INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The Wingless/Wnt pathway plays a crucial role in development and cell cycle control (Cadigan and Nusse, 1997; Huelsken and Behrens, 2000). Dysregulation of the Wingless/(Wnt)/-catenin/Tcf pathway has been implicated in tumorigenesis of diverse types (Polakis, 2000a). Axin/Conductin, together with APC, promote -catenin degradation through serine-threonine phosphorylation of the -catenin N-terminus by GSK3, which targets -catenin for ubiquitination by a SCF^-TRCP (-transducin repeat-containing protein) ubiquitin ligase complex (Fuchs et al., 1999; Winston et al., 1999) and its degradation by the proteasome. On induction of Wnt signaling by extracellular ligands, the Frizzled receptors are activated. The activity of GSK3 and its effect on -catenin is antagonized by Dishevelled, a downstream target of Frizzled, thus preventing the degradation of -catenin by the proteasome. The resulting accumulation of -catenin leads to its nuclear translocation and binding to Tcf/Lef transcription factors to induce target genes including cyclin D1 and c-Myc (He et al., 1998; Shtutman et al., 1999; Huelsken and Behrens, 2000).

In addition to components in the Wnt signaling pathway, several other pathways can regulate -catenin/Tcf signaling and gene expression and confer aberrant cellular growth. The protein encoded by Gas6, a growth factor of the vitamin K-dependent family, which binds members of the Axl receptor tyrosine kinase family, stabilizes -catenin, and induces Tcf signaling (Goruppi et al., 2001). Hepatocyte growth factor/scatter factor (Papkoff and Aikawa, 1998) and oncogenic mutations of RON and MET (Danilkovitch-Miagkova et al., 2001) can also increase cytosolic -catenin and activate Lef/Tcf-responsive reporters. The Xenopus wnt target gene twin is induced by SMAD4 through the -catenin/Tcf complex (Nishita et al., 2000). Conversely, genotoxic stress reduces -catenin abundance in part through p53 signaling and a Siah1/Skp1/Ebi complex, which binds the -catenin N-terminus independently of its GSK3 phosphorylation sites (Liu et al., 2001; Matsuzawa and Reed, 2001).

The c-myc and cyclin D1 genes that encode important regulators of cell proliferation have been identified as transcriptional targets of -catenin (He et al., 1998; Shtutman et al., 1999; Tetsu and McCormick, 1999). Transcription of the cyclin D1 gene is induced through distinct DNA sequences in the promoter by diverse mitogenic and oncogenic signaling pathways including activating mutants of Ras, Src, Stat3, Stat5, and ErbB-2 (Albanese et al., 1995; Bromberg et al., 1999; Matsumura et al., 1999; Pestell et al., 1999; Lee et al., 2000). Distinct binding sites within the cyclin D1 promoter have been characterized for transcription factors including CREB and AP-1 proteins (Albanese et al., 1995; Watanabe et al., 1996a, 1996b; Brown et al., 1998), and a single site at 81 has shown to bind -catenin/Tcf proteins (Shtutman et al., 1999). Although Tcf/Lef proteins can function as either enhancer or repressor elements (Bienz, 1998; Barker et al., 2000), the Tcf binding site of the cyclin D1 promoter at 81 functioned as an enhancer element that conveyed activation of the cyclin D1 promoter by components of the Wnt/-catenin pathway (Shtutman et al., 1999; D'Amico et al., 2000; Lin et al., 2000; Sampson et al., 2001; Soriano et al., 2001). The cyclin D1 gene encodes a regulatory subunit of the holoenzyme that phosphorylates and inactivates the retinoblastoma (pRB) protein. Homozygous deletion of the cyclin D1 gene in mice demonstrated a requirement for cyclin D1 in normal mammary gland development during pregnancy and mouse embryo fibroblasts (MEFs) derived from the cyclin D1^/ animals have both defective induction of DNA synthesis and enhanced cellular apoptosis rates (Fantl et al., 1995; Sicinski et al., 1995; Albanese et al., 1999; Fantl et al., 1999). Cyclin D1 overexpression can enhance DNA synthesis, is required for transformation and contact-independent growth in several cell types and has been implicated in several human cancers including breast, colon, and prostate (Shtutman et al., 1999; Tetsu and McCormick, 1999; Lee et al., 2000). Thus, cyclin D1 plays an important role in tumorigenesis and cell cycle control.

