Ubiquitin manipulation by an E2 conjugating enzyme using a novel covalent intermediate.

Degradation of misfolded and damaged proteins by the 26 S proteasome requires the substrate to be tagged with a polyubiquitin chain. Assembly of polyubiquitin chains and subsequent substrate labeling potentially involves three enzymes, an E1, E2, and E3. E2 proteins are key enzymes and form a thioester intermediate through their catalytic cysteine with the C-terminal glycine (Gly76) of ubiquitin. This thioester intermediate is easily hydrolyzed in vitro and has eluded structural characterization. To overcome this, we have engineered a novel ubiquitin-E2 disulfide-linked complex by mutating Gly76 to Cys76 in ubiquitin. Reaction of Ubc1, an E2 from Saccharomyces cerevisiae, with this mutant ubiquitin resulted in an ubiquitin-E2 disulfide that could be purified and was stable for several weeks. Chemical shift perturbation analysis of the disulfide ubiquitin-Ubc1 complex by NMR spectroscopy reveals an ubiquitin-Ubc1 interface similar to that for the ubiquitin-E2 thioester. In addition to the typical E2 catalytic domain, Ubc1 contains an ubiquitin-associated (UBA) domain, and we have utilized NMR spectroscopy to demonstrate that in this disulfide complex the UBA domain is freely accessible to non-covalently bind a second molecule of ubiquitin. The ability of the Ubc1 to bind two ubiquitin molecules suggests that the UBA domain does not interact with the thioester-bound ubiquitin during polyubiquitin chain formation. Thus, construction of this novel ubiquitin-E2 disulfide provides a method to characterize structurally the first step in polyubiquitin chain assembly by Ubc1 and its related class II enzymes.

The ubiquitin-dependent proteolysis pathway controls the removal of damaged and misfolded proteins in the cell. One of the key steps in this pathway is the assembly of a polyubiquitin chain that ultimately targets a substrate for degradation (1,2). This process involves tagging of a substrate protein with an arrangement of four to eight ubiquitin molecules linked in series through the C-terminal Gly 76 of one ubiquitin molecule and the side-chain ⑀-NH 2 from Lys 48 of another (3). Once tagged, the polyubiquitinated protein is recognized by the 26 S proteasome, where it is degraded. The building of a polyubiquitin chain and subsequent labeling of a substrate protein is a complex process potentially involving the passage of ubiquitin through three different enzymes (1,4,5). Initially, ubiquitin is activated in an ATP-dependent step forming a high energy E1 1 -ubiquitin thioester complex. Ubiquitin is then transferred to an E2 or ubiquitin-conjugating enzyme forming a thioester intermediate. E2 proteins have been demonstrated recently to bind to the ubiquitin-like domain of the E1 providing insight into the mechanism in which the thioester-bound ubiquitin is passed to the E2 (6 -8). Two different E3 ligase proteins can catalyze the final passage of ubiquitin to the substrate. For RING E3 ligases, ubiquitin or a polyubiquitin chain is transferred directly from the E2 to the substrate, whereas HECT E3 ligases form an ubiquitin-E3 thioester prior to ubiquitin transfer to the substrate (5).
The E2 conjugating proteins are the key enzymes in this pathway because they are required to transfer ubiquitin either to the E3 ligase (HECT E3) or to the substrate (RING E3). All E2 proteins have a 150-residue catalytic domain that is structurally conserved throughout many species and contains the cysteine residue necessary for thioester formation with Gly 76 of ubiquitin. Structures of E2 proteins show that the catalytic domain has an ␣/␤-fold that is maintained upon complexation with either HECT (9) or RING (10, 11) E3 ligases. Details on the involvement of the E2 thioester in the polyubiquitin chain-building process or the mechanism for transfer of ubiquitin from an E2 protein to an E3 or substrate are less certain. The E2 proteins Ubc1 and Ubc3 (Cdc34) from Saccharomyces cerevisiae assemble polyubiquitin chains and exhibit autoubiquitination activities (12,13), whereas the mammalian E2 protein E2-25K can assemble polyubiquitin chains in the absence of an E3 enzyme (14). Furthermore, Ubc1 has been shown to be important for the creation of polyubiquitin chains required for protein labeling and subsequent degradation (12).
