pVHL inhibits Wnt/β-catenin signaling and promotes TCF/LEF protein degradation

Our previous study reported that VBP1 regulates TCF/LEF stability through a pVHL-dependent mechanism [19]. To elucidate the underlying mechanism by which pVHL regulates TCF/LEF protein stability, we investigated the effects of pVHL on Wnt-induced TCF/LEF-dependent transcriptional activity. The HCT116 colorectal carcinoma cell line harbors a stabilized mutation in β-catenin that constitutively activates the Wnt/β-catenin signaling pathway [28]. A TOPFlash reporter plasmid, which contained Wnt-responsive TCF/LEF binding sites, was transfected into HCT116 cells, and then the transcriptional activity was measured. We observed that pVHL decreased expression of the TOPFlash reporter in HCT116 cells in a dose-dependent manner (Fig. 1A). These findings suggest that pVHL regulates Wnt signaling downstream of β-catenin. We next investigated whether pVHL modulates Wnt/β-catenin signaling at the TCF/LEF level. To this end, we utilized HEK293T cells, which possess a functional Wnt signaling system with basal Wnt/β-catenin signaling activity (referred to as"Wnt-off"state) [28], were co-transfected with pVHL and VP16-Tcf7l1ΔN, a constitutively active fusion protein derived from Tcf7l1 that lacks the β-catenin-binding domain and is therefore β-catenin-independent. Subsequently, Wnt reporter activity was evaluated. In agreement with the inhibitory effect of VBP1 on VP16-Tcf7l1ΔN-induced Wnt reporter activity previously reported [19], pVHL also inhibited VP16-Tcf7l1ΔN-induced Wnt reporter activity in a dose-dependent manner (Fig. 1B). These results suggested that pVHL inhibits Wnt reporter activity at the TCF/LEF level.

Fig. 1

pVHL inhibits Wnt/β-catenin signaling and destabilizes TCF/LEF protein (A) TOPFlash luciferase assays in HCT116 cells with increasing pVHL overexpression. Expression of Flag-pVHL was confirmed by western blotting. Values are mean ± S.D. (n = 3). One-way ANOVA analysis with Dunnett's multiple comparisons test, **p < 0.01; ***p < 0.001. (B) TOPFlash assays in VP16-Tcf7l1ΔN-treated HEK293T cells with increasing pVHL overexpression. Expression of Flag-pVHL was confirmed by western blotting. Wnt/β-catenin signal was activated by transfection with VP16-Tcf7l1ΔN plasmid DNA (50 ng). Expression of Flag-pVHL was confirmed by western blotting. Values are mean ± S.D. (n = 3). One-way ANOVA analysis with Dunnett's multiple comparisons test, **p < 0.01; ***p < 0.001; ****p < 0.0001. (C) TOPFlash luciferase assays in VHL-knockout HEK293T cells. pVHL protein levels were confirmed by western blotting. Values are mean ± S.D. (n = 3). Unpaired t-test, *p < 0.05; **p < 0.01; ***p < 0.001. (D) TOPFlash luciferase assays in BIO-treated VHL-knockout HEK293T cells. TOPFlash plasmid was cotransfected with Renilla plasmid into control or VHL-knockout cells. Wnt/β-catenin activity was induced by BIO (1 μM) for 4 h. Values are mean ± S.D. (n = 3). Unpaired t-test, *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. (E) The transcriptional levels of Wnt target gene AXIN2, NKD1, and CCND1 in VHL-knockout HEK293T cells were analyzed by qRT-PCR. Values are mean ± S.D. (n = 3). Unpaired t-test, ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001. (F) The transcriptional levels of Wnt target gene AXIN2 and NKD1 in pVHL-overexpressing 786-O cells were analyzed by qRT-PCR. Values are mean ± S.D. (n = 3). Unpaired t-test, ns, not significant; *p < 0.05; **p < 0.01. (G) Endogenous TCF protein levels in HEK293T cells with increasing pVHL overexpression. Western blot analysis detected two distinct TCF7L2 isoforms: TCF7L2E and TCF7L2M/S. (H) Reintroduction of Flag-pVHL downregulated TCF7 and TCF7L2 in 786-O cells. (I) Flag-pVHL promotes endogenous TCF7L2 degradation in HEK293T cells. HEK293T cells were transfected with empty vector or Flag-pVHL, after 48 h, treated with cycloheximide (CHX; 100 μg mL.−1), and harvested at indicated time points (0, 2, 4, 8 h). Quantification of the total protein levels of the two TCF7L2 isoforms, TCF7L2E and TCF7L2M/S were normalized to Histone H3 (right panel). Values are mean ± S.D. (n = 3). Two-way ANOVA analysis with Bonferroni's multiple comparisons test, ns, not significant; *p < 0.05; ***p < 0.001. (J) The protein levels of TCF/LEF in control and VHL-Knockout cells. The expression level of HIF-1α was used as a positive control. Relative protein level normalized to Histone H3 (lower panel). Values are mean ± S.D. (n = 3). Unpaired t-test, *p < 0.05; **p < 0.01; ***p < 0.001. (K) The transcriptional levels of TCF/LEF in control and VHL-knockout cells were analyzed by qRT-PCR. Values are mean ± S.D. (n = 3). Unpaired t-test, ns, not significant. (L) Introduction of Flag-pVHL into VHL-knockout HEK293T cells downregulated TCF/LEF protein levels. Relative protein level normalized to Histone H3 (lower panel). SE, short time of exposure; LE, long time of exposure. Values are mean ± S.D. (n = 3). Unpaired t-test, *p < 0.05; **p < 0.01; ***p < 0.001

