The F-box Protein β-TrCp1/Fbw1a Interacts with p300 to Enhance β-Catenin Transcriptional Activity*

Hyperactivated β-catenin is a commonly found molecular abnormality in colon cancer, and its nuclear accumulation is thought to promote the expression of genes associated with cellular proliferation and transformation. The p300 transcriptional co-activator binds to β-catenin and facilitates transcription by recruiting chromatin remodeling complexes and general transcriptional apparatus. We have found that β-TrCp1/Fbw1a, a member of the Skp1/Cullin/Rbx1/F-box E3 ubiquitin ligase complex, binds directly to p300 and co-localizes with it to β-catenin target gene promoters. Our data show that Fbw1a, which normally targets β-catenin for degradation, works together with p300 to enhance the transcriptional activity of β-catenin, whereas other F-box/WD40 proteins do not. Fbw1a also cooperates with p300 to co-activate transcription by SMAD3, another Fbw1a ubiquitylation target, but not p53 or HIF-1α, which are substrates for other ubiquitin ligase complexes. These results suggest that, although Fbw1a is part of a negative feedback loop for controlling β-catenin levels in normal cells, its overexpression and binding to p300 may contribute to hyperactivated β-catenin transcriptional activity in colon cancer cells.

CBP and p300 are large multidomain proteins that can bind to and activate over 80 different transcription factors thereby affecting a wide spectrum of cellular growth, development, and differentiation pathways (8,9). Mounting evidence indicates that alterations in the normal function of p300 and CBP may underlie the transformation process in certain hematologic malignancies (10 -13) as well as in gastrointestinal and colon cancers (14 -16). Indeed, a recent study found that, similar to such factors as poorly differentiated histology, lymph node metastasis, and large tumor size, overexpression of p300 was an indicator of poor prognosis in colorectal cancer patients (17).
Several lines of evidence show that ubiquitylation machinery may cooperate with co-activator complexes to enhance transcription, and, in some cases, their recruitment to target gene promoters facilitates transcription factor degradation (18 -20). For example, the F-box protein Skp2 serves as a co-activator for c-Myc transcription in addition to its role in ubiquitylating and degrading Myc (21,22). Likewise, the ubiquitin-conjugating enzyme Rsp5 and ubiquitin ligase Ubch7 have both been shown to independently serve as co-activators for steroid hormone receptors, the latter occurring through its interaction with the p160 co-activator, SRC1 (23,24). Additionally, an interaction between the yeast transcription factor, Gal4, and the ubiquitin ligase GCN4 was found to enhance Gal4-mediated gene activation (25).
In this study, we found that p300 and CBP bind to ␤-TrCp1/ Fbw1a, the substrate recognition partner associated with the Skp1/Cullin/Rbx1/F-box (SCF) E3 ubiquitin ligase complex that normally targets ␤-catenin for proteasome-mediated degradation (26). We found that p300/CBP, Fbw1a, and ␤-catenin co-localized to the promoters of target genes and that Fbw1a activated ␤-catenin-mediated transcription. Collectively, the data presented here provide support for the emerging theme that the integration of transcriptional activation complexes with ubiquitination machinery is a fundamental mechanism for regulating gene transcription.
Phage Display Screening-p300 N-terminal and C-terminal fragments (residues 1-595 and 1929 -2261) and CBP N-terminal and C-terminal fragments (1-615 and 1970 -2292) were fused to GST, expressed in BL21 bacteria, and purified on glutathione beads (Amersham Biosciences) according to the manufacturer's instructions. The four p300 and CBP fragments were cleaved from the GST moiety with Precision Protease (Amersham Biosciences). The cleaved fragments were then chemically coupled to Dynal M270 carboxylic acid magnetic beads, according to the manufacturer's instructions. Target proteins, immobilized on beads, were blocked with 1 mg/ml bovine serum albumin in PBS. 10 l of the M13 phage library, (PhD-7, New England Biolabs) diluted in 100 l of TBS plus 0.1% Tween 20 (TBST) was added to each of the 4 samples for 1 h at room temperature, rotating tubes end over end. Samples were then washed 10 times with TBST, and bound phages were eluted with 0.2 M glycine, pH 2.2, for 10 min at room temperature. Eluted phage were neutralized with Tris-HCl (pH9.1), amplified for 5 h, and titered. 2 ϫ 10 10 amplified phage from round 1 panning were used as input for round 2 panning for each sample. Three total rounds of panning were performed as in round 1, except that the concentration of Tween in the TBST was increased from 0.1% to 0.5% in rounds 2 and 3. To increase the diversity of recovered peptides, individual phage from the second round were plaque-purified for sequencing with an M13 reverse primer (New England Biolabs).
