LDL receptor‐related protein LRP6 senses nutrient levels and regulates Hippo signaling

Abstract

Controlled cell growth and proliferation are essential for tissue homeostasis and development. Wnt and Hippo signaling are well known as positive and negative regulators of cell proliferation, respectively. The regulation of Hippo signaling by the Wnt pathway has been shown, but how and which components of Wnt signaling are involved in the activation of Hippo signaling during nutrient starvation are unknown. Here, we report that a reduction in the level of low‐density lipoprotein receptor‐related protein 6 (LRP6) during nutrient starvation induces phosphorylation and cytoplasmic localization of YAP, inhibiting YAP‐dependent transcription. Phosphorylation of YAP via loss of LRP6 is mediated by large tumor suppressor kinases 1/2 (LATS1/2) and Merlin. We found that O‐GlcNAcylation of LRP6 was reduced, and the overall amount of LRP6 was decreased via endocytosis‐mediated lysosomal degradation during nutrient starvation. Merlin binds to LRP6; when LRP6 is less O‐GlcNAcylated, Merlin dissociates from it and becomes capable of interacting with LATS1 to induce phosphorylation of YAP. Our data suggest that LRP6 has unexpected roles as a nutrient sensor and Hippo signaling regulator.

The reduction of the Wnt co‐receptor LRP6 during nutrient starvation leads to the activation of Hippo signaling, revealing an unexpected role for LRP6 as nutrient sensor and Hippo pathway regulator.

Introduction

Hippo signaling, conserved from Drosophila to humans, is one of the major pathways regulating tissue homeostasis and organ growth (Pan, 2010). Aberrant regulation of Hippo signaling leads to various maladies such as cancer and degenerative diseases (Plouffe et al, 2015; Zanconato et al, 2016). Hippo signaling is tightly regulated by upstream factors such as contact inhibition (Zhao et al, 2007), mechanical stress (Dupont et al, 2011), or nutrient conditions (Yu et al, 2012; Mo et al, 2015). The core components of Hippo signaling include mammalian STE20‐like kinase 1/2 (MST1/2), large tumor suppressor kinases 1/2 (LATS1/2), and the transcriptional co‐factor Yes‐associated protein (YAP) and its paralogue TAZ. The activation of Hippo signaling by several upstream factors leads to sequential phosphorylation and activation of MST1/2 and LATS1/2. Activated LATS1/2 phosphorylates YAP/TAZ, leading to their cytoplasmic retention by 14‐3‐3 or their proteasomal degradation by β‐TrCP. In contrast, inactivation of Hippo signaling causes the dephosphorylation, stabilization, and nuclear translocation of YAP/TAZ. Nuclear YAP/TAZ interact with TEA domain family member (TEAD) transcription factors and promote the expression of their target genes, which are mainly involved in regulating cell proliferation and anti‐apoptosis programs (Meng et al, 2016).

Wnt/β‐catenin signaling is an evolutionarily conserved pathway that has an essential role in development and tissue homeostasis (Nusse & Clevers, 2017). If Wnts are not present, the proteins axin, adenomatous polyposis coli (APC), glycogen synthase kinase 3 (GSK3), and casein kinase 1 (CK1) form the β‐catenin destruction complex, which phosphorylates β‐catenin and primes it for degradation by β‐TrCP. When Wnts are present, they bind to a Frizzled receptor and a low‐density lipoprotein‐related receptor 5 or 6 (LRP5/6) co‐receptor, sequestering the β‐catenin destruction complex on the plasma membrane and inactivating it. This allows β‐catenin to be stabilized and to translocate into nucleus, where it induces its target gene expression by binding to T‐cell factor/lymphoid enhancer‐binding factor (TCF/LEF) transcription factors (MacDonald et al, 2009). Abnormal regulation of Wnt signaling is often associated with human diseases such as cancer and tissue degeneration (Nusse & Clevers, 2017).

LRP5 and LRP6, belonging to the LDL receptor‐related protein family and being type I transmembrane proteins, they are highly homologous, with established functions as co‐receptors initiating canonical Wnt signaling (Pinson et al, 2000; Tamai et al, 2000; Wehrli et al, 2000). LRP6 is also involved in other pathways such as mTOR and KRAS signaling (Go et al, 2014; Lemieux et al, 2015). Mutations or dysregulation of LRP6 is associated with developmental defects and diseases such as cancer (Li et al, 2004; Liu et al, 2010), osteoporosis (Li et al, 2014), and metabolic syndromes (Mani et al, 2007; Go et al, 2014).

Several cases have been reported of crosstalk between Hippo signaling and Wnt signaling (Kim & Jho, 2014). It has been reported that YAP and TAZ could be components of the β‐catenin destruction complex (Azzolin et al, 2012, 2014) and that they could control nuclear translocation of β‐catenin (Imajo et al, 2012) and regulate Wnt signaling by affecting the phosphorylation of disheveled (DVL; Varelas et al, 2010). Reciprocally, it has also been reported that Wnt signaling components such as Wnts (Azzolin et al, 2012, 20142014; Park et al, 2015), Frizzled (Park et al, 2015), the destruction complex (Azzolin et al, 2012, 20142014), APC (Cai et al, 2015), and DVL (Lee et al, 2018) could regulate the nuclear localization and activity of YAP/TAZ. Although these data strongly suggest that there is crosstalk between Wnt signaling and Hippo signaling, it is not clear whether the activation of Hippo signaling during nutrient starvation can be modulated by Wnt signaling components.

Merlin, a tumor suppressor belonging to the ERM family (Bretscher et al, 2000), is a well‐known Hippo signaling component (Hamaratoglu et al, 2006; Zhang et al, 2010). Merlin is mainly found at the plasma membrane (Bretscher et al, 2000). Merlin recruits LATS1/2 to the plasma membrane, which allows LATS1/2 to be phosphorylated by MST1/2, resulting in the activation of Hippo signaling (Yin et al, 2013). Besides functioning as a modulator of Hippo signaling, Merlin also acts as a regulator of Wnt signaling. Previous studies from our laboratory suggested that Merlin inhibits Wnt signaling by binding LRP6 and blocking its phosphorylation (Kim et al, 2016). Wnt3a induces the phosphorylation of Merlin by p21‐activated kinase (PAK) in a phosphatidylinositol‐4,5‐bisphosphate (PIP₂)‐dependent manner, and phosphorylated Merlin detaches from LRP6, allowing the phosphorylation of LRP6 and the activation of Wnt signaling. Although we showed that the detachment of Merlin from LRP6 is necessary for the activation of Wnt signaling, the role of Merlin when detached from LRP6 in the regulation of Hippo signaling has not been studied (Kim & Jho, 2016).

Here, we show that LRP6 senses the status of nutrients and regulates Hippo‐YAP signaling. Serum or nutrient starvation causes the degradation of LRP6 via the endocytosis‐mediated lysosomal degradation pathway; the loss of LRP6 leads to the phosphorylation of YAP. The knockdown of LRP6 alone even under nutrient‐rich conditions induces the phosphorylation of YAP, whereas overexpression of LRP6 blocks YAP phosphorylation under starvation conditions. YAP phosphorylation induced by loss of LRP6 is dependent on Merlin and LATS1/2. During serum starvation, Merlin detaches from LRP6 and the interaction between Merlin and LATS1/2 is increased, resulting in LATS1/2 activation and the phosphorylation of YAP. We found that O‐GlcNAcylation on LRP6 is lower during starvation, and the inhibition of O‐GlcNAcylation under nutrient‐rich conditions also reduces the amount of LRP6. Overall, we propose a model that the lower O‐GlcNAcylation of LRP6 during starvation induces a loss of LRP6, which in turn induces YAP phosphorylation in a Merlin‐ and LATS1/2‐dependent manner. The loss of LRP6 during starvation may provide a strategy for efficiently shutting down proliferation, by coupling the inhibition of Wnt signaling with the activation of Hippo signaling.

