Blockade of Fatty Acid Synthase Induces Ubiquitination and Degradation of Phosphoinositide-3-Kinase Signaling Proteins in Ovarian Cancer

The phosphoinositide-3-kinase (PI3K) signaling cascade controls many cellular functions including proliferation, differentiation, tumorigenesis, angiogenesis, autophagy, and apoptosis. A high proportion of ovarian carcinomas harbor molecular aberrations within the PI3K system including, for instance, PI3K itself, AKT, mTOR, and p70S6K (1, 2). The core route of the pathway from PI3K to AKT and mTOR diverges at mTOR (3), which activates either p70S6K and ribosomal protein S6 or the eukaryotic translation initiation factor 4E (eIF4E) inhibitor–binding protein 1 (4EBP1). The S6 protein is a component of the 40S ribosomal subunit, which is required for the synthesis of polypeptides at the ribosome. 4EBP1, on the other hand, modulates the function of eIF4E, which is involved in cap-dependent translation of growth-promoting genes. Thus, both S6 and 4EBP1 are involved in protein synthesis (4).

In addition, the PI3K system is well known to transmit oncogenic signals derived from membrane receptor tyrosine kinases, such as ErbB1 (epidermal growth factor receptor) or ErbB2 (HER2/neu; ref. 5). Notably, both ErbB1 and ErbB2 have been shown to be hyperactivated in 55% and 35% of ovarian carcinomas, respectively (6–9). This would suggest that blockade of ErbB1 and ErbB2 would be a promising strategy for treatment of ovarian carcinomas. However, unfortunately, clinical studies yet reveal that ErbB1- or ErbB2-inhibitory drugs are largely inefficient in ovarian carcinomas (10). Definitive molecular explanations for these failures are still missing. Therefore, novel molecular targeting strategies for the treatment of ovarian carcinomas are urgently needed.

Fatty acid synthase (FASN), the enzyme responsible for de novo synthesis of lipids from sugars, is overexpressed in up to 80% of ovarian carcinomas and signifies unfavorable clinical outcome (11, 12). FASN supports the formation of membrane-anchored lipid rafts. These lipid rafts accommodate growth factor receptors, including ErbB family proteins, and facilitate receptor signaling (13). Thus, FASN is associated with oncogenic ErbB signaling. Most importantly, inhibition of FASN delays disease progression of human ovarian carcinomas (14). Recently, we reported that the FASN inhibitor C75 induces growth arrest in ovarian carcinoma cells, downregulates ErbB1 and ErbB2, and sensitizes the cells against ErbB blocking drugs (15). Here, we show that C75 affects ovarian carcinoma cell growth by attenuating PI3K pathways downstream of ErbB receptors, and we provide first evidence that this effect is caused by ubiquitin-mediated proteolysis of PI3K cascade molecules rather than by dephosphorylation of these proteins. In contrast, the PI3K inhibitor LY294002 blocks phosphorylation but does not stimulate degradation of PI3K and its effector proteins. Therefore, targeting FASN not only abrogates lipogenesis in cancer cells but also efficiently downregulates oncogenic PI3K signaling.

A2780 (kind gift from M. Krainer, Medical University Vienna); OVCAR-3, SKOV-3 (American Type Culture Collection); and HEY (originally from R.N. Buick, University of Toronto, Toronto, Ontario, Canada) were maintained at 37°C, 5% CO₂, and 95% humidity in RPMI-1640, α-MEM, or Dulbecco's Modified Eagle's Media, respectively, containing 10% fetal calf serum (FCS), 100 IU (μg)/mL penicillin–streptomycin, and 2 mmol/L glutamine (Gibco). They were diluted once or twice weekly at a ratio of 1:10 to 1:15 and cultured for up to 20 passages. The cell lines were tested for absence of Mycoplasma (Venor GeM, Minerva Biolabs) and the species origins were confirmed as human by species PCR, and cell line identities and uniqueness were tested by fluorescent nonaplex PCR of short tandem repeat markers (DSMZ). FASN inhibitor C75 (Sigma), PI3K inhibitor LY294002 (Calbiochem), protein biosynthesis inhibitor cycloheximide (Sigma), and proteasome inhibitor bortezomib (ChemieTek) were dissolved in dimethyl sulfoxide (DMSO) and diluted at the ratio of 1:500 or 1:1,000 in culture media immediately before use. Chloroquine (Sigma), a lysosomotropic agent that blocks autophagic flux, is soluble in aqueous solutions and was used at final concentrations of 1.25, 2.5, 5, and 15 μmol/L.

