Phosphatidylinositol 3-Kinase/Akt Activity Regulates c-FLIP Expression in Tumor Cells*

The caspase-8 homologue FLICE-inhibitory protein (FLIP) functions as a caspase-8 dominant negative, blocking apoptosis induced by the oligomerization of the adapter protein FADD/MORT-1. FLIP expression correlates with resistance to apoptosis induced by various members of the tumor necrosis factor family such as TRAIL. Furthermore, forced expression of FLIP renders cells resistant to Fas-mediated apoptosis. Although FLIP expression is regulated primarily by MEK1 activity in activated T cells, the oncogenic signaling pathways that regulate FLIP expression in tumor cells are largely unknown. In this report, we examined the roles of the MAP kinase and phosphatidylinositol (PI) 3-kinase signaling pathways in the regulation of FLIP expression in tumor cells. We observed that the MEK1 inhibitor PD98059 reduced FLIP levels in only 2 of 11 tumor cell lines tested. In contrast, disruption of the PI 3-kinase pathway with the specific inhibitor LY294002 reduced Akt (protein kinase B) phosphorylation and the levels of FLIP protein and mRNA in all cell lines evaluated. The introduction of a dominant negative Akt adenoviral construct also consistently reduced FLIP expression as well as the phosphorylation of the Akt target glycogen synthase kinase-3. In addition, infection of the same cell lines with a constitutively active Akt adenovirus increased FLIP expression and the phosphorylation of GSK-3. These data add FLIP to the growing list of apoptosis inhibitors in which expression or function is regulated by the PI 3-kinase-Akt pathway.


The caspase-8 homologue FLICE-inhibitory protein (FLIP) functions as a caspase-8 dominant negative, blocking apoptosis induced by the oligomerization of the adapter protein FADD/MORT-1. FLIP expression correlates with resistance to apoptosis induced by various members of the tumor necrosis factor family such as TRAIL. Furthermore, forced expression of FLIP renders cells resistant to Fas-mediated apoptosis. Although FLIP expression is regulated primarily by MEK1 activity in activated T cells, the oncogenic signaling pathways that regulate FLIP expression in tumor cells are largely unknown. In this report, we examined the roles of the MAP kinase and phosphatidylinositol (PI) 3-kinase signaling pathways in the regulation of FLIP expression in tumor cells.
We observed that the MEK1 inhibitor PD98059 reduced FLIP levels in only 2 of 11 tumor cell lines tested. In contrast, disruption of the PI 3-kinase pathway with the specific inhibitor LY294002 reduced Akt (protein kinase B) phosphorylation and the levels of FLIP protein and mRNA in all cell lines evaluated. The introduction of a dominant negative Akt adenoviral construct also consistently reduced FLIP expression as well as the phosphorylation of the Akt target glycogen synthase kinase-3. In addition, infection of the same cell lines with a constitutively active Akt adenovirus increased FLIP expression and the phosphorylation of GSK-3. These data add FLIP to the growing list of apoptosis inhibitors in which expression or function is regulated by the PI 3-kinase-Akt pathway.
The p55 TNF 1 receptor, Fas, LARD (DR3), and the TRAIL receptors DR4 and DR5 are members of the TNF receptor family that contain a unique cytoplasmic sequence termed a death domain (DD). Upon engagement by ligand or antibody cross-linking, these receptors trimerize and recruit an adapter protein called FADD (Fas-associated death domain) (1-3), which interacts through its own DD with those of the receptors or other adapters (e.g. TRADD). The resulting multimolecular complex, referred to as a death-inducing signaling complex or DISC, recruits an inactive procaspase called FLICE or caspase-8, which is then cleaved to an active form, initiating a cascade of proteolytic events resulting in the cleavage of numerous proteins essential for DNA repair and the maintenance of cell viability (4). It has been shown that FADD multimerization is sufficient to kill some cells even in the absence of TNF family ligands or receptor cross-linking (5). For example, the apoptosis induced by certain chemotherapeutic agents is thought to rely on ligand-independent FADD multimerization, since it can be blocked by a dominant negative FADD (5). Collectively, these data suggest that FADD plays a critical role in apoptosis induced by the cross-linking of DD-containing membrane receptors of the TNF receptor family (e.g. Fas) as well as that resulting from exposure to cytotoxic drugs.
