The selectivity of the ubiquitin–26 S proteasome system (UPS) for a particular substrate protein relies on the interaction between a ubiquitin-conjugating enzyme (E2, of which a cell contains relatively few) and a ubiquitin–protein ligase (E3, of which there are possibly hundreds). Post-translational modifications of the protein substrate, such as phosphorylation or hydroxylation, are often required prior to its selection. In this way, the precise spatio-temporal targeting and degradation of a given substrate can be achieved. The E3s are a large, diverse group of proteins, characterized by one of several defining motifs. These include a HECT (homologous to E6-associated protein C-terminus), RING (really interesting new gene) or U-box (a modified RING motif without the full complement of Zn2+-binding ligands) domain. Whereas HECT E3s have a direct role in catalysis during ubiquitination, RING and U-box E3s facilitate protein ubiquitination. These latter two E3 types act as adaptor-like molecules. They bring an E2 and a substrate into sufficiently close proximity to promote the substrate's ubiquitination. Although many RING-type E3s, such as MDM2 (murine double minute clone 2 oncoprotein) and c-Cbl, can apparently act alone, others are found as components of much larger multi-protein complexes, such as the anaphase-promoting complex. Taken together, these multifaceted properties and interactions enable E3s to provide a powerful, and specific, mechanism for protein clearance within all cells of eukaryotic organisms. The importance of E3s is highlighted by the number of normal cellular processes they regulate, and the number of diseases associated with their loss of function or inappropriate targeting.
Over the last decade, our knowledge and understanding of the UPS (ubiquitin–proteasome system) has expanded enormously. In particular, we have come to realize the importance of E3s (ubiquitin–protein ligases) for substrate recognition and ubiquitination within the system. Particular interest in these molecules has developed since the discovery that E3s are often found to be mutated, absent or malfunctioning in many diseases, including neurodegenerative disorders and cancer (Table 1) [1–3]. Indeed, the pharmaceutical industry is currently assessing whether E3s are suitable therapeutic targets for such disorders .
E2s (ubiquitin-conjugating enzymes) are characterized by a highly conserved domain adjacent to the cysteine residue that forms the thioester bond with ubiquitin. By contrast, E3s are a much more diverse group of proteins in terms of size and domain structure. Although such diversity made their initial identification difficult, many E3s have now been characterized. These proteins generally contain one of several different defining E3 domain structures. These motifs are found in hundreds of sequences in mammalian, plant and viral genomes, leading to the suggestion that many of these may also be E3s. In general, like most components of the ubiquitin system, they are highly conserved throughout evolution.
Different types of E3 often display different modes of action (Figure 1). Whereas some promote monoubiquitination of substrate, others catalyse polyubiquitination. The specificity and timing of substrate ubiquitination may be dependent on a number of other factors. These may include the requirement for post-translational modification of the substrate, such as phosphorylation or hydroxylation. The respective intracellular locations of E3, substrate and post-translational modifying complex may also be critical. The combination of these factors presents the cell with a powerful and precise mechanism for the specific spatio-temporal clearance of individual proteins.
A major function of E3s is to regulate the polyubiquitination, and subsequent degradation, of target proteins. Complete breakdown of proteins, however, is not always the primary function of the ubiquitination. For example, polyubiquitination by the RING (really interesting new gene) E3, TRAF6 (tumour necrosis factor-receptor-associated factor 6), is required during IκB (inhibitory κB) processing and NF-κB (nuclear factor κB) activation.
Protein monoubiquitination has been shown to play key functional roles that are apparently independent of degradation. In particular, regulation of endocytosis of cell surface receptors, DNA-repair mechanisms and transcription regulation are controlled by monoubiquitination of target proteins .
Interestingly, the levels of E3s themselves can also be regulated by UPS-mediated proteolysis. Indeed, many E3s are capable of self-regulation by ‘autoubiquitination’ [1–4]. Autoubiquitination may have evolved as a gain-of-function, and protective, cellular mechanism. Programmed self-destruction would help to prevent damaging cellular consequences that may occur if intracellular concentrations of E3 became too high. This also raises the possibility that proteins initially identified as E3s as a consequence of their autoubiquitination activity may also display a secondary unrelated function. Loss of E3 autoubiquitination activity through mutation, for example, may not affect its secondary function. This could lead to increased intracellular concentration and secondary function. This property may be particularly relevant in neurological disorders associated with aging, because the UPS becomes less efficient at processing proteins as we age.
