To elucidate novel signaling pathways regulated by ZNRF3/RNF43 in cancer, we conducted integrative proteogenomic analyses of human cancer datasets using LinkedOmics (Vasaikar et al., 2018). Given the critical role of ZNRF3 in adrenal hemostasis (Basham et al., 2019) and its frequent deep deletion in approximately 20% of ACCs (Assié et al., 2014), our initial focus was to examine the ACC dataset from TCGA, comprising 92 samples, to identify proteins exhibiting correlations with ZNRF3 mRNA levels. Among the proteins evaluated by Reverse Phase Protein Array (RPPA), our analysis revealed EGFR protein levels to be most negatively correlated with ZNRF3 mRNA levels (r=–0.50 and p=4.5e-4) (Figure 1A, Figure 1—figure supplement 1A). Notably, ACC tumors with deep deletion in ZNRF3 exhibited the highest EGFR protein levels compared to ACC tumors with other alterations in the ZNRF3 locus (Figure 1B). These findings suggest that disruption of the gene function of ZNRF3 may lead to upregulation of EGFR in ACCs.

Besides frequent deletions in ACC, ZNRF3, and RNF43 are also important tumor suppressors in the more commonly occurring colorectal cancer (CRC) (Koo et al., 2012; Bond et al., 2016); thus, we next analyzed the TCGA colorectal adenocarcinoma (n=629) and CPTAC colon adenocarcinoma (n=110) datasets. By incorporating RPPA-based (TCGA CO/READ) and mass spectrometry-based (CPTAC COAD) proteomic data, we also confirmed EGFR protein to be the one most negatively correlated with ZNRF3 and RNF43 mRNA expression in CRC (Figure 1C, Figure 1—figure supplement 1B–E). Of note, we utilized ZNRF3/RNF43 mRNA abundances rather than protein levels due to the lack of protein measurements for these two low-abundance enzymes in the current proteomic datasets. Mutations in ZNRF3/RNF43 are found in approximately 15% of CRCs overall, with high frequencies observed in the microsatellite instability-high (MSI-H) subtype (60–80%) (Giannakis et al., 2014; Bond et al., 2016; Tu et al., 2019; Vasaikar et al., 2019). In particular, the RNF43 G659Vfs*41 frameshift mutation accounts for 40–50% of all RNF43 mutations in multiple MSI-H cancer types (Tu et al., 2019). Interestingly, CRCs harboring the RNF43 G659Vfs*41 mutation exhibited significantly higher levels of EGFR protein compared to RNF43 wild-type (WT) tumors (Figure 1D, Figure 1—figure supplement 1F). A similar difference was observed in stomach cancer, which also displayed a high frequency of the RNF43 G659Vfs*41 mutation (Figure 1—figure supplement 1G). Additionally, in the microsatellite stable subtype of CRCs, predominantly expressing WT RNF43, we found a negative association between RNF43 mRNA and EGFR protein levels (r=−0.36 and p=0.0018) (Figure 1E). Using cBioPortal (Cerami et al., 2012; Gao et al., 2013), we also detected a negative correlation between ZNRF3/RNF43 expression and EGFR protein in multiple other cancer types, including prostate cancer where ZNRF3 or RNF43 is deleted or mutated with a rate of 5% (Figure 1F). Collectively, these results suggest that disruption of ZNRF3 and RNF43 may upregulate EGFR protein levels in human cancers.

To test whether ZNRF3 and RNF43 regulate EGFR protein, we performed gain-of-function and loss-of-function experiments in several cell lines. In MDA-MB-231, a breast cancer cell line with moderate levels of ZNRF3/RNF43, overexpression of either ZNRF3 or RNF43 substantially reduced EGFR protein levels (Figure 2A). Remarkably, ZNRF3 and RNF43 decreased EGFR as robustly as CBL, the best-known E3 ligase of EGFR (Levkowitz et al., 1998; Waterman et al., 1999). Similarly, in HEK293T cells, ZNRF3/RNF43 overexpression reduced the levels of EGFR (Figure 2—figure supplement 1A). As predicted, the phosphorylated form of EGFR declined as well (Figure 2—figure supplement 1A). Notably, the reduction in both total EGFR and P-EGFR levels reached an extent comparable to that achieved by CBL (Figure 2—figure supplement 1A). Consistent with their known function, ZNRF3/RNF43 overexpression reduced the level of their substrate FZD5 (Hao et al., 2012; Koo et al., 2012; Figure 2—figure supplement 1A). Additionally, these two E3 ligases exhibited selectivity in their regulation of growth factor receptors since overexpression of ZNRF3 or RNF43 did not decrease levels of TGFβ receptor I (Figure 2—figure supplement 1B) or FGFR1 (Figure 2—figure supplement 1C).