To further test the effect of pVHL on TCF/LEF, VHL-knockout HEK293T cells were generated using a CRISPR/Cas9-mediated gene editing approach (Fig. S1A and B). Two pVHL protein isoforms, a long form and a short form, have been previously reported [29]. The VHL-knockout cell lines were established, harboring either premature termination codons in exon 1 (lines #1–4) or a large deletion (line #5). Both types of mutations result in the depletion of both isoforms of VHL (Fig. 1C). Knockout of pVHL enhanced basal Wnt reporter activity (Fig. 1C). The addition of a GSK3 inhibitor, 6-bromoindirubin-3’-oxime (BIO), in HEK293T cells could induce Wnt reporter activity. Under BIO treatment, knockout of pVHL further increased Wnt reporter activity albeit with differently response degrees (Fig. 1D). To confirm this result, we further evaluated the transcriptional levels of Wnt target genes in several pVHL-depleted cell clones that exhibited relatively lower Wnt reporter activity responsiveness under BIO treatment. Indeed, the knockout of pVHL resulted in a significant increase in the mRNA levels of Wnt target genes, including AXIN2, NKD1, and CCND1, in these clones (Fig. 1E). This result further indicated that knockout of pVHL enhanced Wnt/β-catenin signaling activity. In addition, most cases of ccRCCs are associated with the inactivation of VHL [30], we therefore assessed the response of the VHL-deficient 786-O cell line to BIO treatment. BIO treatment significantly increased the mRNA expression levels of AXIN2 and NKD1 in 786-O cells; however, the reintroduction of pVHL attenuated this effect (Fig. 1F). Taken together, these results suggested that pVHL inhibits Wnt/β-catenin signaling.

We speculate that pVHL may promote VP16-Tcf7l1ΔN protein degradation and thus prevent its ability to induce the Wnt reporter activity. To test this hypothesis, we co-transfected Flag-tagged pVHL with Myc-tagged Tcf7, Tcf7l1, Tcf7l2, and Lef1 into HEK293T or HCT116 cells, respectively. The overexpression of pVHL reduced the abundance of Tcf7, Tcf7l1, Tcf7l2, and Lef1 in both cell lines (Fig. S2A and B). Additionally, pVHL also decreased the endogenous protein levels of TCF/LEF family members, including TCF7, TCF7L1, and both major TCF7L2 isoforms (TCF7L2E and TCF7L2M/S) in the HEK293T cells in a dose-dependent manner (Fig. 1G). Likewise, the reintroduction of pVHL into naturally VHL-deficient 786-O cells led to a reduction in the endogenous protein levels of TCF7 and TCF7L2 (Fig. 1H). Therefore, pVHL suppresses Wnt/β-catenin signaling by downregulating TCF/LEF proteins, indicating that this mechanism may play a role in the pathogenesis of ccRCC following the loss of pVHL. Taken together, these results suggested that pVHL negatively regulates Wnt/β-catenin activity and the TCF/LEF protein level.

