ABSTRACT
Low-density lipoprotein (LDL) receptor-related protein 6 (LRP6) was originally identified as a co-receptor of the Wnt signalling pathway and has been shown to be involved in LDL transport. In polarized hepatocytes, many apical proteins are sorted to the basolateral membrane and then internalized and transported to the apical bile canalicular membrane – a process known as transcytosis. We show that LRP6 is transcytosed to the apical membrane of polarized hepatic HepG2 cells via a flotillin-dependent manner in the absence of LDL. LRP6 formed a complex with Niemann–Pick type C1-like 1 (NPC1L1), which is localized to the bile canalicular membrane of the liver and is involved in cholesterol absorption from the bile. LRP6 was required for apical membrane localization of NPC1L1 in the absence of LDL. Clathrin-dependent LRP6 internalization occurred in the presence of LDL, which resulted in trafficking of LRP6 to the lysosome, thereby reducing apical sorting of LRP6 and NPC1L1. These results suggest that LRP6 endocytosis proceeds by two routes, depending on the presence of LDL, and that LRP6 controls the intracellular destination of NPC1L1 in hepatocytes.
INTRODUCTION
Low-density lipoprotein (LDL) receptor-related protein 6 (LRP6) is a member of a gene family that encodes LDL receptors (LDLRs), which reside on the cell surface membrane and participate in endocytosis of their ligands (Schneider and Nimpf, 2003; He et al., 2004). For instance, LRP6 acts as a co-receptor for Wnt, which activates the β-catenin signalling pathway and regulates development and the adult life of animals (Kikuchi and Yamamoto, 2007; Kikuchi et al., 2011). In various cell types, LRP6 is internalized in caveolin- and clathrin-dependent manners in response to Wnt3a and Dickkopf1 (Dkk1), respectively (Yamamoto et al., 2008). Caveolin-dependent LRP6 internalization is required for Wnt3a-dependent signalling (Yamamoto et al., 2006; Kikuchi et al., 2009; Sakane et al., 2010; Jiang et al., 2012; Demir et al., 2013; Özhan et al., 2013). Therefore, it is likely that LRP6 is internalized via different routes depending on ligand availability.
LRP6 is also involved in LDL uptake in NIH3T3, lymphoblastoid and Chinese hamster ovary (CHO) cells (Liu et al., 2008; Ye et al., 2012). Individuals with a rare non-conservative mutation of LRP6 develop high LDL cholesterol levels in their third or fourth decades of life that are comparable to those of patients with heterozygote familial hypercholesterolaemia (Mani et al., 2007). A mutation in the second β-propeller–EGF-like domain pair (EGFP) domain of LRP6 (LRP6R611C) confers reduced LRP6 membrane expression and decreased LDL clearance (Liu et al., 2008). In addition, common variations within the LRP6 gene have been associated with modest elevations in serum LDL in the general population (Tomaszewski et al., 2009). These results suggest that LRP6 is involved in LDL clearance.
Polarized cells generally display two structurally and functionally different plasma membrane domains – apical and basolateral – that are separated by tight junctions. Each domain has a specific protein and lipid composition and is associated with specialized functions. The generation of distinct molecular identities in each domain requires sorting of lipids and proteins and subsequent vectorial traffic to their specific surface destinations (Mellman and Nelson, 2008). Basolateral sorting of secretory and membrane proteins involves trafficking to the membrane in vesicles that are sorted in clathrin- and adaptor protein 1 (AP-1)-dependent manners (Gravotta et al., 2007; Fölsch, 2005; Gravotta et al., 2012). LRP6 has been shown to be trafficked to the basolateral membrane by a process that requires clathrin and AP-1 in polarized epithelial Madin-Darby canine kidney (MDCK) cells (Yamamoto et al., 2015). There are two routes for apical sorting: (1) proteins can be trafficked to the apical surface directly from the Golgi in a process that varies mechanistically depending on the presence of sorting proteins (Kipp and Arias, 2000; Paladino et al., 2006); or (2) proteins can be transported to the apical membrane indirectly by making a detour to the basolateral surface. This latter route is known as transcytosis. Apical transport in epithelial cells relies on these direct and indirect pathways to different degrees, depending on the tissue type (Rodriguez-Boulan et al., 2005).
Hepatocytes, the major epithelial cells in the liver, are highly polarized. Unlike columnar epithelial cells, hepatocyte plasma membranes are separated by tight junctions into sinusoidal–basolateral and bile canalicular–apical domains. The apical domain, which is localized at the cell's apex on the lateral surface, forms a central lumen and delimits a bile canaliculus. In hepatocytes, the predominant transport route to the apical membrane is transcytosis (Aït Slimane and Hoekstra, 2002; Tuma and Hubbard, 2003). Although hepatic HepG2 cells are cancer cells, they are well differentiated, retain many hepatocyte-specific functions (Knowles et al., 1980), and form microfilament-lined vacuoles that resemble bile canaliculi (Chiu et al., 1990).
In this study, we examined the distribution and trafficking routes of LRP6 in a polarized hepatocyte model (HepG2 cells). We found that once LRP6 is trafficked to the basolateral membrane, LRP6 is transcytosed with Niemann–Pick type C1-like 1 (NPC1L1) protein to the apical bile canalicular membrane in a flotillin-dependent manner in the absence of LDL. Hepatic NPC1L1 may play a role in the absorption of cholesterol from the bile (Altmann et al., 2004; Temel et al., 2007; Kurano et al., 2012). In addition, we determined that LRP6 was internalized in a clathrin-dependent manner in the presence of LDL, which decreased trafficking of NPC1L1 to the apical membrane. These results suggest that LRP6 is internalized via different endocytic routes, depending on ligand availability, and that LRP6 controls the intracellular destination of NPC1L1.
RESULTS
LRP6 is trafficked to the apical membrane via transcytosis
The bile canalicular membrane in the liver is highly enriched with actin filaments (Zegers et al., 1998). Ezrin, radixin and moesin (ERM) proteins function as linkers between the plasma membrane and the actin cytoskeleton. Radixin is the dominant ERM protein in hepatocytes (Wang et al., 2006), and phosphoradixin (Thr564) is associated with the bile canalicular membrane similar to actin (Suda et al., 2011). The bile canalicular membrane is developed in polarized hepatic HepG2 cells (Slim et al., 2013). The apical membrane region is identifiable as phospho-ERM- and phalloidin-stained round or oval structures surrounded by two or more cells (Fig. S1). To examine whether endogenous LRP6 is localized to the apical or the basolateral membrane, polarized HepG2 cells were immunostained with anti-LRP6 antibody (Fig. 1A,B). LRP6 was detected at the apical membrane, where F-actin and phospho-ERM were observed (Fig. 1A,B). Approximately 65% of apical membrane regions were positive for LRP6 (Fig. 1A,C). When LRP6 was knocked down by siRNA, LRP6 could no longer be detected at the apical membrane (Fig. 1C). These results suggest that endogenous LRP6 is present primarily at the apical membrane of HepG2 cells.
LRP6 is localized to the apical membrane. (A,B) HepG2 cells were stained with anti-LRP6 antibody (green), Alexa-Fluor-546-conjugated phalloidin (red), and anti-phospho-ERM antibody (blue). Localization of LRP6 at the apical membrane regions is indicated by yellow arrowheads in the leftmost panel. Inset, high magnification. A single apical region is enlarged in B. (C) After HepG2 cells were treated with LRP6 siRNA, lysates were probed with anti-LRP6 and anti-β-tubulin antibodies (left panels), and the percentage of apical membrane regions that accumulated LRP6 was calculated (right panel). (D) After HepG2 cells were treated with CHX for 5 h, cells were further incubated in CHX-free medium for 4 or 8 h. Cells then were stained with anti-LRP6 antibody (green) and Alexa-Fluor-546-conjugated phalloidin (red). Accumulations of LRP6 at the apical membrane regions are indicated by white arrowheads. Scale bars: 20 μm (A,D) and 5 μm (B). (E) The percentage of apical membrane regions accumulating LRP6 in D was calculated by counting at least 100 cells that had developed apical membrane structures. The results shown are means±s.e.m. from three different fields per coverslip. (F) HepG2 cells were treated with 5 μg/ml CHX and were harvested at the indicated times. For the relief condition, cells were treated for 5 h with CHX, washed twice with CHX-free medium, and then were cultured in CHX-free medium for the indicated periods. Lysates were probed with the indicated antibodies. β-Tubulin (C) and HSP90 (F) were used as loading controls.
