Inappropriate regulation of the PI3-kinase/PTEN/Akt kinase-signalling cassette, a key downstream target of insulin/insulin-like growth factor signalling (IIS), is associated with several major human diseases such as diabetes, obesity and cancer. In Drosophila, studies have recently revealed that different subcellular pools of activated, phosphorylated Akt can modulate different IIS-dependent processes. For example, a specific pool of activated Akt within the cytoplasm alters aspects of lipid metabolism, a process that is misregulated in both obesity and diabetes. However, it remains unclear how this pool is regulated. Here we show that the protein phosphatase PP2A-B′ regulatory subunit Widerborst (Wdb), which coimmunoprecipitates with Akt in vivo, selectively modulates levels of activated Akt in the cytoplasm. It alters lipid droplet size and expression of the lipid storage perilipin-like protein LSD2 in the Drosophila ovary, but not in epithelial cells of the eye imaginal discs. We conclude that isoforms of PP2A-B′ can act as subcellular-compartment-specific regulators of PI3-kinase/PTEN/Akt kinase signalling and IIS, potentially providing new targets for modulating individual subcellular pools of activated Akt in insulin-linked disease.
The signalling cassette involving Class I phosphatidylinositol 3-kinase (PI3K), phosphatase and tensin homologue on chromosome 10 (PTEN) and Akt (also known as protein kinase B or PKB) is part of a major intracellular kinase cascade that regulates multiple cellular functions including metabolism, growth, proliferation and survival. It responds to a variety of stimuli, such as insulin, other growth factors including PDGF and FGF, and attachment to the extracellular matrix (Downward, 1998; Liang and Slingerland, 2003; Goberdhan and Wilson, 2003a). Upon activation, PI3K catalyses the formation of phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3] from phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2]. PtdIns(3,4,5)P3 is a lipid second messenger, which in turn recruits the PH-domain-containing Akt protein kinase from the cytosol to the plasma membrane (Lizcano and Alessi, 2002). Here it is activated through phosphorylation at Thr308 by 3-phosphoinositide-dependent protein kinase 1 (PDK1) and at Ser473 (or Ser505 in the unique Drosophila Akt kinase, Akt1) by PDK2, which is thought to be the rictor-mTOR complex (Sarbassov et al., 2005b). Once activated, Akt subsequently phosphorylates multiple targets, leading to its numerous downstream effects (Downward, 1998).
Misregulation of Akt and its cellular targets is linked to several major human diseases. For example, cellular insulin resistance is associated with reduced signalling by the PI3K/PTEN/Akt cassette and is an important defect in individuals suffering from Type 2 diabetes (Jackson, 2006). By contrast, hyperactivation of this cassette, most notably through loss-of-function mutations in the tumour suppressor PTEN, which converts PtdIns(3,4,5)P3 back to PtdIns(4,5)P2, is strongly associated with many forms of human cancer (Sansal and Sellers, 2004).
Molecular genetic studies in Drosophila have given us several fundamental insights into the regulation and functions of the PI3K/PTEN/Akt-signalling cassette (Goberdhan and Wilson, 2003a; Hafen, 2004). Not only has this work highlighted the central importance of nutrient-regulated insulin/insulin-like growth factor signalling (IIS) in controlling the activity of this cassette and cell growth, but it has also revealed a critical downstream link with the nutrient-sensitive mTOR-signalling cascade, which regulates several cellular processes including protein translation and autophagy (Sarbassov et al., 2005a). Furthermore, studies in invertebrates have indicated roles for PI3K/PTEN/Akt and mTOR in ageing (Giannakou and Partridge, 2007), cell polarity (Pinal et al., 2006) and neurodegeneration (Khurana et al., 2006; Ravikumar et al., 2004), functions that all appear to be conserved in mammals and which might involve a combination of cellular and metabolic defects (e.g. Cole and Frautschy, 2006; Martin-Belmonte et al., 2007).
If the role of PI3K/PTEN/Akt in insulin-linked diseases is to be fully understood, it is essential to determine how this single signalling cassette regulates so many different cellular functions. One important part of the explanation is presumably the existence of cell-type-specific downstream-signalling targets that perform different roles. However, recent work, much of it again initiated in flies, has indicated that Akt activity can also be differentially regulated in specific subcellular domains and that these subcellular pools of activated Akt can control different processes. For example, precise regulation of Akt activity at the apical membrane of epithelial cells by localised PTEN is required for normal apical morphology in higher eukaryotes (Pinal et al., 2006; Martin-Belmonte et al., 2007). By contrast, cytoplasmic activated Akt appears to be required for transcription of specific IIS target genes and regulation of lipid metabolism and droplet size in nurse cells of the Drosophila female germ line (Vereshchagina and Wilson, 2006). These observations have highlighted the importance of finding the molecules that regulate different pools of activated Akt in vivo, because their modulation might alter specific functions of IIS in health and disease more selectively.