To further investigate whether pVHL promotes the degradation of TCF/LEF proteins, we conducted a time-course treatment assay using cycloheximide (CHX), a protein synthesis inhibitor. When Myc-tagged pVHL was transfected, the degradation of endogenous TCF7L2 was significantly accelerated (Fig. 1I). In contrast, knockout of pVHL significantly increased not only HIF-1α, but also TCF7, TCF7L1, and TCF7L2 protein levels (Fig. 1J). However, knockout of pVHL did not alter the mRNA levels of TCF7, TCF7L1, TCF7L2, or HIF-1α (Fig. 1K). Moreover, reintroduction of pVHL into VHL-knockout HEK293T cells neutralized this effect, as the TCF7, TCF7L1, TCF7L2, and HIF-1α protein levels were reduced (Fig. 1L). These results indicate that the VHL-knockout is specific and that pVHL promotes the degradation of TCF/LEF proteins.

pVHL interacts with TCF/LEF

We next examined whether TCF/LEF and pVHL interact with each other. Co-IP assay showed that endogenous pVHL interacts with all four TCF/LEF members (Fig. 2A). In addition, co-IP assay also indicated that endogenous TCF7L2 retrieved endogenous pVHL in HEK293T cells (Fig. 2B). Furthermore, purified glutathione-S-transferase (GST)-pVHL protein pulled down all four members of Myc-tagged TCF/LEF in vitro (Fig. 2C). Likewise, a protein–protein interaction between pVHL and endogenous TCF7L2 was also confirmed (Fig. 2D). To further confirm the interaction between pVHL and TCF7L2, surface plasmon resonance analysis was performed using purified recombinant pVHL and TCF7L2(1–456) proteins. As shown in Fig. 2E, pVHL directly binds to TCF7L2(1–456). Collectively, these data revealed that TCF/LEF and pVHL directly interact with each other.

Fig. 2

pVHL directly binds with TCF/LEF (A) Detection of pVHL binding to TCF/LEF in HEK293T cells by Co-IP. Red asterisk indicates the specific band. (B) Co-IP assay revealed the endogenous interaction between TCF7L2 and pVHL in HEK293T cells. (C, D) pVHL pulls down TCF/LEF. Purified GST or GST-pVHL proteins were incubated with extracts of HEK293T cells either transfected with Myc-Tcf/Lef (C) or untransfected (D). Bound proteins were eluted and analyzed by western blot using indicated antibodies. Red asterisk indicates GST-pVHL. (E) Surface plasmon resonance analysis of interactions between pVHL with TCF7L2(1–456) using purified recombinant proteins. (F) Schematic representations of pVHL and truncated mutant proteins. (G and H) Mapping pVHL binding domain interacting with endogenous or exogenous TCF7L2 in transfected HEK293T cells by Co-IP assay. (I) Schematic representation of of TCF7L2 WT and truncated mutant proteins. (J) Mapping TCF7L2 binding domain interacting with pVHL in transfected HEK293T cells by Co-IP assay

To identify the pVHL domain(s) essential for TCF/LEF interaction, various pVHL domain-deleted mutants were generated (Fig. 2F). Using Co-IP assay, we mapped the domain(s) putatively responsible for the interaction between pVHL and TCF7L2. A region comprising amino acid (aa) residues 100–157 was required for its interaction with endogenous TCF7L2 (Fig. 2G). Moreover, structural analysis revealed that the aa (111–120) is essential for pVHL-TCF7L2 binding, as pVHL (1–120), but not pVHL (1–110), was capable of co-immunoprecipitating TCF7L2 (Fig. 2H).