Cells plated in media containing 5% FCS were treated with C75, LY294002, cycloheximide, or bortezomib, or with combinations of C75 with cycloheximide or with bortezomib. Proteins (20 μg/lane) were subjected to SDS-PAGE, blotted onto polyvinylidene difluoride membranes, and immunostained as described (16) using the following antibodies: anti-AKT, anti-phospho-AKT (Ser473), anti-mTOR, anti-phospho-mTOR (Ser2448), anti-p70S6K, anti-phospho-p70S6K (Thr421/Ser424), antiphospho-p70S6K (Thr389), anti-S6 (clone 5G10), antiphospho-S6 (Ser240/244), anti-4EBP1, anti-phospho-4EBP1 (Ser65), anti-MEK1/2, anti-ERK1/2, anti-phopho-ERK1/2 (Thr202/Tyr204), anti-caspase-3, anti-PARP-1, anti-Beclin 1, anti-Atg5 (clone D1G9), or anti-ubiquitin [clone P4D1; all from Cell Signaling Technology, and all at a dilution of 1:1,000, except anti-phospho-S6 (Ser240/244), which was used at 1:2,000], or anti-actin (I-19, Santa Cruz Biotechnology, at 1:1,000), or anti-LC3B (Genetex, at 1:2,000). Secondary antibodies were peroxidase-tagged donkey-anti-rabbit (Promega), donkey-anti-goat (Santa Cruz Biotechnology), or chicken-anti-mouse IgG (Santa Cruz Biotechnology) at the ratio of 1:15,000. Detection was by enhanced chemiluminescence.

SKOV-3 cells (6 × 10⁵/60-mm dish) grown in media containing 5% FCS were treated for 48 hours with 20 μg/mL C75 and then processed for immunoprecipitation with anti-AKT or anti-actin using the Immunoprecipitation Kit (Protein G) from Roche Diagnostics. Sample preparation was done essentially as described in the manufacturer's protocol. Briefly, proteins from cell lysates were precleared and were immunoprecipitated overnight at 4°C on a rocking platform using specific antibodies bound to Protein G Agarose beads. Precipitated proteins were denatured (5 minutes, 95°C), separated from the beads, and then subjected to SDS-PAGE and Western blotting using anti-ubiquitin, anti-AKT, or anti-actin antibodies as described previously.

Cells were plated in 96-well plates. After adhesion, media with or without drugs were added. Cell numbers were estimated after indicated time periods using a formazan dye assay (Biomedica), and results from triplicate determinations were given as mean ± SD.

SKOV-3 cells grown in 35-mm dishes were exposed for 72 hours to 0 to 20 μg/mL C75. Adherent cells were trypsinized, pooled with nonadherent cells, and stained with 0.01% trypan blue. Cell viability was then determined by direct counting of the cells under the microscope. For each drug concentration, the viable (pale) and the nonviable (blue) cell subsets were related to the sum of both populations, which has arbitrarily been set at 100%. Results from triplicate determinations were given as mean ± SD.

SKOV-3 cells, grown to subconfluence in 35-mm dishes, were exposed for 48 hours to 0.2% DMSO (solvent), to 20 μg/mL C75 alone, to 2 μmol/L bortezomib alone, or to C75 and bortezomib together. For detection of acidic vesicles such as late endosomes and lysosomes, the cells were stained for 20 minutes with acridine orange (Sigma, 1 μg/mL) and then subjected to fluorescent microscopy. Chloroquine (24 hours, 15 μmol/L) was used as a lysosomotropic positive control drug.

