The Fbw7/Human CDC4 Tumor Suppressor Targets Proproliferative Factor KLF5 for Ubiquitination and Degradation through Multiple Phosphodegron Motifs^*

Cell Culture and Transfection

Human embryonic kidney 293T cells and human epithelial carcinoma HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum at 37 °C in 5% CO₂. HCT116, HCT116 Fbw7^−/−, DLD1, and DLD1 Fbw7^−/− cells (kindly provided by Dr. Jonathan Cherry, The Johns Hopkins University) were cultured in McCoy's 5A medium (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum. Transfections were performed using calcium phosphate-DNA coprecipitation (for 293T cells), SunbioTrans-EZ (for HeLa cells; Shanghai Sunbio Medical Biotechnology Co., Ltd.), and Lipofectamine 2000 (for HCT116 cells; Invitrogen) according to the manufacturers' instructions.

Plasmids

Expression plasmids of FLAG-Fbw2, FLAG-Fbw5, FLAG-Fbw7, FLAG-Fbw8, and β-TrCP1 were kindly provided by Dr. Michele Pagano (New York University). pcDNA3.1-FLAG-Fbw7α, pcDNA3.1-FLAG-Fbw7β, and pcDNA3.1-FLAG-Fbw7γ were subcloned from glutathione S-transferase (GST)-tagged human CDC4α, CDC4β, and CDC4γ (kindly provided by Dr. Steven I. Reed, The Scripps Research Institute). KLF4 and KLF5 expression constructs were provided by Dr. Hitoshi Niwa (RIKEN Center for Developmental Biology, Kobe, Japan) and Dr. Mukesh K. Jain (Case Cardiovascular Research Institute, Cleveland, OH). Hemagglutinin (HA)-ubiquitin plasmids were described previously (38). Deletion mutants of Fbw7, human KLF5, and mouse KLF5 and KLF4 and their mutants were subcloned into the pcDNA3.1 vector with a FLAG, HA, or green fluorescent protein (GFP) tag at the N terminus by standard cloning methods. GST-KLF5 CPD and GST-Fbw7 WD40 were cloned into pGEX-4T-1. GSK3β plasmids were provided by Dr. Kunliang Guan. WWP1 was amplified from 293T cDNA and cloned into a pcDNA3.1-FLAG vector within the EcoRI and NotI restriction sites. All vectors were confirmed by DNA sequencing.

Immunoprecipitation and Immunoblotting

Transfected cells were lysed in lysis buffer (50 mm Tris-HCl (pH 7.4), 150 mm NaCl, 10% glycerol, 1 mm EDTA, 0.5% Nonidet P-40, and a mixture of protease inhibitors) and cleared by centrifugation. Cleared cell lysates were incubated with 1 μg of anti-HA or anti-FLAG antibody (Sigma) and 16 μl of protein A/G beads (Santa Cruz Biotechnology, Santa Cruz, CA) for 3 h.

To detect endogenous protein interactions, HeLa cells were treated with the proteasome inhibitor MG132 (20 μm; benzyloxycarbonyl-l-leucyl-l-leucyl-l-leucinal, Sigma) for 6 h before harvesting. Cells were lysed in ice-cold lysis buffer. Cleared cell lysates were incubated with 4 μg of anti-KLF5 antibody (AF3758, R&D Systems) or 5 μg of anti-Fbw7 antibody (Ab12292, Abcam, Hong Kong) and 16 μl of protein A/G beads for 3 h at 4 °C. After extensive washing, beads were boiled at 100 °C for 5 min. Proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membranes (Millipore), followed by immunoblotting using anti-HA (1:1000), anti-FLAG (1:1000), or anti-GFP (1:1000; Santa Cruz Biotechnology) antibody. Endogenous KLF5 or Fbw7 was detected using anti-KLF5 antibody (1:1000; sc-22797, Santa Cruz Biotechnology) or anti-Fbw7 antibody (1:500; Ab12292, Abcam), respectively. Immunoblots were analyzed using the Odyssey system (LI-COR Biosciences).

Flow Cytometry-based Protein Degradation Assay

KLF5 was subcloned into a DsRed-IRES-EGFP vector (kindly provided by Dr. Stephen J. Elledge, Harvard Medical School). Two μg of DsRed-IRES-EGFP vector or DsRed-IRES-EGFP-KLF5 were transfected into 293T cells in 6-well plates. Transfected cells were collected and suspended in phosphate-buffered saline (PBS) and analyzed using a FACSCalibur flow cytometer (BD Biosciences).

