The FBXO4 tumor suppressor functions as a barrier to BRAFV600E-dependent metastatic melanoma

Abstract

Cyclin D1-cyclin-dependent kinase 4/6 (CDK4/6) dysregulation is a major contributor to melanomagenesis. Clinical evidence has revealed that p16(INK4A), an allosteric inhibitor of CDK4/6, is inactivated in over half of human melanomas, and numerous animal models have demonstrated that p16(INK4A) deletion promotes melanoma. FBXO4, a specificity factor for the E3 ligase that directs timely cyclin D1 proteolysis, has not been studied in melanoma. We demonstrate that Fbxo4 deficiency induces Braf-driven melanoma and that this phenotype depends on cyclin D1 accumulation in mice, underscoring the importance of this ubiquitin ligase in tumor suppression. Furthermore, we have identified a substrate-binding mutation, FBXO4 I377M, that selectively disrupts cyclin D1 degradation while preserving proteolysis of the other known FBXO4 substrate, TRF1. The I377M mutation and Fbxo4 deficiency result in nuclear accumulation of cyclin D1, a key transforming neoplastic event. Collectively, these data provide evidence that FBXO4 dysfunction, as a mechanism for cyclin D1 overexpression, is a contributor to human malignancy.

Fig 1

Braf^V600E drives melanocyte hyperplasia independent of Fbxo4 or cyclin D1 status. (A and B) Glabrous skin of mice with the indicated genotypes. (C) Benign hyperplastic nevi of the indicated Fbxo4 transgenic mice on a Braf-activated background. (D) Low-power (left) and high-power (right) magnifications of nevus. (E) Amplification of DNA extracts isolated from mice/tissue of the indicated genotypes. (F) Lysates prepared from the indicated tissues/genotypes were assessed for cyclin D1, phospho-ERK1/2, and total ERK1/2 expression by Western blotting. (G) Braf^+/+(left), Braf^V600E/+; cycD1^+/− (middle), and Braf^V600E/+; cycD1^+/+ (right) mice were monitored for melanocyte hyperplasia.

Fig 2

Fbxo4 deficiency induces melanoma in Braf-activated mice. (A) Representative images of Braf-activated mice with the indicated Fbxo4 status at 12 weeks post-4HT treatment. (B) Kaplan-Meier survival analysis of 4HT-treated mice (Braf^CA/+/Tyr::Cre^−/−, n = 20; Braf^+/+/Tyr::Cre^+/o, n = 20; Braf^CA/+/Tyr::Cre^+/o/Fbxo4^+/+, n = 15; Fbxo4^+/−, n = 15; Fbxo4^+/−, n = 10) (*, comparison of Braf^CA/+/Tyr::Cre^+/o/Fbxo4^+/+ versus Braf^CA/+/Tyr::Cre^+/o/Fbxo4^+/−; **, comparison of Braf^CA/+/Tyr::Cre^+/o/Fbxo4^+/+ versus Braf^CA/+/Tyr::Cre^+/o/Fbxo4^−/−). (C) Low-magnification micrograph of a Braf^CA/+/Fbxo4^−/− tumor. Highly invasive amelanotic melanocytes infiltrate skeletal muscle. (D) S-100 neural crest-specific nuclear staining of tumor cells.

Fig 3

Cyclin D1 accumulates in Braf^V600E/Fbxo4-deficient tumors. (A and B) Fbxo4 immunohistochemistry (A) and hematoxylin and eosin staining (B) of Braf^V600E/Fbxo4^−/− tumor sections. (C) Cyclin D1 immunohistochemistry of paraffin-embedded tumor sections derived from Braf^V600E/Fbxo4-deficient mice. Arrows indicate intense nuclear cyclin D1 staining. Arrow indicate cyclin D1 expression in the mitotic layer of the epidermis. Note the relative intensities and the absence of cyclin D1 in the postmitotic layer of the epidermis. (D) Western analysis of the indicated proteins isolated from tumor lysates of Braf^V600E/Fbxo4 WT, heterozygous (Het), or null mice. (E) Phospho-T286 cyclin D1-specific immunohistochemistry of Braf^V600E/Fbxo4-deficient tumors.

