Fbw7 (F-box WD40 protein 7) is a major tumour suppressor, which mediates the degradation of several potent oncogenes. PKC (protein kinase C) comprises a serine/threonine kinase family that can promote transformation when dysregulated. In the present study, we investigated the relationship between Fbw7 and PKC. Multiple members of the PKC superfamily interact with the substrate-binding domain of Fbw7. However, we find no evidence for Fbw7-mediated degradation of PKC. Instead, we demonstrate that Fbw7 is a novel substrate for PKC. Two residues within the isoform-specific N-terminus of Fbw7α are phosphorylated in a PKC-dependent manner, both in vitro and in mammalian cells (Ser10 and Ser18). Mutational analyses reveal that phosphorylation of Fbw7α at Ser10 can regulate its nuclear localization. Cancer-associated mutations in nearby residues (K11R and the addition of a proline residue at position 16) influence Fbw7α localization in a comparable manner, suggesting that mislocalization of this protein may be of pathological significance. Together these results provide evidence for both physical and functional interactions between the PKC and Fbw7 families, and yield insights into the isoform-specific regulation of Fbw7α.
Fbw7 (F-box WD40 protein 7) is a conserved F-box WD40 protein, also known as Fbw6 in mice , archipelago in flies , SEL10 in worms  and Cdc4 in yeast . Fbw7 functions as the substrate recognition component of an SCF (Skp1/cullin/F-box) ubiquitin ligase, binding specifically to a conserved phospho-motif, termed the CPD (Cdc4 phosphodegron), in all reported targets . The CPD sequence comprises Leu/Ile-Leu/Ile/Pro-phosphoThr-Pro-〈Arg/Lys〉4, where 〈〉 refers to disfavoured residues ; the central threonine is usually phosphorylated by GSK3 (glycogen synthase kinase 3) . Through CPD recognition, Fbw7 targets an impressive collection of proteins for ubiquitination and proteasomal degradation, including cyclin E [2,7,8], c-Myc  and c-Jun [10,11]. In this way, Fbw7 regulates diverse cellular functions, including cell growth, proliferation and differentiation . Furthermore, since several of its substrates are potent oncogenes, Fbw7 represents a major tumour suppressor  and is mutated in 6% of human cancers .
Mammalian Fbw7 comprises a family of three differentially spliced isoforms: Fbw7α, Fbw7β and Fbw7γ . In certain contexts, these proteins function co-operatively. For instance, Fbw7α and Fbw7γ operate sequentially to degrade cyclin E in HEK (human embryonic kidney)-293 cells . However, abundant evidence also exists for isoform specificity within the Fbw7 family. For example, Fbw7α is primarily responsible for targeting cyclin E, c-Myc and SREBP (sterol-regulatory-element-binding protein)-1 for degradation in Hct116 cells . The three Fbw7 family members localize to distinct subcellular compartments, as determined by isoform-specific N-terminal regulatory sequences . Fbw7α is predominantly nuclear and Fbw7β is cytosolic, whereas Fbw7γ resides in the nucleolus; the differential distribution of these proteins is likely to underlie important differences in their profiles of substrate degradation. For instance, Fbw7γ specifically mediates the degradation of c-Myc in the nucleolus to control U2OS cell size . In light of the critical roles performed by Fbw7 proteins, it will be important to further characterize the mechanisms through which they are regulated.
PKC (protein kinase C) is a phospholipid-dependent, serine/threonine kinase family which plays a central role in signal transduction . In mammals, there are 13 isoforms, which can be divided into four subgroups based on different co-factor requirements: classical PKCs (isoforms α, βI/II and γ), novel PKCs (isoforms δ, ϵ, θ, η), atypical PKCs (isoforms ι/λ and ζ) and PKNs (protein kinase novel; isoforms 1, 2 and 3) . As a major cellular receptor of the tumour-promoting phorbol esters, PKC has long been associated with cancer . Although only a limited number of PKC mutations have been identified in tumour samples, dysregulation of its kinase activity has been widely linked with disease . For instance, PKCϵ  and PKCι  are both oncogenes, prognostic markers and potential therapeutic targets in cancer. Although the regulatory inputs that control PKC activity have been extensively characterized , our understanding of how PKC-mediated phosphorylation regulates specific substrates remains limited.
In the present study, we explored the relationship between the Fbw7 and PKC protein families. We show that PKC phosphorylates Fbw7α in an isoform-specific manner and can inhibit its nuclear localization. Furthermore, cancer-associated mutations proximal to the PKC phosphorylation sites influence Fbw7α in a comparable manner. We conclude that Fbw7α represents a novel PKC substrate, whose exclusion from the nucleus may be of pathological significance.
All reagents were purchased from Sigma–Aldrich unless otherwise indicated. Yeast SD (synthetic-derived) and tissue culture medium were prepared by CR-UK (Cancer Research UK) Research Services, Opti-MEM® was from Gibco and FBS (fetal bovine serine) was from PAA laboratories. Pharmacological inhibitors were obtained from Calbiochem; BIM1 and MG132 were reconstituted in DMSO; calyculin A was dissolved in ethanol. All peptides were prepared by the CR-UK Peptide Synthesis Laboratory. Purified recombinant PKCα and PKCζ were obtained from Mr Philip Whitehead (London Research Institute, CR-UK, London, U.K.), PKCδ (human) and PKCϵ (human) were purchased from Calbiochem.
