FAM83F regulates canonical Wnt signalling through an interaction with CK1α

FAM83F directs CK1α to the plasma membrane, and through its association with CK1α, FAM83F mediates canonical Wnt signalling.


INTRODUCTION
FAM83F belongs to the FAM83 family of proteins, which is characterised by a conserved Nterminal DUF1669 domain. We have recently shown that the DUF1669 domain mediates interaction with the α, δ or ε isoforms of the CK1 family of Ser/Thr protein kinases (1). The FAM83 proteins direct the CK1 isoforms with which they interact to distinct subcellular compartments, thereby potentially regulating their substrate pools (1). All FAM83 proteins interact with CK1α, albeit with varying affinity, while FAM83A, B, E and H also interact with CK1δ and ε isoforms (1). CK1α, δ and ε isoforms have been implicated in numerous cellular processes including Wnt signalling, mitosis, circadian rhythm and DNA damage responses (2)(3)(4)(5).
There is now increasing evidence that FAM83 proteins regulate the diverse biological roles of CK1 isoforms. For example, FAM83G (a.k.a. PAWS1) regulates canonical Wnt signalling downstream of the β-catenin destruction complex through association with CK1α (6).
Interestingly, two mutations within the DUF1669 domain of the FAM83G gene that cause palmoplantar keratoderma result in the loss of FAM83G-CK1α interaction and attenuation of Wnt signalling (7). FAM83D directs CK1α to the mitotic spindle to ensure proper spindle alignment and timely exit from mitosis (8), and FAM83H mutations that cause amelogenesis imperfecta retain interaction with CK1 isoforms but are mis-localised in cells (9,10).
However, the biological and biochemical roles of FAM83F are poorly understood. High levels of FAM83F protein have been linked to oncogenesis in glioma (11), lung cancer (12), oesophageal cancer (13) and thyroid cancer (14) yet the underlying mechanisms remain unknown. Sequence alignment of the conserved DUF1669 domain reveals that FAM83F most resembles FAM83G and is the only other FAM83 protein to induce Wnt reporter activity in an overexpression assay (Sup. Fig. 1). We therefore sought to explore whether FAM83F is also involved in regulating canonical Wnt signalling.
Wnt signalling plays important roles in embryogenesis and cell proliferation as well as in stem cell and adult tissue homeostasis (15). The key effector of the canonical Wnt signalling pathway is β-catenin. Under basal conditions, most β-catenin protein is located at the adherens junctions, while cytoplasmic β-catenin levels are kept in check by the β-catenin destruction complex. The destruction complex is composed principally of two scaffold proteins, Axin and Adenomatous polyposis coli (APC), and two protein kinases, glycogen synthase kinase-3β (GSK-3β) and casein kinase 1α (CK1α). Phosphorylation of β-catenin at S45 by CK1α primes β-catenin for GSK-3β mediated sequential phosphorylation at T41, S37 and S33, which allows the β-transducin repeat-containing protein (β-Trcp) to ubiquitylate βcatenin and facilitate its degradation through the proteasome (16). Upon binding Wnt ligands, the Wnt receptor Frizzled and co-receptor LRP6 recruit Dishevelled and the βcatenin destruction complex to the plasma membrane. This complex, termed the Wnt signalosome, in turn becomes internalised in multivesicular bodies, thus sparing the degradation of cytoplasmic β-catenin (17). The resultant stabilised β-catenin then translocates to the nucleus, where it binds to its co-transcriptional factor, T-cell factor (TCF), and triggers the transcription of Wnt target genes, such as Axin2, C-myc and Cyclin D1 (18).
Aberrant Wnt signalling is a common feature in various cancers, particularly those of gastrointestinal origin including a vast majority of colorectal cancers (CRC) (19).
In this study we explore the role of FAM83F in driving Wnt signalling in Xenopus embryos and tissue culture cells, including CRC cells.