Cell Culture, Coimmunoprecipitation, and Western Blotting-All cell lines used (HCT116, 293T, and HeLa) were grown in Dulbecco's modified Eagle's medium plus 10% fetal calf serum, in a 37°C, 10% CO 2 incubator. Transient transfections were performed with FuGENE 6 (Roche Applied Science) according to the manufacturer's instructions. Whole cell lysates were made in radioimmune precipitation assay buffer containing protease inhibitors (Complete, Roche Applied Science). Nuclear extracts were made by incubating cells in a hypotonic buffer (10 mM Tris, pH 7.6, 10 mM KCl, 1.5 mM MgCl 2 ), homogenizing with a glass tissue grinder, and centrifuging to isolate nuclear pellets. Nuclear pellets were then placed in a low salt (20 mM KCl) buffer and slowly moved to a high salt buffer (1.2 M KCl) to extract nuclear protein. Dialysis in BC100 buffer (200 mM Tris, 100 mM KCl, 10% glycerol) was performed to allow buffer exchange. Co-immunoprecipitation was performed for 4 h to overnight at 4°C with the indicated antibodies. Protein A/G beads, mouse IgG, and rabbit IgG were from Santa Cruz Biotechnology. For endogenous co-immunoprecipitation, the anti-p300 antibody RW128 (Upstate Biotechnology) was conjugated to protein G beads with the Seize X Protein G IP kit (Pierce) according to the manufacturer's instructions. Washes were performed with radioimmune precipitation assay buffer or PBST (PBS plus 0.5% Triton X-100). Samples were placed in SDS-containing loading dye, subjected to SDS-PAGE, and transferred to nitrocellulose for immunoblotting for Fbw1a (Invitrogen and Santa Cruz Biotechnology), Gal4DBD (Santa Cruz Biotechnology), p53 (Santa Cruz Biotechnology), and the HA tag (Upstate and Covance).
Luciferase Assays and Fbw1a Knockdown-Cells were seeded in 24-well plates and transfections were performed in triplicate wells. A total of 3 g of DNA per triplicate was transfected into cells with FuGENE 6 according to the manufacturer's instructions. Depending on the specific experiment, 0.7-1.2 g of firefly luciferase reporter, 0.05-0.2 g of Renilla luciferase, and between 1.2 and 1.8 g of CMV-promoter vectors were used per triplicate, per experiment. The total amount of CMV-based vectors (pcDNA3.1 empty vector plus pcDNA3.1-Fboxencoding) was always held constant in each experiment. After balancing for CMV levels, the total amount of DNA in each transfection was brought up to 3 g per triplicate by using pBS2. 48 h after transfection, cells were lysed and luciferase assays were performed with a dual luciferase assay system (Promega). Activities of the firefly luciferase reporters were normalized to that of the internal control vector, pCMV-Renilla-luciferase. A Veritas microplate luminometer (Turner Biosystems) was used to collect luciferase readings. Knockdown of Fbw1a was performed with a combination of two lentiviral short hairpin (sh) RNA constructs from OpenBiosystems and one Dharmacon siRNA sequence that had been adapted for shRNA lentiviral infection. Briefly, 293T cells were transfected with lentiviral packaging vectors and a vector containing either a non-silencing or shRNA sequence directed against Fbw1a. 48 and 72 h after transfection, equal volumes of filtered supernatants were added to 293T cells for infection with either non-silencing or Fbw1a knockdown viruses. Puromycin (2 g/ml) was used for selection and fluorescence from the GFP-containing shRNA vectors was used to confirm nearly 100% infection efficiency. 96 -120 h after infection, quantitative reverse transcription-PCR and immunoblotting were performed to confirm Fbw1a knockdown. RNA was extracted using TRIzol and mRNA levels were normalized to GAPDH after quantitative reverse transcription-PCR. Primer sequences for Fbw1a mRNA were fwd (5Ј-GCT GAA CTT GTG TGC AAG GA-3Ј) and rev (5Ј-TAC TGT CCC CAT CCT CTT CG-3Ј); axin2 mRNA, fwd (5Ј-AGT GTG AGG TCC ACG GAA AC-3Ј) and rev (5Ј-CTT CAC ACT GCG ATG CAT TT); cyclinD1 mRNA, fwd (5Ј-GCG GAG GAG AAC AAA CAG AT-3Ј) and rev (5Ј-TGA GGC GGT AGT AGG ACA GG-3Ј). Quantification of immunoblot bands was performed using the Photoshop histogram function. Absolute band intensity was determined using the mean and pixel values for each band; relative band intensity was determined by normalizing to GAPDH.

