Akt Activation Promotes Degradation of Tuberin and FOXO3a via the Proteasome*

Growth factor receptors promote cell growth and survival by stimulating the activities of phosphatidylinositol 3-kinase and Akt/PKB. Here we report that Akt activation causes proteasomal degradation of substrates that control cell growth and survival. Expression of activated Akt triggered proteasome-dependent declines in the protein levels of the Akt substrates tuberin, FOXO1, and FOXO3a. The addition of proteasome inhibitors stabilized the phosphorylated forms of multiple Akt substrates, including tuberin and FOXO proteins. Activation of Akt triggered the ubiquitination of several proteins containing phosphorylated Akt substrate motifs. Together the data indicate that activated Akt stimulates proteasomal degradation of its substrates and suggest that Akt-dependent cell growth and survival are induced through the degradation of negative regulators of these processes.

Growth factor receptors promote cell growth and survival by stimulating the activities of phosphatidylinositol 3-kinase and Akt/PKB. Here we report that Akt activation causes proteasomal degradation of substrates that control cell growth and survival. Expression of activated Akt triggered proteasome-dependent declines in the protein levels of the Akt substrates tuberin, FOXO1, and FOXO3a. The addition of proteasome inhibitors stabilized the phosphorylated forms of multiple Akt substrates, including tuberin and FOXO proteins. Activation of Akt triggered the ubiquitination of several proteins containing phosphorylated Akt substrate motifs. Together the data indicate that activated Akt stimulates proteasomal degradation of its substrates and suggest that Akt-dependent cell growth and survival are induced through the degradation of negative regulators of these processes.
Metazoan cell size, survival, and proliferation are coordinately regulated by the availability of extrinsic growth factors that govern these processes through their control of intracellular signal transduction cascades. The signals emanating from growth factor receptors determine the size, number, and turnover rate of cells within a given tissue (1). A major pathway that controls cell growth and survival in response to the activation of growth factor receptors is the phosphatidylinositol 3-kinase/Akt (PI3K 1 /Akt) pathway (2). Studies of the PI3K/Akt pathway in model organisms such as Drosophila melanogaster and Caenorhabditis elegans have established the importance of this pathway in controlling the number and size of cells within an organism and also in controlling the size and longevity of the entire organism. In vertebrates, constitutively active oncogenic forms of both PI3K and Akt have been characterized in murine and avian retroviruses, and inactivating mutations in PTEN (a phosphatase that opposes the action of PI3K) are prevalent in many types of human tumors (3)(4)(5).
Activation of the PI3K/Akt pathway stimulates growth and survival via its actions on a number of Akt substrates, including the FOXO family transcription factors Bad, GSK-3, mTOR, and tuberin. Active FOXO family transcription factors promote the transcription of genes involved in cell cycle arrest and apoptosis such as p27 kip1 and Bim (6 -8). One mechanism by which Akt promotes cell survival is by phosphorylating FOXO family transcription factors, which inactivates them and prevents the transcription of proapoptotic molecules. Recently Akt has also been shown to phosphorylate and inactivate a negative regulator of cell size, tuberin (9 -12). In conjunction with its binding partner hamartin, increased tuberin protein levels negatively regulate cell size, possibly by inhibiting the phosphorylation and activation of mTOR or p70 S6K . Phosphorylation by Akt prevents tuberin-dependent declines in cell size.