LRP6 is localized to the apical membrane. (A,B) HepG2 cells were stained with anti-LRP6 antibody (green), Alexa-Fluor-546-conjugated phalloidin (red), and anti-phospho-ERM antibody (blue). Localization of LRP6 at the apical membrane regions is indicated by yellow arrowheads in the leftmost panel. Inset, high magnification. A single apical region is enlarged in B. (C) After HepG2 cells were treated with LRP6 siRNA, lysates were probed with anti-LRP6 and anti-β-tubulin antibodies (left panels), and the percentage of apical membrane regions that accumulated LRP6 was calculated (right panel). (D) After HepG2 cells were treated with CHX for 5 h, cells were further incubated in CHX-free medium for 4 or 8 h. Cells then were stained with anti-LRP6 antibody (green) and Alexa-Fluor-546-conjugated phalloidin (red). Accumulations of LRP6 at the apical membrane regions are indicated by white arrowheads. Scale bars: 20 μm (A,D) and 5 μm (B). (E) The percentage of apical membrane regions accumulating LRP6 in D was calculated by counting at least 100 cells that had developed apical membrane structures. The results shown are means±s.e.m. from three different fields per coverslip. (F) HepG2 cells were treated with 5 μg/ml CHX and were harvested at the indicated times. For the relief condition, cells were treated for 5 h with CHX, washed twice with CHX-free medium, and then were cultured in CHX-free medium for the indicated periods. Lysates were probed with the indicated antibodies. β-Tubulin (C) and HSP90 (F) were used as loading controls.
When HepG2 cells were treated with cycloheximide (CHX) for 5 h, the protein level of LRP6 was decreased, and almost no endogenous LRP6 was detected at the apical membrane (Fig. 1D,E). In contrast, the protein levels of ezrin, Rab7 and HSP90 were not changed in HepG2 cells treated with CHX for 10 h (Fig. 1F), suggesting dynamic turnover of LRP6. When cells were subsequently cultured in CHX-free medium for 4 or 8 h, the protein level of LRP6 was restored, and LRP6 was detected in ∼25% and 55% of apical membrane regions, respectively (Fig. 1D,E). Thus LRP6 is trafficked to the apical membrane after protein synthesis.
To examine how LRP6 is trafficked to the apical membrane in HepG2 cells, FLAG–LRP6 was expressed to similar levels as endogenous LRP6 in HepG2 cells (Fig. 2A) and an antibody-trafficking assay was performed as described previously (Aït-Slimane et al., 2003). When cells were exposed to anti-FLAG antibody at 4°C for 1 h, FLAG–LRP6 was found to be exclusively localized to the basolateral membrane (Fig. 2B,C). Cells then were warmed (37°C), and FLAG–LRP6 was found to be internalized when 30 min to 1 h had elapsed. This internalization was represented as a decrease in basolateral membrane staining and a concomitant increase in intracellular punctate staining (Fig. 2B–D). Three hours after internalization was initiated, approximately 65% of apical membrane regions of FLAG–LRP6-expressing cells became FLAG–LRP6-positive (Fig. 2D). Thus transfected LRP6 was localized to the apical membrane, like endogenous LRP6.
LRP6 is trafficked to the apical membrane via transcytosis. (A) Lysates of HepG2 cells or HepG2 cells expressing FLAG–LRP6 were probed with anti-LRP6 and anti-FLAG antibodies. (B,C) Transcytosis of FLAG–LRP6 was examined for the indicated periods with an antibody-trafficking assay. FLAG–LRP6-expressing cells (middle panels) and apical membrane structures (bottom panels) are indicated with green and red arrowheads, respectively. In the merged images (top panels), localization of FLAG–LRP6 at the apical membrane regions is indicated by yellow arrowheads. A single apical region is enlarged in C. (D) The percentage of apical membrane regions that accumulated FLAG–LRP6 in B was calculated. (E) Transcytosis of endogenous CD59 at 3 h was examined with an antibody-trafficking assay. CD59-expressing cells (middle panels) and apical membrane structures (bottom panels) are indicated by green and red arrowheads, respectively. Localization of CD59 to the apical membrane region is indicated by yellow arrowheads in the top panels. (F) The percentage of apical membrane regions that accumulated CD59 in E was calculated. Results shown are means±s.e.m. from three different fields per coverslip. Scale bars: 20 μm (B,E) and 5 μm (C).
LRP6 is trafficked to the apical membrane via transcytosis. (A) Lysates of HepG2 cells or HepG2 cells expressing FLAG–LRP6 were probed with anti-LRP6 and anti-FLAG antibodies. (B,C) Transcytosis of FLAG–LRP6 was examined for the indicated periods with an antibody-trafficking assay. FLAG–LRP6-expressing cells (middle panels) and apical membrane structures (bottom panels) are indicated with green and red arrowheads, respectively. In the merged images (top panels), localization of FLAG–LRP6 at the apical membrane regions is indicated by yellow arrowheads. A single apical region is enlarged in C. (D) The percentage of apical membrane regions that accumulated FLAG–LRP6 in B was calculated. (E) Transcytosis of endogenous CD59 at 3 h was examined with an antibody-trafficking assay. CD59-expressing cells (middle panels) and apical membrane structures (bottom panels) are indicated by green and red arrowheads, respectively. Localization of CD59 to the apical membrane region is indicated by yellow arrowheads in the top panels. (F) The percentage of apical membrane regions that accumulated CD59 in E was calculated. Results shown are means±s.e.m. from three different fields per coverslip. Scale bars: 20 μm (B,E) and 5 μm (C).
Trafficking of glycosylphosphatidylinositol-anchored protein (GPI-AP) CD59, which inhibits the complement membrane attack complex (Rollins et al., 1991), is representative of transcytotic delivery (de Marco et al., 2002). Endogenous CD59 was present at the basolateral membrane at time 0 after keeping the cells at 4°C for 1 h. Approximately 85% of apical membrane regions were CD59-positive 3 h after the temperature was shifted to 37°C and internalization began (Fig. 2E,F). These results suggest that LRP6 and CD59 are trafficked to the apical membrane in HepG2 cells along a transcytotic pathway.
LRP6 in the detergent-resistant microdomain is internalized through a flotillin-mediated route
LRP6 was detected in the detergent-resistant microdomain (DRM) fractions and the non-DRM fractions of human cervical cancer (HeLaS3) cells and of human embryonic kidney (HEK) 293 cells (Yamamoto et al., 2006; Sakane et al., 2010). To examine the distribution of LRP6 in HepG2 cells, cell lysates were fractionated in the presence of Triton X-100 to separate them into DRM (fractions 2–4) and non-DRM (fractions 8–10) layers in a sucrose density gradient (Fig. 3A). Flotillins are considered markers of the DRM fraction (Langhorst et al., 2005); and hepatic cells express flotillin-2 but not flotillin-1 or caveolin (Volonte et al., 1999; Fu et al., 2004). Flotillins have a similar topology to caveolin and are regarded as scaffolding proteins of the DRM fractions. In hepatic cells, transcytosis of GPI-APs is DRM dependent (Aït-Slimane et al., 2003; Nyasae et al., 2003), and basolateral internalization of GPI-APs occurs via a flotillin-2-dependent pathway (Aït-Slimane et al., 2009). Consistent with previous findings (Kenworthy et al., 2004), CD59 was localized to the DRM primarily in the current study (Fig. 3A). Endogenous LRP6 and transiently expressed FLAG–LRP6 were present in the DRM and non-DRM fractions (Fig. 3A).
Transcytosis of LRP6 requires DRM. (A) HepG2 cells or HepG2 cells expressing FLAG–LRP6 were treated with or without 5 mM MβCD or 25 µM MDC for 1 h. Lysates were fractionated by sucrose density gradient centrifugation, and aliquots were probed with the indicated antibodies (top panels). Endogenous flotillin-2 and clathrin indicate the positions of DRM and non-DRM fractions, respectively. The band intensities of LRP6 in the DRM fractions (fractions 2–4) and the non-DRM fractions (fractions 8–10) were quantified. The percentage of LRP6 in the DRM fraction was calculated as (DRM/[DRM+non-DRM])×100 (bottom panel). (B) After HepG2 cells were treated with or without 5 mM MβCD or 25 µM MDC for 1 h, lysates were probed with the indicated antibodies. (C) HepG2 cells expressing FLAG–LRP6 were treated with 5 mM MβCD in the presence or absence of 0.1 mg/ml cholesterol (Chol) or 25 µM MDC. Transcytosis of FLAG–LRP6 was examined at 1 and 3 h with an antibody-trafficking assay. FLAG–LRP6-expressing cells and apical membrane structures are indicated by green and red arrowheads, respectively. Localization of FLAG–LRP6 at the apical membrane regions is indicated by yellow arrowheads in the left panels. Scale bar: 20 μm. (D) After internalization of FLAG–LRP6 from the basolateral membrane was initiated (C), the distribution of FLAG–LRP6 at 1 h was quantified (top panel). In addition, the percentage of apical membrane regions that accumulated FLAG–LRP6 at 3 h was calculated (bottom panel). The results shown are means±s.e.m. from three independent experiments (A) or three different fields per coverslip (C). *P<0.05; n.s., not significant.