Fig 4

Distinct Fbxo4 substrate-binding domain mutants selectively disrupt recruitment of different substrates. (A) 293T cells overexpressing Fbxo4, αB crystallin, and cyclin D1/CDK4 were subjected to MG132 proteasomal inhibition. Fbxo4 was immunoprecipitated from lysates and assessed for substrate/cofactor binding. EV, empty vector. (B) Cell extracts prepared from 293T cells expressing Fbxo4 and TRF1 and treated with MG132 proteasomal inhibition were subjected to immunoprecipitation (IP) for Fbxo4, and substrate binding was assessed by immunoblotting. (C) Lysates prepared from the indicated cells treated with MLN9708 to inhibit SCF function and stabilize SCF-cyclin complexes were subjected to precipitation with a cyclin D1-specific monoclonal antibody (top and middle) or were subjected to direct Western blotting (WB) (bottom). Lysates resolved by SDS-PAGE were transferred onto membranes and blotted with the indicated antibodies.

Fig 5

The FBXO4 I377M mutant fails to regulate cyclin D1 in vivo. (A) Lysates from 293T cells expressing FBXO4, αB crystallin, and cyclin D1/CDK4 were immunoprecipitated for ubiquitin and blotted for polyubiquitylated cyclin D1. (B) Human melanoma 451Lu (WT FBXO4) and 1205Lu (FBXO4 I377M) cells were subjected to MG132 proteasomal inhibition and lysed, and cell extracts were immunoprecipitated for ubiquitin and immunoblotted for polyubiquitylated cyclin D1 (left) or Trf1 (right and middle). (C) Human melanoma 451Lu, WM88, WM983B, WM3918, 1205Lu, and WM793B cells (lanes 1 to 6, respectively) were lysed and analyzed by direct Western analysis for cyclin D1, pT286 cyclin D1, TRF1, FBXO4, and pS780 Rb expression. (D) Quantitative PCR for cyclin D1 mRNA from the indicated cell lines. (E) WM793B cells were infected with empty puro-pBabe-, pBabe-GFP-, WT FBXO4-, or FBXO4 I377M-containing virus and subsequently analyzed by direct Western blotting for cyclin D1, FBXO4, or GFP expression. (F) To assess the cyclin D1-ubiquitylating potential of TRF1-binding-defective mutants, FBXO4, αB crystallin, and cyclin D1/CDK4 were immunoprecipitated for cyclin D1 and probed for ubiquitylated species.

Fig 6

Cyclin D1 turnover is defective, and nuclear accumulation occurs in the presence of the Fbxo4 I377M mutant. (A) Fbxo4-null MEFs were transiently transfected with WT FBXO4 or the FBXO4 I377M mutant, and the endogenous cyclin D1 half-life was assessed by cycloheximide (CHX) chase. (B) Human melanoma cells were treated with cycloheximide and analyzed for endogenous cyclin D1 and FBXO4 expression. (C) Human melanoma cells with or without the FBXO4 mutation were treated with or without MG132 (2 μM for 16 h) and analyzed for endogenous cyclin D1 expression. (D) Small interfering RNA (Si) Fbxo4 knockdown in 1205Lu or 451Lu cells. Immunoblotting antibodies are indicated. (E) 293T cells overexpressing WT FBXO4 or the FBXO4 I377M mutant and cyclin D1/CDK4 were treated with or without MG132 (2 μM for 16 h) and analyzed for cyclin D1 expression. (F) Cells infected with vectors encoding the indicated proteins were seeded in soft agar, and colonies were quantified after 21 days of growth. Quantifications and error bars represent results from 3 independent experiments. (G) Fbxo4-null MEFs were transiently transfected with WT FBXO4 or the FBXO4 I377M mutant, and endogenous cyclin D1 localization was analyzed by immunofluorescence. (H) Human melanoma cells with or without the FBXO4 mutation were analyzed by immunofluorescence for endogenous cyclin D1 localization.