The Ubiquitin Ligase SCFFbw7 Antagonizes Apoptotic JNK Signaling

See allHide authors and affiliations

Science  27 Feb 2004:
Vol. 303, Issue 5662, pp. 1374-1378
DOI: 10.1126/science.1092880


Jun N-terminal kinases (JNKs) are essential for neuronal microtubule assembly and apoptosis. Phosphorylation of the activating protein 1 (AP1) transcription factor c-Jun, at multiple sites within its transactivation domain, is required for JNK-induced neurotoxicity. We report that in neurons the stability of c-Jun is regulated by the E3 ligase SCFFbw7, which ubiquitinates phosphorylated c-Jun and facilitates c-Jun degradation. Fbw7 depletion resulted in accumulation of phosphorylated c-Jun, stimulation of AP1 activity, and neuronal apoptosis. SCFFbw7 therefore antagonizes the apoptotic c-Jun–dependent effector arm of JNK signaling, allowing neurons to tolerate potentially neurotoxic JNK activity.

The Jun N-terminal kinase (JNK) signaling pathway leading to c-Jun phosphorylation is implicated in neuronal apoptosis caused by a variety of conditions, including exposure to excitotoxins and some neurodegenerative disorders (1, 2). Genetic ablation of c-Jun phosphorylation results in protection from neuronal death, but how phosphorylated c-Jun regulates cell death is unclear (3, 4).

To determine whether JNK-mediated phosphorylation of c-Jun modifies the interaction of the c-Jun N terminus with cofactor molecules to modulate neurotoxic JNK signaling, we used a modified yeast-based, cytoplasmic two-hybrid screening to identify phosphorylation-dependent interactions (5) (Fig. 1A). The Ras Recruitment System (RRS) is based on the complementation of a temperature-sensitive yeast cdc25 mutant, deficient in Ras activity, by human oncogenic Ras (RasV12). Ras function requires plasma membrane localization, which can be achieved through interaction between two hybrid proteins. A c-Jun leucine zipper mutant (JunΔLZ) was generated to avoid unwanted interaction with known partner proteins. It was fused to RasV12 (RasV12-JunΔLZ) and used as a bait (Fig. 1B). A constitutively active MKK7-JNK1 fusion protein was placed under the control of the methionine-regulated MET3 promoter, to induce N-terminal phosphorylation of the RasV12-JunΔLZ bait fusion protein in a methionine-dependent manner (Fig. 1C) (68). After transfection of a brain cDNA library in which all encoded proteins were fused to a myristoylation signal and thus associated to the plasma membrane (8), yeast clones were screened for rescue of the cdc25 mutation at the restrictive temperature in a methionine-dependent, and thus JNK-dependent, fashion. Several proteins were identified that either preferentially interacted with unphosphorylated c-Jun, including phosphorylation-dependent interactors of c-Jun 2 (PDJ-2), or bound specifically to N-terminally phosphorylated c-Jun, including PDJ-3 (Fig. 1D).

Fig. 1.

Identification of Fbw7 as a phosphorylation-dependent c-Jun interacting protein. (A) Schematic representation of the modified RRS (detailed description in text). GDP, guanosine diphosphate; pA, polyA signal; P, phosphorylated amino acid. (B) RasV12 control and RasV12-JunΔLZ protein expression in yeast. A Western blot of cell lysates with antibody to Ras is shown. Untr., untransfected. (C) Expression of MKK7-JNK1 fusion protein and c-Jun phosphorylation (p-c-Jun) was induced by withdrawal of methionine. A Western blot of cell lysates with specific antibodies is shown. (D) Methionine-dependent rescue of growth cdc25-2 mutant yeast by PDJs (8).