Immunofluorescence

293T cells cultured on coverslips were transfected with 1 μg of GFP-KLF5 and 1 μg of FLAG-Fbw7 in 24-well plates. The cells were treated with 10 μm MG132 for 4 h before fixation with 4% paraformaldehyde. Fbw7 was stained with mouse anti-FLAG antibody (1:100).

For degradation assays, DLD1 cells, Fbw7-deleted DLD1 cells, or GFP-Fbw7-expressing HeLa cells were fixed with 4% paraformaldehyde, permeabilized using 0.1% Triton X-100, blocked with 1% bovine serum albumin in PBS for 1 h, and stained with rabbit anti-KLF5 antibody (1:100; Santa Cruz Biotechnology), followed by Texas Red-labeled anti-rabbit secondary antibody (1:100; Invitrogen). Fixed permeabilized HeLa cells were also labeled with rabbit anti-KLF5 antibody (1:100) and mouse anti-Fbw7 antibody (1:100; Ab74054, Abcam). The slides were then stained with Texas Red-labeled anti-rabbit secondary antibody and fluorescein isothiocyanate-labeled anti-mouse secondary antibody (1:100; Invitrogen) to detect co-localization of endogenous KLF5 and Fbw7. The nuclei were stained with 0.5 μg/ml 4′,6-diamidino-2-phenylindole (Sigma). Images were analyzed using a Leica SP5 confocal microscope.

Reverse Transcription (RT)-PCR Assay

Total RNA was isolated from cells using TRIzol reagent (Invitrogen) according to the manufacturer's directions, treated with RNase-free DNase, and subjected to reverse transcription with random hexanucleotide primers. The following primer sequences were used: KLF5, 5′-CACTACTGCGATTACCCTG-3′ (sense) and 5′-GGTCTGTCATTTGAGGGAG-3′ (antisense); Fbw7, 5′-GACGCCGAATTACATCTGTC-3′ (sense) and 5′-GTAGCAGGTCTTTGGGTTC-3′ (antisense); KLF4, 5′-CCATTACCAAGAGCTCATGC-3′ (sense) and 5′-TAGTGCCTGGTCAGTTCATC-3′ (antisense); survivin, 5′-GACTTGGCCCAGTGTTTCTTC-3′ (sense) and 5′-CACTTTCTCCGCAGTTTCCTC-3′ (antisense); and GSK3β, 5′-CTCAAATTAAGGCACATCC-3′ (sense) and 5′-CGGTCTCCAGTATTAGCAT-3′ (antisense).

In Vitro Phosphorylation Assay

Wild-type or CPD-deficient (3A) GST-KLF5 CPD protein (amino acids 221–356) was purified from Escherichia coli by standard protocols. FLAG-KLF5 was expressed in 293T cells and purified using 1 μg of anti-FLAG antibody for immunoprecipitation. After washing twice with kinase buffer (25 mm Tris-HCl (pH 7.5), 5 mm β-glycerophosphate, 10 mm MgCl₂, 2 mm dithiothreitol, and 0.1 mm Na₃VO₄), purified FLAG-KLF5 or 2 μg of purified GST-KLF5 CPD protein were incubated with 200 ng of recombinant GSK3β kinase (R&D Systems) in 25 μl of kinase buffer containing 200 μm ATP and 5 μCi of [γ-³²P]ATP at 30 °C for 1 h. The reaction was terminated by the addition of 8 μl of 4× Laemmli SDS sample buffer. The reaction mixtures were boiled and separated by SDS-PAGE and subjected to autoradiography.

GST Pulldown Assay

FLAG-KLF5 was expressed in 293T cells. The transfected cells were lysed and cleared by centrifugation. GST-Fbw7 WD40 was purified from E. coli. Ten μg of purified GST or GST-Fbw7 WD40 protein were incubated with KLF5 cell lysates for 1 h, and GST proteins were purified using glutathione-Sepharose 4B (Amersham Biosciences). Bound KLF5 was detected by immunoblotting.

