Von Hippel–Lindau tumor suppressor (VHL) stimulates TOR signaling by interacting with phosphoinositide 3-kinase (PI3K)

Cell growth is positively controlled by the phosphoinositide 3-kinase (PI3K)–target of rapamycin (TOR) signaling pathway under conditions of abundant growth factors and nutrients. To discover additional mechanisms that regulate cell growth, here we performed RNAi-based mosaic analyses in the Drosophila fat body, the primary metabolic organ in the fly. Unexpectedly, the knockdown of the Drosophila von Hippel–Lindau (VHL) gene markedly decreased cell size and body size. These cell growth phenotypes induced by VHL loss of function were recovered by activation of TOR signaling in Drosophila. Consistent with the genetic interactions between VHL and the signaling components of PI3K–TOR pathway in Drosophila, we observed that VHL loss of function in mammalian cells causes decreased phosphorylation of ribosomal protein S6 kinase and Akt, which represent the main activities of this pathway. We further demonstrate that VHL activates TOR signaling by directly interacting with the p110 catalytic subunit of PI3K. On the basis of the evolutionarily conserved regulation of PI3K–TOR signaling by VHL observed here, we propose that VHL plays an important role in the regulation and maintenance of proper cell growth in metazoans.

In our effort to identify a novel factor that is critical for the regulation of cell growth, we found that VHL is important for maintaining proper cell size control in the fly. We showed that the cell growth-promoting function of VHL is associated with the signaling components of PI3K-TOR pathway. Both genetic and biochemical analyses consistently suggested that VHL activates PI3K-TOR pathway downstream of InR. Further biochemical and pharmacological analyses supported that VHL activates PI3K-TOR signaling pathway via direct interaction with the p110 subunit of PI3K.

dVHL positively regulates cell size in Drosophila
To isolate a novel regulator for cell growth control, we conducted a genetic screen using ϳ2,000 RNAi lines from the Vienna Drosophila Resource Center (Fig. 1A). In the screen, we conducted flip-out recombination experiments to generate RNAi-based knockdown clones in the fat body to overcome possible developmental lethality caused by the RNAi expression. We compared the cell size of RNAi-expressing clones with that of the adjacent cells in third instar larvae (53). As a result, we obtained an RNAi line, v32163 (henceforth, used for all the subsequent experiments) that caused a striking reduction of cell size in the clones with the RNAi expression, compared with the adjacent control cells (Fig. 1B). Later, we found two additional RNAi lines, v32164 and v108920, eliciting similar phenotypes in the clones (Fig. S1). Remarkably, these RNAi lines all targeted the same gene encoding the Drosophila von Hippel-Lindau (dVHL), indicating that dVHL involves in cell growth control in the fly. To examine whether dVHL is also required for other tissues in regulation of cell size, we depleted dVHL in the posterior region of the wing using engrailed-GAL4 (e16E). Consistent with the fat body data, we again observed that the cell size of dVHL-depleted cells (dVHLi) in the wing posterior (P) region was significantly smaller than that of the anterior (A) cells (Fig. 1C), resulting in a ϳ42% increase of cell numbers in the posterior (P) region (Fig. 1D). Taken together, these data strongly suggested that dVHL positively regulates cell growth.

dVHL activates dAkt-dTOR signaling in Drosophila
Considering the fact that cell growth is primarily shaped by TOR signaling (54), we wondered whether any of the conserved components in the TOR signaling cascade is affected by manipulation of dVHL expression. To examine the hypothesis, we depleted dVHL (dVHLi) in the whole body using da-GAL4 and confirmed that the level of dVHL transcript was decreased to 24% compared with the control (ϩ) in quantitative RT-PCR experiments ( Fig. 2A). In the flies with dVHL knockdown, we first examined the changes in the activity of Drosophila ribosomal protein S6 (dS6) and dAkt by measuring the levels of their phosphorylation. The phosphorylation of dS6 by dS6 kinase occurs at Ser-233 and Ser-235, and that of dAkt by dTORC2 occurs at Thr-505 (55). Remarkably, the levels of the phosphorylation on dAkt and dS6 were significantly decreased by dVHL knockdown (Fig. 2B), indicating that dVHL is required for the activity of dTORC2 and dS6 kinase. Supporting this, clonal depletion of dVHL caused nuclear sequestration of dFOXO (Fig. 2C), a robust downstream effector of dAkt, indicating the inhibition of dAkt by dVHL depletion (56). We further observed that lacZ expression from a d4E-BP promoter was markedly enhanced by dVHL depletion (Fig. 2D), substantiating our notion that dVHL is needed for the activity of dAkt by the fact that d4E-BP is a downstream of dFOXO transcription factor (56,57). These results consistently showed that dVHL is critical for the activation of dAkt-dTOR signaling in Drosophila.