RESULTS
p300 Binds Peptides with Homology to F-box/WD40 Proteins-To identify peptides that target proteins for binding to p300 and CBP, we used fragments of p300 and CBP to screen a random 7-amino acid M13 phage-displayed peptide library. Four protein fragments were used as targets for the screen, two from p300 and the corresponding two from CBP (Fig. 1A). The N-terminal fragment of p300 encompassing amino acids 1-595 (and corresponding CBP fragment) includes the N terminus, CH1, and part of the KIX domains that are known binding sites for nuclear receptors, hypoxia-inducible factor 1 ␣ (HIF1␣), and p53, among other transcription factors (8, 30 -33). The C-terminal fragment (p300 amino acids 1929 -2261 or corresponding CBP fragment) includes the SID/IBiD domain and its flanking sequences, which contain known binding sites for the SMAD proteins, p53, ETS2, E1A, and the p160 co-activators SRC1, GRIP1, and AIB1 (34 -37). We did not include the CH3 domain within the C-terminal fragment, because CH3 and CH1 are structurally highly similar and several transcription factors bind redundantly to these two domains (32). After GST purification, the fragments were cleaved of their GST moieties (to avoid nonspecific peptide binding) and chemically coupled to magnetic beads for in solution library screening. In vitro interaction assays confirmed that coupling to magnetic beads did not alter the structure of the target proteins. As expected, immobilization of the N-terminal p300 and CBP fragments to FIGURE 1. Phage display screening isolates a collection of p300/CBP-binding peptides. A, structure of full-length p300 and CBP and their fragments, coupled to magnetic beads, for in solution phage display screening. B, six phage-displayed peptides that were isolated from panning the p300/CBP N-terminal fragments. Alignment reveals consensus motif of K(V/L)WXL. C, schematic of p300/CBP fragments used in peptide interaction assay and lane order for blots shown in D. D, in vitro peptide interaction assay of EGFP-KVWTLNY or EGFP alone and p300/CBP fragments fused to GST.

Fbw1a
Interacts with p300 to Enhance ␤-Catenin Activity MAY 8, 2009 • VOLUME 284 • NUMBER 19 magnetic beads conferred binding to in vitro translated HIF1␣ and coupling of the C-terminal p300 and CBP fragments to beads conferred binding to SRC1 (supplemental Fig. S1). In solution library screening was performed with an M13 phage library whose complexity of ϳ10 11 different peptides encompasses all possible permutations of peptides 7 amino acids in length. Three successive rounds of panning were performed to enrich for p300/CBP-binding peptides, however, to provide for diversity of recovered peptides, phage from the second round of panning were plaque-purified for sequencing. 6 of the 24 peptides sequenced from the screen of N-terminal p300 and CBP fragments contained the motif K(V/L)WXL (Fig. 1B), whereas none of the 27 peptides sequenced from the screen of C-terminal fragments contained it (supplemental Fig. S1 B), indicating a significant enrichment for binding of this motif to the N-terminal fragments (p ϭ 0.007, Fisher's exact test).
Although other motifs emerged in the collection of sequenced peptides (supplemental Fig. S1B), none of these occurred as often as K(V/L)WXL. Of note, the proline-rich motif P(L/R)XXP appeared 3-4 times in the C-terminal fragment screen. These peptides are very similar to the previously characterized proline-repeat motifs in the p53 transactivation domain, which are known to bind p300/CBP and are important for DNA-dependent acetylation of p53 (38). Given the prominence of the K(V/L)WXL motif in our collection of peptides from the N-terminal fragment screen, we chose to focus on elucidating the significance of this motif.
To verify that the K(V/L)WXL peptide was sufficient to direct binding to p300 and CBP, we cloned one of the identified peptides (KVWTLNY) (Fig. 1B) into a solution-exposed loop of the enhanced green fluorescent protein (EGFP). In vitro interaction assays demonstrated that insertion of this peptide conferred binding of EGFP to the N-terminal fragments of both p300 and CBP, as well as a smaller fragment containing only the CH1 domain of p300 ( Fig. 1, C and D).
We searched the proteomics data base at ScanProsite to determine whether any endogenous proteins contain this binding motif. We found that several members of the F-box protein family contain a K(V/L)WXL motif ( Fig. 2A). F-box proteins serve as one of the four subunits in the Skp1/Cullin/ Rbx/F-box (SCF) complex, which functions as an E3 ubiquitin ligase. F-box proteins are further subdivided based on the type of their substrate recognition domain, either a WD40 domain, leucine-rich region, or F-box only with no defined substrate recognition region. In our data base search, all but one of the identified F-box proteins were F-box/WD40 (Fbw) proteins, whereas one F-box only (Fbx18) protein also had the motif. None of the F-box/ leucine-rich repeat (Fbl) family members contained a K(V/ L)WXL motif. WD40 repeat domains fold into a seven-blade propeller, and the crystal structure of Fbw1a (39) reveals that the K(V/L)WXL motif resides within the sixth blade, in a location likely accessible for protein-protein interactions.
p300 and Fbw1a Interact in Vitro and in Cells-To determine if p300 interacts with Fbw proteins, we performed an in vitro interaction assay, which demonstrated that full-length Fbw1a bound to the N-terminal fragment of p300 and to the smaller CH1 domain located within this fragment (Fig. 2B).
