Gene Expression Profiling in Prostate Cancer Cells With Akt Activation Reveals Fra-1 As an Akt-Inducible Gene¹

One of the intracellular signaling pathways frequently activated in cancer cells is the phosphatidylinositol (PI) 3-kinase pathway (1). PI 3-kinases catalyze the formation of the 3′ phosphoinositides, phosphatidylinositol 3,4-diphosphate, and phosphatidylinositol 3,4,5-triphosphate (2). The production of these lipids is stimulated in cells by a variety of growth factors including heregulin and other members of the epidermal growth factor family as well as by the insulin-like growth factors I and II. Increased levels of these lipids are also found in cells which lack the tumor suppressor, PTEN (also called MMAC1) (3). This tumor suppressor is inactivated in a number of cancers including breast, glioblastomas, prostate, etc. and has been shown to be a PIP3 phosphatase (1, 3, 4). In addition, in mouse models in which PTEN has been inactivated by targeted deletion from one allele, there are increases in cancers of the endometrium, thyroid, prostate, breast, liver, and intestine (1).

Extensive studies have therefore been directed at identifying downstream targets of the PI 3-kinase pathway (5–8). In the last few years, a number of proteins which bind these lipids have been identified, including several ser/thr kinases such as the three isoforms of Akt (also called PKB), its activating kinase PDK-1, several isoforms of protein kinase C, serum and glucocorticoid-induced kinase (sgk), and even ERK in some systems. In addition, a number of other proteins have been found to bind these lipids specifically (e.g., GRIP-1) and these may link to other signaling cascades (9). In the case of Akt, increases in PIP3 lead to its membrane translocation and activation by phosphorylation (5, 6). Thus, this enzyme is constitutively active in cells lacking active PTEN. Akt has been shown capable of regulating numerous cellular functions including stimulation of cellular growth and division as well as inhibiting cellular death (apoptosis) (5, 7, 8). In addition, a number of downstream targets of Akt have been identified, including glycogen synthase kinase (GSK)-3, several mammalian homologues of the Caenorhabditis elegans DAF-16 transcription factor (now called Foxo 1, 3, and 4), the anti-apoptotic protein BAD, phosphodiesterase 3B, a Rab GTPase-activating protein, ATP-citrate lyase, and most recently, tuberous sclerosis complex-2, among others (5, 6, 10–12). How many or which of these substrates are involved in the transformation of cells is not known.

To better understand how the PI 3-kinase/Akt pathway can transform cells, we sought a method to study the activation of Akt in such a cell system. Although prior studies have reported on inducible forms of either the PI 3-kinase or Akt (13, 14), we sought a method that could more readily use available cell lines. In the present work, we report on such a system, a prostate cancer cell line lacking PTEN, PC3 cells. Moreover, we have used this system to categorize the genes induced by elevation of PIP3. Finally, we identify a gene, Fra-1, which is directly regulated by Akt and demonstrate that the regulation of Fra-1 transcription can be mapped to a PIP3-regulated binding of Sp1 to a retinoblastoma control element (RCE) in the Fra-1 promoter.

PC3 cells, which contain elevated basal levels of active Akt because they lack PTEN (15), were treated with a reversible inhibitor of PI 3-kinase, LY294002 (16). This treatment with 30 μm LY294002 was found sufficient to decrease Akt enzymatic activity by about 98% (Fig. 1). They were then washed and incubated for different periods of time without drug. On removal of LY294002, the enzymatic activity of Akt was rapidly regained, with about 50% of the enzymatic activity returning within 5 min. This treatment was therefore used to examine which genes were regulated in this system by removal of LY294002 in the hope of identifying genes the transcription of which was directly regulated by Akt.

PC3 cells were therefore treated overnight with 30 μm LY294002, washed and allowed 2, 6, or 10 h to recover. The mRNA from these cells was hybridized to cDNA microarrays containing 17,083 distinct genes and scored against control mRNA from PC3 cells treated overnight with LY294002. The genes which were most consistently induced or repressed in three independent experiments to the greatest extent after LY294002 removal in PC3 cells are listed in Table 1. The increased expression of two genes, CYP1B1 and Fra-1, was confirmed by RT-PCR and Northern analysis (Fig. 2). In addition, an increase in Fra-1 protein could be observed after LY294002 removal (Fig. 2C).

