EGFR RNAi did not block the upregulation of ex-lacZ induced by Ed RNAi, and overexpression of a constitutively active form of MAP kinase (MAPK^CA) by hh-Gal4 failed to activate ex-lacZ expression in wing discs (Fig. S3A–C′). Thus, Ed does not act through the EGFR pathway to regulate Hpo pathway responsive genes.

Ft regulates the Hpo/Wts kinase cassette through the non-conventional Myosin Dachs (Cho et al., 2006). Inactivation of Dachs by RNAi effectively blocked tissue overgrowth and elevated expression of ex-lacZ and Wg induced by Ft RNAi (Fig. S3D–E′). In contrast, Dachs RNAi did not suppress the elevated expression of ex-lacZ and Wg induced by Ed RNAi (Fig. S3F–F′), suggesting that Ed does not act through the Ft/Dachs branch of the Hpo pathway. Consistent with this notion, we found that inactivation of Ds synergized with loss of Ed to activate an Hpo pathway reporter, HREx8-GFP, which contains 8 copies of a 32 bp Hpo responsive element (Fig. 4F–I; Wu et al., 2008). These observations suggest that Ed may act in parallel with Ds/Ft to regulate Hpo signaling.

We next examined the epistatic relationship between Ed and the Hpo kinase cassette. We coexpressed UAS-Sav in hh>Ed-RNAi wing discs and found that Sav overexpression suppressed the tissue overgrowth induced by loss of Ed (Fig. 4E). We also generated MARCM clones in ovarian follicle cells that lack Ed and express UAS-Sav under the control of tub-Gal4 and found that overexpression of Sav suppressed the ectopic cut expression and multi-layer associated with posterior ed mutant follicle cells (Fig. S2C–C″). These observations suggest that Ed acts upstream of or in parallel with Sav to regulate Hpo signaling.

Ed physically interacts with multiple Hpo pathway components and regulates Hpo signaling through its intracellular domain

To understand the molecular mechanism by which Ed regulates Hpo signaling, we examined physical interactions between Ed and membrane proximal Hpo pathway components by coimmunoprecipitation experiments. S2 cells were transfected with a C-terminally Myc-tagged full-length Ed (Ed^FL-Myc) or truncated Ed that lacks the intracellular domain (Ed^ΔC-Myc; Fig. 5A) and HA-tagged Ex, Mer, or Sav, or with HA-tagged Ed^FL or Ed^ΔC (Ed^FL-HA or Ed^ΔC-HA) and Myc-tagged Kibra. Cell extracts were immunoprecipitated with the corresponding antibodies, followed by western blot analysis. As shown in Fig. 5, Ed^FL-Myc but not Ed^ΔC-Myc was coimmunoprecipitated with HA-Sav and HA-Ex (Fig. 5B, D), and to a lesser extent with HA-Mer (Fig. 5E). Myc-Kibra was coimmunoprecipitated with Ed^FL-HA but not Ed^ΔC-HA (Fig. 5F). These observations suggest that Ed interacts with Sav, Ex, Mer and Kibra through its intracellular domain. In addition, we found that Ed^FL-Myc was barely associated with Flag-Hpo but their association was facilitated by coexpression of HA-Sav (Fig. 5C), suggesting that Sav may bring Hpo to Ed. Interestingly, we found that Ed interacted with Yki but not with Sd through its intracellular domain (Fig. 5G and Fig. S4I).

Figure 5. Ed interacts with multiple Hpo pathway components.

(A) A schematic drawing of full-length Ed (Ed^FL) and its deletion constructs. Black boxes denote the transmembrane domain of Ed.

(B–H) S2 cells were transfected with DNA constructs expressing epitope-tagged full-length or indicated truncated Ed and tagged Hpo pathway components including Sav (B–C, H), Hpo (C), Ex (D), Mer (E), Kibra (F), or Yki (G). Cell lysates were immunoprecipitated, followed by western blot analysis with the indicated antibodies (top panels), or directly subjected to western blot analysis with the indicated antibodies (bottom panels). Asterisks in H and J indicate weak bands. Arrows in B indicate the cleaved Ed.

(I) A schematic drawing of full-length Sav and its deletion mutants. The WW and coiled-coil domains are indicated by shaded and black boxes, respectively. The ability of individual Sav constructs to interact with Ed is indicated by “+” or “−”.

