Review article

E2s: structurally economical and functionally replete

Dawn M. Wenzel, Kate E. Stoll, Rachel E. Klevit

This article has a correction. Please see:


Ubiquitination is a post-translational modification pathway involved in myriad cellular regulation and disease pathways. The Ub (ubiquitin) transfer cascade requires three enzyme activities: a Ub-activating (E1) enzyme, a Ub-conjugating (E2) enzyme, and a Ub ligase (E3). Because the E2 is responsible both for E3 selection and substrate modification, E2s function at the heart of the Ub transfer pathway and are responsible for much of the diversity of Ub cellular signalling. There are currently over 90 three-dimensional structures for E2s, both alone and in complex with protein binding partners, providing a wealth of information regarding how E2s are recognized by a wide variety of proteins. In the present review, we describe the prototypical E2–E3 interface and discuss limitations of current methods to identify cognate E2–E3 partners. We present non-canonical E2–protein interactions and highlight the economy of E2s in their ability to facilitate many protein–protein interactions at nearly every surface on their relatively small and compact catalytic domain. Lastly, we compare the structures of conjugated E2~Ub species, their unique protein interactions and the mechanistic insights provided by species that are poised to transfer Ub.

  • RING E3
  • HECT (homologous with E6-associated protein C-terminus) E3
  • ubiquitin ligase enzyme (E3)
  • ubiquitin (Ub)
  • ubiquitination
  • ubiquitin-conjugating enzyme (E2)


In 1983, while studying the regulation of protein degradation, Hershko and co-workers discovered three enzymatic activities required for the ATP-dependent addition of the small modifier protein Ub (ubiquitin) to protein substrates [1]. Over the last three decades, much has been learned about the structures and activities of the E1 Ub-activating enzymes, the E2 Ub-conjugating enzymes and the E3 Ub ligases that comprise the signalling pathway. From this body of work a general paradigm for the stepwise transfer of activated Ub from E1 to a lysine side chain on a protein substrate has been developed, as depicted in Figure 1; for simplicity and clarity, we illustrate only the canonical pathways of RING and HECT (homologous with E6-associated protein C-terminus) E3s. These will be elaborated in the text to describe the emerging more complicated details of the Ub system. Fundamental questions remain: how do E2s and E3s function in a concerted fashion to target specific proteins for modification? What structural factors of E2s, E3s and Ub govern how a protein is modified and which specific Ub signals are generated? A central player in Ub transfer choreography is the E2. Once thought to serve merely as carriers of Ub from E1 to an E3–substrate complex, it has become increasingly clear that E2s must perform multiple functions. An E2 must interact with an E1, form a covalent thioester conjugate with the C-terminal tail of Ub and function with a variety of E3s. Recent work has also demonstrated that E2s play an important role in dictating the final product, be it monoUb or a polyUb chain, of specific lysine linkage [24]. This realization demands that the simple model for E2 function illustrated in the cartoon of Figure 1 be modified and expanded. The fast pace of research in the general area of protein ubiquitination and, in particular, Ub-conjugating enzymes is evidenced by an abundance of recent review articles [510]. In the hope of avoiding redundancy, we present newly emerging themes in E2 structure, function and mechanism not covered in the two recent reviews on E2s [9,10]. In particular, we focus on new non-canonical ways in which E2s interact with other proteins and what additional insights these provide about the mechanism of their activity. While we focus the present review on Ub E2s, when illustrative, we draw parallels with the related Ubl (Ub-like) SUMO (small ubiquitin-related modifier) pathway to better elucidate emerging patterns in the Ub system.

Figure 1 The 1, 2, 3s of protein ubiquitination

The simplified schematic shows the three enzymatic activities associated with the central paradigm of protein ubiquitination: E1, the Ub-activating enzyme; E2, the Ub-conjugating enzyme; and E3, the Ub ligase. Mechanistically, there are two types of E3s, the RING/U-box-type E3s that effect transfer of Ub directly from the active site of an E2 to a lysine residue of a substrate, and the HECT-type E3s that facilitate Ub transfer from an E2 to substrate via a thioester intermediate on the E3. Auto-ubiquitination of the E3 is also observed in some cases and can be used as a proxy signal to assay for the activity of an E2–E3 pair in the absence of a substrate. The best studied fate of polyubiquitinated substrates is degradation by the proteasome (not depicted).

The sheer number of E2s and the fact that many are well-behaved small soluble proteins have combined to generate a large amount of structural information. There are currently ~100 structures of E2 enzymes in the PDB, over half of which are human E2s (see [11], for a nearly up-to-date list of all E2 PDB entries). Most of the structures are of an individual E2, or more specifically, the catalytic domain of an E2 (referred to as ‘Ubc’). There are five structures of E1–E2 complexes [1216], and approximately a dozen structures of E2s in complex with an E3 [1728]. There are nine structures of E2s in complex with other proteins, including UEVs (Ub enzyme variants) [2932], and non-covalently bound Ub or Ubl proteins, such as SUMO [3337]. Until recently, structures of the activated form of E2, the E2~Ub conjugate, have been missing [26,3841]. As discussed below, the growing structural information regarding this mechanistically crucial species is providing new insights into the mechanisms by which E2–E3 pairs transfer Ub. Overall, the wealth of available structural information reveals that E2 Ubc domain structures from yeast to human are remarkably similar. Although these similarities have provided many general insights, a wide array of functional diversity within this enzyme family demands a closer inspection of the distinctive features of E2s and how they interact with other proteins.


E2 enzymes are easily recognized from their evolutionarily conserved ~150-residue catalytic Ubc domains. Ubcs have a compact structure, shaped essentially like a prolate ellipsoid, comprising four α-helices, a short 310 helix near the active site and a four-stranded anti-parallel β-sheet. On the ‘bottom’ surface (in the view shown in Figure 2), the active-site cysteine residue to which Ub (or Ubl) becomes conjugated is nestled in a catalytic groove surrounded by highly conserved amino acids that mediate both thioester formation (reception of Ub from the E1 Ub-activating enzyme) and substrate ubiquitination (isopeptide bond formation with a substrate lysine residue). Most notable is a trio of residues known as the ‘HPN’ motif, usually found 10 residues to the N-terminal side of the active-site cysteine residue. The histidine residue is thought to play a structural role in forming the E2 active site, whereas the asparagine residue is important for mediating the catalysis of an isopeptide bond between Ub and a substrate lysine residue [42]. As the general mechanism for ubiquitination suggests, an E2 must interact with both an E1 and an E3; both interactions use partly overlapping surfaces on one pole of the Ubc ellipsoid (Figure 2). The E1/E3 interaction site comprises residues on the N-terminal helix, or helix 1 (H1), and loops 4 and 7 (L4 and L7; sometimes referred to as L1 and L2). The shared surface implies that an E2 must be free of the E3 before becoming re-loaded with Ub [43].

