Protein proximity networks and functional evaluation of the Casein Kinase 1 g family reveals unique roles for CK1 g 3 in WNT signaling

The WNT/ b -catenin signaling pathway is evolutionarily conserved and controls normal embryonic development, adult tissue homeostasis and regeneration. Aberrant activation or suppression of WNT signaling contributes to cancer initiation and progression, developmental disorders, neurodegeneration, and bone disease. Despite great need and more than 40 years of research, targeted therapies for the WNT pathway have yet to be fully realized. Kinases are exceptionally druggable and occupy key nodes within the WNT signaling network, but several pathway-relevant kinases remain understudied and ‘dark’. Here we studied the function of the CSNK1 g subfamily of human kinases. miniTurbo-based proximity biotinylation and mass spectrometry analysis of CSNK1 g 1, CSNK1 g 2, and CSNK1 g 3 revealed numerous established components of the b -catenin- dependent and independent WNT signaling pathway, as well as novel interactors. In gain-of-function experiments leveraging a panel of transcriptional reporters, CSNK1 g 3 but not CSNK1 g 1 or CSNK1 g 2 activated b -catenin-dependent WNT signaling and the Notch pathway. Within the family, CSNK1 g 3 expression uniquely induced LRP6 phosphorylation. Conversely, siRNA-mediated silencing of CSNK1 g 3 alone had no impact on WNT signaling, though co-silencing of all three family members decreased WNT pathway activity. We characterized two moderately selective and potent small molecule inhibitors of the CSNK1 g family. These inhibitors and a

Here we defined protein-protein proximity networks for the CK1g family and functionally evaluated their contribution to a small panel of disparate signaling pathways.Each CK1g family member co-complexed with various WNT components, including b-catenin and core members of the planar cell polarity complex.Surprisingly, only CK1g3 activated b-catenin-dependent WNT signaling when overexpressed.Our experiments reveal a family-specific role for CK1g3 in (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.
The copyright holder for this preprint this version posted November 30, 2021.; https://doi.org/10.1101/2021.11.30.470617 doi: bioRxiv preprint promoting phosphorylation of LRP6 within the WNT signalosome.Additionally, we characterized two pan-CK1g specific inhibitors in vitro and in cell-based assays.Inhibition of CK1g impairs the ability of WNT signaling to be fully activated by WNT ligand.This finding was extended and confirmed with a kinase dead CK1g3 mutant, demonstrating that CK1g3 kinase activity is required to activate WNT signaling.Overall, this work establishes that CK1g3 positively regulates WNT signaling.