(J) S2 cells were transfected with DNA constructs expressing Myc-tagged Ed^FL and indicated HA-tagged Sav deletion constructs. Cell lysates were immunoprecipitated with anti-HA antibody, followed by western blot analysis with anti-Myc antibody (top panel), or directly subjected to western blot analysis with anti-Myc or anti-HA antibody (middle and bottom panels). Asterisks indicate weak bands.

(K–N′) Wing discs expressing UAS-Ed-RNAi³⁰⁸⁷ (K and K′), or UAS-Ed-RNAi³⁰⁸⁷ + UAS-Ed^TMC-Myc (L and L′), or UAS-Ed-RNAi³⁰⁸⁷ + UAS -Ed^TMCΔ1-Myc (M and M′), or UAS-Ed-RNAi³⁰⁸⁷ + UAS- Ed^{TMCΔ 2}-Myc (N and N′) with hh-Gal4 and immunostained to show the expression of Ci (red), ex-lacZ (green) and Myc (blue). Ed transgenes with comparable Myc expression levels were used for the rescue experiments.

When expressed in S2 cells alone, HA-Sav and HA-Ex exhibited diffused patterns of cytoplasmic staining (Fig. S4A, C). Coexpression of Ed^FL-Myc resulted in the recruitment of HA-Sav and HA-Ex to the plasma membrane where they colocalized with Ed^FL-Myc (Fig. S4B, D). These observations further support the notion that Ed physically interacts with Sav and Ex and suggest that Ed can bring Sav and Ex to the cell membrane. We also confirmed that endogenous Ed and Sav form a complex in wing disc derived cl8 cells (Fig. S4E).

We next generated an Ed variant lacking the N-terminal extracellular domain, which we named Ed^TMC (TMC stands for transmembrane plus cytoplasmic domain; Fig. 5A). We found that C-terminally Myc-tagged Ed^TMC (Ed^TMC-Myc) interacted with HA-tagged Sav, Ex, Mer and Yki in coimmunoprecipitation assays (Fig. 5H, Fig. S4F–H). Deleting aa1217–1332 from Ed^TMC (Ed^TMCΔ1-Myc; Fig. 5A) greatly reduced the binding to Sav, Ex and Mer whereas deleting aa1115–1332 (Ed^TMCΔ2-Myc; Fig. 5A) abolished the binding to these proteins (Fig. 5H, Fig. S4F–G), suggesting that the C-terminal 218 amino acids of Ed mediate its interaction with Sav and Ex/Mer. In contrast, both Ed^TMCΔ1 and Ed^TMCΔ2 interacted with Yki similarly to Ed^TMC (Fig. S4H), suggesting that the membrane proximal region of Ed C-tail mediates Yki binding.

To map the domains in Sav that mediate interaction with Ed, we coexpressed full-length Ed with a number of truncated forms of Sav (Fig. 5I). We found that the N- terminal region and WW domain of Sav are both required for its optimal binding to Ed, as deleting either domain from Sav (SavC1 or SavΔC3) greatly reduced its binding to Ed and combined deletion (SavC2) completely abolished the interaction (Fig. 5J). In contrast, deleting the region C-terminal to the WW domain (SavΔC1 and SavΔC2) did not affect Sav/Ed interaction (Fig. 5J).

Deleting the intracellular domain of Ed (Ed^ΔC) blocked its ability to rescue the loss-of-Ed phenotypes both in wing discs and posterior follicle cells (Fig. S2E–E″ and Fig. S7C–C′). On the other hand, overexpression of Ed^TMC suppressed the loss-of-Ed phenotypes (Fig. 5L–L′; Fig. S2F–F″). These observations suggest that Ed regulates Hpo signaling through its intracellular domain, likely by interacting with multiple intracellular Hpo pathway components. Consistent with this notion, deleting the domain involved in binding to Sav/Hpo and Ex/Mer from Ed^TMC (Ed^TMCΔ1 and Ed^TMCΔ2) abolished its ability to suppress the elevated expression of ex-lacZ induced by loss of Ed in wing discs (Fig. 5M–N′).