The IB/NF-B pathway is another pathway involved in both cell cycle control and inflammation and has recently been implicated in cancer (Karin and Delhase, 2000; Yamamoto and Gaynor, 2001). The NF-B transcriptional activity is normally inhibited by IB proteins that sequester it in the cytoplasm (Karin and Delhase, 2000; Joyce et al., 2001). The IB kinase complex (IKK) that phosphorylates IB contains two functionally distinct kinases, IKK and IKK. IKK plays a dominant role in NF-B regulation by TNF- and IL-1 (Delhase et al., 1999; Li et al., 1999a). In contrast, IKK is required for murine skeletal and keratinocyte differentiation (Li et al., 1999a; Takeda et al., 1999; Hu et al., 2001). IKK cannot compensate for the loss of IKK (Li et al., 1999a), suggesting that distinct targets are regulated by IKK and IKK. Although the IB/NF-B and Wnt/-catenin/Tcf pathways are independent signaling pathways, both IB and -catenin are regulated by phosphorylation at similar consensus N-terminal serines and are targeted for ubiquitination by a similar SCF^-TrCP complex followed by proteasomal degradation. The consequences of this regulation are, however, very different (Fuchs et al., 1999; Winston et al., 1999). Thus, although the SCF^-TrCP-mediated degradation of IB leads to the induction of NF-B activity, the SCF^-TrCP-mediated degradation of -catenin inhibits the activity of the Wnt pathway. In addition, although GSK3 contributes to the degradation of -catenin and represses -catenin/Tcf signaling, the activity of NF-B is enhanced by GSK3 (Hoeflich et al., 2000; Polakis, 2000a).

The IKK complex is regulated by several IKK kinases including the NF-B inducing kinase (NIK), TAK1, MEKK1, Cot/TPL2, and NAK, which coordinate physiological responses to distinct stimuli (Joyce et al., 2001). NF-B activity is also enhanced by the serine threonine kinase Akt (Madrid et al., 2000; Romashkova and Makarov, 1999) that is known to induce cellular proliferation and survival (Datta et al., 1999) in response to PI3K activation (Franke et al., 1997; Klippel et al., 1998). Akt is recruited to IKK by stimulation with growth factors, but not by TNF-. Akt activation by PI3K is inhibited by the tumor suppressor PTEN, a D3 phosphoinositide phosphatase that induces G₁ arrest in prostate cancer cells (Ramaswamy et al., 1999), consistent with both a role for PTEN as a prostate cancer cell tumor suppressor and a role of PI3K-Akt activation in cell cycle progression (Di Cristofano et al., 2001). The Gas6-dependent proliferation and activation of Tcf is also dependent on PI3K (Goruppi et al., 2001), suggesting a role for PI3K signaling in the regulation of -catenin/Tcf signaling. The components of the cell cycle machinery that are regulated by IKK and are required for PI3K-dependent cellular proliferation, however, remain to be determined.

Here we show a novel role for IKK in mitogenic signaling through transcriptional induction of the cyclin D1 gene. We show that the serum induction of cyclin D1 and G₁-S phase progression is PI3K-dependent and that cells lacking cyclin D1 show a reduction in PI3K-dependent S-phase entry. PI3K-dependent induction of cyclin D1 was blocked by an inhibitor of IKK and activation of IKK-induced cyclin D1. PI3K induction of cyclin D1 was inhibited by a dominant negative Tcf, and a single Tcf site in the cyclin D1 promoter was required for its induction by IKK and PI3K. Mouse embryo fibroblasts derived from mice lacking IKK showed reduced phosphorylation of -catenin and reduced Tcf and cyclin D1 abundance and promoter activity. We had previously shown that IKK exists in a complex with endogenous -catenin (Lamberti et al., 2001). Herein we show that purified IKK was sufficient for phosphorylation of -catenin through its N-terminus in vitro, demonstrating that IKK can function as a kinase independently of its heterodimeric partners. Because IKK but not IKK induced cyclin D1 expresion and Tcf activity, these studies indicate that the relative levels of IKK and IKK may alter their substrate and signaling specificities to regulate DNA synthesis through distinct mechanisms.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Construction of Reporter Genes and Expression Vectors

The human cyclin D1 promoter fragments linked to the luciferase reporter gene in the pA₃LUC vector promoters of the c-fos gene (c-fosLUC), TOP-FLASH, FOP-FLASH, cyclinELUC, cyclinALUC, c-MycLUC, Engrailed 2 promoter (EngrLUC), 3xRelLUC, and pGL₃LUC (Promega, Madison, WI) were previously described (He et al., 1998; Joyce et al., 1999; McGrew et al., 1999; D'Amico et al., 2000; Lee et al., 2000). The expression vectors for p110-K227E, p110-CAAX (Matsumura et al., 1999), the p110-kinase dead, the p85, p85iSH2-N, 85iSH2-C, p85bBCR were kind gifts from Dr. J. Downward (Rodriguez-Viciana et al., 1997); pCMV-c-Akt wt, Akt-K179 M, Akt-T308A, were from Dr. A. Bellacosa; and CMV-IB (Super-repressor) [CMV-IB (Sr)] was a gift from Dr. D. Ballard (Brockman et al., 1995). Mammalian expression vectors for IKK (S176/180E and A) and IKK (S177/181E and A) mutants were provided by Dr. F. Mercurio and for IKK (K54 M) and IKK (K44A) were provided by Tularik Inc (South San Francisco, CA).