Small Interfering RNA (siRNA) Knockdown and Cycloheximide Chase Assay

HeLa cells were transfected with siRNA oligonucleotides using Lipofectamine. Two different oligonucleotides against Fbw7 were used: siRNA1 (5′-ACCUUCUCUGGAGAGAGAAAUGCTT-3′) and siRNA2 (5′-GUGUGGAAUGCAGAGACUGGAGATT-3′) (27). Forty-eight h after transfection, cells were treated with 50 μg/ml cycloheximide for 0–4 h. The transfected cells were lysed and analyzed by immunoblotting using anti-KLF5 or anti-tubulin (Sigma) antibody.

Pulse-Chase Assay

HeLa cells were plated in 60-mm dishes at 5 × 10⁵ cells and transfected using Lipofectamine with control siRNA (5′-UUCUCCGAACGUGUCACGUTT-3′) or siRNA against Fbw7 (see above) or GSK3β (5′-GUAUUGCAGGACAAGAGAUTT-3′). Forty-eight h after transfection, cells were washed twice with PBS and cultured in Met/Cys-free Dulbecco's modified Eagle's medium (21013-024, Invitrogen) supplemented with 2 mm l-glutamine (G8540, Sigma) and 5% dialyzed fetal calf serum (30067-334, Invitrogen) for 30 min at 37 °C. The cells were labeled with 50 μl of EXPRE³⁵S³⁵S protein labeling mixture (100 μCi/dish; NEG072007MC, PerkinElmer Life Sciences) for 40 min at 37 °C. After extensive washing, cells were lysed in radioimmune precipitation assay buffer (1× PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 10 mm N-ethylmaleimide, 1 mm phenylmethylsulfonyl fluoride, 1 mm NaV₃O₄, and 1 mm NaF). Endogenous KLF5 was immunoprecipitated with goat anti-KLF5 antibody (2 μg/ml; R&D Systems). Proteins were separated by SDS-PAGE. The gel was treated with Amplify solution (Enlightning^TM, PerkinElmer Life Sciences) to enhance the ³⁵S signal, dried, and autoradiographed.

Ubiquitination Assay

In vivo ubiquitination assays were performed as described previously (38). Briefly, FLAG-KLF5 was transiently transfected into 293T cells with or without HA-ubiquitin and GFP-Fbw7α. Transfected cells were treated with 10 μm MG132 for 6 h before harvesting. Cells were lysed in 500 μl of radioimmune precipitation assay buffer with protease inhibitors, and KLF5 was immunoprecipitated using 1 μg of anti-FLAG antibody for 4 h at 4 °C. Polyubiquitinated KLF5 was detected using anti-HA antibody.

To detect the ubiquitination of endogenous KLF5 under denaturing conditions, HeLa cells were transfected using Lipofectamine with control or Fbw7 siRNA oligonucleotides. Twenty-four h later, cells were transfected with HA-ubiquitin. Transfected cells were treated with 20 μm MG132 for 6 h and lysed in 60 μl of modified radioimmune precipitation assay buffer (1× PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1% SDS, 10 mm N-ethylmaleimide, 1 mm phenylmethylsulfonyl fluoride, 1 mm NaV₃O₄, and 1 mm NaF). Cell lysates were boiled for 10 min, diluted in 10 volumes of lysis buffer without SDS, and subjected to immunoprecipitation using 4 μg of anti-KLF5 antibody (R&D Systems). Ubiquitination of KLF5 was detected by immunoblotting using anti-HA antibody (1:1000). Endogenous KLF5 was detected using anti-KLF5 antibody (1:1000; Santa Cruz Biotechnology).

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium Bromide Assay

HCT116 cells were seeded in 96-well plates and cotransfected with control siRNA or KLF5 siRNA (5′-ACCUUCUCUGGAGAGAGAAAUGCTT-3′) and mouse KLF5 plasmids (HA-KLF5 and HA-KLF5-3A). Forty-eight h after transfection, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (T0793, Sangon Biotech Co., Ltd., Shanghai, China) was added to each well for 4 h. Me₂SO (150 μl) was then added, and the absorbance was measured at 490 nm using a microplate spectrophotometer (SpectraMax 190, Molecular Devices, Inc.).

