ERAD-dependent control of the Wnt secretory factor Evi

Abstract

Active regulation of protein abundance is an essential strategy to modulate cellular signaling pathways. Within the Wnt signaling cascade, regulated degradation of β-catenin by the ubiquitin-proteasome system (UPS) affects the outcome of canonical Wnt signaling. Here, we found that abundance of the Wnt cargo receptor Evi (Wls/GPR177), which is required for Wnt protein secretion, is also regulated by the UPS through endoplasmic reticulum (ER)-associated degradation (ERAD). In the absence of Wnt ligands, Evi is ubiquitinated and targeted for ERAD in a VCP-dependent manner. Ubiquitination of Evi involves the E2-conjugating enzyme UBE2J2 and the E3-ligase CGRRF1. Furthermore, we show that a triaging complex of Porcn and VCP determines whether Evi enters the secretory or the ERAD pathway. In this way, ERAD-dependent control of Evi availability impacts the scale of Wnt protein secretion by adjusting the amount of Evi to meet the requirement of Wnt protein export. As Wnt and Evi protein levels are often dysregulated in cancer, targeting regulatory ERAD components might be a useful approach for therapeutic interventions.

Keywords: CGRRF1; Evi/Wls/GPR177; Porcn; Wnt signaling; regulatory ERAD.

Figure EV1. Evi is not transcriptionally regulated by Wnt

1. A, B

   FPKM‐normalized RNA‐seq. data of the TCGA Research Network (TCGA‐COAD; http://cancergenome.nih.gov/; 09/25/2017) were log‐transformed to illustrate the relative expression of (A) Wnt3 and (B) Evi in healthy colon (41) versus colon adenocarcinoma (456). The distribution into tumor and healthy samples was based on their barcodes as described in TCGA Wiki. Statistical significance of gene expression differences was determined using a Student's t‐test under the alternative hypothesis H₁ that gene expression is higher in tumors compared to healthy tissue. The boxplot diagram shows the median as line within the box, the 25^th and 75^th percentiles as the upper and lower part of the box, the 10^th and 90^th percentiles as error bars and outliers as circles.

2. C

   HEK293T cells were transfected with the indicated expression constructs, treated with 100 ng/ml recombinant mouse Wnt3A (rec. W3A) or with 10 μM GSK3 inhibitor SB216763 for the indicated hours (h). AXIN2 and Evi mRNA levels were analyzed by qRT–PCR and normalized to GAPDH expression. Results are shown as mean ± s.d. from three independent experiments.

3. D, D′

   Twenty‐four hours after reverse transfection with Ctrl or CTNNB1 siRNA, HEK293T cells were transfected with Wnt3A or IGFBP5‐V5 expression plasmids and analyzed (D) for the indicated proteins via immunoblotting or (D′) for canonical Wnt activity using the TCF‐Luciferase Wnt reporter assay. Immunoblotting is representative of three independent experiments, and Wnt reporter activity was calculated as mean from three independent experiments ± s.d.

Source data are available online for this figure.

Figure 1. Wnt ligand production increases Evi protein levels

1. In situ RNA hybridization (red dots) and immunohistochemistry (brown staining) of Evi were performed on sequential FFPE tissue slides of healthy colon and matched colon cancer tissue from five patients. The illustrated example is representative for three patients. Scale bar: 40 μm. Specificity of Evi probes was confirmed in Appendix Fig S1B.

2. Wild‐type (wt) or Evi^KO HEK293T cells were transfected with Wnt3A or IGFBP5‐V5 expression plasmids and subjected to Western blot analysis. Specific Evi bands are indicated by arrows and unspecific bands by asterisks. Endogenous Evi is not only detectable as a monomeric form (46 kDa) but also as SDS‐resistant dimers (80 kDa). Clonal Evi^KO HEK293T cells were generated via CRISPR/Cas9 using Evi sgRNA #2 (Evi^KO2.9) or Evi sgRNA #1 (Evi^KO1.1; Appendix Fig S2).