The three-dimensional structure of Ubc1 from S. cerevisiae, determined by NMR spectroscopy, provides some insights into these biological activities (15). The structure shows that Ubc1 and its related class II conjugating enzymes form a unique two-domain protein having a typical ␣/␤-fold catalytic domain connected via a flexible tether to a C-terminal UBA (ubiquitinassociated) domain. UBA domains, such as the one in Ubc1, are capable of interaction with mono-or polyubiquitin chains in a non-covalent fashion and may result in either inhibition of degradation or transfer enhancement via ubiquitin interaction (16 -19). The E2 conjugating protein Ubc1, is a well positioned candidate to examine the transfer of ubiquitin from the ubiq-uitin-E2 thioester because of its ability to moderate and build polyubiquitin chains.
Mechanistic experiments that examine the role of the ubiquitin-E2 thioester in polyubiquitin chain assembly have been difficult because of the inherent instability of the thioester complex. To date the best details have been garnered from models derived from NMR chemical shift perturbation data for ubiquitin-E2 thioester intermediates (20). Attempts to stabilize this complex for more detailed structural and mechanistic experiments have met with limited success. To circumvent this we have created a novel disulfide-linked ubiquitin-Ubc1 complex that mimics the ubiquitin-E2 thioester intermediate. We show that this ubiquitin-Ubc1 complex can be purified in high amounts, is stable for long periods of time, and has similar structural characteristics to the ubiquitin-E2 thioester intermediate. We have used this complex and NMR spectroscopy to show that the UBA domain can bind ubiquitin in a non-covalent fashion even in the presence of an ubiquitin molecule covalently bound at the catalytic domain. These results provide a glimpse at the first step of polyubiquitin chain formation by Ubc1 and its related class II E2 enzymes.
Ub Cys -Ubc1 Disulfide Complex Formation-Solutions (0.1 mM) of Ub Cys , Ubc1 1-150 , and Ubc1 were fully reduced with 2 mM tris(2-carboxyethyl)phosphine. An excess of either Ubc1 1-150 or Ubc1 was combined with Ub Cys and dialyzed at 4°C against several changes of 100 mM Na 2 HPO 4 /NaH 2 PO 4 , 100 mM NaCl, 10 M CuCl 2 at pH 7.5. The progress of disulfide complex formation was monitored by non-reducing SDS-PAGE and was considered complete when the reduced Ub Cys was exhausted. The protein solution was concentrated and purified by size exclusion chromatography on Sephadex G-75 with 25 mM Tris-HCl, 250 mM NaCl, 1 mM EDTA at pH 7.5. Fractions were analyzed on SDS-PAGE under non-reducing conditions, and those containing pure Ub Cys -Ubc1 1-150 or Ub Cys -Ubc1 were pooled and concentrated for NMR experiments.
NMR Spectroscopy-NMR samples of the Ub Cys -Ubc1 complexes with the following isotopic labeling schemes were used: [ 13 C, 15 (34.8 mM) to a maximum of 4.0 eq Ub:Ub Cys -Ubc1 were titrated into the sample. Chemical shift deviations were calculated for each isolated residue and analyzed by non-linear regression using the software Prism 4. All protein concentrations were determined in triplicate by amino acid analysis.
