Competition between two phosphatases fine-tunes Hedgehog signaling

Hedgehog (Hh) signaling is an evolutionarily highly conserved signaling cascade that coordinates cell fate decisions, tissue patterning, organ growth, and adult homeostasis (Lee et al., 2016). Dysregulated Hh signaling in humans leads to birth defects and cancer (Pak and Segal, 2016; Kong et al., 2019). Thus, the level as well as the activity of core signaling components must be precisely regulated in order to maintain homeostasis of Hh signaling (Pak and Segal, 2016; Liu, 2019). Among identified homeostatic regulatory machineries, posttranslational modifications (PTM) play critical roles in modulating the configuration, subcellular localization, physical interaction, and molecular functions of the core players (Barber and Rinehart, 2018). Thus, various forms of PTM significantly expand cellular properties of key signaling players of the Hh regulatory network in development and adult homeostasis.

Phosphorylation, a common PTM event occurring in >30% of human proteins (Cohen, 2002; Vlastaridis et al., 2017), is tightly regulated by opposing activities of specific protein kinases and phosphatases. They often act together in development to provide dynamic yet robust regulation on the abundance and activity of core players of developmental signaling, forming feedback or feed-forward regulatory networks to maintain signaling homeostasis (Couzens et al., 2013; Xu et al., 2016; Thompson and Williams, 2018). Extensive studies uncovered a stereotypical sequential phosphorylation profile of Hh signaling activator Smoothened (Smo), comprised of basal-, moderate- and highly phosphorylated Smo species, and showed that it is essential for Hh signal transduction in Drosophila melanogaster (Jia et al., 2004; Zhang et al., 2004; Apionishev et al., 2005; Su et al., 2011; Li et al., 2016). Sequential Smo phosphorylation is catalyzed first by cyclic adenosine 3′, 5′-monophosphate–dependent protein kinase A, followed by casein kinase I. In addition, phosphorylation by G protein–coupled receptor kinase 2 is required for maximal Smo activity and subsequent Smo internalization and degradation (Chen et al., 2010; Maier et al., 2014). Two protein serine/threonine phosphatases, namely protein phosphatase 1 and 2A (PP2A), act as Smo-specific phosphatases that respectively antagonize cyclic adenosine 3′, 5′-monophosphate–dependent protein kinase A and casein kinase I activities to maintain sequential phosphorylation of Smo (Su et al., 2011). However, how such Smo-associated kinases and phosphatases sense Hh signaling gradient and how they in turn collectively coordinate graded Smo phosphorylation are not well understood.

Here, we provide genetic and biochemical evidence demonstrating that protein phosphatase V (PpV), the Drosophila orthologue of the catalytic subunit of protein phosphatase 6 (PP6C), is a bona fide Hh signaling target whose activity is essential for regulating PP2A stability in response to differential Hh signaling. Importantly, our study uncovers a noncanonical regulatory function of PpV that does not require its phosphatase activity. Instead, it competes with the catalytic subunit of PP2A for association with Widerborst (Wdb), one of four variable regulatory subunits of PP2A heterotrimeric holoenzyme. The resulting noncanonical Wdb/PpV complex is not functional, leading to Wdb degradation in the proteasome. Our results suggest that PpV senses graded Hh signaling activity to maintain a stereotypical Smo phosphorylation profile essential for Hh signaling. As the expression of PpV directly responds to differential Hh signaling gradient, our study highlights the importance of feedback regulation of PpV on Wdb-mediated PP2A phosphatase activity in homeostatic regulation of Hh signaling.

Hh signaling is one of the major signaling systems that pattern the Drosophila wing, playing a key role in determining the distance between L3 and L4 (L3-L4) longitudinal veins (Fig. 1 A). In an in vivo RNAi screen, we found that knocking down the PpV phosphatase resulted in reduced L3-L4 distance in the adult wing blade (Fig. 1, C and F), a phenotype which has been previously observed in several conditions of reduced Hh signaling, for example, by overexpressing patched (ptc), a negative regulator of Hh signaling (Fig. 1 B and Fig. S1 C). Reduced PpV expression also resulted in loss of anterior cross-vein, a phenotype consistent with a reported role of PpV in JNK-dependent tumor progression (Ma et al., 2017). However, loss of anterior cross-vein, but not reduced L3-L4 distance, was partially suppressed in PpV RNAi adult wings by increased activity of puckered (puc; Fig. 1, D and E), a negative regulator of JNK signaling, suggesting that PpV plays an additional role in Hh signaling. This notion was further supported by observed genetic interactions between PpV and core components of Hh signaling: the wing defects induced by reduced expression of cubitus interruptus (ci; Fig. S1 A), the Hh signaling transcription factor, or by overexpressing ptc (Fig. S1 C), were further enhanced by PpV RNAi (Fig. S1, B and D). Consistently, the fusion of proximal L3 and L4 veins in a temperature-sensitive ci mutant (ci^cell) background (Fig. S1, E–G′; Slusarski et al., 1995; Méthot and Basler, 1999) became more prominent when one copy of PpV was removed in a heterozygous PpV^KO background (Fig. S1, H and H′; cf. Fig. S1, F and F′). Together, the above results establish a requirement of PpV activity in mediating Hh signaling in Drosophila wing development.