Figure 2 Structure of a Ubc domain and its interaction surfaces

The centre panel shows the E2 Ubc13 in ribbon structure (PDB code 2GMI), with the interaction surfaces described in the text coloured. The E2 in each interacting pair is shown in green. Clockwise from the lower-left corner: the E1/E3-binding surface as seen in complex of Ubc13 with the RING E3 TRAF6 (blue) (PDB code 3HCT) and in Ubc12 in complex with the NEDD8-activating E1 (yellow) (PDB code 2NVU); the backside binding surface as seen in the E2-Ub complex UbcH5c-Ub (red) (PDB code 2FUH); the substrate-binding surface as seen in the SUMO E2 Ubc9 in complex with its substrate RanGAP1 (purple) (PDB code 1Z5S); the activated Ub/Ubl surface as seen in the UbcH5b~Ub complex (PDB code 3A33). The inset at the bottom of the Figure shows the highly conserved HPN motif at the active site, with the active-site cysteine residue shown in yellow stick representation.

Many E2s are characterized by additional segments located within and flanking the Ubc domain. For example, Cdc34 (cell division cycle 34) has a 12-residue acidic loop near the 310-helix of its Ubc that is important for catalysing polyUb chain formation [44,45]. More common are E2s that contain variable N- or C-terminal extensions appended to the Ubc domain. Thus historically, the E2 family has been divided into four classes according to the presence of these additional protein segments: class I, Ubc domain only; class II, Ubc domain plus a C-terminal extension; class III, Ubc plus an N-terminal extension; and class IV, Ubc plus both N- and C-terminal extensions. However, although the topology of the Ubc domain is highly conserved, the N- and C-terminal extensions are widely variable in both size and structure. Sizes of extensions vary from a ~50-residue C-terminal extension in the class III E2 Ube2k to a greater than 4000-residue N-terminal extension in the class IV E2 Birc6 [baculoviral IAP (inhibitor of apoptosis) repeat-containing 6] (see [10] for a review of extension-containing E2s). Although a tidy way to group E2s, the classifications are not predictive for functionality, such as which E3s a given E2 will interact with, or whether an E2 transfers monoUb or is capable of building a polyUb chain. In a recent effort to classify E2s, a phylogenetic analysis of similarities within the Ubc domains yielded 17 different families of E2s [46]. In this scheme, five of the ~45 human E2s are assigned to the family that includes the UbcH5 isoforms, whereas other families have only one or two members. Assigning functional significance to each family may have limited utility when there are half as many families as E2s, and additional biochemical and functional characterization will be required to lend predictive value to this fine-grained classification.


How does an E2 bind a particular E3? Crystal structures of the human E2 UbcH7 in complex with the HECT-type E6AP [17] and RING-type cCbl [18] E3 ligase have long provided the framework for our basic understanding of E2–E3 interactions (Figure 3). Although mechanistically distinct, HECT- and RING-type ligases bind a similar surface on the E2 Ubc domain, suggesting that regardless of the Ub destination, be it a HECT active-site cysteine residue or a substrate lysine, E3 binding at this region of an E2 may somehow help to optimally position the activated E2 for transfer of its Ub.

Figure 3 RING- and HECT-type E3s recognize the same surface on the E2 UbcH7

The archetypes for canonical E2–E3 interactions. (A) UbcH7 in complex with the HECT E3 E6AP (PDB code 1C4Z). (B) UbcH7 in complex with the RING E3 cCbl (PDB code 1FBV).

A growing number of E2–E3 structures are available for comparison. Structures that contain the same E2 bound to different E3s allow for both generalizations and distinctions to be drawn. In general, the E2 residues observed to contact E3s include polar and charged residues on H1, a highly conserved hydrophobic residue in L4, usually a phenylalanine (Phe63 in UbcH7; Phe62 in UbcH5c), and hydrophobic residues in L7 (Pro97 and Ala98 in UbcH7). The L4 phenylalanine residue is required for E2s to interact with HECT-type ligases [47], whereas the analogous residue appears unimportant in interactions involving isoforms of UbcH5 and the RINGs of BRCA1 (breast cancer early-onset 1) [2] or CNOT4 [CCR (chemokine receptor) 4–NOT transcription complex] [19]. L7 residues contribute to interactions with both HECT- and RING-type ligases. In the case of BRCA1 and its cadre of ten interacting E2s, the L7 alanine residue (Ala96 in UbcH5c) is a defining feature of the set of interacting E2s, as its mutation to aspartate disrupts the ability of E2s to continue to make productive interactions [2].

Comparison of structures of a single E3 in complex with different E2s reveals plasticity in the E2–E3 interface. Structures of the homodimeric U-box ligase CHIP [C-terminus of Hsc70 (heat-shock complex 70)-interacting protein], in complex with Ubc13 [21] or UbcH5a [23] illustrate how an E3 can accommodate different E2s. The CHIP-interacting surface of UbcH5a includes residues in the N-terminal helix and hydrophobic residues in L4 (Pro61 and Phe62) and in L7 (Pro95 and Ala96), with Phe62 protruding into a hydrophobic groove on CHIP [23]. In Ubc13, the structurally homologous residue to Phe62 is Met64. Whereas mutation of UbcH5a residue Phe62 disrupts CHIP–UbcH5a interactions, replacement of Met64 in Ubc13 does not disrupt the complex [23]. In contrast, the mutation M64A disrupts the interaction between Ubc13 and the RING ligase TRAF6 [tumour-necrosis-factor-receptor-associated factor 6] in a Y2H (yeast two-hybrid) system [48], as well as its interaction with Rad5, another RING-type ligase in yeast [49]. There are salt bridge interactions between Ubc13 and CHIP not observed in the CHIP–UbcH5a complex and these may account for the lowered importance of the L4 hydrophobic residue [21,23]. A salt bridge interaction plays a key role in the complex of the CNOT4 RING domain and UbcH5 loop L4 residue Lys63 [19,50]. Although yet to be exploited, the ability to manipulate charge–charge interactions to yield altered specificity pairs could be a powerful approach for in vivo studies [51,52]. In summary, the E2–E3 structures solved to date identify a modest number of E2 residues that are involved in binding a variety of E3s. It is important to note, however, that there are currently E2–E3 structures for only a small number of Ub E2s, namely isoforms of UbcH5, UbcH7 and Ubc13. As these E2s are quite similar in their E3-binding loops, it may be ill-advised to extrapolate lessons learned from them to other more distantly related E2s.