The ubiquitin-E2 thioester intermediate formed in the ubiquitin-mediated proteolysis pathway occurs between the Cterminal Gly 76 carboxylate of ubiquitin and the side-chain thiol of Cys 88 . Based on a model of the Ubc1-ubiquitin thioester, the length of this linkage from Gly 76 C␣ to Cys 88 C␣ is ϳ4.8 Å. In attempting to stabilize the linkage between Ubc1 and ubiquitin we wanted to preserve this distance while increasing the stability of this junction to a significant extent. To accomplish this, we replaced the ubiquitin Gly 76 with a cysteine residue (Ub Cys ) to form a disulfide linkage between the side chains of Cys 76 in ubiquitin and Cys 88 in Ubc1. In this process the geometry of Cys 88 in Ubc1 should be unaffected. Ubc1 is an ideal candidate to model this disulfide reaction because its sequence contains exclusively the catalytic cysteine preventing any multiple ubiquitin adducts or disulfide exchange between different sites from occurring. Furthermore, in Ub Cys the cysteine side chain, being at the C terminus of the protein, should be flexible enough to adopt a preferred conformation that may be dictated by interaction with Ubc1. In modeling studies, the span between the C␣ atoms of Cys 76 in Ub Cys and Cys 88 in Ubc1 was ϳ4.9 -5.5 Å, a distance similar to that observed in the native ubiquitin-E2 thioester.
Formation of a Stable, Covalent Ubiquitin-E2 "Intermediate"-A disulfide-linked complex between Ub Cys and Ubc1 (Ub Cys -Ubc1) was formed by mixing fully reduced Ub Cys with an excess of fully reduced E2 protein in phosphate buffer containing a catalytic amount of Cu 2ϩ as an oxidizing agent. In both cases the K93R mutant of Ubc1 and the K48R mutant of ubiquitin were used. These mutations are utilized to prevent autoubiquitination and polyubiquitin chain formation reactions from occurring, although these reactions were not a factor in our experiments. A disulfide complex between Ub Cys and a truncated version of Ubc1 lacking the C-terminal domain (Ubc1 1-150 ) was also synthesized in addition to the complex utilizing the full-length Ubc1 protein containing the C-terminal UBA domain. Formation of the Ub Cys -E2 complex occurred for both Ubc1 and Ubc1 1-150 over a period of 10 h at room temperature at which point the Ub Cys had been exhausted ( Fig. 2A). In both experiments two products were observed, the Ub Cys -Ubc1 disulfide (or Ub Cys -Ubc1 1-150 ) as the major species and a minor amount of Ub Cys -Ub Cys disulfide. Over the same time period and under identical conditions there was little difference between the rates or extent of formation for Ub Cys -Ubc1 compared with that of the truncated Ub Cys -Ubc1  indicating that the C-terminal UBA domain in Ubc1 does not appear to enhance or inhibit formation of the disulfide. Both Ub Cys -Ubc1 and Ub Cys -Ubc1 1-150 could be readily purified by size exclusion chromatography, and their identities were confirmed by mass spectrometry (Ub Cys -Ubc1 MW obs 33,270.7, MW calc 33,274.2; Ub Cys -Ubc1 1-150 MW obs 25,421.4, MW calc 25,420.1). Because Ubc1 and Ubc1 1-150 contain a single cysteine residue, which is at the catalytic site, the Ub Cys -Ubc1 (and Ub Cys -Ubc1 1-150 ) disulfide must be formed using the catalytic site cysteine. The complexes were stable in solution for several weeks at room temperature. Formation of a disulfidelinked complex between Ub Cys and other E2 proteins was tested to assess the general applicability of this method. Both of the human E2 proteins, Ubc13 and UbcH7, formed stable disulfide complexes using the identical conditions for synthesis of Ub Cys -Ubc1. The sequence of Ubc13 has exclusively the catalytic cysteine and like Ubc1, the reaction afforded a single product, Ub Cys -Ubc13 (Fig. 2B) identified by mass spectrometry (Ub Cys -Ubc13 MW obs 26,180.8, MW calc 26,180.2). A disulfide reaction between Ub Cys and UbcH7 yielded a single Ub Cys -UbcH7 disulfide, and multiubiquitin adducts were not ob-served by non-reducing gel electrophoresis (Fig. 2B), although UbcH7 has a catalytic cysteine residue and two additional cysteine residues.