FAM83F induces axis duplication in Xenopus embryos through an interaction with
CK1α.
The activation of the canonical Wnt signalling pathway by ectopic expression of Wnt ligands and mediators in early Xenopus embryos causes axis duplication (20). Previously, we showed that injection of Xenopus embryos with FAM83G mRNA into a ventral blastomere at the four-cell stage induced secondary axis formation (6). The expression of mRNA in early Xenopus embryos is thus an efficient method for screening potential regulators of canonical Wnt signalling. Upon injection of axis-inducing mRNA, four possible axial phenotypes can result, including complete secondary axes, partial secondary axes, dorsalised embryos and those resembling wild-type (Fig. 1A). To test the impact of FAM83F on Xenopus embryos, 500 pg of mRNA encoding HA-tagged zebrafish Fam83fa, which closely resembles FAM83F in human and other species and was the only construct available at the time, was injected into a single ventral blastomere at the four-cell stage. Embryos were maintained until approximately stage 35, at which point we counted the embryos displaying each class of axial phenotype. HA-Fam83fa induced secondary axes in >60% of the Xenopus embryos ( Fig. 1B&C). A second zebrafish orthologue of FAM83F, Fam83fb, did not induce axis duplication (Fig. 1B), despite being more robustly expressed than Fam83fa, as shown by Western blot (Fig. 1C).
FAM83F interacts selectively with CK1α through its conserved DUF1669 domain and mutating the phenylalanine residues from a conserved F-X-X-X-F motif to alanine abolishes this interaction (1). In zebrafish Fam83fa, these two phenylalanine residues map at amino acid positions 275 and 279. Mutation of either F275 or F279 to an alanine prevented the induction of a secondary axis (Fig. 1D), as did the double mutant, Fam83fa F275/279A , despite all proteins being expressed, as shown by Western blot (Fig. 1E). This indicates that Fam83fa-CK1α binding is required for Fam83fa to induce axis duplication in Xenopus embryos. When we tested Fam83fa fragments with deletions at the C-terminus for their ability to induce axis duplication, both full-length fam83fa mRNA (HA-fam83fa 1-555aa ) and the DUF1669 domain fragment (HA-fam83fa 1-300aa ) induced secondary axes robustly (Sup. Fig.   2A&B). Interestingly, the C-terminal deletions HA-fam83fa 1-500aa , HA-fam83fa 1-400aa and HA-fam83fa 1-355aa induced secondary axes poorly, indicating that loss of the C-terminal portion of the protein affects Fam83fa structure or function.
Human FAM83F contains a protein prenylation motif, a conserved CAAX box sequence at the C-terminus in which the Cys residue is modified through an addition of either a geranylgeranyl or a farnesyl moiety (21). When FAM83F that has been overexpressed in HEK293 cells was isolated, cleaved with trypsin and subjected to mass-spectrometry, we identified tryptic peptides found to be farnesylated at Cys497 (Sup. Fig. 3). Farnesylation, a posttranslational modification, involves the addition of a 15-carbon farnesyl group to a Cterminal cysteine residue by farnesyltransferase; this plays a role in the regulation of proteinmembrane interactions and in signal transduction circuits (22). Zebrafish Fam83fa, which possesses the CIQS sequence at its C-terminus, is also predicted to be farnesylated (23).
Mutation in the human protein of the CAAX-box invariant cysteine to an alanine, creating FAM83F C497A , prevents farnesylation but injection of mRNA encoding this mutated protein induced secondary axis formation in Xenopus embryos in a similar manner to that of wildtype FAM83F (Sup. Fig. 2C). This indicates that farnesylation of the C-terminus of FAM83F is not required for its ability to activate canonical Wnt signalling in the Xenopus assays. Immunoprecipitation of GFP confirmed that GFP-FAM83F interacts with CK1α but no interaction was detected with GFP only or with GFP-FAM83A ( Fig. 2A). GFP-FAM83F F284/288A does not interact with CK1α, whilst the interaction between GFP-FAM83F D250A and CK1α is severely reduced compared with wild type GFP-FAM83F. The farnesyl-deficient mutant, GFP-FAM83F C497A , still maintains an interaction with CK1α ( Fig.   2A).
Fluorescence microscopy showed that GFP-FAM83F is present predominately at the plasma membrane in U2OS Flp/Trx cells, with some nuclear staining also observed (Fig. 2B). The farnesyl-deficient mutant GFP-FAM83F C497A was detectable only in the nucleus, suggesting that farnesylation of FAM83F directs its localisation to the plasma membrane. Interestingly, the two CK1α-binding deficient mutants, GFP-FAM83F D250A and GFP-FAM83F F284/288A , exhibited cytoplasmic and peri-nuclear localisation away from the plasma membrane and the nucleus. This suggests that the membrane and nuclear localisation of FAM83F is facilitated by its association with CK1α. Co-staining with an anti-CK1α antibody revealed overlapping localisation with GFP-FAM83F and GFP-FAM83F C497A , but not with GFP-FAM83F D250A and GFP-FAM83F F284/288A (Fig. 2B).
Canonical Wnt signalling activity can be measured using a dual luciferase reporter assay in which cells are transfected with a plasmid containing either wild-type TCF binding sites (TOPflash) or mutant TCF binding sites (FOPflash) upstream of a luciferase reporter (24). Tcell factor (TCF) is a co-transcriptional activator of β-catenin, so an increase in canonical Wnt signalling activity causes β-catenin to bind to TCF and induce luciferase expression and hence activity (25). Overexpression of GFP-FAM83F and GFP-FAM83F C497A significantly increased luciferase reporter activity in cells treated with control L-conditioned media compared with GFP controls (Fig. 2C). In contrast, overexpression of the CK1α-binding deficient mutants GFP-FAM83F D250A and GFP-FAM83F F284/288A did not increase luciferase activity under these conditions. Following addition of Wnt3A-conditioned medium, GFP-FAM83F and GFP-FAM83F C497A cell lines had significantly increased luciferase activity when compared with Wnt3A-treated GFP controls (Fig. 2C). Wnt3A-induced luciferase reporter activity in cells expressing the CK1α-binding deficient mutants GFP-FAM83F D250A and GFP-FAM83F F284/288A was substantially lower than in cells expressing GFP-FAM83F and GFP-FAM83F C497A (Fig. 2C). This suggests that FAM83F-induced Wnt reporter activity is mediated through its association with CK1α.