3. HEK293T cells were transfected with Wnt expression plasmids and analyzed for endogenous Evi levels by immunoblotting with a mouse monoclonal Evi antibody (Biolegend, clone YJ5).

4. HEK293T cells were transfected with the indicated overexpression constructs, treated with 100 ng/ml recombinant mouse Wnt3A (rec. W3A) or with 10 μM GSK3β inhibitor SB216763 for the indicated hours (h). The obtained cell lysates were used for Western blot analysis. Representative Western blots of three independent experiments are shown. β‐Actin or N‐cadherin were used as a loading control, and LRP6 served as a reference membrane protein.

5. Scheme showing that Evi is regulated through Wnt proteins within the Wnt‐producing cell. Canonical Wnt signaling can be activated by Wnt ligands, Dishevelled (Dvl) overexpression or by the GSK3β inhibitor SB216763.

Source data are available online for this figure.

Figure 2. Evi stabilization is dependent on Wnt palmitoylation

1. Schematic illustration of the Porcn‐mediated Wnt palmitoylation, which is important for Evi‐Wnt interaction and which is blocked upon Porcn inhibition (LGK974), in Porcn^KO cells and by using a palmitoylation‐deficient S209A Wnt3A mutant.

2. Wild‐type or stable Wnt3‐and Wnt5B‐expressing HEK293T cells were treated with 5 μM LGK974 for 48 h and subjected to Western blot analysis.

3. Western blot analysis of endogenous Evi in wt, Porcn^KO, or Evi^KO HEK293T cells upon overexpression of Wnt3A or IGFBP5‐V5. Porcn^KO1.2 and Porcn^KO1.4 indicate clone #2 and clone #4 of Porcn^KO HEK293T cells generated with Porcn sgRNA1 (Appendix Fig S3). Clonal Evi^KO HEK293T cells were generated with Evi sgRNA2 (Evi^KO2.9; clone #9) or Evi sgRNA1 (Evi^KO1.1; clone #1; Appendix Fig S2). Increase in total β‐catenin protein served as control for Wnt pathway activation.

4. Western blot analysis of endogenous Evi in HEK293T cells transfected with the indicated overexpression plasmids. When indicated, the cells were additionally treated with 5 μM LGK974 for 48 h.

5. Western blot analysis of endogenous Evi in HCT116 or A375 cells treated with 5 μM LGK974 or DMSO for the indicated hours (h). All Western blots are representative of three independent experiments. β‐Actin was used as a loading control, LRP6 as a reference membrane protein and EGF‐Myc and IGFBP5‐V5 as controls for secreted proteins. Specific Evi bands are indicated by arrows, and unspecific bands are marked by asterisks.

6. Scheme: Wnt‐induced Evi stabilization is blocked in the absence of Wnt palmitoylation (Porcn^KO, LGK974, Wnt3A S209A).

Source data are available online for this figure.

Figure EV2. Porcn inhibition does not affect Evi transcription

1. A

   Wild‐type or stable Wnt3‐ and Wnt5B‐expressing HEK293T cells were treated with 5 μM LGK974 for 48 h, and AXIN2 and Evi mRNA levels were quantified by qRT–PCR. Results are shown as mean ± s.d. from three independent experiments.

2. B

   Wild‐type (wt), Porcn^KO1.2, and Porcn^KO1.4 HEK293T cells were transfected with the indicated overexpression constructs and analyzed in Wnt Luciferase activity assays. Results are representative of three independent experiments and shown as mean of seven technical replicates ± s.d.

3. C, D

   HCT116 (C) or A375 cells (D) were treated with 5 μM LGK974 for the indicated hours (h). Evi mRNA was quantified by qRT–PCR and normalized to GAPDH. Results are shown as mean ± s.d. from three independent experiments.

Source data are available online for this figure.