We next mapped the binding domain(s) of TCFL2 to pVHL. A variety of mutants of TCF7L2 were generated based on conserved functional motifs (Fig. 2I and S3). To ensure the nuclear localization of each deletion mutant of TCF7L2, all of the mutants contained an NLS. Co-IP analysis showed that aa 63–201 or aa 328–415, rather than other regions of TCF7L2, binds to pVHL (Fig. 2J and Fig. S4A, B). Immunostaining analysis confirmed that the interactions are specific, as the mutants TCF7L2 (63–201), TCF7L2 (202–327), and TCF7L2 (328–415) are all localized in the nucleus (Fig. S4C). These results suggested that aa 63–201 or aa 328–415 of TCF7L2 is required for interaction with pVHL. The aa 328–415 of TCF7L2 constitutes the HMG DNA-binding domain (HMG DBD), which is composed of HMG domain and NLS. The HMG DBD of TCF/LEF is evolutionarily conserved and nearly identical from invertebrate to vertebrate (Fig. S3) [9]. We therefore investigated whether HMG DBD was downregulated by pVHL. As expected, the Tcf7l1-HMG DBD level was reduced by pVHL overexpression (Fig. S4D). Given that the HMG domain recognizes and binds to specific DNA sequences, we investigated whether pVHL inhibits the binding of TCF/LEF proteins to DNA. Indeed, the ChIP-qPCR analysis revealed that the knockout of pVHL had minimal impact on the binding of TCF7L2 to the target promoter regions of AXIN2 and NKD1 (Fig. S4E). Collectively, our results indicate that aa 63–201 or 328–415 of TCF7L2 is crucial for pVHL binding.

pVHL promotes the proteasomal degradation of TCF/LEF through a mechanism independent of the ubiquitin-mediated pathway

As mentioned earlier, pVHL recognizes and binds to prolyl-hydroxylated substrates, such as prolyl-hydroxylated HIF-α, Akt, and ZHX2, in order to exert its function. Three residues (S111, H115, and W117) in the pVHL hydroxyl-proline binding pocket are critical for pVHL interaction with prolyl-hydroxylated substrates [14]. This triple residue-mutated pVHL was utilized to test whether it has comparable functionality with that of the WT pVHL. Like WT pVHL, the pVHL mutant also downregulated the abundance of Tcf7L2 (Fig. S5A). We next applied a prolyl hydroxylase inhibitor, DMOG, to inhibit the activity of EGLN 1/2/3. It has been reported that DMOG treatment inhibits the binding between HIF-2α and pVHL and therefore stabilizes HIF-2α [14]. However, DMOG treatment did not reverse TCF7L2 protein downregulation induced by pVHL overexpression (Fig. S5B). Therefore, TCF7L2 degradation by pVHL does not depend on prolyl hydroxylation of TCF7L2.

A previous report has indicated that chronic starvation-stimulated autophagy negatively regulates Wnt/β-catenin signaling [31]. We examined the effects of starvation, an autophagy stimulus with nutrient deprivation medium, on the expression of endogenous TCF7L2 in HEK293T cells. Chronic starvation reduced the protein levels of both non-p-β-catenin and total β-catenin but not that of TCF7L2 (Fig. S5C). This result suggested that TCF7L2 is not degraded by autophagy.

To exclude the effect of increased HIFs on TCF7, TCF7L1, and TCF7L2 in VHL-knockout HEK293T cells, we examined the protein levels of TCF7, TCF7L1, and TCF7L2 with enhanced HIF-1α expression upon hypoxia treatment. Hypoxia treatment increased the protein level of HIF-1α, while the protein levels of TCF7, TCF7L1, and TCF7L2 were not increased (Fig. S6A). Dimerization of HIF-1α or HIF-2α with HIF-1β is mediated by their basic helix-loop-helix (bHLH) and PER-ARNT-SIM (PAS) domains, which are required for binding to hypoxia response elements (HREs) and HIF-dependent transcriptional activity [32]. In this case, we generated HIF-1β (ARNT) knockout HEK293T cells, targeting exon 6 to disrupt its bHLH and PAS domains (Fig. S6B and C). Indeed, the cells with absence of HIF-1β did not increase protein levels of TCF7, TCF7L1, and TCF7L2 under hypoxia treatment (Fig. S6D). Therefore, the HIF activity did not upregulate the protein levels of TCF7, TCF7L1, and TCF7L2.