Plasmid rescue and sequencing identified PDJ-3 as Fbw7 (also known as hCDC4 and hAgo). Fbw7 contains an N-terminal F-box and seven WD40 repeats that form a WD40 propeller (912). There are more than 50 F-box–containing proteins in mammalian genomes that function as receptor subunits for Skp1/Cullin/F-box protein (SCF) E3 ubiquitin ligases (1315). The Fbw7-containing SCF complex (SCFFbw7) binds and regulates the abundance of the cyclin E protein and the nuclear form of the activated Notch receptor in a phosphorylation-dependent manner (911, 16, 17). fbw7 mRNA is most abundant in the adult brain, in cerebellar granule cells, and in the neuron-like PC12 cell line (8, 10, 12) (fig. S1A). Lower levels were detected in several other cell lines, and fbw7 expression was normal in fibroblasts and neurons lacking c-Jun or lacking the JNK phosphorylation sites Ser63 and Ser73 (4, 1820) (fig. S1, A and B).

To study the phosphorylation-dependence of c-Jun and Fbw7 interaction, both proteins were translated in vitro (IVT) (8). c-Jun was unphosphorylated in wheat-germ extracts (Fig. 2A, lane 1). However, when proteins were translated in reticulocyte extracts (RT), three c-Jun bands with lower electrophoretic mobility were detected, presumably caused by individual and combined phosphorylation of both N-terminal clusters of phosphorylated residues (Ser63/Ser73 and Thr91/Thr93). Treatment of RT lysates with the phosphatase inhibitor okadaic acid resulted in the hyperphosphorylation of c-Jun (Fig. 2A, lane 3). Immunoprecipitation of IVT epitope-tagged (FLAG-) Fbw7 revealed interaction with phosphorylated c-Jun, but not with unphosphorylated c-Jun (Fig. 2A, lanes 4 to 11). Treatment of RT lysates containing phosphorylated c-Jun with λ phosphatase resulted in a single, unphosphorylated, faster-migrating form that did not interact with FLAG-Fbw7 (Fig. 2A, lanes 12 to 19). FLAG-Fbw7 did not interact with the related activating protein 1 (AP1) transcription factor ATF-2 (Fig. 2B, lanes 7 and 8), although ATF-2, like c-Jun, was also phosphorylated by JNK in RT lysates (Fig. 2, C and D).

Fig. 2.

Fbw7 interacts with and ubiquitinates phosphorylated c-Jun in vitro. (A) 35S-labeled IVT c-Jun and phosphorylated c-Jun, with or without okadaic acid or λ phosphatase (λPPase) treatment, were assayed for FLAG-Fbw7 binding by immunoprecipitation (IP). Autoradiographs are shown. (B) IVT ATF-2 does not interact with FLAG-Fbw7. Nonradiolabeled FLAG-Fbw7 was used for IP because its molecular weight is similar to that of ATF-2. Autoradiographs are shown. (C) 35S-labeled or unlabeled c-Jun was translated in RT extracts, treated with or without λPPase, and analyzed by either autoradiography or Western blot analysis for the indicated specific antibody. P-Ser73, phosphorylated Ser73. (D) RT extracts with or without ATF-2 were used for specific Western blot analysis with the indicated antibodies specific for ATF-2 and phosphorylated ATF-2. (E) SCFFbw7 ubiquitinates phosphorylated c-Jun in vitro. A monoclonal antibody to c-Jun was used for detection of c-Jun-ubiquitin (Ubn) conjugates. A shorter exposure shows that comparable amounts of c-Jun proteins were used as substrates for in vitro ubiquitination.

To examine whether Fbw7 possesses E3 ligase activity toward phosphorylated c-Jun, we assessed c-Jun ubiquitination in vitro (8). Whereas phosphorylated c-Jun was efficiently ubiquitinated in vitro, a mutant c-Jun protein lacking all JNK phosphorylation sites (c-Jun63,73,91,93A) was not (Fig. 2E). Fbw7-mediated ubiquitination of phosphorylated c-Jun was dose-dependent and required an ubiquitin-activating enzyme (E1) and a ubiquitin-conjugating enzyme (E2) (fig. S2). Therefore, SCFFbw7 has properties that are consistent with a function as an E3 ubiquitin ligase that specifically targets phosphorylated c-Jun.