Ed regulates Sav stability and subcellular localization

In the course of characterizing Ed/Sav interaction, we found that coexpression of Ed^FL-Myc stabilized transfected HA-Sav (Fig. 6A). Stabilization of Sav is mediated by the intracellular domain of Ed as Ed^ΔC failed to stabilize Sav whereas Ed^TMC stabilized Sav albeit less effectively than Ed^FL (Fig. 6A). We also measured the half-life of HA-Sav in the absence or presence of Ed^FL-Myc. HA-Sav was transfected into S2 cells with or without Ed^FL-Myc. The levels of HA-Sav were monitored at different time points after treatment with the protein synthesis inhibitor, cycloheximide (CHX). As shown in Fig. 6, the half-life of HA-Sav was less than two hours in the absence of Ed^FL-Myc; however, coexpression of Ed^FL-Myc increased the half-life of HA-Sav to 8 hours (Fig. 6B–C). We also found that treating transfected cells with the proteasome inhibitor MG132 increased the steady state levels of HA-Sav (Fig. 6D), suggesting that Sav was degraded by the ubiquitin/proteasome pathway.

Figure 6. Ed regulates the stability and apical localization of Sav.

(A) Western blot analysis of cell lysates from S2 cells transfected with the indicated constructs. Myc-CFP (arrow) was cotransfected as an internal control. Samples were boiled and loaded immediately after 2X SDS loading buffer was added into cell lysates.

(B) S2 cells were transfected with HA-Sav in the absence or presence of Ed^FL-Myc and treated with cycloheximide for the indicated time. Cell lysates were subjected to western blot analysis using anti-Myc antibody. Myc-CFP (arrow) was cotransfected as an internal control.

(C) Quantification of HA-Sav in the absence or presence of Ed^FL-Myc by western blot analysis performed in B.

(D) Western blot analysis of cell lysates from S2 cells transfected with HA-Sav and Myc-CFP and treated with MG132 for 4 hours before harvesting. Myc-CFP (lower panel) served as a transfection and loading control.

(E–F′) Wing discs expressing UAS-Sav-RNAi with hh-Gal4 were immunostained with an anti-Sav antibody (green) and the nuclear dye, DRAQ5 (red in E′), or an anti-Ed antibody (red in F′). Sav exhibited cytoplasmic staining, which was abolished by Sav-RNAi.

(G–G″) Transverse section view of wild type wing discs immunostained with Ed (red) and Sav antibodies (green). Sav was enriched at the apical plasma membrane where it colocalized with Ed at adherens junctions.

(H–I″) Transverse section view of a wing disc expressing UAS-Ed-RNAi¹⁰⁴²⁷⁹ with hh-Gal4 and immunostained to show the expression of Sav (green), Ed (red in H, H″) or E-cadherin (E-cad; red in I′, I″″) and En (blue in I″, I″″) that marks the P-compartment. Loss of Ed resulted in reduced Sav accumulation at the apical surface.

(J–K″) Wild type (J–J″) and mosaic egg chambers carrying ed mutant clones (K–K″) were immunostained to show the expression of Sav (green), LacZ (red), and Phalloidin (blue). ed mutant follicle cells are marked by the lack of LacZ expression (arrows in K–K″). Posterior ed mutant follicle cells showed reduced Sav signals at the apical surface marked by Phalloidin (arrows in K–K″).

To determine whether Ed regulates Sav stability and subcellular localization in vivo, we generated an anti-Sav antibody (see Experimental Procedures). The specificity of this antibody was confirmed by Sav RNAi (Fig. 6E–F′). We found that Sav was distributed in cytoplasm with an enrichment at the apical surface where Sav colocalized with Ed at adherens junctions (Fig. 6E–G″). Ed RNAi reduced the levels of Sav immunostaining at the apical region (Fig. 6H–I‴). Consistent with Ed regulating Sav at the posttranscriptional level, we found that Ed RNAi also reduced the levels of HA-Sav driven by a UAS transgene at adherens junctions (Fig. S5A–B″). We also found that Sav was enriched at the apical surface of ovarian follicle cells (Fig. 6J–J″″) and that the levels of apically localized Sav were reduced in posterior ed mutant follicle cells (Fig. 6K–K″″).

Because we observed that Sav physically interacts with Ex/Mer, we also examined whether loss of Ed affects the level or subcellular localization of Ex/Mer. We found that levels of Ex and Mer were upregulated by Ed RNAi (Fig. S5C–D″), likely due to the transcriptional upregulation of ex and mer expression in response to loss of Ed. However, loss of Ed did not affect the apical localization of Ex/Mer (Fig. S5C–D″).