Cell growth control by VHL dVHL regulates cell and body size through dPI3K in Drosophila
To reveal the entry point of dVHL into dPI3K-dAkt signaling, we combined different versions of overexpression and RNAi transgenes for Drosophila insulin receptor (dInR), dPI3K, dPTEN, dPDK1, and dAkt with dVHL-depleted flies (dVHLi) and checked the epistatic relationship of dVHL with the components of dPI3K-dAkt signaling. As a result, the decreased cell size in dVHL-depleted clones was fully rescued by the expression of Dp110, dPTEN RNAi, dPDK1, or myr-dAkt (Fig.  3, A and B). However, the expression of a constitutively active form of dInR (dInR CA) did not rescue the decreased cell size of dVHL-knockdown clones (Fig. 3, A and B).
Intriguingly, the co-expression of dVHL RNAi and Dp110 in the wing using engrailed-GAL4 restored the cell size (Fig. 3C). Furthermore, the increased cell number in the wing by expressing dVHL RNAi was also restored by simultaneous expression of Dp110 (Fig. 3D). These results suggested that dVHL controls cell growth via regulation of dTOR signaling downstream of dInR and in parallel with or upstream of Dp110. Another characteristic trait in the growth control produced by inhibition of dTOR signaling particularly in the fat body is marked reduction of the body size (58). Therefore, we wondered whether RNAi-based depletion of dVHL in the fat body elicits similar phenotypes in the body size. Consistent with the previous report (58), dVHL-depleted flies displayed a dramatic reduction of the body size (Fig. 3, E and F). Strikingly, the reduced body size of dVHLdepleted flies was rescued by overexpression of Dp110 (Fig. 3, E and F). These results supported the conclusion that dVHL poses upstream of dPI3K in the dPI3K-dAkt-dTOR signaling pathway to regulate cell growth and body size in Drosophila.

VHL activates PI3K-Akt-TOR signaling pathway in mammalian cells
To understand the relationship of VHL with the signaling components downstream of PI3K such as mTORC1-S6K and mTORC2-Akt in mammalian cells, we measured the activity of mTORC1 in HEK293E cells transfected with S6K and VHL. In this condition, Thr-389 residue on S6K specific for the modification by mTORC1 would be phosphorylated, if mTORC1 is activated by VHL (59,60). Indeed, we observed that a significantly increased level of the mTORC1-mediated phosphorylation on S6K in the presence of VHL expression ( Fig. 4A) (61), indicating specific activation of mTORC1 upon VHL expression. To examine the relationship of the VHL-mediated regulation of mTORC1 with the insulin-dependent mTORC1 signaling, we treated insulin on the cells expressing VHL. Intriguingly, we observed further increased mTORC1-dependent phosphorylation on S6K by VHL, suggesting that VHL overexpression augments insulin-dependent stimulation of mTORC1 (Fig. 4A). Additionally, we found that VHL expression further increased the insulin-induced phosphorylation of Akt at the Ser-473 residue by mTORC2 (Fig. 4B).
As opposed to the stimulated mTOR signaling by VHL overexpression, we wondered whether depletion of VHL inhibits mTOR signaling induced by insulin. To test this, we measured the phosphorylation of S6 and Akt upon siRNA-mediated knockdown of VHL. When HEK293E cells were transfected with VHL siRNA (ϩ), the phosphorylation levels of S6 (the target of S6K) and Akt (the target of mTORC2) were significantly decreased, compared with the control cells transfected with scramble siRNA (Ϫ) (Fig. 4C). Taken together, these results suggested that VHL activates mTOR pathway by functioning upstream of Akt in mammals.