Colony Formation Assay

HCT116 cells (2.5 × 10³) transfected with wild-type KLF5 or its CPD-deficient mutant (KLF5-3A) in the presence or absence of Fbw7 plasmids were mixed with 0.8% agarose in warm 2× Dulbecco's modified Eagle's medium containing 20% fetal bovine serum and plated in each well of a 6-well plate on top of a prepared 1.2% agar base. Colony formation in the soft agar was performed as described previously (18). The number of colonies was counted under a microscope 2 weeks later.

Luciferase Assay

The survivin gene promoter was amplified from human genomic DNA by PCR using primer sequences 5′-CTCGAGCTCAAAGTGTTGGGATTACAGG-3′ and 5′-AAGCTTTGGTCCTTGAGAAAGGGCTGC-3′. Two restriction endonuclease recognition sites, XhoI and HindIII, were attached to the primers for subsequent cloning. The PCR products were purified from agarose gels, digested, and subcloned into a pGL3-Basic plasmid (Promega). The construct was verified by restriction digestion and DNA sequencing. Luciferase activity was measured using the Dual-Luciferase assay system (Promega) according to the manufacturer's protocol.

KLF5 Is Degraded by Tumor Suppressor Fbw7

SCF^Fbw7 ubiquitin ligase can efficiently ubiquitinate and degrade substrates through its association with CPD motifs (24). The protein sequence of KLF5 contains three potential conserved CPDs identified in numerous Fbw7 substrates (Fig. 1A). To determine whether Fbw7 promotes the degradation of KLF5, HA-tagged KLF5 was expressed either alone or in the presence of various FLAG-tagged F-box ubiquitin ligases in 293T cells. Coexpression of Fbw7 significantly decreased KLF5 protein levels (Fig. 1B). In contrast, other F-box proteins (β-TrCP1, Fbw2, Fbw5, and Fbw8) had little effect, suggesting that Fbw7 is a specific E3 ubiquitin ligase of KLF5.

Thus far, three Fbw7 isoforms (Fbw7α, Fbw7β, and Fbw7γ) with different subcellular localizations have been reported (39). Fbw7α is localized mainly in the nucleoplasm, whereas Fbw7β is found in the cytoplasm and Fbw7γ in the nucleolus (40). 293T cells were transfected with KLF5 in the absence or presence of each individual Fbw7 isoform. The nuclear α- and γ-isoforms significantly promoted the degradation of KLF5, but Fbw7β did not (Fig. 1C). Furthermore, the degradation of KLF5 mediated by the α- and γ-isoforms was significantly blocked by the proteasome inhibitor MG132, suggesting that this process is dependent on the ubiquitin-proteasome pathway (Fig. 1C).

The domains of Fbw7α required for KLF5 degradation were also explored (Fig. 1D). Relative to the wild type, expression of the F-box domain deletion mutant (Fbw7αΔF) or the WD40 domain alone could not degrade KLF5, suggesting that the F-box domain is critical for Fbw7-mediated KLF5 degradation.

To illustrate the specificity of Fbw7-mediated KLF5 degradation, we examined whether the protein levels of KLF4 could be affected by Fbw7. KLF4 is a close family member of KLF5 but without any potential CPDs (as analyzed using Scansite of the Massachusetts Institute of Technology). As expected, increasing the amount of Fbw7α greatly promoted KLF5 degradation but had little effect on KLF4 protein levels (Fig. 1E).

To study the degradation of KLF5 by Fbw7 in cells more precisely, a recently reported fluorescence-activated cell sorter (FACS) analysis-based protein degradation system was used (41). KLF5 was fused to GFP in an expression vector, and DsRed was included in the same transcript as an internal control (Fig. 2A). When expressed alone in 293T cells, the relative expression levels of GFP-KLF5 were much lower than those of GFP (Fig. 2B). Treatment with MG132 significantly increased GFP-KLF5 but had little effect on GFP protein levels (Fig. 2C), suggesting that GFP-KLF5 is constitutively degraded by the ubiquitin-proteasome pathway in cells. Coexpression of Fbw7α dramatically decreased KLF5 protein levels, whereas deletion of the F-box produced the opposite effect (Fig. 2, D and E). These results are consistent with the notion that Fbw7αΔF functions as a dominant-negative mutant to block the binding of substrate to endogenous Fbw7. Collectively, these data demonstrate that Fbw7α specifically degrades KLF5 by the ubiquitin-proteasome pathway.

