Dishevelled enables casein kinase 1–mediated phosphorylation of Frizzled 6 required for cell membrane localization

Frizzleds (FZDs) are receptors for secreted lipoglycoproteins of the Wingless/Int-1 (WNT) family, initiating an important signal transduction network in multicellular organisms. FZDs are G protein–coupled receptors (GPCRs), which are well known to be regulated by phosphorylation, leading to specific downstream signaling or receptor desensitization. The role and underlying mechanisms of FZD phosphorylation remain largely unexplored. Here, we investigated the phosphorylation of human FZD6. Using MS analysis and a phospho-state– and -site–specific antibody, we found that Ser-648, located in the FZD6 C terminus, is efficiently phosphorylated by casein kinase 1 ϵ (CK1ϵ) and that this phosphorylation requires the scaffolding protein Dishevelled (DVL). In an overexpression system, DVL1, -2, and -3 promoted CK1ϵ-mediated FZD6 phosphorylation on Ser-648. This DVL activity required an intact DEP domain and FZD-mediated recruitment of this domain to the cell membrane. Substitution of the CK1ϵ-targeted phosphomotif reduced FZD6 surface expression, suggesting that Ser-648 phosphorylation controls membrane trafficking of FZD6. Phospho-Ser-648 FZD6 immunoreactivity in human fallopian tube epithelium was predominantly apical, associated with cilia in a subset of epithelial cells, compared with the total FZD6 protein expression, suggesting that FZD6 phosphorylation contributes to asymmetric localization of receptor function within the cell and to epithelial polarity. Given the key role of FZD6 in planar cell polarity, our results raise the possibility that asymmetric phosphorylation of FZD6 rather than asymmetric protein distribution accounts for polarized receptor signaling.


Frizzleds (FZDs) are receptors for secreted lipoglycoproteins of the Wingless/Int-1 (WNT) family, initiating an important signal transduction network in multicellular organisms. FZDs are G protein-coupled receptors (GPCRs)
, which are well known to be regulated by phosphorylation, leading to specific downstream signaling or receptor desensitization. The role and underlying mechanisms of FZD phosphorylation remain largely unexplored. Here, we investigated the phosphorylation of human FZD 6 . Using MS analysis and a phospho-state-and -site-specific antibody, we found that Ser-648, located in the FZD 6 C terminus, is efficiently phosphorylated by casein kinase 1 ⑀ (CK1⑀) and that this phosphorylation requires the scaffolding protein Dishevelled (DVL). In an overexpression system, DVL1, -2, and -3 promoted CK1⑀-mediated FZD 6 phosphorylation on Ser-648. This DVL activity required an intact DEP domain and FZD-mediated recruitment of this domain to the cell membrane. Substitution of the CK1⑀-targeted phosphomo-tif reduced FZD 6 surface expression, suggesting that Ser-648 phosphorylation controls membrane trafficking of FZD 6 . Phospho-Ser-648 FZD 6 immunoreactivity in human fallopian tube epithelium was predominantly apical, associated with cilia in a subset of epithelial cells, compared with the total FZD 6 protein expression, suggesting that FZD 6 phosphorylation contributes to asymmetric localization of receptor function within the cell and to epithelial polarity. Given the key role of FZD 6 in planar cell polarity, our results raise the possibility that asymmetric phosphorylation of FZD 6 rather than asymmetric protein distribution accounts for polarized receptor signaling.
Signal transduction through G protein-coupled receptors (GPCRs) 4 is fine-tuned by phosphorylation (1,2). Barcoding of receptor function through phosphorylation of intracellular domains of GPCRs is involved in receptor desensitization, recruitment of scaffold proteins such as ␤-arrestins, functional selectivity, and receptor internalization (3). FZDs functionally interact with the scaffold protein Dishevelled (DVL) to mediate WNT/␤-catenin and ␤-catenin-independent signaling (4,5). Mammalian DVL exists in three isoforms, DVL1, -2, and -3, and interacts with FZDs, probably in a multimodal fashion involving DVL's DEP domain and the PDZ domain binding the third intracellular loop and the conserved C-terminal KTXXXW motif of FZDs, respectively (5)(6)(7)(8)(9)(10)(11). Although our understanding of the coupling of Class F receptors to heterotrimeric G proteins is improving (12)(13)(14)(15), the role and underlying mechanisms of FZD phosphorylation remain largely obscure. Direct biochemical evidence for phosphorylation of FZDs is sparse, but phosphorylation of FZD 1 and FZD 3 is involved in negative feedback regulating FZD-mediated signaling along the planar cell polarity pathway, which was found to depend on DVL in the case of FZD 3 (16 -19).
FZD 6 , the focus of this study, is closely related to FZD 3 (14) and is expressed predominantly in the olfactory epithelium, thyroid and parathyroid glands, lung, and uterus in adult mice (Ref. 20; for an overview of expression in human tissues, see https://www.proteinatlas.org/ENSG00000164930-FZD6/ tissue). 5 It has been found amplified in many different forms of cancer, including breast, ovarian, and prostate cancer (21). FZD 6 mediates exclusively ␤-catenin-independent signaling through DVL and as a G i -and G q -coupled receptor (22)(23)(24)(25)(26). When mutated, FZD 6 can cause nail dysplasia in humans, and knockout mice lacking FZD 6 show defects in claw formation, hair patterning, and tissue polarity, phenotypes reflecting dysfunctional planar cell polarity (PCP) signaling (24,25,(27)(28)(29)(30). Polarity of epithelial tissues is defined by asymmetry of epithelial cells perpendicular to the apical-basal axis. WNT/FZD signaling and in particular the closely related FZD 3 and FZD 6 are crucial for the polarity and PCP signaling in cells and tissues (24,25,27,31). However, it remains a conundrum how asymmetry is achieved by a protein that does not necessarily show polarized cellular distribution (32).
In the present study, we investigated the mechanisms and role of C-terminal FZD 6 phosphorylation. Our results identify serine 648 (Ser-648) as a CK1⑀ target site, and we propose a mechanism where DVL and CK1⑀ cooperate to achieve FZD 6 phosphorylation at the plasma membrane. The development of a phospho-Ser-648 -selective polyclonal FZD 6 antibody enabled us to depict polarization of phosphorylated FZD 6 in epithelial cells of the human fallopian tube, suggesting that phosphorylation of FZD 6 rather than asymmetric localization could code for polarized signaling.

