Reduced affinity to and inhibition by DKK1 form a common mechanism by which high bone mass-associated missense mutations in LRP5 affect canonical Wnt signaling

Abstract

The low-density-lipoprotein receptor-related protein 5 (LRP5), a coreceptor in the canonical Wnt signaling pathway, has been implicated in human disorders of low and high bone mass. Loss-of-function mutations cause the autosomal recessive osteoporosis-pseudoglioma syndrome, and heterozygous missense mutations in families segregating autosomal dominant high bone mass (HBM) phenotypes have been identified. We expressed seven different HBM-LRP5 missense mutations to delineate the mechanism by which they alter Wnt signaling. None of the mutations caused activation of the receptor in the absence of ligand. Each mutant receptor was able to reach the cell surface, albeit at differing amounts, and transduce exogenously supplied Wnt1 and Wnt3a signal. All HBM mutant proteins had reduced physical interaction with and reduced inhibition by DKK1. These data suggest that HBM mutant proteins can transit to the cell surface in sufficient quantity to transduce Wnt signal and that the likely mechanism for the HBM mutations' physiologic effects is via reduced affinity to and inhibition by DKK1.

FIG. 1.

Schematic of LRP5 expression constructs and sites of HBM-associated missense mutations. (A) WT and high bone mass-associated mutant LRP5 protein expression constructs. Each construct contained either a WT sequence or a single HBM-associated mutation. Specific amino acid mutations (in single-letter code) and their relative locations within the first EGF-like domain of LRP5 are noted. Constructs were full-length, untagged LRP5, full-length LRP5 that was tagged at the carboxy terminus with a myc-epitope (LRP5-myc), truncated LRP5 protein lacking the transmembrane and cytoplasmic domains but tagged at the carboxy terminus with a myc-epitope (LRP5N-myc). (B) Locations of the HBM mutations modeled on the three-dimensional structure of the EGF-like domain of the low-density-lipoprotein receptor (14) by using the RasMol molecular graphics visualization tool (version 2.6, September 1996 update [http://www.umass.edu/microbio/rasmol/getras.htm]; R. Sayle, Stevanage, United Kingdom). Note that all mutations affect residues near the top surface (side view) and central region (top view) of the β-propeller structure within the EGF-like domain.

FIG. 2.

Synthesis and trafficking of WT-LRP5 and seven different HBM-LRP5 mutants in transiently transfected Cos7 and 293T cells. (A) Western blot analysis of CM and cell lysate (Ly) from Cos7 cells that had been transiently transfected with LRP5N-myc constructs, separated by reducing SDS-PAGE (7.5% gel), and immunodetected by using an anti-myc antibody. Note that the molecular mass of LRP5N-myc in the conditioned medium is greater than the mass of that in the cell lysate, indicative of posttranslational modification during trafficking. Also note that the relative amounts of recombinant protein in conditioned medium in comparison to cell lysate differ between the different mutants. (B) Western blot analysis of WT and HBM-LRP5N-myc expressed in 293T cells and immunodetected after reducing SDS-PAGE (7.5% gel) with an anti-myc antibody. Comparable rates of synthesis occurred for all constructs based upon immunodetection of LRP5N-myc in the cell lysate. As with the Cos7 cells (panel A), the G171R, G171V, and A242T mutants had low abundance in CM of 293T cells compared with that in WT-LRP5N-myc. Equal loading of cell lysate and CM in each lane is demonstrated by immunodetection of cell lysate with an antitubulin antibody and Coomassie staining of conditioned medium. (C) Western blot analysis of affinity purified, biotin-labeled cell surface protein from Cos7 cells transfected with full-length LRP5-myc constructs separated by reducing SDS-PAGE (4 to 15% gradient gel) and immunodetected by using an anti-myc antibody. Note that the ability of different HBM-LRP5-myc proteins to reach the cell surface mirrors the abundance of the LRP5N-myc form of that mutant in the conditioned media of Cos7 and 293T cells (panels A and B). The negative controls comprise cell lysates from nonbiotinylated but LRP5-myc-transfected cells. The positive control in the far right column is a control for the anti-myc antibody. Integrin β1 immunodetection demonstrates equivalent biotin labeling of surface proteins between the different transfected cells. Input LRP5-myc comprises cell lysates from each of the LRP5-myc transfected cells, which demonstrates equivalent levels of expression.

FIG. 3.