To confirm that Hh signaling was indeed altered when PpV was dysfunctional, we examined third instar larval wing imaginal discs for stability of full-length Ci (CiFL), as well as expression of Hh signaling responsive genes decapentaplegic (dpp) and collier/knot (col), which differentially respond to graded Hh morphogen concentration. Low-level Hh signaling was sufficient to stabilize CiFL (Fig. 1 G), whereas dpp (Fig. 1 H) and col (Fig. 1 I) were induced by intermediate and high levels of Hh activity, respectively, consistent with a previous report (Torroja et al., 2005). Considering that the L3-L4 intervein region is largely maintained by col activation (Crozatier et al., 2002, 2003), it is not surprising that the production of Col protein was diminished when PpV activity was reduced (Fig. 1 L), demonstrating that PpV is required for high-level Hh activity. However, contrary to expectations, we found that CiFL protein was stabilized (Fig. 1 J) and the expression of dpp-lacZ was expanded to more anterior cells when PpV RNAi was expressed (Fig. 1 K). These surprising results suggest that PpV is also able to repress the expression of target genes responsive to low to intermediate levels of Hh signaling. To verify these apparently contradictory results, we generated a PpV null allele (PpV^KO) in which the coding sequence of PpV was removed by homologous recombination. Few hemizygous PpV^KO males (<5%) survived to early pupal stage, thereby allowing us to examine the effect of PpV null on Hh signaling targets in larval wing discs. The result observed in PpV^KO hemizygous wing discs was similar to that obtained with PpV RNAi (Fig. 1, M–R), highlighting a surprising role of PpV in differential regulation of Hh responsive genes.

The defect induced by reduced PpV activity in adult wing highly resembled the phenotype caused by wdb overexpression. Wdb was identified in our previous study as a specific regulatory subunit that controls the activity of PP2A in Smo phosphorylation and Hh signaling activity (Su et al., 2011). Following either wdb overexpression or reduced PpV activity, the expression of low to intermediate levels of Hh targets was expanded, while the activation of high-level Hh targets was repressed. The similarity in phenotypes following the two manipulations led us to believe that these genes may function in the same pathway. To explore this possibility, we examined the epistatic relationship between PpV and wdb. We found that loss of both wdb and PpV expression in adult wing phenocopied the defect associated with wdb RNAi alone (Fig. 2, A–D), suggesting that wdb is epistatic to PpV. Consistently, loss of Col expression induced by PpV RNAi (Fig. 2 F′) in wing disc was rescued upon coexpression of a dominant negative form of wdb (wdbDN; Fig. 2 H′).

Previous studies provided extensive evidence for sequential phosphorylation and subsequent cell surface localization of Smo in mediating Hh signaling in Drosophila (Zhu et al., 2003; Jia et al., 2004; Zhang et al., 2004; Apionishev et al., 2005). Consistent with the report that Wdb-mediated PP2A phosphatase activity (Wdb-PP2A) directly dephosphorylates highly phosphorylated Smo to prevent over-activation of Hh signaling, knocking down wdb (Fig. 2, K–K′′ and M) behaved similarly to adding Hh (Fig. 2, I–J′′ and M) to Drosophila Schneider 2 (S2) cells to induce hyper-phosphorylation and translocation of Smo from cytosol to the plasma membrane. Similar effects on Smo were observed when PpV was overexpressed in S2 cells (Fig. 2, L–L′′ and M). Together, the above results demonstrate that PpV acts upstream of wdb to modulate Smo phosphorylation in Hh signaling.

To determine how PpV regulates wdb activity, we first examined wdb expression in wing disc following alteration of PpV activity. We found that the amount of endogenous Wdb was significantly increased in cell lysates extracted from hemizygous PpV^KO larvae (Fig. 3 A). In WT wing disc epithelia, Wdb protein was predominantly localized to apical cytoplasm (Fig. 3, B and B′). However, increased amount of Wdb protein observed in PpV knockdown cells extended to basolateral cytoplasm (Fig. 3, C and C′). To further investigate how PpV regulates Wdb in vivo, we constructed a heat-shock (hs)-wdb-HA transgene where production of Wdb protein was controlled by a hs promoter. Reducing PpV expression in S2 cells by RNAi was still able to increase the amount of Wdb protein produced by the hs promoter (Fig. 3 D), while overexpressing PpV had an opposite effect (Fig. S2 A). Significantly, hs-Wdb protein accumulated in wing disc either following PpV knockdown (Fig. 3 F′) or in loss-of-function PpV^KO mosaic clones (Fig. S2 B′), suggesting that the regulation of Wdb by PpV most likely takes place post-transcriptionally.

A common post-translational regulation mechanism is through protein turnover, via either proteasome- or lysosome-mediated degradation. Wdb protein produced in S2 cells had a relatively short half-life of around 3 h (Fig. S2 C). We found that the degradation of Wdb protein occurred through both proteasome- (Fig. S2 D) and lysosome-dependent pathways (Fig. S2 E), when nascent protein synthesis was blocked by cycloheximide. However, Wdb degradation induced by PpV overexpression was only inhibited when S2 cells were treated with proteasome inhibitor MG132, but not lysosome inhibitor chloroquine (CQ; Fig. 3 G), suggesting that the interaction between PpV and Wdb (Fig. 3, H and I) facilitates Wdb protein degradation in the proteasome. This conclusion was further supported by the observation that elevated PpV expression increased the degree of K48-linked poly-ubiquitination modification on Wdb (Fig. 3, J and K). Consistently, when all the lysine residues in Wdb protein, which may be potentially modified by ubiquitin (Ub) moiety, were mutated to arginine (Wdb^KR), the poly-ubiquitination modification on Wdb was significantly reduced (Fig. 3 L). As a result, Wdb^KR protein became stabilized and was no longer subjected to regulation by PpV (Fig. 3 M; cf. Fig. S2 A).

In vertebrates, the conserved PP6 heterotrimeric holoenzyme is composed of catalytic subunit (PP6C), one of three scaffolding ankyrin repeat domain–containing regulatory subunits and one of three Sit4-associated proteins domain–containing regulatory subunits (PP6R; Fig. 4 A; Ohama, 2019). Over the past decade, PP6C/PpV and its regulatory subunits have been reported to regulate several substrates in diverse cellular processes (Kajino et al., 2006; Mi et al., 2009; Zeng et al., 2010; Hosing et al., 2012; Ertych et al., 2016; Ma et al., 2017; Liu et al., 2019, 2020). Recently, phosphoproteomics in mammalian cells identified >200 potential PP6 substrates, implying that PP6 functions in a broad range of biological processes (Rusin et al., 2015).