Mass spectrometry analysis of FZD 6 phosphorylation
The C terminus of human FZD 6 is 211 amino acids long and contains more than 50 phosphorylatable amino acids (Fig. 1A).
Previous analysis with the MiniMotif Miner predicted the presence of two cAMP-dependent protein kinase (PKA), eight Ca 2ϩ -dependent protein kinase, eight casein kinase 1 (CK1), seven casein kinase 2 (CK2), one Ca 2ϩ /calmodulin-dependent protein kinase II, five ribosomal S6 kinase, one extracellular signal-regulated kinase 1/2, seven glycogen synthase kinase 3 (GSK3), one polo-like kinase, and one epidermal growth factor receptor sites (33). To assess FZD 6 phosphorylation experimentally, we used MS. HEK293 cells expressing FZD 6 -GFP were treated with either control, WNT-3A-, or WNT-5A-containing conditioned medium for 30 min and processed for LC-MS/MS analysis. Six phosphorylated serine residues were detected in the C terminus of FZD 6 ( Fig. 1, B and C). Despite stimulation with conditioned medium, the detected phosphorylation events were rare compared with the amount of predicted kinase sites and phosphorylatable residues in FZD 6 . In total, phosphorylation in six positions was detected on Ser-592, Ser-620, Ser-629, Ser-648, Ser-653, and Ser-656 (Fig.  1C). Although Ser-592, Ser-620, and Ser-629 appeared to be constitutively phosphorylated and we could detect tendencies for agonist-induced phosphorylation in Ser-648, Ser-653, and Ser-656, the total number of detected peptides did not allow firm conclusions on WNT-induced changes.

FZD 6 phosphorylation at Ser-648 is detectable using a phospho-specific antibody
Despite our initial interest in several phosphorylatable residues in the C terminus of FZD 6 , we became particularly interested in the phosphorylation of Ser-648 because we were able to raise a phospho-specific antibody against this residue (SESARpS*EGRISP peptide where pS* is Ser(P)-648; Fig. 2A). The anti-Ser(P)-648 rabbit polyclonal antibody displayed a weak signal when detecting overexpressed FZD 6 (V5-FZD 6 -mCherry; Fig. 2B) in HEK293 cells; however, the signal was specific as it did not detect the FZD 6 -S648A mutant where serine 648 was mutated to alanine (Fig. 2C). The antibody is phospho-selective because alkaline phosphatase (AP) treatment of the lysate abolished the signal (Fig. 2D). Concurrently, we validated the FZD 6 -S648A mutant with regard to its membrane localization (Fig. 2E) and glycosylation (Fig. 2F) compared with WT FZD 6 , indicating that the point mutation does not affect overall maturation of the receptor protein. In contrast, cell ELISA-based assessment of receptor surface expression 5 Please note that the JBC is not responsible for the long-term archiving and maintenance of this site or any other third party-hosted site.  6 . A, amino acid sequence of the FZD 6 C terminus. Putative phosphorylation sites are underlined (determined with the MiniMotif Miner software). The KTXXXW motif is shown in orange. Black/red residues correspond to phosphorylation sites detected by LC-MS/MS analysis presented in Fig. 1B. B, LC-MS/MS data are presented as relative peak areas of all peptides (percentage of maximum) in a sample containing a specific phosphorylated amino acid. Cells lysed for MS were treated with control (ctrl), WNT-3A-, or WNT-5A-conditioned medium. C, schematic depiction of detected phosphorylation sites (symbolized by red dots).

Phosphorylation of FZD 6
indicated that significantly lower levels of FZD 6 -S648A compared with WT FZD 6 are present in the plasma membrane (Fig. 2E).

FZD 6 Ser-648 is phosphorylated in a CK1⑀-dependent manner
In silico phospho-prediction analysis of FZD 6 using the Mini-Motif Miner software (34) suggested that Ser-648 provides a consensus site for casein kinase 1 (Fig. 3A). Furthermore, Ser-645 was predicted to serve as a priming site enabling CK1-dependent phosphorylation of Ser-648.
We therefore decided to test CK1⑀ as a candidate kinase and coexpressed it with FZD 6 . As shown in Fig. 3B, coexpression of CK1⑀ leads to a substantial increase in the Ser(P)-648 FZD 6 signal accompanied by increased binding of CK1⑀ to FZD 6 . To verify that the Ser-648 phosphorylation depends on CK1⑀ activity, we treated the cells with the CK1␦/⑀-selective inhibitor PF670462, which abrogated the Ser(P)-648 signal. Also, the Ser(P)-648 signal was sensitive to AP treatment of the lysates, and the S648A mutant was not phosphorylated upon CK1⑀ coexpression, which further confirmed the specificity of the detected Ser(P)-648 signal.
To explore the priming properties of Ser-645, we created two additional mutants: Ser-645 was mutated to either alanine (S645A) to disable phosphorylation of Ser-645 or to glutamate (S645E), which was intended to mimic the priming phosphorylation event. Both mutants were localized in the membrane, albeit with slightly lower surface expression compared with the WT receptor (Fig. 3C). When coexpressed with CK1⑀, the S645A mutant did not exhibit a Ser(P)-648 signal compared with the WT receptor (Fig. 3D). In contrast, when S645E was coexpressed with CK1⑀, the observed Ser(P)-648 signal was stronger than that of the WT FZD 6 . Overexpressing S645E on A, schematic depiction of the phosphorylated (red, asterisk) human FZD 6 -based (UniProt ID O60353) peptide used for raising a phospho-specific polyclonal (rabbit) antibody (anti-Ser(P)-648 FZD 6 ). B, schematic depiction of the WT FZD 6 construct consisting of signal sequence (SS), V5 tag, FZD 6 , and mCherry. C, HEK293 cells transfected with pcDNA3.1 (control), WT V5-FZD 6 -mCherry, or V5-FZD 6 -S648A-mCherry were lysed, and immunoprecipitation was performed using an anti-RFP (mCherry) antibody. Precipitate (IP) was probed with anti-V5 and the Ser(P)-648 FZD 6 antibodies. Total cell lysate (TCL) was probed with anti-V5 and anti-␤-actin as a loading control. The density of Ser(P)-648 -stained bands, measured using ImageJ, was normalized by dividing the value by the average of the whole experiment. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of three individual experiments. Selected results of the multiple comparison analysis of all data are presented. Results were analyzed with one-way ANOVA and Tukey's multiple comparisons post hoc test. Significance levels are given as *** (p Ͻ 0.001) and ns (not significant). D, cell lysates were treated with AP. E, subcellular distribution of WT V5-FZD 6 -mCherry or V5-FZD 6 -S648A-mCherry was analyzed by confocal microscopy in living HEK293 cells. Scale bars, 10 m. Cell ELISA was used to quantify surface expression of WT V5-FZD 6 -mCherry or V5-FZD 6 -S648A-mCherry using an antibody against the N-terminal V5 tag. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of five individual experiments performed in triplicates. Background fluorescence detected in pcDNA-transfected HEK293 cells was subtracted from all data, and mean values were normalized to WT FZD 6 surface expression. Significance levels are given as * (p Ͻ 0.05). F, for detection of N-linked glycosylation, immunoprecipitates were treated with N-gF, analyzed by PAGE, and probed with anti-RFP. HEK293 lysates expressing WT V5-FZD 6 -mCheery or V5-FZD 6 -S648A-mCherry were divided into three equal parts, and FZD 6 was pulled down using an anti-V5 antibody. The first sample was left untreated (Ϫ), the second was subjected to the deglycosylation treatment but without the enzyme (ctrl), and the third sample was subjected to the deglycosylation treatment including the enzyme (N-gF). WB, Western blot.