In growth factor-dependent cells, the removal of growth factors triggers cellular atrophy and the initiation of programmed cell death (13). Activation of Akt inhibits both of these processes, suggesting that Akt can inhibit mediators of both cell atrophy and cell death (14). In analyzing the effects of Akt activation on downstream targets involved in the regulation of cell size and survival, we show here that the activation of Akt results in a decline in total protein levels of tuberin and FOXO3a. The Akt-stimulated decline in tuberin correlated with a decline in the protein level of its binding partner, hamartin, suggesting that the tuberin-hamartin complex is coordinately regulated by Akt. Inhibition of the proteasome resulted in a recovery in the protein levels of tuberin, FOXO1, and FOXO3a and triggered a general accumulation of Akt substrates. Physiologic activation of the endogenous PI3K/Akt pathway through the IL-3 receptor in the presence of proteasome inhibitors resulted in prolonged phosphorylation of tuberin and FOXO3a. IL-3 stimulation triggered the ubiquitination of a number of potential Akt substrates, indicating that Akt-dependent degradation of substrates may be induced through their ubiquitination. Together the data indicate that activation of Akt targets the substrates of Akt for degradation in the proteasome.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-IL-3-dependent FL5.12 cells were cultured in 350 pg/ml recombinant murine IL-3 (R&D Systems) as described previously (15). Vector control and myristoylated Akt (myrAkt)-expressing cells were transduced with either empty pRevTRE retroviral vector (BD Clontech) or myrAkt pRevTRE, as described previously (14). Bcl-x L was expressed in cells with pSFFV-Bcl-x L , also described previously (16). FOXO1 and FOXO1-3A, generously provided by Drs. Fred Barr (University of Pennsylvania) and Gabriel Nunez (University of Michigan) were cloned into pRevTRE and transduced into an FL5.12 line that carries the rtTA transcriptional activator of tetracycline-regulated promoters (17). Clones were selected by immunoblot for the ability to induce FOXO1 and FOXO1-3A protein expression to comparable levels. For regular passage, all cell lines were cultured in the absence of doxycycline. For induction of myrAkt, doxycycline was added to the medium (1 g/ml, BD Clontech) for 24 h followed by an additional 48 h of culture in doxycycline-supplemented medium with or without IL-3 as appropriate. Bcl-x L -expressing and vector control lines were treated similarly to control for nonspecific effects of doxycycline unless otherwise indicated. Induction of FOXO1 constructs was for 24 h before lysis. Measurement of cellular viability was performed by flow cytometric analysis of cells stained in phosphatebuffered saline containing 2 g/ml propidium iodide.
IL-3 Stimulation-FL5.12 cells were withdrawn from IL-3 for 4 h, pelleted, and resuspended in medium lacking IL-3 at a density of 1 ϫ 10 7 /ml. Vehicle control Me 2 SO or 100 nM final concentration epoxomicin was added, and cells were preincubated at 37°C for 30 min. Recombinant murine IL-3 was added to a final concentration of 4 ng/ml, and each sample was incubated at 37°C for the indicated time. After each incubation, cells were pelleted for 30 s and directly lysed in ice-cold radioimmune precipitation buffer as described above. Cell lysates were kept on ice until the end of the experiment, and equal quantities of protein were resolved by gel electrophoresis without freezing.
Cell Size Measurements-Cell size was assessed in live cells incubated with 10 g/ml cell-permeant DNA dye, Hoechst 33342, for 30 min at 37°C and 2 g/ml propidium iodide. After analysis by flow cytometry (LSR instrument, BD Biosciences), the forward scatter of G 1 -phase live cells was plotted for vector control, Bcl-x L -expressing, and myrAktexpressing cells.
Cell Lysis and Immunoblots-Cells were washed once in phosphatebuffered saline before lysis in ice-cold radioimmune precipitation buffer containing protease inhibitors (Complete Mini, Roche Molecular Biochemicals) and phosphatase inhibitors (10 mM sodium fluoride, 250 M sodium orthovanadate, and 20 mM sodium ␤-glycerophosphate; Sigma). Lysates were sonicated for 5 s, and insoluble matter was removed by centrifugation. Cell extracts were standardized for protein content using the BCA protein assay (Pierce), and 25 g of total cell extract was loaded per lane. Proteins were resolved on Tris acetate 3-8% gradient gels (Invitrogen) and transferred to nitrocellulose for immunoblotting. Anti-tuberin C20, anti-FOXO3a H144, anti-human FOXO1 H128, antiactin I18, and anti-ubiquitin P4D1 were obtained from Santa Cruz; anti-hamartin 2-8a was a kind gift of Dr. Jeff E. DeClue (National Cancer Institute, Bethesda, MD); anti-Akt, anti-Akt pS473, anti-mu-rine FOXO1, anti-Akt substrates, anti-FOXO3a pT26, and anti-tuberin pT1462 were obtained from Cell Signaling Technology. Quantitation of immunoblot data was performed using ImageQuant (Amersham Biosciences).