To test the impact of ZNRF3/RNF43 loss on EGFR levels, we compared WT and Znrf3 knockout (KO) MEFs because the WT MEFs have significant Znrf3 expression but very minimal Rnf43 expression (Lienert et al., 2011). We found that Znrf3 KO enhanced both EGFR and P-EGFR levels, both in the absence and presence of recombinant EGF (Figure 2B). To test the effect of RNF43 KO, we utilized HT29, a CRC cell line that expresses WT RNF43 at a high level but minimal ZNRF3. CRISPR-KO of RNF43 enhanced EGFR and P-EGFR levels regardless of EGF stimulation (Figure 2C). We validated these results using three additional independent RNF43 sgRNAs (Figure 2—figure supplement 1D). Furthermore, since EGFR is mutationally activated in some human cancers, we transfected a common mutant – EGFR L858R – into RNF43 WT versus KO HT29 cells and compared the resulting protein levels. As shown in Figure 2D, EGFR L858R was significantly higher in RNF43 KO cells than in WT cells, indicating that RNF43 can degrade both WT and mutated EGFR and its loss can enhance signaling of both WT EGFR and its oncogenic mutant. In addition, we extended our analysis to HCC1954, a breast cancer cell line that expresses the EGFR family member HER2. We found that KO of either ZNRF3 or RNF43 enhanced the protein levels of HER2, as well as EGFR (Figure 2—figure supplement 1E), suggesting that ZNRF3 and RNF43 downregulate the levels of EGFR and its family members and that the loss of these E3 ligases stabilizes this family of receptor tyrosine kinases.

RSPO1-4 are antagonistic ligands of ZNRF3 and RNF43 (Hao et al., 2012; Koo et al., 2012). They bind to both ZNRF3/RNF43 and LGR4/5/6, resulting in ZNRF3/RNF43 auto-ubiquitination and degradation (Hao et al., 2012; Koo et al., 2012). Approximately 18% of CRCs harbor amplification or mutations of RSPO1-4 or LGR4/5/6, and 10% of CRCs show recurrent RSPO2/3 gene fusions (Seshagiri et al., 2012; Seeber et al., 2019). These genetic alterations potentially lead to the inhibition of ZNRF3/RNF43, thereby activating WNT and EGFR signaling. Therefore, we tested whether RSPO affected EGFR protein levels in CRC. Short-term treatment with recombinant RSPO2 increased EGFR levels in two CRC cell lines (Figure 2E and F) and 293T cells (Figure 2—figure supplement 1F). Further, this impact of RSPO2 depended on the presence of intact ZNRF3/RNF43, as RSPO2 failed to elevate EGFR levels further in HT29 cells knocked out for RNF43, the predominant one compared to ZNRF3 (Figure 2—figure supplement 1G). Conversely, the RSPO2 R65A/Q70A mutant, which cannot bind to ZNRF3/RNF43 (Xie et al., 2013), failed to elevate EGFR levels (Figure 2G). On the other hand, the RSPO2 F105A/F109A mutant, which cannot bind to LGRs (Xie et al., 2013), still enhanced EGFR levels similar to RSPO2 WT (Figure 2F). These data collectively suggest that RPSO2 regulation of EGFR does not rely on LGR-binding but requires its interaction with ZNRF3/RNF43.

After demonstrating that ZNRF3 and RNF43 regulate EGFR protein levels, we next sought to elucidate the underlying mechanism. We first examined whether ZNRF3 and RNF43 regulate EGFR transcripts and found that ZNRF3/RNF43 did not impact EGFR mRNA levels in either MEFs or HT29, as determined by quantitative PCR (qPCR) (Figure 3A and B). Then, we studied the protein stability of EGFR in WT and RNF43 KO HT29 cells. We treated cells with recombinant EGF to induce EGFR internalization/degradation and then collected cell lysates at different time points to measure EGFR protein levels by western blotting. We found that KO of RNF43 delayed EGFR degradation and sustained P-EGFR levels (Figure 3C). Since EGFR clearance is initiated at the cell membrane and accelerated by EGF treatment, we also evaluated the impact of ZNRF3/RNF43 on EGFR levels at the cell surface. Flow cytometry revealed that ZNRF3 KO in MEFs (Figure 3D) or RNF43 KO in HT29 (Figure 3E) increased cell surface EGFR in both unstimulated and EGF-stimulated conditions. These data together suggest that ZNRF3 and RNF43 induce cell surface EGFR internalization and degradation, and that their inactivation leads to EGFR accumulation at the cell surface and thus enhanced EGFR signaling.