Physical and functional evaluation of the CK1g family
To better understand each member of the CK1g family, we defined their protein-protein proximity networks and functionally evaluated their impact on signal transduction.We used the miniTurbo (mT) promiscuous biotin ligase to comprehensively map CK1g proximal proteins, which requires a shorter biotin labeling window compared to BirA* (43,44).Specifically, in the presence of exogenous biotin, a protein of interest fused to the mT biotin ligase will result in biotinylation of surface exposed lysine residues on proximal proteins, enabling their affinity purification with streptavidin and identification by Western blot (W.blot) or mass spectrometry (MS).HEK293T stably expressing mT-CK1g1, mT-CK1g2, mT-CK1g3 were treated with biotin for varying amounts of time before W. blot analysis for biotinylated proteins and for a V5 epitope, the latter of which is cloned in frame with mT (Fig. 1B).Streptavidin affinity purification followed by W. blot analysis confirmed proximal complex formation between CK1g family members and WNT pathway proteins LRP6, AXIN1 and b-catenin (CTNNB1) (Fig. 1C).We next used MS across biological triplicate experiments to identify and quantify proximal proteins to each of the CK1g family members.Resulting identifications were probabilistically scored against controls (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.
The copyright holder for this preprint this version posted November 30, 2021.; https://doi.org/10.1101/2021.11.30.470617 doi: bioRxiv preprint with SAINT and then further ranked with CompPASS before visualization (Table S1, sheet B, C) (45)(46)(47).Heat map representation of protein abundance revealed concordant proximal networks across the CK1g family (Fig. 1D).Of 545 high confidence protein interactions for CK1g1, CK1g2, CK1g3, 326 proteins were seen with all subfamily members.b-catenin, ZDHHC8 (Zinc Finger DHHC-Type Palmitoyltransferase 8), and PIK3R3 (Phosphoinositide-3-Kinase Regulatory Subunit 3) were the most abundant proximal proteins in all three networks.Several core components of the b-catenin dependent and independent signaling pathway were also identified, such as DVL, APC, CELSR2, and VANGL (Fig. 1D, green track).To further explore these networks and their connectivity to the WNT pathway, we annotated all proteins for functional contribution to b-catenin-dependent transcription.From four recently published independent genetic screens, we identified a subset of genes as functionally impactful to b-catenin-dependent transcription (Table S2) (48)(49)(50)(51).Integration of these data with the CK1g proximity networks revealed 32 CK1g proximal proteins that when silenced or CRISPR-deleted impacted b-catenindependent transcription.(Fig. 1D, red track).Last, we visualized the data as a network to examine high confidence prey-prey interactions and subcellular localization (Fig. 1E).The majority of the prey proteins are known to localize to the plasma membrane.
To confirm the MS results, and to test the impact of WNT3A stimulation, we analyzed the CK1g affinity purified proximal proteins by W. blot analysis (Fig. 1F).CK1g2 and CK1g3 robustly interacted with b-catenin-dependent WNT signaling components (LRP6, DVL2, AXIN1 and bcatenin), while CK1g1 interacted with these proteins but to a lesser extent.WNT3A stimulation increased pLRP6 S1490 in all three CK1g pulldowns.Additional WNT3A-dependent changes were not observed.We also confirmed the interaction between the CK1g family and proteins involved in b-catenin independent signaling, CELSR2 and PRICKLE1 (Fig. 1F).
(which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.We next functionally evaluated each member of the CK1g subfamily across a small panel of engineered pathway-specific transcriptional reports.Transient transfection and over-expression of each CK1g subfamily member in HEK293T cells revealed relationships to the WNT (BAR), NOTCH, TGFb (SMAD), NRF2 (hQR41), and RA (retinoic acid, RAR), and TNFa (NFkB) signaling pathways (Fig. 2A) (27,(52)(53)(54).Surprisingly, only CK1g3 activated the WNT reporter, with minimal effects from CK1g1 and CK1g2.Though statistically significant compared to the negative control, none of the CK1g family members strongly regulated the other signal transduction reporters.To further establish the CK1g3 selectivity for b-catenin-dependent activation of the BAR reporter, we performed a dose response overexpression experiment.In contrast to CK1g3, overexpression of CK1g1 and CK1g2 did not activate BAR (Fig. 2B, C).
Similar to previously published reports, in the presence of WNT3A conditioned media (CM), CK1g1 and CK1g2 activated WNT signaling, albeit not to the same extent as CK1g3 (Fig. 2D) (31,32).Additionally, co-overexpression of LRP6 with either CK1g1, CK1g2, or CK1g3 resulted in a significant activation of BAR compared to control and compared to LRP6 overexpression alone (Fig. 2E) (31,32).Together, these data establish CK1g3 as proximal to numerous WNT pathway proteins and as an activator of b-catenin-dependent WNT signaling.CK1g1 and CK1g2 share WNT pathway proximal proteins, but in the absence of WNT3A or co-overexpression of LRP6, they do not regulate b-catenin-dependent transcription.

CK1g3 requires WNT components to activate WNT signaling
To evaluate which WNT pathway proteins are required for CK1g3 activation of b-catenindependent transcription, we studied established inhibitors of WNT signaling and knockout cell (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.
The copyright holder for this preprint this version posted November 30, 2021.; https://doi.org/10.1101/2021.11.30.470617 doi: bioRxiv preprint lines.While overexpression of CK1g3 activated BAR in the absence of WNT ligand, treatment with C59, a PORCN inhibitor (required for WNT ligand palmitoylation and secretion), blocked this activation, which supports a role for autocrine WNT signaling in HEK293T cells (Fig. 2F) (55).We next tested the requirement of LRP6 and DVL.In contrast to HEK293T wildtype cells, overexpression of CK1g3 in LRP5/6 double knockout (DKO) HEK293T cells did not activate WNT signaling (Fig. 2G).HEK293T cells lacking DVL1, DVL2 and DVL3 did not respond to WNT3A CM, as expected (56).Surprisingly, CK1g3 overexpression increased BAR activity in DVL KO cells, but to a lesser extent that in DVL wild type cells (Fig. 2G).These data indicate that WNT3A ligand secretion and expression of LRP5/6 are required for CK1g3-driven activation of b-catenin-dependent transcription; the DVL proteins are involved but are not absolutely required.
CK1g3 overexpression robustly increased LRP6 phosphorylation at T1479, as well as S1490, in both the absence and presence of WNT3A.Together these data suggest that in contrast to CK1g3, CK1g1 and CK1g2 do not activate WNT signaling in the absence of exogenous WNT3A ligand and do not induce LRP6 phosphorylation at T1479.