VHL requires PI3K, TSC2, and mTORC1 to stimulate mTOR pathway in mammalian cells
In parallel with epistatic approaches in Drosophila, to gain a further insight into the molecular mechanism by which VHL activates mTOR signaling, we took a pharmacological approach using inhibitors available for mTOR signaling pathway in mammalian cells. First, we found complete elimination of the VHLmediated S6K phosphorylation with two widely used inhibitors for mTOR, rapamycin and Torin1 (Fig. 5A). S6 phosphorylation was also completely eliminated (Fig. 5A). Subsequently, we monitored the time-course status of the phosphorylation on S6K upon rapamycin treatment. Increased phosphorylation on S6K by VHL could be elicited either by activation of an upstream kinase or inactivation of a phosphatase specific for S6K. However, the elimination rate of phosphorylation on S6K upon rapamycin treatment was not delayed at all both in con-

Cell growth control by VHL
trol and VHL-expressing cells (Fig. S2). These results suggested that the activation of mTOR signaling by VHL depends on the elevated kinase activity of mTORC1 but not on the inactivation by a phosphatase such as PHLPP (62,63).
Considering the fact that mTOR activity is also regulated by serum and amino acid signaling involving TSC2 that inhibits mTORC1 by converting the GTP-bound form of Rheb G-protein into the GDP-bound form (2,(12)(13)(14)(15)(16)(17)(18)(19)(20), we sought to examine the possible involvement of TSC2 in the VHL-dependent regulation of mTOR signaling. To examine the possibility, we employed a constitutively active form of TSC2, TSC2 (3A) that possesses three alanine substitutions on Akt-dependent phosphorylation sites at Ser-939, Ser-981, and Thr-1462 (64) to directly monitor whether any changes occur in the VHL-de-pendent mTORC1 activation by TSC2. When we co-expressed TSC2 (3A) in HEK293E cells transfected with VHL, the stimulatory effects of VHL on mTORC1 were markedly dampened (Fig. 5B). These results indicated that VHL activates mTOR signaling upstream of TSC2.
Also, we used LY294002, a potent inhibitor of PI3K that functions upstream of TSC2. When the cells expressing VHL were treated with LY294002, the VHL-dependent stimulation of mTOR signaling was inhibited (Fig. 5C). However, the VHLmediated enhancement of mTOR signaling remained unaffected by treatment of MAPK inhibitors (Fig. S3).
Because LY294002 can also inhibit mTOR kinase activity (65), we performed similar experiments using other specific inhibitors of p110 catalytic subunit of PI3K. We treated the cells

Cell growth control by VHL
expressing VHL with BYL719, a p110-specific inhibitor, and checked whether any changes occurred in the VHL-mediated stimulation of mTOR signaling. Intriguingly, we observed that BYL719 completely abolished the effect of VHL on mTOR activity (Fig. 5D). Based on these results, we concluded that VHL requires the activity of PI3K, TSC2, and mTORC1 to regulate mTOR signaling in mammalian cells.

VHL activates mTOR signaling pathway independent of HIF1␣
Previous results by others showed that VHL regulates mTOR signaling through the VHL-HIF1␣-REDD1-TSC signaling axis (66,67). However, we observed that VHL could also regulate Akt activity at the upstream of TSC in both mammalian cells and fruit flies. Additionally, we conducted all experiments in normoxic conditions in which HIF1␣ is scarce. Hence, we hypothesized that VHL activates TOR signaling pathway independent of HIF1␣.
In fruit flies, it is possible that VHL regulates the level of REDD1/Scylla that controls TOR signaling (69). However, when VHL was knocked down in the fly, the mRNA expression of REDD1/Scylla was not changed (Fig. 6C). Moreover, the decreased cell size caused by VHL knockdown in the fat body was not rescued by REDD1/Scylla knockdown (Fig. 6, D and E). Through these findings, we confirmed that a novel signaling mechanism that activates mTOR signaling by VHL exists, independent from HIF1␣.