To address the possible pathway of TCF/LEF degradation, we used specific small compound inhibitors, including MG132 (proteasomal inhibitor), NH4Cl (lysosomal proteolysis inhibitor), and 3-MA (autophagy inhibitor), to block the major protein degradation pathway. Addition of MG132 but not of NH4Cl or 3-MA blocked pVHL-mediated TCF7L2 degradation (Fig. 3A). Thus, it is likely that pVHL promotes TCF7L2 degradation via the proteasomal pathway.

Fig. 3

pVHL promotes TCF/LEF degradation by ubiquitin-independent proteasome pathway (A) Changes in endogenous TCF7L2 protein levels in pVHL-overexpressing HEK293T cells treated with indicated inhibitors. The transfected cells were either untreated or treated with MG132 (10 μM), NH4Cl (25 mM), or 3-MA (5 mM) for 8 h. (B) Effects of pVHL-overexpression on Tcf7l2 ubiquitination. Myc-Tcf7l2, Flag-HIF-1α, and HA-Ub were co-transfected with GFP-Vector or pVHL-GFP into HEK293T cells. After 48 h, cells were treated with MG132 (10 μM) for 8 h, and lysed for immunoprecipitation with anti-Myc and anti-Flag antibody. (C) Changes in Tcf7l2-K/R protein levels in pVHL-overexpressing HEK293T cells treated with indicated inhibitors. Western blot analysis of whole cell lysis derived from HEK293T cells transfected with indicated plasmid DNA and either untreated or treated with MG132 (10 μM), NH4Cl (25 mM), or 3-MA (5 mM) for 8 h. (D) Changes in endogenous TCF7L2 or HIF-2α protein levels in pVHL-overexpressing HEK293T cells treated with a specific inhibitor for the ubiquitin activating enzyme or proteasome. The transfected cells were either untreated or treated with TAK243 (1 μM) for 12 h or MG132 (10 μM) for 8 h. (E) Degradation analysis of TCF7L2(1–456) protein in cell-free system. The amount of TCF7L2(1–456) protein degraded by 26S proteasome at each indicated time points (0, 2, 4, 6, 12, 18 h) in cell-free system. The group with MG132 treatment was used as a control. Values are mean ± S.D. (n = 3). One-way ANOVA analysis with Tukey's multiple comparisons test. Groups marked with distinct letters show significant differences from one another (p < 0.05)

Given that pVHL functions as an E3 ubiquitin ligase, we subsequently investigated the effects of pVHL on the ubiquitylation of TCF7L2. Overexpression of pVHL enhanced the polyubiquitination level of HIF-1α but not Tcf7l2 (Fig. 3B). Therefore, during the regulation of TCF7L2 protein stability, pVHL does not exhibit canonical functionality as a conventional E3 ubiquitin ligase, which typically catalyzes the formation of polyubiquitin chains on target proteins. Instead, pVHL likely downregulates TCF7L2 through a mechanism distinct from its regulation of HIF-1α, suggesting a unique mode of action.

We further analyzed evolutionarily conserved lysine residues in several vertebrate TCF/LEF proteins and found approximately 19 conserved lysine residues in total (Fig. S3). We mutated all of them into arginine residues (R) in a Myc-tagged Xenopus Tcf7l2 background (hereafter, Tcf7l2-K/R). Indeed, pVHL also reduced the protein level of the Tcf7l2-K/R mutant. Moreover, addition of MG132 but not of NH4Cl or 3-MA blocked pVHL-mediated Tcf7l2-K/R mutant degradation (Fig. 3C). Collectively, these results suggested that pVHL likely promotes TCF7L2 proteasomal degradation independently of classical lysine-dependent ubiquitin function.

In addition to the classical lysine-dependent ubiquitin modification, lysine-independent N-terminal ubiquitin modification or serine, threonine, cysteine, and tyrosine-dependent ubiquitin modification is also involved in a variety of protein stability regulation. All of the above ubiquitin modifications require ubiquitin activation [33, 34]. To exclude the above non-classical lysine-independent ubiquitin modification, we tested whether reduction of TCF7L2 protein levels under pVHL overexpression were restored by inhibiting activation of ubiquitin. Indeed, addition of TAK-243, a ubiquitin activating enzyme (UAE)-specific inhibitor, restored protein levels of HIF-2α while not TCF7L2 (Fig. 3D). In contrast, addition of proteasomal inhibitor MG132 restored protein levels of both in pVHL-overexpressing cells (Fig. 3D). The results indicate that pVHL likely promotes TCF7L2 degradation in a proteasome-dependent but ubiquitin activation-independent manner.