Given the fact that PpV is the catalytic subunit of PP6 holoenzyme in Drosophila (Fig. 4 A), we reasoned that PpV may rely on its phosphatase activity to modulate Wdb protein stability. Endogenous as well as overexpressed Wdb protein in S2 cells migrated as a doublet with an apparent molecular mass of ∼65 kD. Upon treatment with calf intestine phosphatase, both bands migrated faster when compared with untreated lysates (Fig. S3, A and B), suggesting that both forms of Wdb were phosphorylated, an observation also supported by enrichment of endogenous phospho-Wdb protein (Fig. S3 C). However, to our surprise, manipulating the expression of PpV in PpV knockout larvae (Fig. 3 A) or in wing discs expressing PpV RNAi (Fig. 3 D) or in S2 cells overexpressing PpV (Fig. S2 A) had little effect on the electrophoretic migration of endogenous or ectopically produced Wdb protein, raising a likely possibility that PpV-regulated Wdb protein abundance is independent of its phosphatase activity.

To test this hypothesis directly, we generated a phosphatase inactive form of PpV (PpV*; Fig. 4 B). When introduced into a WT background containing an active allele, PpV* lost its phosphatase activity to dephosphorylate Aurora A kinase, a known PP6 substrate (Fig. 4 C; Zeng et al., 2010). We found that PpV* retained the ability to interact with (Fig. 4 D) and to reduce protein abundance of Wdb in S2 cells (Fig. 4 E). Furthermore, overexpressed PpV* induced hyper-phosphorylation and translocation of Smo from the cytosol to the plasma membrane (Fig. 4, F–G′′). The phosphatase activity of PP6 holoenzyme requires the coordination of its obligate Sit4-associated proteins domain–containing regulatory subunit. fiery mountain (fmt) encodes the sole regulatory subunit of the fly PP6 (Fig. 4 A; Ma et al., 2017). When fmt function was compromised in wing discs either in fmt loss-of-function clones or by RNAi, however, there was no obvious change in Hh target expression (Fig. 4, H–I′; and Fig. S3, D and D′). These results were consistent with the observation that fmt overexpression was not able to reduce Wdb protein abundance in S2 cells (Fig. 4 J). Taken together, the above data indicate that the PpV regulation of Wdb does not rely on its phosphatase activity.

Since the phosphatase activity of PpV was not required for regulation of Wdb stability, we sought alternative mechanisms. It is known that the assembly of PP2A heterotrimeric complex enhances the stability of individual constituent subunits (Li et al., 2002; Silverstein et al., 2002; Janssens et al., 2003; Strack et al., 2004; Chen et al., 2005). Indeed, when Microtubule star (Mts), the catalytic subunit of PP2A (PP2Ac), was removed from the PP2A holoenzyme in S2 cells by RNAi, the amount of endogenous Wdb protein present in cell lysates was significantly reduced (Fig. 5 A). This was likely a consequence of shorter half-life of Wdb in cells treated with mts RNAi when compared with the control Gal80 RNAi (Fig. 5 B). Similarly, the half-life of mutant Wdb protein was reduced (Fig. 5 C) when its binding sites for Mts (Xu et al., 2006) were removed (Fig. 5 D). Concurrently, increased degree of Wdb ubiquitination was observed (Fig. 5 E). As PpV and Mts serve as respective catalytic subunits of PP6 and PP2A holoenzymes, which are two structurally closely related phosphatases (Kong et al., 2009), the above results led us to hypothesize that PpV may compete with Mts for assembly of PP2A holoenzyme, resulting in destabilization of Wdb. To directly test this hypothesis, we examined in S2 cells the interaction between Wdb and Mts following alteration of PpV expression. We found that ectopic PpV expression significantly decreased the amount of Mts coimmunoprecipitating with Wdb protein (Fig. 5 F). More strikingly, this competition between Mts and PpV for Wdb association was dose dependent (Fig. 5, G–I). Taken together, our results uncover a new type of phosphatase regulation in which PpV facilitates Wdb protein degradation through its ability to compete with Mts, thus preventing Wdb from forming a functional stable Wdb-PP2A complex (Fig. 5 J).

Having established how PpV regulates Wdb activity, we next examined the relationship between PpV and graded Hh signaling. We found that both the activity (Fig. 6 A) and the protein abundance of Wdb (Fig. 6, B and D–D′′) were negatively regulated by Hh signaling. We further found this regulation to be post-transcriptional, as altering Hh signaling in the wing disc had little effect on wdb transcription (Fig. 6, C and E–G′′). Since wdb acts epistatically to PpV, we therefore entertained the possibility that Hh signaling may directly regulate PpV expression to control Wdb-PP2A activity. Thus, we generated an antibody specific for endogenous PpV protein (Fig. 3, A and D; and Fig. S4 A′). Immunofluorescence analyses revealed an intriguing PpV expression pattern (Fig. 7 G′′): it accumulated at a higher level in anterior compartment of the wing disc, with the exception of the posterior-most cells of the anterior compartment, where PpV protein accumulation declined sharply (Fig. 7 B). These posterior-most cells (shown in purple in Fig. 7, B and B′) also expressed a higher level of Smo (Denef et al., 2000). We reasoned that these cells must receive the highest concentration of Hh morophogen, as they partially overlap with the cells whose CiFL was converted to a labile form of active Ci (CiA; i.e., lower-level), in response to high-level Hh signaling (shown in violet and purple in Fig. 7, B and B′). Taken together, the expression pattern of PpV led us to propose that Hh signaling activates PpV expression, except for cells receiving the highest levels of Hh signaling, which represses its expression.