Endogenous FAM83F localises to the plasma membrane and interacts with CK1α.
To facilitate the study of endogenous FAM83F protein we screened multiple tissues and cell lines to identify cell lines with detectable levels of endogenous FAM83F protein. Tissuespecific expression of FAM83F from mouse tissue extracts revealed that FAM83F protein was detected in spleen, lung and gastrointestinal tissues, with the highest levels of FAM83F protein detected in the stomach, small intestine, large intestine and intestinal crypts (Fig.   3A). Similar assessment of a panel of routinely studied cell lines showed that FAM83F protein was detected in extracts from the mammary adenocarcinoma cell line MDA-MB-468, and the colorectal cancer cell line HCT116, but was undetectable in many other cell lines (Sup. Fig.4). Separately, we detected FAM83F protein in extracts from HaCaT keratinocytes as well as DLD-1 colorectal cells (Fig. 3A&B). Based on the abundance of FAM83F protein observed in gastrointestinal tissue extracts and colorectal cells, we proceeded with two colorectal cancer cell lines, HCT116 and DLD-1, for further investigation into the role of endogenous FAM83F in canonical Wnt signalling. By using CRISPR/Cas9 technology, we generated FAM83F knockout (FAM83F -/-) HCT116 (clones 1 and 2) and DLD-1 cells and also knocked in a GFP tag N-terminal to the FAM83F gene homozygously in both cell lines (HCT116 GFP/GFP FAM83F and DLD-1 GFP/GFP FAM83F) (Fig. 3B). Both knockouts and GFPknockins were verified by DNA sequencing and Western blotting ( Fig. 3B and Sup. Fig. 5).

