To identify molecular mediators of Jade-1 activity, we screened a human kidney cDNA library with a yeast two-hybrid approach using a transcriptionally inactive truncation of Jade-1 lacking both PHDs (Jade-1 dd) as bait (Supplementary Information, Fig. S1a
). Nine strong interactors were found, including β-catenin, an oncoprotein and the key transcriptional co-activator of canonical Wnt signaling7
The Jade-1-β-catenin interaction was confirmed in mammalian cells by coimmunoprecipitation (). The localization and fate of β-catenin depend on Wnt status7
. Constitutively, in Wnt-off phase, β-catenin is phosphorylated by GSK-3β, binds to the destruction complex in the cytosol and gets degraded. In Wnt-on phase, GSK-3β is inhibited; β-catenin dissociates from the destruction complex and translocates to the nucleus. We therefore examined the binding of endogenous Jade-1 and β-catenin during the different states of Wnt signaling. Wnt signaling was activated using Wnt-3a ligand or lithium chloride (an inhibitor of GSK-3β that mimics Wnt activation) and inhibited using Wnt-3a plus DKK1, a competitive antagonist of Wnt-3a ( and Supplementary Information, Fig. S1b
). Endogenous Jade-1 co-immunoprecipitated with endogenous β-catenin and vice-versa (). However, the Jade-1-β-catenin interaction was increased in vehicle and Wnt-3a-DKK1 treated cells (Wnt-off phase) compared with Wnt-3a treated cells (Wnt-on phase). Co-localization and profile plots were performed to demonstrate the distribution and abundance of the proteins (). In Wnt-off phase (, Vehicle treated), β-catenin was predominantly in the cytosol and cell membrane. Jade-1 was in the cytosol and nucleus, exclusive of nucleoli3,8
. Co-localization of Jade-1 and β-catenin was found in the cytosol. Wnt-3a treatment resulted in nuclear translocation of β-catenin. However, Jade-1 and β-catenin exhibited different sub-compartmental localization in the nucleus (, Wnt-3a treated), resulting in reduction in co-localization. Thus, endogenous Jade-1 and endogenous β-catenin interact, and the interaction is greater in Wnt-off phase than in Wnt-on phase.
Figure 1 Jade-1 and β-catenin interact. (a) In vivo interaction of Jade-1 and β-catenin. Extracts (600 μg protein) from transiently transfected 293T cells were immunoprecipitated (IP) with 1 μg monoclonal Myc-tag or Flag-tag antibodies. (more ...)
Jade-1 specifically interacted with the N terminus of β-catenin (). Interestingly, Jade-1 showed reduced binding to a naturally occurring, cancer-causing, constitutively active (CA) S33A mutant of β-catenin lacking this GSK-3β phosphorylation site (). These findings were confirmed by immunofluorescence microscopy of cells expressing Flag-tagged Jade-1 and a Myc-tagged β-catenin series (Supplementary Information, Fig. S1c-e
). In Wnt-off phase, Jade-1 co-localized extensively with wild-type β-catenin and the C terminus truncation of β-catenin predominantly in cytosol, but not with β-catenin S33A or the N terminus truncation of β-catenin (Supplementary Information, Fig. S1d
). In contrast in Wnt-on phase, wild-type β-catenin localized to the nucleus, thereby reducing co-localization with Jade-1 (Supplementary information, Fig. S1d versus S1e
), consistent with the reduction in endogenous Jade-1-β-catenin binding with Wnt activation. These data indicate that β-catenin N-terminal serine residue 33, or its phosphorylation, is important for optimal binding to Jade-1. We examined the binding of purified recombinant GST-tagged Jade-1 and GST-tagged β-catenin in in vitro
GST pull-down assays. GST-tagged Jade-1 associated with GST-tagged β-catenin (Supplementary Information, Fig. S1f
). However, this binding was substantially increased after in vitro
phosphorylation of β-catenin by kinases CK1 and GSK-3β (Supplementary Information, Fig. S1f
). GST-tagged Jade-1 did not bind to the N terminus deletion of β-catenin. Overall, Jade-1 directly binds the N terminus of β-catenin, and the interaction is enhanced with β-catenin phosphorylation in Wnt-off phase.