To test this hypothesis, we manipulated Hh signaling in wing discs and examined its effect on the level of endogenous PpV protein. Consistent with our hypothesis, when ci was overexpressed along the anterior–posterior (a-p) boundary, elevated PpV expression was observed (Fig. S4, B and B′). This result also holds in ptc^S2 homozygous mutant clones: PpV expression was significantly up-regulated in clones away from the a-p boundary (Fig. 7, C–C1′′′). Furthermore, when the a-p boundary displaying the highest signaling (i.e., with labile CiA) was expanded by overexpressing ptc RNAi in the dorsal compartment of the wing disc (Fig. 7 D), PpV expression was reduced accordingly (arrows in Fig. 7, E–E′′). As the cells receiving the highest Hh signaling activate engrailed (en), which can serve as a repressor for anteriorly expressed genes required for wing development, such as ptc and dpp (Guillén et al., 1995; Sanicola et al., 1995; Tabata et al., 1995), we asked whether En contributed to the repression of PpV. Indeed, knocking down en successfully restored the expression of PpV (arrow in Fig. 7 F′′) in expanded a-p boundary, where cells exhibited highest Hh signaling activity resulted from ptc RNAi (Fig. 7 F′′′). Taken together, the above results demonstrate that PpV expression indeed responds differentially to graded Hh signaling.

As the expression of PpV protein is differentially regulated by Hh signaling, we then asked if this regulation takes place at transcriptional level. We performed in situ hybridization in WT wing discs to reveal the transcription profile of PpV. Similar to the results of PpV antibody staining, PpV transcript was expressed at a higher level in anterior compartment of the wing disc (Fig. 8 B′). Furthermore, mRNA expression pattern altered similarly to protein expression pattern when Hh signaling activity was manipulated (Fig. S5, B and C), highly consistent with PpV serving as a direct transcriptional target of Hh signaling.

Most Hh targets rely on the presence of clustered Ci/Gli consensus sites for binding of Hh signaling transcription factors (Gurdziel et al., 2015). Upon scanning the PpV locus using the Genomatix MatInspector algorithm (http://www.genomatix.de), we identified two clusters of putative Ci/Gli binding sites, as well as three scattered sites spanning a 4-kb region upstream of the PpV transcription start site (PpV 5′; Fig. 8 A and Fig. S5 A). To determine functional importance of these putative binding sites, we first constructed a GFP reporter transgene under control of the PpV 5′ (Fig. 8 A). The expression pattern of the resulting PpV 5′-GFP reporter correlated well with that of the PpV mRNA (Fig. 8 C′). More importantly, the PpV 5′-GFP reporter responded sufficiently to altered Hh signaling (Fig. S5, D–E′′). Having shown the importance of this regulatory region in the PpV locus for PpV transcription, we then examined whether putative Ci/Gli binding sites present in this region are capable of mediating direct interactions with the fly Hh signaling transcription factor Ci. The DNA-binding zinc-finger domain of the Ci protein (Ci ZnF), which binds efficiently to verified Ci binding sites (ptc-5) in the ptc locus (Fig. 8 D; Gurdziel et al., 2015), was purified and used in the electrophoretic mobility shift assay. Our DNA binding and competition assays (shown in Fig. 8 D) showed direct and sequence-specific binding of Ci ZnF to all five consensus sites, supporting a potential role of these Ci/Gli binding sites in the PpV 5′ in mediating the effect of graded Hh signaling on PpV expression.

To conclusively demonstrate the requirement of the Ci/Gli binding sites in the PpV 5′ on PpV expression in vivo, we generated PpV^5′KO, a PpV regulatory sequence deletion mutant in which the PpV 5′ containing all five Ci/Gli binding sites was removed (Fig. 8 A). As predicted, PpV no longer accumulated at the higher level in the anterior compartment of hemizygous PpV^5′KO wing discs (Fig. 8 E′; cf. Fig. 7 A′′). Consequently, PpV^5′KO also lost its ability to respond to heightened Hh signaling (Fig. 8 F′; cf. Fig. S4 B′). Taken together, our results provide sufficient evidence for PpV as a bona fide Hh signaling target whose transcription is maintained by graded Hh signaling.

Taking advantage of fly genetics, we uncover a noncanonical role of protein phosphatase PpV in transducing high-level Hh signaling. The identification of Wdb, the regulatory subunit of PP2A, as a PpV target provides a mechanistic basis for PpV regulation of Hh signaling. PpV interferes with assembly of the Wdb-PP2A complex, in turn facilitating Wdb ubiquitination and its ultimate degradation in proteasome. Importantly, PpV itself is a bona fide transcriptional target of Hh signaling, forming a negative feedback loop between two closely related phosphatases to maintain homeostatic Hh signaling.

The conversion of extracellular Hh morphogen gradient into distinct transcriptional outputs relies on graded Smo phosphorylation. This stereotypical Smo phosphorylation profile is maintained by coordinated actions of kinases and phosphatases. A key unresolved question is whether and how the activities of these kinases and phosphatases are regulated when Hh signaling fluctuates. In our study, we showed that Wdb-PP2A activity negatively correlates with Hh signaling, providing a mechanism for ensuring a sufficient amount of highly phosphorylated Smo species for transducing high-level Hh signaling. However, the regulation of Wdb by Hh is indirect. Here, through an unbiased genetic screen, we identify PpV as the protein that directly responds to the Hh gradient. Importantly, PpV also serves as a direct Hh signaling sensor that coordinates Wdb-PP2A activity and graded Smo phosphorylation. When PpV expression is reduced or Wdb-PP2A activity is enhanced in the developing wing, the Smo phosphorylation profile is shifted from the highly phosphorylated Smo state toward moderate and basal levels of Smo phosphorylation, resulting in elevated low to intermediate levels of Hh signaling at the expense of high-level signaling. By acting both as a sensor and target of Hh signaling, and participating in feedback regulatory loops, PpV contributes to the robustness and reliability of Hh signaling.

