Regulation of Cullin-RING E3 ligase dynamics by Inositol hexakisphosphate

Perhaps in a decade or so, researchers may look back on current times and view it as an inflection point for the development of novel therapeutic modalities for the treatment of various human diseases. One particularly innovative approach involves targeting disease-modifying proteins for degradation by the proteasome, the large multisubunit protease responsible for the lion's share of regulated protein degradation in human cells (1). The technique uses the action of ubiquitin (UB) ligases, the enzymes that append polyubiquitin chains onto proteins that signal for degradation, and unleashes their activities on disease-causing proteins. Inducing proximity between the ubiquitin ligase and the target is accomplished by small molecule effectors that bind in shallow grooves on both the enzyme and the protein target, acting as “glue” and driving substantial affinities between two proteins that normally do not interact in cells (2, 3). While this may sound more like a futuristic treatment than reality, there are several Food and Drug Administration-approved therapies utilizing induced proximity to target proteins to the proteasome, and even more are in clinical trials (4). While one concern had been that molecular glues capable of promoting protein–protein interactions are quite rare, the study in PNAS by Lin et al. (5) is likely to change opinions on the matter.

Before explaining the results from Lin et al. (5), a bit of background is necessary. Most of the current therapeutic molecular glues that target disease-causing proteins for degradation utilize the largest family of ubiquitin ligases in humans, the Cullin-RING (Really Interesting New Gene) ligases (CRLs) (6, 7). Some 200 of these enzymes can be found in our cells and account for at least 20% of all proteasome-dependent protein degradation. This grand number of enzymes associates with seven Cullin scaffold proteins, which bind to RING proteins (RBX1/2) that recruit ubiquitin-carrying enzymes. Cullins also associate with numerous substrate receptors (SRs) responsible for selecting the proteins targeted for ubiquitylation.

Several landmark studies have revealed an elegant mechanism for control of CRL activity in cells (8–12). At the center of this regulation is a ubiquitin-like protein called NEDD8. CRLs covalently bound to NEDD8 are active, whereas the removal of NEDD8 results in the rapid dissociation of the SR and places the CRL in a sleep-like state awaiting reactivation. Recent work has shed light on the critical importance of this mechanism for controlling CRL activity: the concentration of SRs greatly exceeds the levels of Cullin-RING subunits in the cell. As a consequence, the only way for the thousands of CRL-dependent protein substrates to gain access to active enzyme is for the SRs to exist in rapid exchange with each other on the Cullins (11). Notably, SR exchange can only occur when CRLs are inactive, placing an extreme burden on a multisubunit hydrolase, called COP9 signalosome (CSN), that is tasked with the removal of NEDD8 (called deneddylation) from CRLs. Indeed, CSN is of such tremendous importance that knockout mice exhibit embryonic lethality (5).

While CSN has been intensively scrutinized for nearly two decades and much prior literature hinted that inositol polyphosphates might be involved in regulating CRLs, Lin et al. (5) report in PNAS the key observation that one of the CSN subunits is tightly bound to inositol 6-phophate (IP6) (13–16). It had been observed that CSN copurifies with IPTK1, the enzyme responsible for production of IP4 and IP5 (15). Subsequently, IP5K and IP6K1, the kinases responsible for the production of IP6 and IP7, were discovered to interact with CRLs and/or CSN as well (14). Despite these and many other clues, the structural and biochemical mechanisms by which phosphoinositols affect CSN–CRL complexes remained elusive until Lin et al. (5) identified CSN2 as an IP6 binding partner.

Using isothermal titration calorimetry to ascertain the affinities between IP6 and CSN and/or CRL, the authors demonstrate that IP6 binds CSN2 with high affinity. There was little to no detectable binding of IP6 to CUL4A-RBX1. Yet astonishingly, the addition of CUL4A-RBX1 to CSN2 in the presence of IP6 resulted in an order of magnitude increase in the affinity of IP6 for the protein complex as compared with CSN2 binding alone, strongly suggesting that IP6 might act as molecular glue between CSN and CRL4A. Consistent with this notion, IP6 stimulated CSN in vitro deneddylation activity in a dose-dependent manner.

The molecular details of the high-affinity interaction between IP6 and CSN2 were revealed by solving the cocrystal structure of IP6 bound to CSN2. A positively charged surface patch formed by CSN2 helices α1, α3, and α5 forms the binding site for IP6 with several conserved lysine residues coalescing to interact with the phosphate groups from the 2′, 3′, 4′, and 6′ positions of IP6. Mutagenesis of these conserved CSN2 lysine residues and assessing their affinities for CRL4A in pull-down assays validated their role in IP6 recruitment. Importantly, these mutants were demonstrated to be specifically defective in IP6-enhanced CRL4A binding without affecting basal CSN–CRL4A interactions.