EGFR degradation is primarily mediated by ubiquitination (Galcheva-Gargova et al., 1995; Levkowitz et al., 1998). To examine the impact of ZNRF3 and RNF43 on EGFR ubiquitination, we performed EGFR immunoprecipitation (IP) followed by ubiquitin (Ub) immunoblotting (IB). Remarkably, RNF43 overexpression in MDA-MB-231 cells enhanced EGFR ubiquitination as potently as CBL overexpression (Figure 4A). Conversely, in RNF43 KO HT29 cells and MDA-MB-231 cells, anti-Ub IP brought down substantially less EGFR protein (Figure 4B, Figure 4—figure supplement 1A), and anti-EGFR IP produced much less ubiquitinated forms of EGFR (Figure 4C), indicating that RNF43 KO diminished EGFR ubiquitination. Next, we asked whether the E3 ligase activity of ZNRF3/RNF43 is needed to regulate EGFR ubiquitination. The RING domain is required for ZNRF3/RNF43 E3 Ub ligase function (Pickart and Eddins, 2004). Therefore, we compared the impact of ZNRF3/RNF43 versus their RING domain deletion mutants (ΔRING) on EGFR protein levels. We found that both ΔRING mutants failed to decrease EGFR levels (Figure 4D and E) or HER2 levels (Figure 4—figure supplement 1B). Furthermore, while WT ZNRF3 overexpression increased EGFR ubiquitination, which was detectable even in the absence of proteasome or lysosome inhibitors, the ΔRING mutant failed to induce detectable upregulation of EGFR ubiquitination even in the presence of both proteasome and lysosome inhibitors (MG132 and BAF, respectively) (Figure 4F). Together, these results demonstrate that ZNRF3 and RNF43 regulate EGFR ubiquitination and degradation through their E3 ligase activity.

We next investigated whether ZNRF3 and RNF43 interact with EGFR to regulate its ubiquitination. In a co-IP experiment using lysates from MDA-MB-231 cells co-infected with lentivirus carrying EGFR and Myc-tagged ZNRF3/RNF43, anti-EGFR IP pulled down overexpressed ZNRF3/RNF43 (Figure 5A), indicating a complex between EGFR and ZNRF3/RNF43. Mass spectrometry of anti-EGFR immunoprecipitates identified endogenous ZNRF3 protein (Figure 5—figure supplement 1A), confirming ZNRF3 as an interacting partner of EGFR. Importantly, a proximity ligation assay (PLA) provided visual confirmation of the close interaction between co-transfected ZNRF3 and EGFR in situ (Figure 5B). Next, we sought to determine the specific regions involved in the interaction between ZNRF3/RNF43 and EGFR. ZNRF3 and RNF43 are transmembrane proteins comprising a long ECD, a single-span transmembrane domain (TM), and a catalytic intracellular domain (ICD) (Figure 5C). We thus generated constructs expressing ZNRF3 ECD-TM and TM-ICD, with an N-terminal 3× Myc-tag inserted after the signal peptide for detection (Figure 5C). Immunofluorescence (IF) confirmed the membrane localization of these two peptides (Figure 5D), after which we performed co-IP using anti-Myc. EGFR was co-immunoprecipitated with Myc-tagged ZNRF3 full-length (FL) and ECD-TM, but not TM-ICD (Figure 5E). Conversely, anti-EGFR IP pulled down ZNRF3 FL and ECD-TM, but not TM-ICD (Figure 5E). Similar results were obtained when performing anti-HA IP of HA-tagged RNF43 FL and ECD-TM, which also brought down EGFR (Figure 5—figure supplement 1B, C). These data collectively indicate that ZNRF3 and RNF43 interact with EGFR via their ECDs, in accordance with evidence indicating the ZNRF3/RNF43 interaction with Frizzled via ECDs (Tsukiyama et al., 2015). This is in contrast with reports detecting ZNRF3/RN43 interaction with Frizzled via their ICDs and other adaptor proteins (Jiang et al., 2015).

The protease-associated (PA) domain is the only part conserved between the extracellular fragments of ZNRF3 and RNF43 (Tsukiyama et al., 2021). Therefore, we tested whether this domain is necessary for the interaction between ZNRF3/RNF43 with EGFR. We created PA domain-deletion mutants of ZNRF3/RNF43 (Figure 5—figure supplement 2A) and found that these deletion mutants (ΔPA) also interacted with EGFR (Figure 5—figure supplement 2B and C), indicating that the PA domain is dispensable for ZNRF3/RNF43 interaction with EGFR. In accordance with previous reports that the PA domain is dispensable for suppression of WNT signaling (Jiang et al., 2015; Radaszkiewicz and Bryja, 2020), we found that ΔPA ZNRF3 retained the ability to suppress the Top-Flash WNT reporter (Figure 5—figure supplement 2D).

