Solution structure of the flexible class II ubiquitin-conjugating enzyme Ubc1 provides insights for polyubiquitin chain assembly.

E2 conjugating enzymes form a thiol ester intermediate with ubiquitin, which is subsequently transferred to a substrate protein targeted for degradation. While all E2 proteins comprise a catalytic domain where the thiol ester is formed, several E2s (class II) have C-terminal extensions proposed to control substrate recognition, dimerization, or polyubiquitin chain formation. Here we present the novel solution structure of the class II E2 conjugating enzyme Ubc1 from Saccharomyces cerevisiae. The structure shows the N-terminal catalytic domain adopts an alpha/beta fold typical of other E2 proteins. This domain is physically separated from its C-terminal domain by a 22-residue flexible tether. The C-terminal domain adopts a three-helix bundle that we have identified as an ubiquitin-associated domain (UBA). NMR chemical shift perturbation experiments show this UBA domain interacts in a regioselective manner with ubiquitin. This two-domain structure of Ubc1 was used to identify other UBA-containing class II E2 proteins, including human E2-25K, that likely have a similar architecture and to determine the role of the UBA domain in facilitating polyubiquitin chain formation.

Labeling of proteins with the molecule ubiquitin is an important cellular function that is required for protein degradation, cell cycle control, stress response, DNA repair, signal transduction, transcriptional regulation, and vesicular traffic (1)(2)(3)(4). In this process, the number and topology of the ubiquitin molecule chains underscores the fate of the substrate and determines the biochemical pathway followed. For example, labeling the substrate with a single ubiquitin (monoubiqutination) is important for cellular regulation (5,6). On the other hand ubiquitin-dependent proteolysis, the process responsible for the turnover of damaged or misfolded proteins in the cell, involves labeling the substrate protein with a polyubiquitin chain that is subsequently recognized by the 26 S proteasome facilitating substrate degradation. The most common polyubiquitin chains are formed via isopeptide bond linkages between the C-terminal (Gly 76 ) of one ubiquitin molecule and the side chain ⑀-NH 2 from Lys 48 of another. However, other configurations are possible including the Lys 63 linkage, which are important in the postreplicative DNA repair pathway (7).
The ubiquitination degradation pathway is described as a cascade of events in which ubiquitin is passed through three enzymes until it reaches a protein selected for degradation. The first step involves an ATP-dependent activation of ubiquitin by an ubiquitin-activating enzyme (E1) 1 forming a high energy E1-ubiquitin thiol ester complex. The activated ubiquitin is then passed from the E1 to an ubiquitin-conjugating enzyme (E2) forming a second thiol ester intermediate between the E2 and ubiquitin. Labeling the target protein with ubiquitin is catalyzed by an E3 ligase protein, either by direct transfer of the ubiquitin to the substrate from the E2 (RING E3) or by thiol ester formation between ubiquitin to an E3 (HECT E3) and subsequent transfer to the substrate (4). Some details of this ubiquitin-mediated cascade have been garnered from the structures of several E2 enzymes showing a core 150-residue catalytic ␣/␤ fold that is maintained upon complexation with either a HECT (8) or RING (9, 10) E3 ligase or in the E2ubiquitin thiol ester intermediate (11). While these structures show the interactions between the E2, E3, and ubiquitin proteins, details how the transfer of ubiquitin to the substrate occurs and how the polyubiquitin chain is constructed are more uncertain. More recently, three-dimensional structures of the heterodimeric complex between the canonical E2 protein Ubc13 in complex with an E2 variant protein Mms2 have shown how these proteins function together to assemble Lys 63 -linked polyubiquitin chains (12)(13)(14). Similar structural details for construction of Lys 48 -linked polyubiquitin chains are not available, although it has been suggested that dimeric forms of other E2 proteins might also be required (15,16).