Results

The level of LRP6 is reduced in nutrient starvation

It is known that the responsiveness to Wnt ligands is reduced, while Hippo signaling is activated as cell density increases (Zhao et al, 2007; Maher et al, 2009). Previously we showed that LRP6 interacted with Merlin and that this interaction was regulated by Wnt signaling (Kim et al, 2016). Based on these findings, we hypothesized that at a high cell density, there would be less LRP6, and the Merlin not occupied by LRP6 might activate Hippo signaling. Indeed, we found less LRP6 when cells were cultured in high cell density (Fig EV1A), and the medium routinely became yellow, which indicated decreased pH and lower nutrient levels. To determine whether the lower LRP6 level was due to cell contact in high confluency or to the lack of nutrients, we replaced the medium with the fresh medium after cells reached high confluency and measured LRP6. The amount of LRP6 remained low if the medium was not replaced, but it was not reduced at high cell density when the medium was refreshed (Fig EV1B). When the medium pH was changed independently, no loss of LRP6 was observed (Fig EV1C). These data suggest that the amount of LRP6 was reduced due to the lack of nutrients. We found that LRP6 decreased markedly in cells incubated in serum‐free medium; re‐addition of serum rescued the LRP6 level (Fig 1A and B). The loss of LRP6 under serum‐starved conditions was confirmed in different cell lines (Fig 1C), and incubation in a glucose‐free medium also decreased LRP6 (Fig 1D). Despite normalization of the total sugar concentration by adding mannitol, LRP6 decreased with decreasing glucose concentration (Fig EV1D). 2‐Deoxy‐D‐glucose (2‐DG) is a glucose analogue that can mimic glucose‐free conditions, and treatment with 2‐DG also decreased LRP6 (Fig EV1E). Thus, our data suggest that LRP6 is decreased upon nutrient starvation. LRP5 and LRP6 are highly homologous proteins having 71% sequence identity, and both work as Wnt co‐receptors (Joiner et al, 2013). We found that LRP5 was also decreased under serum starvation conditions (Fig EV1F).

Figure EV1. The amount of LRP6 was reduced in nutrient starvation.

• A
  HEK293T cells were seeded to low‐density or high‐density conditions. One day after the cells were lysed and the lysates were analyzed by immunoblotting.
• B
  The loss of LRP6 was due to a lack of nutrients. HEK293T cells were seeded to low‐density or high‐density conditions. One day after, medium for the cells at high density was replaced with fresh medium for 2 h. Cells were then lysed, and the lysates were analyzed by immunoblotting.
• C
  The loss of LRP6 level was not due to lower pH. HEK293 cells were treated in pH‐modified medium for 2 h. Medium was titrated with HCl, and pH was measured by pH meter (Mettler Toledo). Cells were lysed, and the lysates were analyzed by immunoblotting.
• D
  The amount of LRP6 decreased as the glucose level was reduced. HEK293 cells were incubated for 4 h with concentrations of glucose and mannitol as indicated in the figure. Cells were lysed, and the lysates were analyzed by immunoblotting (left panel). The ratio of LRP6/β‐actin of three independent immunoblots was quantified (biological replicates, right panel). Error bars indicate standard deviation of biological triplicate measurements. *P < 0.05. Student's t‐test was used for statistical analysis.
• E
  A competitive inhibitor of glucose processing reduced the amount of LRP6. HEK293 cells were incubated in 2‐DG (25 mM)‐contained medium for 4 h. Cells were lysed, and the lysates were analyzed by immunoblotting.
• F
  Serum starvation reduced the amount of both LRP5 and LRP6. HEK293 cells were incubated in serum‐free medium for 4 h. Cell lysates were immunoblotted with antibodies as indicated in the figure.
• G
  The mRNA level of LRP6 was not reduced under serum starvation. We used a quantitative real‐time PCR assay for measuring the expression of LRP6 and ANKRD1 mRNA in HEK293 cells cultured under serum‐starved conditions for 4 h. Quantification of LRP6 and ANKRD1 mRNA was normalized to the level of β‐actin. Error bars indicate standard deviation of technical triplicate measurements. **P < 0.01 and ***P < 0.001. Student's t‐test was used for statistical analysis.
• H
  The loss of LRP6 during serum starvation was not mediated by the proteasomal degradation pathway. HEK293 cells were incubated in serum‐free medium for 4 h with DMSO or MG132 (25 μM). Cell lysates were analyzed by immunoblotting.
• I, J
  The loss of LRP6 by serum starvation was mediated by the lysosomal degradation pathway. HEK293 cells were incubated in serum‐free medium for 4 h with DMSO, pepstatin A (20 μg/ml)/E64d (10 nM) (I), or ammonium chloride (2 mM) (J). Cells were lysed, and the lysates were analyzed by immunoblotting.
• K
  The loss of LRP6 by serum starvation was not mediated via autophagy. HEK293 cells were incubated in serum‐free medium for 4 h with DMSO or wortmannin (1 μM). Cells were lysed, and the lysates were analyzed by immunoblotting.
• L
  The loss of LRP6 by serum starvation was not mediated via autophagy. HEK293 cells were transfected with GFP or ATG7 siRNA and were incubated in serum‐free medium for 4 h. Cells were lysed, and the lysates were analyzed by immunoblotting.
• M, N
  The mRNA levels of clathrin and caveolin‐1 were properly reduced by the respective siRNAs. We used quantitative real‐time PCR assay for measuring expression of Clathrin and Caveolin‐1 mRNAs. Quantification of the mRNAs was normalized to the level of β‐actin. Error bars indicate standard deviation of technical triplicate measurements. ***P < 0.001. Student's t‐test was used for statistical analysis.
• O
  Components of serum such as cytokines and hormones were dispensable for maintaining LRP6. HEK293 cells were incubated in serum‐free medium with dialyzed FBS for 4 h. Cells were lysed, and the lysates were analyzed by immunoblotting (left panel). The ratio of LRP6/β‐actin of three independent immunoblots was quantified (biological replicates, right panel). Error bars indicate standard deviation of biological triplicate measurements. N.S, non‐significant and *P < 0.05. Student's t‐test was used for statistical analysis.

Figure 1. The level of LRP6 was reduced in nutrient starvation via endocytosis‐mediated lysosomal‐dependent degradation.

• A
  LRP6 level is reduced during serum starvation. HEK293 cells were incubated in serum‐free medium for the indicated hours. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
• B
  Serum starvation reduced the amount of LRP6. HEK293 cells were incubated in serum‐free medium for 3 h. Then, cells were additionally incubated for 1 h after addition of 10% serum. Cells were lysed, and the cell lysates were analyzed by immunoblotting (left panel). The ratio of LRP6/β‐actin and p‐YAP/YAP of three independent immunoblots was quantified (biological replicates, middle and right panel, respectively). N.S.: non‐significant, **P < 0.01 and ***P < 0.001. Error bars indicate standard deviation of biological triplicate measurements. Student's t‐test was used for statistical analysis.
• C
  Loss of LRP6 in serum starvation was seen in different cell lines. Cell lines were incubated in serum‐free medium for 2 h. Cells were lysed, and the cell lysates were analyzed by immunoblotting. S.E., short exposure; L.E., long exposure.
• D
  Glucose starvation decreased LRP6. HEK293 cells were incubated in glucose‐free medium for 4 h. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
• E
  The decrease in LRP6 by serum starvation was mediated by the lysosomal degradation pathway. HEK293 cells were incubated in serum‐free medium for 4 h with DMSO or bafilomycin A1 (100 nM). Cells were lysed, and the cell lysates were analyzed by immunoblotting.
• F, G
  Loss of LRP6 by serum starvation was mediated in a clathrin or caveolin‐1-dependent manner. HEK293 cells were transfected with clathrin (F) or caveolin (G) siRNA and incubated in serum‐free medium for 4 h. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
• H
  Lipid components of serum were essential for maintaining LRP6. HEK293 cells were incubated in serum‐free medium with A2058 BSA (Sigma) or Fraction V BSA for 4 h. Cells were lysed, and the cell lysates were analyzed by immunoblotting (left panel). The ratio of LRP6/β‐actin of three independent immunoblots was quantified (biological replicates, right panel). Error bars indicate standard deviation of biological triplicate measurements. N.S.: non‐significant and *P < 0.05. Student's t‐test was used for statistical analysis.