Transcytosis of LRP6 requires DRM. (A) HepG2 cells or HepG2 cells expressing FLAG–LRP6 were treated with or without 5 mM MβCD or 25 µM MDC for 1 h. Lysates were fractionated by sucrose density gradient centrifugation, and aliquots were probed with the indicated antibodies (top panels). Endogenous flotillin-2 and clathrin indicate the positions of DRM and non-DRM fractions, respectively. The band intensities of LRP6 in the DRM fractions (fractions 2–4) and the non-DRM fractions (fractions 8–10) were quantified. The percentage of LRP6 in the DRM fraction was calculated as (DRM/[DRM+non-DRM])×100 (bottom panel). (B) After HepG2 cells were treated with or without 5 mM MβCD or 25 µM MDC for 1 h, lysates were probed with the indicated antibodies. (C) HepG2 cells expressing FLAG–LRP6 were treated with 5 mM MβCD in the presence or absence of 0.1 mg/ml cholesterol (Chol) or 25 µM MDC. Transcytosis of FLAG–LRP6 was examined at 1 and 3 h with an antibody-trafficking assay. FLAG–LRP6-expressing cells and apical membrane structures are indicated by green and red arrowheads, respectively. Localization of FLAG–LRP6 at the apical membrane regions is indicated by yellow arrowheads in the left panels. Scale bar: 20 μm. (D) After internalization of FLAG–LRP6 from the basolateral membrane was initiated (C), the distribution of FLAG–LRP6 at 1 h was quantified (top panel). In addition, the percentage of apical membrane regions that accumulated FLAG–LRP6 at 3 h was calculated (bottom panel). The results shown are means±s.e.m. from three independent experiments (A) or three different fields per coverslip (C). *P<0.05; n.s., not significant.
Treatment of HepG2 cells with methyl-β-cyclodextrin (MβCD), which disrupts the DRM by removing cholesterol from membranes (Nyasae et al., 2003), decreased the ratios of endogenous LRP6, FLAG–LRP6, flotillin-2 and CD59 in the DRM fractions, under conditions where the total expression levels of LRP6 were unchanged (Fig. 3A,B). One hour after FLAG–LRP6 internalization was initiated in the antibody-trafficking assay, FLAG–LRP6 was detected as punctate structures in the cytoplasm of control cells; in contrast, FLAG–LRP6 remained largely at the basolateral surface membrane in MβCD-treated cells (Fig. 3C,D). Three hours after internalization, FLAG–LRP6 was mainly present at the apical membrane of control cells; however, it persisted at the basolateral surface membrane and in several cytosolic puncta in MβCD-treated cells (Fig. 3C,D). The accumulation of FLAG–LRP6 at the apical membrane was marginal even 6 h after internalization in MβCD-treated cells (Fig. S2). These inhibitory effects of MβCD on internalization and transcytosis of FLAG–LRP6 were rescued by the addition of cholesterol (Fig. 3C,D; Fig. S2).
In contrast, monodansylcadaverine (MDC), which inhibits clathrin-mediated endocytosis, did not affect the ratios of endogenous LRP6, FLAG–LRP6, flotillin-2 and CD59 in the DRM fractions, the total expression level of LRP6, and transcytosis of LRP6 (Fig. 3A–D; Fig. S2). Furthermore, 6 h after internalization in cells treated with MβCD and MDC, FLAG–LRP6 was found to be localized to the basolateral surface membrane with several punctate structures in the cytoplasm, similar to MβCD-treated cells (Fig. S2). These results suggest that the DRM fraction is necessary for transcytosis of LRP6 and that LRP6 can be internalized in a clathrin-independent manner even though the DRM fraction is disrupted. However, under these conditions, trafficking of internalized LRP6 to the apical membrane is severely reduced.
One hour after FLAG–LRP6 was internalized from the basolateral membrane, FLAG–LRP6 was co-localized with flotillin-2, rather than clathrin, as punctate structures (Fig. 4A). Flotillin-2 and clathrin knockdown did not alter the quantities of FLAG–LRP6 localized to the basolateral cell surface (Fig. 4B,C). In flotillin-2-depleted HepG2 cells, FLAG–LRP6 persisted at the basolateral membrane 1 h after internalization was initiated. Three hours post-internalization, FLAG–LRP6 was still detected at the basolateral surface membrane and in punctate structures in the cytoplasm; accumulation of FLAG–LRP6 at the apical membrane was marginal (Fig. 4D,E). These phenotypes were similar to those of cells treated with MβCD. Transcytosis of CD59 to the apical membrane was also suppressed in flotillin-2-depeleted cells (Fig. S3). In contrast, clathrin knockdown did not affect transcytosis of FLAG–LRP6 (Fig. 4B,D,E). Collectively, these results suggest that flotillin-2 is required for transcytosis of LRP6 in HepG2 cells.
Transcytosis of LRP6 requires flotillin-2. (A) One hour after internalization of FLAG–LRP6 at the basolateral membrane was initiated, HepG2 cells were stained with anti-FLAG (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies. Co-localization of FLAG–LRP6 and flotillin-2 or clathrin is depicted in yellow in the top panels. The percentages of internalized FLAG–LRP6 co-localized with flotillin-2 (Flo2)- or clathrin (Cla)-positive punctate were quantified in 20 cells (bottom panel). Inset, high magnification. (B) HepG2 cells were transfected with control (Cont), flotillin-2 or clathrin siRNA. The protein levels of flotillin-2 and clathrin were examined by immunoblotting. HSP90 was used as a loading control. (C) HepG2 cells expressing FLAG–LRP6 were transfected with control, flotillin-2 or clathrin siRNA. After the cells were incubated for 1 h at 4°C, cell surface proteins were biotinylated. Lysates were precipitated with neutravidin–agarose beads to detect basolateral cell surface FLAG–LRP6 (left panels). The signals of cell surface LRP6 were quantified and expressed as arbitrary units (right panel). AP, affinity precipitation. (D) HepG2 cells were transfected with pCS2/FLAG–LRP6 and control, flotillin-2 or clathrin siRNA, and transcytosis of FLAG–LRP6 was examined at 1 or 3 h with an antibody-trafficking assay. Localization of FLAG–LRP6 at apical membrane regions is indicated by yellow arrowheads in the left panels. Scale bars: 5 μm (A) and 20 μm (D). (E) At 1 and 3 h after internalization of FLAG–LRP6 was started in D, the distribution of FLAG–LRP6 was quantified (top panel), and the percentage of the apical membrane regions that accumulated FLAG–LRP6 was calculated (bottom panel). The results shown are means±s.e.m. from three independent experiments (B) or three different fields per coverslip (A and D). *P<0.05; n.s., not significant.
Transcytosis of LRP6 requires flotillin-2. (A) One hour after internalization of FLAG–LRP6 at the basolateral membrane was initiated, HepG2 cells were stained with anti-FLAG (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies. Co-localization of FLAG–LRP6 and flotillin-2 or clathrin is depicted in yellow in the top panels. The percentages of internalized FLAG–LRP6 co-localized with flotillin-2 (Flo2)- or clathrin (Cla)-positive punctate were quantified in 20 cells (bottom panel). Inset, high magnification. (B) HepG2 cells were transfected with control (Cont), flotillin-2 or clathrin siRNA. The protein levels of flotillin-2 and clathrin were examined by immunoblotting. HSP90 was used as a loading control. (C) HepG2 cells expressing FLAG–LRP6 were transfected with control, flotillin-2 or clathrin siRNA. After the cells were incubated for 1 h at 4°C, cell surface proteins were biotinylated. Lysates were precipitated with neutravidin–agarose beads to detect basolateral cell surface FLAG–LRP6 (left panels). The signals of cell surface LRP6 were quantified and expressed as arbitrary units (right panel). AP, affinity precipitation. (D) HepG2 cells were transfected with pCS2/FLAG–LRP6 and control, flotillin-2 or clathrin siRNA, and transcytosis of FLAG–LRP6 was examined at 1 or 3 h with an antibody-trafficking assay. Localization of FLAG–LRP6 at apical membrane regions is indicated by yellow arrowheads in the left panels. Scale bars: 5 μm (A) and 20 μm (D). (E) At 1 and 3 h after internalization of FLAG–LRP6 was started in D, the distribution of FLAG–LRP6 was quantified (top panel), and the percentage of the apical membrane regions that accumulated FLAG–LRP6 was calculated (bottom panel). The results shown are means±s.e.m. from three independent experiments (B) or three different fields per coverslip (A and D). *P<0.05; n.s., not significant.