EGFR is an important tyrosine kinase that mediates many cellular activities, especially cell growth (Wieduwilt and Moasser, 2008; Yue et al., 2021). To study the biological consequence of ZNRF3/RNF43 downregulating EGFR, we first compared cell growth of WT and ZNRF3/RNF43-depleted cells. Znrf3 KO in MEFs led to the upregulation of EGFR (Figure 2B) and increased cell growth (Figure 6A), without affecting canonical WNT signaling activity based on qPCR for WNT target genes (Sox2, Axin2, Ccnd1) (Figure 6B). EGF and EGFR signaling are also critical for the culture and maintenance of organoids derived from Apc-deficient mouse intestinal adenomas (Sato et al., 2011), which exhibit constitutively canonical WNT signaling that is no longer modulated by the RSPO-ZNRF3/RNF43 control at the membrane receptor level (Tsukiyama et al., 2020; Tsukiyama et al., 2021). Therefore, we established intestinal tumor organoids using the intestinal polyps from the Apc^min mice (Evans et al., 1992) and supplemented RSPO1 in the organoid culture medium to investigate the effect of ZNRF3/RNF43 deactivation on organoid growth and EGFR. Remarkably, the addition of RSPO1 increased the size of Apc^min mouse intestinal tumor organoids (Figure 6C and D) in agreement with a previous report (Lähde et al., 2021) although it did not significantly affect the frequency of organoids detected (Figure 6E). IB analysis confirmed elevated EGFR levels after RSPO1 treatment (Figure 6F) while qPCR showed no significant increase of Egfr nor WNT target genes, including Axin2, Myc, Sox2, and Cd44 (Figure 6G). Furthermore, we performed functional assays in HT29 cells, which contain a mutated APC. Overexpression of ZNRF3 in HT29 substantially inhibited cell growth (Figure 6H) and reduced EGFR levels (Figure 6I). Conversely, KO of RNF43 in HT29 activated multiple EGFR downstream effectors, as RPPA detected increased levels of P-AKT, P-ERK, P-GSK3, P-STAT1, and P-PRAS40 (Figure 6—figure supplement 1). Importantly, treatment with the EGFR activity inhibitor erlotinib, which abolished EGFR phosphorylation (Figure 6J), blocked the growth gain caused by RNF43 KO (Figure 6K). We also treated WT and Znrf3 KO MEF cells with erlotinib, the WNT inhibitor Wnt-C59 (a small molecule that suppresses porcupine, which is required for palmitoylation and secretion of Wnt; Proffitt et al., 2013), or their combination. Neither inhibitor impacted WT MEF cell growth. In contrast, erlotinib, but not WNT-C59, significantly blocked the growth gain caused by Znrf3 KO (Figure 6L), suggesting Znrf3 KO induced MEF growth primarily via EGFR signaling rather than WNT signaling. Together, these data indicate that loss of ZNRF3 and RNF43 can promote tumor cell growth through EGFR signaling, sometimes even without substantially engaging canonical WNT signaling.

To demonstrate the function of the ZNRF3/RNF43-EGFR signaling pathway in vivo, we first xenografted luciferase-labeled HT29 cells overexpressing either GFP or ZNRF3 into NSG mice by flank injection and monitored tumor growth by bioluminescence imaging (Figure 7A). We found that overexpression of ZNRF3 significantly inhibited tumor growth (Figure 7A). These tumors exhibited decreased EGFR and P-EGFR levels compared to WT tumors (Figure 7B). However, canonical WNT signaling, based on levels of active-β-catenin (non-phosphorylated at Ser33/37/Thr41; Figure 7B), remained unaffected, suggesting that ZNRF3 suppresses tumor growth through downregulation of EGRF signaling without substantial involvement of canonical WNT signaling.

Furthermore, we validated ZNRF3/RNF43-EGFR signaling in prostate cancer, one of the several other tumors in patients besides CRC that showed a negative correlation between ZNRF3/RNF43 mRNA and EGFR protein levels (Figure 1F). Specifically, we tested whether ZNRF3/RNF43 loss affects EGFR signaling and promotes prostate tumorigenesis in vivo. By breeding mice that contain cre-inactivatable alleles of Znrf3 and Rnf43 (Koo et al., 2012) with mice transgenic for Probasin-Cre (Wu et al., 2001), we generated prostate-specific Znrf3/Rnf43 KO mice. At 1 year of age, these mice exhibited multi-focal prostatic dysplasia and multifocal invasive tumors, as shown by the H&E images (Figure 7C) and pathological scores (Figure 7D, Figure 7—figure supplement 1A–D). Immunohistochemical staining revealed elevated levels of both EGFR and P-EGFR in Znrf3/Rnf43 KO prostatic dysplasia and tumors, compared to WT prostate tissues (Figure 7E), indicating EGFR signaling activation as a result of Znrf3/Rnf43 KO. We also observed more intense β-catenin staining in KO tissues and tumors than in WT mice, although the β-catenin signals were largely excluded from the nucleus in both WT and Znrf3/Rnf43 KO mouse prostate tissues and tumors (Figure 7E). Together, these results suggest that loss of ZNRF/RNF43 elevates EGFR levels and signaling, promoting tumor development, sometimes in conjunction with Wnt signaling activation and other times independent of Wnt signaling.