DNA constructs and site-directed mutagenesis
Molecular biology was performed using standard procedures. All subcloning was achieved by PCR. Mouse PKCϵ cDNA and GFP (green fluorescent protein)–PKCϵ were as described previously . PKCϵ was cloned into pGBKT7 (Clontech) between the NcoI and BamHI sites (primers F, 5′-CAGCACCATGGCAATGGTAGTGTTCAATGGCCTTC-3′ and R, 5′-GTGACAGATCTCTCAGGGCATCAGGTCTTCACC-3′; BglII ends are compatible with BamHI). Full-length PKCϵ (primers PKCϵF, 5′-GAACACACCCGGGATGGTAGTGTTCAATGGCCTTC-3′ and PKCϵR, 5′-CATCTATGGTACCGGGGCATCAGGTCTTCACCAA-3′), and its regulatory (primers PKCϵR, PKCϵF and 5′-CATATAAGGTACCGCTGGCCTGGCCGGACTTCC-3′) and catalytic domains (primers PKCϵC, 5′-CATATTACCCGGGATGGGAATTGCCAAAGTGCTGGCTG-3′, and PKCϵR) were cloned into pMT2-GST between the XmaI and KpnI sites. Wild-type and constitutively active GFP–PKCα and GFP–PKCζ constructs were obtained from Dr Scott Parkinson (CR-UK, London, U.K.) and GFP–PKCδ was as described previously . Wild-type and constitutively active PKCι constructs were a kind gift from Songhai Shi (MSKCC, New York, U.S.A). The SBD (substrate-binding domain) of mouse Fbw7 (Fbw7-SBD; encodes amino acids 283–629 from GenBank® accession number AF391192) was recovered from the pACT2 vector (see below) and subcloned into pEGFP-C1 (between the KpnI and XmaI sites) using primers complementary to the flanking vector sequence (5′-AATACAGGTACCGGTCATATGGCCATGGAGGCC-3′ and 5′-ATAAGAACCGGTCCCAGTATCTACGATTCATAGATCTCTCG-3′; AgeI ends are compatible with XmaI). pCDNA3.1-FLAG was kindly provided by Dr Sylvie Lachmann (CR-UK, London, U.K.), and pcDNA3-FLAG-Fbw7 by Professor Shigetsugu Hatakeyama (Hokkaido University, Japan). FLAG-tagged Fbw7α and Fbw7αΔBD1 were generous gifts from Dr Bruce Clurman (Fred Hutchinson Cancer Research Center, Seattle, U.S.A.) [compared with NCBI (National Center for Biotechnology Information) Reference Sequence NM_033632, these constructs contain the amino acid substitution A626T and also lack the two C-terminal residues]. These sequences were subcloned into pEGFP-C1 (between the BglII and Sal I sites with primers F7αF, 5′-ATTATAAGATCTATGAATCAGGAACTGCTCTCTGTGG-3′, and F7αR, 5′-TAATATGTCGACTCATTCCACATCAAAGTCCAGC-3′). Fbw7α/Fbw7αΔBD1 phosphorylation-site mutants were generated by site-directed mutagenesis, using a two-step PCR approach. The following primers and their reverse complements were employed: S10A, 5′-GCTCTCTGTGGGCGCCAAAAGACGACG-3′; S10D, 5′-GCTCTCTGTGGGCGACAAAAGACGACG-3′; S18A, 5′-CGAACTGGAGGCGCTCTGAGAGG-3′; and S18D, 5′-CGAACTGGAGGCGATCTGAGAGG-3′ (double-mutants were prepared with sequential rounds of mutagenesis). The K11R and +P16 mutants were generated by one-step mutagenesis using Pfu Turbo (Stratagene), with the following primers and their reverse complements: K11R, 5′-CTCTCTGTGGGCAGCAGAAGACGACGAACTGG-3′ and K11R, 5′-CTCTCTGTGGGCAGCAGAAGACGACGAACTGG-3′. All constructs were validated by complete DNA sequencing (CR-UK Sequencing Facility).
Yeast two-hybrid analysis
Yeast two-hybrid analysis was performed using the MATCHMAKER GAL4 System 3, the AH109 yeast strain and a mouse brain cDNA library (in pACT2), according to the manufacturer's guidelines (Clontech). Full-length mouse PKCϵ was used as the bait (pGBKT7-PKCϵ). Basic yeast medium was prepared using SD medium [0.67% yeast nitrogen base, 2% (w/v) glucose, pH 5.8, with or without 2.2% (w/v) agar], 0.59 g/l CSM (complete supplementary dropout medium; −Ade, −His, −Leu, −Trp and −Ura) (Bio 101) and 20 mg/l uracil. This was supplemented with 40 mg/l adenine, 20 mg/l histidine, 100 mg/l leucine, 50 mg/l tryptophan and/or 20 μg/ml X-α-gal (designated X; Clontech), as indicated. Yeast transformations were carried out using a lithium acetate/single-stranded carrier DNA/polyethylene glycol protocol (Agatep, http://research.bmn.com/tto). pGBKT7-PKCϵ expression in AH109 was confirmed by Western blotting and was shown not to auto-activate the HIS3, ADE2 or MEL1 reporter genes. AH109 were co-transformed with pGBKT7-PKCϵ/pGADT7 as negative controls, or pGBKT7-p53/pGADT7-T antigen as positive controls (Clontech). For the screen, yeast were transformed sequentially, first with pGBKT7-PKCϵ and then with the library; a total of 1.4×105 cfu (colony-forming units) were analysed. Transformed yeast were subjected to three rounds of selection. Transformants were first plated on medium stringency nutritionally selective agar plates (CSM −Trp, −Leu and −His) at 30 °C for 5 days. Positive clones were restreaked on to more stringent selective plates (CSM −Trp, −Leu, −His and −Ade) for a further 3 days, and then on to the highest stringency selective media (CSM −Trp, −Leu, −His, −Ade and −X) for 5 days. Plates were imaged by the CR-UK Photographic Department. Library vectors were rescued from positive clones and sequenced using the primers 5′-TACCACTACAATGGATG-3′ (pACT2 5153–5137, sense) and 5′-AGATGGTGCACGATGCACAG-3′ (pACT2 4939–4920, antisense).
