Proteogenomic analysis identifies EGFR as the top candidate protein down-regulated by ZNRF3/RNF43 in cancers.

A, Volcano plot of proteins associated with ZNRF3 mRNA expression in human adrenal cortical carcinoma (ACC), using the TCGA dataset (n = 92).

B, Boxplot of EGFR protein levels in human adrenal cortical carcinomas with different ZNRF3 gene copy number alteration, using the TCGA dataset (n = 92).

C, The top ten proteins negatively correlated with ZNRF3/RNF43 mRNA levels, ranked by Stouffer’s method, using the TCGA colorectal adenocarcinoma (CO/READ) (n = 629) and the CPTAC colon adenocarcinoma (COAD) (n = 210) datasets.

D. Boxplot of EGFR protein levels in human colon adenocarcinomas expressing RNF43 WT or G659Vfs*41 mutant, using the CPTAC dataset.

E. Scatterplot of EGFR protein level versus RNF43 mRNA expression using microsatellite stable (MSS) colon adenocarcinoma in the CPTAC dataset.

F. Bar graph of significant associations between RNF43/ZNRF3 mRNA expression and EGFR protein level in cancer datasets from TCGA PanCancer Altas Study. *, insignificant associations were not shown.

ZNRF3/RNF43 downregulates EGFR protein level.

A. Overexpression of RNF43, ZNRF3, or CBL decreases EGFR protein level compared to GFP control in MDA-MB-231 cells. Cells were infected with lentivirus expressing GFP, or E3 ligases.

B. Znrf3 knockout increases P-EGFR and total EGFR levels in MEFs upon EGF stimulation (50 ng/ml, 10 min).

C. RNF43 knockout increases P-EGFR and total EGFR levels in HT29 cells untreated or treated with recombinant EGF (50 ng/ml, 10 min).

D, E. RSPO2 treatment (50 ng/ml, 2-4 hr) enhances EGFR protein levels in HT29 (D) and LS180

(E) cells.

F. Overexpression of RSPO2 WT or F105A/F109A mutant but R65A/Q70A mutant enhances EGFR protein level, shown by representative western blot images (left panel) and quantification results (right panel).

Means ± SEMs are shown. p-values were calculated by one-way ANOVA uncorrected Fisher’s LSD test. n.s., not significant.

Loss of ZNRF3/RNF43 delays EGFR protein degradation.

A. Knockout of Znrf3 has no impact on Egfr mRNA level in MEFs.

B. Knockout of RNF43 has no impact on EGFR mRNA level in HT29 cells.

C. RNF43 knockout inhibits EGF-induced EGFR protein degradation in HT29 cells. Cells were stimulated with EGF (50 ng/ml) for indicated times. Representative western blot images (left panel) and quantification results (right panel) were shown.

D. Znrf3 knockout increases the cell surface level of EGFR protein in MEFs unstimulated or stimulated with EGF (50 ng/ml, 10 min).

E. RNF43 knockout increases the cell surface level of EGFR protein in HT29 cells unstimulated or stimulated with EGF (50 ng/ml, 10 min). The bars mark the relative peak shifts after EGF stimulation in WT (black) or KO (red) cells.

Means ± SEMs are shown. p-values were calculated by Welch’s t-test (A, B) or two-way ANOVA uncorrected Fisher’s LSD test (C). n.s., not significant.

ZNRF3/RNF43 enhances EGFR ubiquitination through the RING domain

A. Overexpression of RNF43 enhances EGFR ubiquitination level upon EGF (50 ng/ml) stimulation in MDA-MB-231 cells. CBL serves as a positive control. Cells were co-infected with lentivirus expressing EGFR and GFP, RNF43, or CBL.

B, C. Knockout of RNF43 decreases EGFR ubiquitination in HT29 cells. (B) EGFR ubiquitination was examined by Ub IP followed by EGFR IB; (C) HT29 cells were pretreated with 20 µM MG132 and 100 nM Bafilomycin A1 for 4 hr. EGFR ubiquitination after EGF treatment (50 ng/ml, 30 min) was examined by EGFR IP followed by Ub IB.