All experiments using mice were performed utilizing procedures approved by the Institutional Animal Care and Use Committees at Baylor College of Medicine and the Van Andel Institute. The Rnf43-flox; Znrf3-flox were obtained from the laboratory of Hans Clevers (Koo et al., 2012). CMV-Cre animals were ordered from Jackson Laboratories (JAX stock #006054) (Schwenk et al., 1995). C57Bl/6J used for backcrossing were ordered from Jackson Laboratories (JAX stock #000664). Probasin-Cre animals were developed in the laboratory of Pradip Roy-Burman (Wu et al., 2001). NOD.Cg-Prkdc scid Il2rg tm1Wjl/SzJ (NSG) mice were purchased from Jackson Laboratories (JAX stock #005557) and bred in-house.

MDA-MB-231, MDA-MB-468, and HEK293T cells were purchased from ATCC. HT29 and LS180 cells were provided by Q Liu; HCT116 cells were provided by N Shroyer; MEF cells were generated from e12.5 embryos using standard methods (Herrera et al., 1996). MDA-MB-231, MDA-MB-468, HT29, HCT116, HEK293T, and MEF cells were cultured in DMEM (Corning, 10-013-CV) with 10% FBS and 1% penicillin/streptomycin. LS180 cells were cultured in RPMI1640 (Corning, 10-013-CV) with 10% FBS and 1% penicillin/streptomycin. All cell lines were routinely tested for mycoplasma contamination.

Plasmids expressing Flag-FZD5, RNF43, HA-RNF43, or HA-RNF43 (CD8 sp) were provided by Q Liu. Plasmids expressing Myc-RNF43, Myc-ZNRF3, or Myc-ZNRF3 ΔRING were provided by F Cong. Plasmids for Myc-ZNRF3-ECD-TM, Myc-ZNRF3-TM-ICD, Myc-ZNRF3 ΔPA, HA-RNF43 ΔPA, HA-RNF43 ΔRING, HA-RNF43-ECD-TM were constructed by subcloning using In-Fusion (TaKaRa, 639642). Plasmids were further constructed for lentiviral expression by subcloning into pBobi vector by In-Fusion. Flag-EGFR (L858R) is PCR-amplified and subcloned into FUCGW vector at EcoRI site for lentiviral expression.

HT29 RNF43 KO cells were provided by Q Liu (guide RNA sequence, g0: 5’-GGCTGCTGATGGCTACCCTGC-3’). Other CRISPR guide sequences for knocking out RNF43 were designed by http://crispr.mit.edu and cloned into lentiCRISPR v2 (Addgene 52961). Sequences were as follows: g1: 5’-TGGACGCACAGGACTGGTAC-3’; g2: 5’-CAGAGTGATCCCCTTGAAAA-3’; g3: 5’-GGGCAGCCAGCTGCAGCTGG-3’. Lentiviral construct for knocking out ZNRF3 was provided by Q Liu (guide RNA sequence, 5’- AGGACTTGTATGAATATGGC-3’) (Jiang et al., 2015; Tu et al., 2019). Cancer cells were infected with lentivirus-carrying CRISPR constructs or vector, and then selected in the culture medium containing 2 µg/ml puromycin (InvivoGen, ant-pr-1).

Global null alleles of Rnf43 and Znrf3 were generated in vivo by crossing Rnf43-flox; Znrf3-flox mice (Koo et al., 2012) with CMV-Cre (Schwenk et al., 1995). The resulting mice were backcrossed to C57Bl/6J animals to remove the Cre transgene and generate global deletion mouse colonies. Global deletions were confirmed using allele-specific PCR for both Rnf43 and Znrf3. MEFs were generated using standard protocols (Durkin et al., 2013). Briefly, Rnf43^KO/+; Znrf3^KO/+ mice were crossed, and embryos were collected at E12.5. Yolk sacs were collected for genotyping, and embryo heads and internal organs were removed. The remaining tissue was minced using a razor blade and 0.5 ml 0.05% Trypsin-EDTA was added. The suspension was incubated at 37°C for 30 min, then cultured in MEF media (DMEM, 10% FBS, 1× PenStrep) for subsequent experiments.