Many tumors express Fas (6) and TRAIL receptors (7,8) yet are resistant to apoptosis induced by exposure to Fas ligand, TRAIL, or agonist anti-receptor antibodies. It has been proposed that this resistance is due to the presence of one or more inhibitors of apoptosis (9). Numerous cellular proteins, including FLIP, Fas-associated protein (FAP-1) (10), Bcl-2, Bcl-X, Bruton's tyrosine kinase (11), Toso (12), and SODD (silencer of death domain) (13,14) as well as caspase inhibitors such as cIAP (15) and survivin (16,17), have been shown to inhibit the death process, either early on at the level of Fas (FAP-1) or further downstream. FLIP is a cytoplasmic protein that has sequence homology to FLICE. FLIP is capable of binding to FADD yet is unable to be cleaved to an active caspase because of a substitution of a tyrosine for an active site cysteine, thus preventing the initiation of the death pathway (18)(19)(20). FLIP has been demonstrated to play a role in the prevention of apoptosis in a number of systems, especially those involving the immune system. Cells with high levels of FLIP relative to FLICE are generally resistant to apoptosis (21), whereas cells with low levels of FLIP relative to FLICE are more sensitive to apoptosis (21). When naive T cells are initially activated, they express Fas yet are resistant to Fas-mediated apoptosis (22,23). However, when these T cells are rechallenged with antigen, they become sensitive to Fas-mediated apoptosis, even though the expression of Fas on the surface of these cells is unchanged. This change in susceptibility to Fasmediated apoptosis correlated with FLIP levels, which are high during the initial stimulation but significantly reduced upon rechallenge (2,24,4). Interleukin-2 has been shown to markedly enhance the susceptibility of activated T cells to Fas-mediated apoptosis and at the same time down-modulate FLIP (23), the loss of which may account for the interleukin-2 effect. Recently, the protection from Fas-mediated apoptosis in B cells upon cross-linking of the B cell receptor has been shown to be associated with an increase in FLIP expression (26). Thus, at least for B and T lymphocytes, susceptibility to apoptosis induced by death receptor li-gation is governed in part by the level of FLIP.
In addition to the ability of FLIP to act as a competitive inhibitor of the binding of caspase-8 to FADD, it has also been shown to induce the activation of NF-B via TNF receptor-1 through TRADD, TRAF-2, RIP, NIK, and IKK (27), thus providing a second anti-apoptotic mechanism for FLIP.
Although initially described as a viral product (28), a cellular analogue of FLIP has since been discovered, and high levels are commonly found in tumors (18,29). In one study in melanoma cells, the levels of FLIP were found to correlate inversely with susceptibility to apoptosis induced by exposure to TRAIL (30). However, other studies have failed to identify a linkage between FLIP expression and sensitivity to Fas or TRAIL (7).
Very little is known about the signaling pathways or transcription elements that control the expression of FLIP. In activated T cells, FLIP expression has been shown to be dependent on the ERK/MAP kinase pathway because the addition of a dominant active MKK1 (MEK-1) induces FLIP in 293T and Jurkat cells (31). The molecular basis for the high constitutive levels of FLIP observed in many tumor cell lines is unknown. PI 3-kinase has been shown to protect cells from apoptosis in a caspase-dependent manner (32)(33)(34). PI 3-kinase catalyzes the phosphorylation of phosphoinositol-4 phosphate and phosphoinositol-4,5 phosphosphate at the 3 position. Kinases such as PDK1 and pkB/Akt bind to these phosphorylated intermediates via their pleckstrin homology domain. PDK1 in turn phosphorylates and activates pkB/Akt (35), which then phosphorylates several proteins that have been implicated in the control of cell survival (32, 34, 36 -39). The phosphorylation of procaspase-9 by Akt, for example, inactivates this protease, blocking the propagation of death signals originating in the mitochondria (39). Likewise, Akt-mediated phosphorylation of Bad results in the binding of this pro-apoptotic Bcl-2 family member to 14-3-3 proteins and its dissociation from the mitochondria (40 -42). Some targets of pkB/Akt, such as NF-B (43), CREB (44,45), and Forkhead (46), are transcription factors that regulate cell survival. In some cells, the anti-apoptotic effect of NF-B activation is partly attributable to the expression of cIAP, a potent caspase inhibitor (47). The phosphorylation of Forkhead by Akt blocks its activity, reducing the levels of expression of proapoptotic proteins such as Fas ligand (46). These data illustrate the diversity of downstream effectors through which the PI 3-kinase/Akt signaling pathway affects cell survival.