LRP6 possesses a YxxØ (‘x’ represents any amino acid, and ‘Ø’ denotes a large hydrophobic residue) motif in the cytoplasmic C-terminal region. This motif targets surface proteins to clathrin-coated pits (Benmerah and Lamaze, 2007). A LRP6 YxxØ mutant (LRP6Y1522A) failed to form a complex formation with AP-2, a clathrin-associated adaptor protein (Kim et al., 2013). It has been demonstrated that flotillin-1 binds to the sorbin homology (SOHO) domain and the di-leucine of β-site amyloid precursor protein cleaving enzyme 1 (BACE1) (Kimura et al., 2001; John et al., 2014). However, LRP6 does not possess these motifs. To identify the apical targeting region of LRP6, FLAG–LRP6 mutants were generated (Fig. S4A). LRP6(1–1403) includes an intracellular region with two cysteine residues (C1394 and C1399) in the juxtamembranous region. These cysteines are palmitoylated and are required for exit from the endoplasmic reticulum (Abrami et al., 2008). FLAG–LRP6(1–1439) and FLAG–LRP6Y1522A formed complexes with flotillin-2, but not with clathrin. FLAG–LRP6(1–1403) did not form a complex with flotillin-2 or clathrin (Fig. S4B). Transcytosis of FLAG–LRP6(1–1439) and FLAG–LRP6 Y1522A were similar to that of FLAG–LRP6, whereas transcytosis of FLAG–LRP6(1–1403) was suppressed (Fig. S4C). These results indicate that amino acids 1404 to 1439 and the C-terminal YxxØ motif (Y1522RHF) are required for complex formation with flotillin-2 and clathrin, respectively. These results support the hypothesis that the binding of LRP6 to clathrin is not required for apical membrane localization of LRP6.
LRP6 transports NPC1L1 to the apical membrane
The biological role of LRP6 localization to the apical membrane in hepatocytes is unclear. NPC1L1, which functions in the intestinal absorption of free cholesterol, is localized to the bile canalicular membrane in the liver and contributes to the absorption of cholesterol (Altmann et al., 2004; Temel et al., 2007; Kurano et al., 2012). Endogenous NPC1L1 was indeed localized to the apical membrane of HepG2 cells (Fig. 5A). To examine the traffic pathway of NPC1L1, HA–NPC1L1 was expressed at a similar level to endogenous NPC1L1 (Fig. 5B). HA–NPC1L1 was present at the basolateral membrane when HepG2 cells were incubated with anti-HA antibody at 4°C, and approximately 70% of apical membrane regions in HA–NPC1L1-expressing cells became HA–NPC1L1 positive 3 h after internalization was started (Fig. 5C). Thus transfected NPC1L1 was also localized to the apical membrane, like endogenous NPC1L1.
NPC1L1 accumulates at the apical membrane in a flotillin-2-dependent manner. (A) HepG2 cells were stained with anti-NPC1L1 (red) and anti-ezrin (green) antibodies. Co-localization of NPC1L1 with ezrin is depicted in yellow in the left panel. (B) Lysates of HepG2 cells expressing HA–NPC1L1 were probed with anti-HA and anti-NPC1L1 antibodies. (C) Transcytosis of HA–NPC1L1 was examined for the indicated periods with an antibody-trafficking assay. HA–NPC1L1-expressing cells and apical membrane structures are indicated by green and red arrowheads, respectively. Localization of HA–NPC1L1 at the apical membrane regions is indicated by yellow arrowheads in the top panels. The percentage of the apical membrane regions that accumulated HA–NPC1L1 was calculated (right panel). (D) One hour after the start of HA–NPC1L1 internalization, cells were stained with anti-HA (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies (left panels). Co-localization of HA–NPC1L1 with flotillin-2 or clathrin is depicted in yellow in the top panels. The percentages of internalized HA–NPC1L1 co-localized with flotillin-2- or clathrin-positive vesicles were quantified in 20 cells (right panel). Inset, high magnification. (E) After HepG2 cells were transfected with pCS2/HA–NPC1L1 and control, flotillin-2 or clathrin siRNAs, transcytosis of HA–NPC1L1 was examined at 3 h (left panels). The percentage of the apical membrane regions that accumulated HA–NPC1L1 was calculated (right panel). Scale bars: 5 μm (A,D) and 20 μm (C,E). The results shown are means±s.e.m. from three different fields per coverslip. *P<0.05; n.s., not significant.
NPC1L1 accumulates at the apical membrane in a flotillin-2-dependent manner. (A) HepG2 cells were stained with anti-NPC1L1 (red) and anti-ezrin (green) antibodies. Co-localization of NPC1L1 with ezrin is depicted in yellow in the left panel. (B) Lysates of HepG2 cells expressing HA–NPC1L1 were probed with anti-HA and anti-NPC1L1 antibodies. (C) Transcytosis of HA–NPC1L1 was examined for the indicated periods with an antibody-trafficking assay. HA–NPC1L1-expressing cells and apical membrane structures are indicated by green and red arrowheads, respectively. Localization of HA–NPC1L1 at the apical membrane regions is indicated by yellow arrowheads in the top panels. The percentage of the apical membrane regions that accumulated HA–NPC1L1 was calculated (right panel). (D) One hour after the start of HA–NPC1L1 internalization, cells were stained with anti-HA (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies (left panels). Co-localization of HA–NPC1L1 with flotillin-2 or clathrin is depicted in yellow in the top panels. The percentages of internalized HA–NPC1L1 co-localized with flotillin-2- or clathrin-positive vesicles were quantified in 20 cells (right panel). Inset, high magnification. (E) After HepG2 cells were transfected with pCS2/HA–NPC1L1 and control, flotillin-2 or clathrin siRNAs, transcytosis of HA–NPC1L1 was examined at 3 h (left panels). The percentage of the apical membrane regions that accumulated HA–NPC1L1 was calculated (right panel). Scale bars: 5 μm (A,D) and 20 μm (C,E). The results shown are means±s.e.m. from three different fields per coverslip. *P<0.05; n.s., not significant.
Exogenously expressed NPC1L1 has been shown to form a complex and co-localize with flotillin-1 or flotillin-2 in CRL1601 cells, a hepatic cell line (Ge et al., 2011). One hour after the internalization assay was started, HA–NPC1L1 was internalized from the basolateral membrane, and HA–NPC1L1 was co-localized with flotillin-2 rather than clathrin (Fig. 5D). Although flotillin-2 knockdown inhibited the localization of HA–NPC1L1 to the apical membrane, clathrin knockdown did not have this effect (Fig. 5E). Taken together, these findings suggest that NPC1L1 is trafficked to the apical membrane through the flotillin-mediated transcytotic pathway.
LRP6 and NPC1L1 formed a complex at the endogenous level (Fig. 6A). Moreover, FLAG–LRP6 co-localized with HA–NPC1L1 in intracellular vesicles 1 h after FLAG–LRP6 and HA–NPC1L1 were internalized from the basolateral membrane (Fig. 6B). Both FLAG–LRP6 and HA–NPC1L1 were detected in apical membrane regions 3 h post-internalization (Fig. 6B). These results suggest that NPC1L1 is transcytosed to the apical membrane with LRP6.
LRP6 is required for apical membrane localization of NPC1L1. (A) After lysates of HepG2 cells were immunoprecipitated with anti-NPC1L1 antibody, the immunoprecipitates were probed with anti-LRP6 and anti-NPC1L1 antibodies. (B) Co-localization of FLAG–LRP6 and HA–NPC1L1 was examined at 1 and 3 h after internalization was started. Inset, high magnification. (C) HepG2 cells were transfected with indicated siRNAs. The protein levels of NPC1L1 or LRP5 and LRP6 were examined by immunoblotting. HSP90 was used as a loading control. (D) HepG2 cells were treated with control or NPC1L1 siRNAs, and transcytosis of FLAG–LRP6 at 3 h was examined with an antibody-trafficking assay. FLAG–LRP6-expressing cells and apical membrane structures are indicated by green and red arrowheads, respectively. Localization of FLAG–LRP6 at the apical membrane region is indicated by yellow arrowheads in the top panels. The percentage of apical membrane regions that accumulated FLAG–LRP6 was calculated (right panel). (E) HepG2 cells were treated with control or LRP5 and LRP6 siRNAs, and transcytosis of HA–NPC1L1 at 3 h was examined. (F) HepG2 cells were treated with control or LRP5 and/or LRP6 siRNAs and the cells were stained with Alexa-Fluor-488-conjugated phalloidin (left panels). The ratio of apical membrane regions was calculated (right panel). Scale bars: 5 μm (B) and 20 μm (D–F). The results shown are means±s.e.m. from three different fields per coverslip. *P<0.05; n.s., not significant.