CK1g3 kinase activity was required for maximum activation of WNT signaling
We next performed loss-of-function studies to determine if the CK1g family was required for WNT3A-driven b-catenin-dependent transcription.First, we generated a CK1g3 kinase dead mutant (CK1g3-K72A).In contrast to CK1g3-WT, expression of kinase-dead CK1g3-K72A in HEK293T cells did not regulate the activity of the BAR transcriptional reporter (Fig. 4A).In the presence of WNT3A CM, kinase dead CK1g3-K72A expression suppressed BAR activity as compared to wild type CK1g3 or control (Fig. 4A).In an orthogonal experiment, CK1g3 or CK1g3-K72A were expressed in HEK293T cells stably harboring a BAR-GFP fluorescent reporter.Live cell imaging over three days revealed that CK1g3-K72A suppressed WNT3A-driven b-catenindependent transcription (Fig. 4B).Last, we tested whether expression of the CK1g3 kinase dead mutant impacted the phosphorylation of LRP6.CK1g3 overexpression resulted in LRP6 phosphorylation at both S1490 and T1479 sites following WNT3A CM exposure (Fig. 4C-E).In contrast, overexpression of CK1g3-K72A suppressed phosphorylation of LRP6 at both sites.These data demonstrate that kinase activity of CK1g3 is required for phosphorylation of LRP6 and activation of the WNT pathway.
To extend this finding, we tested the effect of CK1g knockdown on BAR reporter activity.
First, we confirmed knockdown by qPCR of two non-overlapping siRNAs for CK1g3, as well as pooled siRNAs targeting CK1g1/2/3 or CK1g1/2 (Fig. 5A).Transfection of the indicated siRNA into HEK293T cells stably harboring BAR revealed that knockdown of CK1g3 alone or a pool of CK1g1/2 did not significantly inhibit WNT activation.However, knockdown of CK1g1/2/3 suppressed WNT activation by approximately 50% (Fig. 5B).Second, we examined b-catenin (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.

CK1g3 inhibition impairs activation of WNT signaling
To complement the genetic study of CK1g family within the WNT pathway, we sought CK1g chemical inhibitors.Molecular tool compounds that target the larger CK1 family of kinases have been studied within WNT signaling in the past (58).However, given the multiple positive and negative roles of various CK1 family members in the pathway, and the lack of selectivity for most tool compounds, the results have been inconclusive (59).We leveraged a library of modestly selective kinase inhibitor tool compounds from the University of North Carolina-Structural Genomics Consortium (UNC-SGC), including a 5-substituted indazole with previously characterized inhibition activity for CK1 (60).To further characterize the lead candidate, we sent AKI00000062a (Fig. 6A, B) (referred to as AKI throughout) for kinome-wide profiling against 403 WT kinases at DiscoverX at 1 μM.This compound demonstrated modest selectivity, with a calculated S10(1 μM) = 0.042, corresponding to 17 kinases with a Percent of Control (PoC) < 10 at 1 μM (Fig. 6C).GSK3α and GSK3β were among those kinases potently inhibited, as well as all three CK1g kinases (Fig. 6D).To confirm this, we collected the corresponding enzymatic assay activity for all three CK1g family members and found AKI to inhibit all three enzymes with IC50 values ≤275 nM (Fig. 6B).Finally, a cellular target engagement assay (NanoBRET) was used to measure the potency of enzymatic inhibition for CK1g2 in live cells.We observed modest suppression of CK1g2 in the cell-based system: IC50 = 991 nM (Fig. 6E).
The kinome-wide profiling of AKI demonstrated potent inhibition of GSK3α and GSK3β; FP1 did not inhibit GSK3α and GSK3β.Using the GSK3β in-cell NanoBRET assay, we found that AKI potently engages GSK3β in cells with an IC50 = 7.14 nM (Fig. 8A).AKI was significantly more potent on GSK3 kinases than on CK1g2.In contrast, FP1 was inactive up to 10 μM in the GSK3β NanoBRET assay (Fig. 8B).As a control, APY69, validated as active in the GSK3 NanoBRET assay by Promega, was included in both the CK1g2 and GSK3β NanoBRET assays.
APY69 had an IC50 = 0.23 nM in the GSK3β NanoBRET assay, and was inactive in the CK1g2 NanoBRET assay.To confirm that AKI inhibits GSK3b, we compared it in a BAR assay with the (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.
Finally, we tested whether FP1 impacted CK1g3 activation of WNT signaling in a BAR assay.We overexpressed CK1g3 and treated with increasing doses of FP1, an observed a decrease in BAR activity (Fig. 8D).In agreement with this, treatment of cells with FP1 and WNT3A CM resulted in diminished b-catenin stabilization and phosphorylation of LRP6 at site S1490, as compared to the DMSO control treated cells (Fig. 8E, F, G).Importantly, when cells were treated with FP1 then exposed to WNT3A, LRP6 phosphorylation at T1479 did not increase (Fig. 8E, H).