Interestingly, Fbw1a also bound to a C-terminal region of p300 that includes the CH3 domain but did not bind to a middle region of p300 that includes the CH2/HAT domains (Fig. 2C). CH1 and CH3 are highly similar in both primary sequence and structure (32) and several transcription factors such as p53, E1A, and Ets-1 bind to both domains. To determine if the K(V/ L)WXL motif is necessary for Fbw1a binding to p300, we mutated the KVWDL sequence within Fbw1a to NERDR. This mutation abolished binding to the N-terminal and CH1 p300 fragments, but it did not abrogate binding to the CH3-to-end fragment (Fig. 2D). This suggests multiple interaction surfaces exist between Fbw1a and p300, similar to those for p53 (34). The Fbw protein, Fbw8, which does not contain a K(V/L)WXL motif, was able to bind to the CH3-to-end fragment but had little or no binding to the N-terminal or CH1 fragments (Fig.  2D). Collectively, these results show that the K(V/L)WXL motif in Fbw1a can direct binding to the N-terminal/CH1 domain of FIGURE 2. F-box proteins contain p300-binding motif. A, alignment of phage peptide consensus with several F-box proteins. B, schematic of p300 fragments fused to GST for use in Fbw interaction assays. C, full-length Fbw1a binds to N-terminal, CH1 and CH3/Q-rich fragments of p300, but not to CH2/ HAT. D, mutation of KVWDL motif within Fbw1a destroys its binding to N-terminal and CH1 fragments of p300, but does not affect its binding to the CH3/Q-rich fragment. Fbw8 (no K(V/L)WXL motif) does not bind N-terminal and CH1 fragments but does bind CH3/Q-rich fragment. Results are representative of three independent experiments.

Fbw1a
Interacts with p300 to Enhance ␤-Catenin Activity p300, but, like certain transcription factors, multiple interaction surfaces mediate the binding between these two classes of proteins.
Next, we wanted to characterize the p300/Fbw interaction within mammalian cells. 293T cells were transiently transfected with HA-tagged p300 alone or together with FLAGtagged Fbw1a, Fbw2, Parkin (a Fbl protein without the K(V/ L)WXL motif), or Fbw8 (40 -42). Forty-eight hours after transfection, immunoprecipitation with an anti-FLAG antibody was performed. Western blotting revealed that HA-p300 was readily co-immunoprecipitated in cells expressing FLAGtagged Fbw1a, Fbw2, and Fbw8 but not from control cells or cells expressing Parkin-FLAG (Fig. 3A). We also wanted to determine if endogenous p300 and Fbw1a could interact. HeLa cell nuclear extracts were prepared, and p300 was immunoprecipitated. Western blotting revealed that Fbw1a was co-immunoprecipitated with p300 but not with a control IgG/mock IP (Fig. 3B). These results verify the binding of Fbw1a to p300 within cells.
To define the region(s) of Fbw1a that can interact with p300, we expressed p300-HA and three Gal4DBD-tagged Fbw1a fragments in 293T cells (Fig. 4A). Co-immunoprecipitation and Western blotting verified that the WD40 domain, which contains the K(V/L)WXL motif, confers binding to p300, whereas other fragments of Fbw1a (F-box, mid) did not (Fig. 4B). This result, along with the in vitro interaction studies (Figs. 1 and 2), demonstrates that the WD40 domain is necessary and sufficient to target Fbw1a for binding to p300.
To confirm the binding surfaces on p300 that are able to bind Fbw1a in cells, six HA-tagged p300 fragments (Fig. 5A) were co-transfected into 293T cells with Fbw1a-FLAG. Immunoprecipitation/Western blotting (Fig. 5, C and D) verified our in vitro interaction data (Fig. 2) indicating that at least two regions of p300, namely CH1 and CH3, can interact with Fbw1a in mammalian cells.
p300 Is a Stable Protein-The substrate recognition domain for many F-box proteins is either a WD40 or leucine-rich region. Given our observation that p300 interacts with the WD40-type substrate recognition domain found in Fbw1a, we wanted to explore the possibility that p300 is a target of SCF/ Fbw1a-mediated degradation. Fbw1a targets, such as such as ␤-catenin, IkBa1, Emi1/2, ATF4, and Per1 are found to rapidly (within 4 h) degrade upon their ubiquitylation by the SCF complex (26,43,44). Overexpression of full-length Fbw1a was found to accelerate the degradation of ATF4 (43), whereas overexpression of an F-box-deleted Fbw1a was found to stabilize ␤-catenin by inhibiting the rest of the SCF complex from associating with it (26). We found that, unlike other Fbw1a targets, p300 is quite stable. Treatment of HeLa cells with the protein synthesis inhibitor, cycloheximide, had no effect on p300 levels over 12 h, whereas p53, a prototypical proteasome target, disappeared within the first 6 h of treatment with cycloheximide (Fig. 6A). Similarly, when HeLa cells were treated with the proteasome inhibitor, MG132, p300 levels exhibited little if any change (Fig. 6B). This is in contrast to p53, which noticeably accumulated within 6 h of MG132 treatment (Fig.

Fbw1a
Interacts with p300 to Enhance ␤-Catenin Activity 6B). Overexpression of Fbw1a or an F-box-deleted Fbw1a (⌬Fbox-Fbwa) in HeLa cells resulted in little to no change in endogenous p300 levels (Fig. 6, C and D). Similar results were obtained in 293T and HCT116 colon cancer cells (data not shown).