Reporter Assays, Cell Culture, and Chemicals

Cell culture and DNA transfection were performed exactly as previously described (Lipofectamine Plus; Life Technologies BRL, Rockville, MD; DiDonato et al., 1997; Zandi et al., 1997). Transfections were normalized using RSV--gal unless otherwise indicated (DiDonato et al., 1997; Zandi et al., 1997). The effect of an expression vector was compared with the effect of an equal amount of vector cassette. The DU145 cells were maintained in DMEM with 10% (vol/vol) calf serum and 1% penicillin/streptomycin. SW480 colon cancer cells and Cos-7 kidney cells were grown in DMEM (5% fetal bovine serum). The IKK^/ mouse embryo fibroblasts (MEFS) and 3T3 cells were a generous gift from Dr. M. Karin. Cells were plated at ~100,000 cells/well in 12-well plates. After 24 h, cells were transfected with the indicated DNA and a Renilla luciferase reporter as an internal control for transfection efficiency. All transfections were done at least in triplicate and were repeated at least three times. Treatments with the PI 3-kinase inhibitor LY294002, the MEK inhibitor PD098059 (10-20 µM), the p38 MAP kinase inhibitor SB203580 (10-20 µM), wortmannin (2, 5, 10 µM) were performed for 24 h, and results were compared with vehicle treatment. Luciferase assays were performed at room temperature using an AutoLumat LB 953 (EG&G Berthold, Natick, MA). Luciferase content was measured by calculating the light emitted during the initial 10 s of the reaction, and the values are expressed in arbitrary light units. Statistical analyses were performed using the Mann Whitney U test with significant differences established as p < 0.05. To select transfected cells, cotransfection experiments were conducted using magnetic separation of transfected cells using CD4 as the marker and the magnetic-activated cell separation system (MACS; Ashton et al., 1999).

Western Blots and Cell Cycle Analysis

Western blotting was performed with antibodies directed to cyclin D1 (DCS-6; NeoMarkers, Fremont, CA), TFIIB (Transduction Laboratories, Lexington, KY), IKK (mAb was from PharMingen, San Diego, CA), IKK, (polyclonal SC7182, Santa Cruz Biotechnology, Santa Cruz, CA) IKK, (polyclonal SC7607, Santa Cruz Biotechnology), -catenin (Transduction Laboratories), phospho--catenin (Cell Signaling, Beverly, MA), Flag, (M2, Sigma Chemical Co., St. Louis, MO) and HA (12CA5, Sigma). Cell homogenates (50 µg) were electrophoresed in an SDS-12% polyacrylamide gel and transferred electrophoretically to a nitrocellulose membrane (Micron Separations Inc., Westborough, MA). After transfer, the gel was stained with Coomassie blue as a control for blotting efficiency. The blotting membrane was incubated for 2 h at 25°C in T-PBS buffer supplemented with 5% (wt/vol) dry milk to block nonspecific binding sites. After a 6-h incubation with primary antibody at a 1:1000 dilution (cyclin D1) or 1:2500 (-tubulin) in T-PBS buffer containing 0.05% (vol/vol) Tween 20, the membrane was washed with the same buffer. For detection of cyclin D1 the membrane was incubated with goat anti-mouse horseradish peroxidase second antibody (Santa Cruz Biotechnology) and washed again. Immunoreactive proteins were visualized by the enhanced chemiluminescence system (Kirkegaard and Perry Laboratories, Gaithersburg, MD). Annexin V staining for apoptosis (Albanese et al., 1999) and cell cycle analysis were performed by flow cytometric analyses using a fluorescence-activated cell sorter (FACStar plus; Becton Dickinson & Co., Lincoln Park, NJ).

In Vitro Kinase Assays

Kinase assays were performed as described (Yamamoto et al., 2000). The baculovirus-produced IKK protein was purified by nickel-agarose chromatography and then immunoprecipitated with 12CA5 mAb (Yamamoto et al., 2000). IKK was added to kinase buffer containing 10 µCi of [-³²P], 1 mM ATP, 1 mM dithiothreitol, 5 mM MgCl₂, 100 mM NaCl, 50 mM Tris-HCl, pH 8.0, and then 1 µg of each of the substrates including GST-IB (1-54) or GST--cat constructs (Lamberti et al., 2001) was incubated for 15 min at 30°C. Reactions were incubated at 30°C for 30 min and stopped by the addition of protein loading buffer and heating to 90°C and SDS-PAGE and autoradiography.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