In Wnt-off phase, β-catenin undergoes degradation, a process that depends on the β-catenin N terminus or ‘degron’7,9
. Jade-1 interacts with the N terminus of β-catenin and, in particular, residue S33, (Supplementary Information, Fig. S1a
). We therefore examined whether Jade-1 regulates β-catenin abundance. Jade-1 down-regulated wild-type β-catenin and a C terminus deletion (). However, Jade-1 had little effect on an N terminus deletion of β-catenin or β-catenin S33A, consistent with the binding pattern of Jade-1 with β-catenin ().
Figure 2 Jade-1 reduces β-catenin protein abundance. (a) Differential regulation of β-catenin constructs by Jade-1. Myc-tagged β-catenin and Flag-tagged Jade-1 (J1) constructs were transiently transfected into 293T cells. β-catenin (more ...)
Endogenous β-catenin exists in three distinct cellular pools7
. Endogenous β-catenin protein in the cytosolic and nuclear fractions was at least 2-fold higher in 3 different Jade-1
silenced cell lines than in empty vector lines ( and Supplementary Information, Fig. S2b-S2d
). The membrane pool of β-catenin was unchanged. Conversely, the amount of β-catenin in the cytosolic and nuclear fractions was substantially lower in Jade-1-expressing stable cell lines than in the empty vector cell lines (). We then examined the half-life of cytosolic β-catenin. A digitonin-extracted fraction10
was enriched for cytosol, as evidenced by the increase in cytosolic markers, but had no membrane contamination (Supplementary Information, Fig. S2e
). The half-life of the digitonin-extracted cytosolic β-catenin was increased from 10 mins to 90 mins in Jade-1
silenced 293 cell lines (). Thus, Jade-1
silencing substantially stabilized cytosolic β-catenin. β-catenin half-life was reduced in the Jade-1-expressing renal cancer cell lines (Supplementary Information, Fig. S2f
). Thus, Jade-1 regulates the stability of the Wnt-responsive pool of β-catenin.
β-catenin degradation depends on GSK-3β. In Wnt-off phase, β-catenin undergoes sequential phosphorylation at threonine 41 and serine 37 and 33 by GSK-3β. Preferential binding of Jade-1 to phospho-β-catenin and lack of binding to β-catenin S33A suggest a possible role for GSK-3β in Jade-1 regulation of β-catenin ( and Supplementary Information, Fig. S1f
). Moreover, full-length Jade-1 reduced total β-catenin due predominantly to reduction in phospho-β-catenin (Supplementary Information, Fig. S3a
). Thus, Jade-1 preferentially regulates phospho-β-catenin. This observation further suggests that GSK-3β may be particularly important for Jade-1 regulation of β-catenin. Indeed, Jade-1 regulation of β-catenin was mitigated by silencing or chemical inhibition of GSK-3β (). Similarly, the effect of Jade-1
silencing on β-catenin abundance was reduced in Wnt-on phase, when GSK-3β activity is inhibited ( and Supplementary Information, Fig. S3b
). Overall, these data indicate that Jade-1 requires intact GSK-3β kinase activity for full inhibition of β-catenin.
β-catenin undergoes proteasomal degradation11
. Proteasome inhibition with MG132 completely abrogated the effect of Jade-1 on β-catenin abundance (Supplementary Information, Fig. S3c
). Moreover, in the presence of Jade-1 and proteasomal inhibition, very high molecular weight species of β-catenin accumulated (Supplementary Information, Fig. S3d
), suggesting that Jade-1 may enhance ubiquitination and degradation of β-catenin.