VHL activates PI3K and mTOR signaling via binding to p110
Having shown that VHL controls PI3K-mTOR signaling axis, we thus wondered whether VHL affects PI3K activity. Interestingly, VHL is also known to function as an adaptor protein that interacts with various molecules to mediate diverse physiological roles (40). Based on this, we hypothesized that VHL physically interacts with one of the PI3K subunits, p110 and p85, to regulate mTOR signaling. We found that VHL was successfully co-immunoprecipitated with both exogenous (Fig.  7, A and B) and endogenous p110 proteins (Fig. 7C). Intriguingly, however, we could not detect any direct interaction between p85 and VHL. Additionally, we used bimolecular fluorescence complementation (Bi-FC) assays to monitor the interaction between VHL and p110 in the cell. Assembly of the two fusion proteins for Venus N-and C-terminal fragments with VHL and p110 generated a green signal in the experiment, demonstrating a specific interaction between VHL and p110 (Fig.  7D). However, VHL could not interact with p85 in the same experimental condition (Fig. 7D). These results strongly suggested that VHL and p110 directly interact to mediate the regulation of mTOR signaling.
To further map the critical domain of p110 for interaction with VHL, we generated five different GST-fusion proteins for p110 (Fig. 7E) (22), comprised of the p85-binding domain (p85-BD), RAS-binding domain, C2 domain, helical domain, and catalytic domain (70). Using these fusion proteins, we performed a series of co-immunoprecipitation experiments and found that VHL co-immunoprecipitates with the p85-BD GST-fusion protein (fragment 1) (Fig. 7F). Based on this observation, we hypothesized that exclusive overexpression of p85-BD in the cell would inhibit the interaction between endogenous p110 and VHL and thus decrease the VHL-dependent activation of mTOR signaling. Indeed, the expression of the p110 p85-BD domain suppressed the VHL-dependent increase of S6K phosphorylation (Fig. 7G).
Finally, using a GFP-fused PH domain of Akt (GFP-PH-Akt), we measured the activity of PI3K in HEK293E cells. GFP-PH-Akt

Cell growth control by VHL
can localize to the plasma membrane (PM) in response to phosphatidylinositol 1,4,5-trisphosphate formation, in other words PI3K activation (71). Through this experiment, we observed that the GFP signals from GFP-PH-Akt were highly enriched in the PM by insulin treatment (Fig. 7, H and I, and Fig. S6). However, the localization of GFP-PH-Akt in the PM induced by insulin treatment was significantly suppressed by VHL knockdown (Fig. 7, H and I, and Fig. S6). Together, these results showed that VHL specifically interacts with p110 and positively regulates PI3K activity and mTOR signaling.

Discussion
In the current study, we have proposed that VHL is involved in cell growth control via its interaction with PI3K-TOR signaling axis in Drosophila and mammalian cells. To support the idea, we have provided lines of evidence: 1) depletion of dVHL produces marked reduction of cell growth and body size in fruit flies, resembling the phenotypes generated by inhibition of TOR signaling in Drosophila, 2) VHL activates the conserved components of TOR signaling cascades such as S6K and Akt, 3) depletion of VHL strongly suppresses TOR signaling represented by decreased S6 and Akt phosphorylation, and 4) VHL physically binds to p110, the catalytic subunit of PI3K, and elevates its activity to stimulate TOR signaling. Collectively, our data established a novel signaling mechanism where PI3K serves as a target point for VHL to trigger TOR signaling cascades for regulation of cell growth.

HIF1␣-independent functions of VHL to induce PI3K and mTOR signaling
There were reports by others that VHL regulates mTOR signaling through the VHL-HIF␣-REDD1 signaling axis (66,67). VHL polyubiquitinates HIF␣, and the polyubiquitinated HIF␣ is degraded by proteasome under normoxic conditions (30). Whole-cell lysates were used for detecting expression of VHL using anti-VHL antibody and TSC2 using anti-FLAG antibody (*, endogenous VHL). Immunoprecipitated samples were detected by anti-p-S6K (pT389) and anti-Myc antibodies. C, HEK293E cells were transfected with S6K-Myc and FLAG-VHL and were changed for fresh medium and treated with DMSO (6 h) or LY294002 (10 M for 6 h) as indicated. The cell lysates were immunoprecipitated by anti-Myc antibody. Whole-cell lysates were used for detecting expression of VHL and tubulin using anti-VHL and anti-tubulin antibodies (*, endogenous VHL). Immunoprecipitated samples were detected with anti-p-S6K (Thr(P)-389) and anti-Myc antibodies. D, HEK293E cells were transfected with S6K-Myc and FLAG-VHL and were changed for fresh medium and treated with DMSO or BYL719 (100 M for 6 h) as indicated. The cell lysates were immunoprecipitated by anti-Myc antibody. Whole-cell lysates were used for detecting expression of VHL, p-Akt, Akt, and tubulin using anti-VHL, anti-p-Akt (Ser(P)-473), anti-Akt, and anti-tubulin antibodies (*, endogenous VHL). Immunoprecipitated samples were detected by anti-p-S6K (Thr(P)-389) and anti-Myc antibodies.
However, we here found that a mutant form of VHL lacking the binding capability to HIF␣ still induces phosphorylation and activation of S6K (Fig. 6A). In addition, we found that VHL also activates TOR signaling in HIF␣-null cells (Fig. 6B) or in normoxic conditions in which HIF␣ is highly unstable. These results suggest that VHL can control TOR signaling in more than two different mechanisms, such as HIF␣-dependent transcriptional mechanism (by inducing gene expression of REDD1) and protein-protein interaction-dependent posttranslational mechanism (by direct interaction with p110).