If Fbw7 is rate-limiting for controlling the abundance of phosphorylated c-Jun, overexpression of Fbw7 should decrease the amounts of phosphorylated c-Jun. Ectopic expression of Fbw7 in HEK 293T, a human cell line that contains constitutively high JNK activity and low endogenous fbw7 expression, did not alter c-jun mRNA levels, but caused a decrease in phosphorylated c-Jun protein (Fig. 3A). This was prevented by treatment of cells with proteasome inhibitors (Fig. 3B) (8). The abundance of total JNK and active phosphorylated JNK protein was unaffected by the presence or absence of Fbw7 (Fig. 3A). However, high levels of JNK activity induced c-Jun phosphorylation despite constitutive Fbw7-mediated degradation of phosphorylated c-Jun (fig. S3).

Fig. 3.

SCFFbw7 degrades phosphorylated c-Jun in vivo. (A) c-Jun and Fbw7 were expressed in HEK 293T cells. Individual samples were split for RNA and protein isolation and subsequent Northern and Western blotting with the indicated antibodies. (B) Lysates of transfected HEK 293T cells overexpressing c-Jun and Fbw7 were treated with proteasome inhibitor I (Prot Inh) for 8 hours. c-Jun and phosphorylated c-Jun levels were determined by Western blotting with specific antibodies. P-Thr91, phosphorylated Thr91. (C) Immobilized c-Jun– or cyclin E–derived peptides with or without phosphorylation were used for IP of Fbw7. The peptide sequence and sites of phosphorylation and Fbw7 detection by autoradiography are shown (24). TNT, RT lysate without Fbw7; beads, magnetic beads not coupled to peptides. (D) Lysates of transfected HEK 293T cells expressing c-Jun phosphorylation mutants, with or without Fbw7, were analyzed by Western blot with the indicated antibodies for c-Jun protein levels. (E) Fbw7 reduces the stability of phosphorylated c-Jun. 35S-methionine–labeled HEK 293T cells overexpressing c-Jun or c-Jun63,73,91,93A, with or without Fbw7, were chased with cold methionine for the indicated times. JNK activity was stimulated with 10mM anisomycin 30min before chase. Autoradiographs are shown.

Phosphorylated peptides encompassing either Ser63 and Ser73 or Thr91 and Thr93 of the c-Jun N terminus interacted with Fbw7 to similar extents as phosphorylated peptides derived from cyclin E, a previously identified Fbw7 substrate (Fig. 3C) (10). A phosphorylated peptide encompassing Thr69 and Thr71 of ATF-2 did not bind Fbw7 efficiently (fig. S4). Mutation of either Ser63 and Ser73 or Thr91 and Thr93 did not result in stabilization of phosphorylated c-Jun, indicating that both phosphorylation clusters contribute to recognition by SCFFbw7 in vivo (Fig. 3D). However, mutation of all four phosphorylation sites rendered the protein insensitive to Fbw7-mediated degradation (Fig. 3D). Phosphorylated c-Jun was stable in HEK 293T cells (21), but coexpression of Fbw7 decreased the steady-state levels of c-Jun and resulted in reduced protein half-life (8) (Fig. 3E). In contrast, the stability of c-Jun63,73,91,93A mutant protein was similar in the presence or absence of Fbw7 (Fig. 3E).