A2780 was transiently transfected with 60 nmol/L FASN siRNAs and SKOV-3 with 75 nmol/L Beclin 1 (BECN1) or Atg5 siRNAs according to the manufacturer's instructions (Dharmacon). Briefly, 3 × 10⁵ cells/35-mm dish were incubated in DharmaFECT 3 siRNA lipid transfection reagent either containing 60 or 75 nmol/L of siGENOME non-targeting siRNA #2 (D-001210-02-05) that targets firefly luciferase mRNA but does not interact with any known human mRNA sequence or containing a set of 4 siRNA species, each at a concentration of 15 or 18.75 nmol/L, that target distinct sequences of the human FASN, Beclin 1, or Atg5 mRNAs, respectively (ON-TARGETplus SMARTpool FASN, L-003954-00; ON-TARGETplus SMARTpool BECN1, L-010552-00; and ON-TARGETplus SMARTpool ATG5, L-004374-00). After 72 hours of culture at 37°C, 5% CO₂, and 95% humidity in RPMI-1640 or α-MEM containing 5% FCS and 2 mmol/L glutamine, untreated or C75-exposed (10 μg/mL, 72 hours) transfectants were subjected to a formazan dye cell proliferation assay (see above). Moreover, untreated transfectants were lysed and further processed for SDS-PAGE and Western blotting to detect protein expression, phosphorylation, and ubiquitination (see above). All data were obtained from unselected polyclonal transfectant populations.

Human constitutively active (CA)-MEK1 lacking nuclear export signal domain and harboring activating amino acid substitutions at 2 phosphorylation sites (kindly provided by H. Kiyama, Osaka City University, Osaka, Japan) and murine CA-AKT1 carrying the myristoylated signal of Lck, which targets it to the membrane (kindly provided by D. Efremov, International Center for Genetic Engineering and Biotechnology, Rome, Italy) were transfected into SKOV-3 cells essentially as previously described (17). Specifically, pcDNA3 containing the gene for GFP was used as a negative control expression plasmid. For transient transfection, 5 × 10⁵ to 8 × 10⁵ SKOV-3 cells were plated in 6-well plates and allowed to attach. Then, 1 μg plasmid DNA diluted in Opti-MEM I (Invitrogen) containing Lipofectamine LTX (Invitrogen) was incubated for 30 minutes at room temperature to allow formation of Opti-MEM–DNA-LTX transfection complexes, added to cells, and left in cultures for 24 hours. Cells were then allowed to grow for another 24 hours in the absence of exogenous plasmid DNA. These transfected, polyclonal cell populations were not undergone any further clonal selection or enrichment but were either directly processed for Western blot analysis or were trypsinized and subjected to C75 treatment and cell growth assays as described above.

The Student t test for unpaired samples was applied to examine the significance in differences of protein expression/phosphorylation or of cell growth. Values of P < 0.05 were considered significant.

Western blots shown in Fig. 1 show that the FASN inhibitor C75 downregulates the expression of the PI3K downstream proteins AKT, mTOR, p70S6K, and 4EBP1 in A2780 ovarian carcinoma cells in a time-dependent manner. The protein levels are reduced after 2 or 4 hours of drug exposure and remain lowered during the observation period (72 hours). Thus, C75 induces silencing of the PI3K pathway. Interestingly, when we determined the relationship between the phosphorylated and the corresponding total protein bands, we found that C75 also inhibits the specific phosphorylation/activity of the PI3K signaling proteins. This was particularly striking for 4EBP1 (Fig. 1). In contrast, C75 decreases total and phosphorylated extracellular signal-regulated kinase (ERK) 1 and ERK2, the crucial signal effectors of the mitogen-activated protein kinase (MAPK) cascade, only slightly and only after prolonged periods of exposure (≥ 48 hours; Fig. 1).

The effects of the FASN inhibitor C75 on PI3K and MAPK signaling were then compared with those induced by the direct PI3K antagonist LY294002. LY294002 blocks specific phosphorylation/activity of PI3K downstream effectors, but unlike C75, it does not downregulate protein steady-state levels (Supplementary Fig. S1). Thus, our data indicate that C75 and LY294002 affect PI3K signaling by crucially different mechanisms of action. While C75 inhibits the PI3K pathway primarily by downregulation of effector protein expression, LY294002 directly targets phosphorylation/activation of these proteins but does not downregulate the protein expression levels. Dose–response experiments (not shown) revealed that C75 and LY294002 are equally efficient in growth inhibition at the concentrations used (7 μg/mL for C75 and 12 μg/mL for LY294002) and block A2780 cell growth completely (Supplementary Fig. S2).

