Distinct Interactions of EBP1 Isoforms with FBXW7 Elicits Different Functions in Cancer

The ErbB3-binding protein EBP1 is a member of the proliferation-associated 2G4 protein family and is ubiquitously expressed in all human tissues and is involved in the regulation of cell growth and differentiation (1, 2). EBP1 encodes two different protein isoforms, the long form P48 and short form P42, due to alternative splicing of pre-mRNA, which are believed to have different cellular activities (3–5). The predominant P48 isoform can promote cell proliferation or survival including cancer cells (4, 6) and localizes to both cytoplasm and nucleus (5). Moreover, correlation of high expression level of P48 with poor clinical outcomes in patients with pancreatic ductal adenocarcinoma (6), brain tumor (4), cervical (7), and prostate cancer (8) suggests that P48 is closely related with cancer progression. In contrast, the P42 isoform, which lacks the N-terminal 54 amino acids, is considered to be a potent tumor suppressor because of its growth-inhibitory function (3, 5, 9). Although it is reported that P48 has an oncogenic activity and P42 plays a suppressive role in various cancer cells, the underlying mechanism regarding the distinctive functions of the 2 isoforms of EBP1 in cancer remains largely unclear.

FBXW7 (F-box and WD40 domain protein 7) functions as a substrate recognition subunit of the SCF (SKP1/CUL1/F-box protein) E3 ubiquitin ligase complex (10) to regulate a network of proteins with central roles in cell division, growth, and differentiation (11–13). FBXW7 contains two important functional domains: the F-box domain, which interacts directly with SKP1 to recruit ubiquitin-conjugating enzymes, and the WD40 domain, which binds to a consensus phosphor-binding motif called the CDC4 phosphodegron (CPD) in its substrates (14, 15). FBXW7 has been characterized as a general tumor suppressor. FBXW7 mutation or deletion is often observed in multiple human cancers including colorectal cancer (16, 17), and loss of FBXW7 function results in tumorigenesis (18–23). Accumulating data indicate that FBXW7 exerts its antitumor function mainly through targeting multiple oncoproteins for ubiquitination and proteasomal degradation, such as cyclin E (24), Notch (25), mTOR (26), Aurora A (27), and c-Myc (28). However, a detailed understanding of the full set of FBXW7 substrates and the mechanisms that link FBXW7 deficiency to tumorigenesis is still lacking.

In this study, we used proteomics approach to globally screen FBXW7-regulated proteins in colorectal cancer cells, and we first identified that EBP1 P48 expression was dramatically upregulated in FBXW7-deficient cells. Further mechanism study showed that FBXW7 physically binds to EBP1 P48 and facilitates the ubiquitination and degradation of P48 through the proteasome pathway in a GSK3β-dependent manner. Interestingly, we also found that FBXW7 is able to interact with EBP1 P42 but fails to result in the degradation of P42. On the contrary, the tumor-inhibitory function of FBXW7 was enhanced by P42 but attenuated by P48. The underlying mechanical protein interaction among the three proteins was further investigated. Therefore, our data demonstrate a physical and functional protein interaction network among FBXW7 and EBP1 isoforms, which provide novel mechanistic insights into the distinctive functions and regulations of EBP1 isoforms.

HCT116 (FBXW7^+/+ and FBXW7^−/−) and DLD-1 (FBXW7^+/+ and FBXW7^−/−) cell lines were kind gifts from B. Vogelstein and cultured in McCoy 5A supplemented with 10% FBS. MDA-MB-468, PC3, DU145, and HEK293T cells were purchased from the ATCC, where they were characterized by DNA fingerprinting and isozyme detection. MDA-MB-468 and HEK293T cells were maintained in DMEM with 10% FBS and PC3 and DU145 cells in RPMI-1640 medium with 10% FBS. All cells were grown at 37°C with 5% CO₂/95% air atmosphere and were revived every 3 to 4 months.