Figure EV3. Evi is ubiquitinated and degraded from the ER by the proteasome

1. HEK293T cells were transfected with the indicated expression constructs and treated with Bafilomycin A (Bafilo) in the indicated concentrations for 24 h. Lysosomal inhibition additionally increased Evi protein on top of Wnt3A expression indicating that Evi is degraded by the lysosome also in the presence of Wnt proteins.

2. HEK293T cells were transfected with Wnt3A or IGFBP5‐V5 expression plasmids and treated for 24 h with DMSO, MG132, or bortezomib in the indicated concentrations.

3. Following MG132 treatment (1 μM for 24 h), poly‐ubiquitinated proteins were precipitated using TUBE2 agarose and incubated with DUB buffer, supplemented with the catalytic domain of the DUB enzyme USP2, if indicated. The precipitates were assayed for endogenous Evi or K48 poly‐ubiquitin. Ctrl agarose beads were used to confirm specificity of the TUBE2 assay for ubiquitinated proteins.

4. HEK293T cells were transfected with the indicated overexpression constructs and treated with 5 μM LGK974 for 48 h, if indicated. Secreted Wnt3A proteins were precipitated from conditioned medium and analyzed via immunoblotting. Compared to wild‐type Wnt3A, Wnt3A‐KDEL is not secreted into the medium affirming cellular retention. HSC70 served as loading control. All Western blots are representative of three independent experiments.

5. Upon VCP knockdown in HEK293T cells using single or pooled siRNAs, Evi and VCP mRNA were analyzed via qRT–PCR. mRNA levels are shown as mean ± s.d. relative to GAPDH mRNA levels from three independent experiments.

Source data are available online for this figure.

Figure 3. Evi poly‐ubiquitination is regulated by the presence of Wnt proteins

1. HEK293T cells were transfected with Wnt3A or IGFBP5‐V5 expression constructs and treated with MG132 at the indicated concentrations for 24 h. Cell lysates were analyzed for endogenous Evi by immunoblotting. Total β‐catenin protein was used to assess MG132 efficiency.

2. Wild‐type (wt), stable Wnt3‐expressing, or Evi^KO2.9 HEK293T cells were treated with 1 μM MG132 for 24 h. TUBE2 immunoprecipitates were assayed for endogenous Evi or K48 poly‐ubiquitin by immunoblotting. To confirm specificity of the TUBE2 assay, Ctrl agarose beads were used as control. The asterisk at the β‐actin blot indicates Wnt3A proteins blotted before membrane stripping.

3. Scheme illustrating ubiquitination and proteasomal degradation of Evi, which is blocked in the presence of Wnt ligands.

4. HEK293T cells were transfected with the indicated plasmids and additionally treated with 5 μM LGK974 for 48 h when indicated. In case of Wnt3A‐KDEL, the ER‐retaining sequence KDEL was C‐terminally fused to Wnt3A. Dvl2‐HA, Dvl3, and β‐catenin‐YFP overexpression was used as negative control to verify that Evi stabilization was not due to downstream activation of Wnt signaling. All Western blots are representative of three independent experiments. β‐Actin was used as loading control. Specific Evi bands are marked by arrows and unspecific bands by asterisks.

Source data are available online for this figure.

Figure 4. The ERAD‐protein VCP is required for Evi destabilization

1. ERAD‐dependent degradation requires recognition, ubiquitination, and subsequent dislocation of the selected substrates through the ER membrane to cytosolic proteasomes. ATP‐dependent dislocation is mediated through the AAA‐type ATPase VCP.

2. Western blot analysis of endogenous Evi and the indicated proteins in HEK293T cells upon knockdown of VCP using single or pooled VCP siRNAs.

3. HEK293T cells were treated for 24 h with the VCP inhibitor NMS‐873 at the indicated concentrations and analyzed for the indicated proteins by immunoblotting.