Unphosphorylated 4EBP1 associates with elF4E. Thereby, 4EBP1 prevents eIF4E from binding to the eIF4F translation initiation complex and precludes eIF4F-dependent protein synthesis. Therefore, we speculated whether C75-mediated downregulation of PI3K and MAPK proteins might be caused by C75-dependent dephosphorylation of 4EBP1. However, 2 observations are against this idea. First, C75 suppresses AKT and p70S6K levels already after 2 hours of treatment, whereas dephosphorylation of 4EBP1 takes longer and requires at least 4 hours of exposure (Fig. 1). Second, LY294002 completely abrogates phosphorylation of 4EBP1 but does not downregulate PI3K and MAPK protein levels (Supplementary Fig. S1). Thus, these data suggest that PI3K and MAPK signaling proteins are not controlled by eIF4F-dependent translational mechanisms in A2780 ovarian carcinoma cells.

We then sought to determine whether the observed C75-mediated downregulation of PI3K and MAPK proteins might result from elevated protein degradation. To test this, A2780 cells were cotreated for 8 to 72 hours with C75 and with the translation inhibitor cycloheximide, which is known to block de novo protein synthesis. Cell proteins were extracted after lysis of the cells and subjected to Western blotting. Separated proteins were then analyzed for the expression of AKT, mTOR, p70S6K, 4EBP1, and ERK1/2. Interestingly, as shown in Fig. 2, C75 markedly accelerates and aggravates cycloheximide-enabled downregulation of each signaling protein tested. Consequently, the FASN inhibitor C75 markedly shortened the period required for cycloheximide to depress the steady-state levels of the PI3K and MAPK proteins to 50% of the original levels (half-lifes, IT₅₀; Supplementary Table S1). Thus, inhibition of FASN silences the PI3K pathway mainly by reducing the stability of the downstream proteins rather than by direct dephosphorylation of these proteins.

Degradation of proteins is usually triggered by attachment of ubiquitin moieties to the client protein. Therefore, we hypothesized that the FASN inhibitor C75, but not the PI3K blocker LY294002, induces protein ubiquitination. To test this, A2780 cells were exposed to vehicle or drugs followed by Western blot analysis using a ubiquitin-specific antibody. Interestingly, inhibition of FASN causes a strong accumulation of ubiquitinated proteins, whereas direct blockade of PI3K does not promote protein ubiquitination (Fig. 3). Thus, C75, in contrast to LY294002, stimulates protein ubiquitination. This is in agreement with the data showing that C75, but not LY294002, decreases the steady-state levels of PI3K and MAPK signaling proteins (Fig. 1 and Supplementary Fig. S1). Therefore, C75-mediated degradation of signaling proteins is, at least in part, caused by elevated ubiquitination of these proteins.

The FASN inhibitory drug C75 has been described as a malonyl-CoA analogue and is known to concomitantly activate fatty acid degradation by antagonizing the allosteric inhibitory effect of malonyl-CoA on carnitine palmitoyltransferase-I (CPT-I), the major rate-limiting enzyme for long-chain fatty acid oxidation (18–20). Therefore, we examined whether the observed C75-induced effects on PI3K downstream protein expression, phosphorylation, and stability were indeed because of blockade of FASN and not because of concurrent activation of CPT-I by transiently transfecting A2780 cells with a set of 4 different FASN-specific siRNAs. Notably, Bandyopadhyay and colleagues (21) recently reported that siRNA-mediated silencing of FASN expression, in contrast to pharmacologic inhibition of FASN by C75, counteracts CPT-I function. In our studies, FASN-targeted RNA interference efficiently abrogated FASN protein expression and blocked cell growth. Most importantly, concordant with C75, FASN siRNAs dramatically downregulated the expression and phosphorylation of the PI3K downstream effectors AKT, mTOR, S6, and 4EBP1. On the other hand, while reducing overall expression of the MAPK proteins ERK1 and ERK2, they did not markedly impede phosphorylation of these signaling proteins. Moreover, corresponding to C75, FASN siRNAs stimulated protein ubiquitination (Fig. 4).

