Ubiquitination, the covalent attachment of the small protein modifier ubiquitin to a substrate protein is involved in virtually all cellular processes by mediating the regulated degradation of proteins. Aside from proteasomal degradation, ubiquitination plays important roles in transcriptional regulation, protein trafficking, including endocytosis and lysosomal targeting, and activation of kinases involved in signalling processes. A three-tiered enzymatic cascade consisting of E1 or ubiquitin-activating enzyme, E2 or ubiquitin-conjugating enzyme, and E3, or ubiquitin ligases, is necessary to achieve the many forms of ubiquitination known to date. In this chapter, we summarize the current knowledge on the enzymatic machinery necessary for ubiquitin activation and ligation, as well as its removal, and provide some insight into the complexity of regulatory processes governed by ubiquitination.
Since its discovery approximately 30 years ago, Ub (ubiquitin) and the process of its covalent attachment to protein substrates, ubiquitination, but also its removal, deubiquitination, have taken centre stage in the research into how proteins are turned over and how their activity can be modulated. Although proteins were looked at as quite static entities with little or no turnover up until the 1930s, pioneering work by Schoenheimer revealed that the body's proteins are subject to constant renewal. Early work into protein turnover were focused on the role of the lysosome and its resident proteases. However, mounting evidence regarding the specificity of protein turnover called also for a more specific protein degradation system.
The realization that the protein Ub, or as it was known at the time APF-1 [1,2], is transferred in an energy-dependent process to other proteins and causes their degradation provided the long-needed epiphany for a lysosome-independent protein degradation system. With the availability of the human genome it became clear that a plethora of gene products is devoted to ubiquitination. In addition to the Ub-activating enzyme, Uba1 or E1, and more than 50 Ub-conjugating enzymes, Ubcs or E2s, several hundred Ub ligases or E3s were identified in the human genome. This large amount of E3s, the fact that more than one E2 can activate a given E3, together with the possibility of differential Ub chain formation and chain editing through DUBs (deubiquitinating enzymes), allows a glimpse into the many layers of regulation and different outcomes that are achieved by ‘simply’ attaching Ub to a substrate protein.
The 76-amino-acid long Ub is quite conserved throughout all eukaryotic organisms, but is absent from bacteria, whereas proteins with similar functions were also found in archaea. For example, budding yeast Ub differs only in three amino acid positions from human Ub, although fungi and humans are separated by approximately one billion years of evolution. This high degree of conservation underlines the importance of ubiquitination.
Ub is made as either a head-to-tail fusion protein between Ub moieties or attached to certain ribosomal subunits from where it is released by the action of certain Ub-specific proteases (see DUBs) and further matured to expose the characteristic di-glycine C-terminus.
Attachment of Ub occurs through an isopeptic link between the glycine residue at its very C-terminus and lysine residues in the substrate protein. Ub itself possesses seven lysine residues (Lys6, Lys11, Lys27, Lys29, Lys33, Lys48 and Lys63) whose ε–amine groups can serve, together with the N-terminal amine group, as acceptor sites for another Ub moiety (Figure 1A). This can lead to the formation of differently shaped polyubiquitin chains (Figure 2) that serve different functions. Chains of at least four Ub moieties linked together via Lys48 are the major form of ubiquitination and are signals for proteolytic degradation of substrate proteins by the 26S proteasome, a large, barrel-shaped, cytosolic protease. Ub moieties linked via Lys63 are the second most common mode for chain formation. Such Lys63 Ub chains do not mediate protein degradation, but are involved in the regulation of DNA repair and signal transduction, as well as trafficking/sorting of membrane proteins. So called non-canonical Ub chains, those not formed via Lys48 or Lys63, are less well studied, but seem to also target proteins to proteasomal degradation . Besides these homotypic chains where all chain links are the same, mixed-linkage chains might also occur and have some as yet unknown function . But chain formation is not the only way for Ub to exert a regulatory function. Another mode of ubiquitination is monoubiquitination or multi-ubiquitination (multiple occurrence of monoubiquitination as opposed to polyubiquitination), which modulates the activity rather than the stability of proteins.
Ubiquitination exerts its regulatory function by attaching docking sites to substrates in order for different UBD (Ub-binding-domain)-containing effector proteins to bind . Such UBDs include the UIM (Ub-interaction motif), the MIU (motif interacting with Ub), the UBA (Ub-associated domain) and others. Most Ub-binding domains bind to Ub via a hydrophobic patch surrounding a crucial Ile44 (Figure 1B). The use of this single interaction surface on Ub prevents binding of different effector proteins to the same substrate, thus triggering a consistent regulatory response for a given substrate ubiquitination.