4. Following reverse transfection with pooled VCP or Ctrl siRNAs (2.5 nM or 25 nM), stable Wnt3‐expressing HEK293T cells were treated with 5 μM LGK974 or DMSO for 48 h and analyzed for the indicated proteins.

5. After reverse transfection with pooled VCP siRNAs (if indicated), HEK293T cells were treated with 50 μM cycloheximide (CHX) for the indicated hours (h). Cell lysates were analyzed by immunoblotting.

6. Evi^KO2.9 or wild‐type (wt) HEK293T cells were reverse transfected with pooled VCP or Ctrl siRNAs and analyzed for ubiquitinated Evi and K48 ubiquitinated proteins by TUBE2 precipitation. All Western blots are representative of three independent experiments. β‐Actin was used as a loading control and LRP6 as a reference membrane protein involved in Wnt signaling. Specific Evi bands are indicated by arrows and unspecific bands by asterisks.

Source data are available online for this figure.

Figure 5. Evi forms a complex with Porcn and VCP simultaneously

1. A

   After VCP knockdown (if indicated) and expression of the indicated proteins in wild‐type (wt) or Evi^KO HEK293T, a FLAG IP was performed and analyzed by Western blotting.

2. B, C

   Wild‐type (wt) or Evi^KO HEK293T cells were transiently transfected with the indicated overexpression constructs and treated with 1.5 μM NMS‐873 for 24 h. (B) GFP co‐immunoprecipitation of GFP‐tagged wt VCP and VCP‐DKO was followed by Western blot analysis. (C) Following FLAG IP, interacting proteins were eluted from the beads (FLAG IP) and subjected to GFP pull‐down (double IP) and analyzed by immunoblotting. All Western blots are representative of three independent experiments. β‐Actin served as loading control. Specific Evi bands are indicated by arrows and unspecific bands by asterisks.

Source data are available online for this figure.

Figure EV4. Specific interaction of Porcn and VCP with endogenous Evi

1. A–C

   Twenty‐four hours after reverse transfection with pooled VCP siRNA, wt or Evi^KO2.9 HEK293T cells were transfected with the indicated overexpression plasmids. Following (A, C) FLAG IP of FLAG‐Porcn or (B) GFP IP of GFP‐tagged VCP versions, the interacting proteins and the input were analyzed by blotting with the corresponding antibodies. VCP‐DKO indicates a catalytically dead mutant form of VCP‐containing mutations in both ATPase domains (E305Q/E578Q). Representative Western blots are shown of three (A, C) or one (B, N‐cadherin blot) experiments. β‐Actin served as loading control.

2. D

   Model of a preformed VCP‐Porcn complex, which waits for newly translated Evi guiding it either into ERAD or the secretory pathway.

Source data are available online for this figure.

Figure 6. UBE2J2 and CGRRF1 are involved in the degradation and ubiquitination of Evi

1. A

   Schematic illustration of UBE2J2 and CGRRF1. The transmembrane (TM) domain of CGRRF1 was predicted using TMHMM server v. 2.0 (Krogh et al, 2001). The core catalytic domain of the E2 conjugating enzyme UBE2J2 (amino acid residues 14–127) is depicted as UBCc. Amino acid residues important for the interaction with E3 ligases are shown in green, residues facing the ubiquitin thioester interaction side in blue, and the active site cysteine in red. The catalytic zinc finger domain of CGRRF1 consists of the typical C₃HC₄ RING type motif and spans the amino acid residues 274–315. Active site cysteine residues are marked in red and the active site histidine in orange.

2. B, C

   Following reverse transfection of HEK293T cells with pooled siRNA against the indicated proteins, the cell lysates were analyzed by immunoblotting.

3. D

   Following UBE2J2 and CGRRF1 knockdown, HEK293T were treated with 1 μM MG132 for 24 h. Poly‐ubiquitinated proteins were TUBE2 precipitated and stained for endogenous Evi or K48 poly‐ubiquitin. Ctrl agarose beads were used to confirm specificity of the TUBE2 assay for ubiquitinated proteins.