Removal of the PI 3-kinase inhibitor LY294002 could result in activation of genes in the PC3 cells by Akt as well as by other mechanisms because overall levels of PIP3 would increase in these cells, thereby resulting in the activation of all downstream targets of PIP3. To directly test whether Akt was responsible for the activation of the transcription of CYP1B1 and Fra-1, reporter constructs containing the 5′ regulatory regions of these two genes were co-transfected with constitutively active Akt into MCF-7 cells. Because PC3 cells contain primarily Akt3 kinase activity (17), we used a constitutively active Akt3 construct. The CYP1B1 reporter construct [containing 2300 nucleotides (nt) of the 5′ regulatory region of this gene] was found not to be induced by co-transfection with the constitutively active Akt3 (Fig. 3A), possible due to this gene being regulated via another downstream target of PI 3-kinase or because the reporter construct used lacked the key Akt regulatory region. A control of another activator of CYP1B1 transcription, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (18), did, however, stimulate the transcription of this reporter construct (Fig. 3A). Attempts to transfect the CYP1B1 reporter construct into PC3 cells and measure the ability of LY294002 washout to induce this transcript also failed, although the expression levels of this construct in these cells was quite low.

A reporter construct containing 749 nt 5′ to the start site of transcription of Fra-1 was induced 4- to 5-fold on co-transfection with constitutively active Akt3 (Fig. 3). This level of stimulation was comparable to the increase in Fra-1 mRNA observed with removal of the PI 3-kinase inhibitor from the PC3 cells. Moreover, an Akt mutant in which the kinase activity was inactivated by mutation did not stimulate an increase in transcription of this Fra-1 reporter construct even though it was expressed at levels comparable to the active Akt3 (Fig. 3). Indeed, it appeared to even decrease the basal level of transcription of this construct, possible by acting as a functional dominant negative. However, it did not decrease the levels of transcription of the control reporter plasmid (pCMVRL) co-transfected in these experiments, indicating that this was not a general inhibitory affect of this plasmid on transcription.

To determine whether the observed stimulation of the Fra-1 reporter construct was unique to Akt3, similar studies were performed with a constitutively active Akt1 (19). Constitutively active Akt1 was found to stimulate the transcription of the reporter construct when it was expressed at low levels (Fig. 4). However, at higher levels, the Akt1 was found to actually inhibit Fra-1 transcription, possibly due to the ability of Akt1 to feedback and inhibit responses (20). This inhibition was not observed with Akt3 even though comparable levels of protein (Fig. 4) and kinase activity were expressed (data not shown).

The region responsible for the Akt3 regulation of the Fra-1 reporter construct was examined by preparing increasing deletions of the regulatory region. A construct which contained 161 nt 5′ to the start site of transcription of Fra-1 was found to still be regulated by constitutively active Akt3 to a level comparable to that observed with the 749 nt construct, whereas a construct which had a further deletion of 56 nt lost the ability to be regulated by Akt3 (Fig. 5).

Prior analyses of the regulation of the Fra-1 gene (21) had identified an RCE site and a serum response element (SRE) in this region that we find critical for Akt3 regulation. In addition, there is a second RCE 3′ to this region (Fig. 5A). We therefore analyzed the ability of active Akt3 to induce mutants of the Fra-1 reporter construct when either RCE1, RCE2, or both RCE1 and 2 were mutated. A mutation in RCE1 reduced the ability of Akt3 to induce the transcription of the reporter construct by 75%, whereas a mutation in RCE2 decreased induction by 40% (Fig. 6A). A construct in which both RCE1 and 2 were mutated exhibited essentially no induction by active Akt3 (Fig. 6B). The basal levels of transcription of this construct were also significantly reduced. In contrast, a mutation in the SRE had no major affect on the ability of active Akt3 to induce the transcription of the reporter construct (Fig. 6A).