Regulation of Ed/Sav interaction by Ed dimerization and membrane localization

Although Ed^TMC interacts with Sav, the soluble Ed intracellular domain (Ed^C; Fig. 7A) that lacks the transmembrane domain failed to bind Sav in our coimmunoprecipitation experiments (Fig. 7B). One possibility is that Ed/Sav interaction may require association of Ed to the membrane. To test this possibility, a myristoylation signal was added to Ed^C (Myr-Ed^C; Fig. 7A). We found that Myr-Ed^C-Myc only weakly interacted HA-Sav, suggesting that tethering Ed C-tail to the membrane is not sufficient to confer efficient binding to Sav.

Figure 7. Regulation of Ed/Sav interaction by dimerization and cell-cell contact.

(A) Schematic drawings of full-length Ed and Ed variants. Black bars indicate the Ed transmembrane domain. Triangle and grey box represent a myristoylation (Myr) signal and a coiled-coil (CC) dimerization domain, respectively.

(B–C) S2 cells were transfected with the indicated constructs. Cell lysates were immunoprecipitated with anti-HA or anti-Myc antibody, followed by western blot analysis with the indicated antibodies (top two panels) or directly subjected to western blot analysis with anti-Myc or anti-HA antibody (bottom two panels). Asterisks in B indicate weak bands. Arrows indicate IgG.

(D) FRET efficiency of indicated pairs of CFP/YFP constructs transfected into S2 cells (mean ± s.d., n≥10).

(E–G′) S2 cells were transfected with tub-Sav-GFP and tub-Ed^FL-Myc or tub-Ed^TMCΔ1-Myc. Immunostaining was carried out after cell aggregation was induced.

(H) A model for Ed in the Hpo signaling pathway. Dashed lines indicate less characterized interactions. See text for details.

A previous study reported that Ed forms a dimer or oligomer (Spencer and Cagan, 2003). Interestingly, we found that Ed^TMC but not Ed^C can self-associate when expressed in S2 cells (Fig. 7C), suggesting that the TM domain of Ed can mediate dimerization/oligomerization. This observation further suggests that Ed/Sav association may require dimerization/oligomerization of Ed C-tail. To test this possibility, we added a coiled-coil dimerization motif (CC; see Experimental Procedures) to generate Myr-CC-Ed^C (Fig. 7A) and found that Myr-CC-Ed^C could self-associate and interact effectively with Sav (Fig. 7B–C). Adding the coiled-coil dimerization motif to the soluble Ed^C (CC-Ed^C) did not result in strong association between CC-Ed^C and Sav (data not shown), suggesting that effective interaction between Ed and Sav requires both membrane localization and dimerization of the Ed intracellular domain. Consistent with this notion, we failed to detect an interaction between the Ed intracellular domain and Sav in a yeast two-hybrid assay (data not shown). Of note, membrane association of Ed^TMC, Myr-Ed^C, and Myr-CC-Ed^C was confirmed by immunostaining of S2 cells transfected with these Ed variants (Fig. S7).

To further characterize the Ed/Sav interaction, we carried out fluorescence resonance energy transfer (FRET) experiments. C-terminally CFP-tagged Ed (Ed-CFP) was transfected into S2 cells together with N- or C-terminally YFP-tagged Sav (YFP-Sav or Sav-YFP). We observed high FRET between Ed-CFP and YFP-Sav but low FRET between Ed-CFP and Sav-YFP (Fig. 7D; Fig. S6), suggesting that Ed is in close proximity with the N-terminal region of Sav. This is consistent with the results from immunoprecipitation experiments indicating that Ed binds to the N-terminal region of Sav (Fig. 5I–J). We also observed high FRET between N-terminally CFP-tagged Sav (CFP-Sav) and YFP-Sav but low FRET between CFP-Sav and Sav-YFP (Fig. 7D; Fig. S6), suggesting that Sav forms a parallel dimer/oligomer. Thus, association between Ed and Sav may involve multimeric (dimer to dimer) interactions.

To determine the effect of C-tail dimerization on Ed activity in vivo, we generated transgenic flies expressing UAS-Ed^C, UAS-Myr-Ed^C and UAS-Myr-CC-Ed^C, and compared their activities by carrying out rescue experiments. We found that overexpression of Myr-CC-Ed^C but not Ed^C or Myr-Ed^C could suppress the elevated ex-lacZ expression induced by Ed RNAi (Fig. S7).