Ub-activating and -conjugating enzymes
To achieve covalent attachment to substrate proteins, Ub needs to be activated. This reaction starts with the C-terminal adenylation of Ub by Uba1 or E1 (Figure 3), followed by the formation of an energy-rich thioester bond between the thiol group of an active site cysteine residue in Uba1 and the carbonyl group of the C-terminal glycine residue of Ub (Figure 3). After recruitment of a second Ub and its adenylation, activated Ub is transferred in a trans-thioesterification reaction to one of the 40–50 Ubcs, again forming a thioester intermediate ready for transfer to a target protein (Figure 3). Ubcs are capable of transferring activated Ub directly to substrate proteins; however, a specificity factor in the form of an Ub ligase is normally needed for substrate recognition. In addition, Ubcs play an important role in the outcome of ubiquitination by influencing the linkage of Ub chains. Thus Ubcs are not merely Ub carriers for Ub ligases, but also greatly shape the outcome of the ubiquitination reaction.
Proteins with E3 activity or Ub ligases catalyse the transfer of activated Ub from an active-site cysteine residue of an E2 on to a lysine residue of a substrate protein. As mentioned above, several hundred potential E3s can be identified in mammalian genomes, thus providing substrate specificity to the process of ubiquitination. E3s come in two different varieties . There is the somewhat smaller class of so called HECT (homologous with E6-associated protein C-terminus) domain proteins and the large class of RING (really interesting new gene) finger proteins. The HECT domain is named after the conserved C-terminus of a cellular Ub ligase (E6-associated protein) that is co-opted by the E6 protein of the human papilloma virus to facilitate the degradation of p53 in a sort of host evasion mechanism. RING finger proteins belong to the larger superfamily of zinc-finger proteins. While zinc fingers are a general interaction motif and confer protein–protein, protein–DNA or protein–lipid binding, RING fingers form a specialized zinc finger subclass that is thought to confer binding to Ubcs, thus RING finger domains are a telltale sign for Ub ligase activity. Recently, a motif related to the RING finger, the U box domain, was described where salt bridges and hydrogen bonds instead of zinc ions stabilize this interaction domain. The U box most probably evolved from the RING finger domain exchanging zinc ions to salt bridges along the way. HECT and RING finger E3 ligases differ in their mode of action. While HECT domain proteins possess an active-site cysteine residue, just like E2 enzymes, and are therefore capable of forming an Ub-intermediate, RING finger proteins are not enzymes in the classical sense, but rather scaffold proteins that recognize substrate proteins and bring them together with activated Ub, thus promoting Ub transfer directly from an E2 on to target proteins.
It is still debated how exactly Ub chains are formed, whether pre-existing chains are attached or whether Ub chains are assembled in a consecutive fashion on the substrate. Certain proteins, referred to as E4s, are thought to be processivity factors aiding and shaping chain formation in concert with Ub ligases. These proteins with E4 activity are E3s themselves that, for specific substrates, co-operatively work with another E3 to enhance Ub chain formation. Whether an E3 has E4 activity is context- and substrate-dependent . For example, one such E3–E4 pair is formed by the Ub ligases Ubr1 (RING) and Ufd4 (HECT). Both ligases act co-operatively in two different pathways, the so-called N-end rule, as well as the Ub-fusion degradation pathway, with reversed functions. While Ubr1 provides initial recognition in the N-end rule pathway, Ufd4 enhances the formation of Ub chains and thus is considered an E4 activity for this pathway. These roles are reversed in the Ub-fusion pathway, where Ufd4 is the initial E3 enzyme and now Ubr1 enhances the processivity of chain formation on these substrate proteins exerting E4 activity. The definition of E4 is less well defined compared with the other components of the ubiquitination system. The concept of distinct E4s proposed earlier is now replaced by a more context-specific assignment of E4-like or E4 activity for certain E3 ligases under specific circumstances.
Although less well studied in comparison with Ub ligases, DUBs represent a large group of proteases with specificity for Ub, Ub conjugates and Ub chains . Over 100 genes potentially coding for DUBs could be identified in the human genome. DUBs come in five different varieties. While the JAMM (JAB1/MPN/Mov34) domain DUBs belong to the superfamily of zinc-dependent metalloproteases, UCH (Ub C-terminal hydrolase), USP (Ub-specific protease), OTU (ovarian tumour domain) and Josephin domain DUBs are classical papain-like cysteine proteases. The large number of DUBs, as well as their great diversity, suggests substrate-specificity on the one hand, while also hinting at important regulatory roles of this protein family. Initially, DUBs were recognized for their role in processing Ub precursor proteins to generate mature Ub, and in the recycling of Ub chains after proteolytic destruction of substrate proteins by the proteasome (Figure 3). Beyond that, DUBs provide reversibility to the process of ubiquitination and allow modifcation, editing or nixing decisions to be made by the ubiquitination machinery. In a sense, DUBs are to E3 ligases what phosphatases are to kinases. And just as the reversibility of phosphorylation is the key to its regulatory flexibility, DUBs make ubiquitination even more complex since they are not only able to remove, but also to edit, ubiquitination. As is the case with all proteases, DUB activity has to be tightly controlled to prevent inadvertent processing of ubiquitinated substrate proteins. Oftentimes, DUB activity is cryptic and the DUB active site is only accessible after the binding of a substrate, mostly through interaction motifs recognizing Ub or Ub chains. In addition to Ub-interacting domains, most DUBs possess other protein–protein interaction domains involved in integrating DUBs into higher-order protein complexes as a means to provide substrate recognition and to restrict DUB activity spatially and temporarily.
