Bicc1 Polymerization Regulates the Localization and Silencing of Bound mRNA

Abstract

Loss of the RNA-binding protein Bicaudal-C (Bicc1) provokes renal and pancreatic cysts as well as ectopic Wnt/β-catenin signaling during visceral left-right patterning. Renal cysts are linked to defective silencing of Bicc1 target mRNAs, including adenylate cyclase 6 (AC6). RNA binding of Bicc1 is mediated by N-terminal KH domains, whereas a C-terminal sterile alpha motif (SAM) self-polymerizes in vitro and localizes Bicc1 in cytoplasmic foci in vivo. To assess a role for multimerization in silencing, we conducted structure modeling and then mutated the SAM domain residues which in this model were predicted to polymerize Bicc1 in a left-handed helix. We show that a SAM-SAM interface concentrates Bicc1 in cytoplasmic clusters to specifically localize and silence bound mRNA. In addition, defective polymerization decreases Bicc1 stability and thus indirectly attenuates inhibition of Dishevelled 2 in the Wnt/β-catenin pathway. Importantly, aberrant C-terminal extension of the SAM domain in bpk mutant Bicc1 phenocopied these defects. We conclude that polymerization is a novel disease-relevant mechanism both to stabilize Bicc1 and to present associated mRNAs in specific silencing platforms.

FIG 1

Bicc1 protein forms cytoplasmic clusters in mouse kidney cells and bile duct cholangiocytes. (A) Frozen sections of WT mouse kidney labeled with anti-Bicc1 antibodies and with the proximal tubule marker LTL on postnatal day 4. (B) Frozen sections of liver of a WT mouse obtained postnatally labeled with anti-Bicc1 and anti-CK19 antibodies. CK19 is an intermediate filament protein of epithelial tissues. The boxed areas in the left panels are magnified in the three panels to the right. Bars, 10 μm (large views) and 2 μm (magnified views).

FIG 2

Bicc1 concentrates an associated reporter mRNA in cytoplasmic foci. (A) Principle of the MS2-YFP colocalization assay. The 3′ extremity of the Luc-AC6 reporter mRNA was fused to 27 MS2 hairpins, which constitute multiple binding sites for the MS2 protein fused to YFP. CDS, coding sequence; prox, proximal. (B) RNA coimmunoprecipitation. The Luc-AC6-MS2×27 reporter mRNA was expressed in HEK293T cells together with HA-Bicc1 or empty vector. After HA immunoprecipitation (IP), the various fractions were analyzed by Western blotting and by RT-PCR. Five percent of the total extract was used as the input. The β-actin mRNA was used as a negative control for the RT-PCR. HA-Bicc1 lacking all KH and KH-like domains (ΔKH) was used as an additional negative control for RNA-binding specificity. (C) Cotransfection of the fluorescent YFP-MS2 fusion protein does not affect the silencing of the Luc-AC6-MS2×27 reporter by HA-Bicc1. β-Galactosidase was used as a control for normalization, and the data represent the percent expression relative to that of a mock-treated control. Error bars show SEMs. *, P < 0.005. (D) Localization by indirect immunofluorescence staining of the Luc-AC6-MS2×27 reporter mRNA and HA-Bicc1 in COS-1 cells. The MS2-tagged mRNA was detected by the relocalization of fluorescent MS2-YFP fusion protein, which binds the MS2 RNA hairpins. Bars, 5 μm.

FIG 3

Molecular modeling of a Bicc1 SAM polymer. (A) ClustalW alignment of mouse (Mus musculus) Bicc1 (mBicc1) and human (Homo sapiens) DGKδ1 (hDGKδ1) SAM domains. The X-ray structure of dimeric DGKδ1 SAM (PDB accession number 3BQ7) served as the template to model the Bicc1 SAM dimer. The two SAM domains share 31% identity and 54% similarity. Predicted α helices are framed. Residues of the ML surface (red), residues from the EH surface (blue), and hydrophobic residues (underlined) are highlighted. Dark and light gray shading corresponds to identical and similar amino acids, respectively, between Bicc1 and the DGKδ1 template. (B) Dimeric Bicc1 SAM model obtained using MODELLER (v9.5) software. The α helix numbers and the side chains of the residues involved in the interface are displayed. (C to E) Magnified views of the main interacting patches in the predicted Bicc1 SAM dimer interface. Acidic and basic residues are displayed in red and blue, respectively. (F) Model of a Bicc1 SAM polymer of 24 units in surface representation. The NH₂ terminus and the COOH terminus of each SAM domain are displayed in green and purple, respectively. (G) Model of the Bicc1 KH domain region in surface representation. Models for individual KH domains were obtained by homology modeling using the SWISS-MODEL work space (48) and templates consisting of the structures with PDB accession numbers 1VIG (KH1), 2CTM (KH2), 1WVN (KH3), and 3N89 (KHL1 and -2). Individual KH domains were then superimposed with their homologous domain in the X-ray structure of ceGLD-3 KH domains (PDB accession number 3N89) (58). The KH domains harboring the GXXG signatures for RNA binding are highlighted in color. Their putative RNA-binding surfaces are darkened, and the identity of their GXXG signature sequence is given in parentheses. The KH-like domains (KHL1 and -2) are displayed in gray. (H) Diagram of a transversal section through a polymer of full-length Bicc1. The SAM polymer is located at the center and displays other Bicc1 domains at its periphery. A schematic representation was used for the other domains. C-ter domain, C-terminal domain. (I) Diagram in longitudinal view of Bicc1 KH domains distributed along the surface of the central SAM polymer.