Discussion
Protein kinases govern information flow though cellular signaling networks, and ultimately impact all of cell biology.Their exceptional druggability and centrality within human disease networks has supported decades of research and therapeutic development (64).However, attention to kinases has not been evenly spread across the protein class.Recently, the NIH Illuminating the Druggable Genome consortium has identified 162 of the most understudied 'dark' kinases (33).In this study, we characterize three of these 'dark' kinases: CK1g1, CK1g2 and CK1g3.Through unbiased protein-protein proximity network derivation and a functional evaluation across a small focused panel of transcriptional reporters, we connect the CK1g family to the b-catenin-dependent WNT signaling pathway.Unexpectedly we found that the CK1g3 family member is uniquely active within b-catenin-dependent WNT signaling.All three family members co-complex with both b-catenin-dependent and independent WNT machinery.All three family members activate bcatenin-dependent transcription in the present of exogenous WNT3A ligand or in LRP6 overexpressing models.CK1g3 appears particularly potent in its activity as it activates b-catenin-dependent transcription in the absence of exogenous WNT3A ligand or LRP6 overexpression.We show though siRNA-based silencing and small molecule inhibitors that functional redundancy likely exists within the CK1g family, as CK1g3 suppression alone had minimal impacts on WNT signaling.Beyond connections to the WNT pathway, the proximity networks, functional data and molecular probes offer a deep resource for the kinase community.
Defining proximity networks for this subfamily of understudied kinases provides insight into their cellular functions.Identification of core WNT signaling components, both b-catenin dependent and independent, further establishes the CK1g kinase subfamily as WNT signaling components.Many of the identified proximal proteins localize to the plasma membrane, which is consistent with the reported localization of the CK1g family, likely due to palmitoylation of their C-terminus (31).One interesting protein identified as abundant and proximal to all three family members is ZDHHC8, a palmitoyltransferase with strong disease connections to schizophrenia (65).In Drosophila, the ZDHHC8 ortholog palmitoylates Scribble, which is a key component of the planar cell polarity pathway and has been shown to negatively regulate WNT/b-catenin signaling in human cell lines (66)(67)(68).It will be important to determine if ZDHHC8 is responsible for palmitoylation and membrane localization of the CK1g family, and reciprocally if CK1g regulates ZDHHC8 activity or localization.A second interesting and abundant discovery across all three family members is PIK3R3 (phosphoinositide-3-kinase regulatory subunit 3).This regulatory subunit of class 1a PI3K binds phosphorylated tyrosine residues to signal downstream of receptor tyrosine kinases and cytokine receptors (69).PI3K-AKT signal transduction is among the most frequently activated and functionally importance pathways in cancer (70).Whether and how CK1g activity influences PI3K signaling and biology is exciting and warrants testing.
Broadly, interrogation of kinase proximity networks like those presented for CK1g offer a powerful resource for kinase substrate prediction and definition.Notably, however, proximity-based networks have caveats that must be considered, including: 1) the approach is exquisitely sensitive to over-expression artifacts, 2) rigorous probabilistic scoring approaches against appropriate controls are needed to illuminate true positives from false positive, 3) as will all of proteomics, absence of identification does not mean absence within the sample but rather below the level of detection; and 4) proximity networks do not identify direct binding relationships.For kinases, motif enrichment queries and integration with phospho-proteomic data can seed experiments for the identification of direct kinase substrates (53).