LRP6 is required for apical membrane localization of NPC1L1. (A) After lysates of HepG2 cells were immunoprecipitated with anti-NPC1L1 antibody, the immunoprecipitates were probed with anti-LRP6 and anti-NPC1L1 antibodies. (B) Co-localization of FLAG–LRP6 and HA–NPC1L1 was examined at 1 and 3 h after internalization was started. Inset, high magnification. (C) HepG2 cells were transfected with indicated siRNAs. The protein levels of NPC1L1 or LRP5 and LRP6 were examined by immunoblotting. HSP90 was used as a loading control. (D) HepG2 cells were treated with control or NPC1L1 siRNAs, and transcytosis of FLAG–LRP6 at 3 h was examined with an antibody-trafficking assay. FLAG–LRP6-expressing cells and apical membrane structures are indicated by green and red arrowheads, respectively. Localization of FLAG–LRP6 at the apical membrane region is indicated by yellow arrowheads in the top panels. The percentage of apical membrane regions that accumulated FLAG–LRP6 was calculated (right panel). (E) HepG2 cells were treated with control or LRP5 and LRP6 siRNAs, and transcytosis of HA–NPC1L1 at 3 h was examined. (F) HepG2 cells were treated with control or LRP5 and/or LRP6 siRNAs and the cells were stained with Alexa-Fluor-488-conjugated phalloidin (left panels). The ratio of apical membrane regions was calculated (right panel). Scale bars: 5 μm (B) and 20 μm (D–F). The results shown are means±s.e.m. from three different fields per coverslip. *P<0.05; n.s., not significant.
Knockdown of NPC1L1 did not affect transcytosis of FLAG–LRP6 (Fig. 6C,D). LRP5 has 71% identity to LRP6 and functions similarly to LRP6 in the Wnt signalling pathway (Joiner et al., 2013), and both LRP5 and LRP6 were expressed in HepG2 cells (Fig. 6C). Knockdown of both LRP5 and LRP6 did not affect the formation of the apical membrane region but inhibited transcytosis of NPC1L1 (Fig. 6C,E,F). Therefore, LRP5 and LRP6 are required for NPC1L1 localization to the apical membrane, whereas NPC1L1 is not necessary for apical localization of LRP6.
LDL decreases the trafficking of LRP6 to the apical membrane
Consistent with previous observations in CHO and splenic B cells (Liu et al., 2008; Ye et al., 2012), LDL induced LRP6 internalization in HepG2 cells (Fig. S5A). 3,3′-Dioctadecylindocarbocyanine (Dil)-LDL had been used in the assays for the LDL binding to LDLR and LDL uptake in HepG2 cells (Stephan and Yurachek, 1993; Matsui et al., 2010). Dil-LDL-induced internalization of LRP6 showed similar patterns to unmodified LDL (Fig. S5A,B). Internalized Dil-LDL was co-localized with FLAG–LRP6 and clathrin (Fig. S5C), and LDL and Dil-LDL increased the co-localization of LRP6 with clathrin to a similar extent (Fig. S5C–E). Therefore, Dil-LDL was used in the following experiments because it showed similar actions to unmodified LDL.
The interaction of LRP6 with clathrin was increased in a Dil-LDL dose-dependent manner, whereas the interaction of LRP6 with flotillin-2 was decreased (Fig. 7A). FLAG–LRP6Y1522A did not interact with clathrin in the presence or absence of Dil-LDL (Figs S4B and S5F). Dil-LDL decreased the interaction of FLAG–LRP6Y1522A with flotillin-2 as well as wild-type FLAG–LRP6, suggesting that LDL inhibits the formation of a complex between LRP6 and flotillin-2 independently of clathrin-mediated endocytosis. One hour after stimulation with Dil-LDL, LRP6 was colocalized with clathrin more frequently than with flotillin-2 (Fig. 7B). When HepG2 cells were treated with Dil-LDL for 1 h, the percentage of LRP6 in the DRM fraction (fractions 2–4) diminished, and the percentage of LRP6 in the non-DRM fraction (fractions 8–10) increased (Fig. 7C), suggesting that LDL causes LRP6 to transit from the DRM fraction to the non-DRM fraction. In addition, Dil-LDL treatment for 1 h reduced LRP6 in the DRM fraction in the presence of MDC (Fig. 7C).
LDL suppresses transcytosis of LRP6. (A) After HepG2 cells were stimulated with the indicated concentrations of Dil-LDL for 1 h at 37°C, lysates were precipitated with anti-LRP6 antibody, and immunoprecipitates were probed with the indicated antibodies (left panels). The signals of clathrin (Cla) and flotillin-2 (Flo) complexed with LRP6 were quantified and expressed as arbitrary units (right panel). (B) HepG2 cells expressing FLAG–LRP6 were treated with Dil-LDL for 1 h, and cells were stained with anti-FLAG (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies (left panels). Co-localization of FLAG–LRP6 and flotillin-2 or clathrin is depicted in yellow in the left panels. The percentages of internalized FLAG–LRP6 co-localized with flotillin-2- or clathrin-positive vesicles were quantified in 20 cells (right panel). Inset, high magnification. (C) After HepG2 cells were treated with or without 25 μM MDC, cells were stimulated with or without 5 μg/ml Dil-LDL for 1 or 3 h. Lysates were fractionated by sucrose density gradient centrifugation, and aliquots were probed with anti-LRP6, anti-flotillin-2 and anti-clathrin antibodies (top panels). The band intensity of LRP6 in the DRM fractions was quantified, and the percentage of LRP6 in the DRM fraction was calculated (bottom panel). (D) After HepG2 cells expressing FLAG–LRP6 were treated with 5 mM MβCD or 25 μM MDC, cells were stimulated with or without 5 μg/ml Dil-LDL for 3 h (top panels). The percentage of the apical membrane regions that accumulated FLAG–LRP6 was calculated (bottom panel). Scale bars: 5 μm (B) and 20 μm (D). The results shown are means±s.e.m. from three independent experiments (A and C) or from three different fields per coverslip (B and D). *P<0.05; n.s., not significant.
LDL suppresses transcytosis of LRP6. (A) After HepG2 cells were stimulated with the indicated concentrations of Dil-LDL for 1 h at 37°C, lysates were precipitated with anti-LRP6 antibody, and immunoprecipitates were probed with the indicated antibodies (left panels). The signals of clathrin (Cla) and flotillin-2 (Flo) complexed with LRP6 were quantified and expressed as arbitrary units (right panel). (B) HepG2 cells expressing FLAG–LRP6 were treated with Dil-LDL for 1 h, and cells were stained with anti-FLAG (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies (left panels). Co-localization of FLAG–LRP6 and flotillin-2 or clathrin is depicted in yellow in the left panels. The percentages of internalized FLAG–LRP6 co-localized with flotillin-2- or clathrin-positive vesicles were quantified in 20 cells (right panel). Inset, high magnification. (C) After HepG2 cells were treated with or without 25 μM MDC, cells were stimulated with or without 5 μg/ml Dil-LDL for 1 or 3 h. Lysates were fractionated by sucrose density gradient centrifugation, and aliquots were probed with anti-LRP6, anti-flotillin-2 and anti-clathrin antibodies (top panels). The band intensity of LRP6 in the DRM fractions was quantified, and the percentage of LRP6 in the DRM fraction was calculated (bottom panel). (D) After HepG2 cells expressing FLAG–LRP6 were treated with 5 mM MβCD or 25 μM MDC, cells were stimulated with or without 5 μg/ml Dil-LDL for 3 h (top panels). The percentage of the apical membrane regions that accumulated FLAG–LRP6 was calculated (bottom panel). Scale bars: 5 μm (B) and 20 μm (D). The results shown are means±s.e.m. from three independent experiments (A and C) or from three different fields per coverslip (B and D). *P<0.05; n.s., not significant.
When HepG2 cells were treated with Dil-LDL for 3 h in the absence of MDC, LRP6 moved to the non-DRM fraction (Fig. 7C), and the localization of FLAG–LRP6 to the apical membrane was reduced (Fig. 7D). MβCD treatment further reduced the localization of FLAG–LRP6 to the apical membrane in the presence of Dil-LDL (Fig. 7D). Thus the routes of transcytosis and LDL-dependent internalization of LRP6 are mutually exclusive. In HepG2 cells treated with Dil-LDL for 3 h in the presence of MDC, after LRP6 moved to the non-DRM fraction, it was redistributed in the DRM fraction (Fig. 7C). This could occur to restore a state of LRP6 dynamic equilibrium due to the block of clathrin-mediated endocytosis by MDC. Therefore, the apical membrane localization of LRP6 is observed in the cells treated with Dil-LDL and MDC for 3 h to a similar level as control cells (Fig. 7D). Taken together, LDL may reduce the amount of LRP6 in the DRM fraction and increase that of LRP6 in the non-DRM fraction, and switch the route of LRP6 internalization from flotillin-2-dependent to clathrin-dependent in HepG2 cells.