Using in vitro growth assays and Western blot analyses, we then examined the effects of C75 on cell growth, on apoptosis, on PI3K and MAPK signaling pathways, and on protein ubiquitination in a panel of additional ovarian carcinoma cell lines (OVCAR-3, SKOV-3, and HEY). Figure 5 shows that C75 effectively reduced the cell numbers (Fig. 5A), downregulated the expression and phosphorylation of AKT and S6, and stimulated protein ubiquitination in all these cell lines. Moreover, similar to A2780, C75 did also lower the ERK1 and ERK2 protein levels but had less effect on the phosphorylation of ERK1/2 in these cells (Fig. 5B and C). Immunoprecipitation for AKT followed by Western blotting against ubiquitinated proteins provided direct evidence that AKT indeed is a client protein of C75-induced ubiquitination in SKOV-3 cells (Fig. 5D). Cytotoxicity of C75 was shown by a relative, dose-dependent increase of trypan blue–stained (dead) cells (Fig. 5E) and by a decrease of caspase-3 and PARP-1, and a correlative occurrence of cleaved PARP-1 indicating apoptotic cell death (Fig. 5F).

Unexpectedly, exposure of SKOV-3 cells to the proteasome inhibitor bortezomib, which abrogates proteasomal proteolysis (22), did not upregulate the steady-state levels of AKT and ERK1/2. Instead, when given alone, bortezomib downregulated AKT just as C75 did. Moreover, AKT expression was almost completely shut down, when both bortezomib and C75 were given together. In contrast, ERK1/2 expression was not significantly affected by bortezomib (Fig. 6A). This suggests (i) that neither AKT nor ERK1/2 protein is subject to proteasomal degradation and (ii) that blockade of the proteasome by bortezomib activates an alternative proteolytic mechanism that causes downregulation of AKT. This pathway can further be stimulated by coexposure to C75. Consequently, we wondered whether C75 might stimulate lysosomal activity. To this end, SKOV-3 cells were stained with the fluorescent dye acridine orange, which labels acidic cell compartments such as lysosomes (23). Figure 6B shows that both bortezomib and C75 alone or together strongly stimulate the formation of lysosomes in SKOV-3 cells. Moreover, Western blot analysis revealed that LC3B-I was converted by C75 to the smaller form LC3B-II, which is a marker for autophagy (Fig. 6C). Interestingly, the lysosomotropic agent chloroquine blocks autophagic flux after generation of LC3B-II but before lysosome–autophagosome fusion. Consequently, chloroquine also causes marked cellular accumulation of lysosomes (Fig. 6B). We observed that inhibition of C75-mediated autophagy, either by coexposure of C75-treated SKOV-3 cells to the autophagic inhibitor chloroquine (Fig. 6D) or by siRNA knockdown of the key autophagic proteins Beclin 1 or Atg5, reverts C75-mediated growth inhibition (Fig. 6E and F). Thus, our data show that C75 stimulates lysosomal accumulation and formation of autophagosomes and that induction of autophagy essentially contributes to the toxic effects of C75.

To assess the potential roles of the PI3K and the MAPK signaling pathways in mediating the anticancer action of C75, we transiently transfected constitutively active variants of AKT1 or MEK1 into SKOV-3 cells. Overexpression/hyperphosphorylation of transgenes or downstream effector molecules (AKT, MEK1/2, and ERK1/2) was confirmed by Western blot analysis of the unselected polyclonal transfectant cell populations. Unlike wild-type SKOV-3 and GFP-transfected control cells, AKT1 and MEK1 transfectants exhibited an excess of the respective unphosphorylated and phosphorylated signal effectors (Fig. 7A). These transfectants were used to examine the relative contribution of MAPK and PI3K signaling to the growth-inhibitory effect of C75. As shown in Fig. 7B, C75 was found to strongly and equally suppress the growth of GFP- and MEK1-transfected SKOV-3 cells, whereas it was significantly less efficient in blocking growth of AKT1-transfected cells. Thus, our data show (i) that forced expression of dominant active AKT1 and MEK1 efficiently stimulates PI3K and MAPK pathways in SKOV-3 cells, respectively, and (ii) that PI3K/AKT, but not MAPK/MEK, hyperactivity confers growth resistance of the cells against C75. Thus, deactivation of the PI3K system seems to be crucial for C75 to exert its anticancer effect in these cells.