E3s: a diverse group of proteins
Three major classes of E3 have been identified, termed the HECT (homologous to E6-associated protein C-terminus), RING finger and U-box (a modified RING motif without the full complement of Zn2+-binding ligands) E3s. In addition, two subclasses of RING E3s have been defined: RIR (RING in between RING–RING) domain, and multi-protein complex E3s [CRL (Cullin-RING E3)] (Figures 2 and 3). However, there is also a minor group of proteins that are not characterized by any of these domain structures, but can act as E3s. For example, the DUB (deubiquitinating enzyme), UCH-L1 (ubiquitin C-terminal hydrolase L1), a protein highly expressed in the brain, displays dimerization-dependent E3 activity in vitro . Why an enzyme should have two opposing functions is somewhat intriguing and warrants further investigation. Whether other DUBs also share this dual function remains to be determined.
HECT domain E3s
One of the first E3s to be identified and characterized was E6-AP1 (E6-associated protein 1) (Figure 2) . Subsequently, many other proteins with a similar 350-amino-acid catalytic domain were identified through database searches. These HECT domain E3s function in a similar manner to E1 and E2s. Each contains a central cysteine residue within the HECT domain, which acts as an acceptor for ubiquitin. The thiol group of the cysteine residue forms a thioester bond with the carboxy group of the C-terminal glycine residue prior to its transfer to substrate (Figure 1). Whereas the C-terminal HECT domain confers E3 activity, the N-terminal domain of these large proteins, many of which are greater than 100 kDa, is involved in substrate binding.
The archetypical HECT E3, E6-AP1, was identified as a protein co-opted by the E6 oncoprotein of the transforming human papilloma virus types 16 and 18 to promote the degradation of the tumour suppressor protein p53 . E6-AP1 is not, however, the physiological E3 that regulates intracellular p53 levels. It appears that MDM2 (murine double minute clone 2 oncoprotein) undertakes this role [2,8]. In addition to being one of the first E3 proteins that was identified, loss of E6-AP1 activity was the first example of an inherited disorder caused by a mutation in a component of the UPS (, but see [9a], ). Angelman syndrome is a neurological disorder characterized by mental retardation, seizures and abnormal gait, with frequent smiling and laughter. The disease is caused by maternal imprinting at 15q11–q13, a region containing UBE3A, the gene that encodes E6-AP1 (, but see [9a], ). Maternal loss of activity of E6-AP1 can occur due to deletions, imprinting mutations, paternal uniparental disomy or point mutations in the UBE3A gene. Imprinting defects cause silencing of the paternal allele in neurons, but not in the glial cells of the brain. However, as the physiological substrates of E6-AP1 remain to be determined, the mechanism by which loss of activity causes the neurological phenotype remains unclear.
It is interesting that the gene encoding the human HECT protein, HERC2 [C-terminal HECT domain and three RCC1 (regulator of chromatin condensation)-like domains], and its pseudogenes also flank the chromosomal deletion that characterizes the Prader–Willi and Angelman syndromes at chromosome 15q11–q13 . Murine Herc2 is a large protein of 4834 amino acid residues that demonstrates a remarkable 95% identity and 99% similarity to human HERC2. Loss of Herc2 function leads to the neurological defects associated with the rjs (runty–jerky–sterile) mouse [11,12].
RING finger E3s
The RING finger family of proteins potentially represents the largest group of E3s, with hundreds of proteins containing a RING finger domain being present in mammalian genomic databases [13,14]. Although their structures suggest an E3 function, such an activity has not been proven in vivo in most instances. Unlike HECT domain proteins, RING finger E3s do not appear to have a direct catalytic role in protein ubiquitination. Instead, RING E3s act as scaffolding partners, facilitating the interaction between an E2 and a substrate (Figure 1) [13,14].