We next used shRNA-mediated knockdown of Fbw1a to assess effects on p300 abundance (Fig. 6E). Quantification of Western blot analysis from three independent experiments showed that knockdown of Fbw1a by ϳ50 -60% (Fig. 6E, graph in lower panel) resulted in an increase (ϳ50%) in ␤-catenin protein levels (Fig. 6F), as would be expected, because ␤-catenin is an Fbw1a target. In contrast, Fbw1a knockdown resulted in a ϳ20% reduction in p300 protein levels, which is opposite of what would be expected if p300 was a degradation target of Fbw1 (Fig. 6F). Although certain studies have shown that the degradation of p300 may be induced by specific stimuli (45)(46)(47)(48)(49), these effects take a minimum of 16 -20 h to occur. Our results are consistent with observations that demonstrate that, under normal growth conditions, p300 is a stable protein whose levels are not significantly altered by inhibition of the proteasome (48). In light of these results, our finding that p300 and Fbw1a readily associate under normal growth conditions suggests that the interaction may serve a purpose other than to target p300 for ubiquitylation and degradation.
Fbw1a and p300 Co-localize with ␤-Catenin at Target Gene Promoters-An increasing body of evidence suggests that ubiquitylation machinery and transcription factor complexes work in concert to control the amplitude of target gene expression. Consequently, we wanted to determine if Fbw1a was able to co-localize with p300 at sites of active transcription. Because Fbw1a normally targets ␤-catenin for ubiquitylation and degradation, we hypothesized that expression of ␤-catenin target genes may be dually controlled by Fbw1a and p300. In normal cells, a cytoplasmic complex consisting of adenomatous polyposis coli (APC), axin, and glycogen synthase kinase 3␤ keep ␤-catenin levels low by inducing its phosphorylation. SCF Fbw1a complexes continually target phosphorylated ␤-catenin for ubiquitylation and proteasome-mediated degradation (50,51). Wnt signaling serves to inactivate glycogen synthase kinase 3␤, thereby allowing nonphosphorylated ␤-catenin to escape destruction, translocate to the nucleus, and activate transcription of target genes, such as cyclin D1 (52). We performed ChIP experiments in HCT116 cells, a colon cancer cell line with a mutation in one of the ␤-catenin phosphorylation sites necessary for its degradation (53). Lithium chloride, which inhibits the glycogen synthase kinase-mediated phosphorylation of ␤-catenin and stimulates its nuclear localization, was added to the cells for 0, 1, 3, and 6 h prior to ChIP. Even though one of its phosphorylation sites is mutated in HCT116 cells, endogenous ␤-catenin still responds to LiCl stimulation (54,55). As shown in Fig. 7 (A-C), ␤-catenin, p300, and Fbw1a were all enriched at the cyclin D1 promoter by comparison to a control region of DNA located 4 kb upstream and to the promoter region of a non-relevant but highly expressed gene, enolase1 (compare black lines to gray lines). In response to 1 h of LiCl treatment, ␤-catenin levels at the cyclin D1 promoter rise but then decrease within 3-6 h of treatment (Fig. 7A). While neither p300 (Fig. 7B) nor Fbw1a (Fig. 7C) show an initial increase in response to LiCl, they both dissociate from the promoter within the same time frame as ␤-catenin, after exposure to LiCl stimulation.
Given that HCT116 cells have highly activated ␤-catenin in the absence of stimulation, we also wanted to examine the dynamics of promoter occupancy in HeLa cells, which contain normal ␤-catenin levels and normal Wnt signaling. In this environment, ␤-catenin was recruited to the promoter of the target gene, axin2 within 1 h of Wnt3a stimulation and began to dis-  ). B, lane order for p300 fragments used in transiently transfected 293T cells shown in C and D. C, Fbw1a-FLAG co-immunoprecipitates N-terminal, CH2 to end and CH3 to end p300 fragments but not KIX-middle fragment. D, Fbw1a-FLAG co-immunoprecipitates CH3 to end, and CH3 fragments but not Q-rich fragment. Results are representative of three to four independent experiments.

Fbw1a
Interacts with p300 to Enhance ␤-Catenin Activity sociate by 5 h of treatment (Fig. 7D). In response to Wnt3a, both p300 and Fbw1a were recruited to the axin2 promoter within 3 h and then dissociated by 5 h of treatment (Fig. 7, E and F). All three factors show Wnt-induced enrichment at the axin2 promoter in contrast to a control region of DNA 3 kb downstream and the promoter of the negative control gene, enolase1.
To verify that p300 and Fbw1a were both co-localized with ␤-catenin, we performed a ChIP-reChIP experiment in HCT116 cells. Chromatin fragments bound by ␤-catenin (or IgG control) was first immunoprecipitated, eluted, followed by a re-immunoprecipitation of p300, Fbw1a, or IgG. PCR of the primary ChIP eluate shows that ␤-catenin is, as expected, enriched at the axin2 promoter as compared with IgG and a negative control region of DNA (Fig. 7G). reChIP shows that p300 and Fbw1a were both present in the ␤-catenin immunoprecipitate and further enriched at the axin2 promoter as compared with negative control regions of DNA (Fig. 7H). These results confirm that both p300 and Fbw1a are co-localized with ␤-catenin transcriptional complexes at the axin2 target gene promoter.
