Ubiquitin binding site of the ubiquitin E2 variant (UEV) protein Mms2 is required for DNA damage tolerance in the yeast RAD6 pathway.

Different ubiquitin modifications to proliferating cell nuclear antigen (PCNA) signal distinct modes of lesion bypass in the RAD6 pathway of DNA damage tolerance. The modification of PCNA with monoubiquitin signals an error-prone bypass, whereas the extension of this modification into a Lys-63-linked polyubiquitin chain promotes error-free bypass. Chain formation is catalyzed by the Mms2/Ubc13 conjugating enzyme variant/conjugating enzyme (UEV.E2) complex together with the Rad5 ubiquitin ligase. In vitro studies of this UEV.E2 complex have identified a ubiquitin binding site that is mainly localized on Mms2. However, the role of this site in DNA damage tolerance and the molecular features of the ubiquitin/Mms2 interaction are poorly understood. Here we identify two molecular determinants, the side chains of Mms2-Ile-57 and ubiquitin-Ile-44, that are required for chain assembly in vitro and error-free lesion bypass in vivo. Mutating either of these side chains to alanine elicits a severe 10-20-fold inhibition of chain synthesis that is caused by compromised binding of the acceptor ubiquitin to Mms2. These results suggest that the ubiquitin binding site of Mms2 is necessary for error-free lesion bypass in the RAD6 pathway and provide new insights into ubiquitin recognition by UEV proteins.

Ub 1 is a highly conserved, 76-amino acid protein that regulates multiple intracellular pathways. The functions of Ub in cell cycle progression, induction of the inflammatory response, antigen presentation, protein trafficking, and other processes all rely on Ub conjugation to lysine residues of target proteins through the carboxyl group of Ub-Gly-76 (1)(2)(3). This modification frequently results in degradation of the target protein by 26 S proteasomes (4), but ubiquitination can also lead to other functional outcomes (1)(2)(3).
Ub conjugation involves the sequential action of activating (E1), conjugating (E2), and ligase (E3) enzymes (2). The E1 and E2 enzymes both form thiol ester adducts with Ub and both intermediates are strictly required for substrate ubiquitination. Most organisms possess a single E1, multiple E2s, and even more E3s. Individual E3s, many of which harbor a zincbinding RING domain, act in conjunction with specific E2s to recognize individual substrate proteins thus imparting specificity in ubiquitination (2,5).
Substrates can be modified with one Ub, several single Ubs, or multiple Ubs in the form of an isopeptide-linked polyUb chain. These various modifications, especially mono-versus polyubiquitination, often lead to qualitatively different outcomes (2,3). Further variation in Ub signaling can result from the use of different lysine residues in polyUb chain assembly (6). For example, Lys-48-linked chains signal targeting to proteasomes (7,8), whereas Lys-63-linked chains constitute a nondegradative signal in several pathways, including ribosomal protein synthesis (9), kinase activation (1), and membrane protein trafficking (3,10). Mechanisms governing selectivity in the synthesis and recognition of these and other non-canonical polyUb signals are only beginning to be understood (6).
Besides the roles mentioned above, Lys-63-linked polyUb chains are required for a DNA damage tolerance pathway that has been most thoroughly characterized in the budding yeast Saccharomyces cerevisiae. The components of the RAD6 pathway are highly conserved from yeast to humans (11)(12)(13)(14); their actions promote the bypass of DNA photo-or alkylation adducts that would otherwise block the progression of the replicative DNA polymerase. All events in this pathway depend on the Ub-conjugating activity of the Rad6⅐Rad18 (E2⅐E3) complex (14 -16), but the pathway subsequently divides into two branches. One branch, defined by MMS2, UBC13, and RAD5, mediates error-free bypass of DNA lesions (17)(18)(19)(20)(21), possibly through the recruitment of an undamaged template strand (12,13). The other subpathway, defined by REV3 and REV7, facilitates mutagenic bypass through use of the damaged strand as a template for a translesion polymerase (12,13). Lys-63-linked polyUb chains signal selectively in the error-free branch of the RAD6 pathway (11,18,22).