To further examine the mechanism for the regulation of Fra-1 gene by the PI 3-kinase pathway, we performed gel shift assays with a digoxigenin-labeled oligonucleotide containing the RCE1 site. Nuclear extracts from PC3 cells induced a shift of the RCE1 oligonucleotides (Fig. 7A, compare lanes 1 and 2) as seen by the formation of the two slower migrating bands (called C1 and C2) that were not observed in the absence of nuclear extracts. The formation of both of these complexes was blocked by an excess of the unlabeled version of this oligonucleotide (lanes 5–7). When the PC3 nuclear extracts were incubated with a labeled version of the oligonucleotide which contained a mutation in the RCE1 site (the same mutation as present in the mutant reporter construct), we still observed the C2 band but not the C1 band (lanes 9–11). Finally, nuclear extracts prepared from PC3 cells treated for 18 h with LY294002 exhibited a reduced amount of the C1 complex (Fig. 7A, lane 3), whereas nuclear extracts from PC3 cells which were treated with this inhibitor, washed, and then incubated for 5 h after removal of the inhibitor, showed a recovery of the C1 complex (Fig. 7A, lane 4). In contrast, no change in the amount of the C2 complex was observed with LY294002 treatment. These results as well as the results with the mutant RCE1 oligonucleotide implicate the components in the C1 but not the C2 complex in the PI 3-kinase regulation of the Fra-1 transcription. A similar PIP3 regulated binding of nuclear components to a labeled oligonucleotide containing the RCE2 element was not observed (data not shown).

Since prior studies have shown that Sp1 can bind to RCE sites in the promoter regions of other genes (22–24), we tested whether the complex which bound to RCE1 contained Sp1. Addition of anti-Sp1 antibodies to the nuclear extracts before the incubation with the labeled oligonucleotide caused a supershift of the C1 complex (this new band is called C3, Fig. 7B, lanes 4–6), whereas control IgG did not cause this supershift (Fig. 7B, lanes 7–9). The amount of this supershifted complex was decreased when nuclear extracts were prepared from LY294002-treated PC3 cells and it was recovered when the LY294002-treated PC3 cells were washed and further incubated for 5 h (Fig. 7B).

In recent years, increasing evidence has implicated the PI 3-kinase/Akt pathway in the transformation of cells (1, 3). This pathway has been shown to regulate cell size, proliferation, and survival as well as tumor metastasis, angiogenesis, and invasiveness (1–3, 6–8, 25). The tumor suppressor, PTEN, has been found to be a key regulator of the PI 3-kinase/Akt by virtue of its ability to degrade PIP3 and has been shown to be mutated in a variety of human cancers. Gene amplification of both the PI 3-kinase as well as some Akt isoforms has also been shown to contribute to the formation of certain tumors. Finally, many growth factors and their receptors (including members of the EGF as well as insulin-like growth factor families) have been shown to induce the PI 3-kinase pathway.

Consequently, numerous studies have sought to identify the key targets of the PI 3-kinase/Akt pathway. A variety of proteins have been identified which are direct targets of the Akt protein kinase and the phosphorylation of these proteins may contribute to the ability of this enzyme to regulate cellular proliferation and survival. In addition, the PI 3-kinase/Akt pathway also appears to regulate gene transcription. To identify such potential targets, gene profiling has been performed in cells overexpressing Akt or in tumor cells lacking PTEN after re-introduction of this enzyme (26–28). These approaches have the potential problem that they require a longer incubation time to allow for the expression of the introduced enzymes and thus one may miss transiently activated or inactivated genes as well as identify genes the activation or inhibition of which represent a secondary event. Alternatively, one may use regulatable versions of PI 3-kinase or Akt which can be rapidly turned on (29). However, this approach can only be performed in cells which have low levels of basal Akt kinase activity to begin with.

In the present work, we describe the application of an alternative approach. In the current strategy, cells lacking PTEN (and consequently having a high basal Akt activity) are treated with a reversible inhibitor of PI 3-kinase, LY294002 (16). This treatment was found to almost completely suppress the elevated basal activity of Akt in these cells. Moreover, the removal of this inhibitor resulted in a rapid (within 5 min) activation of the Akt kinase activity in these cells. Thus, this procedure can be used to attempt to identify novel substrates of Akt as well as to identify genes the transcription of which is directly regulated by the PI 3-kinase/Akt pathway.