There are at least two compelling reasons for studying E2–E3 interactions. First, knowledge of active E2–E3 pairs can guide and inform biological studies. A given E3 may interact with numerous E2s, and the ultimate outcome of the reaction (i.e. the type of product) depends on which E2 is involved [24]. Secondly, structural information on E2–E3 complexes may provide insight into the mechanism and products of Ub transfer reactions. Therefore knowledge of the complete cadre of E2s for a given E3 is critical.

The question of how to best identify active E2–E3 pairs remains a thorny one, with different pathways (i.e. Ub, SUMO etc.) posing different challenges. In the case of the Ubls, including SUMO, the E3s are not as easily recognized as the Ub E3s, which all contain either a RING/U-box or a HECT domain that can be identified through sequence conservation [53]. For Ub transfer, a genome will contain dozens of E2s and hundreds of E3s. Given the relative ease with which E2s and E3s can be recognized directly from protein sequence information and the numerous structural studies of E2–E3 complexes, the goal of defining the E2–E3 interactome might seem straightforward. However, sequence analysis alone has had limited power for predicting uncharacterized E2–E3 pairs. Part of the problem arises from the functional necessity that E2s use overlapping surfaces for binding and recognition of both E1s and E3s, requiring that these surfaces have broadly conserved features (see Figure 2) [43].

The modest affinity and transient nature of E2–E3 complexes pose additional technical challenges for the identification of E2–E3 pairs. Standard techniques such as co-purification, pull-down or co-immunoprecipitation rarely succeed due to the weak affinity of the complexes. Expression analysis in which substrate degradation is monitored when certain E2s or E3s are overexpressed or depleted from cell extracts can be successful in identifying E2–E3 pairs [54,55]. Depletion of the APC (anaphase-promoting complex)-specific E2 Ube2c (UbcH10) from cellular extracts did not completely stabilize APC substrates such as securin, leading to the discovery of another APC-interacting E2 Ube2s [55]. Such approaches may suffer from the pleiotropic or compensatory nature of some E2s/E3s, and their success requires a targeted approach that may not be feasible for large scale de novo determination of interacting E2–E3 pairs.

Directed Y2H approaches have been somewhat more successful in identifying E2–E3 pairs, presumably because a positive read-out can be obtained even for transient interactions. During 2009, two large-scale Y2H screens for E2 partners were published [56,57]. A screen that utilized full-length E2s as bait against approx. 150 RING protein preys yielded putative partners for all but two of 39 E2s (Cdc34 and Birc6) and for approx. 90 of the E3 preys [56]. However, a screen of over 250 RING domain preys with 36 E2 Ubc domain baits failed to identify a binding partner for ten E2s that are known to conjugate Ub and fewer than half the RING domains returned a positive E2 interaction [57]. The different outcomes in the screens may be due in part to the use of full-length or specific E2 and E3 domains, consistent with emerging evidence for non-canonical E2–E3 interactions. There is growing recognition that some RING E3s, such as Rad18 and gp78, utilize regions outside the canonical RING domain for binding to Ubc domains [25,27,58]. In addition, the requirement of some RING E3s to exist as heterodimeric or multi-component complexes may further affect the attainable yield in a Y2H screen. For example, in a targeted Y2H screen aimed at identifying the human E2s that interact with the heterodimeric RING E3, BRCA1–BARD1 (BRCA1-associated RING domain 1), a bait construct in which the RING domains of BRCA1 and BARD1 were fused into a single polypeptide that folds correctly into the E3-active structure identified ten E2s that interact with BRCA1–BARD1 [2]. Screens using baits composed solely of the BRCA1 RING failed to identify any E2s that have been shown biochemically to transfer Ub [2,56,57]. It is reasonable to expect similar behaviour from other heterodimeric E3s, implying that the use of single RING domains as bait is likely to fail for this class of E3. Furthermore, in the same screens, using the individual RINGs from the heterodimeric E3 RING1–Bmi1, (B-cell-specific Moloney murine leukaemia virus integration site 1), only a small subset of the known interacting E2s could be identified [56,57]. Although it is tempting to predict that the RING constructs for which no interacting E2s were identified in the screens reported to date are either heterodimeric or multi-component E3s, many things can conspire to give a negative Y2H result, so caution should be exercised. Within a Y2H screen, the E2 may or may not be conjugated to Ub, depending on whether the endogenous yeast E1 is able to charge the E2 of interest, among other factors. Therefore an issue that may contribute to a failure to identify E2s for an E3 in a Y2H experiment is that Y2H studies may only screen for interactions between an E3 and a free E2, although the functionally relevant interaction involves the E2~Ub conjugate. There are examples of E3s that show detectable binding only to an E2~Ub. For example, SspH2 (a bacterial protein with E3 ligase activity) binds only to an activated E2~Ub conjugate and not to the individual components [59]. Attempts to identify this interaction using Y2H screens with E2s as prey uniformly failed (P. Brzovic and R. Klevit, unpublished work). It remains to be seen whether this feature will be unique to bacterial E3s that have evolved via convergent evolution to work with host E2 enzymes or whether there are eukaryotic E3s that only bind to E2~Ub species. In any case, these examples demonstrate the complicated and context-dependent nature of E2–E3 interactions that confound the ability to identify them.


An emerging interaction surface used in E2s is the so-called ‘backside’ (the top in the view shown in Figure 2), a surface primarily comprising the E2 β-sheet. First identified as a site for non-covalent Ub and SUMO binding with the E2s UbcH5c [33] and Ubc9 respectively [35,37], the list of E2s that use the backside for mediating protein–protein interactions continues to grow and now includes Ube2g2 [25,27,60] (Figure 4).

Figure 4 Ub/Ubl and non-canonical E3 binding to the E2 backside

Upper panel: non-covalent complexes of E2s with the indicated Ub/Ubl (red) bound on the backside. Lower panel: non-canonical E3 (blue) binding uses the same E2 surface. PDB codes are (left to right from the top) 2EKE, 2UYZ, 2FUH, 3A4S, 3H8K and 1Z5S.