Phosphorylation of FZD 6
Phosphorylation of FZD 6 its own led to a Ser(P)-648 signal comparable with the WT receptor in the presence of overexpressed CK1⑀. These findings suggest that Ser-645 indeed serves as a priming site for CK1⑀.
The anti-Ser(P)-648 antibody enabled us to visualize phosphorylated FZD 6 in relation to the FZD 6 -mCherry signal using indirect immunofluorescence in HEK293 cells coexpressing CK1⑀ and FZD 6 . FZD 6 -mCherry is widely distributed in the cell with a distinct fraction in the plasma membrane, similar to what we have reported before for this or similar constructs (22,23,30,35). However, the WT FZD 6 phosphorylated at Ser-648 was predominantly present at the plasma membrane (Fig. 4A). In addition, phosphorylated FZD 6 was detectable in intracellular puncta, most likely early endosomes or lysosomes. To confirm the localization of the phosphorylated FZD 6 to the endosomal pathway, which would be in agreement with previous data (30), we used the anti-Ser(P)-648 antibody in cells coex- A, schematic depiction of a CK1 consensus site matching the sequence of human FZD 6 (gray background). Ser-648, the phospho-acceptor serine in the CK1 consensus site, is highlighted in red and marked with an asterisk. The predicted CK1 priming site Ser-645 is underlined. B, HEK293 cells transfected with WT V5-FZD 6 -mCherry or V5-FZD 6 -S648A-mCherry in combination with CK1⑀ were treated with the CK1 inhibitor PF670462 (10 M; overnight). Cells were lysed, and immunoprecipitation was performed using an anti-V5 antibody. Precipitate (IP) was probed with the anti-Ser(P)-648 FZD 6 , anti-CK1⑀, and anti-RFP (mCherry) antibodies. To verify phosphate specificity of the anti-Ser(P)-648 antibody, lysates were treated with AP. The total cell lysate (TCL) was probed with anti-RFP and anti-CK1⑀. Anti-␤-actin served as a loading control. C, living HEK293 cells are shown expressing either WT V5-FZD 6 -mCherry, V5-FZD 6 -S645A-mCherry, or V5-FZD 6 -S645E-mCherry. Scale bars, 10 m. Cell ELISA was performed to quantify the construct's surface expression compared with the WT V5-FZD 6 -mCherry construct using an antibody against the N-terminal V5 tag. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of four individual experiments performed in triplicates. Background fluorescence detected in pcDNA-transfected HEK293 cells was subtracted, and mean values were normalized to FZD 6 surface expression. ns, not significant. D, anti-RFP immunoprecipitation was done as in Fig. 3B. B and D, the density of Ser(P)-648 -stained bands, measured using ImageJ, was normalized by dividing the value by the average of the whole experiment. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of three individual experiments. Selected results of the multiple comparison analysis of all data are presented. Results were analyzed with one-way ANOVA and Tukey's multiple comparisons post hoc test. Significance levels are given as * (p Ͻ 0.05), *** (p Ͻ 0.001), and ns (not significant). WB, Western blot; ctrl, control.  6 -mCherry or V5-FZD 6 -S648A-mCherry in combination with CK1⑀ were fixed in 4% paraformaldehyde, stained with anti-Ser(P)-648 FZD 6 and anti-CK1⑀, and visualized with a confocal microscope. Arrows point to Ser(P)-648 FZD 6 detection in intracellular puncta. Scale bars, 10 m. B, HEK293 cells cotransfected with WT V5-FZD 6 -mCherry, CK1⑀, DVL2, and Rab5-GFP. Rab5-GFP serves as an early endosome marker. The cells were fixed in 4% paraformaldehyde, stained with anti-Ser(P)-648 and DAPI, and visualized with a confocal microscope. Arrows point to Ser(P)-648 FZD 6 -and Rab5-GFP-positive endosomes. Scale bar, 10 m. 6 pressing CK1⑀, DVL, V5-FZD 6 -mCherry, and GFP-tagged Rab5, a marker for early endosomes. Under these conditions, which allow efficient FZD 6 phosphorylation at Ser-648, we observed punctate, intracellular structures that were positive for the FZD 6 protein (mCherry), Ser(P)-648 (anti-Ser(P)-648 FZD 6 antibody), and the endosomal marker Rab5-GFP (Fig. 4B).