Recent studies have identified the essential DNA polymerase processivity factor PCNA as a major target of Ub modifications during DNA lesion bypass (22)(23)(24). Monoubiquitination of PCNA Lys-164, catalyzed by the Rad6⅐Rad18 complex, signals error-prone repair (22,23), probably by promoting the recruitment of a translesion polymerase (25,26). This same Ub modification also serves as the substrate for extension of a Lys-63linked polyUb chain, which promotes error-free lesion bypass (22). Chain formation requires a ternary complex composed of a RING E3, Rad5, and the heterodimeric Mms2⅐Ubc13 complex (18,19,22). Ubc13 is a canonical E2 enzyme, whereas Mms2 is one of a small family of UEV proteins, which resemble E2s but lack the defining E2 active site cysteine residue (27). The Mms2⅐Ubc13 complex functions as an E2 that is specialized for the assembly of Lys-63-linked polyUb chains (18,28). Interestingly, the same lysine residue of PCNA that receives Ub modifications (Lys-164) is also a target for small ubiquitin-like modifier modification (22,23). The functional consequences of sumoylation remain uncertain.
It is not yet known how the individual biochemical reactions necessary for chain extension from monoubiquitinated PCNA (or other potential targets of Lys-63-polyubiquitination) are apportioned among the three proteins in the Rad5⅐Mms2⅐Ubc13 complex. The binary Mms2⅐Ubc13 complex synthesizes unanchored Lys-63-linked polyUb chains from free Ub in vitro (18,29). Structure-activity studies and molecular modeling identified a likely binding site for the "acceptor" Ub on the Mms2⅐Ubc13 complex (30), where the acceptor Ub is defined as the molecule that donates Lys-63 to the nascent isopeptide bond. Recent solution structural studies carried out by Ellison and co-workers (31,32) have confirmed that Ub indeed binds to this site in a manner similar to that originally predicted. However, the finding that Rad5 is strictly required for PCNA polyubiquitination in vivo (22) challenges the relevance of this site in PCNA polyubiquitination. Here we show that the side chains of Mms2-Ile-57 and Ub-Ile-44 are required for Ub binding to the acceptor site of the Mms2⅐Ubc13 complex, for chain assembly in vitro, and for DNA damage tolerance in vivo. These results strongly suggest that the acceptor Ub binding site is necessary for chain assembly in vivo.
Plasmids and Proteins-Existing centromeric URA3-marked yeast plasmids (18,30) encoding H 10 -Mms2 or Ubc13-HA 3 (the latter including the intron of UBC13) under the control of the respective endogenous promoters were mutated by standard PCR methods. All mutated open reading frames (see also below) were sequenced in their entireties. Sequences of primers used in mutagenesis and cloning are available upon request. URA3-marked 2 plasmids encoding wild type Ub or Ub-I44A were provided by L. Hicke (Northwestern University) (34).
Yeast Extracts and Western Blotting-Cells (2 optical density units) from log-phase cultures of SUB413 (with or without the Ub-I44Aexpressing plasmid) were pelleted, washed, and then sonicated in 90 l of alkaline extraction buffer as previously described (40). Aliquots were analyzed by Western blotting with affinity-purified polyclonal antibodies directed against Ub (produced by these authors). Blots were developed colorimetrically by means of an alkaline phosphatase-conjugated secondary antibody (Bio-Rad).
Assays-UV sensitivity was assayed in triplicate as previously described (18). Chain assembly was assayed at pH 7.6 and 37°C and monitored by SDS-PAGE (29,30). The standard steady-state assay (detection by Coomassie staining) contained 0.1 M E1, 2 M each Ubc13 and Mms2, and 117 M Ub. To determine initial rates of Ub 2 synthesis, Ub 74 or Ub-Asp-77 (117 M) was used as the acceptor in conjunction with full-length 125 I-Ub (2-6 M). Reactions were terminated during the linear phase of 125 I-Ub 2 synthesis, and this product was quantitated (following autoradiography) by band excision/counting or phosphorimage analysis (18). In some cases enzyme concentrations were adjusted, and/or modified forms of Ub were used, as described in the figure legends. Published procedures were used to produce Lys-63linked Ub 4 carrying the K63R mutation in its distal Ub and an extra residue (Asp-77) in its proximal Ub (29); this chain is chemically inert in the Mms2/Ubc13-catalyzed reaction. The integrity of the Mms2/ Ubc13 interaction was assayed by monitoring the binding of untagged Ubc13 to H 10 -Mms2 immobilized on nickel beads as in our previous work (30). Expression levels of H 10 -Mms2 and Ubc13-HA 3 in yeast cell extracts were monitored by Western blotting with antibodies (Santa Cruz) directed against the His 10 tag (Sc-803) and the hemagglutinin tag (Sc-805), respectively.

