The RING protein RBX-1 is implicated in both NEDDylation and ubiquitylation reactions. In this issue, new structural analysis reveals how conformational flexibility of the RBX-1 linker allows for a marked reorientation of the CUL1–RBX1 complex to facilitate transfer of NEDD8 or ubiquitin by closing the gap between the E2 catalytic site and the substrate.
Ubiquitin signaling regulates various cellular processes by controlling the degradation, activity and subcellular localization of crucial regulatory proteins1. The ubiquitin cascade has been understood for some time, and ligation of ubiquitin or ubiquitin-like (UBL) molecules to substrates is mediated by the sequential action of three enzymes, each with E1 (activating), E2 (conjugating) and E3 (ligase) activities2. The E1 activating enzyme creates a thioester bond between its conserved cysteine residue and the C-terminal carboxyl group of monomeric ubiquitin. The activated ubiquitin is then passed by a transthiolation reaction to a reactive cysteine residue of a specific E2 conjugating enzyme, which is transferred through an E3 ligase enzyme to either the N-terminal methionine or internal lysine residues of targeted proteins. In these kinds of cascade reactions, ubiquitin E3 ligases catalyze the final step of the process and govern the specificity of the modification reaction. There are two major classes of E3s that differ in the mechanism by which they mediate the transfer of ubiquitin to substrates. HECT-domain E3s form a thioester intermediate with ubiquitin before its transfer to the substrate, whereas RING ligases facilitate direct transfer of ubiquitin from E2-bound ubiquitin to the substrate. However, despite intense efforts to understand the catalytic mechanism of RING–E3 ligases through structural and biochemical studies, it remains unknown. On page 947 of this issue, Calabrese et al. present evidence that rotation of the RING domain may act as a general mechanism for ubiquitin and UBL transfer for the largest and most intensively studied subclass of RING E3 ligases, the cullin–RING ligases (CRLs)3.
CRLs are modular multisubunit complexes composed of the scaffold cullin protein (CUL1, CUL2, CUL3, CUL4A, CUL4B or CUL5), the E3 RING protein (RBX1 or RBX2), an optional linker protein and a substrate adaptor4. In the SCF complex, the prototype for CRLs, CUL1 and RBX1 form the catalytic core complex, which recruits the E2 enzyme, whereas the ubiquitylation substrate binds the variable F-box protein that is, in turn, linked to CUL1 through the SKP1 adaptor protein. It is well known that the E3 ligase activity of the CRLs is enhanced by post-translational modification of cullin proteins by the ubiquitin-like protein NEDD8 (refs. 5,6). Like ubiquitin, NEDD8 is attached to its substrates by an isopeptide linkage between its C-terminal glycine to a lysine of the target protein. Interestingly, RING protein RBX1 is implicated in both NEDDylation and ubiquitylation reactions. For NEDDylation, it catalyzes ligation of the NEDD8 to CUL1 by binding to UBC12, the NEDD8 E2 enzyme7. The NEDD8-modified CUL1 then cooperates with RBX1, which has now bound an ubiquitin-charged E2 such as UBCH5 or CDC34, to mediate substrate ubiquitylation. Structural studies have revealed the overall extended structure of the CRL complex, in which elongated cullin proteins provide a platform to interact with the adaptor-substrate through the N-terminal helical region and to RBX1 through the C-terminal domain8,9,10,11. However, none of these structures explain the transfer of NEDD8 or ubiquitin to their respective substrates. On the contrary, they show that positioning of the ubiquitin-bound E2 or NEDD8-charged UBC12 creates a gap of ∼50 Å or ∼30 Å between the active site cysteine and either the CRL-bound substrate or the NEDD8 acceptor lysine, respectively (Fig. 1a). Furthermore, the distance between the catalytic site of the E2 enzyme and the substrate lysine is expected to vary during polyubiquitin chain elongation. How these gaps are closed and how the ligase complexes adopt conformations required for catalysis has remained unknown.
Previously, Duda et al. revealed that NEDDylation of cullins induces conformational change in the CRL structure, by transforming the CUL1–RBX1 complex from a 'closed' to an 'open' conformation11. In the open conformation, the RBX1 linker is substantially extended away from the α/β domains of cullin, and the RING is freed from interaction with cullin's winged-helix B (WHB) region. This rearrangement of the CRLs following ligation of NEDD8 closes the gap between the E2 and ubiquitylation substrate11, and hindering the orientational flexibility of RING maintains the closed conformation, resulting in the inaccessibility of the cullin NEDDylation site and inhibition of the E3 ligase activity9,11. To understand the effect of this structural malleability on CRL NEDDylation, Schulman and co-workers undertook a new structural characterization of the CUL1CTD–RBX1 complex, providing the first demonstration of how rotation of the RBX1 RING places the CUL1 acceptor lysine residue in close proximity to the active site cysteine of the NEDD8 E2 UBC12. Through elegant, structurally based disulfide-trapping experiments, the authors confirmed that CUL1 cannot be NEDDylated when the complex is fixed in positions corresponding to previously reported CUL1–RBX1 structures. In fact, constraining CUL1's C-terminal region and RBX1's RING domain impaired NEDD8 transfer to substrate CUL1, suggesting the existence of conformations different from what had been observed so far. Previously, this issue was partially resolved by SAXS data indicating conformational flexibility within the cullin–RING complex11, but those data did not provide conclusive evidence for this mechanism.