D, E. ZNRF3/RNF43 downregulates EGFR protein level through the RING domain. 293T cells were co-transfected with EGFR and Vector, ZNRF3 WT or ΔRING mutant (D), RNF43 WT or ΔRING mutant (E).

F. ZNRF3 regulates EGFR ubiquitination through the RING domain. 293T cells were co- transfected with EGFR and Vector, ZNRF3 WT or ΔRING mutant. EGFR ubiquitination after EGF treatment (50 ng/ml, 10 min) was examined by EGFR IP followed by Ub IB. Cells were pretreated with 20 µM MG132 and 100 nM Bafilomycin A1 for 4 hr.

ZNRF3/RNF43 interacts with EGFR through the extracellular domain

A. Ectopically expressed EGFR is co-immunoprecipitated with Myc-tagged RNF43 and ZNRF3 in MDA-MB-231 cells.

B. Representative images of proximity ligation assay in HCT116 cells co-transfected with EGFR and Myc-ZNRF3ΔRING. Red, PLA signals; Blue, DAPI nuclei staining; Scale bar=20 µm.

C. Schematic diagram of tagged ZNRF3/RNF43 proteins. SP, signal peptide; FL, full-length; ECD, extracellular domain; TM, transmembrane domain; ICD, intracellular domain; RING, E3 ligase RING domain.

D. Immunofluorescence staining for ZNRF3 in 293T cells expressing Myc-tagged ZNRF3 FL, ECD-TM, TM-ICD. Scale bar=40 μm.

E. ZNRF3 extracellular domain is required for ZNRF3 interaction with EGFR. 293T cells were co-transfected with EGFR and Myc-tagged ZNRF3 constructs and the lysate amounts were adjusted to achieve comparable levels of EGFR protein in each IP system (Input, left panel). EGFR interaction with ZNRF3 FL, ECD-TM, or TM-ICD were examined by Myc-tag IP followed by EGFR IB (middle panel) or by EGFR IP followed by Myc-tag IB (right panel). *, IgG heavy chain.

ZNRF3/RNF43 inhibits EGFR-mediated cell growth

A. Knockout of Znrf3 enhances MEF cell growth, as measured by cell counting at the indicated time points.

B. qPCR analysis for WNT target genes in WT and Znrf3 KO MEFs.

C-E. Supplementing RSPO1 promotes Apcmin mouse intestinal tumor organoid growth. The equal number of single cells from Apcmin mouse intestinal tumor organoids were embedded in Matrigel and cultured without or with 10% RSPO1 conditioned medium for 8 days. Representative images (C), and quantification of the size (D) and number (E) of formed Apcmin mouse intestinal tumor organoids are shown. Scale bar = 500 μm.

F. Supplementing RSPO1 enhances EGFR protein level in Apcmin mouse intestinal tumor organoids. Representative images (left panel) and quantification (right panel) of EGFR protein level are shown.

G. qPCR analysis for Egfr and WNT target genes in Apcmin mouse intestinal tumor organoids cultured with or without RSPO1 supplements. Genes with no significant changes after RSPO1 treatment were plotted in grey, genes significantly down-regulated after RSPO1 treatment were plotted in blue. **, p-value < 0.01.

H. Overexpression of ZNRF3 inhibits HT29 cell growth. HT29 cells stably overexpressing GFP or ZNRF3 were seeded in equal numbers and measured by confluence percentage using Incucyte.

I. Overexpression of ZNRF3 reduces EGFR protein level in HT29 cells.

J. Erlotinib treatment blocks EGFR phosphorylation in WT and RNF43 KO HT29 cells. Cells were treated with 5 μM erlotinib for 48 hr.