As the PpV locus contains clustered, functional Ci binding sites, Hh signaling is sufficient to activate PpV transcription in the anterior compartment of the wing disc. However, we noticed that the expression level of PpV is lower in anterior cells immediately adjacent to the a-p boundary. These cells likely exhibit the highest Hh activity, as they maintain higher levels of Smo and coincide with the posterior half of the CiA-expressing region. Consistently, the reduced PpV expression becomes even more apparent when Hh signaling is hyper-activated abutting the a-p boundary. Therefore, in cells with excess Hh signaling, it is necessary to reduce the PpV level, which primes cells for heightened Wdb-PP2A activity to prevent Smo from hyper-phosphorylation. Hence, Hh signaling homeostasis is maintained. How cells sense excess Hh morphogen concentration to tune down PpV transcription is not known. Nevertheless, observation of the same phenomenon in wing discs expressing a PpV 5′-GFP reporter suggests that the control elements differentially regulating PpV expression most likely reside in the 5′ regulatory region. In addition to the Ci consensus sites, an array of binding sites for other transcription factors, including En/EN1, are present (Fig. S5 A). Intriguingly, En can serve as a repressor for anteriorly expressed genes required for wing development, such as ptc and dpp (Guillén et al., 1995; Sanicola et al., 1995; Tabata et al., 1995). As En expression extends to anterior cells immediately adjacent to the a-p boundary, we believe that combinatory activities of Ci and En may potentially facilitate dynamic regulation of PpV expression in response to differential Hh signaling. Thus, it would be interesting to investigate how En activity at the a-p boundary reduces PpV expression.

Coordinated cell signaling is required for development and adult homeostasis. Feedback and feed-forward mechanisms are used to robustly define the behavior of cell signaling, while negative feedback loops are important for homeostatic maintenance of signaling activity. Negative regulators are the key to a feedback regulatory network, as they often contribute sensitivity and dynamic regulation, rather than simply terminating the signal transduction (Kholodenko, 2006; Alon, 2007; Brandman and Meyer, 2008; Lemmon et al., 2016). The most well-known negative feedback loop in Hh signaling is mediated by the Hh receptor Ptc, which is itself a transcriptional target of Hh signaling, shutting down the signal activation through Hh sequestration (Chen and Struhl, 1996; Strutt et al., 2001; Torroja et al., 2004). Even though additional negative regulators have been identified for homeostatic control of the transcriptional factor Ci or essential activator Smo through ubiquination-mediated degradation pathway and kinase/phosphatase cycles (Briscoe and Thérond, 2013; Lee et al., 2016; Pak and Segal, 2016), very few regulators respond directly to graded Hh signaling at the transcriptional level (Atwood et al., 2013). The identification of PpV as a graded Hh signaling sensor provides an innovative feedback regulatory mechanism, with enhanced self-regulatory capacity to buffer fluctuations in Hh morphogen gradient during development.

Increasing evidence demonstrates that PP2A dysregulation is a repeated theme in many diseases, including neurodegenerative disorders, cardiovascular diseases, and cancer (O’Connor et al., 2018). Various endogenous and exogenous inhibitors of PP2A have been shown to contribute significantly to the development of PP2A-related diseases, making them valuable therapeutic targets (Sangodkar et al., 2016; O’Connor et al., 2018). The most prominent mechanism used to inhibit the phosphatase activity in cancer cells is competitive binding of inhibitors to different PP2A subunits, thereby preventing the interaction of the phosphatase active site with substrate proteins (Junttila et al., 2007). Indeed, PP2A reactivation drugs specifically targeting these inhibitors, including cancerous inhibitor of PP2A (CIP2A) and SE translocation oncoprotein (SET), have been successfully developed (Sangodkar et al., 2016; O’Connor et al., 2018).

Our study uncovers a new inhibitory mechanism by which an endogenous catalytic subunit of a PP2A family phosphatase (i.e., PP6C), which is structurally closely related to PP2Ac, competitively blocks the PP2A holoenzyme assembly. Based on our study, we suspect that PpV/PP6C may have a higher affinity for the regulatory subunit Wdb than that of Mts/PP2Ac. Upon PpV binding, Wdb may allosterically change its conformation that interferes with its ability to interact with Mts/PP2Ac. Although the resulting PpV-Wdb complex may be active, its highly unstable nature might prevent it from functioning on PP2A substrates. Our finding highlights a possible use of PP6 as an alternative therapeutic strategy for PP2A activation. Recent studies reveal that allosteric small molecule activators of PP2A specifically stabilize the PPP2R5A (B56α)- and PPP2R2A (B55α)-PP2A holoenzymes in a fully assembled active state to dephosphorylate selective substrates, such as c-Myc (Leonard et al., 2020; Morita et al., 2020), while phenothiazine analogues improved heterocyclic PP2A activators activate PP2A complexes that contain only the PPP2R5E (B56ε) regulatory subunit (Morita et al., 2020). As there are more than 26 regulatory B subunits known to complex with PP2Ac to confer its substrate specificity, identifying the PP2A regulatory subunit whose stability is regulated by PP6 would aid the development of novel therapeutic targets.

Fly cultures and crosses were performed according to standard procedures. Drosophila stocks were ordered from stock centers or obtained from other laboratories and are listed as follows: ap-Gal4 (Bloomington Drosophila Stock Center [BDSC]; 3041), ci^cell (BDSC; 4343), dpp-Gal4 (BDSC; 1553), dpp-lacZ (BDSC; 8411), fmt¹ (BDSC; 82141; Ma et al., 2017), fmt^f02530 (BDSC; 18559), hs-Flp¹²²; Act>y+>Gal4, UAS-nls::gfp; Tub-Gal80, FRT79E (a gift of J. Pastor-Pareja, Tsinghua University, Beijing, China), hs-Flp¹²²; FRT42D, Ubi-gfp (a gift of J.E. Treisman, New York University, New York, NY), MS1096-Gal4 (BDSC; 8860), ptc^S2 (BDSC; 6332), salm-Gal4 (BDSC; 5818), UAS-ci-HA (Domínguez et al., 1996), UAS-ci RNAi (BDSC; 28984), UAS-en RNAi (Vienna Drosophila RNAi Center [VDRC]; 35697; Dietzl et al., 2007), UAS-fmt RNAi (VDRC; 16005), UAS–N-terminal of Hh protein, an active form of Hh [hhN]; Porter et al., 1995), UAS-mCD8::gfp (BDSC; 5130), UAS-ptcB1 (BDSC; 5817; Johnson et al., 1995), UAS-ptc RNAi (BDSC; 28795), UAS-PpV RNAi (VDRC; 101997; and VDRC; 31690), UAS-puc (Ma et al., 2017), UAS-wdbDN (Hannus et al., 2002), UAS-wdb RNAi (VDRC; 27470), UAS-w RNAi (TsingHua Fly Center; HU0583) and yw (BDSC; 1495). Detailed genetic crosses shown in each figure are listed in Table S1.