Protein ubiquitination depends on substrate recognition by a highly selective E3 ubiquitin ligase. PHD proteins such as MEKK1 and MIR exhibit E3 ubiquitin ligase activity12,13
. Jade-1 has two PHDs3
that align well with the PHDs of MEKK1, MIR1, MIR2 and c-MIR (Supplementary Information, Fig. S4a
). Therefore, we reasoned that Jade-1, through its PHDs, might ubiquitinate β-catenin. Deletion of the PHDs reduced the effect of Jade-1 on β-catenin (Supplementary Information, Fig. S4b
). Next, endogenous β-catenin was immunoprecipitated, and its ubiquitination was examined in the presence of Myc-tagged ubiquitin and Jade-1, Jade-1 dd or the well-established β-catenin E3 ubiquitin ligase component β-TrCP6
(). Minimal β-catenin ubiquitination was observed with β-TrCP under these conditions, possibly due to lower expression of β-TrCP or the fact that the other components of the β-TrCP SCF complex were not co-expressed. Interestingly, robust β-catenin polyubiquitination was observed with full-length Jade-1, while deletion of the PHDs substantially reduced β-catenin ubiquitination (). β-catenin ubiquitination appeared as a smear, most prominently in the presence of full-length Jade-1. Since 293T cells are in Wnt-off status under basal conditions, these data indicate that Jade-1 promotes endogenous β-catenin ubiquitination in Wnt-off phase.
Figure 3 Jade-1 ubiquitinates β-catenin. (a) Deletion of the Jade-1 PHDs substantially reduces endogenous β-catenin ubiquitination. Extracts (600 μg protein) of 293T cells transfected and treated with MG132 (10 μM for 1 h) were (more ...) In vitro
ubiquitination of β-catenin was then examined. GST-purified β-catenin was incubated with HeLa cell cytosolic S100 fraction. β-catenin ubiquitination was observed in the presence of Jade-1 (). To address if Jade-1 directly ubiquitinates β-catenin and to map the E3 ubiquitin ligase domain within Jade-1, we reconstituted ubiquitination reactions with all purified components ( and Supplementary Fig. S4c
). β-catenin ubiquitination was observed with Jade-1 in a dose-dependent manner (Supplementary Information, Fig. S4c
, lanes 4-6). Deletion of the Jade-1 PHDs or the β-catenin N terminus abrogated β-catenin ubiquitination (, lanes 5 and 10; and Supplementary Information, Fig. S4c
, lanes 7 and 12). Thus, purified Jade-1 ubiquitinates non-phosphorylated GST-tagged β-catenin, and the Jade-1 PHDs are necessary for E3 ubiquitin ligase activity.
In order to examine the relationship between Jade-1- and β-TrCP-mediated β-catenin degradation, DN β-TrCP lacking the F box was used to antagonize both β-TrCP1 and β-TrCP214
. DN β-TrCP increased endogenous β-catenin expression by 2 fold (). Interestingly, Jade-1 could still down-regulate β-catenin in the presence of DN β-TrCP (), which suggests that Jade-1 regulates β-catenin independently of β-TrCP.
The biological significance of the Jade-1-β-catenin interaction was evaluated in TCF/β-catenin transcription assays. Full-length Jade-1, but not Jade-1 dd, inhibited a TOP-Flash promoter-reporter by 3.5 fold (). Jade-1 had no effect on β-catenin S33A transcriptional activity (). These observations are consistent with the effect of Jade-1 on protein levels of wild-type β-catenin and β-catenin S33A (). Intriguingly, significant inducible endogenous Wnt activity was observed in Jade-1
silenced cell lines (Supplementary information, Fig. S5a
), consistent with the stabilization of endogenous β-catenin in these lines.
Figure 4 Jade-1 inhibits canonical Wnt signaling. (a) Jade-1 suppresses TCF/β-catenin reporter activity. TCF-responsive promoter-reporter TOP-Flash and nonresponsive control reporter FOP-Flash were used. Activity of the Wnt signaling pathway was quantified (more ...)