Ub Cys -Ubc1 Mimics an Ubiquitin-E2 Thioester Intermediate-The instability of the ubiquitin-E2 thioester precluded three-dimensional structure determination of this species and limited characterization of the interface between the two proteins. NMR spectroscopy was used to identify residues in ubiquitin or Ubc1 1-150 that decreased in intensity during formation of the thioester. However, the stability of the thioester (ϳ1 h) did not allow sequence-specific assignment of 1 H-15 N resonances in the Ub-E2 thioester complex thereby limiting this approach (27). Because Ub Cys -Ubc1 is orders of magnitude more stable than its thioester counterpart this has now allowed the 1 H, 13 C, and 15 N resonance (backbone) assignment of the 33-kDa Ub Cys -Ubc1 complex using conventional triple resonance techniques. Fig. 3 shows the 1 H-15 N HSQC spectrum of the Ub Cys -Ubc1 complex using 15 N-labeled Ubc1 and unlabeled Ub Cys . In general the spectrum is similar to Ubc1 alone indicating that disulfide formation between Ub Cys and Ubc1 did not result in global conformational changes in Ubc1. In addition calculation of the Chemical Shift Index (CSI) using CA, CB, and CЈ shifts (28) indicated that there was little change in the secondary structure of Ubc1 upon formation of the disulfide complex.
Although the inter-residue distance between the active site Cys 88 in Ubc1 and the C-terminal Cys 76 in Ub Cys is similar to that in the thioester, it is possible that intermolecular contacts between the two proteins could be modified. A subtle difference also exists between the Ub Cys -Ubc1 disulfide and the Ub-Ubc1 thioester. Because of the difference in covalent linkage, the Ub Cys -Ubc1 disulfide possesses a free C-terminal carboxylate that would normally be used to form the thioester bond. Previously, the ubiquitin-Ubc1 interface for the Ub-Ubc1 1-150 thioester has been partially characterized by following changes in line widths from 1 H-15 N HSQC spectra upon thioester formation (20). The protein-protein interface for the Ub-Ubc1 1-150 thioester complex was compared with that of Ub Cys -Ubc1 1-150 using 15 N-labeled Ub Cys to probe the similarity of the two interfaces for the two complexes. In the absence of Ubc1 1-150 , the 1 H-15 N HSQC for Ub Cys is well resolved with most resonances in similar positions to those found in the wild-type protein (Fig. 4A). Upon formation of Ub Cys -Ubc1 1-150 several resonances shift to new positions including Arg 48 , Val 70 , Leu 71 , Arg 72 , Leu 73 , and Cys 76 in Ub Cys (Fig. 4B). These residues are essentially the same as those identified in the Ub-Ubc1 1-150 thioester (Fig. 4C) (20) and are found in a tight cluster in the 1 H-15 N HSQC spectra. Histograms of the chemical shift perturbations for both Ub Cys (Fig. 4D) and Ub (Fig. 4E) upon complex formation have similar patterns indicating that the same residues are affected in ubiquitin at the ubiquitin-E2 interface in both complexes. Minor variations between the histograms may be a result of the changes in experimental conditions (pH and ionic strength) upon sequential addition of the reaction mixture (E1 and ATP) to the Ub, E2 sample, which is required for the synthesis of the Ub-Ubc1 1-150 thioester in situ (20). The Ub Cys -Ubc1 1-150 on the other hand could be purified, and the pH and ionic strength was controlled. In addition, residual Ub correlations are present in the 1 H-15 N HSQC spectrum of the thioester (Fig. 4C)  are common to those reported to undergo a decrease in peak intensity in the Ub-Ubc1 thioester intermediate (27), indicating that the interacting surface in Ubc1 is similar to that of the thioester intermediate. The interacting residues in both Ubc1 and ubiquitin in the disulfide complex are similar to those observed in the thioester, indicating that the disulfide complex mimics the thioester intermediate, and a minimal influence on the E2 surface is created by the presence of a charged carboxylate group near the active site.