Examples of complex regulations involving ubiquitination
Ubiquitination is in some way or another involved in the regulation of basically all cellular processes. The following examples illustrate the many cellular uses of ubiquitination in maintaining cellular homoeostasis. Giving a complete overview of the many interesting and noteworthy examples for the use of ubiquitination is beyond the scope of this chapter. The reader is referred to the many excellent reviews to gain insight into cell-cycle regulation , kinase activation and its role in transcriptional regulation through NF-κB (nuclear factor κB) , the influence of (de)ubiquitination on the execution of cell death  and its connection to cancer development , maintaining mitochondrial morphology and function [13,14], protein sorting and trafficking , and many other interesting regulatory mechanisms ultimately governed by the enzymes of reversible ubiquitination (Figure 4).
Non-proteolytic histone ubiquitination
Most of the time ubiquitination equals protein degradation, proteasomal or lysosomal; ironically, the first protein found to be ubiquitinated was H2A (histone 2A), a rather stable protein. Histones are the major constituents of chromatin and form the so-called nucleosomes which nuclear DNA is wrapped around. Nucleosomes are general transcriptional inhibitors; however, modification of histone constitutes a major mechanism by which the structure, and therefore function, of chromatin is modulated. Monoubiquitination especially, as part of a so-called histone code, is used by the cell to regulate the availability of DNA for transcription. Monoubiquitination of H2A is generally considered to confer transcriptional repression , whereas monoubiquitination of histone H2B is associated with actively transcribed DNA regions (Figure 5). Transcriptional silencing through H2A ubiquitination is established by PRC1L (polycomb repressive complex 1-like), a protein complex containing the Ub ligase RING2 . Genes under the control of polycomb-response elements are the target of H2A ubiquitination and thus silencing. RING2 is also part of other repressors with different specificities such as BCOR (BCL6 co-repressor) and the E2F6–com-1 complex. Another H2A Ub ligase is 2A-HUB, a subunit of a large repression complex [N-CoR (nuclear receptor co-repressor)/HDAC (histone deacetylase) 1/3] involved in the repression of chemokine genes in response to TLR (Toll-like receptor) activation.
According to the current concept of H2A ubiquitination, different co-repressor complexes containing different Ub ligases are recruited to specific genetic loci on the basis of the specificity of the respective co-repressor.
Although H2A monoubiquitination is found only in higher eukaryotes, H2B monoubiquitination is also found in yeast, supporting a more conserved role for this histone modification. RNF20 and RNF40 (RING finger proteins 20 and 40 repectively) seem to be the major Ub ligases for H2B in mammals . As mentioned above, H2B ubiquitination is associated with actively transcribed genes. In fact, RNF20/40 activity towards histone depends on the activity of RNA polymerase II and occurs during transcription to regulate transcript elongation. Interestingly, H2B Ub marks are not static, but are subjected to constant renewal [19,20]. At least for some genes, optimal transcription seems to require a cycle of ubiquitination/deubiquitination. H2B–Ub supports early transcriptional events, whereas the DUB USP22-associated, so-called SAGA (Spt-Ada-GCN5 acetyltransferase) co-activator complex, mediates H2B deubiquitination and promotes actual transcription. Another DUB, USP7/HAUSP, maintains gene silencing by removing activating H2B Ub marks, for example in loci involved in development. USP7-dependent deubiquitination, however, also influences H2A modification. It was shown to be part of the above-mentioned polycomb complex and reverses RING2 autoubiquitination, causing its stabilization .
Another interesting non-proteolytic role of histone ubiquitination is mediated by the Ub ligases RNF8 and RNF168. Upon a DNA DSB (double-strand break), arguably a major cellular incident, repair mechanisms are set in motion to deal with the damage . RNF8 is one of the first responders on the scene. Immediately after the occurrence of a DSB, histone H2AX is phosphorylated by the kinases ATM (ataxia telangiectasia mutated) and DNA-dependent protein kinases. This initial phosphorylation causes the recruitment of MDC1 (mediator of DNA damage checkpoint 1) and other proteins to the DSB, including RNF8, which binds to phosphorylated MDC1. With the help of the Ub conjugase Ubc13, RNF8 causes a Lys63-linked Ub chain formation on histone H2A(X). This initial ubiquitination serves as a platform for the binding of RNF168 which contains so-called MIUs. RNF168 also uses Ubc13 to form Lys63-linked Ub chains on histones. It is thought that these Ub chains serve as docking sites for further repair proteins  including three more Ub ligases, RAD18, BRCA1 (breast cancer early-onset 1)  and HERC2 .