A Model for Ub Interaction with the Mms2⅐Ubc13 Complex-
Structural and modeling studies have identified two Ub binding sites on the Mms2/Ubc13 heterodimer (30,32). One of these sites ( Fig. 1) interacts with the Ub molecule bound to the active site cysteine residue of Ubc13. We define this site, which is entirely located on Ubc13, as the "donor" site. The other site mainly encompasses a concave lower surface of Mms2 (as oriented in Fig. 1). This site binds the Ub that provides Lys-63 to the isopeptide bond. In the discussion below we refer to this as the "acceptor" site. We proposed the existence of donor and acceptor sites based on the results of modeling and site-specific mutagenesis (30). The donor site is a universal feature of E2 enzymes (32,41), but so far, the acceptor site is unique to the Mms2-containing heterodimer. Solution structural studies of Ellison and co-workers (32) have shown that Ub indeed binds non-covalently to the acceptor site, which is primarily located on Mms2. An Mms2-interacting Residue of the Acceptor Ub (Ile-44) Is Important for Chain Synthesis in Vitro and DNA Damage Tolerance in Vivo-Our goal in the present work was to test the involvement of the acceptor Ub site in the biochemical and biological activities of the Mms2⅐Ubc13 complex. To identify residues of Ub that are important for its binding to this site, we began with three residues that comprise a hydrophobic patch on the surface of Ub. The side chains of Leu-8, Ile-44, and Val-70 are important for the recognition of Lys-48-linked polyUb chains by proteasomes and for several other Ub-dependent processes (34,42,43). In addition, NMR results indicate that the surface encompassing these residues contacts the acceptor Ub binding site of the human Mms2⅐Ubc13 complex (31). To test the functional significance of such contacts, each of the three residues was individually mutated to alanine, and the purified mutant proteins were screened in steady-state chain assembly assays with the yeast Mms2⅐Ubc13 complex. As shown in Fig. 2A, chain synthesis with the L8A and V70A mutants proceeded at a similar rate to the reaction with wild type Ub (lanes 7-14 versus 1-3; the anomalous migration of Ub-L8A seen in lanes 7-10 is frequently observed). In contrast, the Ub-I44A mutation caused a profound defect in chain synthesis ( Fig. 2A, lanes 4 -6).
Because the Ub-I44A mutation is lethal (34), we used an indirect approach to ask whether the in vitro defect conferred by this mutation correlated with a deficiency in error-free DNA lesion bypass. A plasmid specifying wild type Ub or Ub-I44A was transformed into a yeast strain engineered to express Ub-K63R as the sole form of Ub (11). The parent strain, called SUB413, is UV light sensitive because of a specific defect in error-free lesion bypass that is due to the inability of Ub-K63R to be assembled into Lys-63-linked chains (Fig. 2B, open circles) (11,18). As expected, expression of wild type Ub from a control plasmid complemented this phenotype (filled circles). Note that Ub-K63R does not act as a dominant chain terminator, presumably due to rapid disassembly and re-synthesis of chains within cells (8). In contrast, expression of Ub-I44A in SUB413 failed to complement the UV light-sensitive phenotype of this strain (Fig. 2B, filled squares), despite robust expression as evidenced by increased levels of free and conjugated Ub (Fig. 2C). We postulate that the inability of Ub-I44A to promote DNA damage tolerance reflects the inability of this mutant to be used as a substrate for Mms2/Ubc13-catalyzed synthesis of Lys-63linked chains ( Fig. 2A).