Since Jade-1 destabilizes β-catenin and inhibits β-catenin-mediated transactivation, we reasoned that Jade-1 might inhibit the canonical Wnt pathway in vivo
. We used a functional assay for canonical Wnt signaling, formation of an ectopic axis in Xenopus laevis
embryos by ventral injection of Xwnt-8
. Full-length Jade-1
, but not Jade-1 dd
, significantly inhibited both Xwnt-8
- and β-catenin
-induced ectopic axis formation in developing Xenopus laevis
embryos ( and Supplementary Information, Fig. S5b
). These data suggest that Jade-1 has the capacity to suppress Wnt activity during Wnt-on phase in vivo
. This is plausible in view of the appreciable interaction of Jade-1 and β-catenin in Wnt-on phase ( and Supplementary Information Fig. S1b
) and Jade-1 binding and ubiquitination of non-phosphorylated β-catenin (Supplementary Information, Fig. S1f
and ). The specific Wnt target Axin216
was also increased in Jade-1
silenced 293 cell lines (). Other Wnt targets, such as cyclin D1 and c-Myc, were increased in Jade-1
silenced cell lines (Supplementary Information, Fig. S5c
). Full-length Jade-1, but not Jade-1 dd, reduced protein levels of c-Myc (Supplementary Information, Fig. S5d
). Thus, Jade-1 is an inhibitor of canonical Wnt signaling.
Jade-1 protein is stabilized by wild-type pVHL, but not by mutated pVHL associated with renal cancer3,4
. We hypothesized that pVHL might regulate β-catenin through Jade-1. First, we compared β-catenin abundance in VHL
-deficient and VHL
-intact renal cancer cell lines (). Interestingly, endogenous cytosolic and nuclear pools of β-catenin were several-fold lower in VHL
-intact cell lines compared to VHL
-deficient cell lines (). This result can be explained on the basis of our previous observations that Jade-1 levels are significantly lower in VHL
-deficient cell lines5
. Next, pVHL reintroduction in 786-O cells increased endogenous Jade-1 4,5
. and substantially reduced endogenous β-catenin levels ( and Supplementary Information, Fig. S6a
). Knock-down of VHL
with siRNA oligonucleotides resulted in down-regulation of Jade-1 and accumulation of β-catenin ( and Supplementary Information, Fig. S6b
Figure 5 pVHL regulates endogenous β-catenin through Jade-1. (a) β-catenin protein levels depend on VHL status in renal cancer cell lines. Cytosolic and nuclear fractions of renal-cell carcinoma cell lines were probed for endogenous β-catenin (more ...)
To specifically examine whether Jade-1 mediates pVHL down-regulation of β-catenin, we compared β-catenin regulation by wild-type pVHL and by truncated forms of pVHL that do not stabilize Jade-14
(). pVHL del96-122 and naturally occurring, cancer-causing truncations like pVHL 1-143 and pVHL 1-175, which have little or no effect on Jade-1 stability4
, had minimal effect on β-catenin (). We also knocked down Jade-1
in the presence of pVHL. Endogenous Jade-1 levels were increased by pVHL, an effect blocked by Jade-1
knock-down (Supplementary Information, Fig. S6c
). Importantly, down-regulation of β-catenin by pVHL was substantially mitigated by Jade-1
knock-down (Supplementary Information, Fig. S6c
). Furthermore, pVHL wild-type, but not pVHL del96-122, suppressed β-catenin transcriptional activity (Supplementary Information, Fig. S6d
). Specific Wnt targets such as LEF118
and cyclin D1 were reduced in pVHL-expressing renal cancer cell lines (). Wild-type pVHL, but not pVHL del96-122, also reduced Axin2 in 293T cells (Supplementary Information, Fig. S6e
). Overall, these data indicate that pVHL inhibits β-catenin and canonical Wnt signaling and that Jade-1 is a critical mediator of these effects.