In summary, our data suggest that FASN activity is involved in the functionality of proteolytic pathways via a molecular cross-talk involving FASN and the ubiquitination system. Within this scenario, the deactivation of the PI3K system essentially contributes to the anticancer effect of C75 in ovarian carcinoma cells.

The PI3K signaling pathway is crucial for survival and for progression of ovarian carcinoma cells through the G₁ phase of the cell cycle. The p110α catalytic subunit of PI3K (PIK3CA) is one of the most frequently altered oncogenes in cancer. Approximately 40% of ovarian tumors reveal amplification, whereas only 7% reveal mutation of PIK3CA. Conversely, the majority of breast cancers harbor mutations in PIK3CA (40%), whereas only a few reveal PIK3CA amplifications (9%; refs. 1, 24, 25). We and others have recently shown that the PI3K signaling cascade interacts with lipogenic pathways in cancer cells. In fact, FASN, the crucial enzyme in de novo generation of long-chain fatty acids, is overexpressed in approximately 80% of ovarian carcinomas and is linked with the PI3K system in these tumors (13, 15). Unfortunately, the mechanisms of FASN-mediated regulation of PI3K signaling are still not fully characterized. Experimental evidence suggests that multiple pathways are involved (15). For instance, inhibition of FASN has been found to abrogate incorporation of phospholipids, the end products of FASN action, into cell membrane–anchored lipid rafts, which accommodate growth factor receptors. It is, therefore, thought that blockade of FASN indirectly impedes the formation of membrane signaling complexes and prohibits recruitment and activation of downstream effectors such as PI3K (15). Moreover, persistent abrogation of FASN-dependent lipid biosynthesis causes accumulation of reactive oxygen species and FASN substrates as well as depletion of FASN end products (15, 21). These stressful conditions impose a durable insult on the malignant cells, which consequently try to build up a compensatory unfolded protein response that is part of the endoplasmic reticulum stress response. These processes can lead to inhibition of PI3K and to cell growth arrest (26, 27). Because of these properties, FASN has been considered to represent a promising anticancer drug target. C75 is a derivative of the natural FASN blocker cerulenin. It was developed as an anti-FASN agent with a better chemical stability than cerulenin. Unfortunately, both compounds have been shown to affect nutrient uptake and body weight in mice, which hampers their use in clinical oncology (28). However, recently, several novel FASN drugs have been developed, which allegedly overcome these limitations including, but not limited, to C93, C247, and FAS31, a series of 3-aryl-4-hydroxyquinolin-2(1H)-one derivatives, several analogues of epigallocatechin-3-gallate, and a panel of bisamide derivatives. Many of these promising substances are now subject to clinical evaluation and it is hoped that at least some of them will exploit the full potential of FASN as a cancer drug target in the years to come (reviewed in ref. 29). Here, we show that C75- and FASN-specific siRNAs impair both expression and phosphorylation of the PI3K downstream effector proteins AKT, mTOR, p70S6K, S6, and 4EBP1 in ovarian carcinoma cells (Figs. 1, 4, and 5). In contrast, the PI3K-targeting drug LY294002 inhibits phosphorylation of the PI3K downstream effectors but leaves expression of these proteins essentially unaffected (Supplementary Fig. S1). On the other hand, although abrogation of FASN by C75 or by siRNAs, and of PI3K by LY294002 does repress the steady-state levels of ERK1/2 protein to various degrees, phosphorylation of ERK1/2 is yet not consistently inhibited in FASN- or PI3K-blocked ovarian carcinoma cells (Figs. 1, 4, and 5 and Supplementary Fig. S1). This uncoupling between ERK1/2 expression and phosphorylation is in line with reports showing that the PI3K/AKT/mTOR cascade negatively cross-talks with MAPK signaling. For instance, it has been shown that hyperactive AKT curbs the function of C-Raf, which is the first kinase within the MAPK cascade (30). In addition, experimental data indicate that Rheb, which usually activates mTOR, also acts as an inhibitor of MAPK signaling (31). Importantly, exogenous expression of hyperactive AKT1, but not of hyperactive MEK1, significantly lowered the sensitivity of SKOV-3 cells against C75 (Fig. 7). Thus, our data suggest that the anticancer effects of FASN inhibition are, at least in part, mediated by inhibition of PI3K signaling.