In a screen for novel phosphatase regulators of IIS, we identified Widerborst (Wdb) (Hannus et al., 2002), one of the B′ regulatory subunits of the protein phosphatase PP2A, as a negative regulator of the PI3K/PTEN/Akt-signalling cassette. Although wdb is essential for cell viability in some tissues, wdb mutant cells in the germ line and follicular epithelium of the ovary are viable and display phenotypes that are similar to those seen in PTEN mutant ovaries. We show that Wdb and Drosophila Akt1 physically interact in the ovary, and that within this tissue, Wdb regulates the subcellular pool of activated Akt1 in the cytoplasm. This study therefore highlights an important new function for PP2A-B′ subunits in selectively modulating certain IIS-dependent processes by controlling signalling in a specific subcompartment of the cell.
A protein phosphatase overexpression screen identifies Wdb as a novel negative regulator of PI3K/PTEN/Akt-mediated signalling events
We designed an overexpression screen to identify phosphatases that negatively regulate the PI3K/PTEN/Akt signalling cassette. We found that 100% (n⩾50) of flies carrying the hypomorphic PTEN3 allele (Goberdhan et al., 1999) in combination with a deficiency for the PTEN region, Df(2L)170B, die as late third instar larvae and pupae. In the presence of one copy of the strong loss-of-function Akt11 allele (Staveley et al., 1998), 77% of PTEN3/Df(2L)170B transheterozygotes survived to adulthood. Furthermore, 40% were rescued by ubiquitous overexpression of the downstream IIS antagonists Tsc1 and Tsc2 (Gao and Pan, 2001), using the weak constitutive GAL4 driver arm-GAL4 (see Materials and Methods for details) (Brand and Perrimon, 1993). Thus, the survival of this PTEN mutant combination is highly sensitive to levels of downstream IIS signalling.
Protein phosphatases exist as a complex of catalytic and regulatory subunits, with the latter controlling specificity and localisation (Cohen, 2002; Ceulemans et al., 2002). We ubiquitously overexpressed both the catalytic and regulatory subunits of multiple Drosophila phosphatases in this sensitised PTEN mutant background, using the modular GAL4/UAS EP vector (Rorth, 1996; Rorth et al., 1998) and GS (Gene Search) (Toba et al., 1999) misexpression systems. We found that 52% of PTEN mutant flies were rescued by ubiquitous expression of the wdbGS9548 insertion, suggesting that this gene might act as a negative regulator of IIS.
wdb genetically interacts with the PI3K/PTEN/Akt-signalling cassette via a mechanism involving its PP2A regulatory activity
GS and EP insertions in wdb and other phosphatase genes were also tested using a previously developed FOXO interaction screen (Goberdhan et al., 2005). Overexpression of the transcription factor FOXO in differentiating cells of the eye using the GMR-GAL4 driver reduces the size of the eye and alters its shape in a characteristic way (Fig. 1D) (Jünger et al., 2003). Most notably, the posterio-ventral part of the eye is reduced due to increased apoptosis. Phosphorylation of FOXO by activated Akt1 directly inhibits its activity (Kramer et al., 2003; Puig et al., 2003) and co-overexpression of Akt1 (Staveley et al., 1998) with FOXO in this assay completely suppresses the FOXO phenotype (Fig. 1G). Conversely, a relatively modest reduction in PI3K/PTEN/Akt-signalling activity produced by expression of PTEN or a dominant-negative form of PI3K strongly enhances the phenotype (Goberdhan et al., 2005).
The wdb overexpression line wdbGS9548 greatly enhanced the FOXO phenotype in this assay. By itself, overexpression of wdbGS9548 under the control of GMR-GAL4 produced a disorganised eye of slightly reduced size (Fig. 1B). In combination with FOXO, no normal ommatidia were formed and the pigmented eye region was highly reduced (Fig. 1E). This latter phenotype was completely suppressed by co-overexpression of Akt1 (Fig. 1H), giving a phenotype that appeared identical to the one produced by overexpressing Akt1 and Wdb in the absence of FOXO (Fig. 1I). However, Akt1 had no obvious effect on the ommatidial disorganisation or reduced eye size observed when Wdb was expressed alone (Fig. 1B). Interestingly, a previously characterised EP insertion in wdb (wdbEP3559) (Hannus et al., 2002) produced only a very mild effect on ommatidial organisation when expressed under GMR-GAL4 control (Fig. 1C), presumably because it permits much lower expression of wdb (much weaker phenotypes compared to wdbGS9548 were also observed when wdbEP3559 was combined with several other GAL4 drivers; data not shown). Nevertheless, like wdbGS9548, wdbEP3559 strongly enhanced the FOXO phenotype in the eye (Fig. 1F). The ability of wdb overexpression to rescue a phenotype caused by PTEN mutation and synergistically enhance the effects of overexpressed FOXO, two known negative regulators of IIS-dependent functions, indicates that wdb also negatively regulates this pathway.