Our observed physical and functional connections to b-catenin-dependent and independent signal transduction is unsurprising.Previous studies demonstrated that CK1g1 activated b-catenindependent WNT signaling in the presence of exogenous WNT3A ligand, or with cooverexpression of LRP6 or DVL (31,32).Importantly, our data replicate these results (Fig. 2D,   E).Also similar to our results, without co-overexpression the authors observed minimal activation of WNT signaling with CK1g1 overexpression (31).However, subsequent studies to further articulate the role of the larger CK1g family in WNT signaling has been lacking until now, including loss of function characterizations.One of the surprises of this work is the family-unique ability of CK1g3 to activate b-catenin-dependent transcription in the absence of exogenous WNT3A ligand or co-overexpression of LRP6.Interestingly though, the C59 inhibitor experiment (Fig. 2F) establishes that CK1g3 requires an autocrine loop of WNT signaling to impact b-catenin.
Loss-of-function studies for the CK1g family within the WNT pathway have not been reported, perhaps due to modest or lacking effect sizes.Functional redundancy within the WNT signaling pathway is well-acknowledged, and indeed our data support redundancy for the CK1g family (71)(72)(73).This work and previous studies have shown CK1d/e and CK1g can function to phosphorylate LRP6 (31,32).In our siRNA-based loss-of-function studies, knock down of all three CK1g subfamily members was required to observe maximal suppression of the WNT pathway (Fig. 5).Future studies using CRISPR genetic knockouts are needed, as siRNA-based approaches are incomplete, which is particularly problematic for catalytic proteins like kinases.That said, it is possible that with loss of the full CK1g subfamily, CK1d/e may compensate to phosphorylate LRP6.Beyond genetics, we evaluated two tool chemical inhibitors for the CK1 family.The AKI compound also targeted GSK3b which complicates the study of CK1g in WNT signaling (Fig. 6).
The other compound, FP1, did not target GSK3b, but suffered lack of specificity within the CK1 subfamily (Fig. 7).FP1 also targets CK1d/e, as well as CK1a, which governs WNT signaling through phosphorylating b-catenin within the destruction complex (Fig. 7).However, FP1 holds value as a research tool when studying proximal events in signalosome formation.
The highly conserved nature of the CK1g subfamily makes the unique function of CK1g3 activation of WNT signaling intriguing and raises the question why CK1g1 and CK1g2 do not activate WNT signaling with the same potency.Structurally, the CK1g subfamily is similar with one major exception -CK1g3 has a 33 amino acid insertion close to the C-terminal tail (Fig. 1A).
Five b-sheets on the c-terminus of the CK1g subfamily proteins create an open crescent moon shape (74).However, in CK1g3 the 33 amino acid insertion creates an additional, short b-sheet within an intrinsically disordered region (IDR) (74,75).This b-sheet shifts the C-terminal structure so that the b-sheet lies in an opening of the crescent moon shape (74).Additionally, IDRs allow the same peptide sequence to interact differently with different functional outcomes and can have a high number of post-translational modifications (75,76).The functional implications of the altered structure and the role the IDR plays in CK1g3 have not been evaluated.Studies which delete the 33 amino acid insertion are necessary to fully understand the functional and mechanistic implications.
Illuminating understudied proteins, particularly kinases, is critical to furthering our understanding of several signaling cascades.This work highlights the need to understand the role these understudied kinases play in signaling pathways and disease states.Full characterization of the understudied kinases presents a previously untapped pool of therapeutic targets to treat a variety of human diseases and improve patient outcome.