Cell culture, transfection and treatment
COS7 and HeLa cells were cultured as described previously . 293FT cells were obtained from Invitrogen and maintained according to their guidelines. Transfection of plasmid DNA was performed using Lipofectamine™ 2000 (Invitrogen; COS7) or Fugene 6 (Roche; HeLa) as recommended by the manufacturers. Cells were stimulated with 10 μM MG132, 100 nM calyculin A (cells detach and were collected in suspension from the medium) and 400 nM PMA. All untreated controls were performed using an equal volume of DMSO.
RNAi (RNA interference)
RNAi was performed using siRNA (small interfering RNA) duplexes from Dharmacon. siLamin A/C (M-013859–00) was used as a control; siFbw7 targets all three members of the mammalian Fbw7 family and has been described previously . 293FT cells were seeded at 2×105 cells on six-well plates, incubated overnight and then transfected with 50 nM siRNA using Lipofectamine™ 2000, as recommended by the supplier (Invitrogen). Cells were treated or not with 400 nM PMA as indicated and harvested at 3 days post-transfection directly into LDS Sample buffer (see below). Samples were boiled for 5 min and briefly sonicated (power seting 5, Microson sonicator; Misonix) prior to analysis by SDS/PAGE and Western blotting.
GST (glutathione transferase) pull-downs and immunoprecipitations
GST pull-downs and immunoprecipitations were performed as described previously . For FLAG immunoprecipitations, anti-FLAG M2–agarose affinity gel was employed.
SDS/PAGE and Western blotting
SDS/PAGE and Western blotting were performed as described previously . Primary antibodies were diluted as follows: mouse anti-FLAG (M5), 1:1000; mouse anti-GFP antibody (3E1, CR-UK), 1:1000; rabbit anti-GST antibody (sc-459, Santa Cruz), 1:2000; mouse anti-PKCα antibody (MC5, CR-UK), 1:1000; rabbit anti-PKCα phospho-Thr638 antibody (9375, Cell Signaling), 1:1000 plus 5% (w/v) BSA; rabbit anti-PKCϵ antibody (sc-214, Santa Cruz), 1:5000; rabbit anti-atypical PKC antibody (C20, Santa Cruz), 1:5000; and mouse anti-cyclin E antibody (HE12, sc-247, Santa Cruz), 1:1000. Two phosphospecific antibodies were also raised against Fbw7α (see below) and used at 1:1000 with 5% (w/v) BSA and 1 μg/ml of the corresponding, blocking dephosphopeptide: anti-phospho-Ser10 antiserum (PPA-574) and anti-phospho-Ser18 antiserum (PPA-527). The Western blots presented are each representative of results obtained from at least three separate experiments.
In vitro kinase assays
The following peptides were prepared for use as PKC substrates: Fbw7α wild-type 7–14, N-SVGSKRRR-C; Fbw7α-S10A 7–14, N-SVGAKRRR-C; Fbw7α wild-type 11–22, N-KRRRTGGSLRGN-C; and Fbw7α-S18A 11–22, N-KRRRTGGALRGN-C (Peptide Synthesis Facility, CR-UK). Reactions were performed using 50 μl of the following kinase assay buffer: 20 mM Tris/HCl, pH 7.5, 5 mM MgCl2, 1 mg/ml phosphatidylserine (Lipid Products), 0.2% Triton X-100, 1 ng/μl PMA, 50 μM ATP, 5 μCi [γ32P]ATP (Amersham), with 0.2mM peptide substrate and 12.5 ng of PKCϵ. Assays were started with the addition of ATP, incubated at 30 °C for 10 min, and stopped by spotting 15 μl on to P81 paper (Whatman) and washing three times for 5 min each in 30% acetic acid. P81 strips were analysed by Cerenkov counting using a Beckman LS6000IC scintillation machine to quantify 32P incorporation.
Preparation of phosphospecific antibodies
Rabbit polyclonal antisera were raised using the following phosphopeptides as antigens: Fbw7α phospho-Ser10, N-SVGpSKRRR-C; and Fbw7α pS18, N-RTGGpSLRGN-C (corresponding blocking dephosphopeptides were also synthesized). Each phosphopeptide was coupled to KLH (keyhole limpet haemocyanin) using glutaraldehyde and immunized in triplicate (Harlan Sera-Lab). The resulting sera were characterized by ELISA and found to exhibit both site- and phospho-specificity against phospho-Ser10 (serum PPA-574) or phospho-Ser18 (serum PPA-527). An additional peptide was used to determine whether serum PPA-574 could recognize phospho-Ser10 in the context of a K11R mutation (pS10 K11R, N-SVGpSRRRR-C).