Cell growth-promoting property of VHL and its implications on VHL disease
The cell growth-promoting activity of VHL found in this study seems paradoxical because VHL is well-known for its tumor suppressor activity (30). In contrast to our findings, in VHL disease, the activity of mTOR signaling and cell growth are measured to be higher than normal cells, suggesting that additional activation mechanisms for mTOR signaling occur inde-pendent of VHL loss of function in the disease. In the same line with our deduction, VHL-deficient cells have their mTOR signaling rescued by secondary mutations in PTEN or TSC1 (72). In addition, clear-cell renal cell carcinoma patients showed activation of mTOR signaling through epigenetic changes or mutations in the components of mTOR signaling (74, 75).
Therefore, activation of mTOR signaling needs to be carefully examined in the disease induced by VHL loss of function. According to our data, RCC4, a VHL-deleted cell line, becomes more sensitive to serum as a result of VHL overexpression (both WT and Y98N), as does HEK293E cell line (Fig. S4). In consequence, re-expression of VHL may induce aggressive growth of cancer cells by further stimulation of mTOR pathway. Thus, with regard to the treatment of diseases caused by the loss of VHL function, treatment should not target VHL itself but rather should focus on abnormally rewired signaling pathways caused by VHL deficiency, such as abnormally activated mTOR signaling. In summary, the growth-promoting function of VHL through controlling PI3K will provide a new way to understand the underlying molecular mechanisms of VHL disease.

Experimental procedures
Fly stocks

Clone generation and immunostaining
To generate transgene expressing clones in the fat body, the FRT/FLP-mediated flip-out technique was used with Act Ͼ CD2 Ͼ GAL4 and ywhsflp 122 . Embryos were collected for 12 h and incubated at 25°C for 5 days (76,77). Third instar larvae Figure 7. VHL activates mTOR signaling by binding to p110. A, HEK293T cells were transfected with HA-p110 and VHL as indicated. The cell lysates were immunoprecipitated (IP) by anti-VHL antibody. Whole-cell lysates (WCL) were used for detecting expression of p110 and VHL using anti-HA and anti-VHL antibodies. Immunoprecipitated samples were also immunoblotted by anti-HA and anti-VHL antibodies. B, HEK293T cells were transfected with HA-p110 and VHL as indicated. The cell lysates were immunoprecipitated by anti-HA antibody. Whole-cell lysates were used for detecting expression of p110 and VHL using anti-HA and anti-VHL antibodies (*, endogenous VHL). Immunoprecipitated samples were detected with anti-HA and anti-VHL antibodies. C, HEK293T cells were changed for fresh medium for 10 min before cell lysis. HEK293T cell lysates were immunoprecipitated by anti-IgG or anti-VHL antibodies. Whole-cell lysates were used for detecting p110 and VHL using anti-p110 and anti-VHL antibodies. Immunoprecipitated samples were detected for p110 and VHL using anti-p110 and anti-VHL antibodies. D, fluorescent confocal microscopy images of HEK293E cells. HEK293E cells were transfected with FLAG-VHL-Venus N-terminal fragment (Vn), HA-p110 -Venus C-terminal fragment (Vc), and HA-p85-Vc as indicated. The FLAG-tagged VHL proteins were immunolabeled with anti-FLAG antibody (yellow), and the HA-tagged proteins (p110 and p85) were labeled with anti-HA antibody (red). Assembled Venus proteins between Vn and Vc showed a green signal. Hoechst (blue) was used for nuclei staining. Scale bars, 20 m. E, a schematic representation of the domain structure of p110. We made GST-fusion proteins for each p110 domain. F, HEK293T cells were co-transfected with VHL and GST-p110 fragments as indicated. The cell lysates were subjected to GST pulldown. Whole-cell lysates were used for detecting expression of VHL and p110 fragments using anti-VHL and anti-GST antibodies (*, endogenous VHL). Immunoprecipitated samples were detected for VHL and p110 domain fragments using anti-VHL and anti-GST antibodies. G, HEK293E cells were transfected with S6K-Myc, FLAG-VHL, and GST-p110 fragment 1 as indicated and were changed for fresh medium for 1 h. The cell lysates were immunoprecipitated by anti-Myc antibody. Whole-cell lysates were used for detecting expression of VHL and GST using anti-VHL and anti-GST antibodies, respectively. Immunoprecipitated samples were detected by anti-p-S6K (pT389) and anti-Myc antibodies. H, PI3K activity and the localization of GFP-PH-Akt in the PM. HEK293E cells were transfected with GFP-PH-Akt and siRNA (scramble siRNA (siScr) or siVHL) as indicated. The cells were preincubated in serum-starved medium for 2 h. During insulin treatment (2 M for 5 min), GFP-PH-Akt signals were traced by a confocal microscope under 800ϫ magnification. Scale bars, 10 m. I, mean PM/cytosolic GFP ratio changes after insulin treatment for 5 min in H. The intensity of the GFP signals in cytosolic and PM regions was quantitated by ZEN software, and the ratios of intensity (PM/cytosolic) are presented. In each condition, the ratio before insulin treatment (0 min) was normalized to 1.0 (n ϭ 9). Error bars, S.D. ***, p Ͻ 0.001.