PI3K-induction of Cyclin D1 Requires the Tcf Binding Site Activation of phosphatidyl inositol 3'-kinase (PI3K) mediates signaling induced by a number of growth factors and tumor promoters and is required for mitogenic stimulation by specific growth factors during the G₁-S phase of the cell cycle (Klippel et al., 1998; Vanhaesebroeck and Waterfield, 1999). The role of PI3K in serum-induced cyclin D1 expression was examined in mouse embryo fibroblasts (MEFs). In wild-type (wt) MEFs, cyclin D1 protein levels were elevated by 3 h after serum stimulation, and the PI3K inhibitor LY294002 abrogated the induction (Figure 1A). Total ERK levels were unchanged under these conditions in both wt and Cyclin D1^/ MEFs (Figure 1A). Activity of the full-length human cyclin D1 promoter linked to a luciferase reporter gene was induced 2.5-fold by serum addition. The PI3K inhibitor reduced serum-induced activation of the cyclin D1 promoter by 80% (Figure 1B). Activation of PI3K and Akt plays a key role in DNA synthesis in prostate cancer cells (Ramaswamy et al., 1999; Di Cristofano et al., 2001). We therefore examined the role of PI3K in the PTEN containing prostate cancer cell line DU145. Because PI3K plays a role in signaling by diverse growth factors, including Gas6 in density-arrested cells (Goruppi et al., 2001), we examined the regulation of cyclin D1 by PI3K in density-arrested cells. The cyclin D1 promoter (1745 CD1LUC) was induced 10-fold by p110-CAAX compared with the empty vector (Figure 1C). In low-confluence cells the cyclin D1 promoter was induced significantly by p110-CAAX in either high (Figure 1C) or low serum conditions (2.3-fold ± 0.18, n = 11, p < 0.01; Figure 1D). The kinase dead mutant (p110-CAAX KD) did not affect cyclin D1 promoter activity, and the constitutively active p110-K227E mutant induced cyclin D1 2.2-fold (Figure 1D). In contrast with the cyclin D1 promoter, the cyclin E and cyclin A promoters were not induced by p110-CAAX (Figure 1E), suggesting that the induction of cyclin D1 is not an indirect effect of PI3K activity on DNA synthesis and the effect of p110-CAAX is promoter specific. Because cryptic activation sequences, including AP-1, have been identified in several expression vectors, we examined the empty luciferase reporter pA₃LUC in which the cyclin D1 promoter was cloned and found that pA₃LUC was not induced (Figure 1E) in contrast with pGL₃LUC, which was induced threefold by p110-CAAX (Amanatullah et al., 2001). Cyclin D1 promoter activation by PI3K was reduced by the chemical inhibitor LY294002 (Figure 1F) or Wortmannin (our unpublished results). Type 1 PI3K is a heterodimeric holoenzyme, consisting of a regulatory (p85) and a catalytic (p110) subunit, which was initially identified through its role in Src-mediated transformation. p110-CAAX induction of cyclin D1 promoter activity was reduced by the previously described dominant inhibitory mutants of the PI3K regulatory subunit (Rodriguez-Viciana et al., 1997; Figure 1G).

View larger version (27K): [in this window] [in a new window]   Figure 1.   PI3K-induction of cyclin D1. (A) Western blot analysis of MEFs for cyclin D1 derived from wild-type (cyclin D1 wt) or cyclin D1/ mice and treated with serum either with or without the PI3K inhibitor LY294002. The Western blot was probed for cyclin D1 and total ERK. (B) The serum-induced activity of the cyclin D1 promoter in the presence or absence of the PI3 kinases inhibitor LY294002 (20 µM). (C) DU145 cells at either >90% or (D) 30% confluence were transfected with a cyclin D1 promoter luciferase reporter plasmid (1745CD1LUC) and either the p110-CAAX, or (D) the p110 kinase dead mutant (p110-CAAX-KD) or the constitutively active p110-K227E mutant expression plasmid in the presence of either 10% or 0.5% serum. The fold induction of the luciferase reporter activity is shown for nine separate experiments as mean ± SEM throughout. (E) The effect of p110-CAAX on reporter plasmids for cyclin A and the cyclin E promoter, and the luciferase reporter pA3LUC. (F) The p110-CAAX induction of the cyclin D1 promoter activity was inhibited by LY294002 (using 2, 20, and 100 µM). (G) The cyclin D1 promoter activity in the presence of p110-CAAX is shown as 100% and is compared with the effect of cotransfected dominant negative inhibitors of PI3K including p85, p85iSH2-N, 85iSH2-C, or p85BCR (Rodriguez-Viciana et al., 1997). The results are shown compared with equal amounts of empty control vector for each expression vector plasmid.

View larger version (27K): [in this window] [in a new window]   Figure 2.   PI3K-induction of cyclin D1 requires the Tcf site and is dependent upon IB. (A) The effects of the p110-CAAX expression plasmid on the activity of the cyclin D1 promoter (1745 CD1LUC) or of a point mutant of the Tcf site at 81 (1745 Tcf mut) in DU145 cells grown to either > 90% or (B) 30% confluence. Regulation of the cyclin D1 and Tcf-responsive (TOPFLASH) and mutant (FOPFLASH) reporter constructs by either the p110-CAAX plasmid or (C) the activated -catenin point mutant (Y33). The fold induction of the luciferase reporter activity is shown for at least nine separate experiments as mean ± SEM throughout. (D) The cyclin D1 promoter activity in the presence of p110-CAAX is shown as 100% and is compared with the effect of cotransfected dominant negative inhibitors of PI3K including RacN17, Akt wt, AktDN (K179 M), or IBSr. The results are shown compared with equal amounts of empty control vector for each expression vector plasmid. (E) Point mutations of the cyclin D1 promoter Tcf or CRE site were compared with the basal promoter activity of 1745 CD1LUC. The activity of the wild-type promoter construction was set as 1.0. The data are mean ± SEM of five separate transfections. (F) The cyclin D1 promoter activity in the presence of p110-CAAX (100%) is compared with the effect of dominant negative or wild-type Tcf.