Methods and Materials
Cell lines and tissue culture: All cells were cultured at 37 °C and 5% CO2.The following cells were obtained from American Type Culture Collection (ATCC, Manassas, VA): HEK293T/17 (human fetus, cat#CRL-11268), RKO (human, gender N/A, cat# CRL-2577), Lcells (mouse male, cat# CRL-2648), WNT3A expressing Lcells (mouse male, cat# CRL-2648).All cells were grown in DMEM with 10% FBS.Each cell line was tested for mycoplasma contamination and passaged no more than 20 passages from the original ATCC stock.
WNT3A conditioned media: Conditioned media was collected as described by ATCC.Briefly, WNT3A and control Lcells were grown to 100% confluency before fresh media was added conditioned for 48 hrs, and then collected.Immunoblotting: Standard W. blotting techniques were utilized and performed as previously described (27,52).Briefly, cells were lysed in RIPA lysis buffer, flash frozen on dry ice, then high speed cleared for 10 mins at max speed.Protein concentration was determined by a BCA assay following kit specifications.Samples were balanced and NuPAGE 4x SDS loading buffer (ThermoFisher, Waltham, MA, cat# NP0007) containing DTT was added to each sample and heated for 10 mins at 70 °C.Samples were run on a NuPAGE 4-12% Bis-Tris Protein Gel (ThermoFisher, cat# NP0321), then transferred to a nitrocellulose membrane.The membrane was then blocked and incubated overnight with primary antibody at 4 °C with rotation, washed with TBST, incubated with secondary antibody, washed with TBST and imaged.Images were taken using a LiCOR Odyssey CLx imager and quantified with LiCOR software (LiCOR Biosciences, Lincoln, NE).All antibodies were used at a concentration of 1:1000, with the exception of loading controls, which were used at 1:10,000.All primary antibodies utilized are as follows: b- Real-Time Quantitative PCR: Total RNA was extracted from RKO and HT1080 cells using Invitrogen PureLink RNA mini kit (Life Technologies, Cat#12183025) following the manufacture's manual.The AXIN2 and NKD1 primers are described in (Walker et al 2015).All Affinity pulldowns: Cells were incubated for indicated time with 50mM biotin, washed 3x in cold PBS, then lysed in RIPA lysis buffer, flash frozen on dry ice, then thawed and incubated on ice for 30 mins.Lysates were then sonicated for 20 seconds in 5 second pulses.Samples were then high speed cleared and protein concentration was determined by BCA.50uL of bead slurry per sample was washed 5x in RIPA lysis buffer, then incubated with lysate at 4 °C with rotation for either 1 hr for W. blot analysis or overnight for mass spectrometry analysis.Beads were then washed 5x with RIPA lysis buffer.For W. blot analysis samples were then eluted in a 1:1:1 mixture of 1 M DTT, 4x SDS loading buffer and RIPA lysis buffer and heated at 70 °C for 10 mins.
Streptavidin beads were resuspended in 100uL of 0.5% RapiGest and vortexed.1M Dithiothreitol (DTT) in ABC was added to the sample for a final concentration of 5mM and samples were heated at 60°C for 30 mins.After allowing samples to cool to room temperature, chloroacetamide was added to each sample for a final concentration of 15mM and incubated in the dark for 20 minutes at room temperature.The samples were then centrifuged at 400 x g for 2 mins at room temperature to remove the supernatant containing protein from the streptavidin beads.Mass spectrometry data acquisition: Trypsinized peptides were separated via reverse-phase nano-HPLC using an RSLCnano Ultimate 3000 (Thermo Fisher Scientific).The mobile phase consisted of water + 0.1% formic acid as buffer A and acetonitrile + 0.1% formic acid as buffer B.
Peptides were loaded onto a µPACᵀᴹ Trapping column (PharmaFluidics) and separated on a 200 cm µPACᵀᴹ column (PharmaFluidics) operated at 30°C using a 100 min gradient from 2% to 25% buffer B, followed by a 20 min gradient from 25% to 35% buffer B, flowing at 300 nL/min.Mass spectrometry analysis was performed on an Orbitrap Eclipse (Thermo Fisher Scientific) operated in data-dependent acquisition mode.MS1 scans were acquired in the Orbitrap at 120k resolution, with a 250% normalized automated gain control (AGC) target, auto max injection time, and a 375-1500 m/z scan range.Both the linear ion trap and the Orbitrap were used for MS2 scans.MS2 targets with a charge of +2 or +3, and >90% precursor fit at either a 0.8 m/z or 0.4 m/z-wide isolation window were fragmented with by collision induced dissociation (CID) at 35% collision energy and scanned in the linear ion trap at the widest window width that passed the thresholds.
All remaining MS2 targets with charge from +2 to +6, >5e4 intensity, and >10% precursor fit at isolation widths of 1.6 m/z, 0.8 m/z, or 0.4 m/z wide were fragmented with higher-energy collision dissociation at 30% collision energy and scanned in the orbitrap at 15k resolution at the widest window width that passed the thresholds.MS2 AGC targets and maximum injection times were set to standard and auto for their respective analyzers.Dynamic exclusion was set to 60 seconds.
Acquisition was performed with a 2.7 second cycle time.
Mass spectrometry data processing: Raw MS data files were processed by MaxQuant (version 1.6.17.0) (78) using the UniProtKB/SwissProt human canonical sequence database (downloaded Aug. 2019) (79).To facilitate comparison to another 44 bait experiments performed within our laboratory, the files were processed simultaneously with these raw files.The following parameters were used: specific tryptic digestion with up to two missed cleavages, fixed carbamidomethyl modification, variable modifications for protein N-terminal acetylation, methionine oxidation, match between runs, and label-free quantification.Only unique peptides were used for protein quantification due to the high tryptic peptide overlap between CK1g1, CK1g2, and CK1g3.Baits were assigned fractions numbers >1 away from each other to enable match-between-runs within replicates, but not across different baits.Prey proteins were filtered for high-confidence physical interactions and proximal proteins using SAINTexpress (v3.ClusterProfiler using default parameters (80,81).