Fbw7α Interacts with KLF5

We next used a co-immunoprecipitation assay to examine how KLF5 interacts with Fbw7α. GFP-KLF5 was cotransfected into 293T cells with or without FLAG-Fbw7α in the presence of MG132 to prevent degradation of KLF5 by Fbw7α. FLAG-Fbw7α was precipitated using anti-FLAG antibody, and the immunoprecipitates were analyzed by immunoblotting with anti-GFP or anti-FLAG antibody. KLF5 copurified efficiently with Fbw7. In contrast, KLF4 was not detected in Fbw7α immunoprecipitates (Fig. 3A).

To identify the domains of Fbw7 involved in its interaction with KLF5, full-length Fbw7α, Fbw7αΔF, and the WD40 domain were FLAG-tagged and coexpressed with GFP-KLF5. Fbw7α or its mutants were precipitated using anti-FLAG antibody, and copurified KLF5 was detected using anti-GFP antibody (Fig. 3B). Both Fbw7αΔF and the WD40 domain alone interacted efficiently with KLF5, indicating that Fbw7 binds to KLF5 through its WD40 domain. This was confirmed using a GST pulldown assay (Fig. 3C). Relative to GST, GST-Fbw7 WD40 efficiently bound KLF5. Taken together, these data show that Fbw7 binds to KLF5 through its WD40 domain.

Co-immunoprecipitation studies showed that endogenous Fbw7 copurified with endogenous KLF5 (Fig. 3, D and E). The co-localization of Fbw7 and KLF5 in HeLa cells was confirmed by immunofluorescence. Fbw7 co-localized with KLF5 in the nucleus of cotransfected cells (Fig. 3F) and untransfected HeLa cells (Fig. 3G), consistent with the results from the interaction assays.

Fbw7α Degrades and Interacts with KLF5 via Three CPDs

Three putative CPDs have been identified in KLF5 (Fig. 1A). To determine whether these CPDs are involved in Fbw7-mediated KLF5 degradation, key amino acid residues (Thr²²⁷, Ser²⁹², and Thr³¹² in mouse KLF5, which are also conserved in human and rat) were mutated to Ala. These mutants were then subjected to degradation analysis. Surprisingly, mutating these sites individually had only minor effects on Fbw7-mediated degradation (Fig. 4, A and B). However, some Fbw7 substrates, such as PGC-1α and cyclin E, contain more than one degron (34, 42). Therefore, Fbw7 might target KLF5 for degradation via multiple degrons as well. The double mutants T227A/S292A, T227A/T312A, and S292A/T312A and the triple mutant T227A/S292A/T312A (KLF5-3A) were thus created and subjected to the degradation assay (Fig. 4C). The double mutant T227A/S292A was more resistant to Fbw7α (∼35% degradation) compared with T227A/T312A (67% degradation) and S292A/T312A (57% degradation), suggesting that CPD1 and CPD2 might play more important roles than CPD3 in Fbw7-mediated degradation. In contrast, the triple CPD mutant (KLF5-3A) was strongly resistant to Fbw7α degradation (Fig. 4, C and D). Taken together, these data indicate that each individual CPD of KLF5 is a functional Fbw7 degradation signal.

We next examined whether these CPDs are required for the interaction of KLF5 with Fbw7. Compared with wild-type KLF5, the triple KLF5 mutant failed to interact with Fbw7 (Fig. 4E), thereby proving that KLF5 interacts with Fbw7 via these phosphodegrons.

As noted above, WWP1 can promote the degradation of KLF5 (21). We therefore determined if the triple mutation also affects WWP1-mediated degradation. However, KLF5-3A was still efficiently degraded by WWP1 (Fig. 4F), suggesting that Fbw7 and WWP1 use different mechanisms to degrade KLF5.

Fbw7 Targets KLF5 for Ubiquitination

Fbw7 is a component of E3 ubiquitin ligase that promotes the degradation of target proteins through ubiquitination (24). We thus used an in vivo ubiquitination assay to test whether Fbw7 promotes KLF5 ubiquitination. 293T cells transfected with FLAG-KLF5 and HA-ubiquitin in the absence or presence of GFP-Fbw7α were treated with MG132 for 6 h to stabilize the ubiquitinated proteins before lysis. In the absence of Fbw7α, KLF5 was weakly ubiquitinated (Fig. 5A). The addition of Fbw7α greatly enhanced KLF5 ubiquitination, whereas mutation of all three CPDs completely abolished Fbw7α-induced KLF5 ubiquitination (Fig. 5B).