The E2 conjugating proteins are considered key enzymes in the ubiquitination pathway. All E2 proteins have a 150-residue catalytic domain that is structurally conserved through many species. For example, 11 E2 proteins have been identified in Saccharomyces cerevisiae, and at least 25 are known in mammals. E2 proteins are divided into three classes in which class I enzymes are the simplest and are comprised exclusively of the core catalytic domain that contains the active site cysteine residue required for thiol ester formation with ubiquitin. Several E2 proteins are more complex than the class I members and have either N-or C-terminal extensions. Class II E2 proteins have a C-terminal extension or a "tail," whereas class III E2 proteins have an additional N-terminal sequence (17). One of the key functions of the class II E2 conjugating enzymes is the creation of the polyubiquitin chain required for protein labeling and subsequent degradation. For example the mam-* This work was supported by operating and maintenance grants from the Canadian Institutes of Health Research (to G. S. S.) and infrastructure grants from the Canada Foundation for Innovation and the Academic Development Fund of The University of Western Ontario. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  malian E2 protein E2-25K is able to synthesize free Lys 48linked polyubiquitin chains in the absence of an E3 enzyme (18). The E2 proteins Ubc1 and Ubc3 (Cdc34) from S. cerevisiae are able to assemble polyubiquitin chains in conjunction with an auto-ubiquitination activity (16,19). A variety of other mechanistically distinct functions have been identified for the C-terminal extensions in class II E2 enzymes including dimerization (20), substrate recognition (21), and anchoring of specific E2 enzymes to the cystolic side of the endoplasmic reticulum (22).
Despite the diverse range of functions, and the integral nature of class II E2 conjugating proteins in the ubiquitination pathway, a three-dimensional structure of one of these proteins has not been reported. In this work we present the solution structure of the 215-residue protein Ubc1, a class II E2 conjugating enzyme from S. cerevisiae shown to be important for degradation of short-lived proteins especially during the G 0 -G 1 transition accompanying spore germination (23). We show that Ubc1 is a unique two-domain E2 enzyme containing a canonical catalytic domain and a C-terminal UBA domain separated by a flexible tether. Furthermore, we have used NMR chemical shift analysis to determine that ubiquitin binds to the Ubc1 UBA domain with little or no perturbation of the catalytic domain. Our structural results and sequence analysis indicate several other class II E2 proteins, including mammalian E2-25K, likely have this two domain architecture. We have used this information to provide insights into Lys 48 -linked polyubiquitin chain formation by these class II E2 enzymes.
Heteronuclear 15 N{ 1 H} NOEs were measured using the methods of Farrow et al. (34) using a 3-s irradiation period and a 2-s or a 5-s relaxation delay. Experiments were done in triplicate and averaged. 1 H-15 N residual dipolar couplings were measured using IPAP-HSQC experiments (35) using a 0.4 mM sample of 15 N-labeled Ubc1 in 12 mg/ml pf1 phage. NMR spectra were processed using NMRPipe (36) and analyzed using Pipp and Stapp (37) or NMRView (38) software using a Sun Ultra 10 workstation.
Structures of Ubc1 were calculated using the simulated annealing protocol in the program CNS (39). The structures of Ubc1 were generated from 1864 NOEs, including 858 intraresidue, 363 sequential, 357 short range and 286 long range and 63 hydrogen bond distance restraints, 194 dihedral angle restraints, and included 29 1 H-15 N residual dipolar couplings for residues 170 -215. Interproton distances were calibrated for all proton pairs using maximum and minimum nOe intensities for known possible d NH␣ distances. Dihedral restraints were determined from the program TALOS (torsion angle likelihood obtained from shifts and sequence similarity) (40) based on the C␣, C␤, CЈ, H␣, and N chemical shift information for Ubc1 where 9 out of 10 predictions fell in the same region of the Ramachandran plot. The resulting and angles were restricted to two times the error from the TALOS output for the structure calculations. Hydrogen bonds were identified from the slowly exchanging NH residues after 2-h incubation after solvent exchange. For each hydrogen bond, two distance restraints were used, NH-O (1.8 -2.3 Å) and N-O (2.3-3.3 Å). 1 H-15 N residual dipolar couplings were included in the structure calculations with a precision equal to the digital resolution of the HSQC spectra (Ϯ0.75 Hz). Values for D a and R were estimated from the program DYNAMO using a preliminary fold from CNS trials and measured 1 H-15 N residual dipolar couplings. Final and initial force constants were 0.01 and 1.0 kcal/mol, respectively. Structural similarity searches were performed using the DALI web server (41), and sequences were aligned using T-Coffee (42).