Phosphorylation of FZD
As shown above, point mutations of phosphorylatable amino acids revealed phospho-specificity of the antibody and targetselective phosphorylation by CK1. However, we only observed slight reduction in surface expression of these receptor constructs (Fig. 3C). Based on a recent publication reporting on phosphorylation codes spanning three phosphorylatable residues mediating arrestin-GPCR interaction, we reasoned that functional consequences could become more apparent when mutating larger phosphorylation motifs (36). To evoke more serious malfunction, we therefore decided to systematically mutate the phosphorylatable motif, targeting several residues, including the Ser-648 and adjacent serines. We created two triple mutants, S643A/S645A/S648A and S648A/S653A/ S656A, which were expressed well on the protein level. The S643A/S645A/S648A mutant, however, showed dramatically reduced surface expression (Fig. 5, A-C). This effect was phenocopied by the S645A/S648A double mutant, which is deficient in the CK1⑀ priming and target site.

DVL is required to induce phosphorylation of FZD 6 in response to CK1⑀
CK1⑀ is a serine/threonine kinase with well defined function in WNT signaling. Its primary substrates in this pathway are the DVL proteins that strongly interact with CK1 and are phosphorylated in response to WNT proteins. Because DVL interacts with FZD 6 and is phosphorylated and regulated by C⌲1⑀ (22,(37)(38)(39), we asked whether DVL is required for phosphorylation of FZD 6 at Ser-648. We overexpressed FZD 6  Furthermore, we wanted to find out whether the posttranslational modification of FZD 6 induced by overexpression of DVL and resulting in the electrophoretic mobility shift of FZD 6 is evoked by receptor phosphorylation and/or N-glycosylation. Figure 5. Characterization of the phosphorylatable serine motif by the help of Ser double and triple mutants. A, schematic depiction of Ser to Ala mutants, including Ser-648 and adjacent Ser residues. B, HEK293 cells cotransfected with pcDNA3.1 (control), WT V5-FZD 6 -mCherry, V5-FZD 6 -S645A/S648A-mCherry (double mut 1), V5-FZD 6 -S643A/S645A/S648A-mCherry (triple mut 1), or V5-FZD 6 -S648A/S653A/S656A-mCherry (triple mut 2); FLAG-DVL3; and CK1⑀ were lysed and analyzed by immunoblotting using anti-Ser(P)-648 and anti-FZD 6 antibodies. Anti-GAPDH served as a loading control. The density of Ser(P)-648 -stained bands, measured using ImageJ, was normalized by dividing each value by the average of the whole experiment. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of three individual experiments. Selected results of the multiple comparison analysis of all data are presented. Results were analyzed with one-way ANOVA and Tukey's multiple comparisons post hoc test. Significance levels are given as *** (p Ͻ 0.001) and ns (not significant). C, receptor surface expression of HEK293 cells transiently transfected with pcDNA3.1, WT V5-FZD 6 -mCherry, V5-FZD 6 -S645A/S648A-mCherry (double mut 1), V5-FZD 6 -S643A/S645A/S648A-mCherry (triple mut 1), or V5-FZD 6 -S648A/S653A/ S656A-mCherry (triple mut 2) was quantified by cell ELISA using an antibody against the N-terminal V5 tag. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of five individual experiments performed in triplicates. Background fluorescence detected in pcDNA-transfected HEK293 cells was subtracted from all data, and mean values were normalized to WT FZD 6 surface expression. Significance levels are given as * (p Ͻ 0.05) and ns (not significant). WB, Western blot.

Phosphorylation of FZD 6
Therefore, we overexpressed FZD 6 alone or together with CK1⑀ and DVL3 followed by either enzymatic dephosphorylation or deglycosylation. Only treatment with alkaline phosphatase resulted in a decrease of the FZD 6 band at lower mobility, whereas treatment with N-glycosidase F did not (Fig. 6B), confirming the importance of FZD 6 phosphorylation for the electrophoretic mobility shift of the receptor protein. Similar data were obtained when reconstituting the DVL-KO cells with FLAG-DVL1 or FLAG-DVL2 in combination with V5-FZD 6 -mCherry and CK1ε and subsequent treatment of the lysate with AP (Fig. S1).
To investigate whether Ser-648 is phosphorylated as part of the DVL-induced FZD 6 phosphorylation, the same two samples were loaded in parallel and analyzed by antibodies against RFP and Ser(P)-648 (Fig. 7A). We observed that the DVL-induced electrophoretic mobility shift of FZD 6 (marked by a filled arrowhead) corresponds roughly to the Ser(P)-648 signal (marked by two stars), and both these signals migrate at ϳ20 kDa higher molecular mass ("hyperphosphorylated" FZD 6 ) compared with the two species of FZD 6 (glycosylated and unglycosylated; marked by open arrowheads) detected by the anti-RFP antibody. In addition, we observed a weaker and faster migrating Ser(P)-648 FZD 6 band (marked by one star), which could represent unspecific binding of anti-Ser(P)-648 to unphosphorylated FZD 6 . This band was only observed upon very high overexpression of the receptor and was insensitive to AP treatment (Fig. S2).
We confirmed that the DVL-dependent shift of FZD 6 is detectable in the WT HEK293 cells upon DVL overexpression ( Fig. 7, A and B). In Fig. 7B, we show that overexpressing both CK1⑀ and DVL3 had an additive effect on the amount of Ser(P)-648 and the shifted form of FZD 6 compared with expression of CK1⑀ or DVL3 alone. Also, the DVL-induced shift of FZD 6 and phosphorylation of Ser-648 were to a great extent (although not exclusively) dependent on the activity of CK1⑀ (Fig. 7, A and B). Thus, CK1⑀ is crucial for phosphorylation of FZD 6 in the presence of DVL even though we cannot exclude that other kinases participate in this process.
The DVL-dependent electrophoretic mobility shift of FZD is not limited to FZD 6 . We were able to detect a DVL-dependent electrophoretic mobility up-shift of FZD 4 when coexpressed with DVL2 (11). In DVL-KO-HEK293 cells, a DVL2-dependent shift of FZD 4 was observed when WT receptor was expressed. However, when FZD 4 -Y250F, a mutant that is impaired in DVL2 binding (11), was expressed, the DVL2-dependent shift was reduced (Fig. S3), suggesting that FZD-DVL interaction and receptor phosphorylation go hand in hand. Furthermore, a similar DVL-induced, phosphorylation-dependent electrophoretic mobility shift was reported for FZD 3 (17,19).