There are several examples of E3s that utilize the E2 backside surface in addition to the canonical E3-binding surface (H1, L4 and L7). Some E3s contain domains that are structural mimics for Ubls, as seen with the human RENi {Rad60/Esc2/NIP45 [NFAT (nuclear factor of activated T-cells)-interacting protein of 45 kDa]} protein NIP45 and its yeast counterpart Rad60 (Figure 4) [61,62]. NIP45 uses SLDs (SUMO-like domains) to bind the backside of Ubc9, with a similar affinity to SUMO, using conserved electrostatic interactions [62]. The biochemical function of NIP45 remains to be determined; current suggestions are that NIP45 may function as a SUMO E3 ligase or as a target for a SUMO-targeted Ub E3 ligase [62,63]. In another example, the SUMO E3 Nup358/RanBP2 (Ran-binding protein 2) contacts the Ubc9 backside using a domain that recapitulates the contacts of SUMO in a structurally unique way (Figure 4) [20]. In addition to the canonical Ubc9 E3-binding interface comprising residues in H1 and L4, the E3 RanBP2 makes additional contacts with the first three β-strands on the backside of Ubc9, the same surface that binds SUMO. Notably, in contrast with Ub E2s, residues of Ubc9 L7 do not participate in this E2–E3 interaction.

Similar to RanBP2, the Ub E3 ligase gp78 uses a distinct structural domain to interact with the backside of the E2 Ube2g2 [25,27]. In addition to binding Ube2g2 via a RING domain, the E3 gp78 interacts with the backside of Ube2g2 through a short region called the G2BR (Ube2g2-binding region) [25,27]. However, unlike the SUMO E3 that uses a contiguous surface to contact the Ubc9 canonical E3-binding surface (H1/L4/L7) and the backside [20], the gp78 G2BR is located almost 200 residues away from the RING domain [64]. Unstructured in its apo form [25], G2BR adopts an α-helical structure that contacts the Ube2g2 backside surface through mostly hydrophobic interactions bolstered by additional charged contacts [25,27]. Binding of G2BR, even in trans, results in a ~50-fold increase in affinity for the RING domain of gp78 [25]. Functionally, Ube2g2 shows slower rates of thioester formation with Ub in the presence of G2BR [25], as well as an enhanced ability to build chains [25,27]. Although a structure for gp78's RING bound to Ube2g2 does not yet exist, it is reasonable to expect that it will show primarily canonical E2–E3 interactions. Thus the mode of binding for this Ub E2–E3 pair bears a striking resemblance to the mode used by the SUMO E2–E3, using a mix of the canonical E3-binding surface and the backside of the E2. Although no structural information yet exists, a small auxiliary E2-binding module that is 300 residues away from the RING of the E3 Rad18 was identified from Y2H analysis with the yeast E2 Rad6 [58]. It seems reasonable to expect that additional variations on this theme will be discovered for other E2–E3 pairs in the future. In summary, similar to the shared E1/E3-binding surface, E2s appear to use their backside surfaces to provide a multi-functional binding surface for E3s and Ub/Ubls and, possibly, other binding partners.

Interestingly, Ube2g2 binds Ub on its backside surface, although with much weaker affinity than for G2BR [25]. A functional significance, if any, for this non-covalent Ub interaction has yet to be established for Ube2g2. This begs the question, could the binding of Ub via the backside, reported for other E2s, be a harbinger for non-canonical E3 binding or is the E2 using overlapping surfaces to manage multiple functionalities? Given the overlap in the E1/E3-binding surfaces on the E2, the E2 backside may also serve multiple functions binding both Ub and an E3. As functionally significant Ub binding occurs on the backside surfaces of several E2-variant proteins, including the UEVs Mms2, Tsg101 and Vps23 [31,32,34], it remains a possibility that E2 Ub binding on the backside surface may be biologically and mechanistically important.


Available structures of most class II, III and IV E2s are for the Ubc portion only. In a few cases, extensions were included in the crystallization (e.g. Ube2T, PDB code 1YH2), but their structure could not be defined, indicating that they are disordered, at least in the absence of a binding partner [65]. With the exception of yeast Ubc1 and its human counterpart Ube2k (also known as E2–25K), there is very little structural information regarding the non-Ubc portions of E2s. Ubc1/Ube2k are unique in that their C-terminal extensions comprise an identifiable structurally well-characterized domain, the UBA (Ub-associated domain). UBAs appear in a wide variety of proteins and are known to function as Ub-binding domains. Paradoxically, the function of the UBA of Ube2k (and yeast Ubc1), the only non-Ubc domain for which structural information is available, remains enigmatic [66,67]. Although known to bind Ub and polyUb chains, the UBA domain is dispensable for the generation of polyUb chains [3].

Current knowledge points to a role for non-Ubc extensions of E2s in mediating interactions with E3s or substrates. Y2H analysis of the interaction between the yeast E2 Ubc2 and the E3 ligase Ubr1 identified acidic residues in the C-terminal extension of the E2 that interact with a basic region in the E3 [68]. Removal of the extension resulted in some loss of Y2H signal when presenting the RING H2 region alone. The 37-residue C-terminal extension of Cdc34, which contains a high density of acidic residues, helps mediate the interaction with the multi-protein SCF (Skp1/cullin/F-box) E3 complex through a ‘basic canyon’ on a surface located in the C-terminal domain of the Cul1 (cullin 1) subunit [69]. Replacement of basic residues flanking the canyon with acidic residues disrupts ubiquitination mediated by Cdc34, but not by other E2s such as UbcH5c that lack acidic tails. Removal of the Cdc34 acidic tail decreases its activity with the SCF, as does addition of the tail in trans [70]. The interaction of the E3 with the Cdc34 Ubc domain and with its tail are each quite weak, but when presented within a single E2 moiety, the avidity effect results in an orders-of-magnitude increase in Km relative to the Ubc domain alone. In addition to binding E3s, the C-terminal tail of Cdc34 has recently been shown to contain two non-covalent Ub-binding sites composed primarily of two islands of hydrophobic residues in the otherwise acidic tail [71]. Mutations in one of these Ub-binding sites result in defects in SCF-mediated substrate ubiquitination. How Cdc34 co-ordinates Ub binding and E3 binding via the same 37-residue stretch remains to be established. Combined, these observations underscore an emerging theme of dual functionality for E2 regions exhibiting non-covalent Ub-binding.

The unique 30-residue N-terminal extension of Ube2c (UbcH10) is thought to provide this E2 with specificity towards substrates [72]. Deletion of the N-terminus of Ube2c makes the enzyme more active, but less specific for particular lysine residues and degradation signals when transferring Ub with the APC E3 ligase. In contrast with the Cdc34 tail, which provides additional contacts to the E3, the Ube2c N-terminus probably mediates an interaction with the substrate, and is not the primary E3-binding component. Confirmation of this proposal must await further structural characterization of Ube2c in complex with APC and/or a substrate. The N-terminal extensions of members of the Ube2e family (also known as UbcH6, Ube2e2 and UbcM2) are implicated in novel interactions with cullin adapter complexes based on the observation that mutations in the canonical E3-binding region of the E2s do not disrupt the interaction [73]. The functional consequences of these interactions have yet to be resolved. In summary, non-Ubc regions of E2s appear to play a variety of roles, not all of which are yet defined, through interactions with E3s, substrates, adaptors and Ub itself.