Ubc1-Ubiquitin Titrations-The equilibrium dissociation constant for the Ubc1-ubiquitin complex was measured by NMR spectroscopy using 1 H-15 N HSQC experiments. 15
A 22-residue flexible tether links the catalytic domain of Ubc1 to its C-terminal domain. During the structure calculations it became apparent this connecting region contained little regular structure due to the absence of long range NOE contacts to either the catalytic or C-terminal domains and the observation of very few sequential and short range NOEs. Furthermore, no long range NOEs between the catalytic domain and C-terminal domain were observed, indicating that they must be isolated from each other. Previous chemical shift perturbation studies indicated that the C-terminal domain might interact with the catalytic domain, proximal to the start of the tether region (Ser 150 ). The structure of Ubc1 (Fig. 1A) shows that the C-terminal domain adopts many orientations with respect to the catalytic domain, most which envelope the ␤1, ␤2, and ␣4 regions, the sites of the largest chemical shift changes observed previously. Since NOEs were not observed between the two domains it would appear that the chemical shift changes are largely a result of transient interactions or the proximity of the C-terminal domain (in sequence) to the catalytic domain junction. To establish the relationship between the domains, 15 N{ 1 H} heteronuclear NOE experiments were performed to determine whether the two domains had similar relaxation properties that would indicate a more globular structure or different properties that would indicate a more elongated structure with two independent domains. Analysis of the 15 N{ 1 H} heteronuclear NOE measurements indicated there was a clear difference between the average NOE from the catalytic and the C-terminal domains (Fig. 2). The catalytic domain had an average NOE of about 0.72, consistent with a m of about 4.4 ns. This average NOE was very similar to that obtained for UBC9 (0.74), a 15-kDa ubiquitin-like conjugating enzyme that lacks a C-terminal domain (49). UBC9 displays a similar ␣/␤ tertiary fold when compared with the catalytic domain of Ubc1 and also exhibits a more oblate shape, typical of the structures of other E2 catalytic domains. In contrast the C-terminal domain of Ubc1 had an average NOE of about 0.56, indicating that it has a smaller volume than the catalytic domain and is not associated with it. The average NOE for the C-terminal domain translates to a m of about 2.6 ns. From this value, an approximate molecular mass of 5 kDa is estimated, in excellent agreement with that expected for residues His 170 -Lys 215 (4.9 kDa). Furthermore, the heteronuclear NOE experiments showed that residues Asn 154 , Gly 155 , Gln 156 , and Lys 157 had negative NOE values indicative of a region with high mobility on the nanosecond time scale. These observations are similar to those found for other bilobate proteins such as calmodulin (50) and the cytokenesis protein Cdc4p (51). Both of these proteins exhibit 15  The C Terminus of Ubc1 Is a UBA Domain-The C termini of class II E2 enzymes have been implicated in a variety of processes including interactions with specific E3 proteins, modification of ubiquitination patterns, and recognition of substrates targeted for degradation. With the three-dimensional structure of this domain now in hand we probed structural data bases using DALI (41) and VAST (52) to identify structural relatives to the C-terminal domain of Ubc1 that may uncover its function. It was further anticipated that this may establish a general biological role for many other class II E2 conjugating enzymes. Both DALI and VAST analyses identified the UBA domain UBA(2) (53) from the nucleotide excision repair protein HHR23A as the closest structural relative based on a Z-score of 3.3 (DALI) and a backbone r.m.s.d. of 1.9 Å. When only the three helices ␣5-␣7 of Ubc1 are considered the superposition improves to a r.m.s.d. of 1.4 Å. Furthermore, a second UBA domain from HHR23A, UBA(1) (54), was found to have a similar structure (r.m.s.d. of 1.6 Å), using the same helical regions (Fig. 3A). This helical arrangement is also observed in structures of other UBA domains including those from p62 (55) and Swa2p (56).