FIG 4

Screen for Bicc1 SAM polymer mutants. (A) Bicc1 SAM mutant collection. Shading is as for Fig, 3A. Individual electrostatic patches (patches A to F) at the protein surface were replaced by alanines. (B, C) Positions of mutations in the Bicc1 SAM dimer model (B) and on the ML surface and EH surface (C). (D) Table summarizing, for each amino acid patch, the average number of H bonds per time frame during the MD. (E and F) Pulldown of the WT, point mutants, or 8-fold excess bpk mutant HA-tagged Bicc1 from HEK293T cell extracts by glutathione-Sepharose beads coated with a recombinant GST control or GST-Bicc1 SAM. Five percent of total cell extracts were loaded as input.

FIG 5

SAM polymerization is required for Bicc1 clustering. (A) Density fractionation of WT and polymerization mutant Bicc1 on a sucrose gradient. HEK293T cell extracts containing HA-tagged Bicc1 were fractionated on a continuous 15 to 60% sucrose gradient and analyzed by anti-Bicc1 Western blotting. The migration direction from the top to the bottom of the tube is indicated. RPS6 and γ-tubulin (γ-Tub) were used as internal controls. The graph below the gels shows the percentage of Bicc1 compared to the total Bicc1 signal for each fraction. Results represent mean values from 3 independent experiments, and error bars show SEMs. (B) Bicc1 polymer mutants fail to accumulate in cytoplasmic foci. The results of indirect immunofluorescence staining of the HA-Bicc1 WT, mutD, the ΔSAM mutant, or the bpk mutant and the P-body marker GFP-Dcp1a overexpressed in COS-1 cells are shown. Bars, 5 μm. (C) Comparative 3D rendering of the HA-Bicc1 WT and mutD by Imaris software. From the original image (center), z-stacks in two directions (z1 and z2, top and right, respectively) and 3D reconstruction (bottom) are given. Bars, 2 μm. (D) Localization by indirect immunofluorescence staining of the Luc-AC6-MS2×27 reporter mRNA and HA-Bicc1 in COS-1 cells and comparison with that of HA-Bicc1 mutD. Bars, 5 μm.

FIG 6

SAM polymerization is required for Bicc1 accumulation and silencing activity. (A) Time course analysis of HA-Bicc1 WT and mutants expressed in HEK293T cells after CHX treatment. HA-Bicc1 protein levels were compared with γ-tubulin levels by Western blotting at 0, 8, 24, 32, 48, and 56 h. The relative level of each protein is presented in the graph at the bottom, and the estimated half-life is given on the right. AU, arbitrary units. (B) Level of the HA-Bicc1 WT and mutants upon transfection with a single dose (×1) or a double dose (×2) of DNA encoding HA-Bicc1. A double dose of transfected DNA encoding HA-Bicc1 mutD, the ΔSAM mutant, or the bpk mutant is required to obtain a protein level comparable to that of the HA-Bicc1 WT. The relative percentage of WT Bicc1 is indicated for each condition. γ-Tubulin was used for normalization. (C and D) Silencing of AC6 and PKIα 3′ UTR luciferase reporters by WT or polymerization mutant Bicc1 in HEK293T cells. β-Galactosidase was used as a control for normalization. Error bars show SEMs. *, P < 0.005. (E) Induction of the TOPflash reporter of Wnt signaling by Dishevelled 2 (Dvl2) in transfected HEK293T cells is inhibited by both WT and polymerization mutant Bicc1. β-Galactosidase was cotransfected for signal normalization. Error bars show SEMs. *, P < 0.005.

FIG 7

Impaired localization of Bicc1 in bpk mutant mouse kidney. (A) Frozen sections of WT and bpk mutant mouse kidneys labeled with anti-Bicc1 antibodies and with the proximal tubule marker LTL on postnatal day 11. Boxed areas are magnified in the panels to the right. Bars, 20 μm (large views) and 10 μm (magnified views). (B) Bicc1 isoform B (Bicc1-B) accumulates in cytoplasmic foci. The results of indirect immunofluorescence staining of HA-Bicc1 isoforms A and B and the P-body marker GFP-Dcp1a in transfected COS-1 cells are shown. Bars, 5 μm. (C) Silencing of AC6 3′ UTR luciferase reporter by HA-Bicc1 isoform A or B in HEK293T cells. β-Galactosidase was cotransfected for signal normalization. Error bars show SEMs. *, P < 0.005.