Regulation of Ed/Sav interaction by cell-cell contact

A previous study showed that Ed transfected into S2 cells could mediate cell-cell adhesion though homophilic interactions and that Ed was enriched at cell contact sites (Islam et al., 2003). We therefore determined whether cell-cell contact could modulate Ed/Sav interaction. To do this, we transiently expressed low levels of Ed^FL-Myc and Sav-GFP using the tubulin promoter (tub-Ed^FL-Myc and tub-Sav-GFP). In isolated cells, Ed^FL-Myc and Sav-GFP exhibited limited colocalization (Fig. 7E–E″). Consistent with the previous finding (Bai et al., 2001), Ed^FL-Myc accumulated at the cell contact site upon cell-cell contact (Fig. 7F). Strikingly, Sav-GFP also accumulated at the cell contact site where it colocalized with Ed^FL-Myc (Fig. 7F′–F″). Deleting the C-terminal Sav-interacting domain in Ed (Ed^ΔC1-Myc) prevented the recruitment of Sav-GFP to the cell contact region (Fig. 7G–G″). Thus, cell-cell contact facilitates Ed/Sav association.

Discussion

The Hpo signaling pathway has emerged as a conserved regulatory pathway that controls tissue growth and organ size. Although the core pathway components, i.e., the Hpo/Sav/Wts/Mats kinase cassette and its effector Yki/Yap have been well defined, the upstream regulators, especially the membrane receptors that link cell-cell communication to Hpo signaling, remain poorly defined. Here we provide both genetic and biochemical evidence that the transmembrane cell adhesion molecule Ed functions as a upstream regulator of the Hpo pathway. We provide evidence that Ed physically interacts with Sav/Hpo and regulates the abundance of Sav at adherens junctions. Loss of Ed compromises the ability of Hpo/Wts kinase cassette to phosphorylate Yki, leading to elevated nuclear levels of Yki activity that drives tissue overgrowth. We found that Ed/Sav association is facilitated by cell-cell contact, raising an interesting possibility that Ed may serve as a mechanism that links Hpo signaling to cell contact inhibition.

The atypical cadherin Ft functions as a receptor for the Hpo pathway; however, Ft mainly acts through Dachs to control the stability of Wts (Cho et al., 2006). Our genetic study indicated that Ed does not act through Ft-Dachs to regulate Yki activity because inactivation of Dachs did not block Yki activation induced by loss of Ed. Furthermore, loss of Ed synergized with loss of Ds to induce the expression of Hpo responsive genes, supporting a model in which Ed acts in parallel with Ds/Ft in the Hpo pathway (Fig. 7H). Several lines of evidence suggest that Ed regulates Hpo signaling, at least in part, through Sav. 1) Using coimmunoprecipitation, colocalization and FRET assays, we demonstrate that Ed physically interacts with Sav. 2) Deleting the Sav-interacting domain in Ed compromises its in vivo activity. 3) Ed regulates the abundance and subcellular localization of Sav both in vitro and in vivo. 4) Overexpression of Sav suppresses tissue overgrowth induced by loss of Ed. Sav is a binding partner and activator of Hpo. Hence, Ed could influence the Hpo kinase activity by regulating the abundance and subcellular location of the Sav/Hpo complex. How Ed regulates Sav stability is currently unknown; however, we found that Sav is degraded in a proteasome-dependent manner. It is possible that binding of Ed to Sav leads to some modifications of Sav and prevents it from ubiquitin/proteasome-mediated degradation.

Sav binds Ed and Hpo through its N- and C-terminal regions, respectively (Fig. 5; Jia et al., 2003; Udan et al., 2003b; Wu et al., 2003), and thus may function as a bridge to bring Hpo to Ed. Indeed, we observed enhanced Ed/Hpo association in the presence of cotransfected Sav. It has been suggested that apical membrane recruitment of Hpo promotes phosphorylation of Wts (Hergovich et al., 2005; Yu et al., 2010). Thus, it is conceivable that Ed may regulate the Hpo kinase by recruiting Sav/Hpo complex to the apical membrane. We found that Ed/Sav interaction requires membrane association and dimerization/oligomerization of Ed intracellular domain. As Sav also forms a dimer/oligomer (Fig. 7), dimerization/oligomerization of Ed intracellular domain may enhance binding to Sav through multimeric interactions. It is also possible that membrane association and dimerization/oligomerization could lead to a modification of Ed intracellular domain, resulting in increased binding affinity toward Sav.