wdb encodes a protein closely related to members of the mammalian PP2A-B′ regulatory-subunit family (Hannus et al., 2002). To determine whether the effect of Wdb on the FOXO phenotype is mediated via its PP2A-B′ regulatory activity, we altered levels of the only PP2A catalytic subunit present in Drosophila, Microtubule star (Mts) (Wassarman et al., 1996), in the FOXO assay. Although heterozygosity for the strong mts loss-of-function allele mts02496 had little, if any, effect on the FOXO eye phenotype (Fig. 2H), it clearly suppressed the phenotype caused by co-overexpression of FOXO and Wdb (compare Fig. 2G and 2I). Interestingly, expression under GMR-GAL4 control of a dominant-negative version of Mts in which the active site of this enzyme is deleted (Hannus et al., 2002) produced a bulging overgrowth phenotype (Fig. 2E) and completely suppressed the FOXO-induced eye phenotype (compare Fig. 2F and 2J). This suggests that mts normally negatively regulates IIS. Furthermore, since mts and wdb genetically interact in the FOXO assay, the effects of Wdb appear to be mediated through its PP2A-B′ regulatory activity.
Overexpression of wdb with GMR-GAL4 (Fig. 1B) did not produce the clear effects on ommatidial size previously seen when PTEN is overexpressed (Goberdhan et al., 2005), suggesting that only some of the cellular functions that are regulated by the PI3K/PTEN/Akt-signalling cassette, such as FOXO-dependent transcriptional regulation, are controlled by wdb. However, we did observe that when IIS signalling was reduced, changes in wdb activity could modulate growth. Flies carrying a combination of the weak hypomorphic Akt1 allele Akt104226 (Gao et al., 2000) and the stronger Akt11 allele are only about half the weight of phenotypically normal heterozygous Akt11 flies (Fig. 3) (0.61±0.01 mg versus 1.20±0.05 in females; P<0.001). Akt11/Akt104226 flies that are also heterozygous for wdbIP, a strong, loss-of-function wdb allele (Hannus et al., 2002), show a significant increase in body size (Fig. 3) and weight (0.91±0.07; P<0.001) compared with Akt11/Akt104226 flies, even though heterozygous Akt11 wdbIP flies (1.21±0.07; P>0.5) are not significantly different in weight from heterozygous Akt11 flies.
Loss of wdb blocks neuronal differentiation and induces cell loss in the developing eye via an Akt1-independent mechanism
To further assess the role of wdb in regulating the PI3K/PTEN/Akt-signalling cassette, clones of wdb mutant cells were made in the larval eye imaginal disc, a structure that eventually produces the adult eye. Since the wdb14 null mutation in wdb is recessive cell lethal (Hannus et al., 2002), we generated clones mutant for the slightly weaker loss-of-function wdbIP allele using the FLP/FRT system and an eye-specific source of FLP recombinase in combination with a normal third chromosome. No homozygous wdb ommatidial clones were observed in the adult eye (data not shown). When the wdbIP chromosome was combined with an FRT chromosome carrying a recessive mutation that inhibits cell growth, no normal eye tissue was formed (Fig. 4A,B), indicating that wdb mutant cells are unable to form differentiated adult ommatidia.
Our FOXO interaction data suggested that wdb might suppress IIS upstream of Akt1 or in a parallel pathway. Staining of wdbIP mutant eye clones in the larval eye imaginal disc with an antibody against the Ser505-phosphorylated form of Akt1 (pAkt1), which is typically used to detect Akt1 activation (e.g. Vereshchagina and Wilson, 2006), did not reveal any obvious changes in pAkt1 levels compared with wdbIP/+ heterozygous cells (data not shown). The stereotypic accumulation of actin observed near junctional complexes in photoreceptor cells at the centre of each ommatidium was absent in mutant cells (see phalloidin staining in Fig. 4E,Q). Indeed, these cells failed to differentiate and form ommatidia, because no neuron-specific gene expression was observed in mutant patches (Fig. 4D,P), when discs were stained with antibodies against two neuronal antigens, 24B10 (Zipursky et al., 1984) and 22C10 (Fujita et al., 1982). Clones mutant for both wdbIP and Akt11 also failed to differentiate (Fig. 4L,T), suggesting that the effects of the wdb mutation on eye differentiation are not dependent on Akt1 and must involve misregulation of one or more different signalling pathways.