Next, we looked at how LRP6 was decreased during nutrient starvation. We found that LRP6 mRNA was not decreased by serum starvation (Fig EV1G), suggesting that there were changes in LRP6 protein stability. Protein degradation can be categorized as proteasome‐mediated or lysosome‐mediated degradation. We blocked proteasome‐ and lysosome‐mediated degradation by using MG132 and bafilomycin A1, respectively. During serum starvation, bafilomycin A1, but not MG132, blocked the decrease in LRP6 (Fig 1E and Fig EV1H). By using additional lysosomal inhibitors such as pepstatin A/E64d or ammonium chloride (NH₄Cl), we could rescue LRP6 (Fig EV1I and J). Lysosomal degradation can be categorized into autophagy‐ or endocytosis‐mediated degradation. We found that autophagy blockade, by using wortmannin or by knocking down ATG7, did not prevent the decrease in LRP6 during serum starvation (Fig EV1K and L). In contrast, blocking endocytosis by the knockdown of clathrin or caveolin‐1 rescued the LRP6 level (Fig 1F and G); the knockdown efficiency was confirmed by real‐time PCR (Fig EV1M and N). These data suggest that the decrease in LRP6 by serum starvation occurs via the endocytosis‐mediated lysosomal degradation pathway.

Serum contains many substances including cytokines, hormones, and lipids (Yu et al, 2012). We next sought to identify which substances are essential for maintaining the LRP6 level. Using dialyzed fetal bovine serum (FBS) depleted of small molecules such as cytokines and hormones, we could efficiently rescue LRP6 under serum starvation conditions (Fig EV1O), suggesting that neither cytokines nor hormones are essential. To determine whether serum lipids are important, we used two types of BSA: A2058 BSA, which is purified by chromatography and thereby contains lipids, and Fraction V BSA, which is purified by ethanol precipitation and thus is largely devoid of lipids (Yu et al, 2012). Upon serum starvation, A2058 BSA, but not Fraction V BSA, rescued LRP6 (Fig 1H). These results suggest that serum lipids are required for maintaining LRP6.

The amount of LRP6 determines the phosphorylation, localization, and transcriptional activity of YAP

Serum deficiency promotes YAP phosphorylation and inhibits its transcriptional activity (Yu et al, 2012). We observed an inverse correlation between LRP6 level and YAP phosphorylation status during serum starvation (Fig 2A). Interestingly, the knockdown of LRP6 induced YAP phosphorylation even under nutrient‐rich conditions (Fig 2B), and overexpression of LRP6 lessened YAP phosphorylation even in serum starvation conditions (Fig 2C). Because phosphorylated YAP localizes in the cytosol, thus reducing its transcriptional activity, we examined whether LRP6 could regulate YAP localization and activity. Knockdown of LRP6 diminished YAP‐dependent reporter activity (Fig 2D), YAP target gene expression (Fig 2E), and the interaction between YAP and its transcription factor TEAD (Fig 2F). In addition, mouse embryonic fibroblasts (MEFs) from Lrp6 knockout (Lrp6 KO) mice displayed lower mRNA expression from YAP target genes (Fig EV2A). Overexpression of LRP6 during serum starvation partially rescued the YAP‐dependent reporter activity (Fig 2G) and the mRNA expression of a YAP target gene (Fig EV2B). YAP was mainly found in the nucleus under nutrient‐rich conditions and in the cytosol during serum starvation. Knockdown of LRP6 re‐located nuclear YAP to the cytoplasm (Fig 2H), and overexpression of LRP6 during serum starvation increased the nuclear localization of YAP (Fig 2I). We also tested whether LRP5 could similarly regulate YAP activity. Knockdown of either LRP5 or LRP6 reduced the YAP‐dependent reporter activity, and knockdown of both reduced the luciferase activity even further (Fig EV2C). This suggests that LRP5 and LRP6 synergistically regulate YAP activity. The level of phosphorylated YAP was increased at high cell confluence, and knockdown of LRP6 further enhanced YAP phosphorylation (Fig EV2D).

Figure 2. LRP6 regulated the phosphorylation, localization, and transcriptional activity of YAP .

1. The levels of LRP6 and phosphorylated YAP showed inverse correlation under serum starvation. HEK293 cells were incubated in serum‐free medium for the indicated times. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
2. Knockdown of LRP6 induced the phosphorylation of YAP even under nutrient‐rich condition. HEK293T cells were transfected with GFP siRNA as a control and LRP6 siRNA under nutrient‐rich conditions. The ratio of p‐YAP/total‐YAP of three independent immunoblot bands was quantified (biological replicates, right panel). Error bars indicate standard deviation of biological triplicate measurements. *P < 0.05. Student's t‐test was used for statistical analysis.
3. Overexpression of LRP6 blocked the increase in YAP phosphorylation under serum starvation conditions. HEK293T cells were transfected with empty vector (−) or LRP6‐EGFP as indicated in the figure. One day after transfection, cells were incubated in serum‐free medium for 4 h. Cells were lysed, and the cell lysates were analyzed by immunoblotting (left panel). The ratio of LRP6/β‐actin and p‐YAP/YAP of three independent immunoblots was quantified (middle and right panel, respectively). Error bars indicate standard deviation of biological triplicate measurements. *P < 0.05 and ***P < 0.001. Student's t‐test was used for statistical analysis.
4. Knockdown of LRP6 reduced YAP reporter activity. HEK293T cells were transfected with GFP siRNA as a control, LRP6 siRNA, pRL‐TK, and 8XGTIIC luciferase reporter constructs. One day after transfection, luciferase activity was measured. Error bars indicate standard deviation of biological triplicate measurements. ***P < 0.005. Student's t‐test was used for statistical analysis.
5. Knockdown of LRP6 reduced the expression of YAP target genes. Quantitative real‐time PCR assay measuring the expression of CTGF, ANKRD1, INHBA, and LRP6 mRNA in GFP and LRP6 siRNA‐transfected HEK293T cells was performed. Quantification of CTGF, ANKRD1, INHBA, and LRP6 mRNA was normalized against the level of β‐actin. Error bars indicate standard deviation of technical triplicate measurements. *P < 0.05, **P < 0.01 and ***P < 0.001. Student's t‐test was used for statistical analysis.
6. Knockdown of LRP6 reduced the interaction between YAP and TEAD. HEK293T cells were transfected with EGFP‐YAP, Myc‐TEAD4, and siRNA for LRP6. Cells were lysed, and the cell lysates were immunoprecipitated with anti‐Myc antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
7. Overexpression of LRP6 rescued the reduced YAP reporter activity under serum starvation conditions. HEK293T cells were transfected with an empty vector (−) or VSVG‐LRP6, and with pRL‐TK and 8XGTIIC luciferase reporter constructs. One day after transfection, cells were incubated in serum‐free medium for 4 h and luciferase activity was measured. Error bars indicate standard deviation of biological triplicate measurements. ***P < 0.001. Student's t‐test was used for statistical analysis.
8. Knockdown of LRP6 induced cytoplasmic localization of YAP. HEK293A cells were transfected with siRNA for GFP or LRP6, and immunofluorescence analysis was performed. Figure shows representative image of multiple areas. DAPI was used for nucleus staining. Scale bars: 10 μm. Quantification of nuclear YAP is shown in the right panel. Quantification was performed by counting cells that have nuclear‐localized YAP per image (n = 8). Error bars indicate standard deviation. ***P < 0.001. Student's t‐test was used for statistical analysis.
9. Overexpression of LRP6 induced nuclear localization of YAP under starvation conditions. HEK293A cells were transfected with LRP6‐EGFP and 1 day after transfected cells were incubated with or without serum‐containing medium for 4 h, and immunofluorescence analysis was performed. Figure shows representative image of multiple areas. DAPI was used for nucleus staining. Scale bars: 10 μm. Quantification of nuclear YAP is shown in the right panel. Quantification was performed by counting cells that have nuclear‐localized YAP per image (n = 6). Error bars indicate standard deviation. *P < 0.05. Student's t‐test was used for statistical analysis.

Figure EV2. LRP6 regulated the phosphorylation, localization, and transcriptional activity of YAP .