Two major advantages exist for the Ub Cys -Ubc1 disulfide complex. First, the complex contains no side products such as uncomplexed ubiquitin or E2 that can hamper data analysis. For example, in the Ub Cys -Ubc1 complex, resonances resulting from residues in the "tail" region of ubiquitin (Val 70 , Leu 71 , Arg 72 , and Leu 73 ) are completely absent (Fig. 4B) from their original positions in Ub Cys (Fig. 4A). However in the Ub-Ubc1 1-150 thioester (Fig. 4C), remnants of these peaks exist because formation of the thioester only occurs to ϳ90% completion. Second, the stability of Ub Cys -Ubc1 1-150 has allowed backbone NMR assignments to be obtained for the Ub Cys component enabling identification of the new resonance positions in the Ub Cys -Ubc1 1-150 complex. This was not possible in the thioester due to hydrolysis of the Ub-Ubc1 1-150 thioester bond under aqueous conditions. Overall, the formation of the Ub Cys -Ubc1 1-150 complex not only mimics the Ub-Ubc1 1-150 thioester structure but also facilitates its analysis.

Ubc1 Coordinates Two Ubiquitin Molecules at Different
Sites-In addition to the catalytic domain, Ubc1 has a C-terminal UBA domain that non-covalently binds ubiquitin (15). UBA domains bind mono-or polyubiquitin chains with a dissociation constant between 300 and 500 M (29,30). Ubc1 has the potential to coordinate two ubiquitin molecules simultaneously, suggesting that during the Ub-Ubc1 thioester formation process, it is possible that the UBA domain might guide or perturb the interaction of ubiquitin with Ubc1. Alternatively, the UBA domain could also interact with the thioester-bound ubiquitin once the bond is formed at Cys 88 of Ubc1. Identification of either of these steps would provide insight into the first step of polyubiquitin chain formation. The Ub Cys -Ubc1 disulfide allowed us to examine the ubiquitin binding properties of the UBA domain of Ubc1 in the Ub Cys -Ubc1 disulfide complex. A similar investigation of the ubiquitin binding properties of the UBA in the Ub-Ubc1 thioester complex has not been possible because of the transient nature of the thioester conjugate.
We probed the Ub Cys -Ubc1 complex to determine whether this enzyme could coordinate two ubiquitin molecules concurrently. The Ub Cys -[ 15 N]Ubc1 complex was titrated with unlabeled ubiquitin, and each addition was followed by 1 H- 15 5A). In addition, residues Glu 177 , Ser 178 , and Glu 211 in the UBA domain of the Ub Cys -Ubc1 complex are strongly affected by ubiquitin binding (Fig. 5B). Several other residues (His 170 , Asp 199 , Asn 203 , Thr 205 , Ala 206 , Arg 208 , Ile 209 , and Leu 214 ) in the UBA domain of the Ub Cys -Ubc1 complex underwent smaller but significant chemical shift changes. Overall the pattern of residues most affected in Ub Cys -Ubc1, upon interaction with ubiquitin, was specific to the UBA domain of Ubc1. This is remarkably similar to that observed in Ubc1 alone (Fig. 5, B and C). To quantify the interaction of ubiquitin with the Ub Cys -Ubc1 complex, the 15 N chemical shift changes were analyzed by non-linear least-squares fitting (Fig. 6A) for a number of affected residues. This resulted in a dissociation constant of 259 Ϯ 105 M. The dissociation constant calculated from 15 N chemical shift changes for Ubc1 titrated with Ub was 280 Ϯ 69 M using comparable residues. When the data for the ubiquitin interaction with Ub Cys -Ubc1 were normalized and compared with the interaction of ubiquitin with Ubc1 alone, the curves were nearly superimposable, reinforcing the similarity of the strength and stoichiometries of the interactions (Fig. 