It has been shown that the Hpo pathway can mediate cell contact inhibition in mammalian cultured cells although the underlying mechanism has remained poorly defined. Interestingly, we found that cell-cell contact can facilitate the recruitment of Sav to Ed at the contact site. Cell-cell contact may facilitate homophilic interaction in trans and clustering of Ed intracellular domain, or induce post-translational modification of Ed C-tail at the contact site, leading to enhanced Sav association. We propose that regulation of Ed/Sav association may provide a mechanism for cell-cell contact to modulate Hpo signaling and tissue growth.

The mechanism by which Ed regulates Hpo signaling is likely to be more complex than simply regulating Sav/Hpo. For example, we also observed that Ed interacts with Ex/Mer/Kibra as well as Yki. It has been proposed that enrichment of Hpo pathway components to the apical membrane domain may facilitate the activation of the kinase cassette and increase the accessibility of Yki to its kinase (Genevet and Tapon, 2011). Our finding that Ed facilitates the apical localization of Sav lends further support to this notion. Through interacting with multiple components of the Hpo pathway, Ed could function as a molecular scaffold to facilitate Hpo activation and Yki phosphorylation. We found that loss of Ed did not alter the apical membrane localization of Ex and Mer in wing discs even though overexpression of Ed in S2 cells facilitates membrane recruitment of Ex. The apical localization of Ex and Mer is likely to be mediated by other upstream components such as Ft and Crb in the absence of Ed. Indeed, Crb physically interacts with Ex and both loss and gain of function of Crb caused mislocalization of a fraction of Ex to the basal region (Chen et al., 2010; Grzeschik et al., 2010; Robinson et al., 2010). It has been shown that Ex physically interacts with Yki, which may sequester Yki in the cytoplasm independent of Yki phosphorylation (Badouel et al., 2009). Our finding that Ed interacts with Yki through a domain distinct from those mediating its binding to the upstream Hpo pathway components raises a possibility that Ed may also directly sequester Yki in the cytoplasm in addition to regulating its subcellular localization through phosphorylation.

It is interesting to note that Ed is related to TSLC1, a tumor suppressor implicated in human non-small-cell lung cancer and other cancers including liver, pancreatic, and prostate cancers (Kuramochi et al., 2001; Murakami, 2005). Like Ed, TSLC1 also mediates cell-cell adhesion through homophilic interactions (Masuda et al., 2002). TSLC1 interacts with MPP3, a human homolog of Drosophila tumor suppressor Discs large (Dlg) that has been implicated in the Hpo pathway (Fukuhara et al., 2003; Grzeschik et al., 2010), as well as DAL-1, a FERM-domain containing tumor suppressor related to Ex/Mer (Yageta et al., 2002). Therefore, it would be interesting to determine whether TSLC1 inhibits tumor formation through the Hpo pathway.

Experimental Procedures

Mutants and transgenes

ed^1X5 is a strong allele (Bai et al., 2001). To generate ed mutant or control clones in adult eyes, y w ey-FLP; l(2)cl-L3 P[w+] FRT40/Cyo (BL#5622) flies were crossed with ed^1X5 FRT40/Cyo or FRT40 flies. Marked ed clones in imaginal discs were generated using the Minute technique with the following genotype: hs-FLP; FRT40 M (2)24F arm-lacZ/FRT40 ed^1X5. FLP/FRT-mediated mitotic recombination was induced 1–2 days after egg laying by incubation the larvae at 37 for 1 hour. The genotypes for generating clones in ovaries are as follows: ed clones: y w hs-flp; FRT40 ed^1X5 FRT40/arm-LacZ; ed clones expressing various transgenes: y w hs-flp; FRT40 ed^1X5/FRT40 tub-Gal80; UAS-transgenes/tub-Gal4 UAS-GFP. Transgenic RNAi lines used are: UAS-Ed-RNAi⁹³⁷ (VDRC# 937), UAS-Ed-RNAi³⁰⁸⁷ (VDRC# 3087), UAS-Ed-RNAi¹⁰⁴²⁷⁹ (VDRC# 104279), UAS-Ft-RNAi (VDRC# 9396), UAS-Ds-RNAi (VDRC # 36219), UAS-Dachs-RNAi (VDRC# 12556), UAS-EGFR-RNAi (VDRC# 43267), UAS-Yki-RNAi (Zhang et al., 2008), UAS-Sav-RNAi (BL# 28006), and UAS-Ed-RNAi^C targeting the C-terminal region of Ed between aa1216–1332. Other transgenes used in this study include bantam sensor (Brennecke et al., 2003), ex-lacZ (Hamaratoglu et al., 2006), diap1-GFP (Zhang et al., 2008), UAS-Ed (Fetting et al., 2009), and UAS-HA-Sav (Jia et al., 2003). To generate UAS-Myr-Ed^C and UAS-Myr-CC-Ed^C constructs, DNA fragments corresponding to a myristoylation signal, GNKCCSKRQ, and the leucine zipper coiled-coil dimerization domain of the yeast GCN4 (O’Shea et al., 1991; Zhang et al., 2009b), are synthesized and inserted into the N-terminus of Ed^C. Transgenic flies carrying UAS-Ed^FL-Myc, UAS-Ed^TMC-Myc, UAS -Ed^TMCΔ1-Myc, UAS Ed^{TMCΔ 2}-Myc, UAS-Ed^C, UAS-Myr-Ed^C and UAS-Myr-CC-Ed^C were generated by P-element transformation.