The conjugated E2~Ub is the active form of an E2 enzyme. The in vivo steady-state equilibrium of the human E2 Ubc2b favours the conjugated form and this is likely a general feature of the protein ubiquitination system [74]. Thus cellular proteins are more likely to encounter E2~Ubs than free E2s. However, owing to the relative ease with which free E2s can be studied structurally and the technical challenges posed by generating sufficient quantities of purified E2~Ub species, a vast majority of current structural information on E2s are for free E2s. The recent emergence of several E2~Ub structures provides an excellent foundation to begin to understand this functionally important species.

E2~Ub conjugates present a technical challenge for structure determination because the thioester formed between E2 and Ub is inherently unstable, a chemical property that is necessary for catalysis. Manipulation of the unstable thioester bond has enabled structural studies of several E2~Ubs. For many (but not all) E2s, mutation of the active-site cysteine residue to a serine residue allows formation of a longer-lived oxyester, which is more amenable to structural determination [26,39,41,59,75]. The cysteine-to-serine mutation differs by only a single atom in the resulting ester. A disulfide-linked mimic of a UbcH8~Ub (Ube2L6) species has been generated by linking Ub with a cysteine residue added to its C-terminus to the E2 active-site cysteine residue [40]. At present, there are five structures of E2~Ub-like species: Ubc13~Ub (PDB code 2GMI); Ubc1~Ub (PDB code 1FXT); UbcH8~Ub (PDB code 2KJH); and UbcH5b~Ub (PDB codes 3A33 and 3JVZ). Structural information from NMR studies of Ube2b (Rad6b)~Ub is also available, but without a PDB entry [75]. These structures provide the first glimpse of E2~Ubs and the possibility to better understand the mechanism of Ub transfer. Another informative structure representing an E2–E3–product complex in the SUMO system (Ubc9–Nup358–SUMO–RanGAP1, where GAP is GTPase-activating protein) adds to the growing picture of how Ub or a Ubl is transferred from an E2 to a substrate lysine residue [20]. As the Ubc9 active-site cysteine is directed at the isopeptide bond between the RanGAP1 substrate and SUMO, certain features of this structure may be pertinent to a Ubc9~SUMO conjugate.

The first general conclusion that can be drawn from comparison of the structures of the conjugated E2s and their free E2 counterparts is that neither the E2 nor Ub undergoes significant conformational change upon conjugation [26,3841,59]. The comparison of free Ubc13 (PDB code 1JAT) with conjugated Ubc13~Ub (PDB code 2GMI) shows only 1.6 Å (1 Å=0.1 nm) rmsd (root mean square deviation) over all Cα positions [39]. Movements of side chains in or near E2 active sites have been noted, but such changes should be interpreted with care, as the positions of side chains in free E2 structures can be just as variable [41,76]. For example, the location of the proposed catalytic residue Asn77 of the HPN motif, differs by nearly 2 Å in four different free UbcH5b structures depending on whether the structure was solved crystallographically or in solution (PDB codes 2C4O, 2ESK, 2CLW and 1W4U) and dynamics of the side chain of Asn77 in solution are apparent in the NMR ensemble (PDB code 1W4U) [77]. In some cases the resolution of the crystal structure is on par with or lower than the 2 Å changes noted above and therefore lacks sufficient precision to draw conclusions about the position and role of Asn77 (PDB codes 3A33, 2C4O and 2CLW). Furthermore, comparison of the conformation of Asn77 in a variety of E2 structures does not provide a unified theme: Asn77 is in the gauche conformation, pointing away from the active-site cysteine residue, in free Ube2g2 (PDB code 2CYX), free Ubc2 (yeast Rad6) (PDB code 1AYZ) and in Ubc13~Ub (PDB code 2GMI), and is in the trans conformation, pointing towards the cysteine residue and presumably in a better conformation to stabilize the oxyanion intermediate in the solution-state structure of Ube2g2 [78].

Given that neither component undergoes conformational change upon conjugation, it is instructive to compare the available E2~Ub structures as a group. Figure 5 shows five structures with the E2 components superimposed. A striking feature of this display is that Ub/Ubls adopt many positions relative to the E2. It has been proposed that differences in the positioning of Ub in an E2~Ub conjugate may explain the variety of activity observed for the various E2 family members [76], although this remains to be tested experimentally. Observed Ub positions may be caused in part by either crystal-packing effects or additional protein–protein interactions in a given crystal. For example, in the crystal of Ubc13~Ub–Mms2, the UEV protein Mms2 binds the hydrophobic patch on the surface of Ub, centred about Leu8, Ile44 and Val70 of a Ub from another Ubc13–Ub moiety, and positions its Lys63 side chain towards the thioester bond of Ubc13~Ub [39]. This non-covalent interaction involving the ~Ub may contribute to its ultimate position in the crystal lattice. Similarly, in the structure of UbcH5b~Ub, a Ub moiety from one E2~Ub is bound non-covalently on the backside site of the E2 of another E2~Ub moiety [41]. Thus these ‘secondary’ interactions may dictate the ~Ub positions observed in each crystal. In two cases, Ubc1~Ub and UbcH8~Ub, the position of the conjugated Ub is not influenced by a third protein [38,40]. In both cases, a single E2~Ub conformation is reported, despite evidence that the Ub takes many positions in solution in the E2~Ub conjugates [76].

Figure 5 Ub conjugated to an E2 assumes multiple orientations relative to the E2

The E2 Ubc of each structure has been superimposed (green), with the active-site cysteine residue shown as yellow spheres. The Ub moiety in each structure is shown in a different colour: red, Ubc1~Ub (PDB code 1FXT); blue, UbcH5b~Ub in complex with Nedd4L (PDB code 3JVZ); yellow, Ubc13~Ub (PDB code 2GMI); orange, UbcH5b~Ub (PDB code 3A33); and purple, UbcH8~Ub (PDB code 2KJH). The view on the left shows the same E2 orientation as in Figures 2, 3, and 4; the view on the right is rotated 45 degrees about the vertical axis. A three-dimensional interactive version of this Figure is available at

The array of Ub positions observed among the available structures may be taken as an indication that E2~Ub species are dynamic in solution and that each structure represents a snapshot along some trajectory or within an ensemble. NMR studies indicate that conjugated Ub is indeed flexible relative to the E2 and that the two proteins behave as two loosely connected entities rather than one globular protein complex [59,75]. Further investigations of E2~Ub species in solution are needed to more fully understand the nature and relevance of the observed and inferred dynamics.