FZD 6 is phosphorylated in a GRK-dependent manner
The findings that DVL-dependent FZD 6 phosphorylation was not fully inhibited by pharmacologically blocking CK1␦/⑀ and that FZD 4 , which lacks a residue homologous to Ser-648 in FZD 6 , was also shifted by DVL raises the question whether there are other kinases involved in FZD 6 phosphorylation. To  6 investigate other candidate kinases potentially mediating the DVL-induced phosphorylation of FZD 6 , we used inhibitors of GSK3 (CHIR99021), PKA (H89), or GRK2 (paroxetine) (Fig. 8A). The CK1␦/⑀ inhibitor PF670462 was used as a positive control. Although inhibition of PKA or GSK3 did not reduce the Ser(P)-648 FZD 6 signal, paroxetine reduced Ser(P)-648 to a similar extent as PF670462, pinpointing GRK2 as an FZD 6 kinase. To investigate the role of GRK2 and the paroxetine-insensitive GRK5 as FZD 6 kinases in more detail, we overexpressed either CK1⑀, GRK2, or GRK5 with FZD 6 in HEK293 cells endogenously expressing DVL. As shown in Fig. 8B, both GRK2 and GRK5 induced FZD 6 phosphorylation effectively under these conditions.

Phosphorylation of FZD
To investigate the relationship between the CK1-induced and DVL-dependent Ser-648 FZD 6 phosphorylation and membranous expression of FZD 6 , we used a previously characterized nail dysplasia mutant of this receptor, the FZD 6 -R511C mutant (22,30). The R511C mutant results in decreased surface expression and prominent localization of FZD 6 -R511C to the endosomal/lysosomal pathway (30). The comparison of WT FZD 6 and FZD 6 -R511C in the absence and presence of coexpressed CK1⑀ and DVL3 indicates that WT FZD 6 is more efficiently phosphorylated than FZD 6 -R511C at similar total expression levels of the two receptor constructs (Fig. 8C).

The DEP domain of DVL is required for Ser-648 phosphorylation and the shift of FZD 6
To better understand underlying mechanisms of DVL-mediated phosphorylation of FZD 6 , we overexpressed FZD 6 in DVL-KO-HEK293 cells in combination with either DVL3 or various deletion or point mutants of DVL3 (Fig. 9A). The DVL3 ⌬DIX mutant is missing the N-terminal DIX domain important for multimerization and ␤-catenin signaling (41,42). DVL3 ⌬Cterm lacks the very C terminus of DVL. Phosphorylation of this region disrupts the signalosomes observed as puncta and is inhibitory for ␤-catenindependent signaling (38). DVL3 ⌬(DEPϩC) is missing the DEP domain in addition to missing the C terminus, whereas DVL3 DEP represents the DEP domain of DVL3. In the DVL3-K435M mutant, the binding interface of FZD-DVL is disrupted, and thus, this mutant is not recruited to the membrane by FZD (43,44).
The DIX domain and the C terminus of DVL are dispensable for the induction of both Ser(P)-648 and the electrophoretic mobility shift of FZD 6 ( Fig. 9, B and C). However, deleting the C terminus and the DEP domain abolished the effects on FZD 6 (Fig.  9D). In contrast, overexpression of the DEP domain of DVL3 was neither sufficient to induce Ser(P)-648 nor the shift of FZD 6 (Fig.  9E), indicating that the DEP domain is required, but not sufficient, to induce Ser(P)-648 and the FZD 6 shift. It should be noted, however, that both DVL3 ⌬(DEPϩC) and the minimal DVL3 DEP displayed lower expression compared with the WT DVL3. The DVL3-K435M mutant neither induced Ser(P)-648 nor the shift of FZD 6 (Fig. 9F), consistent with the importance of the DEP domain found in Fig. 9D and indicating that the ability of DVL to bind FZD is crucial for its activity toward FZD phosphorylation.
Furthermore, we asked which properties of the DVL DEP domain were required for induction of Ser(P)-648 and the FZD 6 shift. Several mutants of the DVL1 DEP domain (described in the table in Fig. 10A and validated by Paclikova et al. (45)) were used. Instead of WT DVL3 as in Fig. 9, we used DVL1(1-502), a mutant of DVL1 lacking the C terminus, as a positive control for Fig. 10. All the other DVL1 mutants were created in the background of the DVL1(1-502) mutant. In Fig. 10B, we show that the K438M mutation in DVL1 (corresponding to K435M in Figure 7. DVL induces phosphorylation of FZD 6 . A, HEK293 cells were transfected with WT V5-FZD 6 -mCherry in the absence or presence of CK1⑀ and FLAG-DVL3. V5-FZD 6 -mCherry was precipitated using an anti-V5 antibody, and samples were probed with either anti-Ser(P)-648 FZD 6 or anti-RFP (mCherry) antibody. The same two samples were loaded twice, separated by a marker lane. The filled arrowhead depicts hypershifted species of V5-FZD 6 -mCherry. Open arrowheads indicate glycosylated (upper) and unglycosylated (lower) forms of V5-FZD 6 -mCherry. ** marks the phosphorylated receptor positive for Ser(P)-648 FZD 6 . * depicts either an unspecific signal from anti-Ser(P)-648 antibody (unphosphorylated FZD 6 ) or a weakly phosphorylated species. B, HEK293 cells overexpressing V5-FZD 6 -mCherry were cotransfected with combinations of CK1⑀ and FLAG-DVL3 and treated with the CK1⑀ inhibitor PF670462 (10 M; overnight), and V5-FZD 6 -mCherry was precipitated using an anti-V5 antibody. Immunoblots show abundance of proteins in the pulldown (IP) or total cell lysate (TCL) detected by anti-Ser(P)-648, anti-RFP, anti-CK1⑀, and anti-FLAG antibodies. Anti-␤-actin served as a loading control. The density of Ser(P)-648 -stained bands, measured using ImageJ, was normalized by dividing each value by the average of the whole experiment. Data are presented as a scatter plot; bars represent mean and error bars represent S.D. of three individual experiments. Selected results of the multiple comparison analysis of all data are presented. Results were analyzed with one-way ANOVA and Tukey's multiple comparisons post hoc test. Significance levels are given as * (p Ͻ 0.05), ** (p Ͻ 0.01), *** (p Ͻ 0.001), and ns (not significant). WB, Western blot.