Cell culture, transfection, immunoprecipitation, and western blot analysis

Drosophila S2 cells were cultured in Drosophila Schneider’s medium (Invitrogen) with 10% fetal bovine serum, 100 U/ml of penicillin, and 100mg/ml of streptomycin. Transfection was carried out by Calcium Phosphate Transfection Kit (Specialty Media) according to manufacturer’s instructions. The construct of ubiquitin-Gal4 was cotransfected with pUAST expression vectors for all the transfection experiments except for cell-cell contact induced Ed/Sav interaction experiment in which tub-Ed^FL-Myc, tubEd^ΔC1-Myc and tub-Sav-GFP were used. Immunoprecipitation and western blot analyses were carried out using standard protocols as previously described. MG132 (Calbiochem) was used at 50 μM. Transfected S2 cells were treated with Cycloheximide (Sigma) at a final concentration of 100 μM for the indicated time periods before harvesting. For endogenous protein detection, wing imaginal discs were collected from the later third instar larvae and lysed in NP40 cell lysis buffer with protease inhibitor cocktail (Roche). Lysates were cleared by centrifugation and subjected to SDS-PAGE. Western blot was performed by using rabbit anti-Yki (1:1000, K. Irvine), rabbit anti-phospho-Yki (1:1000, D. Pan), mouse anti-α-Tubulin (1:100,000, Sigma), rat anti-Ed (1:1000, J. Hsu), and rabbit anti-Sav (1:1000).

Immunohistochemistry

To generate Sav antibodies, full-length Sav cDNA was cloned into pET30. His-Sav was expressed in BL21 E.coli (Invitrogen) by induction with 0.1 mM IPTG. Sav protein was purified on SDS-PAGE and used to immunize rabbits (Cocalico Biologicals). Other antibodies used were mouse anti-β-Gal (1:1000, Promega), rabbit anti-β-Gal (1:1000, Affinity Bioreagents), mouse anti-Wg (1:100, DSHB), rat anti-Ci, 2A1 (Motzny and Holmgren, 1995), rabbit anti-Yki (1:1000, K. Irvine), Rabbit anti-GFP (Molecular Probes), guinea pig anti-Ex (1:5000) and guinea pig anti-Mer (1:5000, R. Fehon), rat anti-Ed (1:1000, J. Hsu), rat anti-E-cad (1:100, DHSB), rabbit anti-Sav (1:100), DRAQ5 (cell signaling), mouse anti-Grk (DSHB), mouse anti-HA, mouse anti-Myc and mouse anti-Fg (Santa Cruz). Cy2-, Cy3- and Cy5- conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories. Images were captured by Zeiss LSM510 confocal microscopy.

FRET analysis

FRET analysis was carried out as previously described (Zhao et al., 2010). CFP/YFP-tagged constructs were transfected into S2 cells together with an ubiquitin-Gal4 expression vector. Cells were washed with PBS, fixed with 4% formaldehyde for 20 minutes and mounted on slides in 80% glycerol. CFP signals were acquired with 100X objective of Zeiss LSM510 confocal microscope before (BP) and after (AP) photobleaching YFP. Each data set was calculated using 10–20 individual cells. In each cell, four or five regions of interest in photobleached area were selected for analysis. The intensities of CFP signals were quantified by ImageJ software. The FRET efficiency was calculated using the formula: FRET%=[(CFP_AP−CFP_BP)/CFP_AP] ×100. Of note, only CFP signals that colocalized with YFP signals (both membrane and intracellular) were selected for calculation.