To determine the specificity of Fbw1a recruitment to ␤-catenin target gene promoters, we ectopically expressed FLAG-tagged Fbw1a and FLAG-tagged Parkin (which does not contain the K(V/ L)WXL motif) in HCT116 cells. Immunoprecipitation of ␤-catenin from cells transfected with either FLAG-tagged Fbw1a or Parkin showed that binding of ␤-catenin to the axin2 promoter was equivalent in both transfectants (Fig. 7I). However, FLAG immunoprecipitation showed a large difference in the binding of the two F-box proteins. Fbw1a was enriched at the axin2 promoter while Parkin was not associated with it at all (Fig. 7I). We also performed ChIP-reChIP experiments in HCT116 cells transiently transfected with either FLAG-Fbw1a or FLAG-Parkin. ␤-Catenin was immunoprecipitated from each cell type, eluted, and reChIP using an anti-FLAG antibody was then performed. FLAG reChIP showed that Fbw1a was enriched at the axin2 promoter over Parkin (or nonspecific FLAG binding) and in comparison to a nonspecific control (supplemental Fig. S2C). These results confirm that Fbw1a is specifically recruited to ␤-catenin target gene promoters.
Fbw1a Enhances ␤-Catenin Transcriptional Activity-After determining that ␤-catenin, p300, and Fbw1a can co-localize to the promoters of ␤-catenin target genes, we wanted to determine the functional significance of such co-localization. Because our ChIP experiments showed that Fbw1a can localize to the promoter of cyclin D1, we first assessed effects on a cyclin D1 promoter-driven luciferase reporter, CyD1-luc. Overexpression of Fbw1a increased the activity of this reporter in a Fbw1a Interacts with p300 to Enhance ␤-Catenin Activity MAY 8, 2009 • VOLUME 284 • NUMBER 19

JOURNAL OF BIOLOGICAL CHEMISTRY 13039
dose-dependent manner in both HeLa and HCT116 cells, whereas Fbw2 (which does not target ␤-catenin) had no effect in either cell line (Fig. 8, A and B). Interestingly, when we used an Fbw1a construct devoid of its F-box (Fbw1a⌬F) in HeLa cells, its effects on CyD1-luc were greatly diminished. However, in HCT116 cells, Fbw1a⌬F still maintained its ability to enhance CyD1-luc expression (Fig. 8, A and B, middle set of bars).
To confirm that the above observations were due to ␤-catenin activity on the cyclin D1 reporter, we also used pOT-luc, a reporter in which luciferase expression is driven by multimerize copies of the canonical ␤-catenin binding sites and a minimum TATA box. In HeLa cells, exogenous ␤-catenin was added to stimulate the reporter whereas in HCT116 cells, pOT-luc activity was readily detectable due to high levels of endogenous ␤-catenin. In both cell lines, Fbw1a increased pOT reporter activity (Fig. 8, C and D) while Fbw2 did not. Overexpression of Fbw1a had no effect on pOF-luc, in which all canonical ␤-catenin sites have been mutated (Fig. 8, C and D, gray bars), nor did it affect the activity of the internal control, Renilla luciferase (data not shown). The level of Fbw1a expression in these reporter assays was within physiologically relevant levels. Compared with cells transfected with empty vector, the total amount of Fbw1a was less than twice normal physiological levels, even at the highest levels of transfection (supplemental Fig. S3).
To determine if Fbw1a was able to co-activate other transcription factors, we examined the activity of another putative Fbw1a target (SMAD3) as well as two transcription factors that are targets of other degradation pathways (p53 and HIF1␣) (56 -58). As shown in Fig.  8E, ectopic expression of Fbw1a in HeLa cells increased exogenous SMAD3 transcriptional activity on a minimal CAGA(12)-luciferase reporter. This effect occurred both in the absence and presence of TGF␤, which enhances SMAD3 nuclear localization and transactivation. In contrast to ␤-catenin and SMAD3, the transcriptional activity of p53, on a p21-luciferase reporter, was not affected by overexpression of Fbw1a (Fig. 8F). Likewise, HIF1␣ transcriptional activity was not enhanced by Fbw1a, in the absence or presence of desferrioxamine, a chemical that stabilizes HIF1␣ protein levels (Fig. 8G).