Proteasome Inhibitors and Analysis of Ubiquitin Conjugates-ALLN and ALLM (Calbiochem) dissolved in Me 2 SO were added to final concentrations of 50 M; epoxomicin (Calbiochem) dissolved in Me 2 SO was added to a final concentration of 100 nM. Vehicle control Me 2 SO was added to cells that were not treated with proteasome inhibitors. Cells that had been incubated in the presence of proteasome inhibitors were lysed in ice-cold NET-N (10 mM Tris, pH 8.0, 5 mM EDTA, 150 mM NaCl, and 1% Nonidet P-40) containing protease inhibitors and phosphatase inhibitors (10 mM sodium fluoride, 250 M sodium orthovanadate, and 20 mM sodium ␤-glycerophosphate; Sigma). N-ethylmaleimide (freshly prepared as a 1 M stock in 100% EtOH) was added immediately after lysis to a final concentration of 10 mM. Lysates were then sonicated for 5 s, and insoluble matter was removed by centrifugation. Lysates were standardized for protein content, and 500 g of total protein were tumbled with 40 l of agarose beads bound to the UBA domains from Rad23 (Ubiquitinated Protein Enrichment Kit, Calbiochem) for 2-4 h at 4°C. After the beads were washed three times in lysis buffer, PAGE sample buffer was added, and the beads were incubated for 5 min at 95°C. Bound proteins were resolved by gel electrophoresis and probed by immunoblot.

RESULTS AND DISCUSSION
Stable transfectants expressing constitutively active myrAkt under the control of a tetracycline-responsive promoter were established using the nontransformed IL-3-dependent murine hematopoietic cell line FL5.12 with the intent of identifying and purifying Akt substrates involved in the control of growth and survival (Fig. 1, A and B). A low level of myrAkt is expressed in the absence of doxycycline. However, the addition of doxycycline to myrAkt-expressing cells results in a rapid and prolonged increase in Akt activity with concomitant increases in cell size and the induction of growth factor-independent cell survival (Fig. 1C). Disappointingly, using antibodies specific for Akt phosphorylation sites, doxycycline induction of Akt failed to induce an increase in the number or intensity of immunoreactive bands on immunoblots. This result prompted an examination of the expression of specific Akt substrates. The Akt consensus phosphorylation sequence identified tuberin as a potential Akt substrate, which has been confirmed recently in several studies (9,11,18,19). Analysis of tuberin expression revealed a decline in tuberin protein levels as Akt activity was induced (Fig. 1D). A decline in the protein levels of the tuberin binding partner hamartin was also observed in myrAkt cells, and this decline was consistently accompanied by the appearance of a faster migrating species in hamartin immunoblots, suggesting the generation of a hamartin degradation product in cells with activated Akt.
In IL-3-dependent cells, growth factor withdrawal rapidly induces a decline in cell size followed by the induction of programmed cell death. Expression of Bcl-x L prevents the induction of cell death without altering the kinetics of cell atrophy, permitting analysis of cell atrophy in the absence of complicating effects of apoptosis (13). of IL-3 signal transduction still led to a near complete decline in both hamartin and tuberin levels (Fig. 1E). The declines in tuberin and hamartin levels in cells with activated Akt corresponded with increased cell size both in the presence and absence of IL-3 (Fig. 1, F and G).