PpV^KO flies were generated through homologous recombination with donor plasmid pTV-Cherry obtained from J.-P. Vincent (The Francis Crick Institute, London, UK). Details were performed according to procedures previously described (Baena-Lopez et al., 2013). Briefly, ∼3 kb genomic sequence flanking the PpV coding sequence was cloned into pTV-Cherry ashomology arms to generate donor flies. After crossing to flies expressing FLP; I-SceI and then Ub-Gal4(3xP3-GFP), red-eyed F2 homologous recombinant flies were collected and confirmed by PCR. PpV^5′KO flies were generated by CRISPR/Cas9-mediated homologous recombination (Port et al., 2014). hs-wdb-HA and PpV 5′-gfp reporter transgenic flies were generated by P-element–mediated germline transformation. Primers used for generating fly transformation constructs are listed in Table S2.

For conventional immunofluorescence staining, wing discs dissected from third instar larvae were fixed in 4% paraformaldehyde, blocked in 0.2% BSA for 1 h, and incubated overnight at 4°C with primary antibodies: mouse anti-β-galactosidase (1:200; Developmental Studies Hybridoma Bank [DSHB]; 40-1a), rat anti-Ci full-length (1:10; DSHB; 2A1), rat anti-Col (1:2,000; generated in this study), rabbit anti-GFP (1:1,000; Thermo Fisher Scientific; A-11122), mouse anti-En (1:500; DSHB; 4D9), mouse anti-HA (1:1,000; Cell Signaling Technology [CST]; 2367S), rabbit anti-PpV (1:100; generated in this study), mouse anti-Smo (1:20; DSHB; 20C6), and guinea pig anti-Wdb (1:500; a gift of A. Sehgal, University of Pennsylvania, Philadelphia, PA). For cell staining, Drosophila Schneider S2 cells (American Type Culture Collection [ATCC]; CRL-1963) grown on coverslips were fixed with 4% paraformaldehyde before direct visualization. The coverslips seeded with S2 cells were precoated with poly-L-lysine (Sigma-Aldrich). To visualize transiently transfected S2 cells expressing gfp-smo, we used mouse anti-Smo (DSHB; 20C6) antibody to stain fixed but not permeabilized cells. Under this condition, 20C6 was able to detect GFP-Smo on the surface of S2 cells. The wing discs or cells were then incubated with goat anti-mouse/rabbit/rat/guinea pig Alexa Fluor–conjugated secondary antibodies (1:400; Thermo Fisher Scientific; A-11001, A-11004, A-11006, A-11008, A-11011, A-11075, A21247, and A-31553) for 1 h at room temperature before mounting in VECTASHIELD (Vector Laboratories; H-1200).

The in situ hybridization was processed with a standard protocol described previously (Su et al., 2011). Briefly, third instar larval wing discs with corresponding genotypes were dissected, fixed, dehydrated, hydrated, and then incubated with specific RNA probes synthesized using a Digoxigenin RNA Labeling Kit (Roche; 11175025910). Discs were incubated overnight with HRP-conjugated anti-Digoxigenin antibody (Roche; 11207733910) at 4°C followed by development using a TSA Plus Fluorescence Kit (PerkinElmer; NEL741001KT) at room temperature. Primers for RNA probe synthesis are listed in Table S2.

Images were obtained on a Leica TCS SP8 confocal microscope (Leica 20×/0.70 and 63×/1.40 oil objective lenses) or a Zeiss Axio Imager Z2 microscope (Zeiss 20×/0.80 objective lens) equipped with an ApoTome as well as an AxioCam Mrm Camera. Images were processed with LAS AF X (Leica) or AxioVision 4.8.1 (Zeiss) and Adobe Photoshop CS5 to adjust brightness and/or contrast. 3D reconstitutions were used to show the side view of fly wing discs in AxioVision 4.8.1.

Adult wings were dissected and mounted in Euparal mounting medium (BioQuip; 6372A). The images were acquired with a Leica DMIL LED inverted microscope (Leica 4×/0.10 and 10×/0.22 objective lenses) with a QImaging MicroPublisher RTV-5.0 CCD Camera (QImaging).

Drosophila Schneider S2 cells were cultured at 25°C in Schneider’s Drosophila medium (Thermo Fisher Scientific; 21720024) supplemented with 10% FBS (Lanzhou Bailing; 20130507) and 100 U/ml of penicillin/streptomycin (Thermo Fisher Scientific; 15140122). HEK293T cells (ATCC; CRL-1573) were maintained in DMEM (Thermo Fisher Scientific; C11995500BT) with 10% FBS and 100 U/ml of penicillin/streptomycin at 37°C with 5% CO₂. DNA transfections were performed using a standard calcium phosphate protocol, Effectene Transfection Reagent (Qiagen; 301425), or Lipofectamine 2000 Transfection Reagent (Thermo Fisher Scientific; 11668019). Cell cycle arrest was induced by treating HEK293T cells for 24 h with colchicine (30 µM; JiSiEnBei; JS0420).

In protein degradation assays, S2 cells were treated for up to 5 h with cycloheximide (50 mg/ml; Sigma-Aldrich; 5087390001) before harvest to inhibit nascent protein synthesis. MG132 (20 µM; Sigma-Aldrich; SML1135) was used to inhibit proteasome activity, while chloroquine (50 µM; Sigma-Aldrich; C6628) was used to inhibit lysosome function. S2 cells transfected with hs-wdb-HA, and indicated vectors were heat-shocked for half an hour at 37°C after transfection for 36 h. The cells were then recovered at 25°C for 1 h followed by drug treatment. 1 mM CuSO₄ was used to induce the expression of PpV, PpV*, or OsTIR1 cloned into pMT vector. 1 mM auxin (Sigma-Aldrich; I2886) was used to induce the formation of OsTIR1-SCF E3 complex.

dsRNA was generated with the RiboMAX Large Scale RNA Production System (Promega; P1280) according to the manufacturer’s instructions. DNA templates targeting PpV (encoding amino acids 3–149), wdb^#1 (encoding amino acids 246–346), wdb^#2 (encoding amino acids 268–315), and mts (encoding amino acids 93–211) were generated by PCR and used for dsRNA synthesis. dsRNA targeting yeast Gal80 coding sequence was used as a negative control. For RNAi knockdown in S2 cells, dsRNA transfection was performed using a standard calcium phosphate protocol. Primers used to generate dsRNAs are listed in Table S2.