The UBA (ubiquitin-associated) domain was first identified by Hofmann and Bucher (57) as a potential ubiquitin binding domain. Elegant structural work by Mueller et al. (54) has shown that UBA domains possess a conserved hydrophobic core comprised of residues primarily found on the interior faces of the three helices. When the C-terminal UBA domain from Ubc1 was superimposed with both UBA(1) and UBA(2) (Fig. 3A), it was apparent that analogous hydrophobic interactions in Ubc1 were observed. In particular residues in helix ␣5 for Ubc1 (Ile 173 , Phe 176 , and Phe 181 ) lie in similar positions as Leu 167 , Ile 170 , and Tyr 175 from UBA(1) and Ile 324 , Leu 327 , and Phe 332 from UBA (2). The third residue in this triad (Phe 181 ) is preceded by a glycine (Gly 180 ). This GF pair, initially identified by sequence alignment, is highly conserved in all UBA domains (57) (58), a close structural relative that also forms a three-helix bundle.
The structure of the UBA domain from S. cerevisiae Ubc1 was used along with its sequence to identify other class II E2 conjugating enzymes that may possess this domain. An initial group of 25 class II enzymes was considered. This was consolidated to those E2 enzymes that, in addition to the conserved catalytic domain, possessed a C-terminal extension of more than 30 residues. Alignment of these proteins using the conserved residues that comprise the hydrophobic core of the UBA domain in Ubc1, UBA(1) and UBA (2), and the GF couple near the C terminus of helix ␣5 revealed six other class II E2 conjugating enzymes that contain UBA domains (Fig. 3B). It is interesting that most of these UBA domains were initially predicted by Hofmann and Bucher (57), based on sequence alignment of a broad range of proteins involved in ubiquitination, although S. cerevisiae Ubc1 fell below the threshold. However, based on the structure of Ubc1 and the other E2 sequences several new observations and differences can be made. For example, in helices ␣5 and ␣6 it appears that the first hydrophobic core position favors a ␤-branched residue (V, I). Furthermore, only two of the E2 proteins have extended loops between helices ␣6 and ␣7, where there is little sequence similarity. The position for the start of the UBA domain in each of these E2 proteins also differs, indicating that the linker between the catalytic domain and the UBA domain is of differing lengths. Although this region is almost certainly flexible, based on the Ubc1 structure shown here, this could indicate that differing roles for the UBA domain in conjunction with the catalytic domain.
Regiospecific  (59), Mms2, and Uev1a (60) have been shown to interact in a non-covalent fashion with ubiquitin or ubiquitin homologues. So in principle, ubiquitin could interact non-covalently with both the N-terminal catalytic domain and the C-terminal UBA domain. To determine whether ubiquitin interacts with Ubc1 and to distinguish which domain it might bind to, a titration of 15 N-labeled Ubc1 with unlabeled ubiquitin was followed using a series of 1 H-15 N HSQC spectra (Fig. 4A). In these spectra, only resonances from the 15 N-labeled Ubc1 protein are visible as a function of added ubiquitin. The spectra indicated that only a subset of the 1 H-15 N cross-peaks for Ubc1 shifted upon sequential additions of ubiquitin. Residues that underwent the most significant changes included Gly 180 and Phe 181 in the linker region between helices ␣5 and ␣6, residues Glu 177 and Ser 178 at the C terminus of helix ␣5, Asn 201 and Asp 202 in the loop between helices ␣6 and ␣7, and Glu 211 and Glu 212 near the C terminus of helix ␣7. In addition changes were noted for His 170 in helix ␣5 and Asn 207 , Arg 208 , Ile 209 , Leu 213 , and Leu 214 in helix ␣7 (Fig. 3B). Generally an increased magnitude in chemical shift change, as a function of ubiquitin, is indicative of residues on Ubc1 that are proximal to the binding site. The resulting plot of chemical shift changes as a function of residue number shows a skewed arrangement of residues that have larger perturbations and ones that are minimally affected (Fig. 4B). The most striking feature of this plot is that only residues in the UBA domain (His 170 -Lys 215 ) from Ubc1 are significantly perturbed. Furthermore, all residues measured in the catalytic domain fell well below the threshold shift (0.05 ppm) for the entire protein. These two observations indicate that the non-covalent interaction of ubiquitin is restricted to the UBA domain at the C terminus of Ubc1 and provides further evidence for the separated nature of the UBA and catalytic domains in the structure. A specific interaction of ubiquitin with the UBA domain in Ubc1 is supported by earlier results that showed a truncated version of Ubc1 (residues 1-150) comprising the catalytic domain, but lacking residues 151-215 of the C-terminal domain, does not interact in a non-covalent manner with ubiquitin (11). The average dissociation constant for the non-covalent interaction of ubiquitin with the UBA domain of Ubc1 was 228 Ϯ 69 M derived from non-linear curve fitting of the chemical shift changes (Fig.  4, C and D).