The Ubc9–Nup358–SUMO–RanGAP1 structure is presumably a snapshot of an E2–E3–product complex [20]. Here, the E3 contacts the β-sheet surface of SUMO that is now attached to the product. The interaction between the ligase and SUMO may provide a hint into the role an E3 plays in catalysing Ub/Ubl transfer. The SUMO is in a unique position relative to the E2 in the admittedly small number of E2~Ub structures currently available, and may represent the orientation of an E2 and Ub just after formation of the isopeptide bond.

Ub itself binds to a remarkable array of proteins, doing so mainly via a hydrophobic surface centred on residue Ile44. Thus when an E2 is conjugated with Ub, there is the potential to utilize this domain as a site for protein–protein interactions. The ability to distinguish between free E2s and E2~Ub conjugates would seem to offer a functional advantage. Recently several E3s have been shown to possess domains that bind Ubls [26,59,7981], in some cases using this site to specifically recruit E2~Ubls [26,59], thereby enhancing productive associations.

Most E3s appear to bind both free and conjugated forms of E2s. Although there is anecdotal evidence in support of the general expectation that binding is stronger with an E2~Ub, this property has only been quantitatively demonstrated in a few instances. A disulfide-linked mimic of activated UbcH7~Ub showed no increased affinity for the HECT E3 E6AP compared with the affinity measured for free E2 [82]. In contrast, a kinetic analysis of the yeast ligase E3α and the E2 Ubc2 revealed that binding to Ubc2~Ub is ten times stronger than to free Ubc2 [74]. Similarly, kinetic analysis showed that Cdc34~Ub binds Cul1–Rbx1 (RING-box 1) about twice as tightly as free Cdc34 [83]. These relatively modest differences in binding energy make the measurement technically challenging; kinetic measurements in which Ki is compared for a free E2 and an oxyester form of E2~Ub are a better choice than thermodynamically based approaches.

Although most E3s appear to bind free E2 with a low, but detectable, affinity it is the E2~Ub–E3 complex that represents an enzyme poised to react. The first structure of an E2~Ub–E3 complex was recently reported for the HECT E3 NEDD4L (neuralprecursor-cell-expressed, developmentally down-regulated 4-like) and UbcH5b~Ub (Figure 6) [26]. The ability to capture this important species in a crystal was made possible by inclusion of several mutations in the E2: an active site cysteine-to-serine mutation allowed formation of the oxyester with Ub and mutations were made in the E3-binding region to create a stronger-binding complex amenable to structural investigation [26,84]. The structure is characterized by a combination of canonical and novel binding surfaces on the E2 and on Ub. The N-lobe (N-terminal lobe) of the HECT domain binds the canonical E3-binding site on UbcH5, much as previous structures have revealed. However, the HECT domain C-lobe (C-terminal lobe), which contains the E3 active-site Cys922, is in a dramatically different position relative to previous structures [17,26]. Contact between the C-lobe and ~Ub moiety closes the distance between the E2 and E3 active sites to about 8 Å, much closer than distances observed in any previous structures of HECT E3s in complex with an E2 [17,26]. NEDD4L C-lobe residues Leu916 and Met943 make hydrophobic contacts with Ub residues Ile36, Leu71 and Leu73. Mutations in the E3 or Ub at this interface decreased the ability of the E2 to transfer Ub to the HECT active-site cysteine residue. On the basis of this structure, it is reasonable to propose that HECT E3s can be described in general as consisting of an E2-binding domain (the N-lobe) and a ~Ub-binding domain (the C-lobe), even though an interaction between the C-lobe and free Ub has never been detected [80]. However, with only a single E2~Ub–E3 structure currently available, it is wise to avoid the temptation to draw many generalizations. There are indications that various active E2~Ub–E3 complexes interact in different ways: Ub mutations I36D and L71A decrease formation of a HECT~Ub in NEDD4L and other family members, such as yeast Rsp5 and ITCH (itchy E3 ubiquitin protein ligase), but the HECT E3 E6AP is not impaired by the same mutations [26]. We can therefore expect that even among HECT E3s, the details of how they recognize and bind to E2~Ubs will be distinct.

Figure 6 Structure of a E2~Ub–E3 complex

The HECT E3 Nedd4L (blue) is shown in complex with UbcH5b~Ub (green/red). Left-hand panel: the HECT N-lobe contacts the E1/E3 surface on UbcH5, as observed in previous structures (see Figure 3) and the HECT C-lobe makes contacts with the ~Ub moiety. The NEDD4L active site is shown as yellow spheres (PDB code 3JVZ). Right-hand panel: the view is rotated by 180 ° to show NEDD4L residues analogous to those observed in Rsp5 and SMURF2 (orange spheres) that are important for non-covalent Ub binding in the N-lobe.

In contrast with eukaryotic E3s, the bacterial HECT-type E3 SspH2 binds only to E2~Ub, with no detectable binding to the individual components [59]. The E3-interacting surface mapped by NMR studies only modestly overlaps with the canonical binding surface of eukaryotic E3s, as binding on H1 and L4 of the E2 were not detected. Instead, the interaction involves a surface on UbcH5c not observed previously in E2–E3 complexes, with the SspH2-binding site comprising residues in regions around the active site, the cross-over helix and L7. In addition, the E3 makes contacts to the hydrophobic Ile44 surface of the Ub moiety in the E2~Ub. Mutation of the L4 phenylalanine residue of UbcH5c, known to be essential for HECT E3 binding [47], does not disrupt the interaction with SspH2 [59], consistent with the non-overlapping nature of the SspH2-binding site relative to other HECT E3s. Thus this effector protein from a pathogenic bacterium has evolved by convergent evolution to specifically recognize the active form of a critical host enzyme, i.e. an E2~Ub, and accomplishes the task using a surface on the E2 different from host E3s. Whether this novel binding surface is used by other bacterial effectors that hijack host E2s is an open question.