Wild-type (WT) HA-FBXW7α, FBXW7α mutants (ΔWD, ΔF,), shFBXW7 A, and shFBXW7 B plasmids were previously constructed by our laboratory (Supplementary Fig. S1C; refs. 21, 26, 29). The shRNA sequences for FBXW7 are as follows: shFBXW7 A: GGCAACAACGACGCCGAAT; shFBXW7 B: GAGTTGGCACTCTATGTGC.EBP1 P48 and P42 human cDNA were subcloned using the PrimeSTAR Max DNA polymerase (TaKaRa) into the Myc-CMV₁₀ vector and Flag-CMV₁₀ vector. FBXW7αNΔF were subcloned into the pCGN-HA vector. Flag-EBP1 P48 S40A, Flag-EBP1 P48 S44A, HA-FBXW7α mutants (R465C, R465H, R479H, R505C), and HA-FBXW7αΔF^R465C were generated using PCR-directed mutagenesis method. Ubiquitin gene was amplified and subcloned into Myc-CMV₁₀ vector and pCGN-HA vector. The shRNA plasmids specific for EBP1 P48 were purchased from Genechem with the following target sequences: shP48.1: GAGCAACAGGAGCAAACTA; shP48.3: GGTGGAAGCATCTAGCTCA.

The antibodies used in our experiments are listed in Supplementary Table S1. CHX, MG132, GSK3β inhibitor VIII, and protein G/A were from Calbiochem. Protease inhibitor cocktail was from Roche. TRIzol and lipo-2000 were from Invitrogen. RIPA was from Beyotime Biotechnology.

Whole-cell lysates were prepared using RIPA lysis buffer supplemented with protease inhibitors. Nuclear and cytoplasmic extracts were prepared using a nuclear and cytoplasmic protein extraction kit (Beyotime Biotechnology) according to the manufacturer's instructions. Western blotting and immunoprecipitation were performed as described previously (30).

RNA was isolated using the TRIzol reagent. The reverse transcription reaction was performed using the RevertAid First Strand cDNA Synthesis Kit (Thermo). qRT-PCR was performed using the SYBR Green PCR Master Mix (Thermo) and the ABI PRISM 7900HT Real-time PCR Detection System (Eppendorf).

Fifty samples of colorectal adenocarcinoma tissues and corresponding adjacent non-cancerous tissues were obtained from patients undergoing surgical excision of tumors in Qilu hospital of Shandong University (Jinan, China). The samples were fixed with 10% formalin and embedded in paraffin and sliced into 5-μm sections. The sections were processed and stained as described previously (30). Antibodies against N-EBP1 P48 and FBXW7 were diluted 1:300. Staining was observed in 5 randomly selected high-power fields. The staining intensity was based on the average percentage of positive cells. The scoring results were analyzed by 2 investigators.

MTT and colony formation assays were performed to assess cell proliferation. Wound-healing and invasion assays were used to examine cell motility as described (30). Cells for motility assays were cultured in a low-serum concentration (0.2%) to exclude the affection of proliferation on cell motility.

Nude mice (6 weeks old) were purchased from Shanghai Slac Laboratory Animal Co. Ltd. and maintained in microisolator cages. All animals were used in accordance with institutional guidelines, and the current experiments were approved by the Use Committee for Animal Care of the institute. For metastatic assays, HCT116 FBXW7^+/+, FBXW7^−/− and FBXW7^−/− shEBP1 P48.1 cells (3 × 10⁶) were injected into the tail veins. After 40 days, the mice were sacrificed by anesthesia with chloral hydrate. The lung and liver were dissected out for hematoxylin and eosin staining.

Results are expressed as mean ± SD from at least three independent experiments. SPSS17.0 statistical software package (SPSS Inc.) was used for statistical analysis. Statistical differences between groups were assessed using the Student t test. Association between FBXW7 and EBP1 P48 expression in colorectal cancer tissue was evaluated by the Spearman rank correlation test. P < 0.05 was considered statistically significant.