Interestingly, DUBs are also actively involved in the response to a DSB, such as BRCC36 , USP16 and USP3, and are thought to counteract the DSB response and to eventually mediate reactivation of genes repressed in reaction to a DSB event.
ER (endoplasmic reticulum)-associated degradation (ERAD)
The most acknowledged function of ubiquitination is the activation of proteasomal degradation for a given substrate protein. The ER is a major site for cellular protein production and folding, but has equally important roles in protein quality control. One-third of all cellular proteins, among them most membrane proteins, are routed through the ER creating a crowded challenging environment for protein folding. In order to keep the cellular proteome in order, rigorous quality control is necessary. Especially under stress conditions with a large amount of unfolded, misfolded and damaged proteins, the cell evokes the UPR (unfolded protein response), meaning a shutdown of general protein production, an increase in chaperone expression, and ERAD to remove misfolded and damaged proteins . Protein ubiquitination through ER-membrane-anchored Ub ligases plays a central part in ERAD. With 24 membrane-anchored RING finger Ub ligases , the ER membrane possesses the highest E3 density of all cellular membranes. This is a testament to the high need for protein turnover in the ER. Even under non-stress conditions, it is estimated that many proteins, e.g. CFTR (cystic fibrosis transmembrane conductance regulator), are poor folding substrates, and that in some cases over 90% of the proteins produced never reach a productive fold and get degraded in the ER. Chaperones in the ER membrane and the ER lumen, such as calnexin and calreticulin respectively, monitor successful protein folding and protect folding proteins from degradation, or shunt terminally misfolded proteins towards degradation. Major Ub ligases responsible for the ubiquitination of such misfolded proteins are Hrd1/synoviolin  and gp78/Doa10, together with the E2 Ubc7/UBE2G2 (Figure 6). At least in the model system budding yeast, Hrd1 seems to favour substrate proteins with lesions in the luminal or membrane-spanning part of the protein, whereas substrates for Doa10 are predominately damaged in their cytosolic portions [29,30]. In order for ubiquitinated proteins to be degraded, they have to be removed from the ER since the ER lumen is devoid of E1, E2 and proteasomes, which are all located in the cytosolic compartment. In a process called retrotranslocation, ubiquitinated proteins are unfolded and pulled through a protein conduction channel by the hexameric AAA (ATPase associated with various cellular activities)-ATPase p97 with the help of the adaptor proteins Ufd1 and Npl4. Interestingly, DUBs interacting with p97 seem to be necessary for the completion of protein translocation from the ER to the cytosol for degradation, potentially to allow the unfolded protein to pass through the p97 complex by removing a bulky polyubiquitin chain . There is still debate about the nature of the protein conducting channel involved in ERAD. There is very convincing evidence that the protein import channel formed by Sec61 is also involved in the export of proteins for degradation. However, recent data suggest that Derlin-1, an ER membrane protein, or the membrane-spanning ERAD ligases themselves can form protein-conducting channels for protein retrotranslocation . It seems that there exist several, maybe interdependent, pathways for protein export from the ER during ERAD.
UPR and ERAD gone wrong can cause severe diseases, as exemplified by cystic fibrosis where mutated forms of the anion channel CFTR are subject to exaggerated quality control, and are thus unable to reach their final target membrane to fulfill their function. In cancer cells burdened with ER stress due to hypoxia or nutrient deprivation , a deregulated overactive UPR and ERAD prevents these cells from undergoing ER stress-induced programmed cell death.
Conclusions and perspectives
The all-inclusive role of Lys48-linked ubiquitination in protein degradation is widely acknowledged and, so it seems, well researched. However there are still several hundred suspected Ub ligases encoded in the human genome without proper characterization, a big challenge here is the straightforward identification of physiological Ub ligase substrates. In addition, further research into the role of non-canonical Ub linkages will most probably prove to be a fruitful line of investigation.
• The attachment of Ub to substrate proteins depends on a three-tiered enzymatic cascade.
• All lysine residues of Ub are used to form Ub chains.
• Ubiquitination influences virtually all cellular processes.
• Ubiquitination often, but not always, equals proteolytic degradation of substrate proteins.
The work in our laboratory is supported by the Swiss National Foundation, the Grieshaber Foundation for Eye Research and the Messerli Foundation, Zurich. We also want to thank Charles Hemion for careful reading of the chapter.
- © The Authors Journal compilation © 2012 Biochemical Society