S2, HEK293T cells, and third instar larval wing discs were lysed in NP-40 buffer (1% NP-40, 150 mM NaCl, and 50 mM Tris-HCl, pH 8) or radioimmunoprecipitation assay buffer (1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate, 150 mM NaCl, and 50 mM Tris-HCl, pH 7.4) supplemented with protease inhibitor cocktail (Roche; 11697498001). Phosphatase inhibitors (25 mM NaF and 400 mM Na₃VO₄) were added to the lysis buffer when necessary. Immunoblotting was performed using standard protocols. The following antibodies were used for immunoblotting: mouse anti-β-tubulin (1:5,000; DSHB; E7), mouse anti-human GAPDH (1:1,000; DSHB; 2G7), mouse anti-Flag (1:2,000; CMC-Tag; A0022), rabbit anti β-actin (1:1,000; ABclonal; AC026), rabbit anti-Phospho-Aurora A (Thr288; 1:500; ABclonal; AP0523), rabbit anti-GFP tag (1:1,000; Thermo Fisher Scientific; A-11122), mouse anti-HA tag (1:2,000; 6E2; CST; 2367S), mouse anti-Myc tag (1:2,000; 9B11; CST; 2276S), rabbit anti-PP2A A subunit (1:1,000; CST; 2039S), rabbit anti-PP2A C subunit (1:1,000; CST; 2038S), rabbit anti-PpV (1:2,000; generated in this study), guinea pig anti-Wdb (1:2,000; a gift of A. Sehgal), mouse anti-Ub (1:2,000; P4D1; Santa Cruz; sc-8017), mouse anti-p44/42 MAPK (Erk1/2; 1:1,000; ENZO; BML-MA1366-0025), mouse anti-Hsp60 (LK-1; 1:1,000; ENZO; ADI-SPA-806-D), and HRP-conjugated goat anti-mouse/rabbit/guinea pig IgG (H+L; 1:10,000; ABclonal; AS003, AS014, and AS025). Immunoblots presented in all figures are representatives of at least three independent experiments.

Immunoprecipitation was performed using either agarose anti-HA (Vector Labs MB-0734) or Dynabeads (Thermo Fisher Scientific; 10006D) according to the manufacturer’s instructions. For ubiquitination assays, S2 cells transfected with indicated vectors were hot-lysed in denaturing buffer (1% SDS, 50 mM Tris, pH 7.5, and 0.5 mM EDTA) by boiling for 5 min. Lysates were then diluted 10-fold with NP-40 lysis buffer and subjected to immunoprecipitation. In some experiments, mutant forms of Ub (Ub^K48 or Ub^K63) were used to investigate the specificity of Ub modification on substrates. In Ub^K48 or Ub^K63, all the lysine residues were mutated to arginine residues except at amino acid position K48 or K63, which allowed the poly-Ub linkage only occurring at K48 or K63, respectively.

For phospho-protein enrichment, S2 cells were lysed in NP-40 buffer. Phospho-proteins in the lysates were then purified using PhosphoProtein Purification Kit (Qiagen; 37101) according to the manufacturer’s instructions.

To measure the PP2A activity, third instar larvae were dissected, and 50–80 pairs of imaginal wing discs were collected and lysed in NP-40 buffer followed by PP2A activity measurement. Normalized PP2A activity of 1 µg wing disc protein lysate was determined by measuring phosphate release using a nonradioactive, malachite-green based Serine/Threonine Phosphatase Assay System (Promega; V2460). A synthetic phosphopeptide RRA(pT)VA was used as a specific PP2A substrate. 50 mM NaF was used to inhibit the PP2A activity.

To purify Ci ZnF protein, nucleotides encoding the zinc finger domain (amino acids 349–516) of Ci were cloned into pGEX-6P-1 (Addgene; 27–4597-01). GST-tagged protein was purified by glutathione agarose (Thermo Fisher Scientific; 16100) in equilibration/wash buffer (50 mM Tris and 150 mM NaCl, pH 8.0). After wash, fusion protein was eluted with 100 mM glutathione in equilibration/wash buffer. Protein concentration was estimated by Coomassie-stained gels. Known concentrations of BSA were used as protein concentration standards.

For the electrophoretic mobility shift assay, purified GST-tagged Ci ZnF protein and 50 fmol biotin-labeled (or 5 pmol un-labeled “cold”) double-stranded DNA probes were incubated at 25°C in 20 µl reaction buffer (20 mM Hepes, pH 7.9; 50 mM KCl; 0.1 mM EDTA; 2 mM DTT; 6 mM MgCl₂; 0.1 mg/ml BSA; 50 µg/ml poly[dI-dC], Thermo Fisher Scientific; 20148E; and 5% glycerol) for 30 min. The reaction mixture was loaded and resolved in 8% Tris-borate-EDTA (TBE) gel. About 400 ng of GST–Ci ZnF recombinant protein were used per reaction.

Total RNA of fly wing imaginal discs was extracted using Magzol reagent (Magen; R4801-02). Residual genomic DNA was removed by gDNA remover supplied by Promega Eastep RT Master Mix Kit (Promega; LS2050). First-strand cDNA was synthesized using oligo-dT primer (Thermo Fisher Scientific; 18418012) and SuperScript III reverse transcription (Thermo Fisher Scientific; 18080093).