Cyclin D1 Is Required for PI3K-dependent S-Phase Entry in Primary Cells

The current studies suggest cyclin D1 is a distal target of PI3K in serum-induced DNA synthesis. Cyclin D1 is known to play a role in the entry of cells into the DNA synthetic (S) phase induced by several growth factors and mitogens. The role of PI3K in serum-induced DNA synthesis through cyclin D1 is not known and was therefore further examined. In wt MEFs, serum-induced entry into S phase, increased from 10 to 26% (Figure 3A). LY294002 reduced the S-phase proportion from 26 to 7% at 24 h, indicating that serum-induced DNA synthesis is substantially PI3K dependent in MEFs (Figure 3, A and B). LY294002 treatment reduced serum-induced DNA synthesis by a mean of 39% at 12 h after serum addition but did not affect the serum-induced entry into the S-phase fraction in the cyclin D1^/ MEFs (Figure 3C, mean for n = 4 separate experiments). To confirm that LY294002 was effective at inhibiting signaling downstream of PI3K in both the cyclin D1 wt and cyclin D1^/ MEFs, western blotting was performed for phosphorylated Akt using a specific antibody, and the membrane was probed for total ERK as a control (Figure 3D). Serum-induced phosphorylation of Akt was reduced by LY294002 in both cell types (Figure 3D). Similar analyses of serum-induced DNA synthesis were performed in 3T3 cells derived from the cyclin D1^+/+ and cyclin D1^/ MEFs with similar results (our unpublished results). To determine the role of PI3K in apoptosis mediated by serum deprivation, annexin V staining and sub G₁ analysis was performed on the MEFs. Cyclin D1^/ MEFs exhibited a fivefold greater level of annexin V staining compared with wt MEFs, indicating increased basal apoptosis as previously shown (Albanese et al., 1999) that was rescued by serum (Figure 3E). LY294002 did not affect the level of apoptosis in either wt or cyclin D1^/ MEFs as determined by either annexin V staining (Figure 3E) or sub G₁ analysis (our unpublished results). These studies suggest that a substantial component of serum-induced expression of cyclin D1 is PI3K dependent and that MEFs derived from animals deleted of the cyclin D1 gene show reduced PI3K-dependent induction of DNA synthesis.

View larger version (40K): [in this window] [in a new window]   Figure 3.   Involvement of cyclin D1 in PI3K-dependent S-phase entry in primary cells. (A) MEFs were treated either with vehicle (DMSO) or LY294002 (20 µM) and DNA synthesis assessed by FACS. (B) FACS analysis of wt MEFs in the presence and absence of LY294002 (LY; 20 µM) 12 and 24 h after serum stimulation. (C) The effects of LY294002 on S-phase wt and cyclin D1/ MEFs are shown after serum stimulation as mean of four separate experiments. (D) Western blotting for phosphorylated Akt or total ERK of MEFs treated with serum and either vehicle or LY294002. (E) The level of apoptosis was determined by Annexin V staining in serum-starved-stimulated cyclin D1wt and cyclin D1/ MEFs.

IKK, but not IKK Induces Cyclin D1 through -Catenin/Tcf

The studies described above indicate that the PI3K activation of cyclin D1 involves Akt and IB (Ozes et al., 1999). As IKKs regulate IB activity, we assessed the role of IKKs in PI3K-dependent activation of cyclin D1 using previously characterized dominant negative IKK mutants (Delhase et al., 1999). These expression vectors behaved as previously described in cultured cells (below). We found that both the dominant negative and kinase dead IKK constructs reduced PI3K-induced cyclin D1 promoter activity and the basal promoter activity in a dose-dependent manner (Figure 4A). The constitutively active mutant IKKCA(S176/180E) induced the cyclin D1 promoter 4.2-fold (Figure 4B). The IKKCA expression vector was previously well characterized and was shown to integrate in the IKK kinase using the identical transfection approach (DiDonato et al., 1997; Zandi et al., 1997). In contrast with IKK, the constitutively active IKK mutant (IKKCA) decreased the cyclin D1 promoter activity (see below). Using a series of 5' promoter deletion constructions the IKK responsiveness was confined to the proximal 163 base pairs (our unpublished results). Mutation of the Tcf site in the context of the 1745-base pair fragment abolished induction of cyclin D1 by IKKCA (Figure 4C). IKKCA induced TOP-FLASH threefold but did not induce a reporter with mutations of the Tcf site (FOP-LUC; Figure 4C). IKKCA also activated the canonical NF-B-responsive sequences (3xRelLUC) to the same extent (Figure 4C). Consistent with previous studies, in which PI3K and Akt induced NFB activity in response to IL-1 (Madrid et al., 2000, 2001), the IKK kinase dead and dominant negative mutants reduced the activity of the NFB-responsive reporter gene 3XRelLUC in the presence of p110-CAAX (Figure 4D).