Phosphorylation of FZD 6
DVL3) does not promote phosphorylation of Ser(P)-648 and FZD 6 shift. Fig. 10C shows that Abl-mediated phosphorylation of tyrosine 494 in DVL1, implicated to have a positive effect on DVL function in the PCP pathway, was dispensable for the induction of Ser-648 phosphorylation and the overall FZD 6 electrophoretic mobility shift. Mutating clusters of positively charged residues important for DVL binding to the phospholipids in the membrane (mutants RRRKA and HKA) neither affected the DVL1-mediated Ser(P)-648 nor the FZD 6 shift ( Fig. 10, D and E). Lastly, the deletion of the N-terminally positioned LPDSG motif in the DEP domain (mutant ⌬LPDSG) rendered DVL1 unable to induce Ser(P)-648 or FZD 6 shift (Fig. 10F). In all our experiments, the electrophoretic mobility shift of FZD 6 was accompanied by the phosphorylation of FZD 6 at Ser-648 (and vice versa), suggesting that the two events are interconnected.
In summary, our data using DVL mutants show that the DVL-induced hyperphosphorylation of FZD 6 , including

Phosphorylation of FZD 6
Ser(P)-648, requires (i) the recruitment of DVL to FZD, (ii) DVL-CK1 binding, and (iii) an intact N-terminal LPDSG motif of the DEP domain. Conversely, FZD hyperphosphorylation is neither dependent on the DIX domain of DVL nor on the ability of DVL DEP to be attached to the membrane or to be phosphorylated by Abl.

Localization of Ser(P)-648 FZD 6 in human fallopian tube epithelium
Despite the detailed mechanisms that we have identified leading to the DVL-and CK1-dependent phosphorylation of FZD 6 , its role for cell surface expression of the receptor and the full functional role of Ser(P)-648 remain obscure. Although -mCherry either alone or together with FLAG-DVL3 or one of the FLAG-DVL3 mutants listed above. Cell lysates were precipitated using anti-V5 or anti-RFP. IP and total cell lysate (TCL) were analyzed by immunoblotting and probed with anti-Ser(P)-648, anti-RFP, anti-FLAG, and anti-V5 antibodies. Anti-␤-actin served as a loading control. Quantification of the Ser(P)-648 FZD 6 signal from three independent experiments for each condition is shown in Fig. S4. WB, Western blot. 6 deletion of the CK1⑀ phosphomotif impairs surface expression, we asked now where Ser(P)-648 FZD 6 is localized compared with the total FZD 6 pool in human epithelial tissue. According to the protein expression mapping efforts of the Human Protein Atlas (46), FZD 6 is widely expressed in epithelial tissues. We focused here on human fallopian tube tissue, which showed a distinct membranous FZD 6 immunoreactivity using a polyclonal antibody with a large C-terminal epitope. Parallel staining of adjacent paraffin sections of the same fallopian tube tissue with the anti-Ser(P)-648 FZD 6 antiserum revealed a more restricted and polarized immunoreactivity (Fig. 11A). Although total FZD 6 immunoreactivity was evenly distributed in the  6 plasma membrane of all epithelial cells both on the apical and basal side of the cells, Ser(P)-648 FZD 6 immunoreactivity was predominantly apical, associated with cilia in a subset of epithelial cells (Fig. 11B). Also, double labeling and confocal imaging supported the polarized localization of the phosphorylated FZD 6 in fallopian tube epithelial cells (Fig. 11, B and C).
Initially, we identified phosphorylation sites in the C terminus of FZD 6 by MS/MS in cell lysates exposed to WNT-conditioned medium. Despite enrichment of phosphorylated peptides in cells exposed to WNT-conditioned medium compared with control stimulation, we were at a later time point not able to verify the WNT dependence of the Ser-648 phosphorylation. Because the conditions for the MS/MS experiment did not include overexpression of DVL and CK1 to efficiently enhance FZD phosphorylation, we speculate that Ser-648 FZD 6 phosphorylation was a rare event and therefore not efficiently detectable by MS/MS under all conditions.
GRKs represent a family of serine/threonine kinases that regulate GPCR function by direct phosphorylation and by the formation of a ternary complex (3,54). Here, we have identified GRK2 and GRK5 as two additional kinases able to efficiently phosphorylate FZD 6 in the presence of endogenous levels of DVL. Further studies are now required to shed light on the interaction of GRKs and DVL, the requirement of DVL for GRK-mediated FZD phosphorylation, GRK-mediated receptor barcoding regulating desensitization, internalization, and functional selectivity in resemblance to what is known for other GPCRs (1, 3). Interestingly, GRK5 and GRK6 are involved in ␤-catenindependent WNT signaling through phosphorylation of LRP6 (55).
The central tool that we present here is the polyclonal antiserum for detection of Ser-648 phosphorylation of FZD 6 . We have validated the antibody with regard to phosphorylation state and target selectivity using point mutations of FZD 6 and phosphatase treatment of the cell lysates. Furthermore, an efficient and specific signal of the Ser(P)-648 antibody in HEK293 cell lysates is only visible when DVL and kinase are coexpressed, further underlining antibody selectivity. The phosphorylation event on Ser-648 can be seen as a consequence of FZD-DVL interaction. This interaction is intrinsically labile and short-lived and can experimentally be stabilized by chemical cross-linking (11,56). The antibody provides a new tool to detect the pool of FZD 6 that has been in functional contact with DVL and to monitor this transient interaction, which results in Ser-648 phosphorylation. Furthermore, the suitability of the Ser(P)-648 FZD 6 antibody for immunohistochemistry could open new possibilities. It remains to be thoroughly validated whether detection of phosphorylated FZDs could present a useful diagnostic tool by screening patientderived cancer tissue microarrays.
Using the anti-Ser(P)-648 FZD 6 antibody, we provide evidence in human fallopian tube epithelium that receptor phosphorylation can contribute to or mirrors cellular asymmetry and polarization of receptor pools. This notion, however, should be taken with a grain of salt because our findings do not distinguish between the possibility that differential localization of Ser(P)-648 FZD 6 is the cause or consequence of the phosphorylation event at Ser-648. The experiments with the nail