6B). DISCUSSION Polyubiquitin chains linked through the C-terminal glycine of one ubiquitin to Lys 48 of another are a required recognition motif for substrate degradation by the 26 S proteasome. The mechanism in which these polyubiquitin chains are assembled is unclear. Two possibilities exist in which either the polyubiquitin chain is constructed on the E2 protein and transferred to an E3 or substrate, or the chain is built on the target substrate. In both cases the ubiquitin-E2 thioester complex is a key intermediate in this process. The transient nature of this thio-ester species has not allowed its three-dimensional structure to be determined or permitted direct experimentation showing how the thioester participates in the assembly of polyubiquitin chains. To date, the best structural analysis of the ubiquitin-E2 thioester complex has been found from peak intensity changes in NMR experiments upon thioester formation. This has indicated that the surface on ubiquitin at the protein-protein interface encompasses residues Val 70 -Gly 76 and Arg 48 (27). In the current work, the ubiquitin-E2 disulfide is the first thioester intermediate synthesized that mimics this surface. In the disulfide the ubiquitin surface includes residues Arg 48 , Val 70 -Leu 73 , and Cys 76 , which are common not only to those in the thioester but also in the UBA domain interface. This indicates the ubiquitin-E2 disulfide is a suitable mimic of the transient ubiquitin-E2 thioester.
The ubiquitin-Ubc1 disulfide complex is straightforward to synthesize in high yields and may be purified from any remaining starting materials by size exclusion chromatography. The product is stable for several weeks at room temperature. This is a remarkable improvement from other attempts to stabilize the ubiquitin-E2 linkage. For example, mutation of the active site cysteine in the E2, HsUbc2b, to a serine residue has allowed formation of an ester linkage between ubiquitin and the E2 (31)(32)(33). Although the resulting complex could be purified, it was short-lived at pH 6.7 and unstable at alkaline pH impeding any attempts to model the association of the two proteins or to use the complex for biochemical characterization (31). However, chemical shift analysis revealed that residues Val 70 -Gly 76 and Lys 48 on ubiquitin were most significantly perturbed upon HsUbc2b binding, a similar observation to that obtained for the ubiquitin-Ubc1 thioester and the disulfide complex described here. An alternative procedure traps the thioester intermediate by reducing it to a hemithioacetal with sodium borohydride (NaBH 4 ). This approach was successful in stabilizing the complex between ubiquitin and an ubiquitin hydrolase for several days allowing for structural characterization by NMR spectroscopy (34). Our method for mimicking the ubiquitin-E2 thioester by forming a stable disulfide complex with S. cerevisiae Ubc1, human Ubc13, and UbcH7 E2 proteins can be used for mechanistic or structural studies with ubiquitin and other ubiquitinconjugating enzymes. Approximately 20% of E2 conjugating enzymes contain exclusively the catalytic cysteine residue (35), and this procedure can be used as described in this work. For other conjugating enzymes such as UbcH7, which contains multiple cysteine residues, site-directed mutagenesis may be required to mutate the non-catalytic cysteine residues. In addition, the generality of this procedure illustrates that the disulfide complex designed could be used to investigate the intermediates between ubiquitin and other ubiquitin pathway enzymes, for example, E1 enzymes or HECT E3 ligases.
