By screening a siRNA library targeting all human ubiquitin E3 ligases, we have identified RNF146 as a new positive regulator of Wnt signaling. RNF146 acts with tankyrase in the constitutive turnover of Axin proteins that maintains low levels of Axin and allows the Wnt receptor-coreceptor complex to initiate signaling by further repressing Axin's function in degrading β-catenin. Paradoxically, RNF146 also destabilizes through proteasomal degradation the tankyrase proteins, which are also positive regulators of Wnt signaling 
. Upon RNF146 RNAi, the stabilized tankyrase protein relocalizes to a centrosomal location, which might represent aggregated and inactive protein in aggresomes 
. Transgenic overexpression of tankyrase does not result in centrosomal puncta (
, and data not shown) however, interestingly, a tankyrase small-molecule inhibitor induced a similar stabilization and relocalization of tankyrase protein, indicating that PARsylation and ubiquitylation act together to maintain both proteasomal degradation and cytoplasmic localization of tankyrase (). Since RNF146 seems to be localized to centrosomes, as well as the cytoplasm, this may be a site of tankyrase ubiquitylation and degradation, as it can be for misfolded proteins 
Model of RNF146 activity that leads to degradation of tankyrase, Axin, and RNF146 proteins.
RNF146 protein is reciprocally destabilized by tankyrase activity, indicating that PARsylation and ubiquitylation target all three proteins in the RNF146-tankyrase-Axin complex to the proteasome for degradation. Since tankyrase inhibition prevents Axin ubiquitylation 
, PARsylation seems to be a signal for subsequent ubiquitylation, as has been proposed for proteasomal degradation of tankyrase substrate TRF1 
and tankyrase itself 
. Indeed, purified PAR stimulates RNF146 auto-ubiquitylation in vitro
During the preparation of this manuscript, Zhang et al
. reported an independent identification of RNF146 as a regulator or Wnt signaling 
. Consistent with our findings, Zhang et al
. propose that RNF146 ubiquitylates Axin that has been PARsylated by tankyrase to target Axin for proteasomal degradation. We significantly expand the mechanistic understanding of this branch of the pathway to include: (1) RNF146 regulation of tankyrase ubiquitylation, protein stability, and subcellular localization; (2) tankyrase regulation of RNF146 degradation; (3) polyubiquitin linkage types specified by RNF146 on tankyrase, Axin, and itself; (4) RNF146 regulation of autocrine Wnt signaling in teratocarcinoma cells, but not signaling activated by APC or β-catenin mutation in multiple colorectal cancer cell lines; and (5) specificity of RNF146 for ubiquitylating and regulating tankyrases and not the related PARP1 protein.
We show that RNF146 can operate as a single-subunit RING-type E3 catalyzing both ubiquitin-substrate ligation and ubiquitin chain elongation in combination with E1 and E2 enzymes in vitro, and that the RING domain is required for function in vivo. RNF146 directs the ubiquitylation of tankyrase, Axin, and itself, and the polyubiquitin chains specified are of multiple or mixed linkage types, containing at least K48 and K63 linkages. These different linkage types are consistent with RNF146 controlling both the proteasomal degradation and subcellular localization of tankyrase. We also identified the HECT-type ubiquitin E3 ligase HUWE1 (also known as ARF-BP1 or MULE) as a RNF146 interacting protein that depends on the RING domain of RNF146 for binding and seems to associate with polyubiquitylated RNF146. This suggests that HUWE1 may function as an E4 enzyme in ubiquitin chain elongation for RNF146 substrates. Interestingly, HUWE1 contains a putative ubiquitin-binding UBA domain and a WWE domain.
While RNF146 RNAi can inhibit autocrine Wnt signaling in teratocarcinoma cells and also stabilize tankyrase proteins in colorectal cancer cells with APC mutation-driven signaling, we find that RNF146 knockdown does not significantly affect Wnt signaling or Axin protein stabilization in the colorectal cell lines tested. In HCT-15 cells, tankyrase RNAi or small-molecule inhibitors increase Axin levels and partially block signaling, suggesting that there may be a functionally redundant ligase in some cell types. While there are other predicted E3s in the human genome with WWE domains, we find no clear RNF146 paralog. Testing combination RNAi of RNF146 and HUWE1 in colorectal cancer cells may be informative.
RNF146 can also bind PARP1 and PARP2, although apparently with lower affinity than for tankyrase association. PARP1 protein levels are not affected by RNF146 and, since PARP1 protein is localized to the nucleus 
whereas RNF146 protein is cytoplasmic, RNF146 may not mediate degradation of all PARP family members or of all PARsylated proteins. It will be important to further define the specificity of RNF146 in regulating the activity of different PARP enzymes, as well as to determine whether the other functions of tankyrases in different subcellular locations, such as telomere elongation, are regulated by RNF146.
The function of RNF146 in mouse development or human disease is currently unknown. The chromosomal region of RNF146 has been linked to breast cancer risk in a Jewish Ashkenazi population, although no mutations in the protein coding region have been identified 
. While RNF146 is expressed in all human tissues examined in one study, expression was upregulated in the brain of Alzheimer's disease patients 
. It will be interesting to specifically investigate a role for RNF146 in Wnt-dependent developmental and disease processes once mouse knockout strains have been generated.