1. Conditional knockout of LRP6 in MEF resulted in down‐regulation of endogenous expression of YAP target genes. For conditional knockout of Lrp6, Lrp6‐floxed MEF cells were transduced with Cre‐expressing adenovirus for 48 h. GFP‐expressing adenovirus was used as a control. Quantitative real‐time PCR assay for measuring the levels of Ctgf, Cyr61, and Lrp6 mRNA in WT and Lrp6 KO MEF cells was performed. Error bars indicate standard deviation of four measurements (technical replicates). *P < 0.05 and ***P < 0.001. Student's t‐test was used for statistical analysis.
2. Overexpression of LRP6 rescued the reduced expression of YAP target gene under serum starvation. We used quantitative real‐time PCR assay for measuring expression of ANKRD1 and LRP6 mRNA in empty vector (−) or VSVG‐LRP6 transfected HEK293T cells. After transfection, serum was removed as indicated in the figure. Quantification of ANKRD1 and LRP6 mRNA was normalized to the level of β‐actin. Error bars indicate standard deviation of four measurements (technical replicates). *P < 0.05 and ***P < 0.001. Student's t‐test was used for statistical analysis.
3. Knockdown of either LRP5 or LRP6 reduced YAP reporter activity. HEK293T cells were transfected with siRNA for GFP, LRP5, and LRP6 and with pRL‐TK and 8XGTIIC luciferase reporter constructs. One day after transfection, luciferase activity was measured. Error bars indicate standard deviation of biological triplicate measurements. *P < 0.05, **P < 0.01 and ***P < 0.001. Student's t‐test was used for statistical analysis.
4. Knockdown of LRP6 enhanced the level of phosphorylated YAP. HEK293T cells were transfected with siRNA for GFP and LRP6. One day after transfection, cells were dissociated and seeded again at low‐density or high‐density condition. For the low‐density condition, cells were seeded on a 60‐mm dish to be 50% confluent; for high‐density condition, twice the number of cells were seeded on a 35‐mm dish. One day after seeding, cells were lysed and the lysates were analyzed by immunoblotting.
5. Overexpression of DVL1 enhanced Wnt‐reporter activity in LRP6 knockdown. HEK293T cells were transfected with siRNA for GFP and LRP6, empty vector (−), and Flag‐DVL1, and with pRL‐TK and 8XGTIIC luciferase reporter constructs. One day after transfection, luciferase activity was measured. Error bars indicate the standard deviation of biological triplicate measurements. ***P < 0.001. Student's t‐test was used for statistical analysis.
6. Overexpression of DVL1 did not rescue the reduced YAP reporter activity caused by LRP6 knockdown. HEK293T cells were transfected with siRNAs for GFP and LRP6, empty vector (−), and Flag‐DVL1, and with pRL‐TK and pSuper‐TOP luciferase reporter constructs. One day after transfection, luciferase activity was measured. Error bars indicate standard deviation of biological triplicate measurements. N.S: non‐significant and ***P < 0.001. Student's t‐test was used for statistical analysis.

LRP6 interacts with Wnts, thereby transducing the signal to the cytoplasm. We thus investigated whether the regulation of YAP activity was due to LRP6 itself or to changes in Wnt signaling. Disheveled (DVL) is a key cytoplasmic component of Wnt signaling. Overexpression of DVL leads its polymerization, and polymerized DVL creates binding sites for Axin that inhibit its function (Schwarz‐Romond et al, 2007). Therefore, overexpression of DVL activates Wnt signaling. We overexpressed DVL concomitant with LRP6 knockdown and, as expected, Wnt‐reporter activity increased (Fig EV2E), but the YAP‐dependent reporter activity remained low regardless of DVL overexpression (Fig EV2F). Taken together, these data suggest that LRP6 regulates the phosphorylation, localization, and transcriptional activity of YAP, and that the regulation of YAP activity is due to LRP6 itself, not to overall changes in Wnt signaling.

Regulation of YAP activity by loss of LRP6 requires Merlin and LATS

Previously we showed that LRP6 can interact with Merlin and that treatment with Wnt3a dissociated Merlin from LRP6. We then hypothesized that under starvation conditions, Merlin unbound due to loss of LRP6 might regulate Hippo signaling. In a parallel Hippo signaling model proposed by Pan's group, Merlin promoted Hippo signaling without stimulating the kinase activity of MST1/2 at the plasma membrane (Yin et al, 2013). Therefore, we examined whether regulation of YAP activity by the level of LRP6 requires both Merlin and LATS. Since LATS is a major kinase of YAP and LATS activity is controlled by its phosphorylation state, we examined whether LRP6 regulates the phosphorylation of LATS. Knockdown of LRP6 induced LATS phosphorylation (Fig 3A), and overexpression of LRP6 blocked LATS phosphorylation even under serum starvation conditions (Fig 3B). Overexpression of DVL1 concomitant with LRP6 knockdown did not diminish the phosphorylation of LATS (Fig EV3A), suggesting that an increase in LATS phosphorylation is not due to reduced Wnt signaling. Knocking down LRP6 alone induced YAP phosphorylation, diminished YAP‐dependent reporter activity, and localized YAP to the cytoplasm (Fig 2 and Fig 3C–E). However, knockdown of both LRP6 and Merlin, or of both LRP6 and LATS1/2, blocked the increase in YAP phosphorylation, the decrease in YAP‐dependent reporter activity, and the cytoplasmic localization of YAP (Fig 3C–E and Fig EV3B). Furthermore, under dual knockdown of LRP6 and Merlin, overexpression of wild type of Merlin, but not of the FERM domain‐truncated mutant form of the protein, which does not interact with LATS (Yin et al, 2013), rescued the phosphorylation of YAP (Fig 3F). These results indicate that the increase in YAP phosphorylation induced by the decrease in LRP6 requires an interaction between Merlin and LATS. Because LRP6 knockdown induced phosphorylation of LATS, we examined the involvement of MST1/2 and MAP4K (Meng et al, 2015; Zheng et al, 2015), which are the main upstream kinases of LATS. Knockdown of both MST1/2 and LRP6, or of both MAP4K4/6/7 and LRP6, blocked the phosphorylation of LATS, suggesting that MST1/2 and MAP4Ks are required for the induced LATS phosphorylation by loss of LRP6 (Fig 3G and H; Fig EV3C). Next we examined whether either the knockdown of LRP6 or serum starvation could enhance the activity of MST1/2 or MAP4Ks. Knockdown of LRP6 or serum starvation could not induce a mobility shift of MAP4K4 (Meng et al, 2018; Fig EV3D and E), suggesting that these two processes do not activate MAP4K4. Similarly, knockdown of LRP6 or serum starvation did not elevate phosphorylation of MST1, showing that these two processes do not activate MST1 (Fig EV3F and G).

Figure 3. Regulation of YAP phosphorylation and activity by LRP6 requires Merlin and LATS1/2.

1. Knockdown of LRP6 activated LATS1. HEK293T cells were transfected with siRNA for GFP and LRP6. After transfection, cells were lysed and the cell lysates were analyzed by immunoblotting.
2. Overexpression of LRP6 reduced the amount of the active form of LATS1 that was induced by serum starvation. HEK293T cells were transfected with empty vector (−) or VSVG‐LRP6. One day after transfection, cells were incubated in serum‐free medium for 4 h. After incubation, cells were lysed, and the cell lysates were analyzed by immunoblotting.
3. Merlin and LATS1/2 were necessary for the LRP6‐knockdown‐mediated YAP phosphorylation. HEK293T cells were transfected with siRNA for GFP, LRP6, Merlin, and LATS1/2. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
4. Merlin and LATS1/2 were necessary for the LRP6‐knockdown‐mediated decrease in YAP reporter activity. HEK293T cells were seeded and transfected with siRNA for GFP, LRP6, Merlin, and LATS1/2. After 1 day, these cells were transfected again with pRL‐TK and 8XGTIIC luciferase reporter constructs. One day after transfection, cells were lysed and luciferase activity was measured. Error bars indicate standard deviation of biological triplicate measurements. **P < 0.01 and N.S: non‐significant. Student's t‐test was used for statistical analysis.
5. Merlin and LATS1/2 were necessary for the LRP6 knockdown‐mediated‐cytoplasmic localization of YAP. HEK293A cells were transfected with siRNA for GFP, LRP6, Merlin, and LATS1/2, and immunofluorescence analysis was performed. The figure shows representative images of multiple areas. DAPI was used for nucleus staining. Scale bars: 10 μm. Quantification of cells that have nuclear YAP is shown in Fig EV3B.
6. YAP phosphorylation by loss of LRP6 was dependent on the interaction between Merlin and LATS1/2. HEK293T cells were transfected with siRNA for GFP, LRP6, and Merlin, and after 1 day were transfected again with HA‐Merlin and HA‐MerlinΔFERM. Cells were lysed, and the cell lysates were analyzed by immunoblotting. The arrow indicates endogenous Merlin and HA‐Merlin, and the arrowhead indicates HA‐MerlinΔFERM.
7. MST1/2 were necessary for the LRP6‐knockdown‐mediated activation of LATS1/2. HEK293T cells were transfected with siRNA for GFP, LRP6, and MST1/2. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
8. MAP4K4/6/7 were necessary for the LRP6 knockdown‐mediated activation of LATS1/2. HEK293T cells were transfected with siRNA for GFP, LRP6, and MAP4K4/6/7. Cells were lysed, and the cell lysates were analyzed by immunoblotting.