A unique feature of Ubc1 is that it is a flexible two-domain protein that consists of catalytic and UBA domains, representative of several other class II E2 proteins. Upon titration of Ub Cys -Ubc1 with Ub, the UBA domain is able to non-covalently bind a second ubiquitin molecule. Therefore, Ubc1 has the ability to bind an ubiquitin molecule as a thioester at the active site of the catalytic domain and a second ubiquitin at the C-terminal UBA domain. The ability of Ubc1 to interact with two ubiquitin proteins may explain how Ubc1 and its related class II E2 proteins can build or manipulate polyubiquitin chains. The presence of a UBA domain affects the chain-building properties of two homologous class II E2 proteins in vitro, Ubc1 and E2-25K. Ubc1 undergoes autoubiquitination by assembling polyubiquitin chains at Lys 93 near the Cys 88 active site (12). Polyubiquitin chains comprising up to 10 -12 ubiquitin molecules are assembled in the absence of the UBA domain. Shorter chains averaging only four ubiquitin molecules are built when full-length Ubc1 is utilized. A flexible tether (22 amino acids) links the catalytic and UBA domains in Ubc1. This tether is long enough to allow the UBA domain to reach the thioester-bound ubiquitin thereby either aiding or interfering with thioester formation or polyubiquitin chain formation. However, our data show that the UBA domain does not interfere with disulfide formation. Furthermore, the affinity of the UBA domain for ubiquitin is not affected by the presence of a covalently attached ubiquitin at the catalytic site. This indicates that any interaction between the thioester-bound ubiquitin and the UBA domain is either very weak or does not occur. It is possible that the UBA domain may perturb the chain elongation step by potentially interacting with the growing thioester-bound polyubiquitin chain upon assembly of longer polyubiquitin chains at the thioester or on Lys 93 . A related class II E2 protein, E2-25K, has a flexible two-domain structure similar to Ubc1 and does not autoubiquitinate itself. Instead unanchored polyubiquitin chains are detected in vitro (14). Deletion of the UBA domain in E2-25K results in termination of polyubiquitin chain formation, although ubiquitin-E2 thioester formation proceeds (36). Both of these observations are consistent with our observations that the UBA domain in Ubc1 does not affect initial thioester formation and must be involved in construction of polyubiquitin chains. Further structural characterization of the ubiquitin-Ubc1 disulfide should provide some insight into how the UBA domain influences polyubiquitin chain assembly by Ubc1 and other related class II E2 proteins.
From a mechanistic perspective, the ability of Ubc1 to bind two ubiquitin molecules may allow the thioester-bound acceptor ubiquitin (Gly 76 ) to be in close proximity to the UBA-bound (donor) ubiquitin, which contains the nucleophilic side-chain ⑀-NH 2 of Lys 48 . Recently, mechanisms for the construction of polyubiquitin chains have been proposed in which the acceptor ubiquitin is near the donor ubiquitin in other E2 complexes. Assembly of Lys 63 -linked polyubiquitin chains by the canonical E2 (Ubc13) proceeds through a heterodimer of Ubc13 with an inert E2 variant (Mms2) that non-covalently binds ubiquitin (37). A thioester formed between Ubc13 and ubiquitin accepts a second ubiquitin from Mms2, building up a Lys 63 -linked chain. The Mms2-Ubc13 complex binds the two ubiquitin molecules required for chain building, keeping them in close proximity. A mechanism for multiubiquitin chain synthesis has been proposed for S. cerevisiae Cdc34 (Ubc3), which has been observed to self-associate in vitro. Dimerization of the Ubc3 is driven by formation of the ubiquitin thioester, and this Ub-Ubc3 thioester dimer with or without E3 participation directs polyubiquitin assembly (38). Both of these mechanisms involve association of two ubiquitin molecules by dimerization of an ubiquitin-E2 thioester or an E2 thioester and an inert E2 variant. Similar to the mechanisms discussed above, self-association of a glutathione S-transferase-tagged E2-25K protein lacking its UBA domain initiates assembly of polyubiquitin chains, thus reversing the effect of deleting the UBA domain (36). This may indicate that in E2-25K and Ubc1 the UBA domain and ubiquitin binding could facilitate dimerization of these E2 proteins in a manner similar to Ubc13-Mms2 or Ubc3. Although dimerization of Ubc1 has not been observed (21) our design of the ubiquitin-E2 disulfide thioester mimic is well positioned to probe the potential association of the Ub-Ubc1 intermediate and provide further information on polyubiquitin chain building.