We noticed that in large wdbIP mutant clones, cells were frequently extruded basally from the eye imaginal disc (Fig. 4I), a phenomenon previously reported when apoptotic cell death events are induced in imaginal discs (e.g. Shen and Dahmann, 2005). Staining of eye imaginal discs that were almost completely mutant for wdb with an antibody against activated caspase 3 revealed elevated levels of cell death throughout these discs (supplementary material Fig. S1). These observations presumably explain the absence of eye tissue in adults with this genotype (Fig. 4B).
Wdb physically interacts with Akt1 in the ovary and controls cytoplasmic levels of activated Akt1 and lipid droplet size in nurse cells
Recent data suggest that specific subcellular pools of activated pAkt1 can be differentially regulated in different fly tissues (Pinal et al., 2006; Vereshchagina and Wilson, 2006). Our FOXO interaction data indicated that wdb might be involved in the control of IIS-dependent transcriptional events. We therefore tested the function of this gene in nurse cells, where cytoplasmic pAkt1 appears to upregulate transcription of the lipid storage protein LSD2 (Vereshchagina and Wilson, 2006; Teixeira et al., 2003), a functional homologue of mammalian perilipin, which promotes storage of lipid in droplets within adipocytes and other cell types (Londos et al., 2005).
Staining of nurse cell clones mutant for two different wdb alleles, wdbIP and wdbdw, with anti-phosphoSer505-Akt1 antiserum revealed elevated levels of cytoplasmic pAkt1 compared with heterozygous wdb/+ cells (Fig. 5B,F). Co-staining with Hoechst 33258, demonstrated that nuclear pAkt1 was not obviously increased (supplementary material Fig. S2).
Cytoplasmic pAkt1 regulates intracellular lipid droplet size in late-stage Drosophila nurse cells (Vereshchagina and Wilson, 2006). These droplets consist of neutral lipids, particularly triacylglycerol and sterol esters (Teixeira et al., 2003). To visualise neutral lipids, ovaries were stained with Nile Red. wdb mutant nurse cells accumulated highly enlarged lipid droplets compared with neighbouring cells (Fig. 5J,P). Co-staining with Nile Red and Hoechst 33258 in Stage 11 egg chambers, shortly before nurse cells normally activate the apoptotic cell death pathway, showed absence of DNA fragmentation in cells that display the lipid droplet phenotype (Fig. 5Q), indicating that these cells are not dying prematurely. Importantly, the droplet phenotype was strongly abrogated in clones mutant for both wdb and Akt1, suggesting that this phenotype is mediated via regulation of Akt1 activity (Fig. 5M). No effect on lipid distribution was observed in wdb mutant clones in the eye (data not shown), which is consistent with our observation that pAkt1 levels are not noticeably affected in this tissue. Increased cytoplasmic pAkt1 and Akt1-dependent accumulation of large lipid droplets are both seen in nurse cells mutant for PTEN (Vereshchagina and Wilson, 2006). These cells also express increased levels of LSD2 protein. We found that LSD2 was similarly upregulated in wdbIP mutant clones (Fig. 5T).
Based on our genetic and immunostaining results, and previous analysis of mammalian PP2A-B′ isoforms (Li et al., 2003; Ugi et al., 2004; Strack et al., 2004; Van Kanegan et al., 2005), we reasoned that Wdb might act directly on Akt1 to regulate its phosphorylation and activity. Indeed, Drosophila Akt1 and Wdb were identified as interacting partners in a previous genome-wide yeast two-hybrid screen (Giot et al., 2003). To test whether this interaction occurs in vivo, antibodies against each protein were used in coimmunoprecipitation experiments with wild-type tissue extracts. Even with large quantities of third instar larval extract (up to 3.5 mg), there was no indication that significant levels of Wdb and Akt1 stably interact. However, these molecules did coimmunoprecipitate from adult ovary extracts (Fig. 6), the tissue in which Wdb and Akt1 display regulatory and genetic interactions (Fig. 5). Wdb was immunoprecipitated by both anti-pan-Akt1 and anti-pAkt1 antisera. The anti-Wdb antibody detects two proteins on western blots, both of which have been shown to be the products of the wdb gene by RNAi knockdown in S2 cells (Sathyanarayanan et al., 2004) (Fig. 6). Interestingly, only the higher molecular mass species interacted with immunoprecipitated Akt1. Since the wdb gene is predicted to encode a single protein product, one of these two species presumably corresponds to a post-translational modification of Wdb, a modification that seems to regulate binding to Akt1. This finding also very clearly demonstrates the high specificity of the coimmunoprecipitation in this experiment.