Interacting Surfaces between the UBA Domain in Ubc1 and Ubiquitin-Binding of ubiquitin to Ubc1 indicated that only the C-terminal UBA domain of Ubc1 was involved in the interaction. As a result the chemical shift perturbations determined from titration experiments were mapped to the surface of the UBA domain in Ubc1 (Fig. 5A). This analysis showed that residues that exhibited the largest chemical shift changes (Glu 177 , Ser 178 , Gly 180 , and Phe 181 ), near the C terminus of helix ␣5, are located on the Ubc1 surface near residues from helix ␣7 (Ala 206 , Asn 207 , Arg 208 , Ile 209 , Glu 211 , Glu 212 , Leu 213 , and Leu 214 ) and the loop between helix ␣6 and ␣7 (Asp 199 , Asn 201 , Asp 202 ). On the Ubc1 surface these residues form a contiguous patch centered about Gln 179 , Gly 180 , Phe 181 , and Leu 214 . The location of the residues involved in this surface is very similar to that uncovered for the UBA domains from HHR23A (Fig. 5, B and C). Each of these UBA domains presents a four-residue patch comprising Ser 172 , Met 173 , Gly 174 , and Tyr 175 in UBA(1) and Ala 329 , Leu 330 , Gly 331 , and Phe 332 in UBA(2) (61). Of these the GF(GY) sequence found in Ubc1 is common to other UBA containing E2 conjugating enzymes (Fig.  3B). The region comprising residues Ala 206 , Asn 207 , and Arg 208 shares features common to both UBA(1) (His 192 , Arg 193 ) and UBA(2) (Glu 348 , Asn 349 ) (61). Sequences of UBA domains reveal this region is variable with the exception of the central residue (Asn 207 in Ubc1), which is typically polar. In Ubc1 analysis of this ubiquitin-binding surface indicates that it is remarkably one-sided showing no residues affected by ubiquitin binding on the reverse side of the orientation depicted in Fig. 5A. Furthermore, it is apparent that the sequences of other UBA-containing E2 conjugating enzymes (Fig. 3B) share many of these interacting residues indicating these class II E2 enzymes likely utilize a similar surface for ubiquitin interaction.
The observation that residues Asp 199 , Asn 201 , and Asp 202 in the extended loop between helices ␣6 and ␣7 undergo some of the largest chemical shift changes and are part of the ubiquitin-binding surface is unique to Ubc1. Although other UBA domains have variable length linkers between these two heli-ces only one other UBA-containing E2 enzyme (Ubc_SCHPO, Fig. 3B) appears to share this feature. Two of the interacting residues in this loop are acidic (Asp 199 , Asp 202 ) and along with several other acidic residues affected most by ubiquitin binding (Asp 174 , Glu 177 , Glu 211 , Glu 212 ) make up a largely acidic band that encircles the hydrophobic patch comprising Phe 181 , Ile 209 , and Leu 213 (Fig. 5A). While Asp 199 is unique to Ubc1, the location of Asp 202 is frequently occupied by polar residues in the UBA domains shown in other E2 conjugating enzymes (Fig.  3B) or those described by Mueller and Feigon (54). Given the clustering of lysine and arginine residues toward the C terminus of ubiquitin, this additional acidic surface in Ubc1 might be expected to stabilize the Ubc1-ubiquitin complex. Consistent with this the dissociation constant for Ubc1 with ubiquitin (228 Ϯ 69 M) is among the tightest observed compared with ubiquitin binding to UBA domains from HHR23A and HHR23B (300 -500 M) (61,62).
