Interplay between ubiquitylation and SUMOylation: Empowered by phase separation

Abstract

Ubiquitin and the ∼20 human ubiquitin-like proteins regulate numerous aspects of cell biology via interlinked mechanisms that have not been fully elucidated. Sha et al. now explore the interplay between ubiquitylation and SUMOylation, finding that inhibition of ubiquitylation enhances SUMOylation of hundreds of newly synthesized proteins and that the resultant pools are stored in phase-separated protein condensates called PML nuclear bodies. These unexpected outcomes identify a new role for SUMOylation and raise new questions about cell behavior under normal and stress conditions.

Introduction

Ubiquitin is a small protein, with its molecular weight only at ∼8.6 kDa, but has a mighty effect on the cell: Virtually all life activities are regulated by ubiquitylation in one way or another (1, 2). It exerts its functions through conjugation to protein substrates via an E1-E2-E3 enzymatic cascade called ubiquitylation. Most ubiquitylated proteins are directed to the 26S proteasome to be degraded (1, 2), but ubiquitylation can also initiate sorting to other destinations, alter protein activity, or cause other outcomes. The human genome also contains about 20 ubiquitin-like (Ubl)² proteins (3). Like ubiquitin, these proteins can be conjugated to protein substrates via their own E1-E2-E3 enzymatic cascades, and, like ubiquitin, these modifications regulate a variety of biological processes. Because Ubls use the same lysine residues on protein substrates to mediate attachment, Ubl modifications can compete for the same sites (4–6), with the different modifications differentially altering the functions of the modified proteins. For instance, during NF-κB activation, IκBα is ubiquitylated and degraded to release its inhibition of NF-κB (5). Conversely, SUMOylation at the same site blocks ubiquitylation and turnover of IκBα (6), therefore inhibiting NF-κB activation. Ubls have also been shown to promote the activation of ubiquitin ligases. In the case of Cullin-RING ubiquitin ligases (CRLs), conjugation of Nedd8 to the Cullin scaffold proteins is required for their maximum activity (7).

Whereas these overlapping sites and functions undoubtedly add to the rich regulatory potential of the cell, dissecting this cross-talk to understand the role of each Ubl has been difficult. One Ubl of particular interest is SUMO, which modifies more than 1000 target proteins. SUMO's enzymatic cascade includes the SUMO-activating enzymes SAE1 and SAE2, the conjugating enzyme Ubc9, and E3 ligases such as PML, leading to modification by any of the three SUMO proteins, SUMO1–3. SUMOylation initiates binding to proteins containing SUMO-interaction motifs, often in multivalent arrays, and the resultant extended complexes form phase-separated protein condensates, such as the PML nuclear body. SUMOylation and the nuclear bodies increase upon oxidative stress and proteasome inhibition as well as other stimuli, but the role of these condensates has not been clear.

A new outstanding story by Sha et al. (8) provides new evidence to demonstrate a connection between SUMOylation and proteolysis by directly examining the interplay between SUMOylation and ubiquitylation. The authors first tested the effect of TAK243, a specific pharmacological inhibitor of the ubiquitin-activating enzyme Uba1, on protein SUMOylation in several cell lines (8). They found that ubiquitylation inhibition by TAK243 enhanced SUMOylation. Conversely, SUMOylation inhibition did not lead to increase of ubiquitylated proteins. Moreover, modifications by other Ubls, such as ISG15, Nedd8, or Ufm1, were not changed by TAK243, indicating some specificity. Interestingly, SUMOylated proteins start to accumulate 3 h after TAK243 treatment. Cycloheximide, a ribosome inhibitor, almost completely blocked the accumulation of SUMOylated proteins, confirming that they are newly synthesized proteins. Unexpectedly, cycloheximide treatment had a larger impact on accumulation of proteins conjugated to SUMO2/3 than on those conjugated to SUMO1, indicating a differential SUMO-specific outcome upon ubiquitylation inhibition. Overall, these observations echo the earlier discovery that SUMOylated proteins build up in cells treated with proteasome inhibitors and such an accumulation of SUMOylated proteins can be blocked by inhibitors of protein synthesis as well (9). However, this is very different from what has been observed with heat shock–induced accumulation of SUMOylated proteins, where the modified sequences are not newly synthesized proteins (8). In addition, Sha et al. confirmed the roles of SAE and Ubc9 in amassing SUMOylated proteins using a SUMO E1 inhibitor and small interference RNA oligomers targeting Ubc9. Using an MS approach, the authors identified that SUMOylation of 1058 proteins in HeLa and 736 in U2OS cells was enhanced upon TAK243 treatment, with 543 proteins increased in both cell lines. The increase of SUMOylation in over 80% of proteins could be blocked by cycloheximide, again confirming that the majority of the proteins with increased SUMOylation are newly synthesized. The buildup of SUMOylated proteins could simply be due to a defect in ubiquitylation and turnover. Very surprisingly, however, gene ontology analysis suggested that there is some specificity to the modified proteins: SUMOylated proteins with the greatest increase are mainly involved in transcription, the DNA damage response, and DNA replication as well as cellular stress responses, including ER stress. In fact, all three major transcription factors (ATF6, ATF4, and XBP1) related to ER stress were at the top of the list. STRING analysis further indicated that SUMOylation of proteins related to DNA repair and the G₂ checkpoint pathway, DNA replication, and translational control was increased, whereas SUMOylation of mRNA splicing factors was decreased. Together, these data suggest that inhibition of ubiquitylation causes certain stress responses, but further work is needed to understand why these particular signaling pathways were affected the most.

Finally, the authors observed a most interesting result: These newly SUMOylated proteins are localized in PML nuclear bodies in a PML-dependent manner. PML nuclear bodies are believed to antagonize certain hazardous conditions (8), and, as Sha et al. note, PML nuclear bodies are important niches for protein quality control in the nucleus. Perhaps, under certain stress conditions, PML nuclear bodies are temporarily employed to store potentially harmful nuclear proteins such as SUMOylated conjugates by phase separation before proteolysis by the ubiquitin-proteasome pathway. But what is the mechanism for selectivity? Are there any signals leading to SUMOylation of selective proteins by PML, both among existing and newly synthesized proteins? And how are the newly synthesized proteins trafficked to the nucleus to join the PML bodies? Furthermore, does PML contribute to the SUMO2/3 selectivity observed for the new modifications? These SUMO variants are more prone to form extended chains of SUMOs by conjugating incoming SUMO sequences to existing SUMO modifications (10); would these multivalent platforms help to initiate or stabilize the phase-separated PML nuclear body condensates? It will be fascinating to dig into these results more deeply. In addition, it will be important to determine the impact of the enhanced SUMOylation on cellular physiology. If SUMOylation is simply a mechanism to store proteins that cannot be degraded and would otherwise stress the cell, why did not the simultaneous inhibition of the SUMO and ubiquitin pathways significantly influence cell viability? There is clearly much to learn about this unexpected connection between ubiquitylation and SUMOylation.