Phosphorylation can target proteins for degradation by the proteasome (24). To determine whether proteasomal degradation was involved in the decline in tuberin protein levels in myrAkt-expressing cells, the proteasome inhibitor ALLN was added to cultures. In vector control and Bcl-x L -expressing cells, the addition of ALLN had little effect on tuberin protein levels. However, ALLN triggered a substantial recovery of tuberin protein in myrAkt-expressing cells, suggesting that proteasomal degradation of tuberin was triggered by activated Akt (Fig.  2A). In addition to inhibiting proteasomal degradation, ALLN inhibits the activities of other proteases within the cell. To verify the involvement of the proteasome in the Akt-dependent decline in tuberin levels, cells were treated with ALLM, an analogue of ALLN that shares the ability to inhibit proteases but has poor affinity for the proteasome (25). In contrast to ALLN, ALLM treatment did not result in the recovery of tuberin protein levels, indicating that the ability to inhibit the proteasome is required to prevent degradation of tuberin (Fig.  2B). Further evidence supporting a role for the proteasome in Akt-dependent declines in tuberin levels was obtained by treating cells with epoxomicin, an irreversible inhibitor of the proteasome with a mechanism of action that is distinct from ALLN (26). Similar to ALLN, the addition of epoxomicin to Akt-expressing cells resulted in a recovery in tuberin protein levels (Fig. 2B). Thus, Akt stimulates proteasomal degradation of tuberin.
Akt can phosphorylate tuberin in vitro and in vivo at a number of sites, suggesting that Akt might trigger tuberin degradation by phosphorylation (9,11). To determine whether other Akt substrates are also targeted for degradation by the proteasome, vector control, Bcl-x L -expressing, and myrAkt-expressing cells were treated with ALLN and probed in an immunoblot with an antiserum specific for phosphorylated residues in the context of the consensus Akt substrate motif (9). Surprisingly, the number of immunoreactive protein species was lower in Akt-expressing cells than in control or Bcl-x Lexpressing cells. However, a number of bands containing potential Akt phosphorylation sites were induced by treatment with ALLN in all three lines, and the magnitude of induction was greatest in cells expressing activated Akt (Fig. 3A). The accumulation of Akt substrates in control cells cultured in ALLN and IL-3 suggests that the IL-3 receptor activation of the endogenous PI3K/Akt pathway is sufficient to cause degradation of Akt substrates. These data suggest that Akt may target a number of substrates for proteasomal degradation, and the extent of proteasomal degradation correlates with the extent of Akt activation.
Because ALLN promoted the accumulation of potential Akt substrates in the 50 -90-kDa range, the effects of proteasome inhibitors on protein levels of FOXO family transcription factors were evaluated as a test of this hypothesis. Expression of activated Akt triggered a decline in protein levels of both FOXO1 and FOXO3a, and treatment of cells with ALLN triggered a recovery in protein levels for both transcription factors (Fig. 3B). In lysates from vector control cells the addition of ALLN triggered a significant increase in FOXO3a protein levels, specifically in a slower migrating form of the protein. This result suggested that IL-3 receptor stimulation was sufficient to trigger endogenous Akt phosphorylation and degradation of FOXO3a. To test this possibility, vector control and myrAkt cells were cultured in the presence and absence of ALLN and the PI3K inhibitor LY294002. As shown previously, culture in the presence of ALLN caused an accumulation of a slow migrating form of FOXO3a. The addition of LY294002 prevented IL-3-induced phosphorylation of FOXO3a and caused an increase in the mobility of total FOXO3a protein (Fig. 3C). The addition of LY294002 had little effect on the phosphorylation or mobility of FOXO3a in cells expressing activated Akt. These data indicate that the effects of LY294002 on FOXO3a protein mobility in response to IL-3 depend on inactivation of the PI3K/Akt pathway. Taken together, the results suggest that protein levels of both FOXO1 and FOXO3a are regulated in response to the activation of Akt in a proteasome-dependent manner.
The effects on FOXO3a protein mobility suggested that the activation of endogenous Akt by the IL-3 receptor can influence protein stability. To further explore whether proteasomal degradation of Akt substrates is induced at physiologic levels of Akt activation, we examined the effects of IL-3-induced Akt activation on the expression and phosphorylation status of tuberin and FOXO3a after the addition of IL-3 to growth factordeprived cells. IL-3 was added to cells that had been pretreated for 30 min with vehicle control or epoxomicin. We have previously demonstrated under these conditions that IL-3 causes a rapid induction of PI3K-dependent Akt activity that is detectable within 5 min and persists for more than an hour (14). Total cell lysates were isolated at several time points and probed in immunoblots with antisera that recognize Akt phosphorylation sites within tuberin and FOXO3a. IL-3-stimulated phosphorylation of tuberin and FOXO3a was maximal at 10 min followed by a decay in phosphorylation and protein levels over time (Fig.  4A). Treatment of cells with epoxomicin resulted in the accumulation of phosphorylated forms of both tuberin and FOXO3a at later time points, suggesting that phosphorylated tuberin and FOXO3a are targeted for proteasomal degradation after IL-3 withdrawal (Fig. 4A). At later time points, IL-3 stimulation caused a mild decline in tuberin protein levels that was abrogated by the addition of epoxomicin (compare Fig. 4A,  lanes 9 and 10, with lanes 1 and 2). As seen in ALLN-treated cells, FOXO3a was characterized by heterogeneous mobility at early time points after IL-3 stimulation, and this persisted in epoxomicin-treated cells (Figs. 3, B and C, and 4A).