The SUMO E3, Pc2 (proprotein convertase 2), was recently shown to contain a cryptic SIM (SUMO-interacting motif) that is distal to its Ubc9-binding domain [79]. Point mutants in the Pc2 SIM fail to recruit Ubc9 to subnuclear foci, and a Ubc9 active-site mutant incapable of conjugating SUMO also fails to be localized by Pc2. These observations are consistent with the E3 preferentially binding Ubc9~SUMO. Interestingly, Ubc9 mutants defective for non-covalent backside SUMO-binding also fail to be relocalized by Pc2, suggesting Pc2 may bind both activated and non-covalently-bound SUMO on the E2. Whether these interactions have mechanistic consequences in addition to their demonstrated subcellular localization functions remains to be explored.

There is a growing list of proteins that do not contain canonical RING, U-box or HECT domains that exhibit preferential binding of E2~Ub over free E2. The nuclear import protein importin-11 interacts with the conjugated UbcM2~Ub (Ube2e3), but not with the free E2 [85]. Thus preferential binding of conjugated E2~Ub results in a specific subcellular localization of an E2. UbcM2~Ub is restricted to the nucleus, whereas free UbcM2 cannot enter the nucleus, a mechanism that may spatially regulate the activity of UbcM2. Likewise, the kinase OspG from Shigella bacteria exclusively binds human UbcH5b~Ub, but not free UbcH5b or Ub alone [86]. OspG binding to UbcH5b~Ub presumably disrupts Ub transfer activity in the host as observed by decreased Ub-dependent degradation of IκBα (inhibitory κB α).

In summary, the observed range of Ub positions relative to the E2 in E2~Ub complexes may play a variety of roles. Certain orientations may help to recruit specific binding proteins. Alternatively, the dynamics of conjugated Ub itself may play a role in catalysis by sampling numerous states which could alter the environment of the active site and modulate reactivity. Additional structural information regarding E2~Ubs alone and in complex with E3s will help to fill in the blanks in the not-too-distant future.


Several HECT E3s contain an additional Ub-binding site in their N-lobe, distinct from the E2-binding region [80,81]. Ub binding to the N-lobe of the HECT E3 Rsp5 was detected by GST (glutathione transferase) pull-downs and scanning alanine mutagenesis [80]. A similar site has been identified in HECTs NEDD4 [80] and SMURF2 (SMAD-specific E3 ubiquitin protein ligase 2) [81], but not in Tom1 (target of Myb1) [80], again suggesting that not all HECT E3s recognize the components of ubiquitination in the same fashion. The canonical hydrophobic Ile44 Ub surface is responsible for the HECT N-lobe–Ub non-covalent interaction, which is different from the surface used by the C-lobe of NEDD4L (Figure 6) [81]. The functional significance of HECT N-lobe Ub-binding has yet to be settled unambiguously. Rsp5 mutants with decreased affinity for Ub appear to synthesize Ub chains more efficiently than the wild-type enzyme [80]. In contrast, SMURF2 mutants deficient for Ub binding have disrupted polyUb chain synthesis [81]. SMURF2 can bind its substrate more efficiently when the substrate is mono-ubiquitinated, and mutations in the E3 Ub-binding site preclude this interaction [80]. In neither case was a role for the N-terminal Ub-binding domain in recruiting E2~Ub demonstrated [80,81]. As the interactions required to recruit a ubiquitinated substrate and a Ub-laden E2 may be quite similar, E2~Ub recruitment remains a formal possibility.


Considering the wealth of structural and biochemical information regarding E2s, it is remarkable how much remains unclear about the mechanism by which Ub is transferred from E2~Ub. Fundamental questions such as ‘How does an E3 catalyse transfer?’ are not yet answered. Studies with the SUMO E2 Ubc9 show that the ‘natural’ hydrolysis (in absence of any other nucleophile or E3) is slow, suggesting that thioesters in native E2s are relatively stable and must be coaxed into discharging their Ub on to the substrate [87]. Hypotheses invoking allosteric effects on the E2 are pervasive but neither well-documented nor supported. Studies utilizing E2~Ub conjugates (or their mimics) are likely to finally shed light on Ub transfer mechanisms.

To understand the mechanism of Ub transfer, it is sensible to focus on the E2 active site and its surroundings. The current paradigm for an E2 active site includes the cysteine residue to which Ub/Ubl will be conjugated and the conserved HPN motif (see above). However, these residues have been insufficient to completely explain the mechanism, suggesting that we must look further afield for clues. In the following section, we suggest some additional features that may play direct roles in E2 function.

Dynamic behaviour of E2s may play a role in catalysis and mechanism. Some E2s have flexible loops near their active sites: for example, there are two flexible loops (residues 95–107 and 130–135) surrounding the active site of the E2 Ube2g2, identified by comparison of solution and crystal structures [78]. Similar loops are present in yeast Ubc7, human and yeast Cdc34, but not in the large UbcH5 family. Two acidic residues in the Ube2g2 loops (Asp98 and Asp99) are required for activity [78]. Current hypotheses regarding the role of the acidic loop residues include: (i) that they polarize the histidine residue of the conserved HPN motif to create a general base [78]; or (ii) that they guide a lysine residue into the active site [78]. Either case requires a fixed orientation of the acidic residues in question and therefore predicts that the flexible loops become more ordered either as a result of formation of the E2~Ub conjugate or as a result of E3 binding. Further investigations will be required to sort out these possibilities.

The position of conjugated Ub within an E2~Ub conjugate likely plays a role in catalysis, perhaps by influencing the approach of the nucleophilic substrate lysine residue. Early studies demonstrated specificity for Ub transfer to lysine residues over other amines, indicating that the E2~Ub protects its thioester from other reactive species [88]. The reactivity of the E2~Ub thioester is significantly lower than that of a free thioester in solution, which may be due to limited access to the thioester within the active site [88]. In the structure of the oxyester mimic of Ubc13~Ub−Mms2, the substrate Ub's Lys63 is ‘poised’ to attack the ester and the direction of nucleophilic attack is defined by the geometry of the E2~Ub–UEV complex [39]. The ability to form such a structure may explain the ability of Ubc13–Mms2 to build free Lys63-linked polyUb chains in the absence of an E3. Intriguingly, in the structure of the E2–E3–product mimic Nup38–Ubc9–SUMO–RanGAP1 (PDB code 1Z5S) the now reacted lysine residue is observed at nearly the same angle as the about-to-be-reacted Lys63 discussed above (Figure 7) [20]. Whether the direction for the substrate lysine residue approach for other E2s will be the same as the two examples currently available cannot be predicted, but it is tempting to speculate that the orientation observed represents one of perhaps several that allow for optimal reactivity. Access of substrate to the optimal approach path might also be modulated by E2~Ub dynamics in which the preferred route is blocked much of the time. In such a scenario, E3 binding could either limit the available E2~Ub conformations and/or stabilize the most reactive conformer, thereby effecting catalysis.