LDL switches the internalization route of LRP6 and determines the destination of NPC1L1
It has been reported that LRP6 binds to and colocalizes with Niemann–Pick type C1 (NPC1) protein (Liu et al., 2008), which is involved in late endosomal LDL trafficking and is responsible for NPC disease (Carstea et al., 1997). Endogenous NPC1 was colocalized with LAMP1, a late endosomal marker (Fig. 8A) (Liu et al., 2008). To examine the traffic pathway of NPC1, Myc-NPC1 was expressed to a similar level as endogenous NPC1 (Fig. 8B). Dil-LDL induced the colocalization of FLAG–LRP6 with Myc-NPC1 (Fig. 8C). NPC1 knockdown inhibited Dil-LDL-dependent colocalization of FLAG–LRP6 with LAMP1 (Fig. 8D,E). Myc-NPC1 was localized to the late endosome (Higgins et al., 1999) and Dil-LDL did not affect its subcellular distribution (Fig. 8F). Knockdown of LRP5 and LRP6 did not affect the localization of Myc-NPC1 to the late endosome (Fig. 8F). Therefore, localization of LRP6 to the late endosome is dependent on NPC1, whereas NPC1 is localized to the late endosome independently of LRP5 and LRP6.
LDL determines the destination of LRP6 and NPC1L1. (A) HepG2 cells were stained with anti-NPC1 (red) and anti-LAMP1 (green) antibodies and phalloidin (white). Co-localization of NPC1 with LAMP1 is depicted in yellow in the top panel. (B) Lysates of HepG2 cells expressing Myc-NPC1 were probed with anti-Myc and anti-NPC1 antibodies. (C) HepG2 cells expressing FLAG–LRP6 and Myc-NPC1 were stimulated with or without 5 μg/ml Dil-LDL for 1 h. Subsequently, internalization of FLAG–LRP6 (green) and Myc-NPC1 (red) was examined with an antibody-trafficking assay (left panels). The percentage of FLAG–LRP6 puncta that were co-localized with Myc-NPC1-positive vesicles was quantified (right panel). Inset, high magnification. (D) HepG2 cells were transfected with control or NPC1 siRNA. The protein level of NPC1 was examined by immunoblotting. HSP90 was used as a loading control. (E) HepG2 cells were transfected with pCS2/FLAG–LRP6 and control or NPC1 siRNAs and were stimulated with 5 μg/ml Dil-LDL for 1 h. Subsequently, co-localization of FLAG–LRP6 (green) with LAMP1 (red) was examined (left panels), and the percentage of FLAG–LRP6 co-localized with LAMP1 was quantified (right panel). (F) HepG2 cells were transfected with pCS2/Myc-NPC1 and control or LRP5+LRP6 siRNAs and stimulated with or without 5 μg/ml Dil-LDL for 1 h. Co-localization of Myc-NPC1 (green) with LAMP1 (red) was then examined (left panels), and the percentage of Myc-NPC1 co-localized with LAMP1 was quantified (right panel). (G) HepG2 cells were treated with control or LDLR siRNA and the formation of a complex between NPC1L1 and the indicated proteins was examined in the presence or absence of 5 μg/ml Dil-LDL for 1 h (top panels). The signals of LRP6, clathrin, LDLR and flotillin-2 complexed with NPC1L1 were quantified and expressed as arbitrary units (bottom panel). (H) One hour after stimulation with or without Dil-LDL, HepG2 cells expressing HA–NPC1L1 were stained with anti-HA (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies (left panels). Co-localization of HA–NPC1L1 with flotillin-2 or clathrin is depicted in yellow in the top panels. The percentages of internalized HA–NPC1L1 co-localized with flotillin-2- or clathrin-positive vesicles were quantified in 20 cells (right panel). (I) After HepG2 cells expressing HA–NPC1L1 were stimulated with 5 μg/ml Dil-LDL for 3 h, transcytosis of HA–NPC1L1 was examined (left panel), and the percentage of apical membrane regions that accumulated HA–NPC1L1 was calculated (right panel). Scale bars: 5 μm (A,C,E,F,H) and 20 μm (I). The results shown are means±s.e.m. from three independent experiments (G) or from three different fields per coverslip (C, E, F, H and I). *P<0.05; n.s., not significant.
LDL determines the destination of LRP6 and NPC1L1. (A) HepG2 cells were stained with anti-NPC1 (red) and anti-LAMP1 (green) antibodies and phalloidin (white). Co-localization of NPC1 with LAMP1 is depicted in yellow in the top panel. (B) Lysates of HepG2 cells expressing Myc-NPC1 were probed with anti-Myc and anti-NPC1 antibodies. (C) HepG2 cells expressing FLAG–LRP6 and Myc-NPC1 were stimulated with or without 5 μg/ml Dil-LDL for 1 h. Subsequently, internalization of FLAG–LRP6 (green) and Myc-NPC1 (red) was examined with an antibody-trafficking assay (left panels). The percentage of FLAG–LRP6 puncta that were co-localized with Myc-NPC1-positive vesicles was quantified (right panel). Inset, high magnification. (D) HepG2 cells were transfected with control or NPC1 siRNA. The protein level of NPC1 was examined by immunoblotting. HSP90 was used as a loading control. (E) HepG2 cells were transfected with pCS2/FLAG–LRP6 and control or NPC1 siRNAs and were stimulated with 5 μg/ml Dil-LDL for 1 h. Subsequently, co-localization of FLAG–LRP6 (green) with LAMP1 (red) was examined (left panels), and the percentage of FLAG–LRP6 co-localized with LAMP1 was quantified (right panel). (F) HepG2 cells were transfected with pCS2/Myc-NPC1 and control or LRP5+LRP6 siRNAs and stimulated with or without 5 μg/ml Dil-LDL for 1 h. Co-localization of Myc-NPC1 (green) with LAMP1 (red) was then examined (left panels), and the percentage of Myc-NPC1 co-localized with LAMP1 was quantified (right panel). (G) HepG2 cells were treated with control or LDLR siRNA and the formation of a complex between NPC1L1 and the indicated proteins was examined in the presence or absence of 5 μg/ml Dil-LDL for 1 h (top panels). The signals of LRP6, clathrin, LDLR and flotillin-2 complexed with NPC1L1 were quantified and expressed as arbitrary units (bottom panel). (H) One hour after stimulation with or without Dil-LDL, HepG2 cells expressing HA–NPC1L1 were stained with anti-HA (green) and anti-flotillin-2 (red) or anti-clathrin (red) antibodies (left panels). Co-localization of HA–NPC1L1 with flotillin-2 or clathrin is depicted in yellow in the top panels. The percentages of internalized HA–NPC1L1 co-localized with flotillin-2- or clathrin-positive vesicles were quantified in 20 cells (right panel). (I) After HepG2 cells expressing HA–NPC1L1 were stimulated with 5 μg/ml Dil-LDL for 3 h, transcytosis of HA–NPC1L1 was examined (left panel), and the percentage of apical membrane regions that accumulated HA–NPC1L1 was calculated (right panel). Scale bars: 5 μm (A,C,E,F,H) and 20 μm (I). The results shown are means±s.e.m. from three independent experiments (G) or from three different fields per coverslip (C, E, F, H and I). *P<0.05; n.s., not significant.
Knockdown of LRP6 but not NPC1L1 decreased the internalization of Dil-LDL, suggesting that LRP6 but not NPC1L1 is required for LDL uptake at the basolateral membrane (Fig. S6). NPC1L1 formed a complex with LRP6 and flotillin-2 at the endogenous level and HA–NPC1L1 was colocalized with flotillin-2 (Fig. 8G,H). Dil-LDL treatment decreased the interaction of NPC1L1 with LRP6 and flotillin-2 and the co-localization of HA–NPC1L1 and flotillin-2 (Fig. 8G,H). Consistently, Dil-LDL inhibited transcytosis of HA–NPC1L1 to the apical membrane (Fig. 8I), suggesting that LDL suppresses the formation of a complex between LRP6, flotillin-2 and NPC1L1 by reducing LRP6 in the DRM fraction, resulting in the inhibition of the transport of NPC1L1 to the apical membrane region. However, the formation of a complex between HA–NPC1L1 and clathrin was unchanged in the presence or absence of Dil-LDL (Fig. 8G,H). It has been reported that LDLR forms a complex with LRP6 in the absence of LDL (Ye et al., 2012). LDLR was not observed in the HA–NPC1L1 immune complex with or without Dil-LDL treatment (Fig. 8G). Therefore, although LDLR is in the same complex as LRP6, LDLR could not be in the same complex as NPC1L1.