To confirm that direct Akt phosphorylation is required for Akt-induced protein degradation, we experimentally introduced tetracycline-inducible constructs of either wild type human FOXO1 or a human FOXO1 in which Akt phosphorylation sites had been mutated to alanine (FOXO1-3A) into FL5.12 cells (17). These cells were then infected with a retrovirus encoding GFP alone or both GFP and myrAkt. GFP-positive cells were isolated by cell sorting and expanded in the presence or absence of doxycycline. In cells expressing wild type FOXO1, expression of myrAkt led to increased levels of phosphorylated Akt and a concomitant decline in the level of FOXO1 protein. In cells expressing FOXO1-3A in which major Akt phosphorylation sites have been mutated, the effect of myrAkt on the proteins levels of the FOXO1-3A protein was blunted compared with the effect of Akt on wild type FOXO1 (Fig. 4B). These data indicate that Akt-induced declines in FOXO protein levels require the presence of Akt phosphorylation sites in the substrate.
The 26 S proteasome degrades proteins that are conjugated to polyubiquitin chains. Akt-induced degradation of substrates in the proteasome may be stimulated by an increase in the ubiquitination of these proteins. To determine whether tuberin ubiquitination is increased in cells expressing activated Akt, ubiquitinated proteins were enriched using an affinity matrix consisting of glutathione S-transferase fused to the ubiquitin-binding UBA domains of Rad23 (27,28). Tuberin was enriched among ubiquitinated proteins isolated from epoxomicin-treated myrAkt cells compared with vector control cells (cultured in the absence of IL-3), indicating that increased Akt activity is sufficient to cause an increase in tuberin ubiquitination (Fig. 5A).
To determine whether activation of the endogenous PI3K/ Akt pathway by the IL-3 receptor can cause a generalized increase in the ubiquitination of Akt substrates, vector control and myrAkt cells were stimulated with IL-3 in the presence and absence of epoxomicin and LY294002 for 60 min, a time point at which degradation of tuberin and FOXO3a was observed previously (Fig. 4A). Akt substrates were detected using the Akt substrate antiserum in fractions enriched for ubiquitinated proteins using the Rad23 UBA domains (Fig. 5B). In vector control cells, ubiquitinated Akt substrates were detected in lysates from epoxomicin-treated cells but not from cells treated with vehicle control, suggesting that ubiquitinated Akt substrates are processed through the proteasome on stimulation of the endogenous PI3K/Akt pathway with IL-3. The addition of LY294002, a PI3K inhibitor that blocks the activation of Akt by the IL-3 receptor, prevented the accumulation of ubiquitinated Akt substrates in epoxomicin-treated cells, suggesting that activation of Akt was required for ubiquitination of these proteins. Expression of activated Akt-enriched ubiquiti- FIG. 5. Akt induces the ubiquitination of tuberin and other putative substrates. A, vector control (Vec.) and myrAkt (Akt) cells were removed from IL-3 for 1 h and incubated in the presence and absence of epoxomicin for an additional hour. Ubiquitinated proteins were isolated from total cell lysates using the Rad23 UBA domains bound to agarose. After three washes, bound proteins were resolved by gel electrophoresis and probed in an anti-tuberin immunoblot. The blot was stripped and reprobed in an anti-ubiquitin immunoblot to control for efficiency of enrichment of ubiquitinated proteins. B, vector control (V) and myrAkt (A) cells were removed from IL-3 for 2 h followed by a 30-min incubation in vehicle control (Me 2 SO), 100 nM epoxomicin (Epox.), or epoxomicin and 10 M LY294002 (LY ϩ Epox.). Cells were stimulated with IL-3 for 1 h, and total cell lysates were prepared. Putative Akt substrates were detected using the Akt substrate antiserum (top). Ubiquitinated Akt substrates that were detectable in epoxomicin-treated vector control cells and epoxomicin-treated myrAkt cells, but not vector control