The RING finger domain is a cysteine/histidine-rich, zinc-chelating domain that promotes both protein–protein and protein–DNA interactions. It is defined as Cys1-Xaa2-Cys2-Xaa9–39-Cys3-Xaa1–3-His4-Xaa2–3-Cys/His5-Xaa2-Cys6-Xaa4–48-Cys7-Xaa2-Cys8 (where Xaa can be any amino acid residue) . Two zinc atoms are complexed by the cysteine/histidine residues in a ‘cross-brace’ manner, to provide correct folding and biological activity of the RING domain (Figure 4A). RING finger motifs are further subdivided, depending on whether a cysteine or histidine residue is found at Cys/His5 within the motif. Thus they are classified as being either a RING-HC (Cys5) or a RING-H2 (His5) type. The presence of Cys5 or His5 appears to be structurally important for E2 recognition, because substituting cysteine for histidine prevents E2–E3 binding for several RING-containing proteins, including HHARI (human homologue of Drosophila ariadne) . It is of note that the second RING domain that characterizes the RIR E3s displays a modified structure that is capable of binding only one Zn2+ ion .
c-Cbl is a RING finger E3 required for endocytosis and degradation of RTKs (receptor tyrosine kinases) [17,18]. It is required for the early-endosome-to-late-endosome/lysosome sorting step of EGFR (epidermal growth factor receptor) down-regulation. As many membrane receptors signal via a ‘triple membrane passing signal’ through the EGFR, the control of its levels and activity is fundamental to many signalling pathways. Endocytosis of EGFR by c-Cbl is dependent on phosphorylation. c-Cbl recognizes phosphorylated tyrosines on RTKs through its N-terminal SH2 (Src homology 2) domain. When the ligand binds to EGFR, its tyrosine kinase activity is activated, catalysing autophosphorylation. In the presence of endophilin and Cbl-interacting protein of 85 kDa (also called SH3-domain kinase binding protein 1), c-Cbl then causes rapid endocytosis of EGFR into endosomes. This process is facilitated by receptor ubiquitination. The receptor can then be recycled or processed by the ESCRT (endosomal sorting complex required for transport) system.
Another member of the Cbl family is Cbl3. This protein is structurally similar to c-Cbl, but it lacks the UBA (ubiquitin-associated) domain and proline-rich domains. Cbl3 also regulates the EGFR; however, in this instance it requires an interaction with the HECT E3, ITCH (itchy homologue E3 ubiquitin–protein ligase), before ubiquitination can occur . Interestingly, ITCH itself does not interact directly with EGFR. This suggests that Cbl3 may be required for the initial ubiquitination step with ITCH, enhancing polyubiquitination of the receptor. Cbl proteins have also been described as suppressors of other EGFR superfamily members, including the HER2/Neu oncogene. Up-regulation of EGFR-superfamily signalling is a major contributor to uncontrolled proliferation in many malignant conditions.
As noted above, p53 is usually targeted for cellular removal by the RING finger protein, MDM2 [2,8]. Ubiquitination of p53 by MDM2 may also regulate the nuclear export of p53 to the cytoplasm for proteasomal degradation, although this role of MDM2 remains controversial . MDM2 regulation of p53 turnover is complex. Under normal cellular conditions, p53 up-regulates MDM2 gene expression while MDM2 down-regulates p53. Other proteins, including cyclin-dependent kinase 2A (CDKN2A) and MDMX, and post- translational modifications (such as phosphorylation of both p53 and MDM2) add extra layers of complexity to the regulation of p53 by MDM2. Whereas ARF acts to inhibit MDM2 activity, MDMX enhances its activity. MDMX is a RING finger protein highly related to MDM2. The MDM2–MDMX dimer forms via RING domain interactions between the two proteins and prevents autoubiquitination of MDM2. The resultant increased stability of MDM2 was expected to result in lower levels of p53. However, it also caused the stabilization of p53. This occurs because the MDM2–MDMX complex ubiquitinates p53, but the latter polyubiquitinated product accumulates in the nucleus and is not degraded by the 26 S proteasome. Interestingly, when CDKN2A blocks MDM2's activity towards p53, MDM2 targets the degradation of MDMX.