It has been reported that LRP6 is required for the binding of LDL to LDLR and LDL internalization (Ye et al., 2012). In control HepG2 cells Dil-LDL inhibited the interaction of LRP6 with flotillin-2 and increased that of LRP6 with clathrin (Fig. S7A). Dil-LDL suppressed the co-localization of FLAG–LRP6 with flotillin-2 and enhanced that of FLAG–LRP6 and clathrin (Fig. S7B). Knockdown of LDLR did not affect the formation of a complex between LRP6, NPC1L1 and flotillin-2 in the presence or absence of Dil-LDL (Fig. 8G). However, Dil-LDL-dependent changes in the association of LRP6 and flotillin-2 or clathrin were not observed in LDLR-depleted HepG2 cells (Fig. 8G; Fig. S7A). Therefore, although LRP6 forms a complex with NPC1L1 independently of LDLR, the switching of LRP6 from flotillin-2 to clathrin in response to LDL would require the binding of LDL to LDLR.
Finally, we examined the effects of Wnt signalling on transcytosis of LRP6. Although Wnt3a enhanced LRP6 internalization in HepG2 cells until 30 min after stimulation, the ratios of internalized LRP6 in the presence or absence of Wnt3a were almost the same at 1 h (Fig. S8A). When HepG2 cells were treated with Wnt3a for 1 h, Wnt3a signalling did not affect the interaction of LRP6, flotillin-2 and NPC1L1, but reduced the association of LRP6, clathrin and NPC1 (Fig. S8B). Consistently, Wnt3a stimulation did not affect the co-localization of LRP6 with flotillin-2 at 1 h after stimulation but diminished that of LRP6 with clathrin (Fig. S8C). This might be due to Wnt3a decreasing the amount of LRP6 in the non-DRM fraction. Treatment with Wnt3a for 3 h did not influence the localization of LRP6 to the apical membrane (Fig. S8D). Therefore, Wnt3a signalling alone is not involved in transcytosis of LRP6 but might compete with LDL for this novel LRP6 traffic pathway.
Collectively, these results suggest that LDL controls the localization of the LRP6 and NPC1L1 complex to the apical membrane by inducing trafficking of LRP6 to the late endosome.
DISCUSSION
Distinct trafficking routes of LRP6 in flotillin-2- and clathrin-dependent manners
In the absence of LDL, LRP6 is internalized via a flotillin-2-dependent pathway from the basolateral membrane and is transcytosed to the apical membrane of HepG2 cells. In the presence of LDL, LRP6 moves to a non-lipid raft microdomain where it is internalized with clathrin and trafficked to the lysosome. Our findings in the current study are similar to our previous observations that LRP6 is internalized from the cell surface membrane through distinct internalization routes triggered by Wnt3a or Dkk1 stimulation (Yamamoto et al., 2006, 2008; Sakane et al., 2010). After Wnt3a binds to LRP6 in the lipid raft microdomain, LRP6 is internalized via a caveolin-dependent pathway; this pathway is required for Wnt3a-dependent signalling. In contrast, upon binding to Dkk1, LRP6 moves to the non-lipid raft microdomain and is internalized through a clathrin-mediated route, thereby inhibiting β-catenin signalling by the removal of LRP6 from the cell surface membrane. Although we do not know what triggers the internalization of LRP6 in the absence of LDL in HepG2 cells, LRP6 is probably internalized from the cell surface membrane via different routes, depending on ligand availability.
FLAG–LRP6(1–1439) bound to flotillin-2 and localized to the apical membrane, whereas FLAG–LRP6(1–1403) was less capable of forming a complex with flotillin-2 and accumulating at the apical membrane, suggesting that LRP6(1404–1439) is necessary for LDL-independent apical localization of LRP6. Taken together with our previous observation (Yamamoto et al., 2008), LRP6 may possess the caveolin- and flotillin-binding motifs in its cytoplasmic domain. However, LRP6(1404–1439) does not possess the typical flotillin-binding motif; LRP6 may form a complex with flotillin-2 through other protein(s), or directly binds to flotillin-2 through unknown motif(s).
Several endocytic signals (e.g. di-leucine, NPxY and YxxØ) that target cell surface proteins to clathrin-coated pits have been identified (Benmerah and Lamaze, 2007; Traub, 2009). We found that the Y1522RHF motif is important for LDL-induced basolateral internalization of FLAG–LRP6, which is mediated by binding to AP-2 and clathrin. Thus LRP6 amino acids 1404–1439 and LRP6 amino acids 1440–1613 – particularly Y1522 – could be responsible for flotillin-mediated and clathrin-mediated internalization of LRP6, respectively. However, LRP6 internalization was still observed in HepG2 cells treated with MβCD and MDC, suggesting the existence of a flotillin- and clathrin-independent internalization route for LRP6.
Physiological relevance of LRP6 transcytosis in hepatocytes
The physiological basis of LRP6 transcytosis in hepatic cells is not well understood. We demonstrated that LRP6 forms a complex with NPC1L1 and is required for apical membrane localization of NPC1L1. NPC1L1 has been reported to regulate cholesterol absorption from the intestine and liver. Free cholesterol in the intestine (from dietary intake) is taken up by enterocytes via apically localized NPC1L1 (Jia et al., 2011). Hepatic NPC1L1 is localized to the canalicular membrane and inhibits cholesterol excretion by its absorption from canalicular bile (Kurano et al., 2012).
As another function, NPC1L1 co-localizes with NPC2 in pre-lysosomal compartments and promotes its degradation (Yamanashi et al., 2012). NPC2 is a secretory protein that transfers its bound cholesterol to NPC1 in the late endosome and lysosome and plays a role in cholesterol delivery to the endoplasmic reticulum and plasma membranes. In addition, NPC2 is secreted into the bile and biliary NPC2 enhances secretion of cholesterol by stimulating cholesterol efflux through adenosine triphosphate-binding cassette G8 (ABCG8) and ABCG5 (Yamanashi et al., 2011). At the transcriptional level cholesterol induces the expression of NPC2 mRNA through activators of liver X-receptor and downregulates NPC1L1 mRNA through sterol regulatory element binding protein-2 (Rigamonti et al., 2005; Alrefai et al., 2007). Thus NPC1L1 and NPC2 have opposite actions in the regulation of intracellular cholesterol levels, and the expression levels of NPC1L1 and NPC2 are tightly regulated by various steps. Therefore, LRP6 might regulate cholesterol absorption in the liver via NPC1L1 and NPC2 to maintain lipoprotein homeostasis.
In hepatocytes, LDLR is present at the basolateral–sinusoidal membrane. The process of LDLR-dependent LDL clearance starts with its internalization, mediated by clathrin-coated vesicles. LRP6 has been shown to form a complex with LDLR in CHO cells, and knockdown of LRP6 impairs LDL-dependent LDLR endocytosis (Ye et al., 2012). We showed that LDL induces trafficking of LRP6, which requires LDLR, to the late endosome in HepG2 cells through increase in the co-localization of LRP6 with NPC1 or clathrin. This process inhibited the accumulation of LRP6 and NPC1L1 to the apical membrane through the decrease in the interaction of LRP6, NPC1L1 and flotillin-2.
Our findings suggest the following hypothetical mechanism for LRP6 in cholesterol metabolism. When dietary intake is limited, the amount of free cholesterol and LDL is low. After LRP6 and NPC1L1 are transcytosed to the apical membrane via a flotillin-2-dependent manner, the NPC2 protein level may be stabilized post-translationally. However, in the condition that the intracellular cholesterol level is low the NPC2 mRNA level is decreased and the NPC1L1 mRNA level is increased. In contrast, when dietary intake supplies sufficient free cholesterol, the amount of LDL is increased. LDL consequently induces clathrin-mediated internalization of LDLR and LRP6 and inhibits transcytosis of NPC1L1 to the apical membrane. This might downregulate the NPC2 protein level. When intracellular cholesterol levels are high, the NPC2 mRNA level is increased and the NPC1L1 mRNA level is decreased. Taken together, the balance between the apical distribution of LRP6, NPC1L1 and NPC2 and their protein levels could regulate the influx and efflux of hepatic cholesterol to maintain lipoprotein homeostasis. Additional studies are necessary to further elucidate LDL-dependent and -independent endocytosis of LRP6 and to explore hypocholesterolaemic drugs as a means to control related metabolic diseases.