Some proteins, such as retinoblastoma protein and HIF, have been reported to be degraded through the ubiquitin-independent proteasome pathway. These proteins can be directly degraded when incubated with purified proteasomes in cell-free systems [18, 35]. To further confirm that TCF7L2 undergoes ubiquitin-independent proteasomal degradation, we investigated whether purified TCF7L2 protein with any modifications could be directly degraded by the 26S proteasome in a cell-free system. To this end, the recombinant TCF7L2(1–456) protein was incubated with the purified 26S proteasome in a cell-free system. We assessed the degradation rate of recombinant TCF7L2(1–456) protein in the presence of purified 26S proteasome. The abundance of TCF7L2(1–456) decreased gradually in a time-dependent manner, indicating that TCF7L2(1–456) was directly degraded by the 26S proteasome in the cell-free system (Fig. 3E). Collectively, these findings suggest that the downregulatory effect of pVHL on TCF/LEF is independent of ubiquitin-mediated mechanisms.

The pVHL downregulates the TCF/LEF proteins through a VBC complex-independent pathway

To further validate that pVHL-mediated TCF/LEF degradation does not rely on E3 ubiquitin ligase activity, we tested the effects on TCF/LEF by naturally occurring and cancer-associated pVHL point mutants L158P and R167W and the truncated mutant pVHL (1–157). All three mutants reduced or diminished elongin B/C binding capability and abolished E3 ligase activity [36,37,38]. They all exhibited the same effects on the abundance of Tcf7l2 protein and VP16-Tcf7l1ΔN-induced Wnt reporter activity as WT pVHL (Fig. 4A and B). Likewise, they all reduced the Tcf7l2-K/R mutant protein levels (Fig. 4C). Similar effects were observed when each mutant was co-expressed with Tcf7l1-HMG DBD (Fig. 4D).

Fig. 4

pVHL promotes TCF degradation in an E3 ubiquitin ligase-independent manner (A) Tcf7l2 protein levels in HEK293T cells with overexpression of WT, site-mutated, or truncated pVHL. (B) TOPFlash reporter assays in VP16-Tcf7l1ΔN-transfected HEK293T cells with overexpression of WT, site-mutated, or truncated pVHL. Wnt/β-catenin signal was activated by transfection with VP16-Tcf7l1ΔN (50 ng). Values are mean ± S.D. (n = 3). Unpaired t-test. *p < 0.05; **p < 0.01; ***p < 0.001. (C) Tcf7l2-K/R protein levels in HEK293T cells with overexpression of WT, site-mutated, or truncated pVHL. (D) Tcf7l1-HMG DBD protein levels in HEK293T cells with overexpression of WT, site-mutated, or truncated pVHL. (E) pVHL truncation mutant pVHL (1–157) promotes endogenous TCF7L2 degradation in HEK293T cells. HEK293T cells were transfected with empty vector or Flag-pVHL (1–157), after 48 h, treated with cycloheximide (CHX; 100 μg mL−1) and harvested at indicated time points (0, 2, 4, 8 h). The total protein levels of the two TCF7L2 isoforms, TCF7L2E and TCF7L2M/S, were normalized to Histone H3 (lower panel). Values are mean ± S.D. (n = 3). Two-way ANOVA analysis with Bonferroni's multiple comparisons test. ns, not significant; *p < 0.05; **p < 0.01; ****p < 0.0001. (F) Overexpression of Flag-pVHL and Flag-pVHL (1–157) reduced TCF7, TCF7L1, and TCF7L2 protein levels in VHL-KO cells. HIF-1α was downregulated in VHL-KO after transfection with Flag-pVHL but not with Flag-pVHL (1–157). Relative protein level normalized to Histone H3 (right panel). Values are mean ± S.D. (n = 3). Unpaired t-test, ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001. (G) Overexpression of pVHL or pVHL (1–157) decreased endogenous Tcf7l2 protein level in wide-type zebrafish embryos at 24 hpf. Protein samples of 4 zebrafish embryos were added in each well. (H) vhl-null mutant zebrafish embryos exhibited elevated protein levels of Tcf7l2 at 120 hpf (upper panel). Reintroduction of pVHL or pVHL (1–157) into vhl-null mutant zebrafish embryos reduced Tcf7l2 protein level at 48 hpf (lower panel). Protein samples of 4 zebrafish embryos were added in each well. (I) Changes in endogenous TCF7L2 protein levels in pVHL (1–157)-overexpressing HEK293T cells treated with indicated inhibitors. The transfected cells were either untreated or treated with MG132 (10 μM), NH4Cl (25 mM), or 3-MA (5 mM) for 8 h. (J) The protein levels of TCF/LEF in control or ELOC-Knockdown cells. The expression level of HIF-1α was used as a positive control