To screen new substrates recognized by FBXW7, we performed two-dimensional gel electrophoresis (2-DE) and mass spectrometry (MS) in human colorectal cancer cell line HCT116 with homozygous deletion of the FBXW7 gene (HCT116 FBXW7^−/−; ref. 31). Several differentially expressed proteins were separated and identified in HCT116 FBXW7^−/− and FBXW7^+/+ cells, of which, EBP1 protein was most significantly elevated in HCT116 FBXW7^−/− cells (Supplementary Fig. S1A and S1B). The altered expression of EBP1 was validated in HCT116 and DLD-1 FBXW7^−/− cells by Western blotting using 2 EBP1 antibodies, anti-C-EBP1 to detect both EBP1 isoforms and anti-N-EBP1 P48 to specific P48. Interestingly, EBP1 P48 but not P42 is significantly elevated in the FBXW7-deficient cells (Fig. 1A and B). Restoration of FBXW7α WT reduces the endogenous P48 level, which was not found in the cells restored with truncated dominant negative form (FBXW7F; Fig. 1C, the diagrammatic drawings of FBXW7α mutants were shown in Supplementary Fig. S1C). Moreover, elevated P48 protein levels were also observed in other cancer cell lines such as MDA-MB-468, DU145, and PC3 cells when silencing FBXW7 expression by two different FBXW7 shRNA constructs (shFBXW7A and B; Supplementary Fig. S1D–S1F). To confirm the differential regulation of FBXW7 on EBP1 isoforms, Myc-tagged P48 (Myc-P48) or P42 (Myc-P42) were cotransfected into HEK293T cells along with HA-tagged FBXW7α or FBXW7F. FBXW7α dramatically reduced while FBXW7F increased the protein level of Myc-P48 compared with the empty vector (Fig. 1D). In contrast, neither FBXW7α nor FBXW7F had influence on the protein levels of Myc-P42 (Fig. 1E). No change in P48 mRNA level was observed after depleting or silencing FBXW7 expression in the cell lines used (Supplementary Fig. S1G–S1J). These results indicated that FBXW7 downregulates P48 expression at posttranscriptional level.

To recognize any clinical correlation of FBXW7 and EBP1 P48 expression, immunohistochemistry (IHC) was performed on 50 colorectal cancer tissue samples (11 of them with paired adjacent non-tumorous tissues) to analyze FBXW7 and P48 expression. As expected, FBXW7 expression was significantly downregulated, and P48 was highly expressed in colon cancer tissues compared with their adjacent normal colon tissues (Fig. 1F). Statistical analysis showed that P48 expression was negatively correlated with FBXW7 level in colon cancer tissues (Fig. 1G).

To evaluate whether FBXW7 affects the protein stability of EBP1, we overexpressed FBXW7α and Myc-tagged P48 or Myc-tagged P42 in HEK293T cells and measured the exogenous P48 and P42 protein degradation by cycloheximide chase assay. We found that Myc-P48 but not Myc-P42 was more quickly degraded when FBXW7 was present (Fig. 2A and B). The effect of FBXW7 on the turnover of endogenous P48 protein was verified in HCT116 and DLD-1 cells (Fig. 2C and D). These results indicate that FBXW7 facilitates P48 protein degradation.

Next, we determined whether the degradation of P48 by FBXW7 is proteasome-dependent. HEK293T, HCT116, and DLD-1 cells were treated with proteasome inhibitor MG132. Proteasome inhibition increased not only exogenous P48 protein levels (Fig. 2E) but also endogenous P48 protein levels in an FBXW7-dependent manner (Fig. 2F and G). Proteins targeted for proteasome destruction are usually polyubiquitinated. To examine whether FBXW7 influences the ubiquitination status of P48, Flag-tagged P48 was transfected into HEK293T cells with HA-tagged ubiquitin (HA-ub) or with HA-ub and Myc-tagged FBXW7α. Immunoprecipitation with Flag-EBP1 P48 followed by HA immunoblotting showed that overexpression of FBXW7 increased the ubiquitination of Flag-P48 (Fig. 2H). In addition, we compared P48 ubiquitination in HCT116 FBXW7^+/+ and FBXW7^−/− cells and found that the P48 ubiquitination is dramatically decreased in HCT116 FBXW7^−/− cells (Fig. 2I).

As FBXW7 mediates the degradation of most substrates in a GSK3β-dependent manner, we then examined whether FBXW7 targets P48 in a GSK3β-dependent manner. GSK3β was found to coimmunoprecipitate with Flag-tagged P48 (Supplementary Fig. S2A). Moreover, GSK3β inhibitor VIII reversed the reduced expression of ectopic and endogenous P48 protein mediated by FBXW7 (Supplementary Fig. S2B and S2C). We then investigated whether GSK3β inhibitor treatment could inhibit the ubiquitination of P48. HEK293T cells were transfected with HA-ub and treated with or without GSK3β inhibitor. Immunoprecipitation of HA followed by immunoblot analysis of P48 showed that P48 ubiquitination was dramatically reduced in the cells following GSK3β inhibitor treatment (Supplementary Fig. S2D). Taken together, these results indicate that FBXW7 promotes the ubiquitination and proteasomal degradation of P48 in a GSK3β-dependent manner.