Complex RING E3s
The RIR E3 family consist of a central Cys/His-rich region called the IBR [in between RING fingers; alternatively termed the double RING linked (DRIL)] domain. It links a classic RING finger domain to a second RING-finger-like or ACC (accessory C-terminal zinc finger) domain [20,21]. This specialized group of E3s appear to be RING-HC type E3s. They rely on the highly homologous E2s, UbcH7 (ubiquitin-conjugating enzyme H7) or UbcH8, for their activity. The RING1 domain is structurally similar to many other RING-finger-containing proteins, such as c-Cbl (Figure 4B). However, it is generally a little larger and is essential for E2 and substrate binding. The IBR domain is characterized by the canonical structure, C6HC . The purpose of the IBR remains unknown. It may promote protein–protein interactions or, alternatively, it may act as a spacer, or flexible region, that accurately positions the RING1 and ACC domains for E2/substrate recognition and binding. Indeed, a 20-amino-acid essential spacer region is required for UbcH7 binding to the RIR of HHARI . The RING2, or the ACC, domain is smaller (but similar in size between family members) than RING1. Although this was originally thought to be an additional RING finger, structural studies have now established that this domain has a distinctly different topology from classical RING finger domains . The Cys1, Cys2, Cys4 and Cys6 residues are required for binding of only a single zinc atom. As a consequence, the unique structure of the ACC domain resembles a zinc ribbon motif with a ‘criss-cross’ appearance, similar to that of subunit Rbp12 of RNA polymerase II .
Whereas it is generally accepted that RING1 is essential for E3 activity, the ACC may also be required for the recognition and binding of both E2 and substrate. This is the case with Parkin. Although the ACC domain in isolation can interact with E2s and cause ubiquitination in in vitro assays , they have not been identified independently of the RIR. Hence, the significance of this observation to in vivo activity remains unclear.
RIR E3s, such as Parkin, can act both alone with the E2 to facilitate ubiquitination, or with cullin-containing proteins, as part of a multi-protein complex (Figure 5). Parkin mutations are responsible for more than 50% of all cases of AR-JP (autosomal recessive juvenile Parkinsonism) [22–24]. AR-JP patients suffer from an early onset form of Parkinson's disease, with symptoms often manifesting in the third decade of life. Because of the importance of Parkin, many studies have been undertaken to identify substrates of this E3, with the aim of understanding why it causes disease. Substrates include synphilin-1, synaptotagmin XI and Pael-R (Parkin-associated endothelin-receptor-like receptor) [24–27]. It is proposed that Pael-R accumulates in the ER (endoplasmic reticulum) of dopaminergic neurons to cause ER stress and neurodegeneration in Parkin-associated AR-JP patients. Under normal conditions, Parkin degrades unfolded Pael-R before it is able to accumulate . This mechanism involves the chaperone hsp70 (heat-shock protein 70), and CHIP (C-terminus of the hsp-70-interacting protein) (Figure 5B) . Both CHIP and hsp70 interact with Parkin to regulate Pael-R degradation. The E3 activity of Parkin is inhibited when hsp70 transiently forms a complex with unfolded Pael-R and Parkin. The presence of CHIP, however, causes the dissociation of hsp70 from the Pael-R–Parkin complex, enhancing Parkin activity and breakdown of the Pael-R protein. The interplay between hsp70, Parkin and CHIP is particularly interesting, given that CHIP can independently act as both a co-chaperone and an E3 (see the section on U-boxes below). These combined activities provide a direct link between the chaperone system and the UPS.
By contrast, Parkin has also been demonstrated to be part of an SCF (Skp1–Cul1–F-box protein)-like E3, composed of Cullin-1 and the F-box protein Fbw7, which targets cyclin E for degradation (Figure 5C) . Interestingly, Skp1 is not required for E3 activity in this complex . Since elevated levels of cyclin E cause apoptosis in post-mitotic neurons, AR-JP-associated Parkin mutations may also cause neuronal loss because of increased cyclin E-mediated apoptosis.
The flexible nature of Parkin E3 activity may explain why it can target many different substrates. Furthermore, the broad range of cellular functions of these targets may explain why the molecular mechanisms underlying Parkinson's disease have been so difficult to elucidate.