Epifluorescent and confocal microscopy
Immunofluorescence was performed as described previously . For propidium iodide staining of the nuclei, coverslips were incubated with 1 mg/ml RNase (Qiagen)/PBS for 15 min, followed by 0.25 μg/ml propidium iodide/PBS for a further 10 min. Alternatively, 1 μg/ml Hoechst 33342 (Invitrogen)/PBS was used for 10 min to stain DNA. Coverslips were washed three times for 10 min each in PBS and once in water, and mounted on to glass slides using Mowiol 4–88 (Calbiochem) or Aqua Poly/Mount (Polysciences). Confocal images were acquired using a confocal, laser scanning microscope (LSM 510, Zeiss) equipped with a 63×/1.4 Plan-Apochromat Oil Ph3 objective. Individual channels were scanned sequentially, line by line, with averaging set at eight. Each image (12 bit, 1024 pixels) represents a single 1.0 μm optical z-section with a pixel size of approx. 0.1 μm. Epifluroescent images were acquired using an Imager.A1 upright microscope (Zeiss) equipped with a Plan-Apochromat 63×/1.4 Oil objective and an Orca-ER Digital Camera (Hamamatsu) using Axiovision software (Release 4.6.3). All images presented are representative of results obtained from at least three independent experiments.
Quantification of Fbw7α nuclear localization
Confocal images were captured as described above except that complete z-stacks were acquired for each field of view. Each z-stack was compressed and analysed using Metamorph. The nuclear compartment was defined using the propidium iodide signal; the GFP–Fbw7α signal was then quantified both inside and outside of this region. Nuclear localization was expressed as the ratio of the mean integrated GFP signal intensity outside compared with inside the nucleus. At least 20 cells were analysed per experiment to obtain a mean ratio, and the S.E.M. was determined across three separate experiments. Student's t tests were performed to determine statistical significance using Prism software.
Fbw7 amino acid sequences from a variety of species were retrieved from the NCBI Entrez-Protein Database using the following accession numbers: Homo sapiens, NP_361014; Macaca mulata, XP_001084190; Rattus norvegicus, EDM00812; Mus musculus, AAH85184; Gallus gallus, XP_420447; and Xenopus laevis, NP_001089186. The accession number for human cyclin E is NP_476530 and the different isoforms of human PKC are as follows: PKCα, NP_002728; PKCβI, NP_997700; PKCβII, NP_002729; PKCδ, NP_997704; PKCϵ, NP_005391; and PKCζ, NP_002735. Sequence alignments were performed using ClustalW software, with default settings, according to EMBL-EBI recommendations.
The catalytic domain of PKC interacts with the SBD of Fbw7
PKCϵ regulates a diverse range of cellular processes, including proliferation, survival, migration and invasion . Aberrant PKCϵ activity is associated with a transformed phenotype and its expression is a biomarker for aggressive breast cancer . To gain further insights into the physiological and pathophysiological functions of PKCϵ, we performed a yeast two-hybrid screen to identify novel binding partners.
Full-length PKCϵ bait was used to screen a mouse brain cDNA library, yielding 28 in-frame positive clones, including a known PKCϵ-interacting protein, Nell2 , which validated the approach. Among the other hits, ‘clone 19’ was identified as the C-terminal SBD of Fbw7 (Figures 1a and 1b). Fbw7 is a tumour suppressor that mediates the degradation of several potent oncoproteins to control cell division, growth, differentiation and transformation . As such, Fbw7 was selected as an interesting candidate for further analysis.
Fbw7 interacts with, but does not down-regulate, multiple members of the PKC superfamily
To verify and map the interaction between PKCϵ and Fbw7 in mammalian cells, GST pull-downs were performed (Figure 1c). Cells were co-transfected with the GFP–Fbw7-SBD, and full-length or truncated PKCϵ–GST. Pulling down with a GST control or the regulatory domain of PKCϵ recovers only background amounts of GFP–Fbw7-SBD, whereas significantly elevated levels are recovered when using either full-length PKCϵ or its catalytic domain. Since the latter two constructs are more weakly expressed, the efficiency of these pull-downs is particularly striking. We conclude that the catalytic domain of PKCϵ associates with the SBD of Fbw7 in mammalian cells.
The catalytic domain is highly conserved among PKC superfamily members, raising the possibility that Fbw7 may also interact with other isoforms. To test this, cells were co-transfected with full length FLAG–Fbw7 and a panel of different GFP-tagged PKC isoforms, and analysed by co-immunoprecipitation (Figure 1d). Since PKC represents a potential substrate for the SCFFbw7 ubiquitin ligase, immunoprecipitations were performed in the presence or absence of MG132, a proteasome inhibitor, to protect any ubiquitinated PKC from degradation. Fbw7 was immunoprecipitated using anti-FLAG antibody, and immunoprecipitates were probed for associated PKC using anti-GFP antibody. All the PKC isoforms tested bound to Fbw7; the strongest interaction detected was with the atypical PKCζ. Total levels of PKC (in cell lysate) did not change upon Fbw7 co-expression or MG132 treatment, suggesting that PKC is not degraded under these conditions. The increase in immunoprecipitated PKC following MG132 treatment is explained by the stabilization of Fbw7 itself (which is known to undergo auto-ubiquitination).
