Identification and characterization of a novel folliculin-interacting protein FNIP2

Abstract

Birt-Hogg-Dube' syndrome characterized by increased risk for renal neoplasia is caused by germline mutations in the BHD/FLCN gene encoding a novel tumor suppressor protein, folliculin(FLCN), which interacts with FNIP1 and 5'-AMP-activated protein kinase(AMPK). Here we report the identification and characterization of a novel FNIP1 homolog FNIP2 that also interacts with FLCN and AMPK. C-terminally-deleted FLCN mutants, similar to those produced by naturally-occurring germline mutations in BHD patients, were unable to bind FNIP2. These data taken together with our previous results that demonstrated FNIP1 binding to the C-terminus of FLCN suggest that FLCN tumor suppressor function may be facilitated by interactions with both FNIP1 and FNIP2 through its C-terminus. Furthermore, we demonstrate that FNIP1 and FNIP2 are able to form homo- or heteromeric multimers suggesting that they may function independently or cooperatively with FLCN. Differential expression of FNIP1 and FNIP2 transcripts in some normal tissues may indicate tissue specificity for these homologs. Interestingly FNIP1 and FNIP2 were oppositely expressed in human clear cell renal cell carcinoma (RCC), and coordinately expressed in chromophobe RCC and oncocytoma, suggesting their differential function in different histologic variants of RCC.

Figure 1

Identification of FNIP2 and evolutionary divergence among FNIP1 and FNIP2 homologs. (A) Amino acid sequence alignment of human FNIP1 and FNIP2 with 49% identity and 74% similarity. Asterisk, identity; dots, similarity. (B) Evolutionarily conserved sequence blocks in FNIP2. The sequences were aligned with ClustalX (1.8). Seven conserved sequence blocks were identified with similarities > 30%. (C) The phylogenetic tree for the FNIP1 and FNIP2 proteins. The tree was constructed from a multiple alignment of FNIP1 and FNIP2 protein sequences from human, mouse, Xenopus, Drosophila and Caenorhabditis by the neighbor-joining method. A distance scale bar is indicated.

Figure 2

Relative mRNA expression levels of FLCN, FNIP1 and FNIP2 in human tissues analyzed by qRT-PCR. (A) FLCN (black), FNIP1 (red) and FNIP2 (blue) expression levels. Mean+standard deviation (SD). (B) FNIP2 immunohistochemistry staining in tubules of normal human kidney.

Figure 3

FNIP2 interaction with FLCN and AMPK and its subcellular localization. (A) Co-immunoprecipitation of endogenous FLCN and AMPK subunits with HA-FNIP2. Doxycycline-inducible HA-FNIP2 expressing HEK 293 cell lysates were immunoprecipitated with anti-HA antibody and blotted with antibodies as indicated, showing that FLCN and AMPK subunits interact with FNIP2. (B) Endogenous interaction of FNIP2 with FLCN and AMPK. HEK293 cell lysates were immunoprecipitated with control IgG, anti-FNIP2 (FNIP2-3G) and anti-FLCN (FLCN-105) antibodies followed by western blotting with indicated antibodies. (C) Cytoplasmic co-localization of FLCN and FNIP2 with a reticular pattern. HA-FLCN and FNIP2-Flag co-transfected into MCF7 cells were detected with anti-HA and anti-FNIP2 (FNIP2-3G) antibodies. Red, HA-FLCN; Green, FNIP2-Flag; Blue, DAPI. (D) C-terminus of FLCN is critical for FNIP2 binding. Recombinant GST-FLCN and GST-FLCN deletion mutants expressed in Sf9 cells were immobilized on glutathione-Sepharose beads and incubated with ³⁵S-labeled in vitro translated (IVT) FNIP2. FNIP2 binding was evaluated by SDS-PAGE followed by autoradiography. CBB (Coomaissee Brillant Blue) staining shows relative expression of the GST-FLCN fragments.

Figure 4

FNIP1 and FNIP2 multimer formation. (A) Doxycycline-inducible HA-FLCN-expressing HEK293 cells co-transfected with V5-FNIP1 and Flag-FNIP2 were immunoprecipitated with anti-V5 or anti-Flag antibody followed by western blotting with indicated antibodies. V5-FNIP1 and Flag-FNIP2 were reciprocally co-immunoprecipitated independent of FLCN overexpression. (B) Endogenous FNIP1 and FNIP2 form a multimer. Control IgG or anti-FNIP2(FNIP2-3G) immunoprecipitates from HEK293 cells were separated by SDS-PAGE followed by western blotting with anti-FNIP1(FNIP1-189) and anti-FNIP2(FNIP2-3G) antibodies. (C) FNIP1 conserved domains contribute to FNIP2 binding. Recombinant GST-FNIP1 and GST-FNIP1 deletion mutants expressed in Sf9 cells were immobilized on glutathione-Sepharose beads and incubated with ³⁵S-in vitro translated (IVT) FNIP2. FNIP2 binding was evaluated by SDS-PAGE followed by autoradiography. CBB staining shows relative expression of the GST-FNIP1 fragments. The numbered blocks in the diagram indicate FNIP1 sequence blocks conserved across species (Baba et al., 2006). (D) All homo- and hetero-multimers of FNIP1 and FNIP2 are able to bind FLCN and AMPK. To purify all possible FNIP1/FNIP2 molecular complexes, HA-FNIP1, Flag-FNIP1, Flag-FNIP2 and HA-FNIP2 expression vectors were transfected into HEK293 cells in combinations as indicated. HA-immunoprecipitates of cell lysates were eluted by HA peptide competition followed by sequential immunopurification by anti-Flag antibody. Final immunoprecipitates were boiled with SDS-sample buffer followed by western blotting with antibodies indicated.

Figure 5

FLCN, FNIP1 and FNIP2 mRNA expression analysis in sporadic renal cell carcinoma and normal kidney. FLCN(A), FNIP1(B) and FNIP2(C) mRNA levels in human clinical samples were evaluated by qRT-PCR. The gene expression levels are shown by dot and box-whisker plots. FLCN mRNA expression in normal kidney and tumors of various histologies was not significantly different. FNIP1 expression was significantly higher in clear cell and chromophobe RCC relative to normal kidney. Marginally higher FNIP1 mRNA levels were detected in collecting duct carcinoma compared to normal kidney. FNIP2 mRNA levels were significantly lower in clear cell RCC and higher in oncocytoma relative to normal kidney. Box, borders of the 25% and 75% quartiles; horizontal bar, median value; whiskers, ranges in each group; vertical axis, relative gene expression by qRT-PCR; horizontal axis, tumor categories.