Polar Effect of Ub-I44A Mutation-We next sought to determine whether Ub-I44A was defective in binding to one or to both of the Ub binding sites in the Mms2⅐Ubc13 complex. To address this question we created two versions of Ub-I44A, each of which could perform only one role (i.e. donor or acceptor). Blocking the C terminus of Ub via biosynthetic introduction of an extra residue (Asp-77) prevents activation of Gly-76; this blocks the donor function of Ub (35). To ablate the acceptor function of Ub we blocked Lys-63 (and other lysines) through reductive methylation (39,44). Prior studies with several conjugation systems have shown that the presence of Asp-77 does not inhibit the acceptor activity of Ub and that reductively methylated-Ub 76 is functional as a donor (18,35,39,44). As expected, the Ub-Asp-77 and reductively methylated-Ub 76 proteins carrying the (wild type) isoleucine side chain at position 44 were efficiently conjugated to each other by the Mms2⅐Ubc13 complex to generate Ub 2 (Fig. 3A, lanes 1-4). Introducing the I44A mutation into the Ub-Asp-77 acceptor virtually eliminated Ub 2 synthesis (Fig. 3A, lanes 5-8), whereas introducing this mutation into the reductively methylated-Ub 76 donor did not detectably inhibit chain formation (lanes 9 -12). Thus, the Ub-I44A mutation has a polar effect on the Mms2/Ubc13-catalyzed reaction; it profoundly inhibits acceptor Ub functionality but has no effect when Ub is bound at the donor site. (Although we cited this result in an earlier publication (30), these data have not been presented previously.) Quantitative reactions with a radiolabeled donor Ub confirmed these conclusions. Introducing the I44A mutation into the acceptor Ub inhibited by 90 -95%, almost equivalent to the effect of removing the reaction site altogether through introduction of the Ub-K63R mutation (Fig. 3B, open circles and squares versus filled circles). In marked contrast, introducing the I44A mutation into the labeled donor Ub had a negligible effect (Յ20% inhibition, data not shown).
Competition experiments were used to confirm that the I44A Error bars that cannot be seen are smaller than the diameter of the symbol. C, Ub-I44A is strongly expressed (anti-Ub Western blot). Extracts from log-phase SUB413 cells transformed with empty vector (lanes 1 and 3) or the Ub-I44A-expressing plasmid (lanes 2 and 4) were analyzed (lanes 1 and 2, 2 l of extract; lanes 3 and 4, 4 l of extract). Migration of molecular mass standards is shown at the left. Coomassie staining of duplicate lanes confirmed equal loading of the sample pairs (not shown). Ub 1 and Ub n refer to mono-Ub and Ub-conjugated proteins (the latter including free polyUb chains), respectively. mutation inhibited acceptor Ub binding and not a downstream catalytic step. As shown by the filled circles in Fig. 3C, mono-Ub carrying the K63R,Asp-77 double mutation inhibits Ub 2 synthesis in a concentration-dependent manner as a result of the binding of this (chemically inert) mutant to the acceptor site. The value of K i(app) ϳ100 -200 M is consistent with previously estimated K d values of 150 -400 M for acceptor Ub binding to this (yeast) complex (29). A stronger interaction, K d ϳ30 M, has been reported for the acceptor site of human Mms2 (45). When Ub-I44A,D77 was used as a competitor, inhibition was greatly reduced (Fig. 3C, open circles). The data indicate an increase of Ն4-fold in K i(app) , although the inhibition is too weak to allow a reliable determination of the new value. The Ub-I44A competitor carries a lysine side chain at position 63. If labeled Ub were transferred to this site, inhibition would be underestimated. However, data shown in Fig. 3B indicate that there is negligible transfer to Ub-I44A bound at the acceptor site.
Taken together, the data presented in Fig. 3 indicate that the Ub-I44A mutation selectively compromises the interaction of Ub with the acceptor site of the Mms2⅐Ubc13 complex. These results reinforce the notion that the inability of Ub-I44A to support error-free bypass in vivo (Fig. 2) reflects compromised binding to this site. Our standard catalytic assays were done at a Ub concentration (117 M) that is below saturation for the acceptor site. Thus, the ability of Ub-L8A and Ub-V70A to support chain synthesis ( Fig. 2A) suggests that unlike Ile-44, the Leu-8 and Val-70 side chains do not make important contributions to the interaction of Ub with the acceptor site of yeast Mms2. (Note that the donor site is saturated in this assay, because its occupancy is mainly determined by the upstream interaction of Ub with E1, which follows K m ϳ1-2 M for wild type Ub (46,47)).