The results above suggest that PKC is unlikely to be a substrate for Fbw7. However, it is striking that the CPD motif, recognized in all Fbw7 targets, shares significant homology with the ‘TP’/turn motif of PKC (Figure 1e), one of three critical ‘priming’ phosphorylation sites required for its activation . To test directly whether endogenous Fbw7 can regulate the degradation of PKC, an RNAi-based approach was employed. For these experiments, we reproduced conditions established by van Drogen et al. , who designed an siRNA duplex to target the Fbw7 common region, and demonstrated co-depletion of the mammalian isoforms from HEK-293T cells [HEK-293 cells expressing the large T-antigen of SV40 (simian virus 40)], by qRT-PCR (quantitative real-time PCR) (for endogenous Fbw7) and by Western blotting (for exogenous tagged Fbw7). To validate the siRNA duplex, we sequentially transfected cells, first with siControl or siFbw7, and then with GFP alone or GFP–Fbw7α. As expected, the siFbw7 duplex led to a specific depletion of GFP–Fbw7α (Figure 1f).
Using these conditions, cells were co-depleted of all three Fbw7 isoforms and subjected to Western blotting for endogenous PKC isoforms, with cyclin E as a positive control (Figure 1g). Lysates were also probed with a phosphospecific antibody against the PKCα turn motif (as if this site acts as a CPD, it may be expected to provide a more sensitive marker of degradation). Finally, to analyse both basal and regulated degradation, these experiments were performed in the presence or absence of phorbol ester (PMA), which stimulates the activation-induced down-regulation of classical PKCs (e.g. PKCα) and novel PKCs (e.g. PKCϵ) isoforms . As shown in Figure 1(g), depletion of Fbw7 led to a clear stabilization of its substrate cyclin E. However, levels of both total and phospho-PKC were unchanged by Fbw7-knockdown, suggesting that none of these isoforms represent Fbw7 substrates. PMA treatment induced the complete degradation of PKCα, and a partial reduction in PKCϵ levels after 20 h; however, neither of these events was inhibited by depletion of Fbw7, indicating that Fbw7 is not responsible for PMA-stimulated PKC degradation.
We conclude that Fbw7 interacts with multiple members of the PKC superfamily but does not target them for either basal or activation-induced degradation.
PKC phosphorylates Ser10 and Ser18 within the unique N-terminal domain of Fbw7α in vitro and in mammalian cells
We next addressed whether Fbw7 may be a substrate for PKC-mediated phosphorylation. First, Fbw7 protein sequences were analysed using a predictive database for PKC phosphorylation, MPR PKC Scan . Two of the highest scoring sites identified were Fbw7α residues Ser10 and Ser18, which are in the top 0.06 and 0.2 percentiles respectively for PKCζ phosphorylation. These residues lie within the isoform-specific N-terminal region of Fbw7α, which plays an important role in determining the unique nuclear localization of this isoform; NLS1 (nuclear localization signal 1) lies nearby, between residues 11 and 14 (Figure 2a). Furthermore, this region is altered in certain cancer samples; the mutations K11R and +P16 have been reported . We reasoned that PKC-dependent phosphorylation within this regulatory region may be of functional significance and sought to study these candidate sites further.
PKC phosphorylates the isoform-specific N-terminal region of Fbw7α in vitro and in cells
To test whether PKC can phosphorylate Fbw7α, in vitro kinase assays were performed (Figure 2b). Peptides were synthesized corresponding to residues 7–14, to test for Ser10 phosphorylation, and residues 11–22, for Ser18 analysis (in control peptides the candidate serine residue was substituted with a non-phosphorylatable alanine residue). Peptide kinase assays were carried out using purified PKCϵ and [γ-32P]ATP and incorporation of 32P was detected by Cerenkov counting. Both wild-type peptides were efficiently phosphorylated compared with the corresponding alanine mutants, confirming the prediction that Fbw7α residues Ser10 and Ser18 can be phosphorylated directly by PKC. Similar results were obtained using recombinant PKCα, PKCδ or PKCζ (results not shown), suggesting that Fbw7α is a common substrate for PKC superfamily members.
To determine whether Fbw7α is phosphorylated in mammalian cells, phosphospecific antibodies were raised. Cells were transfected with GFP–Fbw7α constructs and analysed by Western blotting (Figure 2c). As a broad stimulus, cells were treated with or without calyculin A, an inhibitor of the PP1 and PP2A phosphatases (this compound protects sensitive sites from dephosphorylation, and may also up-regulate certain signalling pathways by preserving activating phosphorylations). As revealed by the phosphospecific antisera, phosphorylation of both Ser10 and Ser18 was induced by calyculin A treatment in all constructs, except for the corresponding alanine mutants. These results confirm that Fbw7α residues Ser10 and Ser18 can be phosphorylated in mammalian cells in a mutually independent fashion, and also serve to illustrate the site specificity of the two antisera.
To test whether Fbw7α phosphorylation is dependent upon PKC in a cellular context, a pharmacological approach was taken. Since Fbw7α can be phosphorylated by multiple PKC isoforms, we employed the pan-PKC inhibitor Gö6983. Upon pre-treatment with Gö6983, the calyculin A-induced phosphorylation of Ser10 and Ser18 decreased by approx. 20% and 40% respectively, consistent with a partial requirement for endogenous PKC activity (Figure 2d). We note that calyculin A will tend to cause phosphorylation to ratchet up inexorably, such that the observed response to inhibition of the upstream kinase(s) may be compromised, given that inhibition is never complete. Notwithstanding this caveat, because these phosphorylation events are only partially inhibited by Gö6983, it is possible that multiple kinase activities may converge on Fbw7α phosphorylation under these calyculin-A-triggered conditions.