Erlotonib (Selleck Chemicals) was dissolved in ethanol, and Wnt-C59 (Cellagen Technology) was dissolved in DMSO.

MEF cells were seeded at 5000 cells in 96-well and the media was refreshed with drugs the next day. On day 4 post-drug treatments, cells were incubated with Cell Counting Kit-8 (CCK-8) reagents (ApexBio) for 4 hr, and the growth was determined at 450 nm by spectrophotometer.

Intestinal polyps from 6-month-old female Apc^min mice (Evans et al., 1992) were isolated and maintained using previously described protocols (Sato et al., 2011; Xue and Shah, 2013), embedded in Matrigel (Corning, 356231), and cultured in basal culture medium (advanced Dulbecco’s modified Eagle medium/F12 [Invitrogen, 12634-028], penicillin/streptomycin [Invitrogen, 15140-122], 10 mmol/l HEPES [Invitrogen, 15630-080], 2 mM GlutaMAX [Invitrogen, 35050-061], 1×N2 [Invitrogen, 17502-048], 1×B27 [Invitrogen, 17504-044], 1 mmol/l N-acetylcysteine [Sigma-Aldrich, A9165-5G]) supplemented with 50 ng/ml mouse recombinant EGF (Invitrogen, PMG8043). The organoids were trypsinized with 0.05% Trypsin-EDTA (Invitrogen, 25300054) to obtain single cells, counted, and embedded into Matrigel in equal numbers, cultured in basal culture medium supplemented with EGF or basal culture medium supplemented with EGF and 10% RSPO1 conditioned medium for 8 days. The organoids were imaged using a Leica DMi8 Inverted Microscope and then harvested for immunoblot analysis.

Cells were lysed with TRIzol reagent (Invitrogen, 15596-026). Total RNA was extracted by chloroform and isopropanol precipitation. cDNA was obtained using the SuperScript III First-Strand Synthesis System (Invitrogen, 18080-051). qPCR analyses were performed with primers listed below using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, 1725270) with at least duplicate samples in three independent experiments. Plotted are data normalized to ACTB or Actb and relative to the control. The following primers were used: ACTB: 5’-ACTCTTCCAGCCTTCCTTCC-3’, 5’-CAGTGATCTCCTTCTGCATCC-3’; EGFR: 5'-TGCCCATGAGAAATTTACAGG-3', 5'-ATGTTGCTGAGAAAGTCACTGC-3'; Actb: 5’- CATTGCTGACAGGATGCAGAAGG-3’, 5’-TGCTGGAAGGTGGACAGTGAGG-3’; Egfr: 5′- GGACTGTGTCTCCTGCCAGAAT-3′, 5′-GGCAGACATTCTGGATGGCACT-3′; Axin2: 5’-ATGGAGTCCCTCCTTACCGCAT-3’, 5’-GTTCCACAGGCGTCATCTCCTT-3’; Sox2: 5’-AACGGCAGCTACAGCATGATGC-3’, 5’-CGAGCTGGTCATGGAGTTGTAC-3’; Ccnd1: 5’- GCAGAAGGAGATTGTGCCATCC-3’, 5’-AGGAAGCGGTCCAGGTAGTTCA-3’. The primer sequences for Tgfbr1, Tgfbr2, Smad4, Ep300, Bach1 were from OriGene.

Cells were trypsinized and resuspended in antibody dilution buffer (0.5% BSA in PBS). 5×10⁵ cells were incubated with EGFR (D1D4J) XP Rabbit antibody (PE Conjugate) (Cell Signaling Technology, 48685) for 1 hr at 4°C. Cells were then washed with antibody dilution buffer and resuspended in antibody dilution buffer. The cells were then subjected to flow cytometry using a BD FACSCanto II (BD, NJ, USA). The resulting data were analyzed using the BD FACSDiva software.

HEK293T cells were seeded on poly-D-lysine-coated chamber slides (Corning, 354632) and transfected with empty vector, or constructs expressing WT or fragments of RNF43 or ZNRF3. 48 hr after transfection, cells were fixed in 3.2% paraformaldehyde in PBS at RT for 15 min, permeabilized in 3% BSA/0.1% saponin (Sigma-Aldrich, 47036) in PBS at RT for 30 min, and then incubated in the secondary antibody anti-Rabbit Alex 488 (Invitrogen, A32731) at RT for 1 hr. Afterward, the cells were stained with DAPI (Thermo Fisher Scientific, 62248) and mounted.