The multi-component, modular CRLs are a large and diverse group of ligases [3,4,14,29]. They comprise at least a Cullin and a RING protein (Figure 3). The principal groups are the F-box (in SCF), BTB (broad complex, tramtrak, bric a brac) and SOCS (suppressor of cytokine signalling)/BC (elongin B, elongin C) types of E3 complexes (Figures 3A and 3B) [4,13,29]. The Cullin proteins (CUL1, CUL3 and CUL2A, respectively) provide the central scaffolding component of each of these complexes. For example, for the SCF complexes, Skp1 interacts with its N-terminal sequence to recruit the substrate receptors (F-box proteins). By contrast, Rbx1 (RING box protein 1) (or ROC1/HRT1) binds the C-terminal domain to recruit the E2 to form the active complex. The C-terminal region is also modified by the ubiquitin-like protein, NEDD8 (neural precursor cell expressed developmentally down-regulated 8) [4,14,29]. This post-translational modification appears to improve the ligase activity of the E3, possibly by enhancing the binding of ubiquitin-loaded E2s.
The presence of multiple F-box, BTB and SOCS/BC box proteins in genome databases suggests that changing just one component of these complexes can potentially allow the formation of many different E3s, each with their own targeting specificity. For example, SCFβTrCP (SCF β-transducin repeat-containing protein) can target the degradation of β-catenin or IκBα, whereas substrates of SCFSKP2 include p27KIP1, c-Myc and cyclin E. Recognition of these substrates often requires their phosphorylation prior to their targeted ubiquitination.
In the case of β-catenin, phosphorylation requires a complex of GSK-3β (glycogen synthase kinase 3β), axin and the APCp (adenomatous polyposis coli tumour suppressor protein). UPS-mediated breakdown of β-catenin is essential to limit the amount of cytosolic β-catenin entering the nucleus, thereby inhibiting its role as an activator of TCF4 (T-cell factor 4)-driven transcription. Without this tight regulation, increased levels of transcriptional activation of downstream genes occur. Mutations in APCp or β-catenin that prevent their interaction lead to the malignant transformation of colonic epithelium.
The VCB [VHL (von Hippel–Lindau) E3 or VHL elongin C–elongin B] complex is made up of the VHL tumour suppressor, elongins B and C (ubiquitin-like and Skp1-related, respectively), Rbx1 and an E2 (Figure 3B) [4,14]. Despite its ubiquitin-like homology, the role of elongin B within the complex is unclear. The VCB is involved in down-regulation of the HIF-1α (hypoxia-inducible factor-1α) and HIF-2α. Under normoxic conditions, HIF-1α is hydroxylated and targeted for degradation by VCB. However, under hypoxic conditions, hydroxylation of the key proline residue in HIF-1α/2α does not occur. This has the consequence that HIF-1α/2α is stabilized, as it is no longer recognized by the VCB complex. Mutations in VHL also result in a loss of VCB activity. The consequence of stabilization of HIF-1α/2α is increased transcription of downstream genes, such as that encoding vascular endothelial growth factor. This can lead to the development of a variety of highly vascularized tumours, such as those observed in renal cell carcinoma.
The APC (anaphase-promoting complex) is key to many cell cycle events, including chromosome segregation and exit from mitosis. It is a large multi-component complex consisting of at least eleven subunits (Figure 3C) [3,4,14]. Its core structure is highly similar to that of the SCF complexes. The APC2 subunit contains a cullin-like domain and APC11 is a RING finger protein which binds the E2. The interchangeable regulatory subunits Cdc20 (cell-division cycle 20) or Cdh1 (Cdc20 homologue 1; also known as fizzy) are required for its activation . They act in a similar way to the F-box proteins, providing substrate specificity. The additional subunits are likely to act as scaffolding molecules, promoting interactions that bring a substrate and the E2–E3 core complex into sufficient proximity to promote ubiquitination of its various substrates. Indeed, recent data suggest that APC1, APC4 and APC5 are involved in polyubiquitin chain assembly, APC3 and APC7 can recruit Cdh1 to the APC complex, APC10 is involved in substrate recognition, and APC9 provides structural stability to the complex [30,31].