To investigate whether PKC activity is sufficient to promote the cellular phosphorylation of Fbw7α, we tested the effect of co-expressing wild-type, or constitutively active, PKCα or PKCι. Importantly, the overexpression and overactivation of PKCα  and PKCι [22,30] is observed in range of human cancers, so these experiments effectively model the status of certain tumours . Expression of wild-type PKCα or PKCι induced Fbw7α phosphorylation at both Ser10 and Ser18 (Figures 2e and 2f), and this was dramatically increased by the constitutively activated kinases; similar results were also observed using PKCζ (Figure 2g). Together, these results confirm that PKC activity is able to drive Fbw7α phosphorylation in mammalian cells.
Finally, to establish whether nearby cancer-associated mutations may influence these phosphorylation events, GFP–Fbw7α-K11R and +P16 mutants were analysed as described above. Fbw7α +P16 was phosphorylated by the atypical PKCζ at both Ser10 and Ser18, as efficiently as the wild-type protein. Fbw7α-K11R was also phosphorylated at Ser18 (Figure 2g). The phosphorylation of Fbw7α-K11R at Ser10 could not be monitored, as this mutation impairs the recognition of phospho-Ser10 by our antiserum (demonstrated by ELISA; results not shown). However, since the K11R substitution preserves a basic charge at this residue, it is unlikely to affect PKC-mediated phosphorylation; in fact, Ser10 scores even more highly as a candidate phosphorylation site in a K11R mutant (0.05 percentile) than in the wild-type protein (0.06 percentile) according to the MPR PKC Scan database.
We conclude that PKC phosphorylates residues Ser10 and Ser18 in the isoform-specific N-terminal region of Fbw7α, both in vitro and in mammalian cells. These phosphorylation events are not mutually dependent or affected by the cancer-associated mutations K11R and +P16.
Phosphorylation of Ser10 can regulate Fbw7α nuclear localization
We next sought to examine the functional consequences of Fbw7α phosphorylation at Ser10 and/or Ser18. These residues lie in close proximity to NLS1 (Figure 2a) and we reasoned that phosphorylation may thereby influence Fbw7 localization. Regulation of Fbw7 localization is of particular interest because it is intimately linked to its profile of substrate down-regulation, determining which of many potential targets the ubiquitin ligase is exposed to. Fbw7α contains two NLSs, NLS1 in the isoform-specific N-terminus and NLS2 in the shared SBD (Figure 3a). Either of these signals can confer nuclear localization and both must be lost to shift Fbw7α to the cytosol in U2OS cells . Thus, to analyse regulation of NLS1 by phosphorylation, we employed a mutant lacking NLS2 (ΔNLS2) so that the nuclear localization of this protein is completely dependent on a functional NLS1, and any regulatory inputs should manifest clearly.
Ser10 phosphorylation can regulate Fbw7α nuclear localization
Cells were transfected with GFP–Fbw7α constructs in which Ser10 and/or Ser18 were mutated to either non-phosphorylatable alanine or phosphomimetic aspartate residues, in the presence or absence of an intact NLS2. GFP–Fbw7α localization was then analysed by confocal microscopy (Figure 3b and results not shown). Quantitative analyses were performed by measuring the GFP signal intensity inside and outside of the nucleus, as judged by co-localization with DNA staining; the ratio between these values represents Fbw7α localization (Figure 3c). Consistent with previous reports, wild-type Fbw7α was localized almost exclusively in the nucleus (the same pattern of localization was observed using FLAG–Fbw7α; results not shown and ). None of the phosphomutants showed an altered pattern of localization in an otherwise wild-type context (NLS2wt constructs). The ΔNLS2 mutant was also predominantly nuclear, although a weak cytosolic signal was detected, suggesting that NLS1 alone is not completely sufficient for proper localization of the protein (the difference between the wild-type/wild-type and wild-type/ΔNLS2 localization is statistically significant; P=0.001). In the context of the ΔNLS2 protein, the phosphomimetic aspartate mutation of Ser10, but not Ser18, led to a dramatic cytosolic redistribution of the protein, indicating that this modification severely impairs the activity of NLS1. Furthermore, the modest cytosolic signal detected with the ΔNLS2 mutant was significantly reduced when Ser10 was mutated to an alanine residue; S10A/ΔNLS2 mutants localize more like the wild-type protein (Figure 3c). This observation suggests that endogenous phosphorylation of Ser10 may underlie the reduced efficiency of NLS1 as a single nuclear localization signal.
We conclude that phosphorylation of Fbw7α at Ser10 strongly inhibits the activity of its isoform-specific nuclear localization signal, NLS1. However, for this modification to result in a cytosolic redistribution of the protein, NLS2 activity must also be suppressed.
Cancer-associated mutations can regulate Fbw7α nuclear localization
Since the cancer-associated Fbw7α mutations, K11R and +P16, lie close to NLS1, we reasoned that these modifications may also regulate nuclear localization. To test this, GFP–Fbw7α constructs harbouring these mutations, in the presence or absence of NLS2, were transfected and visualized by epifluorescent microscopy (Figure 4a). Both mutants localized to the nucleus in the presence of an intact NLS2 (NLS2wt). However, in the absence of NLS2 (ΔNLS2), these mutations relocalized the protein; both K11R and +P16 mutants were distributed throughout the cell and resembled the S10D mutant with respect to localization.
Cancer-associated mutations can regulate Fbw7α nuclear localization
We conclude that mutations in the isoform-specific N-terminal of Fbw7α (K11R and +P16) can regulate the activity of NLS1 in a manner comparable with phosphorylation of Ser10. Furthermore, since these mutations were identified in tumours, relocalization of Fbw7α may be of pathophysiological significance.