In the present study, we have applied this approach to identify genes the transcription of which is regulated by the PI 3-kinase/Akt pathway in the context of the prostate cancer cell line PC3. A screen of 17,083 genes by microarray analysis identified 6 genes which were consistently induced more than 2-fold after 2 h and 5 genes the levels of which were consistently decreased more than 2-fold at this time. Two of the induced genes were selected for follow up analyses, CYP1B1 and Fra-1. The former is a member of the family of P450-containing enzymes which has been proposed to play a role in the metabolism and activation of various environmental carcinogens (30). Fra-1 is a member of the Fos family of transcription factors (31). It has been previously demonstrated to be induced by insulin, serum, retinoic acid, the human T-cell leukemia virus, as well as by β-catenin (21, 31–34). Although Fra-1 does not possess transforming activity in focus formation assays (35), it can promote anchorage-independent growth in vitro, tumor development in athymic mice, motility, and invasiveness (36, 37). In addition, increased levels of Fra-1 have been observed in neoplastic thyroid disease and has been shown to predict the aggressiveness of breast cancer cells (38, 39).

Attempts to demonstrate direct activation of a CYP1B1 reporter construct by active Akt were unsuccessful although a control of a CYP1B1 inducer (TCDD) did stimulate transcription of this construct. These results could indicate that another downstream target of the PI 3-kinase pathway is responsible for the increase in CYP1B1 mRNA in the PC3 cells after removal of the LY294002. Alternatively, it is possible that the reporter construct is missing the region of the CYP1B1 which is regulated by Akt. Attempts to distinguish between these two possibilities by transfections into the PC3 cells failed due to the low level of expression in these cells. In contrast to these results with the CYP1B1 reporter construct, active Akt3 was found to induce a Fra-1 reporter construct 4- to 6-fold, an amount which is consistent with the increase in Fra-1 mRNA levels observed in the PC3 cells after removal of LY294002. In contrast, an inactive mutant of Akt3 did not induce this reporter construct but instead actually appeared to decrease its basal level of transcription. Because all cells contain a basal level of active Akt, it is possible that this inactive Akt3 construct is acting as a functional dominant negative. The inhibitor of PI 3-kinase LY294002 also lowered the basal levels of transcription of this reporter construct. Quite surprising was the finding that expression of high levels of Akt1 also suppressed the expression of this reporter construct. This finding is, however, consistent with the ability of Akt1 to feedback and inhibit a number of biological responses (20).

The region of the Fra-1 promoter that is responsible for its stimulation by Akt3 was mapped to a region between −161 and −105 nt from the start site of transcription of Fra-1. This region contains a RCE (RCE1) which has previously been identified as contributing to the regulation of Fra-1 by serum and the Tax1 protein of human T-cell leukemia virus (21). In the present work, we have demonstrated that RCE1 also plays a key role in the induction of transcription of the Fra-1 reporter construct by Akt3. Moreover, we have been able to show that nuclear extracts from cells treated with the PI 3-kinase inhibitor LY294002 exhibit a decreased ability to shift an oligonucleotide containing this RCE and that this activity returns in nuclear extracts from cells that have been allowed to recover after removal of the inhibitor. Because prior studies have implicated Sp1 as one of the components able to bind to the RCE (22–24), we also examined whether this molecule was part of the complex that was regulated by PI 3-kinase. A supershift of this PI 3-kinase-dependent complex was demonstrated by the addition of anti-Sp1 antibodies to the nuclear extracts. Because prior studies have indicated that insulin can regulate Sp1 activity through either phosphorylation or heterodimerization with other factors like Sp3 or the retinoblastoma protein (40), it is possible that this is how Akt is inducing transcription of Fra-1.

An analysis of the promoter of two other genes (CYP1B1 and PTGS2) which are up-regulated on LY294002 withdrawal did not reveal any RCE sites. This is consistent with the lack of regulation of the CYP1B1 with the co-transfection with Akt. In the case of PTGS2, Akt has been recently shown to regulate the stability of its mRNA (41), and this could result in the observed increase in its mRNA in the present studies. In addition, we did not observe an increase in some other genes with a RCE (e.g., MN/CN 9), consistent with prior studies indicating that RCEs can be regulated in a cell type-specific manner (42).