HA-tagged wdb was generated by cloning the coding as well as the 5′ untranslated regulatory sequence amplified from pUAST-wdb transgenic fly (Hannus et al., 2002) into pCaspeR-hs vector (Drosophila Genomics Resource Center [DGRC]; 1215) derived from pCaSpeR.

The PpV-Myc construct was generated by fusing a Myc tag at the C terminus of the full-length PpV cDNA and then cloned into the pMT vector (DGRC; 1145).

To generate PpV*-Myc plasmid, two mutations (H53Q and R83A) were induced by primer-based mutagenesis (I-5; Molecular Cloning Laboratories; I5HD-200).

The full-length PpV and PpV* were cloned into the pcDNA3.1 vector (ATCC; V790-20) to generate pcDNA3.1-PpV and pcDNA3.1-PpV*.

The OsTIR1-Myc and mAID were cloned from pMK289 (mAID-mClover-NeoR; Addgene; 72827) and pMK232 (CMV-OsTIR1-PURO; Addgene 72834), respectively. Then the OsTIR1-Myc and mAID were cloned into the pMT-gfp vector. Then the full-length PpV was cloned into this vector to generate pMT-OsTIR1-Myc-P2A-PpV-mAID-gfp vector.

The PpV 5′-gfp reporter construct was generated by fusing a GFP tag in frame after the PpV 5′ regulatory sequence (−3684 +3) and then cloned into pB-ARE-Green (a gift of D.P. Bohmann, University of Rochester, Rochester, NY).

To generate Wdb^ΔMts, nucleotides that encode amino acids 129–135, 268–270, 304–310, 351–353, and 394 were deleted from hs-wdb-HA plasmid.

The AURKA-HA construct was generated by fusing a HA tag at the C terminus of the full-length AURKA cDNA and then cloned into the pcDNA3.1 vector.

The pUAST-Myc-Ub, pUAST-Myc-Ub^K48, and pUAST-Myc-Ub^K63 were gifts from A. Plessis (Institut Jacques Monod, Paris, France).

A rabbit polyclonal antibody was raised against a synthetic peptide (AVPDAERVIPKQNTTP) corresponding to amino acids 285–300 of fly PpV (Abclonal). Its specificity was confirmed by immunoblotting (1:1,000) and immunostaining (1:100). Note this PpV antibody was able to cross-react with human PP6C.

A rabbit polyclonal antibody was raised against full-length fly Col protein (Abclonal). Its specificity was confirmed by immunostaining (1:2,000).

For quantification of the distance between two points in adult wing blade, the mounted wing blade was imaged, and the distance between distal ends of L3 and L4 longitudinal veins and the distance between distal ends of L2 and L5 veins were measured by QCapture Pro software (QImaging).

For quantification of the amount of PpV or Mts protein in association with Wdb after coimmunoprecipitation, the amounts of Wdb, Mts, or PpV were determined by the National Institutes of Health ImageJ software on the basis of the mean pixel intensity of each band shown in the blot. Relative amounts of PpV or Mts protein associated with Wdb were then measured by calculating the pixel intensity ratio between the PpV or Mts bands and relevant Wdb bands in each lane.

For quantification of PP2A activities, 1 µg protein lysate extracted from wing discs expressing UAS-gfp or UAS-hhN was determined by measuring phosphate release using a nonradioactive, malachite-green–based phosphatase assay system. A synthetic phosphopeptide RRA(pT)VA was used as the specific PP2A substrate. The reading from wing discs expressing UAS-gfp was set as 100% of the PP2A activity (basal activity).

For quantification of the intensity of antibody staining, images were taken with the same confocal settings and the mean fluorescence intensity was measured with the National Institutes of Health ImageJ software.

Two-tailed Student’s t tests were used to analyze the difference between two different genotypes.

Fig. S1 shows genetic interactions between PpV and Hh signaling components. Fig. S2 shows two degradation pathways used by exogenous Wdb protein in S2 cells. This figure also shows the alteration of Wdb protein stability upon manipulating PpV expression in S2 cells and in wing discs. Fig. S3 shows the phosphorylation pattern of Wdb protein in S2 cells. Fig. S4 shows the characterization of PpV antibody in immunofluorescence staining and the alteration of PpV protein abundance upon manipulating Hh signaling. Fig. S5 shows the PpV mRNA transcription and activity of PpV 5′ regulatory elements when Hh signaling was manipulated in wing discs. Table S1 shows all the genetic crosses for figures and supplemental figures. Table S2 lists all primers used in the study.

We thank Drs. D.P. Bohmann, J. Pastor-Pareja, A. Plessis, A. Sehgal, J.E. Treisman, J.-P. Vincent, Bloomington Drosophila Stock Center (BDSC) at Indiana University, Bloomington, Tsinghua Fly Center at Tsinghua Universisty (THFC), Beijing, Vienna Drosophila RNAi Center (VDRC) at Vienna Biocenter Core Facilities, Vienna, and Developmental Studies Hybridoma Bank (DSHB) at the University of Iowa, Iowa City (created by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health and maintained at the University of Iowa, Iowa City) for fly stocks, antibodies, and plasmids. We also thank Dr. C. Shan at the National Center for Protein Science at Peking University for assistance with microscopic imaging.

This work was supported by grants from the National Natural Science Foundation of China (31725019, 31830058, and 31671512 to A.J. Zhu, and 31701274 to Y. Su), the Peking-Tsinghua Center for Life Sciences (to A.J. Zhu and M. Liu), and the Ministry of Education Key Laboratory of Cell Proliferation and Differentiation (to A.J. Zhu). J. Wang, Y. Zhang, and Y. Li, are Peking University President's Scholarship awardees. BDSC and DGRC, which are supported by the grants from National Institutes of Health P40OD018537 and 2P40OD010949, respectively, provided some fly lines used in this study.

The authors declare no competing financial interests.

Author contributions: M. Liu, A. Liu., Y. Su, and A.J. Zhu designed the study; M. Liu, A. Liu, J. Wang, Y. Zhang, Y. Li, and Y. Su performed experiments and data analyses; and M. Liu, A. Liu, and A.J. Zhu wrote the paper.