In the present study, we explored the relationship between two protein families, Fbw7 and PKC, whose dysregulation has been associated with cancer. Through a series of biochemical and cell-based assays, we uncover a physical and functional relationship between these proteins that may play a role during transformation.
Fbw7 was identified as a PKCϵ-binding partner by yeast two-hybrid analysis, indicating that these proteins interact directly. In mammalian cells, the interaction was mapped to the catalytic domain of PKC and the SBD of Fbw7, which is an unusual eight-bladed β-propeller composed of WD40 repeats . The PKC kinase domain is highly conserved, and consistent with this, all isoforms tested are able to interact with Fbw7. In the absence of an effective anti-Fbw7 antibody, we were unable to confirm the interaction between endogenous proteins. However, we note that in a large-scale study of tight-junction components, Fbw7β was detected by MS following immunoprecipitation of endogenous PKCζ from T84 cells . These results support the conclusion that PKC and Fbw7 interact under physiological conditions.
We were initially intrigued by the possibility that Fbw7 may target PKC for proteasomal degradation. PKC is well known to undergo activation-induced down-regulation and this may contribute to transformation under certain conditions . However, to date, only two ubiquitin ligases have been identified that can mediate degradation of selected PKC isoforms, VHL (von Hippel–Lindau protein)  and RINCK (RING-finger protein that interacts with C kinase) . The PKC TP/‘turn’ motif bears striking similarity to the CPD sequence recognized by Fbw7 in its substrates, consistent with the notion it may represent a novel target. However, despite this, several lines of evidence suggest that PKC does not represent an Fbw7 substrate: (i) all isoforms of PKC interact with Fbw7, although only some possess a well-conserved CPD; (ii) expression of the Fbw7 SBD does not stabilize PKC expression, although it should act as a dominant-negative with respect to the SCFFbw7 ubiquitin ligase; (iii) expression of full-length Fbw7 does not down-regulate PKC; and (iv), most importantly, depletion of endogenous Fbw7 does not protect PKC from either basal, or activation-induced, down-regulation.
It is important to note that our results do not completely exclude Fbw7 as a possible PKC ubiquitin ligase. First, interfering with Fbw7 function does not always manifest in detectable changes in steady-state substrate levels; for instance, although c-Myc turnover is compromised in an Fbw7α-null Hct116 cell line, protein levels do not change, due to a compensatory decrease in mRNA levels . Secondly, the profile of Fbw7 substrate down-regulation also exhibits some cell-type specificity, such that PKC may be a substrate under different conditions to those tested. For instance, Fbw7γ degrades nucleolar c-Myc in U2OS , but not in Hct116 cells , and Fbw7 degrades c-Jun in HEK-293T , but not in Hct116 cells . Nevertheless, in the absence of clear evidence for Fbw7-mediated PKC degradation, we focussed our attention on the reciprocal relationship and investigated whether PKC can phosphorylate Fbw7.
We have identified Fbw7α as a novel PKC substrate. Two residues, Ser10 and Ser18, are phosphorylated in a PKC-dependent manner, both in vitro and in mammalian cells. In vitro, Fbw7α can be phosphorylated by multiple PKC isoforms, suggesting it is a common substrate for the PKC superfamily. In mammalian cells, the overexpression of either active classical PKCα or atypical PKCι is sufficient to drive robust Fbw7α phosphorylation. Since the overexpression and overactivation of PKCα  and PKCι  is observed in a range of human tumours, we conclude that Fbw7α phosphorylation may be elevated in certain cancers. Notably, PKCι is a known oncogene  and its amplification is one of the most common somatic alterations in human cancer . As such, it will be important to investigate whether aberrant Fbw7α phosphorylation contributes to the pathophysiological function of PKCι.
The two novel Fbw7α phosphorylation sites identified in the present study, Ser10 and Ser18, lie within the isoform-specific N-terminus of Fbw7α, which is the most abundant Fbw7 family member and bears primary responsibility for the degradation of several critical substrates . Importantly, both of these phosphorylation sites are evolutionarily conserved (Figure 5a), suggesting that they may be of functional significance. Future work will aim to determine whether phosphorylation of Fbw7α influences its dimerization , assembly into an SCF complex, auto-ubiquitylation  or ubiquitylation of other substrates. In the present study, we investigated the possibility that Fbw7α phosphorylation may influence its localization.
Model of Fbw7α regulation
Ser10 is adjacent to NLS1, one of the two NLS motifs in Fbw7α that confer nuclear localization to the protein. It is well-established that phosphorylation can influence the activity of an NLS, for instance, PKC-catalysed phosphorylation proximal to the AhR (aryl hydrocarbon receptor) NLS inhibits its nuclear import . Consistent with this, mutation of Ser10, but not Ser18, to a phosphomimetic aspartate residue severely inhibits the activity of NLS1. Strikingly, cancer-associated mutations in nearby residues (K11R and +P16) disrupt Fbw7α NLS1 function in a comparable manner. Although the K11R substitution preserves a basic charge, structural and thermodynamic studies confirm that a lysine residue is strictly required in the +1 position of a monopartite classical NLS . The +16 insertion of a rigid proline residue would yield a significant change in local structure and so it is perhaps not surprising that this mutation also interferes with the function of NLS1. Importantly, since these mutations were identified from somatic events in tumours, these observations indicate that aberrant function of NLS1 may be of pathological significance. By extension, we surmise that dysregulation of Fbw7α by PKC-mediated phosphorylation may also contribute to transformation under certain conditions.