View larger version (31K): [in this window] [in a new window]   Figure 4.   IKK induces the cyclin D1 gene through the -catenin/Tcf site. (A) DU145 cells were transfected with the cyclin D1 promoter luciferase reporter plasmid (1745 CD1LUC) and the p110-CAAX, with the dominant inhibitors (IKKKM(K54 M), IKKDN(S176/180A)) or (B) constitutively active plasmid (IKKCA(S176/180E) or IKK(S177/181E)). (C) Identification of the IKK-responsive sequences in the cyclin D1 promoter. The IKKCA(S176/180E) expression plasmid was coexpressed with the luciferase reporters shown and fold induction determined compared with equal amounts of empty expression vector cassette. (D) Effect of IKK kinase dead (IKKK54 M) and dominant negative mutants (IKK(S176/180A)) on p110-CAAX induced NF-B activity assessed using the 3XREL LUC reporter.

To provide genetic evidence for the involvement of IKK activity in regulating cyclin D1, MEFs from IKK^/ mice were selected by the 3T3 protocol. Cells were serum starved for 24 h and western blotting was performed to determine cyclin D1 levels. Immunostaining for IKK showed the presence of IKK in the wt 3T3 and the absence of staining in the IKK^/ 3T3 cells (Figure 5A). We found that cyclin D1 abundance was reduced by 85% in the IKK^/ cells (Figure 5A), and the activity of the cyclin D1 promoter in the IKK^/ cells was lower by 67% compared with IKK^+/+ cells (Figure 5B). Serum treatment induced cyclin D1 abundance in wt MEFs by two- to threefold after 3 h, whereas in the IKK^/ cells, induction was delayed until 6 h after serum stimulation (Figure 5C), suggesting a role for IKK in both the basal level of cyclin D1 expression and in serum-induced cyclin D1 abundance. Because IKK induced cyclin D1 through the Tcf site and serum-induction of cyclin D1 protein abundance was defective in the IKK^/ cells, we assessed the role of the cyclin D1 promoter Tcf site in serum-induced activity. In wt 3T3 cells, serum-induced activation of the cyclin D1 promoter was reduced more than 90% by mutation of the Tcf site (Figure 5D). Furthermore, serum-induced activation of the cyclin D1 promoter was defective in the IKK^/ cells (Figure 5D).

View larger version (34K): [in this window] [in a new window]   Figure 5.   Reduced mitogen-induced cyclin D1 expression in IKK/ cells involves Tcf signaling. (A) 3T3 cells from either wt or IKK/ mice were examined by Western blotting for cyclin D1, using equal amounts of total protein. Immunostaining for IKK in the wt or IKK/ 3T3 cells. (B) The cyclin D1 promoter luciferase reporter plasmid (1745 CD1LUC) was transfected into either wt or IKK/ 3T3 cells along with the -galactosidase control reporter. Relative cyclin D1 promoter activity is shown as mean ± SEM for n = 3. (C) Western blotting for cyclin D1 of wt or IKK/ 3T3 cells treated with serum for the time points indicated. The fold change in cyclin D1 protein levels is shown normalized to GDI as a loading control. The data is representative of three separate experiments. (D) The cyclin D1 promoter (1745CD1LUC) or the point mutant of the Tcf site (1745Tcfmut) were compared for relative activity in wt or IKK/3T3 cells. The data are mean ± SEM, n = 9.

Consistent with the reduced abundance of cyclin D1 in the IKK^/ 3T3 cells and the ability of cyclin D1 overexpression to promote DNA synthesis in fibroblasts (Pagano et al., 1994), serum-induced DNA synthesis was reduced in IKK^/ 3T3 cells (Figure 6A). To determine whether the reduction in IKK abundance was important in the reduced levels of cyclin D1, the IKK^/ 3T3 cells were transfected with the IKKCA expression vector and MACS-sorted, and the cell extracts were subjected to Western blotting. IKK protein levels were increased in the IKKCA-transfected IKK^/ 3T3 cells (Figure 6B, lane 3). Although the relative levels of IKK in the IKK^/ 3T3 cells transfected with the IKKCA expression vector were substantially less than the wt 3T3 cells, cyclin D1 levels were increased threefold compared with the IKK^/ 3T3 cells, demonstrating a key role for IKK in inducing cyclin D1 levels. Activity of the cyclin D1 promoter was also increased threefold in IKK^/ 3T3 cells transfected with the IKKCA expression vector. Furthermore, the induction of cyclin D1 by IKKCA required the Tcf site (Figure 6C).