To test whether endogenous Fbw7 can ubiquitinate endogenous KLF5, HeLa cells were transfected with control or Fbw7 siRNA. HA-ubiquitin was then expressed in the transfected cells. After treatment with MG132, endogenous KLF5 was purified under denaturing conditions, and ubiquitination was detected by immunoblotting (Fig. 5C). Knockdown of Fbw7 significantly decreased the ubiquitination of endogenous KLF5. Taken together, these results support the hypothesis that Fbw7 negatively regulates the stability of KLF5 through ubiquitination in a CPD-dependent manner.

Endogenous KLF5 Is Degraded by Endogenous Fbw7 in Cancer Cell Lines

To examine whether endogenous KLF5 is regulated by Fbw7, GFP-Fbw7α was transfected into HeLa cells expressing high levels of endogenous KLF5 protein. Immunofluorescence staining showed that expression of wild-type Fbw7α dramatically decreased endogenous KLF5 levels relative to the GFP control, whereas Fbw7αΔF significantly increased KLF5 levels (Fig. 6A), suggesting that the mutant might function as a dominant-negative form to block the Fbw7-mediated degradation of endogenous KLF5.

To determine whether endogenous KLF5 could be targeted for degradation by endogenous Fbw7, KLF5 protein levels were measured when endogenous Fbw7 was knocked down. Transfection of siRNA against Fbw7 into HeLa cells significantly inhibited its expression (as measured by RT-PCR). Knockdown of Fbw7 in turn efficiently elevated the amount of KLF5 protein but had little effect on its mRNA levels (Fig. 6B). In addition, the half-life of KLF5 protein increased significantly in Fbw7 knockdown cells compared with controls as assessed by both pulse-chase and cycloheximide chase assays (Fig. 6, C and D), suggesting that the elevated KLF5 protein levels in Fbw7-deficient cells are indeed the result of reduced protein degradation.

To confirm this, the Fbw7-deficient colon carcinoma cells HCT116 and DLD1 (43) were used to examine the expression of Fbw7 and KLF5. Deletion of Fbw7 weakly increased KLF5 mRNA levels in HCT116 cells but produced a modest reduction in DLD1 cells (Fig. 6E). In contrast, the protein levels of KLF5, but not KLF4, significantly increased in both Fbw7-deficient cell lines compared with their parental cells. These data were confirmed by immunofluorescence assays: deletion of Fbw7 in DLD1 cells significantly increased nuclear KLF5 protein levels (Fig. 6F). Taken together, our data indicate that endogenous KLF5 protein levels are regulated by endogenous Fbw7 ubiquitin ligase activity in cancer cell lines.

GSK3β Is Required for Fbw7-mediated Degradation of KLF5

Many known Fbw7 substrates (e.g. c-Myc, c-Myb, SREBP-1, PGC-1α, etc.) are phosphorylated by GSK3β. Phosphorylation by GSK3β promotes the binding of substrate to Fbw7 and enhances substrate degradation (25, 27, 34, 44). Interestingly, Scansite software analysis revealed that KLF5 CPDs are also potential GSK3 phosphorylation motifs (Fig. 1A). To investigate whether KLF5 is a substrate of GSK3β, we performed in vitro phosphorylation assays. HA-KLF5 was expressed in 293T cells and purified with anti-HA antibody. KLF5 was indeed significantly phosphorylated by recombinant GSK3β (Fig. 7A). We next tested whether the CPDs of KLF5 are phosphorylated directly by GSK3. The KLF5 fragment containing wild-type or mutant CPDs was fused with GST, purified from E. coli, and subjected to in vitro phosphorylation assays. As expected, phosphorylation of wild-type KLF5 by GSK3β was much stronger than that of the CPD-deficient fragment (Fig. 7B). These data indicate that KLF5 is a substrate of GSK3 and that its CPDs are potential GSK3 phosphorylation sites.