As noted above, Fbw7α bears two redundant nuclear localization signals (NLS1 and NLS2) and it is necessary to suppress the activity of both to shift the protein outside of the nucleus . We have identified three distinct mechanisms through which NLS1 can be inhibited: phosphorylation of Ser10 by PKC, substitution of residue Lys11 for an arginine residue, and insertion of a proline residue at position 16 (Figure 5b). Under conditions where NLS2 activity is suppressed, these regulatory inputs result in a dramatic relocalization of Fbw7α to the cytosol. Further work will be required to explore whether NLS2 is also subject to regulation, by mutation, post-translational modification or masking by protein–protein interactions . Since it lies within the conserved SBD, an interesting possibility is that engagement of certain substrates may mask NLS2. Related to this, we note that the +P16 mutant, which is nuclear in HeLa cells (Figure 4), is reported to localize to the cytosol in HEK-293T cells, even in the absence of other alterations . On the basis of these results, we hypothesized that the presence of large T-antigen, which binds competitively to the SBD , may ameliorate the activity of NLS2 in these cells. However, we have been unable to detect cytosolic +P16 Fbw7α in HEK-293T cells, so could not explore this possibility further. Nevertheless, these results suggest that conditions exist in which NLS1 is solely responsible for Fbw7α nuclear localization, and that the regulatory inputs that influence NLS2 activity warrant further investigation.
The mammalian Fbw7 family members are localized to different subcellular compartments by regulatory signals in their isoform-specific N-terminal regions . This suggests that each isoform performs a unique function in a distinct region of the cell. Indeed, Fbw7γ is localized to the nucleolus, and in this subcellular compartment degrades a specific pool of c-Myc in order to control U2OS cell size . Furthermore, Fbw7α degrades full-length cyclin E in the nucleus, but does not down-regulate certain low-molecular-mass forms that are localized in the cytoplasm . Mislocalization of Fbw7α to the cytosol, a compartment usually occupied by the much less abundant Fbw7β isoform [14,16], is therefore expected to have significant effects on its pattern of substrate down-regulation. Under these conditions, full-length cyclin E and other nuclear substrates may be spared from degradation, while cytosolic targets, such as Notch, may be exposed to higher levels of the ubiquitin ligase. Future work will be aimed at investigating such changes by differential proteomics, for instance by employing a SILAC (stable isotope labelling by amino acids in cell culture)-based strategy. The combined effects of elevated nuclear and decreased cytosolic substrate expression would be expected to have profound consequences for cell proliferation, growth and differentiation. Furthermore, since many nuclear Fbw7 substrates are potent oncoproteins, relocalization of Fbw7α is also likely to drive transformation, ultimately with a similar outcome to loss or mutation of the protein. Related to this, we note that overexpression/activation of PKCι occurs concurrently with a substantial up-regulation of cyclin E in certain ovarian tumours  and it will be interesting to explore whether mislocalization of Fbw7α through Ser10 phosphorylation contributes to this phenotype.
Our current findings integrate Fbw7α into a novel PKC-dependent signal transduction pathway. In light of this, it is interesting to consider the place of Fbw7 within a wider network of signalling events. For instance, we note the potential for interplay between PKC, Fbw7 and GSK3. GSK3 is the kinase responsible for phosphorylating most, if not all, of the CPD motifs recognized by Fbw7 in its substrates , and phosphorylation of GSK3 by PKC has been shown to inhibit its activity . As such, it is plausible that PKC activity may inhibit GSK3-mediated CPD phosphorylation, thereby protecting potential substrates from Fbw7 recognition. Notably, PKC activation results in reduced phosphorylation of c-Jun by GSK3 in the region of Ser243 , the central phosphoserine of the c-Jun CPD . Thus we can conceive two possible mechanisms through which PKC activity may antagonize Fbw7 function: (i) indirectly, by inhibiting the GSK3-mediated phosphorylation events which target substrates for Fbw7 recognition; and (ii) directly, by phosphorylating Fbw7α itself, to relocalize the protein and protect nuclear substrates from ubiquitination and degradation. Thus stabilization of Fbw7 substrates may partly underlie the tumour-promoting properties of PKC activity.
In conclusion, our current results provide evidence for both physical and functional interactions between PKC and Fbw7α, controlling the function of NLS1 in the latter case. Future work will address the regulation of Fbw7α NLS2, in order to define the combinatorial events which may control its nuclear uptake.
Cancer Research UK
complete supplementary dropout medium
F-box WD40 protein 7
green fluorescent protein
glycogen synthase kinase 3
human embryonic kidney
HEK-293 cells expressing the large T-antigen of SV40 (simian virus 40)
National Center for Biotechnology Information
nuclear localization signal
protein kinase C
small interfering RNA
Joanne Durgan and Peter Parker conceived the experiments and wrote the manuscript; Joanne Durgan performed all of the experiments.
We thank Professor Shigetsugu Hatakeyama, Dr Songhai Shi and Dr Bruce Clurman for providing us with plasmids. We are grateful to Phil Whitehead for preparing purified recombinant PKCs, and to Peter Jordan and Alastair Nichol for their assistance with confocal microscopy and image analysis respectively. We also express thanks to Axel Behrens, Annett Jandke and members of the Parker laboratory for helpful discussions, and to Professor Alan Hall for supporting this work.
This work was supported by a PhD fellowship from the Medical Research Council (to J.D.) and by CR-UK.