Figure EV3. Regulation of YAP phosphorylation and activity by LRP6 requires Merlin and LATS1/2.

1. Overexpression of DVL1 did not inhibit LRP6‐knockdown‐mediated activation of LATS1. HEK293T cells were transfected with siRNA for GFP and LRP6, empty vector (−), and Flag‐DVL1. Cells were lysed, and the lysates were analyzed by immunoblotting.
2. A quantification was performed by counting cells that have nuclear‐localized YAP per image (n = 8). Error bars indicate standard deviation. N.S: non‐significant and ***P < 0.001. Student's t‐test was used for statistical analysis.
3. The mRNA levels of MAP4K4/6/7 were properly reduced by siRNAs. We used quantitative real‐time PCR assay for measuring expression of MAP4K4, MAP4K6, and MAP4K7 mRNA in GFP and MAP4K4/6/7 siRNA‐transfected HEK293T cells. The quantification was normalized to the level of β‐actin. Error bars indicate standard deviation of technical triplicate measurements. *P < 0.05 and ***P < 0.005. Student's t‐test was used for statistical analysis.
4. Knockdown of LRP6 did not induce a mobility shift of MAP4K4. HEK293 cells were transfected with siRNA for GFP or LRP6. Cells were lysed and the lysates were analyzed by immunoblotting.
5. Serum starvation did not induce a mobility shift of MAP4K4. HEK293 cells were incubated in serum‐free medium for 4 h. Cells were lysed, and the lysates were analyzed by immunoblotting.
6. Knockdown of LRP6 did not induce phosphorylation of MST1. HEK293 cells were transfected with EGFP‐MST1 and siRNA for GFP or LRP6. Cells were lysed, and the lysates were analyzed by immunoblotting (left panel). The ratio of p‐MST1/MST1 of three independent immunoblots was quantified (biological replicates, right panel). N.S: non‐significant. Error bars indicate standard deviation of biological triplicate measurements. Student's t‐test was used for statistical analysis.
7. Serum starvation did not induce phosphorylation of MST1. HEK293 cells were transfected with EGFP‐MST1 and incubated in serum‐free medium for 4 h. Cells were lysed, and the lysates were analyzed by immunoblotting (left panel). The ratio of p‐MST1/MST1 of three independent immunoblots was quantified (biological replicates, right panel). N.S: non‐significant. Error bars indicate the standard deviation of biological triplicate measurements. Student's t‐test was used for statistical analysis.

To activate Hippo signaling, Merlin switches binding partners from LRP6 to LATS

Since Merlin interacts with LRP6 and since the interaction between Merlin and LATS is a crucial step for LATS activation (Yin et al, 2013), we hypothesized that Merlin changes its binding partner from LRP6 to LATS under nutrient starvation conditions, which would activate Hippo signaling. Using serum starvation or LRP6 knockdown, we studied the interaction between LRP6 and Merlin and between LATS and Merlin. Serum starvation significantly diminished the interaction between overexpressed (Fig 4A) or endogenous LRP6 and Merlin (Fig 4B and C), but the interaction between LATS1 and Merlin was increased (Fig 4D). Knocking down LRP6 strengthened the interaction between overexpressed LATS1 and Merlin (Fig 4E) and between endogenous LATS1 and Merlin (Fig 4F). These data suggest that Merlin switches its binding partner from LRP6 to LATS during starvation (Fig 4G).

Figure 4. Merlin changed binding partners from LRP6 to LATS under serum starvation.

• A
  The interaction between overexpressed Merlin and LRP6 was reduced under serum starvation condition. HEK293T cells were transfected with Flag‐Merlin and LRP6‐EGFP. After transfection, cells were incubated in serum‐free medium for overnight with bafilomycin A1 (20 nM). The cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting with the indicated antibodies. WCL, whole‐cell lysates.
• B, C
  The interaction between endogenous Merlin and LRP6 was reduced under serum starvation. HEK293T cells were incubated in serum‐free medium for 4 h with bafilomycin A1 (100 nM). The cell lysates were immunoprecipitated with anti‐LRP6 (B) or anti‐Merlin (C) antibody and analyzed by immunoblotting with the indicated antibodies. WCL, whole‐cell lysates.
• D
  The interaction between overexpressed Merlin and LATS1 was increased under serum starvation. HEK293T cells were transfected with Myc‐LATS1 and EGFP‐Merlin. After transfection, cells were incubated in serum‐free medium for overnight and then were lysed. The cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
• E
  Knockdown of LRP6 enhanced the interaction between overexpressed Merlin and LATS1. HEK293T cells were transfected with Myc‐LATS1, EGFP‐Merlin, and LRP6 siRNA. Cells were lysed, and the cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
• F
  Knockdown of LRP6 enhances the interaction between endogenous Merlin and LATS1. HEK293T cells were transfected with siRNA for GFP or LRP6. After transfection, cells were lysed and the cell lysates were immunoprecipitated with anti‐Merlin antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
• G
  Schematic summary of Fig 4 experimental results.

O‐GlcNAcylation on LRP6 is reduced during nutrient starvation, and modulation of global O‐GlcNAcylation affects the interaction between Merlin and LATS

Our findings show that LRP6 stability is heavily influenced by the nutrient status, but how LRP6 senses and responds to that status remains to be answered. We speculated that metabolic changes due to nutrient starvation alter the post‐translational modification (PTM) of LRP6, which may cause conformational changes of LRP6 and release Merlin. Subsequently, after dissociation from Merlin, LRP6 could be downregulated by lysosomal degradation. We focused on protein O‐GlcNAcylation, the attachment of O‐linked β‐N‐acetylglucosamine (O‐GlcNAc) to Ser/Thr residues of target proteins, a process mediated by O‐GlcNAc transferase (OGT; Slawson & Hart, 2011). This process is reversed by O‐GlcNAcase (OGA), which detaches O‐GlcNAc from target proteins. Because UDP‐GlcNAc, the donor sugar, is generated through the hexoamine biosynthesis (HBP) pathway using glucose as building blocks (Yang & Qian, 2017; Cork et al, 2018; Akella et al, 2019), and because membrane localization of OGT is influenced by serum (Yang et al, 2008), we asked whether nutrient starvation could affect LRP6 O‐GlcNAcylation. As shown in Figs 5A and EV4A, endogenous or overexpressed LRP6 and OGT interacted with each other. We next examined the O‐GlcNAcylated status of LRP6. The inhibition of OGA by Thiamet G or the overexpression of OGT induced LRP6 O‐GlcNAcylation (Figs 5B and EV4B). We also checked LRP6 O‐GlcNAcylation by using succinylated wheat germ agglutinin beads (sWGA) that can pull down GlcNAcylated proteins. Immunoblotting the precipitates with the anti‐LRP6 antibody showed a positive signal, which was lessened by incubating with free GlcNAc acting as a competitor, further confirming that LRP6 is O‐GlcNAclyated (Fig. 5C).

Figure 5. O‐GlcNAcylation of LRP6 was regulated by nutrient status and modulation of global O‐GlcNAcylation affected the interaction between Merlin and LATS .

1. Endogenous LRP6 and OGT interact with each other. HEK293T cells were lysed, and the cell lysates were immunoprecipitated with anti‐LRP6 antibody and blotted with OGT. WCL, whole‐cell lysates.
2. LRP6 is O‐GlcNAcylated. LRP6‐EGFP-transfected HEK293T cells were treated in Thiamet G (30 μM) or OSMI‐1 (50 μM) for 6 h with bafilomycin A1 (100 nM). After treatment, cells were lysed and the cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
3. LRP6 is O‐GlcNAcylated. Cell lysates from HEK293T cells were incubated with sWGA beads in the absence or presence of free GlcNAc (25 mM) for 3 h. After incubation, sWGA beads were precipitated and analyzed by immunoblotting. WCL, whole‐cell lysates.
4. Blocking of O‐GlcNAcylation reduced the amount of LRP6. HEK293 cells were treated with OSMI‐1 (50 μM) and bafilomycin A1 (100 nM) for 6 h. Cells were lysed, and the cell lysates were analyzed by immunoblotting.
5. The O‐GlcNAcylation on LRP6 was reduced under serum starvation condition. HEK293T cells were transfected with LRP6‐EGFP. Cells were incubated in serum‐free medium with bafilomycin A1 (100 nM) for 4 h. After incubation, cells were lysed and the cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates. The ratio of O‐GlcNAc/EGFP from three independent immunoblots was quantified (biological replicates, bottom panel). *P < 0.05. Error bars indicate standard deviation of biological triplicate measurements. Student's t‐test was used for statistical analysis.
6. Increase in O‐GlcNAcylation enhanced the interaction between ectopically expressed LRP6 and Merlin. LRP6‐EGFP and Flag‐Merlin transfected HEK293T cells were incubated in serum‐free medium with bafilomycin A1 (20 nM) and Thiamet G (30 μM) overnight. Cells were lysed, and the cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
7. Enhancement of global O‐GlcNAcylation blocked the serum starvation‐mediated increase of interaction between overexpressed LATS1 and Merlin. HEK293T cells were transfected with Myc‐LATS1 and Flag‐Merlin. After transfection, cells were incubated in serum‐free medium with Thiamet G (30 μM) and bafilomycin A1 (20 nM) overnight. Cells were lysed, and the cell lysates were immunoprecipitated with anti‐Myc antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.