The effect of Fbw7 on AP1 activity was determined with an AP1 luciferase reporter gene assay. Ectopic c-Jun expression in HEK 293T cells increased reporter activity; this was reduced by overexpression of Fbw7 (Fig. 4A) (8). Reporter gene activation by c-Jun63,73,91,93A mutant protein was reduced compared to the wild type and unaffected by Fbw7 expression (Fig. 4B). To determine whether Fbw7 is a physiologically relevant, nonredundant regulator of phosphorylated c-Jun, Fbw7 was depleted by expression of small interfering RNA (siRNA). Knock-down of Fbw7 expression was efficient, increased the levels of phosphorylated c-Jun, and augmented AP1 reporter gene activity (Fig. 4, C to E). Fbw7 depletion increased the stability of phosphorylated c-Jun (fig. S5), and degradation of phosphorylated c-Jun was mediated mainly by the β isoform of Fbw7 (fig. S6). Fbw7 knock-down caused cell death; this effect was reduced by overexpressing either the JNK inhibitor JIP-1 or c-Jun63,73,91,93A, indicating that apoptosis induced by Fbw7 depletion requires JNK activity and phosphorylated c-Jun (8) (Fig. 4F). Mismatched or unspecific siRNAs induced neither c-Jun phosphorylation nor cell death (fig. S7). Fbw7 was also required to prevent apoptosis of differentiated PC12 cells and augmented apoptosis induced by the withdrawal of nerve growth factor (NGF) (8) (Fig. 4G). Moreover, primary cerebellar neurons isolated from mice with a Cre/LoxP–mediated brain-specific c-jun deletion (c-junΔn) were protected from cell death induced by Fbw7 depletion, providing genetic evidence for c-Jun as an essential Fbw7 target (Fig. 4H) (8, 18).

Fig. 4.

Fbw7 antagonizes AP1 activity and apoptotic JNK signaling. (A) HEK 293T cells were transfected to express c-Jun, Fbw7, the AP1 firefly luciferase reporter gene (AP1-Fluc), and the tk-renilla luciferase transfection control reporter gene (tk-Rluc). Luciferase activity of mock-transfected cells was arbitrarily set to 1 (8). (B) c-Jun, c-Jun63,73,91,93A, and Fbw7 were expressed in HEK 293T cells with reporter genes as in (A). Luciferase activity induced by c-Jun is shown as 100%. (C) PC12 cells were transfected to express siFbw7 with empty vector as the control (25). Individual samples were split for RNA, protein isolation, and subsequent Northern and Western blot analysis. Transfection efficiency was 80%; thus, some residual Fbw7 mRNA and protein is still detectable. gapdh, glyceraldehyde-3-phosphate dehydrogenase. (D) PC12 cells were transfected to express siFbw7, control vector, and JIP-1 as indicated. Cell lysates were analyzed by Western blot with the indicated antibodies. (E) PC12 cells were transfected to express siFbw7 or control vector, together with both AP1-Fluc and tk-Rluc, and luciferase activity was determined (8). Luciferase activity of mock-transfected cells is arbitrarily set to 1. (F) PC12 cells were transfected to express siFbw7 or control vector and JIP-1 or c-Jun63,73,91,93A as indicated. We measured cell death using DNA fragmentation analysis by flow cytometry. The percentage of cells with a sub2N (<2N ploidy) DNA content is shown (8). (G) Apoptosis was induced by withdrawal of NGF from differentiated PC12 cells transfected to express siFbw7 or control vector. Cell death was determined after 20 hours by determining sub2N DNA content and is shown as percent of total cells analyzed. (H) Primary cerebellar neurons from c-junf/f and c-junΔn mice were transfected with pSUPER vectors. Cell death was measured after 36 hours (8, 18) and is shown as percent of total cells analyzed. (I) A model illustrating the specific inhibition of apoptotic JNK signaling by Fbw7 in neurons.

JNK signaling is required to maintain normal brain function, because mice lacking JNK1 show progressive axonal degeneration (22). JNK regulates neuronal microtubule assembly by phosphorylating the microtubule-associated proteins MAP1B and MAP2 (22). MAP1B and MAP2 phosphorylation is unaffected by Fbw7 depletion (fig. S8), indicating that Fbw7 specifically inactivates the apoptotic effector arm of JNK signaling, thereby allowing neurons to tolerate potentially neurotoxic JNK signaling (Fig. 4I). Furthermore, Fbw7 has tumor suppressor function and is mutated in a high percentage of endometrial cancers with high cyclin E expression (23). Accumulation of the phosphorylated, transcriptionally active form of c-Jun in tumors that lack Fbw7 may contribute to cancer development.

Supporting Online Material

Materials and Methods

Figs. S1 to S9

References and Notes

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article