Synthesis
were purchased from ATCC.These cells were grown in Dulbecco's Modified Eagle's medium (DMEM, Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS, Corning).They were incubated in 5% CO2 at 37°C and passaged every 72 hours with trypsin and not allowed to reach confluency.Constructs for NanoBRET measurements of CK1g2 (NLuc-CK1g2) and GSK3β (NLuc-GSK3β) were kindly provided by Promega.NanoBRET assays were carried out as described previously (83).Preferred NLuc orientations, both N-terminal in this case, are indicated in parentheses after each construct.Assays were carried out in dose-response as described by the manufacturer using 0.5 μM of tracer K-10 for CK1g2 and 0.063 μM of tracer K-8 for GSK3β.Two biological replicates were executed with two technical replicates to produce the graphs with error bars shown in Fig. s 6E, 7E, 8A, 8B.Tracer titration curves for these kinases that motivated our tracer selection and working concentration can be found on the Promega website.
(which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.

CK1γ3
Hit in WNT screens WNT-related    (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.
The copyright holder for this preprint this version posted November 30, 2021.; https://doi.org/10.1101/2021.11.30.470617 doi: bioRxiv preprint   (which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.
The copyright holder for this preprint this version posted November 30, 2021.
not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.

( 3 '
which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.otherprimers were designed using the National Center for Biotechnology Information's (NCBI) Primer-BLAST platform.Primer sequences are as follows: CK1g1 Forward 5Reverse 5' GCCCTCCAATCAGTCTTCTG 3'.Before reverse transcription, RNA was quantified using a Nanodrop One (Thermo, Waltham, MA), and 1 μg of RNA was used to generate cDNA with the iSCRIPT Clear kit (Bio-Rad, Hercules, CA, cat# 170-8891).For the qRT-PCR, PowerUP SYBR Green (Thermo, cat# A25741) was used, and data were analyzed on a QuantStudio 5 Real Time PCR machine (Applied Biosystems, Foster City, CA).ΔCT values were normalized to housekeeping gene RPL13a.
The supernatant was transferred to a new tube and 2.5 µg of mass spectrometry grade trypsin was added to each sample.The samples were incubated at 37°C overnight, then Hydrochloric acid (HCl) was added to the sample at a final concentration of 250mM and incubated for 45 mins at 37°C to deactivate the trypsin.Pierce C18 spin columns (ThermoFisher, cat#89870) were placed in a 2 mL centrifuge tube, the column was activated by adding 200 µL of 100% acetonitrile (ACN) and centrifuged at 3000 xg for 2 mins.Columns were equilibrated by adding 200 µL of 0.5% Trifluoroacetic acid (TFA) and centrifuged at 3000 xg for 2 mins.This step was then repeated.The sample was resuspended in 200 µL of 0.5% TFA, then loaded onto the column and centrifuged for 5 mins at 1000 xg.The sample was reloaded onto the column and centrifuged a second time for 5 mins at 1000 xg.The column was washed twice with 200 µL of 5% ACN/0.5% TFA and centrifuged for 2 mins at 3000 xg.The samples were eluted by adding 50 µL 70% ACN and centrifuged for 5 mins at 1000 xg.A second elution step was performed by adding 50 µL 70% ACN and centrifuged for 2 mins at 3000 xg.Following a C18 clean up, an ethyl acetate clean-up was performed.The sample was resuspended in 100 µL of 0.1% TFA, then 1 mL of water saturated ethyl acetate was added to each sample, vortexed and centrifuged at max speed for 5 mins.The upper layer of ethyl acetate was removed.This process was repeated for a total of 3 times.Samples were then dried down in the speed vac and resuspended in 25 µL of 98/2 Buffer A (Water + 0.1% FA)/Buffer B (ACN + 0.1% FA).

CK1g family enzymatic assays:
Eurofins kinase enzymatic radiometric assays were carried out at the Km = ATP in dose-response (9-pt curve in duplicate) for each CK1g kinase listed in Fig.s 6B and 7B.Details related to the substrate used, protein constructs, controls, and assay protocol /www.eurofinsdiscoveryservices.com.Kinome-wide profiling: The scanMAX assay platform was used to assess the selectivity of FP1-24-2 and AKI00000062a at 1 µM at Eurofins DiscoverX Corporation.As described previously, this commercial assay platform screens against 403 WT human kinases in binding assays and provides PoC values (82).All WT kinases that demonstrated PoC values < 10 are plotted on the respective kinome trees in Fig.s 6C, 7C.These same WT kinases with PoC value < 10 are tabulated in Fig.s 6D, 7D.

Figure 1 -
Figure 1 -CK1γ proximity-based interaction networks identify WNT components.(A) Schematic of CK1γ family highlighting conserved amino acids, sequence similarity between family members and mutations.(B) W. blot analysis of stable HEK293T cells expressing mini-Turbo CK1γ or V5 control incubated with 50 μM of biotin over a time course (untreated, 15 min, 30 min, 45 min, 1 hr, 2 hr).(C) W. blot analysis of streptavidin affinity pulldown from stable HEK293T cells expressing mini-Turbo CK1γ or V5, treated as indicated.(D) Heat map demonstrating changes in log2(LFQ intensity) of protein interactions of CK1γ subfamily AP-MS.Red track indicates hits in previously published WNT screens (Table S1, S2). Green track indicates proteins involved in WNT/β-catenin-dependent and independent signaling.(E) Affinity purification mass spectrometry protein-protein interaction networks for CK1γ1, CK1γ2, CK1γ.HEK293T cells stably expressing mT-V5-CK1γ construct were treated with 50 μM biotin for 30 mins.Blue circle indicates localization to the plasma membrane, and orange circles represent alternate subcellular localizations.Grey lines represent bait-prey interactions, while purple lines indicate prey-prey interactions determined by BioGrid.(F) W. blot validation of mT-CK1γ protein-protein interaction networks treated with biotin and WNT3A CM for 30 mins.

