Structural Studies of RNF146-Mediated PARylation-Dependent Ubiquitylation

Abstract

The ubiquitylation of proteins is involved in nearly every cellular process. Hence, precise control of the ligases that facilitate the attachment ubiquitin (Ub) to target substrates is of the utmost importance to eukaryotic biology. My thesis work on the RING E3 ubiquitin ligase RNF146 has defined a new mechanism by which a ligase can be allosterically activated. I have provided the first example of a RING domain (E3 domain) that has a conformational change, with my discovery that RNF146 is conformationally “switched on” by binding the post translational modification poly(ADP-ribose) (PAR). My work on RNF146 (and other proteins) builds on a growing theme of RING E3 ligase regulation, upending the notion that all single subunit RING E3 domains are constitutively able to bind and activate an E2~Ub conjugate to transfer ubiquitin to substrates. Protein poly(ADP-ribosyl)ation (PARylation) is involved in many cellular processes including DNA repair, cell division, and cell death. It was recently shown that PARylation catalyzed by the PARPs tankyrase-1 and -2 can mark proteins for ubiquitylation and subsequent degradation via the proteasome. The E3 ubiquitin ligase RNF146 (a.k.a. Iduna) is the only E3 identified thus far that participates in PARylation-dependent ubiquitylation (PARdU) with tankyrases (TNKSs). Through PARdU, RNF146 controls the levels of important regulatory proteins including Axin and 3BP2. We provide a structural basis for the role of RNF146 in PARdU and how RNF146 achieves substrate specificity. First, we show that the smallest internal poly(ADP-ribose) (PAR) structural unit, iso-ADPr, binds between the WWE and RING domains of RNF146 acting as an allosteric signal to switch the RING domain from an auto-inhibited conformation to an active one. In the absence of PAR/iso-ADPr, RNF146 only weakly binds an E2 and cannot activate it for ubiquitin transfer. In this unliganded state, a loop of the RING domain blocks the E2-E3 interaction. PAR binding causes the incorporation of this loop into the main helix of the RING domain, generating a productive RING E3 ligase. Second, we demonstrate that RNF146 forms a complex with tankyrases via its disordered c-terminus. This complex appears to be mediated by at least four elongated tankyrase-binding motifs (TBMs). We have solved a structure of one of these motifs in complex with a fragment of tankyrase-1 confirming that these non-canonical TBMs are bona fide binding motifs. Mutants that disrupt either the allosteric activation of the RING domain or the RNF146-TNKS interaction inhibit Axin turnover in vivo. Third, we show through structural modeling that tankyrases oligomerize through their sterile alpha motif domains in a head-to-tail manner. We can disrupt this interaction with site-directed mutagenesis and oligomerization-deficient mutants cannot promote tankyrase- mediated Wnt signaling enhancement. Hence, we show that RNF146 represents a new class of E3 ligases in which binding of ligands causes conformational changes in the RING domain, that PARdU substrate specificity is likely facilitated by the TNKS-substrate interaction, and that tankyrases oligomerize via a head-to-tail SAM-SAM interaction which likely affects the tankyrase-substrate interaction and the role tankyrases play in Wnt signaling.