The ability of a constitutively active MEK1 to induce FLIP expression in T cell lines and the multiplicity of anti-apoptotic proteins in which expression and/or function are regulated through the PI 3-kinase pathway suggest that the constitutive expression of FLIP by tumor cells might be attributable to the activation of either the Ras-Raf-MEK-ERK and/or PI 3-kinase pathways. To test these hypotheses, we carried out a series of studies with drugs and a dominant negative kinase that inhibit these pathways. The results of our studies clearly implicate the PI 3-kinase/Akt pathway as the predominant regulator of FLIP expression in tumor cells.

EXPERIMENTAL PROCEDURES
Cell Lines and Reagents-The human colon adenocarcinomas HT29 and DLD-1, the prostate carcinomas DU145 and PC-3, the breast carcinoma MCF7, the renal carcinomas 786-0 and 769-P, and the melanoma G-361 cell lines were purchased from the ATCC. The human colon carcinoma cell line Clone A was obtained from Dr. Ian Summerhayes of the Lahey Clinic (Burlington, MA), and the breast carcinomas MDA-MB231 and T47D were obtained from Dr. Hava Abraham of the Beth Israel Deaconess Medical Center (Boston, MA). The PC-3 cells were maintained in McCoy's 5A media supplemented with 10% fetal bovine serum, and the G-361 cells were maintained in Kaign's modification of Ham's F12 media supplemented with 10% fetal bovine serum. All of the remaining cell lines were grown in Dulbecco's modified medium supplemented with glutamine, gentamicin, and 10% fetal bovine serum.
The MEK inhibitor PD98059 and the PI 3-kinase inhibitor LY294002 were purchased from New England Biolabs (Beverly, MA) and Calbiochem (Lajolla, CA), respectively. The vinculin antibody was purchased from Sigma. The hemagglutinin (HA) antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and the rabbit anti-pERK and anti-p-GSK3a/b antibodies were obtained from New England Biolabs. The p-Akt (Ser-473) antibody was purchased from BD/PharMingen. The FLIP antibody was generated in New Zealand White rabbits against a peptide spanning amino acids 2-27 (SAEVIHQVEEALDT-DEKEMLLFLCRD) of the human protein (48).
Western Analyses-Cells were treated with 50 M PD98059 or LY294002 in medium supplemented with 5% fetal bovine serum for the indicated times and then lysed with Passive Lysis Solution (Promega, Madison, WI) supplemented with protease (phenylmethylsulfonyl fluoride and leupeptin) and phosphatase (NaF, Na 3 VO 4 , glycerophosphate, and sodium pyrophosphate) inhibitors. Lysates were separated on 12% SDS-polyacrylamide gels, and the fractionated proteins were transferred to nitrocellulose. The top portions of the blots were removed and probed for the cytoskeletal protein vinculin to ensure equivalent loading of all samples. The bottom portions of the blots were probed either for FLIP, pERK, HA, or p-GSK3a/b. The blots were further probed with goat anti-rabbit (FLIP and pERK) or goat anti-mouse (vinculin and HA) conjugated to horseradish peroxidase and then exposed to SuperSignal chemiluminescent substrate (Pierce). The blots were then exposed to Kodak X-Omat Blue XB-1 film. The films were analyzed by densitometry using a Bio-Rad densitometer and Molecular Analyst software.