Immunochemistry staining was performed using the Vectastain Elite ABC system (Vector Laboratories, Burlingame, CA, USA) and developed using DAB as chromogen (Agilent Dako, Santa Clara, CA, USA). Primary antibodies used in the study included EGFR (D1P9C) rabbit antibody (Mouse Preferred, Cell Signaling Technology, 71655), P-EGFR (Y1068) XP rabbit antibody (Cell Signaling Technology, 2234), non-phospho (Active) β-catenin (Ser33/37/Thr41) (D13A1) rabbit antibody (Cell Signaling Technology, 8814).

HCT116 cells were seeded onto poly-D-lysine-coated chamber slide (Corning) and co-transfected with Myc-ZNRF3ΔR and EGFR. Two days after transfection, the cells were processed following the standard protocol provided by Sigma-Aldrich (Duolink In Situ Red Starter Kit Mouse/Rabbit). The primary antibodies, EGFR (D38B1) XP rabbit antibody (Cell Signaling Technology, 4267) and Myc-tag (9B11) mouse antibody (Cell Signaling Technology, 2276), were used to detect ZNRF3 and EGFR interaction. For negative control, no primary antibody was used. DAPI was used to stain nuclei. The stained cells were imaged using Leica DMi8 Fluorescence Microscope.

Cells were lysed on ice using RIPA lysis buffer supplemented with protease inhibitors (Sigma-Aldrich, P8340) and phosphatase inhibitors (Sigma-Aldrich, P5726 and P0044). Cell lysates were then centrifuged at 14,000 rpm for 15 min at 4°C. Supernatant was collected, and the concentration of protein lysate was quantified by BCA assay (Thermo Fisher Scientific, 23225). Cell lysates were mixed with Laemmli sample buffer (Bio-Rad, 1610747) and β-mercaptoethanol (Thermo Fisher Scientific, 21985023) before boiling for 5 min at 95°C. Equal amounts of protein lysates were loaded and run in 8–12% SDS-PAGE gel. Gels were transferred onto nitrocellulose membrane (Thermo Fisher Scientific, 88018) at 100 V for 90 min at 4°C. Membranes were then blocked with 5% non-fat milk in TBST at RT for 1 hr, incubated with primary antibodies overnight at 4°C and secondary antibodies at RT for 1 hr, and scanned using the Odyssey LI-COR imaging system. Primary antibodies used in the study included EGFR (D38B1) XP rabbit antibody (Cell Signaling Technology, 4267), EGFR (A-10) mouse antibody (Santa Cruz, sc-373746), EGFR (D1P9C) rabbit antibody (Mouse Preferred, Cell Signaling Technology, 71655), P-EGFR (Y1068) XP rabbit antibody (Cell Signaling Technology, 3777), HA-tag mouse antibody (BioLegend 901501), HA-tag (C29F4) rabbit antibody (Cell Signaling Technology, 3724), Myc-tag (9B11) mouse antibody (Cell Signaling Technology, 2276), Myc-tag (71D10) rabbit antibody (Cell Signaling Technology, 2278), Flag-tag M2 mouse antibody (Sigma-Aldrich, F1804), Ub (P4D1) mouse antibody (Santa Cruz, sc-8017), CBL mouse antibody (Santa Cruz, sc-1651), non-phospho (Active) β-catenin (Ser33/37/Thr41) (D13A1) rabbit antibody (Cell Signaling Technology, 8814), β-catenin mouse antibody (BD Biosciences, 610153), GAPDH rabbit antibody (Santa Cruz, sc-25778), β-actin (Sigma, A5441), and α-tubulin mouse antibody (Sigma-Aldrich, T9026). Secondary antibodies used included anti-Mouse IRDye 680RD (LI-COR, 926-68070) and anti-Rabbit IRDye 800CW (LI-COR, 926-32211).

Cells were lysed on ice using NP-40 or RIPA lysis buffer supplemented with protease inhibitors (Sigma-Aldrich, P8340). Cell lysates were then centrifuged at 14,000 rpm for 15 min at 4°C. Supernatant was collected, and the concentration of protein lysate was quantified by BCA. Protein lysates were incubated with either EGFR (Ab-13) mouse antibody (Thermo Fisher Scientific, MS-609-P1) or Ub (P4D1) mouse antibody (Santa Cruz, sc-8017), and Protein G Sepharose beads (GE Healthcare, 17061801) overnight at 4°C. Beads were washed with NP-40 or RIPA lysis buffer. Whole-cell lysates and immunoprecipitates were analyzed by western blot analysis.