Figure EV4. LRP6 is O‐GlcNAcylated, and the level of LRP6 O‐GlcNAcylation is reduced in nutrient starvation.

1. LRP6 binds to OGT. HEK293T cells were transfected with LRP6‐EGFP, Flag‐OGT, and Flag‐Merlin. Cells were lysed, and the lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates. The interaction between LRP6‐EGFP and Flag‐Merlin was used as a positive control.
2. LRP6 is O‐GlcNAcylated. HEK293T cells were transfected with VSVG‐LRP6, Flag‐OGT, and Flag‐OGA. Cells were lysed, and the lysates were immunoprecipitated with anti‐VSVG antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
3. O‐GlcNAcylation of LRP6 was detected after the removal of N‐glycosylation. HEK293T cells were transfected with LRP6‐EGFP. Cells were lysed, and the lysates were incubated with PNGase F. The cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. Upper and lower arrowheads indicate glycosylated and un‐glycosylated LRP6, respectively.
4. Inhibition of global O‐GlcNAcylation leads to less LRP6. HEK293 cells were treated with OSMI‐1 (50 μM) for indicated times. Cells were lysed, and the lysates were analyzed by immunoblotting.
5. The decrease in LRP6 by OSMI‐1 treatment is mediated by the lysosomal degradation pathway. HEK293 cells were treated with OSMI‐1 (50 μM), bafilomycin A1 (100 nM), and MG132 (25 μM) for 6 h. Cells were lysed, and the lysates were analyzed by immunoblotting.
6. Membrane OGT was reduced during serum starvation. HEK293 cells were incubated in serum‐free medium for the indicated times. Cells were lysed, and membrane fractionation was performed. Proteins were analyzed by immunoblotting.
7. The interaction between overexpressed LRP6 and OGT was decreased under serum starvation conditions. HEK293T cells were transfected with LRP6‐EGFP and Flag‐OGT. Cells were incubated in serum‐free medium with bafilomycin A1 (20 nM) overnight. Cells were lysed, and the lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
8. The O‐GlcNAcylation of LRP6 was reduced under glucose starvation. HEK293T cells were transfected with LRP6‐EGFP. Cells were incubated in glucose‐free medium with bafilomycin A1 (100 nM) for 4 h. After incubation, cells were lysed and the lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
9. The phosphorylation of LRP6 at S1490 site was unchanged by serum starvation. HEK293T cells were transfected with LRP6‐EGFP. Cells were incubated in serum‐free medium with bafilomycin A1 (100 nM) for 4 h. After incubation, cells were lysed and the cell lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.
10. The serine phosphorylation of LRP6 is reduced by serum starvation. HEK293T cells were transfected with LRP6‐EGFP. Cells were incubated in serum‐free medium with bafilomycin A1 (100 nM) for 4 h. After incubation, cells were lysed and the lysates were immunoprecipitated with anti‐EGFP antibody and analyzed by immunoblotting. WCL, whole‐cell lysates.

It is known that besides O‐glycosylation sites, LRP6 also contains several N‐glycosylation sites (Khan et al, 2007; Matoba et al, 2017) and that O‐GlcNAc antibody used in this study can also detect N‐glycosylation (Isono, 2011; Tashima & Stanley, 2014). To prove that the O‐GlcNAc band of LRP6 is truly O‐GlcNAcylation as well as N‐glycosylation, PNGase F was used to remove N‐glycosylation from glycoproteins. After the PNGase F treatment, the mobility of LRP6‐EGFP was down‐shifted, but the O‐GlcNAc band was still detected (Fig EV4C), suggesting that the LRP6 is O‐GlcNAcylated as well as N‐glycosylated. To check whether deficiency of LRP6 O‐GlcNAcylation affected its stability, we inhibited OGT by using OSMI‐1. Interestingly, treatment with OSMI‐1 reduced LRP6 O‐GlcNAcylation (Fig 5B) and the amount of LRP6 (Fig EV4D). The loss of LRP6 was rescued by bafilomycin A1 but not by MG132 (Figs 5D and EV4E), suggesting that upon removal of LRP6 O‐GlcNAcylation, the protein is degraded through the lysosomal degradation pathway. Since the localization of OGT to the plasma membrane is enhanced in the presence of serum (Yang et al, 2008), we tested whether serum starvation could reduce O‐GlcNAcylation of LRP6. We confirmed that serum starvation reduced the membrane localization of OGT (Fig EV4F), diminished the interaction between LRP6 and OGT (Fig EV4G), and decreased the amount of O‐GlcNAcylation on LRP6 (Fig 5E). Further, as with serum starvation, glucose starvation lessened LRP6 O‐GlcNAcylation (Fig EV4H). Since O‐GlcNAcylation and phosphorylation may occur in a reciprocal fashion as, e.g., as in YAP O‐GlcNAcylation (Peng et al, 2017; Zhang et al, 2017), we examined this possibility regarding LRP6. During serum starvation, phosphorylation of the LRP6 S1490 site was unchanged, while the lessening of LRP6 O‐GlcNAcylation was still observed (Fig EV4I). Interestingly, the overall phosphorylation of serine sites on LRP6, as detected by a p‐Ser‐specific antibody, was less during serum starvation (Fig EV4J), suggesting that LRP6 O‐GlcNAcylation and serine phosphorylation are not regulated reciprocally during serum starvation.

There is less interaction between Merlin and LRP6 during starvation (Fig 4A–C), and we tested whether the reduction was due to less O‐GlcNAcylation on LRP6. An increase in global O‐GlcNAcylation by Thiamet G enhanced the interaction between LRP6 and Merlin even under starvation conditions (Fig 5F). We also found that the loss of O‐GlcNAcylation on LRP6 affected the Merlin–LATS interaction, being significantly enhanced by serum starvation, but being reduced when O‐GlcNAcylation was increased by Thiamet G (Fig 5G). These data suggest that reduction in global O‐GlcNAcylation weakens the interaction between Merlin and LRP6 and enhances the interaction between Merlin and LATS1 under serum starvation conditions.

Nutrient starvation reduces the level of LRP6 and its O‐GlcNAcylation in vivo

The liver is an essential organ that strictly controls metabolism by sensing nutrients (Rui, 2014). Under nutrient stress conditions, Hippo signaling is activated in liver tissue, and YAP/TAZ activity is inhibited (Wang et al, 2015). Having shown that LRP6 and its O‐GlcNAcylation are reduced during serum starvation in vitro, we asked if this were true in vivo, using liver tissues of both fed and starved mice. Liver tissues of starved mice showed lower levels of YAP/TAZ, and refeeding rescued the YAP/TAZ level. Similarly, the amount of LRP6 was reduced during starvation and was rescued by refeeding (Fig 6A and B). However, YAP/TAZ and LRP6 levels were not reduced in brain tissue, where they might be preferentially spared from the initial starvation stress (Fig 6C; McCue, 2010). A positive signal from immunoblotting with the anti‐LRP6 antibody in precipitates with sWGA beads suggested that LRP6 is O‐GlcNAcylated in mouse liver tissues (Fig 6D). The addition of free GlcNAc during the incubation with sWGA resulted in a reduced LRP6 signal in the immunoblot, further confirming that LRP6 is O‐GlcNAcylated (Fig 6D). Finally, the O‐GlcNAcylation of LRP6 was reduced even when twice as much lysate from the livers of starved mice was loaded on the gel (in order to equalize the total LRP6 between the starved and non‐starved samples; Fig 6E). This suggests that the level of O‐GlcNAcylation, as well as the total level of LRP6, was reduced during starvation. Cumulatively, these data suggest that the level of LRP6 and its O‐GlcNAcylation change with changes in the nutrients status in vitro and in physiological settings.