Wdb also controls cytoplasmic pAkt1 levels in follicle cells
Upon analysis of wdb mutant ovaries, we noticed that mutant follicle cells also showed defects in pAkt1 regulation. Unlike in the eye, large wdb mutant clones were frequently observed within the follicular epithelium, where they contributed to the follicular monolayer surrounding the oocyte. pAkt1 levels were highly elevated in the cytoplasm of cells within these clones (Fig. 7B,F) (100% of wdb mutant follicle cell clones have elevated levels of pAkt1; n>30). Although rare clones appeared to accumulate excess levels of Nile-Red-positive lipid, the vast majority showed no associated lipid phenotype (data not shown). However, LSD2 expression was upregulated in 100% of clones (n>30) (Fig. 7J). Surprisingly, unlike in nurse cells, LSD2 was most strongly overexpressed in the nucleus of follicle cells (Fig. 7J-L), where it would not be predicted to affect lipid droplet formation.
We tested whether the effects of wdb on LSD2 in follicle cells were dependent on Akt1 by generating double-mutant clones. LSD2 levels were reduced relative to wdb clones, but still higher than in wild-type neighbouring cells (Fig. 7N) (interestingly, nuclear LSD2 was nearly completely lost). An incomplete suppression of LSD2 expression was also observed in wdbIP Akt11 double-mutant nurse cells (Fig. 5X), even though the enlarged lipid droplet phenotype caused by loss of wdb function in this cell type is completely suppressed by Akt11 (Fig. 5M). The partial suppression of LSD2 overexpression by Akt11 reflects the possibility that other signalling molecules are also involved in mediating the effects of wdb on LSD2 levels. An alternative explanation, which we favour, is that although the Akt11 allele used in this analysis is the strongest allele currently available, it does still retain some residual activity (Staveley et al., 1998), which could then be hyperactivated by loss of wdb. In strong support of this idea, wdbIP Akt11 double-mutant follicle cells and their nuclei are similar in size to their normal neighbours (Fig. 7O), even though in the absence of a wdb mutation, the Akt11 allele drastically reduces cell size in homozygous clones within multiple tissues, including the follicular epithelium (Cavaliere et al., 2005).
Since PTEN and wdb both play a critical role in controlling cytoplasmic pAkt1 in nurse cells, we tested whether PTEN is also involved in cytoplasmic pAkt1 regulation in follicle cells. A surprisingly variable phenotype was observed. Three of 11 clones showed high levels of pAkt1 (Fig. 7R) and increased LSD2 expression was found in 2 of 12 clones (data not shown). All other clones showed no change in pAkt1 and LSD2 levels (Fig. 7V,Z) (even though levels of pAkt1 and LSD2 were affected in all nurse cell clones from the same preparations). The frequencies of PTEN clones with upregulated pAkt1 or LSD2 are similar. We therefore suspect that these two phenotypes are connected, but could not verify this because both available antibodies were generated in rabbit. Overall we conclude that Wdb has an essential role in controlling cytoplasmic pAkt1 and LSD2 levels in follicle cells, although this does not appear to be important in regulating gross lipid droplet morphology. By contrast, absence of PTEN in follicle cells is generally insufficient to alter levels of pAkt1 and LSD2, suggesting that other negative regulators such as Wdb play a more dominant role in controlling these molecules in this cell type.
Recent studies have indicated that regulation of activated Akt in individual subcellular compartments is critical in controlling specific IIS-dependent functions. However, the mechanisms involved remain largely unexplored. Here, we report the identification of Wdb, a highly conserved PP2A-B′ subunit, as a novel tissue-specific negative regulator of cytoplasmic activated Akt1 in Drosophila. Loss of wdb in ovarian nurse cells and follicle cells increases levels of cytoplasmic pAkt1 and the lipid storage molecule LSD2. In nurse cells this is associated with defects in lipid storage. Wdb therefore appears to represent the first member of an important new group of molecules that modulates specific functions of IIS by regulating signalling in a particular subcellular domain within specific cell types.
Wdb complexes with Akt1 and regulates its PP2A-dependent dephosphorylation
Several lines of evidence confirm that Wdb controls IIS activity and Akt1 phosphorylation state. First, when overexpressed, wdb genetically modifies phenotypes produced by altered IIS signalling, rescuing a lethal PTEN mutant combination and modifying the effects of FOXO in the eye. Second, loss-of-function wdb mutations produce very similar phenotypes to PTEN mutations in nurse cells, elevating levels of cytoplasmic pAkt1 and LSD2, and inducing an abnormal accumulation of lipid droplets. Third, although wdb mutations do not independently appear to have strong effects on growth, they do suppress growth phenotypes produced by reduced Akt1 signalling both in mutant follicle cells homozygous for the Akt11 allele and in animals carrying a hypomorphic viable combination of Akt1 alleles. Genetic interactions with the PP2A catalytic subunit Mts in the eye indicate that these effects are dependent on the PP2A regulatory activity of Wdb.