To test whether FBXW7 physically interacts with EBP1 isoforms, HEK293T cells were cotransfected with HA-tagged FBXW7α and Flag-tagged P48 or Flag-tagged P42, respectively. Immunoprecipitation and subsequent immunoblot analyses showed that FBXW7 coprecipitated with both P48 and P42 (Fig. 3A). Moreover, endogenous FBXW7 coprecipitated with endogenous P48 (Fig. 3B). To identify the sites of FBXW7 that interact with P48 and P42, HEK293T cells were cotransfected with Flag-tagged EBP1 isoforms and HA-tagged FBXW7α, FBXW7F (F-box deletion), FBXW7WD (WD domain deletion; Supplementary Fig. S1C), or WD mutants in the residues 465, 479, and 505 of FBXW7α (R465C, R465H, R479H, and R505C), which are essential for the substrate interaction (32). Coimmunoprecipitation analysis showed that both FBXW7 and FBXW7F could precipitate with P48 and P42. However, FBXW7WD and WD mutants failed to bind to P48 but still had the binding activity to P42 (Fig. 3C and D and Supplementary Fig. S3A–S3C). Consistently, functional analysis revealed that neither FBXW7WD nor WD mutants could increase the ubiquitination of P48 (Fig. 3E) and had effects on the protein level of P48 (Fig. 3F and Supplementary Fig. S3D). These results indicate that FBXW7 binds to P48 solely through its WD domain but to P42 in a more complex mechanism.

To further investigate the functional region of FBXW7 to bind with P42, we cotransfected HEK293T cells with Flag-tagged P42 and HA-tagged FBXW7NF with the deletion of both F-box and WD domain (Supplementary Fig. S1C) or HA-tagged FBXW7F^R465C with F-box domain deletion and a point mutation in WD domain, which lose its binding ability to its substrates. Subsequent coimmunoprecipitation showed that neither FBXW7NF nor FBXW7F^R465C could precipitate with EBP1 P42 (Fig. 3G). Taken together, these data indicate that FBXW7 differentially interacts with EBP1 isoforms, FBXW7 binds to P48 through its WD domain and promotes P48 degradation, but P42 binds to both FBXW7 F-box domain directly and to WD domain mediated by FBXW7 substrates.

FBXW7 is known to target substrates containing a consensus motif termed CPD comprising residues that are often phosphorylated by GSK3β. Using the PhosphoMotif Finder software, we found the N-terminal region of P48 harbors 2 potential CPDs at residues 40–44 and 88–94. To test whether these motifs in P48 are critical for its regulation by FBXW7, Myc-tagged P4840–44 and Myc-tagged P4888–94 constructs were generated and cotransfected with HA-tagged FBXW7α into HEK293T cells to examine the binding activity. Both full-length P48 and P4888–94 coprecipitated with FBXW7; in contrast, P4840-44 did not form a detectable complex with FBXW7 (Fig. 4A). Consistent with these findings, we found that overexpression of FBXW7α significantly decreased the protein level and accelerated the degradation of P4888-94 but had no effect on those of P4840-44 (Fig. 4B–D). In addition, in the presence of proteasome inhibitor MG132, a significant increase in P4888–94 protein level occurred whereas EBP1 P4840-44 protein level did not change (Fig. 4E). Together, these results suggest that the motif at residues 40–44 is a functional FBXW7 degradation signal.

We then investigated the significance of individual site for regulation by FBXW7. As the sequence of EBP1 P48 Ser40 to Ser44, S⁴⁰SGVS⁴⁴ is highly similar to the putative GSK3β conserved phosphorylation sequence (Ser/Thr-X-X-X-Ser/Thr), we hypothesized that P48 Ser40 and Ser44 are crucial for its GSK3β-dependent degradation by FBXW7. We generated the mutant EBP1 P48S40A where serine at 40 was substituted by alanine. As expected, FBXW7 as well as GSK3β inhibitor VIII cannot affect the protein level of EBP1 P48S40A (Fig. 4F). Co-immunoprecipitation showed that EBP1 P48S40A mutant completely lost the capacity to bind to FBXW7 (Fig. 4G), and in turn, the ubiquitination of EBP1 P48S40A mutant was dramatically reduced compared with that of WT EBP1 P48 (Fig. 4H). As GSK3β usually requires priming phosphorylation on Ser or Thr at position +4 (33, 34), we then constructed the mutant EBP1 P48S44A where serine at 44 was substituted by alanine. Co-immunoprecipitation showed that the mutant EBP1 P48S44A cannot bind to FBXW7 (Supplementary Fig. S4A) or GSK3β (Supplementary Fig. S2A). Moreover, EBP1 P48S44A cannot be efficiently ubiquitinated by FBXW7 (Supplementary Fig. S4B). These results suggested that EBP1 P48S44 may serve as a priming phosphorylation site to recruit GSK3β and then S40 is phosphorylated by GSK3β. Taken all together, we concluded that the residual S40 and S44 sites of P48 are responsible for the physical interaction with FBXW7 and for the FBXW7-mediated proteasomal degradation of EBP1 P48.