The new crystal structure reveals a relative repositioning of the subunits largely due to rotation of the RBX1 RING domain, placing it in proximity of the site of NEDDylation. The authors suggest that this represents the active form of RBX1 and that other RING E3s may work in a similar manner. A linker between the N-terminal CUL1 binding site and the C-terminal RING domain is able to adopt multiple conformations, thereby accounting for the structural flexibility of the SCF complex11,12. In a similar manner, conformational changes in the new CUL1–RBX1 structure determined by Calabrese et al. are mainly mediated by the large extension of the RBX1 linker and rotation about the hinge centered at residues Val38 and Val39, located in the linker region3. Notably, deletion of even a single residue of this linker abrogates CUL1 NEDDylation by preventing linker extension and subsequent formation of the CUL1–RBX1 open conformation, which is regarded as the active form of the ligase complex10. Modeling the CUL1–RBX1–UBC12 complex by docking other E2–RING structures onto the new CUL1–RBX1 complex clearly shows that RBX1-associated UBC12 approaches CUL1, and that the catalytic Cys111 of UBC12 becomes juxtaposed to CUL1's NEDD8 acceptor Lys720 (ref. 3, Fig. 1b).
Conformational flexibility in the SCF complex resulting in relative subunit rearrangement is thus crucial to regulation of the RING E3 ligase activity. Inactive forms of the SCF complex are defined by closed conformations of CUL1–RBX1 in which RBX1's C-terminal domain makes extensive interactions with the WHB region of the C-terminal domain of CUL1 (CUL1-CTD)7,8. In the inactive form, the large gap prevents NEDDylation of the CUL1 acceptor lysine. However, in the active form of the complex, because of the relative repositioning of CUL1 and the RING domain of RBX1, Cys111 of the associated UBC12 moves as close as ∼3 Å to the CUL1 acceptor lysine, thereby allowing transfer of NEDD8 to the substrate3. Further reorientation of the cullin and RBX1 RING domain is essential for activation of the RING E3 ligase complex to bridge the ∼50 Å gap between the ubiquitylation substrate and the ubiquitin-bound E2 enzyme6,11. Both NEDDylation and ubiquitylation are impaired by restricting the conformational flexibility of the SCF complex. This explains the inhibitory mechanism of cullin-associated and NEDDylation-dissociated 1 (CAND1)9,11, which locks CUL1–RBX1 in a closed conformation. Thus, despite the considerable diversity of the CRLs, they are regulated by similar mechanisms13, and it seems likely that conformational variability results in similar actions in many other RING E3 ligases to mediate ubiquitin and UBL ligations to the substrates.
Given the inherent flexibility of RBX1, certain factors are needed to restrict cullin–RBX1 conformations to those favoring either the NEDDylation or ubiquitylation reactions. So far, Dcn1 and NEDD8 can be regarded as such factors, as they help confine SCF's structural orientations. Although RBX1 is critical for efficient NEDDylation14, a more recent study indicates that Hrt1 (RBX1 in mammals) and Dcn1 work synergistically as a dual NEDD8 E3 ligase15. It is suggested that Dcn1 places the UBC12-associated Hrt1 adjacent to the cullin NEDD8 acceptor lysine, thereby enhancing NEDDylation16. Therefore, this dual E3 mechanism might function to limit orientations that would allow UBC12's catalytic site to encounter the cullin's acceptor lysine15. On the other hand, ligation of NEDD8 to cullin might also assist SCF to adopt conformations that allow recruitment of distinct E2 conjugation enzymes to a particular substrate. It is likely that once NEDDylation occurs, the RING domain is unable to reach this conformation, so the Cdc34-bound complex uniquely selects conformations that favor substrate ubiquitination rather than cullin NEDDylation.
Although it is well established that spatiotemporal organization of the E2 and E3 enzymes is implicated in the control of specific substrate modifications17, how the dynamic changes of E2 and E3 are regulated in vivo remains unclear. The study by Calabrese et al.3 reveals the critical role of conformational flexibility in the RBX1 RING ligase for controlling NEDDylation of cullins as a signal to switch on cullin–RING ubiquitin ligase activity. Clearly, there are more challenges ahead for those who seek a better understanding of the regulatory mechanisms within the CRL-assembled complexes, and future studies will certainly soon reveal new surprises.