Coimmunoprecipitation experiments revealed that Akt1 and Wdb form a complex in ovaries, the tissue in which we see the most obvious effects of wdb on pAkt1 levels. Our data suggest that one isoform of Wdb affects IIS within a complex containing Akt1, presumably by directly modulating the phosphorylation state of this molecule. This regulatory interaction appears to be evolutionarily conserved, because several studies in mammalian cell culture have shown that a PP2A-type activity controls Akt phosphorylation at Ser473, the equivalent position to Ser505 in Drosophila Akt1 (Ugi et al., 2004; Strack et al., 2004). PP2A-B′ activity has been implicated in this process (Van Kanegan et al., 2005). Furthermore, mammalian PP2A can dephosphorylate Akt in vitro (Li et al., 2003). The phosphorylation state of Thr308 might also be affected by PP2A (Ugi et al., 2004). However, current tools do not allow us to determine the phosphorylation state of Thr342 (the equivalent position to Thr308 in mammalian Akt) in wdb mutant cells in ovaries. Nevertheless, our study adds to our current understanding of the effects of PP2A on Akt by showing for the first time that at least one PP2A-B′ isoform can act as a pool-specific suppressor of activated Akt. We believe that this property is likely to be shared by some mammalian PP2A-B′ isoforms.
Wdb selectively regulates cytoplasmic pAkt1 in specific tissues
Unlike several other previously characterised components of the IIS cascade (Goberdhan and Wilson, 2003a), the effects of wdb mutations on IIS appear to be tissue specific. Although pAkt1 levels are strongly upregulated in wdb mutant nurse cells and follicle cells, they appear unaffected in clones within the eye. PP2A is a broad-specificity protein phosphatase, which is selectively targeted to specific signalling molecules by regulatory subunits such as Wdb. Wdb has already been shown to be involved in several signalling events, including those regulating apoptosis (Li et al., 2002) and the Hedgehog (Hh) pathway (Nybakken et al., 2005), pathways that might be implicated in the wdb mutant phenotype we observe in the eye imaginal disc (Hsiung and Moses, 2002).
How can Wdb have such a central IIS-regulatory role in the ovary, but show no detectable effect on this pathway in the developing eye? It seems unlikely that wdb mutant cells in the eye die too rapidly to observe changes in Akt1 phosphorylation, because we see wdb clones in posterior positions within eye imaginal discs, which must have formed many hours previously (e.g. Fig. 4C-F). The IIS cascade is active in this tissue, because mutations altering IIS produce significant effects on growth in the eye disc (Böhni et al., 1999; Goberdhan et al., 1999). However, unlike in nurse cells, activation of IIS in the developing eye primarily leads to cell surface accumulation of pAkt1, at least in pupae (Pinal et al., 2006). Surface-localised activated Akt1 may normally be sufficient to promote eye growth, since a myristoylated membrane-anchored form of Akt1 dominantly induces overgrowth in this and other tissues (Stocker et al., 2002). One possible explanation for our data is therefore that cytoplasmic pAkt1 levels in the eye are restricted by other unknown molecules in addition to Wdb in this tissue, so loss of wdb here has little effect, whereas increased expression can still modify the FOXO phenotype.
In this context, at least two other phosphatases might be involved in Akt1 regulation. First, there is a second isoform of PP2A-B′ in flies [called PP2A-B′, CG7913 or Well-rounded (Wrd); (Viquez et al., 2006)] that is most closely related to mammalian PP2A-B′γ and PP2A-B′δ isoforms. Simian virus 40 small t antigen acts as a specific inhibitor of mammalian PP2A-B′γ, stimulating phosphorylation of Akt and other targets, and thereby promoting growth (Janssens et al., 2005). Reduced PP2A-B′γ activity has also been linked to the establishment and progression of melanomas (Deichmann et al., 2001; Ito et al., 2000; Ito et al., 2003a; Ito et al., 2003b).
Surprisingly, a recent report suggests Wrd is nonessential. Unless it acts redundantly with Wdb, it cannot therefore play a significant role in growth regulation (Viquez et al., 2006). Our analysis of the PP2A catalytic subunit Mts, using a dominant-negative construct, indicates that this enzyme enhances the effects of FOXO and is important in normal growth regulation in the eye, perhaps consistent with the idea that the two PP2A-B′ isoforms do act redundantly. Alternatively, Mts may perform some of its growth regulatory functions independently of PP2A-B′ (Bielinski and Mumby, 2007).