Cell growth control by VHL
were turned inside out in PBS and fixed with 4% paraformaldehyde for 20 min. After several washes with PBS-containing 0.1% Triton X-100 (PBST), the samples were permeabilized in PBS with 0.5% Triton X-100 for 5 min and then washed twice with PBST. The samples were incubated in blocking solution with primary antibody at 4°C overnight. The blocking solution contained PBST with 3% BSA. The samples were washed three times with PBST and then incubated in blocking solution with secondary antibody and Hoechst 33258 for 2 h. After three washes in PBST, the fat body tissues were isolated and mounted in SlowFade (Life Technologies, S36936) on slide glasses. The prepared samples were observed with an LSM 710 confocal microscope (Zeiss). Anti-dFOXO rabbit antibody, a gift from Dr. Kweon Yu (Korea Research Institute of Bioscience and Biotechnology, Daejeon, Korea), and anti-␤-gal mouse antibody (JIE7, DSHB, IW) were used as primary antibodies. To compare cell size of clones with adjacent control cells without immunostaining, the fat body tissues were dissected, fixed, and stained with Hoechst.

Cell culture and transfection
HEK293E, HEK293T, Hif1␣ MEF (ϩ/ϩ, Ϫ/Ϫ), and RCC4 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific) at 37°C in a humidified atmosphere with 5% CO 2 . Transfection of mammalian expression plasmid was performed using polyethylenimine (Sigma) according to the manufacturer's instructions. For RCC4 cell line transfection, plasmid transfection was performed by Lipofectamine Plus reagent (Invitrogen) according to the manufacturer's instructions. siVHL (human and mouse) was purchased from Bioneer in Korea. siRNA was transfected using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. Co-transfection of DNA and siRNAs was performed by Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.

Preparation of lysate, immunoprecipitation, and immunoblotting
For preparation of cell lysate, the cells were washed with cold PBS and lysed with buffer A (20 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA, 2 mM EGTA, 50 mM ␤-glycerophosphate, 50 mM NaF, 2 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 1 g/ml pepstatin A, and 1% Triton X-100) and subjected to immunoprecipitation and immunoblotting according to the standard procedures. The blots were developed and visualized using LAS-4000 (Fujifilm, Japan). For immunoblotting of the lysates from larvae, 15 third instar larvae were collected and homogenized in ice-cold buffer A using a homogenizer.

Confocal microscopy of live samples
Round cover slides were coated with poly-L-lysine and seeded with HEK293E cells. HEK293E cells were transfected with GFP-fused PH domain fragment of Akt (GFP-PH-Akt) and siRNAs by Lipofectamine 2000. After 2 days, the cells were serum-starved for 2 h and traced for GFP signals. The slides were loaded onto a cell perfusion chamber, and the GFP-expressing cells were identified and visualized with a confocal microscope. After insulin treatment (2 M for 5 min), laser images were taken every 5 s with an excitation track of 488 nm.
Statistics p values were obtained by Student's t test or one-way analysis of variance Tukey's test, and Ͻ0.05 was considered to be significant.