The surface of ubiquitin involved in the interaction with the UBA domain from Ubc1 was probed using chemical shift perturbation experiments. Since Ubc1 is only soluble to ϳ500 M, a complete reverse titration of 15 5D). The surface is strikingly similar to that obtained for the ubiquitin-binding surfaces for the interaction with UBA(1) and UBA(2) from HHR23A (61) and in the thiol ester E2 complex (63) (Fig. 5E). One potential difference is the apparent lack of interaction of Thr 9 , Gly 10 , and Lys 11 in ubiquitin with Ubc1. In fact these residues may be involved in the binding surface, but their rapid amide exchange with water under our experimental conditions made their changes in chemical shift impractical to measure. DISCUSSION E2 conjugating enzymes are the key intermediary enzymes in the ubiquitin-mediated proteolysis pathway. It has been well established that the catalytic domain of E2 conjugating enzymes is responsible for the formation of the thiol ester intermediate, required for transfer of ubiquitin to the fated substrate protein, in conjunction with an E3 ligase. While all E2 proteins possess this catalytic domain, many contain a C-terminal extension (class II) that is not required for thiol ester formation. These E2 proteins have been suggested to have additional roles in the ubiquitination mechanism that can be attributed to their C termini including dimerization of the E2, substrate recognition, and involvement in ubiquitination chain formation. The solution structure of Ubc1 is the first reported structure of a class II E2 conjugating enzyme and allows an examination of each of these functional roles.
Ubc1 reveals a two-domain protein comprising an N-terminal catalytic domain and a UBA domain separated by a flexible tether. The three-helix bundle and tether for Ubc1 total about 65 amino acids where about one-third of these arise from the tether region. Sequence similarity based on the Ubc1 structure has identified several other class II E2 conjugating enzymes (Fig. 3B) that likely have a similar two-domain architecture as Ubc1. The human class II E2 enzyme E2-25K is perhaps the most widely studied of these. Gel filtration experiments have shown this E2 protein to be an elongated asymmetric molecule that is susceptible to proteolysis in a similar region defined by the flexible tether in Ubc1 (64). Other E2 proteins displaying this behavior include S. cerevisiae Ubc2 (Rad6, 20 kDa) (44) and Ubc3 (CDC34, 32 kDa) (15), which have Stoke's radii much higher than expected for a globular protein. While this could be attributed to oligomerization, sedimentation equilibrium studies of Ubc1 (24), Ubc2 (44), and Ubc3 (15) show these proteins to be monomeric. Thus, the combination of the two-domain structure of Ubc1 shown here, and the asymmetric shapes determined from gel filtration indicates the class II E2 proteins E2-25K, Ubc2, and Ubc3 likely have monomeric, elongated structures analogous to Ubc1. Of these, only E2-25K appears to have a C-terminal UBA domain based upon sequence comparison to Ubc1 (Fig. 3B). The C terminus of Ubc2 appears to be too short (23 residues) and would likely contain only the flexible tether region, while Ubc3 must have a tether and an as yet unidentified domain structure.