View larger version (55K): [in this window] [in a new window]   Figure 6.   IKK regulates mitogen-induced DNA synthesis and is required for Tcf signaling to natural Tcf-responsive genes. (A) The serum-induced S-phase fraction of wt or IKK/ 3T3 cells was compared. The data are mean ± SEM, N = 7. (B) Western blotting of wt or IKK/ 3T3 cells transfected with an expression vector for IKKCA or empty expression vector. IKK and cyclin D1 immunoblotting is shown with GDI as an internal control for loading. (C) Relative activity of the cyclin D1 promoter or the corresponding Tcf point mutant in IKK/ 3T3 cells transfected with either IKK expression vector or the control vector. The data are mean ± SEM, n = 8. (D) Relative activity of the c-Myc and Engr promoters (n = 6) in randomly cycling wt or IKK/ 3T3 cells. The activity of the promoter is set as 1 in the IKK/ 3T3 cells. (E) Activity of the Engr promoter in the presence of serum stimulation. The data is mean ± SEM of n = 12 separate experiments. (F) Nuclear and cytosolic fractions of Cos-7 cells were analyzed by Western blotting for the abundance of IKK or IKK in the nuclear (NE) and cytosolic (S100) fractions. Internal controls for (nuclear; TFIIB) and cytoplasmic (actin) markers are shown. Substantially more IKK than IKK was found in the nuclear extracts (NE) of Cos-7 cells. (G) Immunostaining for IKK and IKK. IKK is predominantly extranuclear, whereas IKK was found in both nuclear and cytoplasmic compartments.

To determine whether the activity of other known -catenin responsive promoters were regulated by IKK, the relative activity of the c-Myc (He et al., 1998) and Engrailed (McGrew et al., 1999) promoters were compared in the wt and IKK^/ 3T3 cells, with relative activity normalized to an internal control of renilla luciferase activity. The relative activity of the Engr and c-Myc promoter activity was reduced 10- to 12-fold in the IKK^/ 3T3 cells (Figure 6D). Furthermore, as with the cyclin D1 promoter, the serum-induced activity of the Engr promoter was substantially reduced in the IKK3T3 cells (Figure 6D). Together, these studies demonstrate that the activity of both heterologous and natural Tcf responsive genes is dependent on IKK in vivo. Furthermore, these studies demonstrate an important role for IKK in regulating the kinetics of serum-induced expression of -catenin/Tcf-responsive genes.

IKK Associates with and Phosphorylates -Catenin and Increases -Catenin Abundance

In addition to the differences in Tcf-mediated activation of gene promoters, several lines of evidence suggest that IKK and IKK fulfill distinct cellular functions. Thus, homozygous deletion of the IKK and IKK genes results in distinct phenotypes (Hu et al., 1999, 2001; Li et al., 1999a; Takeda et al., 1999), and IKK acts more potently on IB proteins and plays a more significant role in the NF-B pathway in response to activation with TNF- and IL-1 than IKK (Delhase et al., 1999; Li et al., 1999a, 1999b). To investigate further the basis for these diverse functions, we determined the subcellular localization of IKK and IKK and their cell-type expression patterns. Western blot analysis of nuclear and cytoplasmic extracts showed a differential localization of IKK and IKK in Cos-7 cells with IKK present in both the nuclear and cytoplasmic fractions (marked by TFIIB and actin, respectively), whereas IKK was predominantly cytoplasmic (Figure 6A), consistent with the immunohistochemical analysis (Figure 6B).

Consistent with a role for IKK in regulating -catenin phosphorylation and/or abundance, the total level of wt -catenin and of a higher molecular weight form of -catenin were increased in cells coexpressing IKKCA and -catenin expression vectors (Figure 7A). Point mutation of -catenin at Ser33 to alanine abrogated the induction of the higher molecular weight form of -catenin (Yost et al., 1996). The abundance of the -catenin S37A mutant and the higher molecular weight form were also increased in cells transfected with IKKCA, suggesting a dominant role for S33 in the generation of the high molecular weight form. Because IKKCA induced cyclin D1 and Tcf reporter activity, we hypothesized that IKK may regulate -catenin abundance and/or phosphorylation. In our previous studies, IKK immunoprecipitation on fractionated Cos-7 cell extracts cotransfected with HA-tagged -catenin and FLAG-tagged IKK showed that -catenin is present in IKK immunoprecipitates and IKK was also present in -catenin immunoprecipitates (Lamberti et al., 2001). We had also demonstrated an association between endogenous -catenin and IKK by reciprocal IP-Western blotting of SW480 cell extracts (Lamberti et al., 2001). Consistent with these findings in cultured cells, we found that GST--catenin fusion proteins were efficient substrates for phosphorylation by IKK in vitro in which IKK was immunoprecipitated from cultured cells and used as the enzyme source (our unpublished results). The minimal region of -catenin sufficient for phosphorylation by immunoprecipitated IKK included the N-terminal portion of the molecule between aa 30 and 55 (our unpublished results). IKK bound to and phosphorylated -catenin in vitro with an efficiency that was similar to that of IB as recently shown (Lamberti et al., 2001).