FLIP Expression-RNA was isolated from the treated tumor cells using Trizol reagent (Life Technologies, Inc.). RT-PCR was performed (22) using oligonucleotides specific for human FLIP (5Ј-GAAGTTGACT-GCCTGCTGGCTTTCT and 5Ј-TGGGGCAACCAGATTTAGTTTCTCC) and the ribosomal protein S9 (5Ј-GATGAGAAGGACCCACGGCGTCT-GTTCG and 5Ј-GAGACAATCCAGCAGCCCAGGAGGGACA). The FLIP oligonucleotides were selected using DNA STAR software to give an annealing temperature of at least 60°C and to amplify FLIP specifically (i.e. not the homologous caspase-8). [ 32 P]dCTP was incorporated into the products (550 base pairs for FLIP and 431 for S9). The radioactive products were resolved on 4% polyacrylamide gels, dried, and exposed to Kodak XAR-5 film. The films were analyzed by densitometry using a Bio-Rad densitometer and Molecular Analyst software.
Infection of DN-Akt-Cell lines were infected with adenoviral constructs bearing HA-tagged catalytically inactive (49) or constitutively active Akt (49) or with the same construct bearing the gene for ␤-galactosidase at the specified multiplicities of infection. After overnight incubation, the medium was replaced. Six or 24 h later, the cells were then lysed in preparation for Western blot analysis.

FLIP Expression Is Dependent on MEK Activation in Two Colon Cancer Cell Lines but Not in Other Tumor Cell Lines
Examined-The human colon carcinoma cell lines HT29, clone A, and DLD-1, the prostate carcinoma cell lines DU145 and PC-3, the breast cancer cell lines MCF7, MDA-MB231, and T47D, the renal carcinoma cell lines 786-0 and 769-P, and the melanoma cell line G-361 all constitutively express FLIP (Fig.  1). Because of prior reports linking FLIP expression in T cells to the activation of MEK1 (31), we sought to determine the extent to which the MAP kinase pathway contributed to the constitutive expression of FLIP in tumor cells. To examine the role of the Ras-Raf-MEK-ERK pathway in FLIP expression, tumor cells were first cultured overnight in medium containing 1% FCS and then placed in medium containing 5% FCS with or without the MEK inhibitor PD98059. As shown in Fig. 1A, ⌭RK phosphorylation was readily detectable in both the HT29 and clone A colon carcinoma cell lines after 24 h of serum starvation (zero time point). The readdition of serum transiently enhanced ERK phosphorylation in clone A and, to a lesser extent, in HT29 cells. In both cell lines, the addition of PD98059 (50 M) for even 15 min completely eliminated ERK phosphorylation as determined by Western blot analysis. The FLIP level declined with time in HT29 but not Clone A cells. Similar studies with all of the other cell lines listed above except the colon carcinoma DLD-1 demonstrated that PD98059, at concentrations sufficient to suppress ERK phosphorylation (50 M), had no effect on FLIP levels (Fig. 1B). The DLD-1 cells were similar to the HT29 cells in that their FLIP levels declined, albeit modestly, in response to PD98059. Thus, of the 11 tumor cell lines studied, HT29 and DLD-1 were the only cells in which FLIP expression appeared to be linked to sustained MEK activity.
FLIP Expression Is Dependent on PI 3-Kinase Activity-To assess the role of the PI 3-kinase pathway in the regulation of FLIP levels, four of the tumor cell lines were cultured in fresh medium containing or lacking the PI 3-kinase inhibitor LY294002. As shown in Fig. 2, the addition of fresh medium increased the levels of p-Akt to a variable extent in each cell line examined, and this increase was blocked by 50 M LY294002. In each of the tumor cell lines tested, LY294002 down-modulated FLIP expression. In the case of the HT29 cells, this suppressive effect was apparent within 6 h (not shown), whereas in Clone A, MCF7, and DU145 cells, no significant decrement was observed until 24 h. To determine whether the suppressive effect of LY294002 on FLIP protein levels was associated with a decrement in the level of FLIP transcripts, FLIP mRNA levels were analyzed by RT-PCR at various time points after placing the cells in either control medium or medium containing 50 M LY294002. As shown in Fig. 3, the addition of LY294002 resulted in a decline in the level of FLIP mRNA levels. The extent of inhibition at 6 h ranged from 50% of the control for the DU145 and MCF7 cells to as much as 70 and 85% for the Clone A and HT29 cells, respectively, as determined by densitometry. In Clone A, MCF7, and DU145 cells, the FLIP mRNA levels were reduced after only 3 h of exposure to LY294002.