K. Erlotinib treatment inhibits cell growth in RNF43 KO HT29 cells.

Means ± SEMs are shown. p-values were calculated by two-way ANOVA uncorrected Fisher’s LSD test (A, H), Welch’s t-test (B, D-G), one-way ANOVA uncorrected Fisher’s LSD test (K). n.s., not significant.

ZNRF3/RNF43 loss enhances EGFR signaling and promotes tumorigenesis.

A. Overexpression of ZNRF3 suppresses HT29 tumor growth in vivo. Representative bioluminescence images (top panel) and quantification (bottom panel) of flank-injected HT29 cells expressing either GFP or Myc-tagged ZNRF3.

B. Overexpression of ZNRF3 inhibits P-EGFR and total EGFR levels in HT29 tumors. Representative images (left panel) and quantification of P-EGFR (right-top panel) and total EGFR (right-bottom panel) protein levels are shown.

C. Representative H&E images of prostate tissues from WT or prostate-specific Znrf3/Rnf43 knockout mice. Prostate tissue or tumor samples were collected at 1-year-old. Scale bar=300 μm.

D. Total pathological scores of prostate tissues from WT or prostate-specific Znrf3/Rnf43 knockout mice. n=6 mice per group.

E. Representative images of immunochemistry staining for EGFR, P-EGFR, and active-β-Catenin in prostate tissues from WT or prostate-specific Znrf3/Rnf43 knockout mice. Scale bar=40 μm. Means ± SEMs are shown. p-values were calculated by two-way ANOVA uncorrected Fisher’s LSD test (A), Welch’s t-test (B), Mann-Whitney (D).

EGFR protein level is negatively associated with ZNRF3/RNF43 mRNA expression in cancers.

A. Scatterplot of EGFR protein level (signal) versus ZNRF3 mRNA expression (RSEM, Log2 (Val+1)) using the TCGA adrenal cortical carcinoma (ACC) dataset.

B. Scatterplot of EGFR protein level (z-normalized) versus ZNRF3 mRNA expression (z- normalized) using the TCGA colorectal adenocarcinoma (CO/READ) dataset.

C. Scatterplot of EGFR protein level (z-normalized) versus ZNRF3 mRNA expression (z- normalized) using the CPTAC colon adenocarcinoma (COAD) dataset.

D. Scatterplot of EGFR protein level (z-normalized) versus RNF43 mRNA expression (z- normalized) using the TCGA colorectal adenocarcinoma (CO/READ) dataset.

E. Scatterplot of EGFR protein level (z-normalized) versus RNF43 mRNA expression (z- normalized) using the CPTAC colon adenocarcinoma (COAD) dataset.

F. Boxplot of EGFR protein levels (signal) in human colorectal adenocarcinomas expressing RNF43 WT or RNF43 G659Vfs*41, using the TCGA dataset.

G. Boxplot of EGFR protein levels (signal) in human stomach adenocarcinomas (STAD) expressing RNF43 WT or RNF43 G659Vfs*41, using the TCGA dataset.

ZNRF3/RNF43 negatively regulates EGFR protein level.

A. Overexpression of RNF43 or ZNRF3 decreases P-EGFR, total EGFR, and FZD5 protein levels in 293T cells. GFP, EGFR, Flag-FZD5 constructs were co-transfected with E3 ligase constructs or vector control. CBL overexpression serves as a positive control for EGFR downregulation. RNF43 or ZNRF3 were expressed with their native signal peptides or CD8 signal peptide.

B. Overexpression of RNF43 or ZNRF3 has no impact on TGF-β receptor I (TGFβRI) protein level in 293T cells. TGFβRI-Flag construct was co-transfected with RNF43, ZNRF3 or vector control.

C. Overexpression of ZNRF3 or RNF43 does not decrease FGFR1 protein level in 293T cells. Flag-FGFR1 construct was co-transfected with ZNRF3, RNF43, or vector control.