As EBP1 P48 is identified as a target of FBXW7, we questioned whether P48 is responsible for the cell proliferation and motility phenotypes induced by FBXW7 deficiency. Depletion of P48 expression in HCT116 FBXW7^−/− cells by specific P48 shRNA (Supplementary Fig. S5A) abolished FBXW7 loss–induced cell proliferation using MTT and colony formation assays (Fig. 5A and B). In addition, knockdown of P48 significantly alleviated the increase of cell migration and invasion induced by FBXW7 deficiency in HCT116 cells (Fig. 5C and D). Similar results were observed in DLD-1 FBXW7^−/− cells and prostate cancer DU145-shFBXW7 cells when silencing P48 expression (Supplementary Fig. S5B–S5G). These data collectively showed that knockdown of P48 effectively suppressed FBXW7 loss–driven cell proliferation, migration, and invasion, indicating that FBXW7 executes tumor suppression function at least partly through regulating P48 level.

To verify whether P48 mediates FBXW7 loss–induced metastasis in vivo, HCT116 FBXW7^−/− cells with or without silencing P48 expression were injected into nude mice through the tail vein. Silencing P48 not only significantly decreased the number of mice with distant metastasis (Fig. 5E) but also dramatically decreased the number of metastatic tumors in both lung and liver of each mouse (Fig. 5F and G). To distinguish the colonized cell population between human and mouse, anti-human GAPDH antibody, which does not react with mouse GAPDH, was used to detect the lung and liver tissues of mice by IHC. The results showed that the colonized cell populations were from human colon cancer cells (Fig. 5H and I). Therefore, the in vivo results further demonstrated the critical role of P48 in the colon cancer cell metastatic behavior induced by loss of FBXW7.

In our efforts to explore the modulation of FBXW7 function, EBP1 P48 was observed to be able to influence the activity of FBXW7. We first found that overexpression of P48 in HEK293T cells inhibited the nuclear localization of FBXW7α (Fig. 6A and Supplementary Fig. S6A) but had no effect on WD or WD domain mutant forms of FBXW7 (Fig. 6B and C and Supplementary Fig. S6B and S6C). As nuclear localization of FBXW7α is essential for nuclear oncoprotein degradation, and FBXW7α mislocalization is of pathologic significance (35), we then subsequently examined the protein levels of FBXW7α nuclear substrates, including c-Myc, cyclin E, and Aurora A. These protein levels were increased after ectopic expression of P48 in an FBXW7-dependent manner (Fig. 6D). Further coimmunoprecipitation in HEK293T cells showed that EBP1 P48 could decrease the binding of FBXW7 with its substrates, including c-Myc, Aurora A, and cyclin E (Fig. 6E). mRNA levels of these analyzed FBXW7 substrates were not affected by ectopic expression of P48 (Supplementary Fig. S6D). Collectively, we concluded that P48 may suppress FBXW7-mediated substrate degradation through relocalizing FBXW7α to the cytosol. Thus, FBXW7α and P48 form a negative regulatory loop, FBXW7 induces P48 degradation, and meanwhile P48 partially inhibits the degradation of nuclear target proteins of FBXW7α.

Next, we transfected EBP1 P48 into HCT116 FBXW7^+/+ and HCT116 FBXW7^−/− cells, respectively, and detected the cell proliferation and motility phenotypes induced by EBP1 P48. As a resultant, overexpression of EBP1 P48 enhanced the proliferation and migration of both HCT116 FBXW7^+/+ and HCT116 FBXW7^−/− cells, but the promoting effect of EBP1 P48 was more significant in HCT116 FBXW7^+/+ cells, indicating that P48 executes its oncogenic activity at least partly dependent on its interaction with FBXW7 (Supplementary Fig. S6E and S6F).