Knockout of FAM83F reduces canonical Wnt signalling in colorectal cancer cells.
Constitutively active Wnt signalling is a hallmark of many colorectal cancers (CRCs) (19).
Mutations in the APC gene, which encodes a central component of the β-catenin destruction complex and facilitates the destruction of free cytoplasmic β-catenin, are among the most common in CRCs (26). DLD-1 cells express APC truncated at residue 1417, whilst HCT116 cells have a heterozygous mutation of β-catenin at S45, which prevents β-catenin from being phosphorylated and then ubiquitinated and degraded (27,28). Thus, constitutive activation of Wnt signalling in HCT116 and DLD-1 cells occurs through perturbations at different stages of the pathway, and the responses to Wnt3A-ligand stimulation are also likely to differ.
Canonical Wnt signalling activity can be measured by the transcript expression of Wnt target genes such as Axin2 (18) as demonstrated by the >2-fold increase in Axin2 mRNA expression following the addition of Wnt3A-CM to wild-type HCT116 cells (Fig. 4A). In both HCT116 FAM83F -/clones, the Wnt3A-induced increase in Axin2 mRNA abundance was significantly reduced compared with wild-type HCT116 cells. Neither wild-type DLD-1 cells nor DLD-1 FAM83F -/cells were responsive to treatment with Wnt3A-CM, but DLD-1 FAM83F -/cells had a slight but significant reduction in basal Axin2 mRNA abundance. In addition to the two colorectal cancer cell lines, we generated a FAM83F knockout in the osteosarcoma cell line U2OS (Sup. Fig. 7A). Endogenous FAM83F protein abundance is low in U2OS cells, with detection only possible after immunoprecipitation with an anti-FAM83F antibody. Canonical Wnt activity was determined using the dual luciferase assay (Sup. Fig.   7B) and the endogenous Wnt target gene Axin2 (Sup. Fig. 7C). U2OS wild-type and U2OS FAM83F -/cells respond to Wnt3A stimulus with increased luciferase activity and Axin2 mRNA abundance respectively, but the extent of both responses was significantly reduced in U2OS FAM83F -/cells compared to U2OS wild-type cells (Sup. Fig. 7B&C).

FAM83F acts upstream of the β-catenin destruction complex.
We sought to investigate where within the canonical Wnt/β-catenin signalling pathway FAM83F was acting. If the inhibition of Wnt signalling upon FAM83F loss can be restored by GSK-3 inhibitors, which prevent the phosphorylation and subsequent ubiquitin-mediated proteasomal degradation of β-catenin, this would imply that FAM83F acts upstream of the βcatenin destruction complex. As previously noted, Wnt3A-induced Axin2 mRNA abundance was lower in HCT116 FAM83F -/cells compared to wild-type HCT116 cells (Fig. 4A).
However, treatment of wild-type HCT116 and HCT116 FAM83F -/cells with 5 µM CHIR99021, a selective GSK-3 inhibitor, enhanced Axin2 mRNA abundance to a similar extent in both cell lines regardless of Wnt3A stimulation (Fig. 5A). In a similar assay, when U2OS wild-type and U2OS FAM83F -/cells were treated with 5 µM of CHIR99021, Axin2 mRNA abundance increased to a similar extent in both cell lines regardless of Wnt3A stimulation (Sup. Fig. 7C). These observations indicate that FAM83F acts to modulate Wnt signalling upstream of the β-catenin destruction complex.
These results suggest that FAM83F directs CK1α to the plasma membrane.

signalling.
To investigate the importance of membrane bound FAM83F in canonical Wnt signalling, we introduced a point mutation into the farnesylation motif of FAM83F, which should prevent membrane anchorage of FAM83F. Using CRISPR/Cas9, we sought to introduce a C497A mutation into the FAM83F gene in HCT116 cells. Unfortunately, we were only able to isolate a single heterozygous clone, HCT116 FAM83F WT/C497A . Nonetheless, when we separated wild-type HCT116, HCT116 FAM83F WT/C497A and HCT116 FAM83F -/-(cl.2) cell extracts into cytoplasmic, nuclear and membrane fractions, FAM83F protein was detected predominately in the nuclear fraction in HCT116 FAM83F WT/C497A cells, with a smaller proportion at the membrane, in a manner resembling the localisation detected with overexpression of GFP-FAM83F C497A (Fig. 6A and Fig. 2B). Quantification of membrane localised FAM83F and CK1α protein abundance showed a significant reduction in both proteins in HCT116 FAM83F WT/C497A cells compared to wild-type HCT116 cells, while a further reduction was noted in HCT116 FAM83F -/cells (Fig. 6B). CK1α IPs from wild-type HCT116 and HCT116 FAM83F WT/C497A cell extracts co-precipitated FAM83F but not from HCT116 FAM83F -/cells Fig. 6C), confirming that farnesylation of FAM83F does not affect CK1α binding. Wnt3Ainduced expression of Axin2 mRNA was slightly but significantly reduced in HCT116 FAM83F WT/C497A cells compared to wild-type HCT116 cells (Fig. 6D), indicating that membrane bound FAM83F is required for canonical Wnt signalling.