We next examined the interaction between GSK3 and KLF5 by co-immunoprecipitation. FLAG-KLF5 was cotransfected with or without Myc-GSK3 into 293T cells, and KLF5 was immunoprecipitated using anti-FLAG antibody. GSK3β copurified with KLF5 only when they were cotransfected (Fig. 7C), indicating that GSK3β interacts specifically with KLF5.

We further explored whether GSK3β promotes KLF5 degradation. KLF5 was coexpressed with Fbw7α in the presence or absence of constitutively active GSK3β (GSK3β-S9A) in 293T cells, and KLF5 protein levels were monitored by immunoblotting. KLF5 protein levels decreased in the presence of Fbw7α and were further reduced when GSK3β-S9A was coexpressed (Fig. 7D). In contrast, both Fbw7α and GSK3β-S9A had little effect on protein levels in the KLF5-3A mutant, suggesting that the GSK3β promotion of Fbw7α-mediated KLF5 degradation depends on KLF5 CPDs.

The role of endogenous Fbw7 in GSK3-promoted KLF5 degradation was also examined by siRNA knockdown. Expression of GSK3β-S9A significantly reduced endogenous KLF5 protein levels, but the reduction was inhibited by Fbw7 knockdown (Fig. 7E), suggesting that Fbw7 is required for GSK3β-promoted KLF5 degradation.

To confirm that GSK3β activity is required for KLF5 degradation, we took advantage of the FACS-based degradation system (Fig. 7F). Treatment with the GSK3 inhibitor LiCl significantly elevated KLF5 protein levels, and this was increased further by the addition of MG132. Furthermore, GSK3β knockdown using specific siRNAs clearly extended the half-life of endogenous KLF5 as measured by pulse-chase assay (Fig. 7G). Similar results were obtained when the cells were treated with LiCl (data not shown). Treatment with LiCl also significantly inhibited the ubiquitination of endogenous KLF5 (Fig. 7H), indicating that endogenous GSK3 activity is required for KLF5 ubiquitination and degradation.

To further understand the mechanism by which GSK3 affects KLF5 stability, we examined the effect of GSK3 activity on the interaction between Fbw7 and KLF5. Treatment with LiCl clearly decreased the interaction between endogenous Fbw7 and KLF5 (Fig. 7I), suggesting that GSK3 regulates KLF5 stability by promoting the binding of KLF5 to Fbw7.

Fbw7 Negatively Regulates KLF5 Biological Activity in Colorectal Cancer Cells

It has been reported that expression of KLF5 promotes the growth and proliferation of colorectal cancer cells (18). We thus tested whether Fbw7-mediated degradation inhibits KLF5-dependent cell proliferation using a soft agar formation assay. Compared with empty vector, more colonies were formed when wild-type KLF5 was overexpressed, whereas the Fbw7-resistant mutant KLF5-3A was able to stimulate more HCT116 cells to form colonies compared with the wild type (Fig. 8A). In addition, coexpression of Fbw7α significantly inhibited KLF5-mediated cell proliferation but had little effect on KLF5-3A-mediated proliferation (Fig. 8B). Furthermore, knocking down endogenous KLF5 using human KLF5-specific siRNA (which has little effect on mouse KLF5) significantly inhibited the proliferation of HCT116 cells (Fig. 8C). Re-introducing mouse KLF5 rescued the activity induced by KLF5 knockdown, suggesting that KLF5 is critical for HCT116 cell proliferation. As expected, expression of mouse KLF5-3A had a stronger effect on cell proliferation compared with the wild type. Thus, these data support an inhibitory role for Fbw7α-mediated degradation in the KLF5-stimulated anchorage-independent growth of cancer cells.

On the basis of our results thus far, we speculated that Fbw7 might inhibit the expression of KLF5-transactivated genes. We therefore measured by RT-PCR the expression levels of the known KLF5-targeted genes platelet-derived growth factor-A (45), survivin (46), and cyclin D₁ (47) in wild-type or Fbw7-deficient HCT116 and DLD1 cancer cells. Survivin mRNA levels in particular were elevated significantly in both types of Fbw7-deficient cells (Fig. 8D). Subsequent luciferase reporter assays showed that the expression of KLF5 significantly enhanced the activity of the survivin essential promoter in HCT116 cells, whereas this effect was inhibited efficiently by coexpression of Fbw7α (Fig. 8E). Thus, Fbw7 appears to down-regulate the transcriptional activity of KLF5.