View larger version (39K): [in this window] [in a new window]   Figure 7.   IKK phosphorylates -catenin and increases -catenin abundance. (A) The IKKCA(S/E) expression plasmid was coexpressed in cells transfected with either wild-type or mutants (S33A, S37A) of -catenin. Western blotting analysis showed an increase the total amount of -catenin, including a higher molecular weight form (upper arrow). The S33A-catenin shows no increase in the amount of the high molecular weight form. (B) IKK kinase assays conducted using baculovirus expressed purified IKK and the GST--catenin constructs as shown. Kinase activity (left panel) and the Coomassie stained gel for the substrate are shown. (C) Western blot analysis of IKK/ or IKKwt 3T3 cells with antibodies to phosphospecific -catenin, total -catenin, Tcf-4, and PCNA.

Because the IKK immunoprecipitation may coprecipitate other components of the IKK complex to phosphorylate -catenin, IKK was produced in baculovirus, purified, and used as the enzyme source in IKK kinase assays with -catenin as substrate (Figure 7B). Purified IKK was sufficient for phosphorylation of GST--catenin 1-400. Deletion of the N-terminus of -catenin (130-400) abolished phosphorylation by IKK, and the N-terminus from 1-91 was sufficient for phosphorylation by IKK (Figure 7B). To determine if the endogenous IKK is involved in the phosphorylation of -catenin, equal amounts of proteins from IKK^/ and wt MEFs were compared using an antiphospho--catenin antibody. The results shown in Figure 7C demonstrated that phosphorylated -catenin exists in wt MEFs but with a significantly reduced abundance in the IKK^/ cells. Interestingly, the levels of the nuclear effector of -catenin, Tcf were also lower in the IKK^/ cells. The abundance of the nuclear protein PCNA was similar between the IKK^/ and wt MEFs.

Our findings that IKK phosphorylates -catenin and that IKKCA increases Tcf activity and -catenin abundance suggests that -catenin phosphorylation by IKK may contribute to the regulation of -catenin-mediated Tcf-dependent gene transcription. The consequent induction of cyclin D1 by PI3K-IKK-Tcf signaling contributes to the induction of DNA synthesis.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This study demonstrates for the first time a requirement for IKK in response to mitogens and DNA synthesis and the induction thereby of cyclin D1 abundance and promoter activity through a -catenin/Tcf pathway. IKK selectively and directly induced cyclin D1 but not cyclin E or cyclin A. Reintroduction of IKK into IKK-deficient cells restored cyclin D1 expression and promoter activity in a Tcf-dependent manner. Using a dominant negative mutant of Tcf activity we showed that IKK induction of cyclin D1 requires -catenin/Tcf activity. IKK was shown to be a key genetic determinant of the activity of several other Tcf responsive genes (c-Myc, Engr, TcfLUC). IKK-deficient cells demonstrated a delayed induction of serum-induced DNA synthesis and a delayed induction of serum-induced activity of the cyclin D1 and Engr promoters. Together these studies indicate a key role for IKK in coordinating the kinetics of mitogen responsiveness to a subset of cellular targets. These studies are consistent with an evolving view that separate components of the IKK complex may subserve distinct functions to convey signal transduction specificity (Ghosh and Karin, 2002).

Serum induction of DNA synthesis and cyclin D1 expression was PI3K dependent, and cyclin D1 was required for the PI3K-dependent induction of DNA synthesis. PI3K-dependent, serum-induced DNA synthesis was substantially reduced in cyclin D1-deficient cells, indicating a key role for cyclin D1 in this signaling pathway. Although serum deprivation increased apoptosis in cyclin D1^/ MEFs, the inhibition of apoptosis by serum addition was not affected by PI3K inhibition, demonstrating distinct functions of cyclin D1 in PI3K-dependent proliferation versus apoptosis. Although the upstream effectors of IKK that contribute to the induction of -catenin remain to be identified, the current studies demonstrate that the PI3K-dependent induction of cyclin D1 involves IKK. PI3K is involved in a PDGF-regulated pathway that activates Akt, leading to an association with and activation of IKK in cultured cells (Romashkova and Makarov, 1999), which is consistent with a role for PI3K in activating a subset of IKK functions. Although IB-independent effects of Akt on NF-B have been reported (Madrid et al., 2000; Reddy et al., 2000) and IKK phosphorylation by Akt is not essential for IKK activation of NF-B signaling (Delhase and Karin, 2000), increasing evidence suggests IKK conveys important kinase-dependent and -independent functions. Because the dominant inhibitors of Akt, IKK, and Tcf reduced the induction of cyclin D1 by constitutively active PI3K mutants, it appears that PI3K may be an important upstream inducer of IKK in the context of -catenin/Tcf signaling.

The current studies identify the cyclin D1 Tcf site as the common target of activated PI3K, IKK, and -catenin and establish, using dominant negative mutants, a colinearity of these components to regulate cyclin D1 expression in cultured cells. Wnt family ligands and Frizzled family receptors define one important mechanism that can induce -catenin/Tcf signaling (Polakis, 2000a