Conversely, we used shRNA knockdown to determine the effects of Fbw1a depletion on ␤-catenin transcriptional activity. As determined by reverse transcription-PCR and consistent with Western blot results (Fig. 6E), we were able to achieve significant knockdown of Fbw1a in 293T cells (Fig. 8H). Interestingly, lithium chloride treatment caused a reduction in Fbw1a mRNA levels in both control and Fbw1 knockdown FIGURE 7. ␤-Catenin, Fbw1a, and p300 are co-localized on target promoters. A-C, time course of endogenous factors binding to DNA in response to 0, 1, 3, and 6 h of LiCl treatment in HCT116 cells. A, ␤-catenin ChIP; B, p300 ChIP; and C, Fbw1a ChIP followed by quantitative PCR (ChIP-QPCR) was used to assess amount of each factor at the promoter of cyclin D1 (black squares), a control region 4 kb upstream from the promoter (gray triangles), and a non-related but highly expressed promoter, enolase1 (gray circles). PCR was performed in triplicate, and average C t values were normalized to input using the 2 Ϫ⌬Ct method (1). Normalized values are expressed as the binding of each factor relative to its binding at the Ϫ4 kb control region at the 0-h time point. D-F, time course of endogenous factors binding to DNA in response to 0, 1, 3, and 5 h of Wnt3A treatment in HeLa cells. D, ␤-catenin ChIP; E, p300 ChIP; and F, Fbw1a ChIP followed by QPCR shows binding of each factor to the promoter of axin2 (black squares), a control region 3 kb downstream from the promoter (gray triangles), and the negative control enolase1promoter (gray circles). Data are normalized and expressed as in A-C. G and H, ChIP-reChIP experiment in HCT116 cells. G, primary ChIP-QPCR of IgG or ␤-catenin at axin2 promoter and the indicated negative controls. Data are normalized and expressed as in A-C. H, reChIP of Fbw1a and p300 from ␤-catenin immunoprecipitate. QPCR shows significant enrichment of both Fbw1a and p300 at the axin2 promoter (black bars) as compared with the ϩ3kb control (gray columns) and nonspecific promoter control, enolase 1 (white columns). I, parallel ChIP experiments in HCT116 cells transfected with either FLAG-Fbw1a or FLAG-Parkin. QPCR detects factors at axin2 promoter and a ϩ3-kb control. An anti-FLAG antibody was used to immunoprecipitate Fbw1a and Parkin. Results are normalized to input using the 2 Ϫ⌬Ct method and expressed as -fold enrichment over IgG. Results are representative of two to four independent ChIP assays.

Fbw1a
Interacts with p300 to Enhance ␤-Catenin Activity cells. To specifically address the effects of Fbw1 knockdown on ␤-catenin transactivation, we measured the activity of the pOT and pOF-luc reporters in response to lithium chloride. As shown in Fig. 8I, both control and Fbw1 knockdown cells had low basal levels of pOT-luc reporter activity in the absence of Wnt-signaling. In response to 6 h of LiCl treatment, activity of pOT-luc, but not pOF-luc increased in both cell lines, indicating ␤-catenin transactivation. However, the induction of pOT-luc in Fbw1 knockdown cells was only ϳ50% of that obtained in control cells. Activity of the control reporter, pOF-luc, did not significantly change in response to LiCl or Fbw1 knockdown. In HCT116 cells, shRNA-mediated knockdown of Fbw1 only resulted in a transient (2-3 days) reduction in its expression. However, during this time, Fbw1 knockdown resulted in reduced ␤-catenin transactivation at target gene promoters in response to lithium chloride (supplemental Fig. S4, C-E). Collectively, these results indicate that Fbw1a co-activates ␤-catenin-mediated transcription and that these effects are not simply due to nonspecific transcriptional activation.

DISCUSSION
Several recent studies in diverse model systems have reported that ubiquitylation machinery plays a role in the activation and regulation of gene transcription. The F-box protein, Fbw1a, serves as the substrate recognition component of the SCF E3 ubiquitin ligase complex that is normally responsible for ␤-catenin degradation. In the current study, we have characterized a novel interaction between Fbw1a and the transcriptional co-activators p300 and CBP, revealing a previously undiscovered role for Fbw1a in ␤-catenin transcriptional activation. We demonstrate both in vitro and in cells (with ectopic and endogenous proteins) that Fbw1a can directly bind to p300/CBP, and co-localizes with p300 and ␤-catenin at the promoters of target genes. Highlighting the importance of the Fbw1a/p300 interaction, we found that Fbw1a can enhance ␤-catenin transcriptional activity not only in cells with normal Wnt signaling, but also in cells where a mutation in ␤-catenin hinders its degradation by Fbw1a/SCF. In addition to our demonstration of Fbw1a directly interacting with p300 and CBP, two recent reports have shown direct interactions between p300 and other E3 ubiquitin ligase components, namely APC5/7 of the APC/C complex and Skp2/p45, the related Fbl protein (59,60). Regions of p300 encompassing the CH1 and CH3 domains are involved in the physical interaction of p300 with APC5/7, Skp2, and as our results show, Fbw1a (Figs. 2 and 5). However, each E3 ligase component uses a different surface for binding to p300. APC5 and APC7 utilize sequences within their C termini that are highly similar to the viral protein, E1A, whereas neither Skp2/p45 nor Fbw1a contain such sequences. Instead, Skp2 binds p300 through a region in between its F-box domain and leucine-rich region substrate recognition domain. As we show here, Fbw1a uses yet another unique region, within its WD40 substrate recognition domain, to bind p300 (Fig. 4). We noted that the short peptide sequence KVWDL is sufficient to mediate an interaction with p300, but that it is not the sole sequence within the WD40 domain that binds p300 (Fig. 2). The unique surfaces through which each E3 ligase component interacts with p300 may provide specificity for carrying out different functions in different cellular environments.