In conclusion, the present studies demonstrate that Akt3 can induce Fra-1 gene transcription. Although prior studies had implicated other signaling pathways (i.e., ERK) in the induction of Fra-1 (43), Akt had not previously been identified as playing a role in this process. Because Fra-1 has been shown to contribute to the progression of certain cancers (including thyroid and breast), it is possible that this is another mechanism whereby activation of Akt can contribute to tumor formation and/or the progression of a particular cancer.

Cell culture media, RT-PCR kit, random priming kit, and the restriction endonucleases were from Life Technologies, Inc. (Gaithersburg, MD). Total RNA was isolated using the RNeasy kit from Qiagen (Chatsworth, CA) and the poly(A)⁺ fraction was isolated with the FastTrack 2.0 kit (Invitrogen, Carlsbad, CA). [γ-³²P]ATP (3000 Ci/mmol) and [α-³²P]dCTP (3000 Ci/mmol) were from NEN Life Science Products, Boston, MA. Anti-HA monoclonal antibody and Fugene6 transfection reagent were from Roche Molecular Biochemicals, Indianapolis, IN. Rabbit polyclonal anti-Fra-1 and anti-Sp1 antibodies were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Hybond nylon membrane and horseradish peroxidase conjugated to donkey anti-rabbit IgG or anti-mouse IgG were from Amersham Biosciences (Amersham, United Kingdom). Luciferase reporter vector pGL3, pCMVRL, and the Dual-Luciferase reporter assay kit were from Promega (Madison, WI). The CYP1B1 reporter construct was a kind gift of Dr. William Greenlee (CIIT, Research Triangle Park, NC) (18). The constitutively active Akt1 and Akt3 plasmids (myrAkt1 and myrAkt3 in pcDNA3.1) were as previously described (19, 44). To construct the kinase dead myrAkt3, the active site lysine was changed to methionine by base substitution using the Quickchange site-directed mutagenesis kit (Stratagene, Cedar Creek, TX).

The Fra-1 promoter (bases −749 to +23 bp relative to the transcription start site) was amplified by PCR from genomic DNA isolated from PC3 cells (sense primer: GGCAATACGGCGAGACCC, antisense primer: GTACACGGCTGCTGGGTTCTG). The resulting PCR product was cloned into pGL3 basic luciferase reporter vector (Promega) to generate plasmid pFra1. 5′ Deletion mutants of the Fra-1 promoter were amplified by PCR with specific primers and were placed back into the pGL3 basic vector. Base substitution mutations were made for RCEs and the SREs by using Quickchange site-directed mutagenesis kit (Stratagene) and specific primers. The sequence of the SRE was changed from CCAAGTTCGGG to CGCGGTTCGCG, RCE1 was changed from GGGGGTGG to GGCGCTCG, and RCE2 was changed from CCCCCACCCCC to CCCCGAGCGCC.

PC3 cells were treated with 30 μm LY294002 overnight, washed, and then incubated for 2, 6, or 10 h in the absence of the drug. Cells were then lysed, poly(A)⁺ RNA isolated, reverse transcribed, labeled, and hybridized to microarrays as described (45). The arrays contained 22,648 different human cDNAs representing 17,083 different UniGene clusters. Hybridized arrays were imaged using a Genpix 4000 scanner and the data are available at: http://genome-www5.stanford.edu/MicroArray/SMD/.

PC3 cells were maintained in a humidified atmosphere of 5% CO₂ in RPMI 1640 supplemented with 5% fetal bovine serum, 100 μg/ml streptomycin, and 100 units/ml penicillin. Cells were serum starved and treated overnight with 30 μm LY294002 and then lysed in 1 ml lysis buffer [50 mm Tris-HCl (pH 7.5), 150 mm NaCl, 1% Triton X-100, 10% glycerol, 1 mm EDTA, 1 mm DTT, 1 mm benzamidine, 1 mm phenylmethylsulfonyl fluoride, 1 mm Na₃VO₄, 30 mm NaPPi, 10 mm NaF, 100 nm okadaic acid]. Lysates were centrifuged for 15 min at 15,000 × g and immunoprecipitated for 2 h at 4°C with 2.5 μl of anti-Akt3 antibody. Immunoprecipitates were washed three times with the lysis buffer and twice with the kinase assay buffer [50 mm Tris-HCl (pH 7.5), 10 mm MgCl₂, 1 μM DTT] and assayed using GSK-3 (GRPRTSSFAEG) peptide as substrate as described (46). Following the kinase reaction, the phosphorylated peptide was separated from unincorporated [γ-³²P]ATP on a 40% polyacrylamide gel containing 6% urea. The phosphopeptide spots were excised and counted.