Role of Mms2-Ile-57 in Acceptor Ub Binding-Based on our model for acceptor Ub binding to the Mms2⅐Ubc13 complex, it appeared possible that Ub-Ile-44 could contact Mms2 near the side chain of Mms2-Ile-57 (Fig. 1). As a first test of this model, we introduced the I57A mutation into Mms2 and compared the purified mutant protein to wild type Mms2 in a steady-state chain synthesis assay. The I57A mutant displayed strongly reduced activity (Fig. 4A). Quantitative assays with 125 I-labeled donor Ub showed that the initial rate of Ub 2 synthesis was reduced by 15-fold (data not shown). Control experiments showed that wild type Mms2 and the I57A mutant bound with similar efficiencies to wild type Ubc13 (data not shown). Therefore the defect in chain synthesis seen with Mms2-I57A cannot be attributed to a weakened interaction with Ubc13, consistent with the large distance between Mms2-Ile-57 and the heterodimer interface ( Fig. 1 (30, 48)).
To test whether the heterodimer containing Mms2-Ile-57 was impaired in its ability to bind the acceptor Ub, we again turned to competition assays. Here we used catalytically inert Lys-63-Ub 4 as the competitor (see "Materials and Methods"). The weak competition seen in assays involving Mms2-I57A (Fig. 4B, open circles), in comparison to the result with wild type Mms2 (filled circles), indicates that Mms2-Ile-57 plays an important role in acceptor Ub binding. The I57A mutation in Mms2 led to a Ն6-fold increase in the value of K i(app) .
Mms2-Ile-57 Is Important for DNA Damage Tolerance-We next tested whether the in vitro defect conferred by the Mms2-I57A mutation (Fig. 4, A and B) translates into compromised DNA damage tolerance in vivo. A low copy yeast plasmid specifying Mms2-I57A under the control of the endogenous MMS2 promoter was transformed into an mms2⌬ strain of S. cerevisiae. As shown in Fig. 4C, expression of Mms2-I57A afforded minimal rescue of the UV-light sensitive phenotype of the mms2⌬ strain (filled squares versus open circles) in comparison to results obtained for wild type Mms2 in the same vector backbone (filled circles). Blotting experiments confirmed that the mutant protein was expressed at a comparable level to wild type Mms2 (data not shown). Thus, Mms2-Ile-57 is important for chain synthesis in vivo.
Mutation of Ubc13-Asp-81 Inhibits DNA Damage Tolerance-Ubc13-Asp-81 is predicted to lie near the boundary of the acceptor site in the Mms2⅐Ubc13 complex ( Fig. 1) (30). We found in earlier work that mutating Ubc13-Asp-81 to arginine or alanine inhibited in vitro chain synthesis completely or severely, respectively (30). To address the importance of Ubc13-Asp-81 in vivo, we expressed each of these mutant Ubc13 proteins in a ubc13⌬ strain. As shown in Fig. 4D, the phenotype of the ubc13-D81R strain (filled squares) was indistinguishable from that of the null strain (open circles), whereas expression of the D81A mutant (filled triangles) provided weak rescue. Both proteins were expressed comparably to wild type Ubc13 expressed in the same vector backbone (data not shown). These phenotypes correlate well with the biochemical effects of the corresponding mutations.

DISCUSSION
The ability of chemically distinct polyUb chains to act as functionally distinct signals can be appreciated by comparing the properties of Lys-48-linked with Lys-63-linked chains. Lys-48-linked chains are the principal signal for targeting to proteasomes (7,8), whereas Lys-63-linked chains are nonproteolytic signals in several different pathways. For example, the modification of TRAF6 with Lys-63-linked chains activates a specific protein kinase upstream of IB␣ kinase, ultimately leading to phosphorylation and degradation of IB␣, translocation of NFB into the nucleus, and the induction of inflammatory responses (1). In the Ub-dependent DNA damage tolerance pathway studied here, the modification of PCNA with a Lys-63-linked chain promotes an errorfree mode of DNA lesion bypass (11,22). Because TRAF6 and PCNA are metabolically stable proteins, Lys-63-linked chains do not elicit target protein degradation in these signaling pathways. Instead of being recognized by proteasomes, the non-canonical TRAF6-linked chains bind to the kinase adaptor proteins TAB2 and TAB3 (49); the receptor of the PCNA-bound chain signal is not yet known. Although the target protein that is modified by the non-canonical chain differs between the two signaling pathways, there are notable similarities in the biochemistry of signal generation. In both cases, a heterodimeric UEV⅐Ubc13 complex collaborates with a RING domain E3 to generate the chain. Here we have identified specific molecular determinants of the UEV⅐Ub interaction in the DNA damage tolerance pathway.