As shown above that EBP1 P42 interacts with FBXW7 F-box domain and FBXW7–substrate complex, we questioned the possibility that P42 functions as a bridge to enhance the binding of substrates to the WD domain of FBXW7. Consistently, when HEK293T was cotransfected with Myc-tagged and HA-tagged FBXW7α and Flag-tagged P42, coimmunoprecipitation results confirmed that P42 did enhance the binding of FBXW7 with its substrates such as c-Myc, Aurora A, cyclin E, and P48 (Fig. 7A). We also examined whether P42 has any influence on the dimerization of FBXW7, as the dimerization regulates its interaction with substrates and the robustness of substrate degradation (36, 37). Coimmunoprecipitation results showed that P42 enhanced the binding between Myc-tagged and HA-tagged FBXW7α (Fig. 7A), which may also contribute to the enhanced binding between FBXW7α and its substrates.

We then investigated the biologic significance of this interaction between P42 and FBXW7. Although overexpression of P42 had no significant effects on FBXW7 protein level and localization (Fig. 6A), overexpression of WT P42 led to a more significant decrease of several FBXW7 substrates including c-Myc, cyclin E, Aurora A, and P48 in FBXW7^+/+ HCT116 cells compared with that in FBXW7^−/− HCT116 cells without affecting the mRNA levels of these substrate (Fig. 7B and Supplementary Fig. S7), suggesting that P42 can promote FBXW7-mediated substrate degradation.

To verify whether P42 has any effect on the tumor-suppressive function of FBXW7, combinations of P42 and FBXW7 expression were constructed in HCT116 FBXW7^−/− cells. Overexpression of P42 dramatically enhanced FBXW7-induced inhibition of cell proliferation using MTT and colony formation assays (Fig. 7C and D). In addition, scratch-healing and Boyden chamber assays showed that overexpression of P42 promotes the inhibitory effect of FBXW7 on cell migration and invasion (Fig. 7E and F). These data illustrated that P42 could enhance the tumor-suppressing role of FBXW7.

Furthermore, although overexpression of P42 led to a significant decrease in the cell growth, migration, and invasion activity of HCT116 FBXW7^+/+ cells but to a less extent in that of HCT116 FBXW7^−/− cells (Supplementary Fig. S8), suggesting that P42 plays its tumor-suppressive activity partially through enhancing the tumor-suppressive role of FBXW7.

Together, our data demonstrated that FBXW7 differentially regulates EBP1 isoforms. In turn, the activity of FBXW7 is differently regulated by EBP1 isoforms, which was attenuated by P48 but enhanced by P42. The mutual interactions among FBXW7, P48, and P42 may play important roles in tumor initiation and progression (Fig. 7G).

EBP1 was originally identified as an ErbB3-binding protein, which was considered as an oncogene or tumor suppressor in different circumstances (38–41) and has prognostic significance in a panel of cancers such as cervical (7, 42), pancreatic (6), hepatocellular (43), brain (44), and other tumors. Recent study has clarified that EBP1 possesses two isoforms, the longer form P48 and the shorter form P42, which distinctively regulate cell survival and differentiation (5). The isoform P48 with the oncogenic potential promotes cell proliferation and facilitates cancer development (4, 45), whereas the isoform P42 is considered as a tumor suppressor by inhibiting cell proliferation and suppresses tumorigenesis (3, 9, 46). However, the molecular mechanism(s) underlying these differential functions of the 2 isoforms especially in cancers remain largely unknown. It was previously reported that P48 and P42 bind to ErbB3 with different affinities, thus differentially regulate AKT and PI3K activities to regulate cell proliferation and differentiation (5), and interact differentially with nucleophosmin/B23 to regulate cell proliferation and apoptosis in PC12 cells (2). In this study, through searching novel targets of FBXW7, we discovered the mechanism by which EBP1 isoforms execute differential functions in tumor development through distinctive interactions with FBXW7.

We first identified that the oncogenic isoform P48 is a substrate of FBXW7. Our results show that FBXW7 binds to the P48 via its WD40 domains and mediates its ubiquitination and subsequent proteasomal degradation in GSK3β-dependent manner, which is consistent with the stereotyped pattern of a vast majority of characterized substrates recognized by FBXW7 (47–49). In addition, depletion of P48 alleviated FBXW7 loss–induced motility and invasiveness in colon cancer cells in vitro and in vivo. These experimental findings provide a mechanistic framework to explain the inverse correlation between FBXW7 and P48 abundance in colon cancer clinical samples.