We next test the effects of pVHL (1–157) on protein levels of TCF/LEF at endogenous level. As WT pVHL, pVHL (1–157) also significantly downregulated TCF7L2 protein and shortened its half-life (Fig. 4E). Likewise, introduction of pVHL (1–157) into VHL-knockout HEK293T cells remarkably decreased TCF7, TCF7L1, and TCF7L2 protein accumulation by depleting pVHL, while the HIF-1α protein level was reduced by WT pVHL rather than by the pVHL (1–157) mutant (Fig. 4F). Therefore, pVHL (1–157) and WT pVHL had comparable effects on TCF/LEF downregulation. In addition, we used developing zebrafish embryos to determine the effects of human pVHL and pVHL (1–157) on the promotion of Tcf7l2 degradation in vivo. We generated in vitro transcribed GFP, VHL-P2A-GFP, or VHL (1–157)-P2A-GFP mRNA and injected them into zebrafish embryos. The Tcf7l2 protein levels were reduced in the zebrafish embryos injected with either VHL-P2A-GFP or VHL (1–157)-P2A-GFP mRNA (Fig. 4G). Zebrafish pVhl is an ortholog of the short human pVHL isoform [39]. Knockout of pVhl in zebrafish caused accumulation of Tcf7l2 at the larval stage (120 hpf) (Fig. 4H, upper panel). Introduction of human VHL mRNA also decreased Tcf7l2 protein levels in vhl-null mutant background at 48 hpf (Fig. 4H, lower panel). The pVHL (1–157) mRNA had a similar effect (Fig. 4H, lower panel). Additionally, the endogenous protein levels of TCF7L2 were also reduced by pVHL (1–157), and this reduction was blocked by addition of MG132, but not of NH4Cl or 3-MA (Fig. 4I). Taken together, these results suggested that pVHL (1–157) is sufficient to reduced protein levels of TCF/LEF via proteasomal degradation.

We next test this possibility at endogenous level. As mentioned above, pVHL functions as a subunit of an E3 ligase complex to recognize substrates. In this complex, pVHL, Elongin B/C, and RBX1 in association with ubiquitin-conjugated E2 component, are assembled by CUL2 to ubiquitinate pVHL-bound HIF-α proteins [40]. Depletion of Elongin C leads to disassembly of the E3 ligase complex, which resulted in accumulation of HIF-α proteins [41]. We tested whether the proteasomal degradation of TCF/LEF depend on the whole VBC complex as HIF-α proteins. In agreement with previous results, ELOC-knockdown increased HIF-1α proteins; however, the protein levels of TCF7, TCF7L1, and TCF7L2 did not accumulate simultaneously (Fig. 4J). These results implied that the whole VBC complex is not necessary for the maintenance of TCF/LEF protein stability.