Because most if not all ubiquitylation targets of SCF Fbw1a bind to the WD40 domain of Fbw1a, we sought to determine if p300 was also an ubiquitylation and degradation target. We found that, unlike other Fbw1a targets, p300 does not turn over rapidly or in response to overexpressed Fbw1a (Fig. 6). Moreover, unlike ␤-catenin, which is an Fbw1 degradation target, p300 levels do not increase, but may actually decrease, in response to Fbw1 knockdown. Our results do not rule out the possibility that Fbw1a may target p300 for degradation upon certain physiological or pathological stimuli. For example, a study by Sanchez-Molina and colleagues shows that continual treatment of NIH 3T3 cells with platelet-derived growth factor stimulated a slow decline in p300 and CBP expression levels over the course of 1-3 days. The ubiquitylation and degradation of p300 could be expedited by the activation/overexpression of H or N-Ras and occurred in an MDM2-dependent manner (45). Likewise, another report shows that, in cardiac cells, p300 undergoes a p38-dependent phosphorylation in response to doxorubicin treatment, which ultimately triggers degradation of p300 by the proteasome (46). Nonetheless, our results suggest that p300 is not a target of proteasomal degradation under normal growth conditions. In both of the previous studies that reported p300/E3 ligase interactions, the E3 ligase was able to affect transcription of p53. APC5/7 binding to p300 was found to enhance p53 transcriptional activity (59), whereas in contrast, Skp2 was found to inhibit p53 activity by titrating p300 away from p53 (60). In our studies, we found that Fbw1a did not affect the activity of p53 or the hypoxia-inducible transcription factor, HIF1␣. On the other hand, we observed that Fbw1a was able to enhance the transcriptional activity of both ␤-catenin and SMAD3 (Fig. 8) while the highly related F-box protein, Fbw2 did not. Consistent with these findings, we also observed that Fbw1 specifically co-localizes with p300 to ␤-catenin target gene promoters in response to Wnt signaling, in cells with normal and hyper-activated ␤-catenin (Fig. 7). We speculate that the specificity of Fbw1a for binding to ␤-catenin and SMAD3 is the basis for both its degradation and transcriptional co-activation functions. Indeed, a previous report has shown that p300 facilitates the interaction of SMAD3 with ROC1-containing SCF Fbw1a complexes and aids in the ubiquitylation of SMAD3 by this complex (56).
Of interest, Fbw1 expression is integrally linked to Wnt/␤catenin signaling in a feed-forward pathway that serves to autoregulate ␤-catenin levels in normal cells. The ␤-catenin target gene, CRD-BP, produces an RNA-binding protein that binds to and stabilizes Fbw1 mRNA, resulting in an increase in Fbw1 protein (61). In cells with normal Wnt signaling, this feed-forward mechanism results in the enhanced turnover of ␤-catenin and among other substrates, the NFB inhibitor, IkB␣ (61). In primary human colorectal cancers, elevated levels of Fbw1a have been found to correlate with increased ␤-catenin expression (62,63). In these cells, hyperactivated Wnt signaling leads to continual production of the Fbw1-stabilizing protein, CRD-BP, and may explain why we were only able to get transient knockdown of Fbw1 in HCT116 cells. Additionally, continuous degradation of IkB through elevated Fbw1a is thought to promote the anti-apoptotic effects of NFB in proliferating colon cancer cells (63,64). Our results suggest that elevation of Fbw1a, in the absence of ␤-catenin degradation may contribute to heightened ␤-catenin transcriptional activity. Specifically, mutations that make ␤-catenin not degradable by Fbw1a are apparently still able to be co-activated by Fbw1a and p300 (e.g. in HCT116 cells). Consistent with this, we observed that knockdown of Fbw1 reduced the level of ␤-catenin transactivation in response to lithium chloride ( Fig. 8 and supplemental Fig. S4). Notably, high expression levels of p300 may lead to poor prognosis in colon cancer (17) in part due to its ability to recruit proteins like Fbw1a to ␤-catenin target genes, thereby enhancing transcriptional activation.
The various ways in which ubiquitin ligase machinery may fundamentally contribute to transcriptional activation remains to be determined, however some studies have begun to unravel the details. Association of RNA polymerase II with the ubiquitin ligases elongin B/C, which occurs through the polymerase II mediator complex component Med8, is thought to help stimulate transcriptional elongation (65). The ubiquitylation and subsequent degradation of Gal4 by the ubiquitin ligase Dsg1 was shown to be critical for disassembly of initiation complexes and allowing co-transcriptional RNA processing of mRNA into protein (66). Transcription-coupled DNA damage repair machinery relies on CSA/DDB1/Cul4A/Roc1 E3 ubiquitin ligase activity to promote the ubiquitylation and clearance of stalled RNA polymerase II complexes from sites of active transcription (67). In a non-proteolytic manner, mono-ubiquitylation of Gal4 inhibits its removal from promoters (68) while ubiquitylation of estrogen receptor promotes its dynamic on/off cycling from promoters and nuclear mobility (69,70). Lastly, components of the 19 and 20 S proteasome have been found to bind to sites of highly transcribed genes in yeast, suggesting that proteolytic processing plays a pivotal role in transcription (71). Given the diversity of mechanisms whereby ubiquitylation machinery can facilitate transcription, the recruitment of F-box proteins to p300 and CBP may enhance the transcription co-activator function of these proteins through a multitude of mechanisms.