To determine if pVHL is able to reduce endogenous Wnt signaling in vivo
, and to examine the difference in the suppression of Wnt activity by wild-type pVHL and pVHL del96-122, we injected wild-type VHL
or VHL del96-122
mRNA into the dorsal blastomeres of Xenopus laevis
embryos, in which canonical Wnt signaling is necessary for dorsal development19
. Inhibition of dorsal axis formation in Xenopus laevis
embryos is evidenced by reduction of dorsoanterior structures (small eyes, microcephaly, or anencephaly) measured with a dorsoanterior index (DAI) scale19
. Wild-type pVHL suppressed dorsal axis formation significantly more than pVHL del96-122 at the same level of protein abundance (Supplementary Information, Figs. S6f and S6g
). As expected, inhibition of endogenous axis by wild-type pVHL was associated with greater suppression of β-catenin/Tcf target genes Xsiamois
than by pVHL del96-122 (). Thus, pVHL inhibits canonical Wnt signaling in vivo
, and pVHL del96-122, which binds and regulates Jade-1 only minimally4,5
, had little effect on dorsal axis formation and Wnt target genes in Xenopus laevis
embryos. These data further support the role of Jade-1 as a critical mediator of pVHL regulation of β-catenin.
In this study, we have demonstrated that Jade-1 is a single-subunit E3 ubiquitin ligase for β-catenin and that the Jade-1 PHDs are essential for E3 ubiquitin ligase function. Moreover, Jade-1 is a critical mediator of pVHL inhibition of β-catenin and canonical Wnt signaling. Jade-1 is anti-proliferative and pro-apoptotic, while β-catenin is pro-proliferative, anti-apoptotic and oncogenic7
. Therefore, the tumor suppressor activity of Jade-1 may be due in part to inhibition of β-catenin.
Several ubiquitin ligases for β-catenin have now been identified9,20,21
, but only β-TrCP and Jade-1 show Wnt responsiveness, suggesting they are both important in the physiology and pathophysiology of canonical Wnt signaling. Endogenous β-TrCP resides in the cytosol and is capable of binding and ubiquitinating phosphorylated β-catenin6,9,22,23
. In contrast, endogenous Jade-1 resides in the cytosol but is found primarily in the nucleus3,8
and is capable of binding and ubiquitinating both phosphorylated and non-phosphorylated β-catenin. Thus, Jade-1 and β-TrCP have only partially overlapping subcellular locations and have differing specificities for the forms of β-catenin. These differences may explain why Jade-1
silencing cannot be completely compensated for by β-TrCP ( and Supplementary Information, Figs. S2b-d
). Moreover, Jade-1 seems to act more distally in the canonical Wnt cascade, affecting β-catenin in the nucleus. This may make Jade-1 responsible for fine control of β-catenin levels.
Jade-1 functions as a single-subunit E3 ubiquitin ligase for β-catenin, whereas β-TrCP requires formation of a multi-subunit protein complex. The PHD functions as an adaptor-type E3 ubiquitin ligase directly transferring the ubiquitin moiety to the substrate13
. Control of Jade-1 may therefore be a simpler and perhaps more efficient way of regulating β-catenin abundance. It is likely that the cell exploits the distinct functional and contextual differences between β-TrCP and Jade-1 to ensure effective regulation of Wnt signaling.
β-catenin is emerging as a key molecule in the pathogenesis of renal cancer and renal cystic disease. For example, increased β-catenin activity in renal epithelium in mice results in robust renal cyst and tumor formation24-26
. In renal cancer, methylation of the APC
gene promoter is common27
. pVHL was recently shown to inhibit HGF-mediated tyrosine phosphorylation of β-catenin28
. These observations further strengthen the role of Wnt signaling in renal-cell carcinoma. Jade-1 and pVHL may also participate in other forms of cystic kidney disease, in which evidence of dysregulated Wnt signaling is mounting26,29
Our data suggest that Jade-1 inhibition of Wnt signaling represents a new tumor suppressor axis for pVHL. Furthermore, these findings directly link the kidney-specific pVHL tumor suppressor pathway and the Wnt signaling cascade, a more general tumorigenesis pathway. Jade-1 and beta-catenin may therefore represent therapeutic targets in renal-cell carcinoma.