Phosphorylation of FZD 6
dysplasia mutant FZD 6 -R511C, a mutant that has an intact CK1 phosphorylation motif but shows a mislocalization phenotype with decreased presence at the cell surface (30), suggest that Ser(P)-648 on FZD 6 could mark FZD 6 at the cell surface.
It should also be underlined that receptor-overexpressing HEK293 cells apparently are not a suitable model to mimic receptor distribution in "real" cells such as fallopian epithelial cells. Yet undefined mechanisms appear to be in action to sort subpopulations of receptor pools in a delicate manner, mechanisms that are most likely not present or overridden by receptor overexpression in HEK293 cells. In contrast, the observed changes in receptor surface expression upon single, dual, and triple serine mutation of the CK1-targetd FZD 6 motif underline that FZD 6 phosphorylation is important for efficient membrane embedding or for stabilization of the membrane-expressed receptor pool. In the human fallopian epithelium, it was remarkable that distinct Ser(P) FZD 6 expression was observed only in ciliated cells.
Previously, it was suggested that differential sorting of PCP proteins such as Vangl2 and FZD 6 could contribute to a polarized expression pattern in asymmetric cells (57). Our findings indicate now that receptor phosphorylation could play a complementary role in sorting proteins in a polarized manner in asymmetric cells, providing novel means of cellular polarization by posttranslational modification.
The overall importance of FZD phosphorylation is not yet understood. We have pinpointed a novel mechanism of FZD 6 phosphorylation through CK1 and DVL or GRK2/5 with implications for receptor polarization and cell surface expression. Despite the fact that the C termini across the Class F receptors are poorly conserved, consensus sites of various Ser/Thr kinases are frequent (33), especially when it comes to the closely related FZD 3 and FZD 6 , which present the longest C termini in Class F receptors. Thus, CK1-mediated or, more generally speaking, Ser/Thr kinase-mediated phosphorylation of the C termini of FZDs could represent a new concept for FZD regulation and cellular polarization. In support of our findings, previous studies have identified a role of DVL in phosphorylation of FZD 3 with functional implications for receptor hyperphosphorylation, surface expression, and signaling (17,19). In those studies, the DVL-mediated phosphorylation of FZD 3 evoked a negative feedback on receptor function.
Given the recent information of arrestin barcodes of phosphorylated GPCRs (36), it will be interesting to follow up on our findings by investigating putative arrestin interaction with phosphorylated FZD 6 . Although previous findings suggest that DVL bridges FZD-arrestin interaction (58), the DVL-assisted phosphorylation of FZD could play a role in direct arrestin binding to FZDs, resembling the well established GPCRarrestin interaction (3). Functionally, previous publications point at a central role of arrestins in the regulation of ␤catenin-dependent and -independent signaling, and thus, Ser/Thr phosphorylation of FZDs could play an important part in the modulation of WNT signaling (49,50,58,59). Further experiments are needed to better understand the divergent functions of FZD and particularly FZD 6 phosphorylation with regard to polarity, signal specification, desensitization, and arrestin and DVL interaction.

LC-MS/MS analysis
After the pulldown using the anti-GFP antibody (rabbit; 20R-GR-011, Fitzgerald), samples were incubated with 10 mM DTT at 56°C for 45 min. After removal of excess DTT, samples were incubated with 55 mM iodoacetamide at room temperature in darkness for 30 min. The trypsin digestion was performed for 2 h at 40°C on a Thermomixer (750 rpm; Eppendorf). Digested peptides were extracted using 50% acetonitrile solution with 2.5% formic acid (FA) and concentrated in a SpeedVac concen-trator (Eppendorf). An aliquot of a concentrated sample was directly analyzed by LC-MS/MS for protein identification. The rest of the sample was used for phosphopeptide analysis. The sample was diluted with acidified acetonitrile solution (80% acetonitrile and 2% FA). Phosphopeptides were enriched using the Pierce Magnetic Titanium Dioxide Phosphopeptide Enrichment kit (Thermo Fisher Scientific) according to the manufacturer's protocol with slight modifications (samples were mixed with binding solution at a 1:2 ratio before loading, and one additional washing step with binding solution after phosphopeptide capture was implemented). Eluates were concentrated under vacuum and then diluted in 10 l of 0.1% FA solution before LC-MS/MS analysis. The LC-MS analysis was performed on a nanoACQUITY UPLC (Waters Corp.) coupled to a LTQ-Orbitrap Velos mass spectrometer (Thermo Electron Corporation) equipped with a nanoelectrospray ionization source and installed with a Picotip Emitter (New Objective).
For LC separation, the digested peptides were first enriched on a nanoACQUITY UPLC 2G-V/Mtrap Symmetry C 18 precolumn (2-cm length, 180-m inner diameter, and 5-m particle size) from Waters Corp. and separated using a nanoACQUITY BEH130 C 18 column (10-cm length, 100-m inner diameter, and 1.7-m particle size from Waters Corp.). The separation was achieved with the formation of 92-min gradient containing buffer A (2% acetonitrile in water with 0.1% acetic acid) and buffer B (acetonitrile with 0.1% acetic acid; gradient, 2-5% buffer B in 2 min, 5-25% buffer B in 63 min, 25-60% buffer B in 25 min, 60 -99% buffer B in 1 min). The peptides were eluted at a flow rate of 400 nl/min. The eluted peptides were analyzed first in the Fourier transform MS operated in positive and profile mode. The second scan event is an MS/MS scan performed in data-dependent mode to fragment peptides and acquire data in positive and centroid mode. The MS automatically switches between Orbitrap-MS and LTQ-MS/MS acquisition to carry out the MS and MS/MS events. Survey full-scan MS spectra (from m/z 300 to 1700) were acquired in the Orbitrap with resolution R ϭ 30 000 with a target value of 1 ϫ E6. The method allowed sequential isolation of a maximum of the 20 most intense ions (depending on signal intensity), which were subjected to collision-induced dissociation fragmentation with an isolation width of 2 Da and a target value of 1 ϫ E4 or with a maximum ion time of 100 ms. Target ions already selected for MS/MS were dynamically excluded for 60 s. General MS conditions were electrospray voltage of 1.65 kV, no sheath and auxiliary gas flow, and capillary temperature of 300°C. Ion selection threshold was 1500 counts for MS/MS with an activation time of 10 ms and activation energy of 35% normalized (also applied for MS/MS). Only doubly and triply charged ions were triggered for tandem MS analysis.
The data were analyzed by Thermo Proteome Discoverer 1.4 and searched against the human UniProt fasta database using Sequest HT. The search parameter settings included trypsin as protease with two miscleavages; oxidation of methionine and phosphorylation of serine, threonine, and tyrosine as dynamic modifications; and carbamidomethylation of cysteine as a static modification. A peptide error tolerance of 10 ppm and fragment tolerance of 0.8 Da were used for peptide identifications with a false discovery rate of 1%. Peptides were validated by Phosphorylation of FZD 6 node Percolator, and the phosphopeptides were validated by node PhosphoRs 3.0. The output results were later filtered to yield high-confidence peptides and enable protein grouping. The MS proteomics data have been deposited to the Pro-teomeXchange Consortium via the PRIDE partner repository (61) with the data set identifier PXD011173.