Figure 6. Nutrient starvation reduced the level of LRP6 and its O‐GlcNAcylation in vivo .

1. Nutrient starvation reduced the level of LRP6 in mouse liver. Mice were starved for 36 h, and liver tissues were analyzed by immunoblotting (left panel). The ratio of LRP6/Vinculin of three independent immunoblots was quantified (biological replicates, right panel). Error bars indicate standard deviation of biological triplicate measurements. **P < 0.01. Student's t‐test was used for statistical analysis.
2. Refeeding of nutrients rescued LRP6 in mouse liver. Mice were starved and then refed for 6 h, and liver tissues were analyzed by immunoblotting (left panel). The ratio of LRP6/Vinculin of three independent immunoblots was quantified (biological replicates, right panel). Error bars indicate standard deviation of biological triplicate measurements. *P < 0.05. Student's t‐test was used for statistical analysis.
3. The level of LRP6 was not reduced during nutrient starvation in mouse brain. Mice were starved, and brain tissues were analyzed by immunoblotting.
4. LRP6 was O‐GlcNAcylated in mouse liver. Liver tissues were incubated with sWGA beads in the absence or presence of free GlcNAc (25 mM) for 3 h. After incubation, sWGA beads were precipitated and analyzed by immunoblotting. WCL, whole‐cell lysates.
5. O‐GlcNAcylation of LRP6 was reduced during starvation in mouse liver. LRP6 in liver tissue lysates from both fed and starved mice was equilibrated by loading twice as much lysate from the livers of starved mice on the gel. Liver tissues were incubated in sWGA beads for 3 h. After incubation, sWGA beads were precipitated and analyzed by immunoblotting. WCL, whole‐cell lysates.

Discussion

Hippo signaling is known as a crucial controller of organ size and tissue homeostasis (Huang et al, 2005), but it is still unclear how extracellular factors and membrane receptors coordinate to regulate the Hippo pathway. In this study, we report that the Wnt co‐receptor LRP6 regulates Hippo signaling by sensing nutrients (Fig EV5), which extends the role of LRP6 beyond merely a Wnt signaling component.

Figure EV5. Graphical summary of LRP6‐YAP signaling pathway.

Under nutrient‐rich conditions, LRP6 is O‐GlcNAcylated and it maintains its interaction with Merlin. In this situation, YAP is activated and its target genes are expressed. In nutrient starvation, the O‐GlcNAcylation of LRP6 is lower and Merlin is detached from LRP6, followed by lysosomal degradation of LRP6. Merlin then binds to LATS1/2, allowing LATS1/2 to be activated by upstream kinases such as MST1/2 and MAP4K4/6/7. Activated LATS1/2 phosphorylates and inactivates YAP, thereby decreasing the expression of YAP target genes.

Several studies have shown the capacity of LRP6 to regulate Hippo signaling. Azzolin et al reported that overexpression of LRP6 activated YAP activity. They showed that the activation of Wnt signaling by ectopic expression of LRP6 induced nuclear localization of YAP released from the β‐catenin destruction complex (Azzolin et al, 2014), but they did not study whether loss of LRP6 could induce phosphorylation of YAP under regulation by Merlin and LATS1/2. Park et al (2015) showed that Wnt5a/b and Wnt3a inhibited the phosphorylation of YAP/TAZ in a β‐catenin‐ and LRP6‐independent manner, but they did not test whether changes in the LRP6 level alone could regulate that phosphorylation. More recently, it was shown that a high level of Dickkopf‐3 (DKK3) increased the level of LRP6 and YAP activity in cancer‐associated fibroblasts (CAFs; Ferrari et al, 2019). In the current study, we provide a novel regulatory mechanism of Hippo signaling by LRP6. More importantly, we show that O‐GlcNAcylation of LRP6 measures the status of nutrients and regulates Hippo signaling.

We observed that nutrient starvation reduced the level of LRP6 O‐GlcNAcylation (Fig 5G), leading to lysosomal degradation of LRP6. How the deficiency of nutrients reduces LRP6 O‐GlcNAcylation and LRP6 protein level is still unclear. We found that global O‐GlcNAcylation was not reduced during 4 h of serum starvation (Fig 5E; Fig EV4I). We know that OGT acts as a target gene of YAP and that expression of OGT is significantly reduced under YAP‐ or TEAD‐deficient conditions (Peng et al, 2017). In Fig 5G, serum starvation was induced overnight (16 h) to observe significant interaction changes between Myc‐LATS1 and Flag‐Merlin. We think that the level of global O‐GlcNAcylation is decreased in lane 4 of Fig 5G because expression of OGT may be reduced with the decrease in the YAP activity due to the long‐term serum starvation. Therefore, we believe that LRP6 O‐GlcNAcylation is reduced through localization of OGT during short‐term serum starvation, while in long‐term serum starvation, both localization of OGT and reduction in global O‐GlcNAcylation play a role. Reduced O‐GlcNAcylation on LRP6 may cause conformational changes and the release of Merlin. Future studies are needed to determine how the reduction in LRP6 O‐GlcNAcylation occurs; to find specific sites of LRP6 O‐GlcNAcylation; and to determine the importance of these sites in the regulation of LRP6 stability and its interaction with Merlin.

Several studies have reported that dysregulation of LRP6 expression or mutation of LRP6 leads to various human disease such as cancer (Li et al, 2004; Liu et al, 2010), neurodegeneration (Abe et al, 2013; Liu et al, 2014), osteoporosis (Li et al, 2014), and metabolic diseases (Mani et al, 2007; Go et al, 2014). Such diseases due to the dysregulation of LRP6 so far have been interpreted as misregulation of Wnt/β‐catenin signaling. However, our data strongly suggest that these diseases should be reexamined in the view of the induced misregulation of Hippo signaling. In cancer, the expression of LRP6 is upregulated, producing hyperactivation of Wnt signaling. Upregulated protein O‐GlcNAcylation and hyperactivated YAP are also usually found in cancer (Peng et al, 2017; Zhang et al, 2017). Therefore, we expect that upregulated LRP6 O‐GlcNAcylation, resulting in elevation of YAP activity and Wnt signaling, will also be found in cancer. Identifying small molecules that block LRP6 O‐GlcNAcylation may be a good strategy for cancer therapy.

Materials and Methods

Plasmids, siRNAs, and reagents

LRP6‐pCS2‐VSVG was generously provided by Dr. Xi He (Harvard Medical School, USA). YAP sequence was amplified using cDNA of HEK293T cells and inserted into pEGFP‐C1 (Clontech) vector. pRK5‐Myc‐TEAD4 was generously provided by Dr. Kun‐Liang Guan (University of California San Diego, USA). pCS2‐LRP6‐EGFP was generously provided by Dr. Akira Kikuchi (Osaka University, Japan). HA‐Merlin, HA‐MerlinΔFERM, pFlag‐Merlin, and pEGFP‐Merlin were constructed by PCR‐based cloning methods. MST1 coding sequence was amplified by PCR and cloned into pEGFP‐C1. pcDNA3.1‐Myc‐LATS1 was generously provided by Dr. Erich Nigg (University of Basel, Switzerland). pFlag‐OGT and pFlag‐OGA were generously provided by Dr. Jin Won Cho (Yonsei University, Korea). 8xGTIIC‐Luciferase and pSuper‐TOP were generously provided by Dr. Dae‐Sik Lim (KAIST, Korea) and Randall T. Moon (University of Washington, USA), respectively. siRNAs were purchased from Genepharma and ST Pharm and the list of siRNAs is in Appendix Table S1. The sources of reagents are in Appendix Table S2.