cells treated with LY294002, are marked with a star. A potential ubiquitinated Akt substrate detected in only myrAkt cells is marked with an arrow. The blot was stripped and reprobed with anti-ubiquitin as a loading control (bottom). nated proteins detected using the Akt substrate antiserum (Fig. 5B). The addition of LY294002 reduced but did not eliminate the detection of Akt substrates among ubiquitinated proteins in myrAkt cells, suggesting that the effect of LY294002 in vector control cells required inactivation of Akt. By combining the data obtained from vector control and myrAkt cells, ubiquitinated Akt substrates can be identified as proteins that are 1) detected in epoxomicin-treated vector control cells, 2) not detected in vector control cells treated with LY294002 and epoxomicin, and 3) detected in myrAkt cells treated with either epoxomicin or LY294002 and epoxomicin. Proteins meeting these criteria are marked with an asterisk (Fig. 5B). An additional potential ubiquitinated Akt substrate was also identified only in lysates from myrAkt cells. These data suggest that a number of proteins are phosphorylated and ubiquitinated in an Akt-dependent manner in response to stimulation with IL-3 or expression of myrAkt, resulting in proteasome-dependent degradation of Akt substrates.
Previous reports have suggested that Akt phosphorylation of substrates can alter their subcellular distribution, induce their association with 14-3-3 proteins, and/or inhibit their catalytic activity (29). The data presented here show that an additional effect of Akt phosphorylation is the ubiquitination and proteasomal degradation of the substrates of Akt. Prolonged expression of myrAkt caused proteasome-dependent declines in tuberin and FOXO protein levels, which were recapitulated to a lesser degree by stimulation of the PI3K/Akt pathway by the IL-3 receptor. Differences in the magnitude of the effects of myrAkt and the IL-3 receptor on tuberin and FOXO protein levels may be a result of the relative levels of Akt activation between the two stimuli, as shown in Fig. 1A. The induction of proteasomal degradation of tuberin and FOXO3a by Akt may serve as prototypic examples of a general mechanism used by Akt in the coordinated regulation of cell growth and survival. This hypothesis is supported by the data presented in Fig. 3A, which indicate that the phosphorylation of a number of Akt substrates is stabilized by the addition of proteasome inhibitors. Furthermore, endogenous levels of Akt are sufficient to cause both ubiquitination and degradation of substrates (Figs. 3B, 4A, and 5B), indicating that these effects can be induced in a physiologic context.
Recent findings show that Akt can control the expression levels of proteins, such as p53, the androgen receptor, and insulin receptor substrate-1 (30 -32). The mechanism by which Akt targets substrates such as tuberin and FOXO proteins for ubiquitination is unknown. Akt has been shown to activate the ubiquitin ligase activity of murine double minute-2, and murine double minute-2 activity has been implicated in Akt-stimulated increases in androgen receptor and p53 degradation (30,31,33). Alternatively, an E3 ubiquitin ligase may specifically target proteins with phosphorylated residues within Akt substrate motifs in a manner analogous to the regulation of protein stability by ubiquitin ligases capable of binding to proteins phosphorylated by glycogen synthase kinase-3 (34).
The PI3K/Akt pathway plays a critical role in promoting cell growth and survival downstream of growth factor receptors. Much of the growth and survival activities of Akt have been attributed to the ability of Akt to inhibit the function of proteins that mediate cellular atrophy and programmed cell death. Using tuberin and FOXO3a as examples, we have shown that Akt targets its substrates for degradation via the proteasome. We propose that proteasomal degradation of substrates may therefore represent a central mechanism by which Akt acts to promote cell growth and survival.