The above results prompted us to hypothesize whether any naturally occurring isoforms or homologues of pVHL exist that abolish E3 ligase activity. We noted that a previous study reported the existence of a human pVHL homologue, the pVHL-like protein (pVHLL; also known as pVLP), which lacks the domain responsible for Elongin C binding (corresponding to aa 1–160 in pVHL) (Fig. S7A, B) [42]. Unlike pVHL, pVHLL has an incapability for assembling an E3 ubiquitin ligase complex that consists of Elongin B/C, RBX1, and CUL2. Consequently, pVHLL acts as a dominant-negative pVHL to promote the accumulation of HIF-1α since it binds to HIF-1α independent of prolyl hydroxylation status [42]. We speculate that pVHLL exerts a similar effect on the protein stability of TCF/LEF as both the WT and mutated forms of pVHL. Indeed, endogenous TCF7L2 is associated with Flag-tagged pVHLL in HEK293T cells (Fig. S7C). As expected, the endogenous protein levels of TCF7, TCF7L1, and TCF7L2 were reduced by pVHLL in HEK293T cells (Fig. S7D). These results suggested that pVHLL reduce protein levels of TCF/LEF. Thus, substrate recognition by pVHL as a component of E3 ubiquitin ligase complex is not required for TCF/LEF protein degradation.

Collectively, these data implied that pVHL does not function as an E3 ligase complex adaptor in VBC complex to promote TCF degradation.

pVHL directly interacts with the 26S proteasome to mediate TCF7L2 degradation

The proteasome primarily functions in the degradation of ubiquitin-modified proteins and is also capable of processing non-ubiquitinated substrates [43]. Several proteins, including Parkin, Rad23, and midnolin, are likely to interact with the 26S proteasome and form complexes that mediate substrate degradation [44,45,46,47]. Given that pVHL (1–157) is sufficient to promote the proteasomal degradation of TCF/LEF and that purified, unmodified TCF7L2 protein is directly degraded by the purified 26S proteasome in a cell-free system, we hypothesized that pVHL mediates ubiquitin-independent proteasomal degradation by bridging TCF/LEF to the proteasome, thereby facilitating their degradation. To test this hypothesis, we conducted IP-MS analysis using HEK293T cells expressing Flag-pVHL to determine whether proteins of the 26S proteasome could be enriched by a Flag antibody. The 26S proteasome consists of the 19S regulatory particle and the 20S core particle and each particle contains multiple subunits [43]. As expected, 19 subunit proteins of the 19S regulatory particle and 14 subunit proteins of the 20S core particle were identified through IP-MS (Fig. 5A). Furthermore, we performed Co-IP and GST pull-down assays to validate the IP-MS results using representative proteins identified from the 19S regulatory particle and the 20S core particle, respectively. Co-IP assay showed that PSMD4 of 19S regulatory particle was retrieved by Flag-pVHL (Fig. 5B). In consistent, GST-pulldown assay with HEK293T cell lysis showed that recombinant GST-pVHL pulled down PSMD4 of 19S regulatory particle and PSMA4 of 20S core particle, respectively (Fig. 5C). Moreover, a cell-free system pulldown assay on purified human 26S proteasome and recombinant GST-pVHL also showed that pVHL pulled down PSMD4 or PSMA4 (Fig. 5D). Therefore, pVHL do directly interact with the 26S proteasome.

Fig. 5

pVHL directly interacts with 26S proteasome to promote TCF protein degradation (A) IP-MS analyzing the proteins interacting with pVHL in HEK293T cells. After transfecting pCS2-Flag or pCS2-Flag-pVHL plasmids into HEK293T cells for 48 h, cells were treated with MG132 (10 μM) for 8 h. Flag-pVHL was then immunoprecipitated using anti-Flag antibody and analyzed by mass spectrometry. Coomassie blue staining of Flag immunoprecipitates revealed proteins interacting with Flag-pVHL. Red arrow indicates Flag-pVHL. Mass spectrometry analysis identified multiple subunits of the 26S proteasome, including PSMD4 (19S regulatory particle) and PSMA4 (20S core particle), which are highlighted in red in the table (lower panel). (B) Co-IP assay revealed that Flag-pVHL interacted with 26S proteasome, as indicated by detecting 19S regulatory subunit PSMD4 in HEK293T cells with an antibody against PSMD4. (C) pVHL pulls down 26S proteasome, as indicated by detecting 19S regulatory subunit PSMD4 and 20S core particle PSMA4 in HEK293T cells. Purified GST or GST-pVHL protein was incubated with extracts of HEK293T cells. Red arrow indicates GST-pVHL. (D) In vitro cell-free GST pulldo