How then does IP6 glue together CSN and CRL4A? To address this question, the authors revisited a previous cryoelectron microscopy map of the CSN–CRL4A complex that revealed a patch of density not assignable to CSN or CUL4A, which was also consistent with the location of IP6 observed in the IP6–CSN2 structure (17) (Fig. 1). This helped to identify two conserved lysine residues from the RING domain protein RBX1 that is located opposite of CSN2 and poised to interact with IP6. Indeed, mutation of these RBX1 residues reduced CSN–CRL interactions in cells and specifically abolished IP6 enhanced binding in vitro.

Fig. 1.

Cartoon schematic for the role of IP6 in regulating CRL dynamics. Following dissociation of polyubiquitinated substrate, IP6 exerts its molecular glue properties to enhance CSN–CRL interactions to promote removal of NEDD8 and protect bound SRs from spurious polyubiquitination and degradation.

With the structural role of IP6 binding to the CSN–CRL complex firmly established, the authors next set out to uncover the functional importance of the interaction. They hypothesized that IP6 is a critical cofactor for efficiently terminating CRL4A activity through displacement of the ubiquitin-carrying enzyme CDC34. This displacement may protect CRL4A and its associated SRs from autoubiquitylation, a process where ubiquitins are transferred to the various CRL subunits when SRs are bound to CRLs but not protein substrates, leading to their misguided degradation. Consistent with this idea, competition binding assays demonstrated the ability of IP6 to significantly bias CRL4A binding for CSN2 over CDC34. Furthermore, halting production of IP6 by knockdown of IP5K promoted instability to the CUL4A SR DDB2 (DNA damage-binding protein 2). Similar effects were observed on expression of an RBX1 mutant incapable of IP6-mediated binding. In light of recent reports demonstrating that the ubiquitin-carrying enzyme UBE2G1, but not CDC34, is required for ubiquitination of CRL4A–cereblon-based neosubstrates, it is unclear what the precise role of CDC34 is in ubiquitinating CRL4A-based substrates or adaptors (18, 19). Nevertheless, the enhanced affinity of CSN for CRL4A, mediated through IP6, would be expected to further preclude RBX1’s RING domain from recruiting ubiquitinating enzymes capable of imparting instability to bound SRs (Fig. 1).

The findings by Lin et al. (5) provide significant structural and biochemical insights into IP6-dependent modulation of CRL dynamics, but it still remains unclear whether the role of IP6 is regulated in some manner. IP6 levels in cells have been estimated to be at least ∼20 μM to perhaps even an order of magnitude higher (20). While the pool of bioavailable material may be lower due to binding by other proteins or due to flux in IP6 levels from the action of both kinases and/or phosphatases, CSN2's binding affinity for IP6 (∼0.3 μM) suggests total saturation of CSN2 unless truly drastic changes in IP6 levels occur. Thus, if changes in cellular IP6 levels are not regulating CSN activity, it is tempting to speculate that a single molecular glue used to enhance CSN–CRL interactions may reflect a simple solution for achieving a binding threshold for CSN to the hundreds of distinct CRL–SR assemblies. For instance, there is considerable amino acid sequence variation among the seven human Cullin subunits, and it can be imagined that happening on multiple, equivalent CSN–Cullin interfaces of sufficient binding energy may have been too onerous during evolution. Instead, using a molecular glue bridging CSN's primary Cullin binding subunit and a highly conserved basic region on the CRL partner RBX1 or RBX2 is both elegant and simple.

Yet, even more perplexing is why the kinases that produce IP6 and IP7, with IP7 having approximately threefold less affinity to CSN2 than IP6, associate with both CSN and CRL components (13–15). These observations may indicate that a complex sequence of events may regulate IP6 and IP7 production on CSN–CRL subunits, thereby directly regulating CSN binding and dissociation. Future work will be needed to clarify why these complexes form and how they regulate both CSN as well as CRL activities.

Another important aspect to consider in light of the discovery of Lin et al. (5) involves the development of drugs that hijack E3 ligases and promote targeted protein degradation. These drugs, such as the bivalent degraders termed proteolysis-assisted chimeras (PROTACs) or molecular glues like palmolidimide that induce proximity between a CRL–SR and the disease-modifying protein, have garnered tremendous interest in recent years. Given the newly defined role of IP6 in regulating CRL dynamics, inhibitors that target IP6 production might also lead to deactivation of CSN and the supercharging of targeted proteins degraders like PROTACs.

IP6 itself has been shown to possess anticancer activity (21). Several studies have demonstrated that IP6 treatment induces G1 arrest and increases the levels of the cyclin-dependent kinase inhibitors p21 and p27 in cancer cells. The anticancer properties of IP6 can be reversed through simultaneous knockdown of p21 and p27, suggesting that they are critical targets for IP6 efficacy (22). Since both p21 and p27 are well-defined substrates for CRL1- and CRL4-based ubiquitin ligases, it is tempting to speculate that at least some degree of IP6's anticancer activity might come from increased down-regulation of CRL function, thus stabilizing p21 and p27 to promote G1 arrest.

In summary, molecular glues are beginning to show up in greater and greater numbers of important biological roles in the cell. As is so often the case, taking lessons from nature is leading to new therapeutic approaches for the treatment of human disease.