Cell culture, generation of Lrp6 KO MEFs, and transfection

HEK293, HEK293T, L929, HeLa, and NIH3T3 cells were cultured in Dulbecco's modified Eagle medium (DMEM, HyClone) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1X antibiotic‐antimycotic (Gibco) at 37°C in a humidified 5% CO₂ incubator. HEK293 and HEK293T cell lines were tested for Mycoplasma contamination using BM‐Cyclin (Roche) and used within 15 passages after thawing. The Lrp6‐floxed mice were originally generated by Dr. B. O. Williams’ lab. Lrp6‐flox/flox MEFs were cultured in high‐glucose DMEM containing 15% FBS. Cells were plated 16 h before they were transduced with Cre‐expressing adenovirus (MOI = 50; GFP‐expressing adenovirus was used as a control). They were then passaged to 6‐well plates for treatments 48 h after transduction. Transfections were performed using a calcium phosphate precipitation method, Lipofectamine 2000, or Lipofectamine RNAiMAX (Invitrogen). Cells were transfected for 24–48 h prior to collection of samples for analysis.

Immunoblotting and immunoprecipitation

Cells were scrapped and centrifuged at 14,200 g for 5 min in phosphate‐buffered saline (PBS) at 4°C. After centrifugation, pellets were placed in a lysis buffer (20 mM Tris (pH 7.4), 150 mM NaCl, 1% Triton X‐100, 1 mM EDTA, 1 mM EGTA, 2.5 mM sodium pyrophosphate, 1 mM β‐glycerophosphate, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and 1 μg/ml of leupeptin) for 30 min at 4°C. Lysates were centrifuged at 14,200 g for 10 min at 4°C, and the supernatant was collected and used for immunoblotting and immunoprecipitation. For immunoblotting, each lysate was weighted equally and 4× SDS loading dye (200 mM Tris (pH 6.8), 8% SDS, 0.05% bromophenol blue, 40% glycerol, and 200 mM β‐mercaptoethanol) was added and boiled for 10 min. Samples were separated by SDS–PAGE and transferred to PVDF membrane. Membranes were incubated with primary antibodies in 5% skim milk or bovine serum albumin overnight at 4°C. After incubation, membranes were washed five times in 1× TBST buffer for 10 min per each wash. Membranes were then incubated with secondary antibodies in 5% skim milk for 1 h at room temperature and washed three times for 15 min each in 1× TBST buffer. ECL (ELPIS or Millipore) was sprayed on the membranes, and signals were detected by X‐ray film or MicroChemi 4.2 (DNR Bio‐Imaging Systems). For immunoprecipitation, cell lysates were incubated with antibodies overnight in a 4°C rotator. After incubation, Protein A/G plus agarose beads (Fast Flow, 50% slurry, Millipore) was treated and incubated for 1 h in 4°C rotator. Washing with lysis buffer was performed five times, 10 min each in a 4°C rotator, and then, beads were boiled with 4× SDS loading dye for 10 min. The antibodies used in the immunoblotting and immunoprecipitation experiment are listed in Appendix Table S3.

RNA extraction, cDNA synthesis, and quantitative real‐time PCR

The medium was removed, and the cells were treated with 1 ml of TRI Reagent (MRC) for 5 min on rocker at 25°C. After treatment, solutions were collected in tubes and 0.2 ml of chloroform was added and mixed vigorously. Solutions were incubated for 3 min at 25°C and centrifuged at 12,000 g for 15 min at 4°C. After centrifuging, 0.2–0.3 ml of the upper aqueous solution was separated and 0.5 ml of isopropyl alcohol was added and mixed. Solutions were incubated for 10 min at 25°C and centrifuged at 12,000 g for 10 min at 4°C. Supernatants were removed and RNA pellets were washed with 75% ethanol and centrifuged at 7,500 g for 5 min at 4°C. RNA pellets were dried and dissolved in DEPC by pipetting, and incubated for 10 min at 60°C. Dissolved RNA was synthesized to cDNA using ReverTra Ace^® qPCR RT Kit (Toyobo) according to the manufacturer's instructions. Quantitative real‐time PCR was performed using THUNDERBIRD^® SYBR^® qPCR Mix (Toyobo) according to manufacturer's instruction. The threshold cycle (Ct) value for β‐ACTIN or Gapdh was used for normalizing the Ct value for each gene. The ΔΔCt method was used for calculating relative mRNA expression. The primer sequences for the human cell lines and mouse cells used in the experiment are in Appendix Table S4.

Luciferase Reporter Assay

HEK293T cells were seeded into 12‐well plates and transfected with 100 ng of reporter plasmid (8xGTIIC‐Luciferase for a YAP reporter and pSuper‐TOP for a canonical Wnt reporter), and 10 ng of thymidine kinase promoter‐Renilla luciferase reporter plasmid (pRL‐TK) for a control, and the plasmids or siRNAs indicated in the figures. After transfection, cells were placed in passive lysis buffer (Promega) for 15 min at RT. Lysates were centrifuged at 7,500 g for 5 min at 4°C, and 10 μl of supernatants was used for the assay. Luciferase activity was measured by a dual‐luciferase reporter assay kit (Promega) according to manufacturer's instructions. Renilla luciferase activity was used to normalize firefly luciferase activity.

Immunofluorescence analysis

HEK293A cells were seeded on glass cover slips in 6‐well plates and transfected with plasmids or siRNAs as indicated in the figures. After transfection, cells were fixed with 4% paraformaldehyde (Biosesang) for 20 min at 25°C and washed three times with PBS briefly. After washing, cells were permeabilized with 0.1% Triton X‐100 in PBS for 15 min at 25°C and washed three times with PBS. After washing, blocking was performed by 5% BSA in PBS for 1 h at 25°C, and cells were incubated with primary antibodies diluted in 5% BSA in PBS overnight at 4°C. An antibody for YAP (Cell signaling, #14074S) was used as the primary antibody. After incubation, cells were washed three times with 1% BSA in PBS for 10 min, and cells were incubated with secondary antibodies (Alexa Fluor 546 goat anti‐Rabbit; Invitrogen, A11035 and Alexa Fluor 488 goat anti‐Rabbit; Invitrogen, A11034) diluted in 5% BSA in PBS for 1 h at 25°C. After incubation, cells were washed with three times with 1% BSA in PBS for 10 min. DAPI (300 ng/ml; Sigma, D9542) staining was performed during the second wash step. Glass cover slips were mounted on slide glass with mounting solution (SouthernBiotech), and fluorescence was measured with an LSM 800 confocal microscope (Zeiss). Quantification of immunofluorescence assays was done using MetaMorph image analysis software (Molecular Devices, USA).

Subcellular fractionation assay

Subcellular fractionation assays were performed as described (Yin et al, 2013). Cells were incubated for 20 min at 4°C in S100/P100 buffer (20 mM Tris (pH 7.4), 150 mM NaCl, 2.5 mM EDTA, and 1 mM EGTA) supplemented with 1 mM phenylmethylsulfonyl fluoride and 1 μg/ml of leupeptin. Cell lysates were passed through a 26 gauge needle 20 times, and then centrifuged at 1,000 g, 4°C for 2 min to separate and remove nuclei. The supernatant was centrifuged at 20,000 g, 4°C for 60 min to separate cytoplasmic pool (supernatant) and membrane pool (pellet). The pellet was placed in lysis buffer and analyzed by immunoblotting assay, as described above.

Preparation of mouse tissue

Mice were euthanized by carbon dioxide inhalation, and tissues were dissociated. Tissues were homogenized in a lysis buffer (20 mM Tris (pH 7.4), 150 mM NaCl, 1% Triton X‐100, 1 mM EDTA, 1 mM EGTA, 2.5 mM sodium pyrophosphate, 1 mM β‐glycerophosphate, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and 1 μg/ml of leupeptin) at 4°C and centrifuged at 12,000 g, 4°C for 15 min. The supernatant was collected and centrifuged again at 12,000 g, 4°C for 10 min. The supernatant was collected and analyzed for immunoblotting. All experimental procedures were performed according to animal care and ethics legislation; the protocols were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Seoul (UOS IACUC‐2020‐01‐TA).

Author contributions

WJ and SK carried out most of the experiments and wrote the manuscript. UL, ZAZ, MS, HK, JK, and TL performed the experiments. JWC provided intellectual input and all materials for O‐GlcNAcylation. BOW, VLK, and E‐hJ supervised the project and wrote the manuscript.

Conflict of interest

The authors declare that there are no conflicts of interest regarding the publication of this paper. BOW is a member of the Surrozen scientific advisory board and receives research support from Janssen Pharmaceutica.

Supporting information

Appendix

Expanded View Figures PDF

Review Process File

Source Data for Expanded View and Appendix

EMBO Reports (2020) 21: e50103

Data availability

All data are available in the manuscript text and supporting information. No primary datasets have been generated or deposited.