HBM-causing mutations differentially affect the interaction between LRP5 and MESD-C2. Full-length LRP5-myc constructs were coexpressed with Flag-tagged MESD-C2 (MESD-flag) in 293T cells. Coimmunoprecipitation was carried out using the cell lysates. Cells that had been transfected only with MESD-Flag or LRP5-myc expression vectors served as negative controls. Cell lysate was immunoprecipitated by use of an anti-myc antibody (A) or an anti-Flag antibody (B). (A) Western blot of anti-myc-immunoprecipitated protein that was immunodetected by use of an anti-Flag antibody (IB: Flag) after reducing SDS-PAGE (4 to 15% gradient gel). Note that immunoprecipitation of WT-LRP5-myc at high stringency also precipitated MESD-Flag. Several (D111Y, A214V, and A242T) but not all HBM-LRP5-myc mutants were able to precipitate MESD-Flag at high stringency. To ensure equal immunoprecipitation of LRP5-myc proteins in the assay, LRP5-myc was immunodetected on the same blot by use of the anti-myc antibody (IB: myc). To ensure equal expression of MESD-Flag, cell lysates were immunodetected with anti-Flag antibody (Input: MESD-flag). MESD-Flag migrated as an ∼30-kDa protein in this experiment. (B) Western blot of anti-Flag-immunoprecipitated protein, immunodetected by use of an anti-myc antibody (IB: myc) following reducing SDS-PAGE (4 to 15% gradient gel). Note that immunoprecipitation of MESD-Flag at high stringency comparably precipitated WT-LRP5-myc and several HBM mutants (D111Y, A214V, and A242T), while other HBM mutants were less efficiently precipitated. To ensure equal immunoprecipitation of MESD-Flag in the assay, MESD-Flag was immunodetected on the same blot by using the anti-Flag antibody (IB: flag). To ensure equal expression of LRP5-myc, cell lysates were immunodetected with anti-myc antibody (Input: LRP5-myc). Note that both assays identified the same HBM-LRP5-myc mutants as having affinities to MESD-Flag that are comparable to that of WT-LRP5-myc.

FIG. 4.

Wnt-induced canonical signaling in 293T cells expressing WT-LRP5 or HBM-LRP5. Wnt-induced signaling via the canonical pathway is depicted as the increase (n-fold) in firefly luciferase (Topflash) activity normalized to Renilla luciferase activity. (A) Wnt induction by transfecting cells with LRP5 alone (black bars) or cotransfecting LRP5 with Wnt1-V5 and then culturing in the presence of pcDNA3-CM (gray bars) or DKK1-CM (open bars). (B) Wnt induction by coculturing LRP5-transfected cells with Wnt1-expressing Rat2 cells in the presence of pcDNA3-CM (shaded bars) or DKK1-CM (open bars). (C) Wnt induction by adding conditioned medium from Wnt3a-expressing L cells to LRP5-transfected cells and then culturing in the presence of pcDNA3-CM (shaded bars) or DKK1-CM (open bars). (D) Wnt induction by cotransfecting cells with LRP5 and mouse Wnt10b and then culturing in the presence of pcDNA3-CM (shaded bars) or DKK1-CM (open bars). (E) Representative Western blot of cell lysates from transfected cells separated by reducing SDS-PAGE and immunodetected with anti-LRP5 antibody, demonstrating equal levels of LRP5 expression (cell lysates from the experiment depicted in panel A are shown, but similar results were observed in all experiments). To control for equal loading, the same blot was immunodetected by antitubulin antibody. Panels A through D depict results for single experiments performed in triplicate. However, each experiment was performed on at least three separate occasions, and similar results were obtained.

FIG. 5.

All HBM-LRP5 mutants have reduced affinities for DKK1 compared to WT-LRP5. (A) Conditioned medium containing recombinant human DKK1-Flag protein was mixed with conditioned medium containing secreted LRP5N-myc or control conditioned medium from cells transfected with empty vector (pcDNA3). LRP5N-myc was immunoprecipitated from the mixed medium by use of an anti-myc antibody and coprecipitation of DKK1-Flag was immunodetected by use of an anti-Flag antibody (IB: DKK1-flag) after reducing SDS-PAGE (4 to 15% gradient gel). DKK1-Flag migrated as an ∼40-kDa band in this experiment. To demonstrate equal immunoprecipitation of LRP5N-myc, the same Western blot was immunodetected with anti-myc antibody (IB: LRP5N-myc). Note that none of the HBM-LRP5N-myc proteins could coprecipitate DKK1-Flag as efficiently as WT-LRP5N-myc or the T173M-LRP5N-myc. The greater-molecular-weight immunodetectable band in all samples, including the control, is IgG heavy chain. (B) Conditioned medium containing recombinant human DKK1-V5 protein was mixed with conditioned medium containing LRP6N-Fc or control conditioned medium from cells transfected with empty vector (pcDNA3). LRP6N-Fc was immunoprecipitated by use of anti-mouse IgG agarose beads. The coprecipitated DKK1-V5 protein was immunodetected by anti-V5-HRP (IB: DKK1-V5). Equal immunoprecipitation of LRP6N-Fc protein was demonstrated by immunoblotting the same Western blot with goat anti-mouse IgG (IB: LRP6N-Fc). Note that WT-LRP6N-Fc was able to precipitate DKK1, whereas G158V-LRP6N-Fc was not.