A second candidate negative regulator of Akt is the novel phosphatase PHLPP, which directly dephosphorylates human Akt at Ser473 and Drosophila Akt1 at Ser505 in cell culture (Gao et al., 2005), a function that may be disrupted in some tumours. Drosophila PHLPP could therefore control pAkt1 accumulation at the cell surface and perhaps reduce the amount of pAkt1 that can diffuse into the cytoplasm in tissues such as the eye. Since loss of wdb in either follicle cells or nurse cells is sufficient to elevate levels of cytoplasmic pAkt1, PHLPP presumably does not play such an important role in these cell types (microarray data suggest that PHLLP is not expressed at detectable levels in the adult ovary) (Chintapalli et al., 2007).
Interestingly, our data in the ovary suggest further variable tiers of pAkt1 control. In nurse cells, loss of PTEN leads to accumulation of pAkt1 and LSD2 in the cytoplasm, but most PTEN mutant follicle cell clones do not show these phenotypes, presumably because other pAkt1 regulators such as Wdb play a more dominant role in these cells. We do not have a good explanation for how genetically identical clones can show such phenotypic variability. There is no obvious correlation with clone size or position in the small minority of PTEN-mutant follicular clones where pAkt1 and LSD2 upregulation is observed.
Because perilipin, the mammalian LSD2 orthologue, is thought to be regulated via insulin-dependent transcriptional and post-translational mechanisms (Holm, 2003; Prusty et al., 2002; Akimoto et al., 2005), we proposed that the increased LSD2 expression seen in PTEN mutant nurse cell clones results from similar effects of IIS on this molecule in flies (Vereshchagina and Wilson, 2006). An alternative explanation is that increased IIS promotes excess triacylglyceride (TAG) synthesis and that LSD2 is only indirectly upregulated to permit proper packaging of these TAGs into lipid droplets. Our analysis of wdb mutant follicle cell clones does not support this latter model, since these clones strongly upregulate LSD2 expression, but do not show obvious changes in lipid droplet accumulation.
Cytoplasmic pAkt1 might have different effects on IIS-dependent functions when IIS signalling levels are modulated
When wdb is overexpressed in the differentiating eye, the external structure of the eye becomes more disorganised and there is a slight reduction in overall eye size (Fig. 1). Since this effect is not noticeably suppressed by co-overexpressing Akt1, it seems unlikely to be caused by reduced IIS. Unlike PTEN mutant follicle cells (Fig. 7B′), wdb mutant follicle cells are not noticeably larger than their wild-type neighbours (Fig. 7L). Furthermore, although low level constitutive expression of Wdb in a pupal-lethal PTEN mutant background can rescue these flies to viability, the rescue may be explained by altered metabolism, because the rescued flies are still larger than normal. All these observations are consistent with our model that Wdb modulates cytoplasmic pAkt1 and has less of an effect on cell surface pAkt1, which is thought to be the primary regulator of normal growth. Wdb shows a relatively strong genetic interaction with the IIS-regulated transcription factor FOXO and this is completely suppressed by Akt1, raising the possibility that low levels of pAkt1 in the cytoplasm may play an important part in controlling FOXO activity.
Although wdb does not appear to modulate growth significantly under normal IIS-signalling conditions, we have found that mutations in wdb do enhance growth when Akt1 activity is reduced. Viable Akt1 mutant animals are larger in the presence of a heterozygous wdb mutation (Fig. 3), while the Akt11 recessive growth phenotype in follicle cells (Cavaliere et al., 2005) is strongly suppressed by wdb (Fig. 7P). Interestingly, Jünger et al. (Jünger et al., 2003) have reported that mutations in foxo have no effect on growth in otherwise normal animals, but that when IIS is reduced in chico mutants, which produce small adults, this phenotype is partially suppressed by loss of foxo function. Our data are consistent with this result, and may indicate that growth regulation in chico flies relies more on cytoplasmic pAkt1 and its effects on downstream targets like FOXO than it does in normal flies.
In conclusion, the identification of a PP2A-B′ subunit as a novel cell-type-specific regulator of IIS within a specific subcellular compartment highlights the importance of studying the subcellular control of this signalling pathway in multiple cell types in vivo. Akt activation also promotes lipid synthesis and droplet formation in many mammalian cell types (Magun et al., 1996; Schwertfeger et al., 2003; Bouzakri et al., 2006). This is likely to involve similar regulatory control mechanisms for cytoplasmic pAkt to those we have uncovered in flies. Our work therefore raises new issues concerning the underlying causes of IIS-associated disease. For example, excess accumulation of lipid and obesity could be linked to selective changes in cytoplasmic pAkt control and might therefore be modulated by specific PP2A-B′ subunits. Developing a better understanding of this form of regulation could therefore suggest new strategies for disease-specific treatments of IIS-linked disorders in the future.