The structure of Ubc1 shows that it is the first E2 conjugating enzyme that possesses both a UBA domain and catalytic domain. The binding modes of these two ubiquitin-interacting domains are significantly different, since the catalytic domain is responsible for accepting the ubiquitin molecule from the E1 activating enzyme, forming a covalent thiol ester linkage at Cys 88 . This domain has been shown to have little or no inherent affinity for ubiquitin that would lead to a non-covalent interaction. Quite the opposite, the UBA domain from Ubc1 interacts in a non-covalent fashion with ubiquitin with little or no perturbation of the catalytic domain. So, in principle class II E2 enzymes such as Ubc1 and E2-25K might be able to interact with two separate ubiquitin molecules, or a thiol ester-bound ubiquitin could undergo further interactions with a UBA domain. How then does the UBA domain work with the catalytic domain to build or manipulate polyubiquitin chains? In Ubc1 the UBA domain affects the manner in which this protein is auto-ubiquitinated (19 (61). D, surface diagram of ubiquitin where the blue regions represent surface residues that underwent chemical shift changes, ͚⌬␦ Ͼ 0.03 ppm, upon Ubc1 binding. E, surface diagram of ubiquitin in which the magenta regions represent the residues that underwent a change in chemical shift upon binding UBA(2) as described by Mueller et al. (61). F, surface diagram of ubiquitin in which the green regions represent the amide cross-peaks that underwent a decrease in intensity upon thiol ester formation with truncated Ubc1 as described by Hamilton et al. (63). affected by thiol ester formation (Fig. 5F) and interaction with the UBA domain (Fig. 5D) both utilize the His 68 , Val 70 , Leu 71 , Arg 72 , and Leu 73 region. This could indicate that polyubiquitin chain formation in Ubc1 is perturbed through interaction of the UBA domain with the non-anchoring ubiquitin molecule. Alternatively, interaction of the anchoring ubiquitin at Cys 88 or Lys 93 might be a more flexible arrangement allowing preferential interaction of the ubiquitin "face" with the UBA domain and inhibiting further chain building.
In contrast to Ubc1, the human enzyme E2-25K requires the C-terminal UBA domain to build polyubiquitin chains, linked at Lys 48 (64). The structure of Ubc1 lends some insight into this process, which requires nucleophilic attack of the thiol ester ubiquitin (donor) by the side chain ⑀-NH 2 from Lys 48 from an acceptor ubiquitin (65). The flexible tether linking the catalytic domain to the UBA domain in Ubc1 is perfectly adapted to allow the acceptor ubiquitin bound to the UBA domain to be brought into close proximity with the donor ubiquitin in the thiol ester allowing formation of a polyubiquitin species bound to the UBA domain. Several lines of evidence support this idea. First, deletion of the UBA domain in E2-25K does not impede thiol ester formation but results in elimination of polyubiquitin chain formation (64). Second, chemical shift experiments and the multiple orientations of the UBA domain with respect to the catalytic domain show that transient association of the two domains is possible. For E2-25K this might account for protection to alkylation of the catalytic domain by its C terminus (64). Furthermore, this mechanism would be facilitated by a higher affinity of the UBA domain for longer ubiquitin chain length. While not clearly shown for E2-25K, other work shows the UBA affinity for ubiquitin increases as the chain length increases (66,67). This mechanism bears a strong analogy to that recently uncovered for the Mms2/Ubc13 heterodimer involved in Lys 63 polyubiquitin chain formation (12)(13)(14). In this complex a thiol ester is formed between the canonical E2 Ubc13 and ubiquitin. Ubiquitin chain formation occurs by transfer of this ubiquitin to a second ubiquitin molecule non-covalently bound by Mms2 (an inert E2 variant).
UBA domains have been suggested to have two possible, yet seemingly opposite, roles in the ubiquitin-mediated degradation pathway. Interaction of monomeric ubiquitin with a UBA domain is suggested to inhibit further chain elongation leading to a culling of the degradation of the substrate (66,67). In contrast, polyubiquitin chains linked via the Lys 48 isopeptide bond have a 1000-fold greater affinity than a single ubiquitin molecule for a UBA domain and this might facilitate transfer of a polyubiquitinated substrate to the proteasome (67,68). The structure of the class II conjugating enzyme Ubc1 shows how this and related enzymes such as E2-25K might contribute to both of these affects. It is anticipated that future structural and biochemical experiments will delineate and clarify the roles of these important enzymes in protein degradation.