Cell ELISA
For quantification of cell surface receptor expression, 5 ϫ 10 4 HEK293 cells were plated in 96-well plates coated with 0.1 mg/ml poly-D-lysine. Cells were transfected with 0.1 g of the indicated constructs and maintained for an additional 24 h. Cells were then incubated with an anti-V5 antibody (mouse; 1:1000; R960-CUS, Thermo Fisher Scientific) in 1% BSA in PBS for 1 h at 4°C. Following incubation, cells were washed five times with 0.5% BSA in PBS and probed with an horseradish peroxidase-conjugated goat anti-mouse antibody (1:4000; 31430, Thermo Fisher Scientific) in 1% BSA in PBS for 1 h at 4°C. The cells were washed five times with 0.5% BSA in PBS, and 100 l of the peroxidase substrate 3,3Ј,5,5Ј-tetramethylbenzidine (T8665, Sigma-Aldrich) was added (30 min at room temperature). After acidification with 100 l of 2 M HCl, the yellow color was read at 450 nm using a POLARstar Omega plate reader (BMG LABTECH).

Human tissue samples and generation of tissue microarrays
The use of Human Protein Atlas (HPA) tissue was covered by the HPA ethical permit (EPN Uppsala, Sweden, 2002/577, 2005/338, 2011/473) following the Declaration of Helsinki principles. Tissue was acquired from the archives at the Department of Pathology of Uppsala University Hospital. In brief, formalin-fixed and paraffin-embedded tissue samples were collected based on hematoxylin-and eosin-stained tissue sections showing representative normal histology, quality-controlled by a certified pathologist. Tissue was sectioned in 4-m-thick sections using waterfall microtomes (Microm HM 355S, Thermo Fisher Scientific), dried at room temperature overnight, and baked at 50°C for 12-24 h prior to immunohistochemical staining.

Immunohistochemistry
Tissue sections were deparaffinized in xylene, hydrated in graded alcohols, and blocked for endogenous peroxidase in 0.3% hydrogen peroxide diluted in 95% ethanol. For antigen retrieval, a Decloaking Chamber (Biocare Medical) was used. Slides were immersed and boiled in citrate buffer, pH 6 (Lab Vision), for 4 min at 125°C and then allowed to cool to 90°C (the whole cycle was ϳ40 min). Automated immunohistochemistry was performed using an Autostainer 480 instrument (Thermo Fischer Scientific). The primary anti-FZD 6 (rabbit; 1:300; HPA017991, Human Protein Atlas project-Atlas Antibodies AB) and the anti-Ser(P)-648 FZD 6 (rabbit; 1:200; Moravian Biotechnology) were diluted in UltraAb Diluent (Thermo Fisher Scientific) followed by incubation for 30 min at room temperature. The slides were further incubated with the secondary reagent anti-rabbit/mouse horseradish peroxidaseconjugated UltraVision (Thermo Fischer Scientific) for 30 min at room temperature and developed for 10 min using diaminobenzidine Quanto (Thermo Fisher Scientific) as chromogen.
All incubations were followed by a rinse in Wash Buffer (Thermo Fisher Scientific) two times for 5 min. Slides were counterstained in Mayer's hematoxylin (Histolab) and coverslipped using Pertex (Histolab) as mounting medium. The stained slides were dried in oven overnight and digitalized using Scanscope AT2 (Aperio) using a 20ϫ objective.

Statistical analysis
All immunoblot experiments are representative of n ϭ 3. Statistical and graphical analyses were done using GraphPad Prism 6 software. Quantification of the Western blots was performed by ImageJ software. Briefly, the density of each Ser(P)-648 -stained band (Figs. 2C; 3, B and D; 5B; 6A; 7B; 8, A and B; 9, B-F; and 10, B-F) was measured and normalized by dividing the value by the average of the whole experiment. Normalized values were plotted as a scatter plot, and bars represent mean Ϯ S.D. of three independent experiments. Differences among the groups were analyzed by one-way ANOVA with Tukey's multiple comparisons post hoc test. Significance levels are given as follows: *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001. For analyzing cell ELISA data, the basal fluorescence detected in pcDNA-transfected HEK293 cells was subtracted from all data, and mean values were normalized to WT FZD 6 surface expression. Differences among the groups were analyzed by two-tailed t test or one-way ANOVA with Dunnett's multiple comparisons post hoc test. Significance levels are given as follows: *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001. All data points represent normalized values, each performed in triplicates, and values Phosphorylation of FZD 6 were plotted as a scatter plot. Bars represent mean Ϯ S.D. of four to five independent experiments.