Besides the reported mechanistically oncogenic functions of P48 (4), we found that P48 promotes FBXW7α nuclear export, and the relocalization of FBXW7α has been reported to be of pathologic significance because it is intimately linked to its role of substrate downregulation (35). Consistent with this finding, the levels of some nuclear substrates of FBXW7 are accumulated in the cells, and subsequently the inhibitory effects of FBXW7 on cell migration were attenuated. Our findings further support the oncogenic role of P48 in cancer development. FBXW7α contains two nuclear localization signals (NLS), one in the isoform-specific N-terminus and another in the shared WD40 domain. Our preliminary data showed that P48 can regulate nuclear localization of FBXW7 through binding to its WD40 domain, raising a possibility that P48 may mask the NLS in WD40 domain thereby influencing its localization. Thus, this study reveals a mechanism by which P48 provokes cancer through regulating FBXW7α activities.

We have identified that the residues S40 and S44 within the N-terminus of P48 are critical for binding to FBXW7 and ubiquitination mediated by FBXW7. P42 lacks N-terminal 54 amino acids but is still able to bind to FBXW7 and cannot be degraded by FBXW7. The lack of N-terminal 54 amino acids causes P42 to have different crystal structure from P48 (50), thus P48 and P42 bind to different domains of FBXW7 with different motifs. The deletion of 40–44 aa or mutation of key amino acid of Ser40 or Ser44 in P48 results in the destruction of the structure of P48, which is necessary for binding to FBXW7 WD domain. However, P42, without the N-terminal 54 amino acids of P48, still maintains its own structure and the binding mode with FBXW7. This is also consistent with previous studies that show different binding partners of P42 and P48 (2, 5).

Detailed analysis of protein interaction revealed that P42 binds to FBXW7 F-box directly and to WD40 domain in a substrate binding–dependent manner. Such interactions potentiate the binding between FBXW7 and its substrates and subsequently lead to reduced levels of these substrates. Our findings truly indicate that P42 interacts with FBXW7–substrate complex and functions as a bridge to enhance the binding of substrates to the WD40 domain of FBXW7. Moreover, our mechanism study further demonstrates that P42 fortifies the dimerization of FBXW7. Emerging evidences have suggested that FBXW7 dimerization plays a critical role in regulating FBXW7 stability and its function on substrate degradation (36, 37). As a result, P42 promotes FBXW7-mediated substrate degradation and enhances the tumor-suppressing role of FBXW7 in an FBXW7-dependent manner. Our study reveals a mechanism by which P42 suppresses tumor development by promoting FBXW7 activities.

In conclusion, our study uncovers a physical and functional relationship among FBXW7 and EBP1 isoforms (P48 and P42). Our results demonstrate that two EBP1 isoforms distinctively modulate the role of FBXW7 by different and independent mechanisms. The knowledge of the role of FBXW7 and EBP1 isoforms in the molecular mechanisms governing tumorigenesis sheds new lights on improving current anticancer therapies.

No potential conflicts of interest were disclosed.

Conception and design: J.-H. Mao, G. Wei

Development of methodology: Y. Wang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Zhang, Y. Wang, C. Liu, J.-H. Mao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Wang, P. Zhang, C. Liu, J.-H. Mao

Writing, review, and/or revision of the manuscript: Y. Wang, P. Zhang, Y. Wang, J.-H. Mao, G. Wei

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Zhan, G. Wei

Study supervision: G. Wei

This work was supported by National Natural Science Foundation of China (nos. 31271461, 81472583, 81528017) and the Taishan Scholar Program of Shandong Province (G. Wei); by the NIH, National Cancer Institute R01 CA116481, and the Low Dose Scientific Focus Area, Office of Biological & Environmental Research, US Department of Energy (DE-AC02-05CH11231) to J.-H. Mao; by National Natural Science Foundation of China (no. 81470127), the Natural Science Foundation of Shandong Province (no. ZR2014HM032), China Postdoctoral Science Foundation Funded Project (nos. 2011M501136 and 2012T50616) to P. Zhang; by the China Postdoctoral International Exchange Program 2015, National Natural Science Foundation of China (no. 81402193), Postdoctoral Innovation Project of Shandong Province (no. 147751), and Postdoctoral Science Foundation of China (no. 2015M570597) to Y. Wang.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.