(Figure 2 -
Figure 2 -Low throughput reporter screen identifies CK1γ3 as an activator of WNT signaling.(A) HEK293T cells transfected for 30 hrs with 100 ng of indicated control (β-catenin, NRF2, NCID) construct or pGUS control or indicated CK1γ construct then 40 ng TK-Renilla and 40 ng Response Element-Luciferase.For treatments (TGFβ 10 ng/mL, TNFα 10 ng/mL, RA), cells were treated for 6 hrs (TGFβ, TNFα) or 16 hrs (RA).Inset W. blot demonstrates overexpression for BAR luciferase assay.Each condition was normalized to pGUS.Error bars represent standard deviation and statistics are compared to pGUS control.(B) BAR luciferase assay from HEK293T B/R (stable BAR-firefly luciferase, TK-Renilla expressing cell line) cells transfected for 24 hrs with indicated doses of construct.Statistics are compared to pGUS control.(C) Confirmation of overexpression for luciferase assay in panel B. (D) HEK293T B/R cells were transfected with the indicated constructed for 14 hrs, then treated with Lcell or WNT3A CM for 16 hrs.All statistics are compared to Flag-Control Lcell sample.(E) HEK293T B/R cells were transfected with the indicated constructed for 24 hrs then analyzed by luciferase assay.Statistics are compared to pGUS control.(F) HEK293T B/R cells were transfected with the indicated constructed for 14 hrs, then treated with 10 μM C59 for 18 hrs.(G) Wildtype, LRP5/6 DKO, DVL TKO HEK293T cells transfected with 20 ng BAR-luciferase, 10 ng TK-Renilla and 70 ng of indicated constructs for 14 hrs, then treated with either Lcell CM or WNT3A CM for 18 hrs.Statical significance presented are compared to Lcell treated Flag-Control cells for each cell type.All panels *** p<0.0005, ** p<0.005, * p<0.05 and all data are representative of biological triplicates, unless otherwise noted.Error bars represent S.E.Bars represent average Firefly/Ren (RFU) from 3 technical replicates +/-standard error (S.E.), unless otherwise stated.

(Figure 3 -Figure 4 -
Figure 3 -Overexpression of CK1γ3 activates WNT signaling by increasing LRP6 phosphorylation.(A) HEK293T cells were transfected with the indicated constructed for 14 hrs, then exposed to WNT3A CM for indicated time and analyzed by W. blot.(B, C) Box-and-whisker plot quantification of W. blot from panel A for indicated time points, pLRP6 T1479 (panel B) and pLRP6 S1490 (panel C) across 4 replicates, normalized to total LRP6.For all panels: *** p<0.0005, ** p<0.005, * p<0.05 and all data are representative of biological quadruplicates.All statistics are compared to Flag-Control untreated sample.

(Figure 5 -
Figure 5 -Loss of CK1γ1/2/3 impairs WNT activation.(A-C) RKO B/R cells were transfected with indicated siRNA or pooled siRNA for 48 hrs, then split for the following experiments: 24 hrs post-split gene expression was analyzed by RT-qPCR (panel A), 6 hrs post-split cells were treated with WNT3A for 18 hrs then analyzed by Luciferase assay (panel B), 24 hrs post-split cells were exposed to a WNT3A time course, harvested at the indicated time post WNT3A and analyzed by W. blot.For panel A all statistics are compared to siControl expression for the specific gene.For panel B all statistics are compared to siControl WNT3A treated cells.(D-F) Box-and-whisker plot quantification of W. blot from panel E, β-catenin (panel D), pLRP6 T1479 (panel E) and pLRP6 S1490 (panel F) across 4 replicates, normalized to ACTIN.Statistics are compared to siControl untreated, unless otherwise stated.For all panels: *** p<0.0005, ** p<0.005, * p<0.05 and all data are representative of biological triplicates.Bars represent average Firefly/Ren (RFU) from 3 technical replicates +/-S.E.

Figure 8 -
Figure 8 -CK1γ inhibition suppresses β-catenin stabilization and LRP6 phosphorylation.(A) Nanoluc-GSK3β NanoBRET curve and corresponding IC50 value for AKI.(B) Nanoluc-GSK3β NanoBRET curve and corresponding IC50 value for FP1.(C) HEK293T B/R cells were transfected with the indicated constructed for 14 hrs, then treated with 10 uM of the indicated compound for 18 hrs and analyzed by luciferase assay.(D) HEK293T B/R cells were transfected with the indicated constructed for 14 hrs, treated with DMSO, 0.1, 1.0 or 10 μM of FPI for 18 hrs, and then analyzed by luciferase assay.(E) RKO cells were pre-treated with FPI or DMSO for 1 hr, treated with WNT ligand for 30 or 60 mins, then analyzed by W. blot.(F-H) Box-and-whisker plot quantification of W. blot from panel E, β-catenin (panel F), pLRP6 S1490 (panel G) and pLRP6 T1479 (panel H) across 4 replicates, normalized to ACTIN.For all panels: *** p<0.0005, ** p<0.005, * p<0.05 and are compared to DMSO control unless otherwise indicated.All data are representative of biological triplicates, unless otherwise noted.Error bars represent S.E.Bars represent average Firefly/Ren (RFU) from 3 technical replicates +/-standard error (S.E.).
(which was not certified by peer review) is the author/funder.All rights reserved.No reuse allowed without permission.