MATERIALS AND METHODS
Materials and reagents
The plasmid pCS2/FLAG–LRP6 was kindly provided by Christof Niehrs (Division of Molecular Embryology, DKFZ, Heidelberg, Germany) (Davidson et al., 2005). Mouse cDNAs encoding NPC1 (AK155379) and NPC1L1 (AK078947) were purchased from Danaform (Kanagawa, Japan). Standard recombinant DNA techniques were used to construct pCS2/Myc-NPC1, pCS2/HA–NPC1L1, pCS2/FLAG–LRP6(1–1439), pCS2/FLAG–LRP6(1–1403) and pCS2/FLAG–LRP6Y1522A. The anti-Myc antibody for immunoblotting was prepared from 9E10 cells. Other primary antibodies used in this study are listed in Table S1. Dil-LDL and LDL were purchased from Life Technologies (Carlsbad, CA, USA) and Meridian Life Science (Memphis, TN, USA), respectively. EZ-Link sulfo-NHS-biotin and neutravidin–agarose were purchased from Pierce Biotechnology (Rockford, IL, USA). RNA duplexes used in this study are depicted in Table S2. Transfection was performed with lipofectamine RNAiMAX (Life Technologies) and FuGene HD (Roche, Basel, Switzerland). Other materials were obtained from commercial sources.
Immunofluorescence analysis
HepG2 (hepatocellular carcinoma cell line) was provided by Dr H. Ohdan (Hiroshima University, Hiroshima, Japan) in December 2008 (no authentication was done by the authors). HepG2 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 25 mM glucose and 10% fetal bovine serum (FBS). Cells were grown on glass coverslips and were fixed for 10 min in phosphate-buffered saline (PBS) containing 4% (w/v) paraformaldehyde (PFA). Cells then were permeabilized with PBS containing 0.2% (w/v) Triton X-100 and 0.2% (w/v) bovine serum albumin (BSA) for 5 min. For analysis of co-localization of flotillin-2 with LRP6 or NPC1L1, the cells were fixed with paraformaldehyde for 1 min at 4°C and subsequently were permeabilized with methanol for 10 min at 4°C (Galmes et al., 2013). After blocking with PBS containing 0.2% (w/v) BSA for 30 min, cells were incubated for 1 h at room temperature with primary antibodies. Cells were washed with PBS three times and were incubated for 1 h with goat Alexa-Fluor-488-conjugated anti-rabbit or anti-mouse IgG and Alexa-Fluor-546-conjugated phalloidin (Life Technologies) to visualize actin.
Antibody-trafficking assay for endocytosis and transcytosis
We conducted antibody-trafficking assays to evaluate endocytosis and transcytosis of FLAG–LRP6 and its mutants, HA–NPC1L1, Myc-NPC1 and CD59, as described previously (Aït-Slimane et al., 2003). Briefly, HepG2 cells were seeded onto 18-mm glass coverslips, and pCS2/FLAG–LRP6, pCS2/HA–NPC1L1 or pCS2/Myc-NPC1 was transfected. At 72 h post-transfection, cells were incubated with ice-cold binding medium (DMEM with 20 mM HEPES-NaOH, pH 7.4) containing anti-FLAG, anti-HA or anti-CD59 antibodies (0.5 μg/ml) for 1 h at 4°C. Unbound antibodies were removed by washing with cold PBS three times, and internalization was initiated by adding warm DMEM and transferring the dishes to a heated chamber (37°C, 5% CO2) for the indicated periods. For examination of LDL-dependent internalization of FLAG–LRP6, warm DMEM plus 5 μg/ml Dil-LDL was added. Cells were fixed and permeabilized, and antibodies bound to FLAG–LRP6, HA–NPC1L1, Myc-NPC1 and CD59 were detected with goat Alexa-Fluor-488- or -546-conjugated anti-rabbit or anti-mouse IgG.
The apical membrane region was defined as phalloidin-stained round or oval structures surrounded by two or more cells. When the staining of endogenous LRP6, CD59, FLAG–LRP6 or HA–NPC1L1 was dominant at the apical membrane region (compared with any other region of the apical membrane-forming cell), localization at the apical membrane was defined as positive. To quantify localization of endogenous LRP6, CD59, FLAG–LRP6 or HA–NPC1L1 at the apical membrane, the percentage of apical membrane regions that exhibited an accumulation of endogenous LRP6, CD59, FLAG–LRP6 or HA–NPC1L1 was calculated by counting at least 100 cells that had developed an apical membrane structure between adjacent cells.
To quantify the distribution of FLAG–LRP6, localization was classified into three types with regard to the distribution and number of cytosolic puncta. The ‘membrane’ type exhibited clear localization at the cell surface, with a few puncta in the cytosol. The ‘membrane/cytosol’ type included localization of puncta at the cell surface and puncta in the cytosol. The ‘cytosol’ type lacked a cell-surface distribution and contained more than 20 puncta in the cytosol. More than 100 cells were evaluated in each experiment.
Cycloheximide treatment
After HepG2 cells were grown to 90% confluency on 35 mm plates, cells were treated for 5 h with 5 μg/ml cycloheximide. For the relief condition, cells were treated for 5 h with CHX, washed twice with CHX-free medium, and then cultured in CHX-free medium for the indicated periods.
Preparation of DRM fractions
HepG2 cells (in a 100 mm dish) were lysed in 0.7 ml of ice-cold TNE buffer (25 mM Tris-HCl, pH 7.5; 150 mM NaCl; 5 mM EDTA-NaOH, pH 8.5) containing 0.2% Triton X-100, 2 μg/ml leupeptin, 2 μg/ml aprotinin and 1 μM phenylmethylsulfonyl fluoride. Cell lysates were further homogenized using a Dounce homogenizer (40 strokes) and by subsequent passage through a 25-gauge needle (Brown and Rose, 1992). Lysates (0.65 ml) were mixed with 0.65 ml of 80% (w/v) sucrose in TNE and were overlayered with 2.6 ml of 30% sucrose in TNE. Subsequently, 1.3 ml of 5% sucrose in TNE was added. The gradients were centrifuged at 190,000 g for 18 h at 4°C in a SW55Ti rotor (Beckman Coulter, Porterville, CA, USA). Fractions (520 μl) were harvested from the top of the gradient. Aliquots were probed with the indicated antibodies.
Labelling of cell-surface proteins
HepG2 cells were incubated with 0.5 mg/ml sulfo-NHS-LC-biotin for 30 min at 4°C (Yamamoto et al., 2006). After quenching excess biotin with 50 mM NH4Cl, cells were lysed in 0.2 ml of TNE buffer containing 1% Triton X-100, 0.4% sodium deoxycholate, 2 μg/ml leupeptin, 2 μg/ml aprotinin and 1 μM phenylmethylsulfonyl fluoride. The lysates were precipitated using neutravidin–agarose beads, and the precipitates were probed with the indicated antibodies.
LDL uptake assay
After HepG2 cells were treated with control, LRP6 or NPC1L1 siRNA for 48 h, cells were incubated in DMEM with 0.1% BSA for 12 h to serum starve the cells. After the cells were rinsed with DMEM and LDL uptake was initiated by adding warm DMEM including 10 µg/ml Dil-LDL, the dishes were transferred to a heated chamber (37°C, 5% CO2) for indicated periods. The binding of Dil-LDL to cell membrane was detected by the incubation of cells for 1 min at 37°C and internalization of Dil-LDL was examined for 10 min. After cells were fixed for 10 min in PBS containing 4% PFA, cells were incubated with PBS containing 0.2% BSA for 5 min. Cells were stained with anti-integrin β1 to detect the cell membrane. Localization of Dil-LDL was classified into two types with regard to the distribution and cytosolic puncta. The ‘membrane’ type exhibited clear localization at the cell surface with a few puncta in the cytosol. The ‘membrane/cytosol’ type included localization of puncta at the cell surface and puncta in the cytosol.
Statistical analysis
Experiments were performed at least in triplicate, and results were expressed as means±standard error of the mean (s.e.m.). Statistical analysis was performed with StatView software (SAS Institute, Inc., Cary, NC, USA). Significant differences were assessed with Student's t-test. Two-tailed P-values less than 0.05 were considered statistically significant.
Acknowledgements
We thank Dr C. Niehrs for donating plasmid.
Footnotes
Author contributions
Conceptualization: H.Y., D.U., A.K.; Methodology: H.Y., D.U., A.K.; Validation: H.Y., D.U., A.K.; Formal analysis: H.Y., D.U., S.M.; Investigation: H.Y., D.U.; Data curation: H.Y., D.U., A.K.; Writing - original draft: H.Y., D.U., A.K.; Writing - review & editing: H.Y., D.U., A.K.; Visualization: H.Y., D.U., A.K.; Supervision: A.K.; Project administration: A.K.; Funding acquisition: H.Y., A.K.
Funding
This work was supported by Grants-in-Aid for Scientific Research on Innovative Areas to A.K. (2011–2015; no. 23112004), for Scientific Research (2011–2013; no. 23590333) to A.K., and for Scientific Research (2014–2016; no. 26460365) to H.Y. from the Ministry of Education, Science and Culture of Japan.
References
Competing interests
The authors declare no competing or financial interests.