Activation of the PI3K/AKT/mTOR cascade leads to phosphorylation and dislodgement of 4EBP1 from eIF4E. This enables eIF4E to bind to the scaffolding protein eIF4G, which contains the RNA helicase eIF4A as second binding partner. The assembly of eIF4A with eIF4E and eIF4G is designated eIF4F and represents the active translation initiation complex (4). PI3K/AKT/mTOR also activates ribosomal protein S6 via phosphorylation of p70S6K. Both events, activation of eIF4F and phosphorylation of S6, stimulate translation of genes that regulate cell growth and apoptosis. Therefore, it is tempting to speculate that C75- and FASN siRNA–mediated downregulation of PI3K signaling occurs in 2 steps including a negative autoregulatory loop. Inhibition of FASN would first inhibit the phosphorylation/activity of PI3K downstream effectors. This would result in reduced activity of 4EBP1 and S6 and, consequently, in reduced translation of oncogenic genes including the PI3K signaling proteins. However, our findings argue against this idea because we observed that the direct PI3K inhibitor LY294002 diminishes phosphorylation but does not reduce the expression of PI3K downstream proteins (Supplementary Fig. S1). Instead, we actually show that inhibition of FASN rather stimulates the degradation of PI3K and MAPK signaling proteins in ovarian carcinoma cells. Therefore, C75-induced downregulation of oncogenic signaling proteins likely is caused by increased proteolysis (Fig. 2). This is further supported by our finding that inhibition of FASN by C75- or by FASN-specific siRNAs markedly increases the ubiquitination of the proteins in ovarian carcinoma cells and that AKT is a direct client for ubiquitination (Figs. 3–5). Coupling of the short modifier protein ubiquitin to a lysine residue of a protein targets this protein for lysosomal or proteasomal degradation. In the ovarian carcinoma cells analyzed, it seems that the proteasome inhibitor bortezomib does not stabilize AKT and ERK1/2. Hence, these proteins are most likely degraded along the lysosomal rather than the proteasomal route (Fig. 6A), which correlates with our finding that inhibition of FASN causes striking accumulation of acidic vesicles, such as late endosomes, lysosomes, and autophagosomes, and ultimately induces autophagy in ovarian carcinoma cells (Fig. 6B and C). Interestingly, blockade of C75-mediated autophagy by drugs, such as chloroquine, or by siRNAs that silence crucial autophagic proteins, such as Beclin 1 or Atg5, efficiently abrogates the growth-inhibitory action of C75 (Fig. 6D–F). This shows that autophagy is a cytotoxic rather than a cytoprotective response of ovarian carcinoma cells to inhibition of FASN. Generally, the activity of signaling pathways can be regulated either by site-specific phosphorylation of the effector proteins or by modulation of their expression level. Degradation of signaling proteins can occur either transiently in response to activation (32) or constitutively, independent of the activation state of the client protein. This has recently been shown for several PI3K downstream effectors including serum- and glucocorticoid-induced kinase-1 (SGK1) and p70S6K (31, 33–36). Our data reveal for the first time that blockade of FASN stimulates degradation of PI3K and MAPK signaling proteins, which is most likely caused by induced ubiquitination and subsequent autophagic decomposition of the effector proteins. This is corroborated by a recent report by Knowles and Smith (37), who showed that FASN activity is inversely related to the expression of several ubiquitination enzymes. Thus, we provide first evidence that inhibition of FASN reduces PI3K signaling not because of direct interference of FASN pathways with PI3K downstream effector phosphorylation but rather because of elevated ubiquitin-dependent proteolysis of these PI3K effectors.

In summary, our data show that growth arrest and cytoreduction, which is induced by inhibition of FASN activity, is mediated, at least in part, by indirect silencing of the PI3K signaling pathway because of increased ubiquitination and degradation of the PI3K effector proteins. The observed molecular cross-talks between FASN, PI3K, and proteolytic pathways provide the rationales for a number of novel combination targeting approaches that may ultimately improve the clinical management of ovarian carcinomas.

No potential conflicts of interest were disclosed.

The work was supported by Medical Scientific Fund of the Mayor of the City of Vienna (to C.F. Singer and T.W. Grunt, #08037), Austria.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.