Figure 7 Position of substrate lysine approach into the E2 active site

Lys63 of a putative substrate Ub in the formation of a Lys63-linked Ub chain, as observed in a structure of Ubc13~Ub–Mms2 (PDB code 2GMI), is shown as a red stick representation. The substrate lysine residue involved in isopeptide linkage to the Ubl (SUMO), as observed in a E2–E3–product structure (Ubc9–Nup358–SUMO–RanGAP1) (PDB code 1Z5S), is shown as a purple stick representation. The E2s of each structure were superimposed for this Figure and are shown in green, with the active-site cysteine residues shown as yellow spheres (the conjugated Ub in Ubc13~Ub is shown in cyan).

Another factor that may control thioester reactivity is the ability of the E2~Ub active site to favourably position the atoms of the tetrahedral intermediate, formation of which is likely the rate-limiting step of the transfer reaction. Computational calculations of transition-state models of thioesters indicate that electron delocalization and the dihedral angles of the tetrahedral oxyanion intermediate contribute significantly to activation energies and hence thioester reactivity [89]. These factors have yet to be considered in the context of the E2 active site cradling a thioester, but it is likely that the geometry of the active site plays a role in the reactivity of Ub transfer by influencing the dihedral angles of the tetrahedral oxyanion intermediate. Since every E2~Ub structure is a snapshot of the reactant ester before lysine residue attack, it is virtually impossible to test the implications of steric dihedral angle regulation of the oxyanion intermediate by the E2 active site. Trapping the Ub transfer reaction at the oxyanion intermediate structurally will be challenging, but computational simulations of the intermediate in the context of an E2 active site may well provide insights.


Over the past few years, investigation of the roles E2s play in the Ub transfer pathway have increased appreciation for their importance. We are beginning to understand the functional diversity the E2 family provides, a remarkable feat considering the high similarity within this class of enzyme. In the present review, we have highlighted the economy of E2s in their ability to facilitate many protein–protein interactions at nearly every surface on their compact ellipsoid Ubc domain. In fact, several surfaces are used for more than one protein–protein interaction including the E3/E1-binding site, as well as the backside Ub/E3-binding site. We have focused on non-canonical E2–protein interactions, and predict that new ones will be discovered, possibly involving the few remaining Ubc surfaces yet to be identified as mediating protein binding. Finally, we stressed the importance of studying the conjugated E2~Ub, the active form of the enzyme, to fully inform function.

In the future, several discoveries could quickly advance the understanding of Ub transfer. Mechanistically, the structure of a RING E3–E2~Ub will be highly informative as such a structure may provide insight into the catalytic role of the RING E3, beyond its ability to recruit target substrates to an E2~Ub. As ~95% of E3s in the human genome are RING/U-box-type E3s (including multi-subunit ligases, such as APC and SCF) [53], advancements in our understanding of RING-mediated Ub transfer stands to impact almost every field of biology. Secondly, there is a need to develop approaches that incorporate non-canonical E2–E3 interactions in efforts to comprehensively identify E2–E3 partners. Thirdly, since traditional Y2H screens for E2–E3 pairs cannot parse novel interactions with the conjugated E2~Ub compared with the free E2, new techniques are needed to uncover E2~Ub-specific interactions. The ability to identify binding partners for E2~Ubs (or a reasonable mimic of the conjugate) will likely result in the discovery of different and potentially more relevant E3 partners than have been found by screening with an E2 alone. Fourthly, identification of non-Ub-pathway binding partners for E2s may provide important new breakthroughs in the future. A large-scale Y2H screen in which 39 full-length E2s were used as bait against high complexity prey libraries identified 229 total interactions, only 30% of which involved known E3 ligase domains [56]. Although the ~160 E2–non-E3 interactions remain to be confirmed, it is unlikely that such a large percentage of hits all correspond to false positives, suggesting that there are even more E2-binding partners than we might have been led to expect. The nature of these proteins and of their interactions with E2s is likely to be one of the newly emerging themes of E2 investigation in coming years.

E2s function at the centre of the Ub transfer pathway, and provide much of the diversity the Ub signal transmits in the cell. From E3 selection to the type of Ub transfer (mono, poly and the type of lysine residue linkage), the E2 is at the crux of substrate regulation, and arguably, many disease pathways. Considering this responsibility, it is perhaps not surprising that the human genome contains in excess of 35 E2 genes. Many of these have yet to be characterized even at a rudimentary level. There are sure to be more surprises on the horizon as new insights into the mechanisms of E2-catalysed Ub transfer and a complete description of the cellular roles of E2s emerge.


The work of our laboratory is supported by the National Institute of General Medical Sciences [grant numbers 5R01 GM088055 (to R.E.K.), T32 GM008268 (to K.E.S.), T32 GM07270 (to D.M.W.)]; and the National Science Foundation [grant number MCB0615632 (to R.E.K.)].


We thank P. Brzovic, N. Zheng, R. Gardner and members of the Klevit laboratory for careful reading of this manuscript and many helpful comments prior to submission.

Abbreviations: APC, anaphase-promoting complex; BRCA1, breast cancer early-onset 1; BARD1, BRCA1-associated RING domain 1; Birc6, baculoviral IAP (inhibitor of apoptosis) repeat-containing 6; cCbl, Casitas B cell lymphoma; Cdc34, cell division cycle 34; CHIP, C-terminus of Hsc70 (heat-shock complex 70)-interacting protein; C-lobe, C-terminal lobe; CNOT4, CCR (chemokine receptor) 4–NOT transcription complex; Cul1, cullin 1; E6AP, E6-associated protein; G2BR, Ube2g2-binding region; GAP, GTPase-activating protein; HECT, homologous with E6-associated protein C-terminus; NEDD4, neural-precursor-cell-expressed, developmentally down-regulated 4; NEDD4L, NEDD4-like; NIP45, NFAT (nuclear factor of activated T-cells)-interacting protein of 45 kDa; N-lobe, N-terminal lobe; Pc2, proprotein convertase 2; RanBP2, Ran-binding protein 2; SCF, Skp1/cullin/F-box; SUMO, small ubiquitin-related modifier; SIM, SUMO-interacting motif; SMURF2, SMAD-specific E3 ubiquitin protein ligase 2; TRAF6, tumour-necrosis-factor-receptor-associated factor 6; Ub, ubiquitin; UBA, Ub-associated domain; Ubl, Ub-like; UEV, Ub enzyme variant; Y2H, yeast two-hybrid


View Abstract