The IP-MS experiments were performed as previously described (Chen et al., 2019). BCG823 cells were starved overnight and then treated with 50 ng/ml EGF for 120 min before sample collection. For each IP experiment, 1 mg protein lysate was incubated with 5 µg EGFR (Ab-13) antibody for 2 hr at 4°C and cleared by ultracentrifugation (100,000×g, 15 min). The supernatant was then incubated with 30 µl 50% protein A-Sepharose slurry (GE Healthcare, 17-0780-01) for 1 hr at 4°C. The bead-bound complexes were washed four times with NETN buffer (20 mM Tris pH 7.5, 1 mM EDTA, 0.5% NP-40, 150 mM NaCl) and eluted with 20 µl 2× Laemmli buffer and heated at 95°C for 10 min. IP samples were resolved on a NuPAGE 10% Bis-Tris gel (Life Technologies, WG1201BX10) in 1× MOPS running buffer; the gel was cut into 3 molecular weight regions plus the IgG heavy and light chain bands. Each band was in-gel digested overnight with 100 ng of trypsin, which cleaves peptide chains at the C-termini of lysine or arginine in 20 µl of 50 mM NH₄HCO₃ at 37°C. Peptides were extracted with 350 µl of 100% acetonitrile and 20 µl of 2% formic acid, and dried in a Savant Speed-Vac. Digested peptides were dissolved in 10 µl of loading solution (5% methanol containing 0.1% formic acid) and subjected to LC-MS/MS assay as described previously (Dharmat et al., 2018).

RPPA assays were carried out as described previously with minor modifications (Creighton and Huang, 2015). Protein lysates were prepared with modified TPER buffer supplemented with a cocktail of protease and phosphatase inhibitors. Protein lysates at 0.5 mg/ml of total protein were denatured in SDS sample buffer containing 2.5% β-mercaptoethanol at 100°C for 8 min. The Aushon 2470 Arrayer (Aushon BioSystems) with a 40 pin (185 μm) configuration was used to spot samples and control lysates onto nitrocellulose-coated slides (Grace Bio-labs) using an array format of 960 lysates/slide (2880 spots/slide). The slides were probed with a set of 216 antibodies against total and phospho-proteins using an automated slide stainer Autolink 48 (Dako). Each slide was incubated with one specific primary antibody, and a negative control slide was incubated with antibody diluent instead of primary antibody. Primary antibody binding was detected using a biotinylated secondary antibody followed by streptavidin-conjugated IRDye680 fluorophore (LI-COR Biosciences). Total protein content of each spotted lysate was assessed by fluorescent staining with Sypro Ruby Protein Blot Stain according to the manufacturer’s instructions (Molecular Probes). Fluorescence-labeled slides were scanned on a GenePix 4400 AL scanner, along with accompanying negative control slides, at an appropriate PMT to obtain optimal signal for this specific set of samples. The images were analyzed with GenePix Pro 7.0 (Molecular Devices). Total fluorescence signal intensities of each spot were obtained after subtracting the local background signal for each slide and were then normalized for variation in total protein, background, and non-specific labeling using a group-based normalization method. For each spot on the array, the background-subtracted foreground signal intensity was subtracted by the corresponding signal intensity of the negative control slide (omission of primary antibody) and then normalized to the corresponding signal intensity of total protein for that spot. Each image and its normalized data were evaluated for quality through manual inspection and control samples. Antibody slides that failed the quality inspection were either repeated at the end of the staining runs or removed before data reporting. Samples (in four biological replicates) were extracted from the normalized data and then log2 transformed. The median value of the three technical replicates was used for statistical analysis. Welch’s t-tests were used for group comparisons. FDR-adjusted p-value<0.05 was considered statistically significant.

5×10⁵ luciferase-labeled HT29 cells were suspended in 0.1 ml serum-free DMEM and injected into the right back flank of 21- to 24-week-old NSG mice. Bioluminescence was measured once a week, by injection of 100 µl 15 mg/ml D-luciferin (LUCNA-1G, Goldbio) via the intra-orbital sinus. Mice were imaged using IVIS Lumina II (Advanced Molecular Vision). The acquired bioluminescence signals were normalized to the day 0 bioluminescence intensity.

All statistical analyses in our study were conducted using GraphPad Prism or R. Two-tailed t-test without equal variation assumption was used for data comparison between two experimental groups. For data comparison with at least three groups, one-way ANOVA test was performed first to assess the overall difference among groups. If differences existed, uncorrected Fisher’s LSD test was performed to assess the significance of differences between two groups. Two-way repeated measures ANOVA and uncorrected Fisher’s LSD test were used to assess the difference between datasets with time series measurements, including growth curves of cell proliferation and in vivo BLI signals. p-Values less than 0.05 were considered significant.