4. E

   CGRRF1‐FLAG‐HA (FH) and Hrd1‐FLAG‐HA (FH) inducible Flp‐IN HEK293T cells were treated with 1 ng/ml doxycycline for 18 h and 10 μM MG132 for 4 h prior to FLAG IP and subsequent immunoblotting with mouse monoclonal Evi antibody (clone YJ5). All Western blots are representative of three independent experiments. β‐Actin was used as loading control. Specific Evi bands are indicated by arrows and unspecific bands by asterisks.

Source data are available online for this figure.

Figure EV5. The E2 UBE2J2 and the E3 CGRRF1 are involved in Evi regulation

1. A

   Following reverse transfection of HEK293T cells with siCtrl or pooled siRNAs against VCP and the indicated E2 and E3 enzymes, knockdown was confirmed by qRT–PCR. mRNA levels are relative to GAPDH and shown as mean ± s.d. from three independent experiments.

2. B, C

   Western blot analysis of endogenous Evi in HEK293T cells upon knockdown of (B) UBE2J2 or (C) CGRRF1 using single or pooled siRNAs. All Western blots are representative of three independent experiments. β‐Actin served as loading control.

3. D, E

   HEK293T cells were transfected with the indicated overexpression plasmids (FH: FLAG‐HA tag) and subjected to FLAG IP. The interacting proteins and the input were analyzed by immunoblotting with the corresponding antibodies. Western blots are representative of three independent experiments.

4. F

   Schematic model illustrating the involvement of the E2 conjugating enzyme UBE2J2 and the RING E3 ubiquitin ligase CGRRF1 in conjugating ubiquitin (Ub) to Evi.

5. G, H

   FPKM‐normalized RNA‐seq data of the TCGA Research Network (http://cancergenome.nih.gov/; 09/25/2017) were log‐transformed to illustrate the relative expression of (G) CGRRF1 and (H) Evi in healthy colon (41) versus colon adenocarcinoma (TCGA‐COAD, 456 samples) and in healthy endometrium (35) versus endometrial carcinoma (TCGA‐UCEC; 552 samples). Statistical significance of gene expression differences was determined using a Student's t‐test under the null hypothesis H₀. The boxplot diagram shows the median as line within the box, the 25^th and 75^th percentiles as the upper and lower part of the box, the 10^th and 90^th percentiles as error bars and outliers as circles.

Source data are available online for this figure.

Figure 7. Wnt secretion is modulated by ERAD

1. A–C

   Twenty‐four hours after reverse transfection with Ctrl siRNA or pooled siRNAs against VCP, UBE2J2, or CGRRF1, HEK293T cells were transfected with (A) Wnt5A or (B, C) Wnt3A expression plasmids. Using Blue Sepharose, secreted Wnt proteins were precipitated from conditioned medium 24 h after plasmid transfection. HSC70 also binds to Blue Sepharose and served as loading control. Western blot quantification of three (B) and six (C) replicates is shown in Appendix Fig S4A and B.

2. D

   Model of ERAD‐dependent regulation of Wnt protein secretion. In the absence of Wnt proteins (− Wnt production), Evi is poly‐ubiquitinated and degraded via ERAD, which involves the E3 ligase CGRRF1 and the E2 conjugating enzyme UBE2J2. Poly‐ubiquitinated Evi is translocated via the ATPase VCP into the cytosol and degraded by the proteasome. In the presence of mature Wnt proteins (+ Wnt production), Evi is directed through Porcn into the Wnt secretory pathway and not degraded via ERAD. Stabilized Evi is capable to cope with the need of increased Wnt protein secretion. Having reached the plasma membrane, Evi can be recycled back to the Golgi and ER and Wnt proteins are secreted directly or after endocytosis through exovesicles. Despite the existence of a complex containing Evi, Porcn, and VCP, Porcn and VCP do not necessarily need each other to exert their functionality.

Source data are available online for this figure.