DISCUSSION
We know little about the function of FAM83F. In this study, we show that FAM83F mediates the canonical Wnt/β-catenin signalling pathway both in developing Xenopus embryos and in human cancer cells. We show that FAM83F is localised at the plasma membrane through farnesylation and directs CK1α to the plasma membrane. The FAM83F-CK1α complex appears to act upstream of the β-catenin destruction complex and we show that the interaction between FAM83F and CK1α and their membrane localisation is essential for driving Wnt signalling.
CK1 isoforms have been implicated in both positive and negative regulation of Wnt/β-catenin signalling (3). This suggests that the activity of CK1 isoforms is tightly regulated in a spatiotemporal manner. As key regulators of the CK1 isoforms, it is highly likely that the FAM83 proteins play a role. Any perturbation of the homeostatic balance of endogenous CK1 pools in cells, caused for example by overexpression of interacting proteins, is thus likely to disrupt coordinated roles of CK1 isoforms in Wnt signalling. This could explain the apparent contradictory observations we made when the overexpression of FAM83F C497A in Xenopus embryos and in cells caused an increase in canonical Wnt signalling, while the replacement of an allele of endogenous FAM83F with FAM83F C497A attenuated Wnt signalling. Expressing high levels of FAM83F C497A protein, which redirects much of the endogenous CK1α protein to the nucleus, potentially causes a reduction in the cytoplasmic pool of CK1α, thus removing inhibitory CK1α from the β-catenin destruction complex and thereby activating the canonical Wnt signalling pathway.
We have shown that FAM83F and FAM83G/PAWS1 both activate Wnt signalling, but FAM83F acts upstream of the β-catenin destruction complex, while FAM83G/PAWS1 acts downstream (6). The specific CK1α substrates within the canonical Wnt signalling pathway which mediate the effects of both FAM83F and FAM83G/PAWS1 are unknown.
Phosphorylation of β-catenin by CK1α at Ser45 is the most established regulatory role of CK1α within the Wnt pathway and is critical for priming subsequent GSK-3 phosphorylation, ubiquitination and degradation of β-catenin and thus inhibition of Wnt signalling (29). P120catenin is one of the few reported Wnt-dependent CK1α substrates located at the plasma membrane. The sequential phosphorylation of P120-catenin, at the adherens junctions by CK1ε and CK1α is required for the internalisation of the Wnt signalosome (17). The contradictory nature of the CK1 isoforms in Wnt signalling can be demonstrated by the CK1δ/ε isoforms which have been reported to phosphorylate multiple proteins within the Wnt signalling pathway including Dishevelled (30) and a co-transcriptional regulator, lymphoid enhanced binding factor 1 (Lef-1) (31). Interestingly, phosphorylation of Dishevelled at the membrane following Wnt stimulation promotes signalling whilst phosphorylation of Lef-1 in the nucleus is inhibitory, thus the same kinase can have opposing actions depending on subcellular localisation. It will be interesting to determine whether and how different FAM83 proteins coordinate the phosphorylation of these and other CK1 substrates to finetune Wnt signalling.
FAM83F has been implicated in oncogenesis in several cancers (12,14,32), and increased abundance of FAM83F protein is also associated with a more aggressive phenotype and poor prognosis in oesophageal carcinoma (33). However, a mechanistic explanation for these oncogenic effects has not been forthcoming. We show here that overexpression of FAM83F protein in cells increases both basal and ligand-dependent canonical Wnt signalling which may explain the reported increase in cell proliferation. FAM83F has also been implicated in the stabilisation of p53 protein, a crucial tumour suppressor, by decreasing p53 ubiquitination and degradation (34). This stabilisation is also apparent with mutant p53 protein, thus FAM83F may have a tumour suppressor or an oncogenic role depending on the p53 mutational status of the cell (34). Interestingly, CK1α has also been reported to influence p53 stabilisation through binding to the E3 ligases MDM2 (35) and MDMX (36) which inhibit p53 activity through ubiquitination and direct binding respectively. Therefore, these reported p53 effects may potentially be another function of the FAM83F-CK1α complex. The potential regulation of both canonical Wnt signalling and p53 activity indicates that further investigation of the FAM83F-CK1α complex is required and given the common dysregulation of both Wnt and p53 signalling during oncogenesis, the FAM83F-CK1α complex may be an attractive therapeutic target in cancer progression.