FLIP Expression Is Dependent upon pkB(Akt) Activity-To determine the role of pkB(Akt) in FLIP expression, tumor cells were infected with an adenoviral construct containing an HAtagged dominant negative form of pkB/Akt. A similar adenovirus containing a ␤-galactosidase insert was used as a negative control. For the HT29, Clone A, and MCF7 cell lines, infection was carried out with a multiplicity of 300 and an overnight incubation followed by the addition of fresh medium and an additional 24 h of incubation. This protocol had no affect on the viability of these cell lines but proved toxic for the DU145 cells, which were therefore infected at a lower multiplicity of infection (150) and a shorter incubation time (6 h) to prevent significant cell death. In each case, infection of the tumor cells was confirmed by Western blot analysis using an HA-specific antibody to detect the HA-tagged dominant negative pkB/Akt (Fig.  4A). To confirm that this DN-Akt virus actually inhibited the activation of endogenous Akt, we first carried out Western blot analyses using commercially available anti-phospho-Akt antibodies as we had done previously in experiments with LY294002 (Fig. 2). These studies were largely unsuccessful . Cells were first serum-starved in 1% FCS overnight and then exposed to 5% FCS in culture media Ϯ PD98059 for 15 or 30 min or 24 h. The lysates were then probed for pERK, FLIP, and vinculin as described under "Experimental Procedures." B, DU145, PC-3, DLD-1, MCF7, MDA-MB231, T47D, 769-P, 786-0, and G-361 cells were similarly treated either with complete media (5% Y/designated media) alone or complete media and PD98059 for 24 h. The lysates were probed for pERK, FLIP, and vinculin. because of unexpected cross-reactivity of the antibodies with the DN-Akt virus. To avoid this confounding artifact, we chose a downstream Akt target, glycogen synthase kinase-3, as an indicator of Akt activity in our cell lines (50). As shown in Fig.  4, GSK-3 phosphorylation was reduced in cells infected with the DN-Akt virus but not in the uninfected control cells or cells harboring the ␤-galactisodase virus. The extent of inhibition was variable, ranging from almost complete inhibition in the HT29, Clone A, and DU145 cells to 40% for the MCF7 cells, as determined by densitometry. These data indicate that the DN-Akt virus effectively blocks the activity of endogenous Akt. Infection with the DN-Akt virus down-regulated FLIP expression in all of the tumor cell lines tested (Fig. 4A). This finding corroborates the results of studies with LY294002 implicating the PI 3-kinase/Akt pathway as a key regulator of FLIP expression in tumor cells. To verify that the down-regulation of FLIP was not a general effect of adenoviral infection and to further verify the influence of activated Akt on FLIP expression, the same cell lines were infected with an adenovirus containing a constitutively active Akt insert. As shown in Fig. 4B, infection of each cell line with the constitutively active Akt adenovirus resulted in increased phosphorylation of GSK-3. FLIP expression in each cell line was also increased. In addition to confirming a direct link between Akt activity and FLIP expression, these experiments confirm that the adenovirus was not responsible for observed changes in FLIP expression.
The mechanism by which the PI 3-kinase/Akt pathway regulates FLIP expression is yet to be determined. As mentioned previously, Akt has been shown to phosphorylate several transcription factors (e.g. CREB, NF-B, Forkhead), many of which have been implicated in the expression of genes whose protein products regulate susceptibility to apoptosis. CREB, for example, plays a major role in T cell survival (51) and is phosphorylated by Akt at the same site phosphorylated by its structurally relative protein kinase A (44). The Akt substrate Forkhead is involved in the regulation of Fas ligand expression (46) and NF-B in the expression of the caspase inhibitor survivin (47). The enhanced expression of survivin is, in fact, thought to be the dominant factor underlying the protective effect of NF-B activation against apoptosis induced by exposure to tumor necrosis factor (47). It appears likely that one or more factors may be involved in the regulation of FLIP, and studies to determine the role played by these factors in FLIP expression are currently under way.
Our results indicate that sustained MEK activity is not essential for FLIP expression in most tumor cell lines as it appears to be in activated T cells. However, FLIP expression is critically dependent upon PI 3-kinase and Akt activity. These data add FLIP to the growing list of survival-related proteins regulated by the PI 3-kinase pathway.