D. Expression of Cas9-CRISPR RNF43 guide RNA enhances EGFR protein level in HT29 cells, shown by representative western blot images (left panel) and quantification results (right panel).

E. Expression of Cas9-CRISPR RNF43 guide RNA or ZNRF3 guide RNA enhances both EGFR and HER2 protein levels in HCC1954 cells.

F. RSPO2 treatment (50 ng/ml, 2-4 hr) enhances EGFR and LRP6 protein levels in 293T cells infected with lentivirus expressing FUCGW-EGFR.

G. Overexpression of RSPO2 WT does not enhance EGFR protein level in HT29 RNF43 knockout cells.

Means ± SEMs are shown. p-values were calculated by one-way ANOVA uncorrected Fisher’s LSD test. n.s., not significant.

RNF43 loss decreases EGFR ubiquitination.

A. Depletion of RNF43 by CRISPR/Cas9 decreases EGFR ubiquitination in MDA-MB-231 cells. EGFR ubiquitination was examined by Ub IP followed by EGFR IB.

B. RNF43 downregulates HER2 protein level through the RING domain. 293T cells were co- transfected with HER2 and Vector, RNF43 WT or ΔRING mutant.

ZNRF3/RNF43 interacts with EGFR

A. Relative abundance of the EGFR-associated proteins. BGC823 cells were starved overnight and then treated with 50 ng/ml EGF for 0 min (control) or 120 mins (EGF-stimulated). Endogenous EGFR was immunoprecipitated by anti-EGFR antibodies. EGFR-associated proteins with over 10^5 iBAQ (intensity Based Absolute Quantification) and over 2-fold increase in iBAQ after EGF stimulation were plotted. The areas of the circles indicate the abundance of iBAQ of EGFR- associated proteins obtained in EGF-stimulated IPs. The y axis indicates the fold change of iBAQ of EGFR-associated proteins identified in EGF-stimulated versus control in the log2 scale, which are arranged from low to high along the x axis.

B. Immunofluorescence staining for HA-tagged RNF43 FL and RNF43 ECD-TM in 293T cells transfected with RNF43 constructs. Scale bar=40 μm.

C. RNF43 ECD-TM interacts with EGFR. 293T cells were co-transfected with EGFR and HA- tagged RNF43 constructs. EGFR interaction with RNF43 FL and RNF43 ECD-TM were examined by HA-tag IP followed by EGFR IB.

The protease associate domain of ZNRF3/RNF43 is dispensable for EGFR interaction.

A. Schematic diagram of tagged ZNRF3/RNF43 wild-type and mutant proteins. SP, signal peptide; WT, wild-type; PA, protease associate domain; TM, transmembrane domain; RING, E3 ligase RING domain.

B, C. EGFR interacts with the ΔPA mutant of Myc-tagged ZNRF3 (B) and HA-tagged RNF43

(C). 293T cells were co-transfected with EGFR and ZNRF3/RNF43 WT or ΔPA constructs.

D. Overexpression of ZNRF3 WT or ΔPA mutant inhibits canonical WNT signaling in 293T cells in Super Top-Flash assay. ZNRF3 ΔRING mutant serves as a negative control.

Means ± SEMs are shown. p-values were calculated by one-way ANOVA uncorrected Fisher’s LSD test. n.s., not significant.

RPPA identifies EGFR downstream signaling molecules upregulated by RNF43 knockout in HT29 cells.

Pathological assessment on eccentric thicken wall (A), dysplasia (B), micro-invasion (C), and frank invasion (D) of WT and Znrf3/Rnf43 KO mouse prostate tissues.

qPCR analysis for TGF-β signaling relevant genes in Apcmin mouse intestinal tumor organoids cultured with or without RSPO1 supplements. Genes with no significant changes after RSPO1 treatment were plotted in grey, genes significantly down-regulated after RSPO1 treatment were plotted in blue. Means ± SEMs are shown. Welch’s t-test was used to assess statistical significance. **, p-value < 0.01.