The U-box is a 74-amino-acid domain that is structurally similar to the RING finger domain [32,33]. However, it lacks the key residues required for metal chelation. It has been proposed that it utilizes salt bridges to maintain its structure. The U-box was first described in the yeast protein Ufd2 (ubiquitin fusion degradation protein 2), a protein involved in ubiquitin-chain elongation in conjunction with an E1, E2 and a HECT E3. This new type of enzyme was termed an E4. Initially, all U-box proteins were thought to be auxiliary proteins, supplementing the activity of E2–E3 interactions. However, subsequent studies have demonstrated that other U-box proteins can interact with E2s and display E3 activity independently of another E3 [32,33]. Whether these proteins act as an E3 or E4 may depend on which components of the UPS are present, and the nature of the substrate. Given their similarity in structure, it is perhaps not surprising that U-box E3s show similar modes of action to the RING E3s, sometimes acting alone, and sometimes as part of multi-protein complexes.
CHIP is the best characterized of the U-box proteins [32,33]. It was first identified as a co-chaperone that interacts with hsp70 or hsp90 through the N-terminal tetratricopeptide repeat regions. It is involved in quality control, regulating decisions between folding of the protein by the chaperones or degradation by the UPS [26,32,33]. Recently, CHIP was demonstrated to be an E3 that uses UbcH5 as its E2 partner. Its substrates include: the cystic fibrosis transmembrane conductance regulator; the Alzheimer's-disease-associated protein, tau; the glucocorticoid receptor; and the transcription factor E2A [32–35]. Hsp70 or hsp90 appear to direct the substrate specificity of CHIP.
Analogous to RING E3s, CHIP also interacts with other U-box proteins. For example, the Caenorhabditis elegans orthologues of CHIP and Ufd2 interact to degrade unc-45, a (co-)chaperone required during thick myosin filament assembly . Furthermore, as mentioned above, CHIP has a role in enhancing Parkin E3 activity .
E3 proteins are now recognized as performing a pivotal role in ubiquitin-mediated intercellular protein degradation. The majority of E3s are proteins that contain HECT, RING or U-box motifs within their structure. Different families of E3s have members which can act alone, with other proteins or E3s, or as part of larger, multi-protein complexes. Under certain circumstances, additional proteins may also be required to enhance the activity of an E3. E4 proteins may act in conjunction with E3s, but some also have E3 properties themselves, making the characterization of these molecules difficult. Other proteins containing DOC (deleted in oral cancer), SOCS or CUE (coupling of ubiquitin conjugation to ER degradation) domains (involved with substrate recognition and interaction with E2s respectively) can also enhance ubiquitination reactions.
E3 proteins, with their myriad of combinations and their need for post-translational modifications and/or additional proteins, ensure the precise spatio-temporal targeting and breakdown of individual substrates within cells. Much is still to be learnt about the regulation of protein levels by this complex group of proteins, especially in vivo.
E3s are essential for substrate recognition and ubiquitination.
Substrate recognition may require post-translational modification of the substrate.
The major classes of E3 proteins contain either HECT, RING or U-box domains.
HECT domain proteins are directly involved in ubiquitination of substrate proteins. By contrast, RING and U-box proteins act as facilitators of ubiquitination reactions.
Many RING finger proteins are core components of multi-protein E3 complexes.
Many RING finger E3s interact with other E3s, thereby promoting greater targeting specificity or enhanced ubiquitination.
Many E3s possess the potential to regulate their own levels through autoubiquitination.
Literature covering the topic of E3s is increasing at an exponential rate. Consequently, it has been impossible to cite many exceptional articles within the constraints of this review. We apologize to those whose work we have not quoted. However, many additional E3 references can be found at: http://www.leeds.ac.uk/medicine/res_school/mol_med/res_robinson.htm
H.C.A is a recipient of a Research into Ageing Fellowship Award. Work in the authors' laboratory is also supported by The Parkinson's Disease Society (UK) and Yorkshire Cancer Research. H.C.A. would like to dedicate this essay to Jessica and Thomas Carey.
- © 2005 The Biochemical Society