Total RNA was isolated from LY294002-treated PC3 cells using RNeasy kit (Qiagen). First strand cDNA was synthesized from 1 to 5 μg total RNA with random primers and Superscript II reverse transcriptase (Life Technologies). The forward and reverse primers for amplification of human Fra-1 were based on the human Fra-1 sequence (Genbank accession no. NM005438). The conditions used for amplification were: 94°C for 1 min followed by 55°C for 1 min and 72°C for 1 min. For Northern blot analyses, 10 μg RNA were denatured and separated on 1% formaldehyde/agarose gels. The RNA was transferred to nylon membranes by capillary action overnight using 20× SSC and cross-linked by UV light (1200 J/cm², Stratalinker). The blots were probed at 42°C overnight with a random primed ³²P-labeled probe of human Fra-1 cDNA as described (47).

PC3 cells, treated as indicated, or transiently transfected MCF-7 cells were washed in ice-cold PBS and then lysed by scraping in 500 μl of lysis buffer (as described above). After centrifugation at 15,000 × g for 30 min, the supernatants were resolved on 10% SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Fra-1 and Akt3 were detected by Western blotting with anti-Fra-1 (1:1000) and anti-HA (1:1000) antibodies, respectively.

MCF-7 cells were grown to 70% confluence in six-well plates. The cells were incubated for 15 min with a mixture of 5 ng of pRLCMV, 0.5 μg of the indicated Akt construct (or its empty vector, pcDNA3.1), and 0.5 μg of pFra-luc construct and Fugene6 reagent (Roche Molecular Biochemicals) using a ratio of 3 μl of Fugene6 reagent per microgram of plasmid DNA in serum-free media, according to the manufacturer's instructions. After 48 h of transfection, the cells were washed twice in ice-cold PBS, extracted in passive lysis buffer (Promega) and assayed sequentially for the firefly luciferase and Renilla luciferase activities, according to the manufacturer's instructions.

Nuclear extracts were prepared from PC3 cells as described (48). Electrophoretic mobility shift assays (EMSAs) were performed using DIG gel shift kit (Roche), according to the manufacturer's instructions. Binding reactions contained 4 μl binding buffer (Roche), 5 μg nuclear extracts, 0.4 ng of digoxigenin-labeled probe, 1 μg poly(dI·dc), 1 μg poly l-lysine in a total reaction volume of 20 μl. Oligonucleotides, RCE1 (GCGCGTCTCGGGGGTGGAGCCTGGAGG), its mutant (GCGCGTCTCGGCGCTCGAGCCTGGAGG), and RCE2 (GCAACGCCCCCACCCCCCGCGGTCGCA) were labeled at the 3′ end with digoxigenin-11-ddUTP. For competition experiments, 125-fold molar excess of unlabeled oligonucleotide was added to the reaction mixture before the addition of the probe. For super shift assays, anti-Sp1 antibody was pre-incubated with the nuclear extracts for 1 h at 4°C before the initiation of the binding reaction. Mobility shift reactions were resolved on 6% nondenaturing polyacrylamide gels for 3 h at 100 V. Following the electrophoretic separation, the complexes were transferred to nylon membrane by contact blotting and visualized by enzyme immunoassay using anti-Digoxigenin-alkaline phosphatase antibody and chemiluminescent alkaline phosphatase substrate.

We thank Dr. Daisy De Leon (Loma Linda University) for the MCF-7 cells used in these studies, Dr. William Greenlee for the CYP1B1 reporter construct, and Dr. Jon Horowitz for Sp1 constructs.