Materials and Methods
The genotypes of rescued PTEN mutant flies are: w; PTEN3/Df(2L)170B; Akt11/+: w; PTEN3/Df(2L)170B; P[w+, arm-GAL4]/P[w+, UAS-Tsc1], P[w+; UAS-Tsc2]: w; PTEN3/Df(2L)170B; P[w+, arm-GAL4]/wdbGS9548. Screening of 149 fly genes encoding catalytic, inhibitory and regulatory phosphatase subunits, revealed that GS inserts were available upstream of 61. For inserts on the first and third chromosomes, the corresponding GS chromosomes were combined with the PTEN3 chromosome and crossed to w; Df(2L)170B; P[w+, arm-GAL4]/SM5.TM6 flies. For second chromosome inserts, the GS insert was first recombined on to the PTEN3 chromosome before setting up the same cross. The insertion position in the wdbGS9548 line was confirmed by PCR.
Genotypes of flies used to create wdb mutant clones in the eye were: y w eyFLP GMR-lacZ; FRT82B wdbIP/FRT82B RpS3Plac92 for imaginal discs that lack wild-type tissue, and y w eyFLP; FRT82B wdbIP/FRT82B GMR-myrGFP and y w eyFLP; FRT82B wdbIP Akt11/FRT82B GMR-myrGFP for discs that contain wild-type tissue.
PTEN1 mutant clones in the ovary were generated as described in Vereshchagina and Wilson (Vereshchagina and Wilson, 2006). Ovaries containing wdbIP, wdbdw and wdbIP Akt11 mutant clones were also generated using the FLP/FRT system. For example, males hemizygous for a hsp70-flp122 X chromosome insertion, which produces leaky expression of flp even in the absence of heat shock (Britton et al., 2002), and heterozygous for FRT82B wdbIP were crossed to w1118; FRT82B P[w+; Ubi-GFPnlsS65T]/TM6B Hu females, which preferentially express nuclear GFP. These crosses were maintained at 18°C. Newly eclosed yw1118 hsp70-flp122/w1118; FRT82B wdbIP/FRT82B P[w+; Ubi-GFPnlsS65T] females were heat shocked in a water bath for 1 hour at 37.5°C on the day of collection and aged at 25°C for an additional 24 hours before dissection. For nurse cells, only cells where no GFP fluorescence had persisted over the time since the clone was formed were scored as loss-of-function clones (some clones containing one or two cells still expressed some GFP and were not considered in our analysis). To assess animal size, flies were weighed within 2 days of eclosion in at least three batches of five.
Rabbit anti-Drosophila pSer505 Akt1 antibody (Cell Signalling) was used at 1:500 for immunostaining. Rabbit anti-LSD2 antiserum (Welte et al., 2005) and guinea pig anti-Wdb antibody (Sathyanarayanan et al., 2004) were diluted as described. Goat anti-rabbit and donkey anti-guinea pig IgG-HRP-conjugated antibodies were obtained from GE Healthcare and Abnova, respectively. MAb22C10 and 24B10 were from the Developmental Studies Hybridoma Bank.
Coimmunoprecipitation and western blotting
Tissues were homogenised in lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% CHAPS) supplemented with a cocktail of protease inhibitors (Sigma) and left on ice for 30 minutes. Following removal of cell debris by centrifugation at 20,000 g at 4°C for 30 minutes, protein lysates were quantified using the BCA assay (Pierce) and immunoprecipitated overnight at 4°C with 2 μg of the indicated antibodies. Immunocomplexes were bound to protein G-Sepharose 4B fast flow (Sigma) for 2 hours at 4°C, washed extensively in lysis buffer, resolved by SDS-PAGE, and analysed by western blot using the same primary antibodies followed by appropriate HRP-conjugated secondary antibodies, according to standard procedures. Proteins were visualized by ECL detection (GE Healthcare) following the manufacturer's instructions.
We thank Suzanne Eaton, Armen Manoukian, Toshiro Aigaki, Duojia Pan, as well as the Kyoto and Bloomington Stock Centres, for fly stocks. We also are grateful to Steve Gross, Michael Welte, Amita Sehgal and the Developmental Studies Hybridoma Bank (University of Iowa) for providing antibodies, and Hannah Robertson for her technical advice. We thank Kay Davies for providing support for this work, and also acknowledge the financial support of Diabetes UK (ref. BDA:RD02/0002540), the Association for International Cancer Research and the Medical Research Council.