Plasmids
All constructs are available for request from the MRC-PPU reagents website (http://mrcppureagents.dundee.ac.uk) with the unique identifier (DU) numbers providing direct links to cloning strategy and sequence information. Sequences were verified by the DNA sequencing service, University of Dundee (http://www.dnaseq.co.uk). Constructs

Generation of L-and Wnt3A-conditioned media
Conditioned media were generated from mouse fibroblast L-cells and mouse fibroblast Lcells that stably overexpress Wnt3A. L-cells and L-Wnt3A cells were grown in DMEM in 15cm diameter dishes for 3 days before medium was filtered (45 µm) and stored as Lconditioned media (L-CM) and Wnt3A-conditioned media (Wnt3A-CM). Conditioned medium was diluted 50:50 in DMEM containing 10% FCS before use.

CHIR99021 (Tocris), a highly selective Glycogen Synthase Kinase 3 (GSK-3) inhibitor, was
added to cells at a concentration of 5 µM for 6 h prior to lysis.

Protein extraction from cells
Cells were washed in PBS twice, scraped in PBS and pelleted. For whole cell protein extractions, cell pellets were resuspended in total lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na3VO4, 1x complete EDTA-free protease inhibitor cocktail (Roche)). Lysates were incubated on ice for 30 min and vortexed regularly then clarified at 13000 rpm for 20 min. For cellular fractionation, cell pellets were washed in PBS twice, scraped in PBS, pelleted and then separated into cytoplasmic, nuclear and membrane lysates using a subcellular protein fractionation kit (Thermo Fisher Scientific) following the manufacturer's protocol. Briefly, cell pellets were resuspended in sequential buffers and clarified to isolate specific cellular compartments (cytoplasmic, nuclear, membrane and cytoskeletal).

Protein extraction from mouse tissue
Mouse tissue samples were obtained from a single male C57BL/6j mouse, which was obtained from the MRC-PPU after it was designated as surplus to current breeding requirements and culled by schedule one methods. Tissue samples were dissected, washed in PBS and snap frozen in liquid nitrogen. Frozen tissue samples were ground using a mortar and pestle until the sample was a fine powder which was then resuspended in PBS and pelleted. This pellet was resuspended in total lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na3VO4, 1x complete EDTA-free protease inhibitor cocktail (Roche)). Lysates were incubated on ice for 45 min and vortexed regularly before clarifying at 13000 rpm for 20 min.
For the isolation of single crypts from the mouse small intestine, a section of proximal small intestine was washed in PBS then cut into small fragments. Further washing of the fragments in PBS was completed until minimal contamination remained. The intestinal fragments were incubated in 3 mM EDTA for 30 min at 4ºC on a rotating wheel to dissociate the crypts. The EDTA was removed and the fragments gently washed in PBS before washing the fragments more aggressively in PBS to dislodge the crypts. Crypts were collected as a component of the supernatant. Crypts were pelleted at 800 rpm for 5 min. The pellet was resuspended in total cell lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% (v/v) Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na3VO4, 1x complete EDTA-free protease inhibitor cocktail (Roche)). Lysates were incubated on ice for 30 min with regular vortexing and clarified at

SDS-PAGE and Western blotting
The protein concentration of lysates was measured using the Pierce Coomassie Bradford protein assay kit (Thermo Fisher Scientific). Final protein concentrations were adjusted to 1-

Dual luciferase reporter assays
Cells were plated in 6-well plates and grown to approximately 70% confluence in complete culture medium. Cells were transfected with either 500 ng of Super TOPFlash (Addgene) or Super FOPFlash (Addgene) luciferase plasmids plus 10 ng of Renilla (Addgene) luciferase plasmids. Plasmids were diluted in 1 ml OptiMem (Gibco) and 20 µl of PEI (1 mg/ml) was added. The transfection mixture was vortexed then incubated for 20 min at room temperature before adding dropwise to a 6-well plate of cells in complete culture medium.        (C) qRT-PCR was performed using cDNA from U2OS wild-type and U2OS FAM83F -/cell lines following treatment with L-CM or Wnt3A-CM with or without 5 µM CHIR99021 for 6 h, and primers for Axin2 and GAPDH genes. Axin2 mRNA expression was normalised to GAPDH mRNA expression and represented as fold change compared to L-CM treated cells.
Data representative of six biological repeats with bar graph representing mean  standard error. Statistical analysis of (B) and (C) was completed using a students unpaired t-test and    Luciferase assay: U20S Flp/Trx GFP-FAM83 proteins