The Ub-based determinant, the Ile-44 side chain, is essential for viability in budding yeast (34). This side chain is important for the recognition of mono-Ub signals in the non-essential process of endocytosis (34) and for the recognition of Lys-48linked chains by proteasomes, which is an essential process (8,43). Ub-Ile-44 is known to make an important contribution to the affinity of mono-Ub for representative members of two families of Ub-interacting elements, the ubiquitin-interacting motif (50,51) and CUE (52)(53)(54) domains. The present results add the UEV protein Mms2 to this list. This conclusion is based on the profound inhibition of Mms2/Ubc13-catalyzed chain synthesis caused by the Ub-I44A mutation ( Fig. 2A), an effect that is specifically because of compromised interaction of this mutant Ub with the acceptor site of the heterodimeric UEV⅐E2 complex (Fig. 3, A and B). Although the weak affinity of wild type mono-Ub for this site of the yeast Mms2⅐Ubc13 complex precluded a direct determination of the affinity of Ub-I44A, the results of competition studies (Fig. 3C) indicate that reduced binding to this site is the principal cause of the catalytic defect. The other major constituents of the hydrophobic patch of Ub do not seem to contribute importantly to the Mms2/Ub interaction, because the individual Ub-L8A and Ub-V70A mutations have no detectable effect on the kinetics of chain synthesis at a subsaturating Ub concentration ( Fig. 2A).
NMR chemical shift perturbation studies indicate that the hydrophobic patch of Ub contacts Mms2 in the human Ub⅐Mms2 complex (31). Our finding that the Ile-44 side chain plays a unique role in mediating the interaction is consistent with these results but represents a significant refinement. Further studies will be necessary to determine whether Ub-Ile-44 also plays a predominant role in the interaction of Ub with human Mms2. The higher affinity of the latter interaction, important energetic contribution of these side chains to binding, as seen in recent studies of ubiquitin-associated domains (55). Indeed, we found that Ub-I44A is fully functional when covalently bound at the donor site in Ubc13 (Fig. 4A) even though NMR data suggest that the Ub-Ile-44 side chain contacts human Ubc13 and yeast Ubc1 in their respective Ub thiol ester complexes (32,41). Ub-Ile-44 is also located at the interface in complexes of mono-Ub with representative NZF and ubiquitin-associated domains (55)(56)(57)(58). To our knowledge, an important energetic contribution of the Ile-44 side chain to binding has not been established in these cases.
Although we have not systematically addressed how affinity depends on chain length, data obtained in the present study suggest that Lys-63-Ub 4 binds only ϳ2-fold more tightly than mono-Ub to the acceptor site ( Fig. 3C versus 4B). This is not surprising given the limited area of this site (Fig. 1). Because Ub-Ile-44 remains exposed in the extended, open conformation of Lys-63-linked Ub 2 (59), each Ub in Lys-63-Ub 4 is likely to be accessible for interaction; this will enhance affinity through a simple concentration effect. These considerations, as well as the positions of the donor/acceptor sites, argue for a distributive (versus processive) mechanism of chain extension by the isolated Mms2/E2 complex (32). Earlier kinetic data are largely in agreement with this model (29). It remains possible that additional relevant conjugation factors (Rad5 and the Rad6⅐Rad18 complex) increase processivity in vivo.
Based on our model for acceptor Ub binding (30), we postulated that Mms2-Ile-57 might be part of the acceptor Ub binding site. The current results confirm this hypothesis. The I57A mutation has no effect on the Mms2/Ubc13 interaction, but it caused a severe inhibition of chain assembly (Fig. 4A) that could be assigned, through competition studies, to reduced binding of Ub at the acceptor site (Fig. 4B). Ile-57 is positioned at the beginning of the third ␤-strand of yeast Mms2, and there is an isoleucine residue at the orthologous position in human Mms2 (30). Although the chemical shift of this amide nitrogen of isoleucine is not perturbed upon Ub binding to human Mms2, the amide nitrogen of the next residue is strongly perturbed (32). Thus, this region of Mms2 is likely to constitute a conserved Ub interaction surface. Mms2-Ile-57 is also conserved in mammalian Uev1A, which plays an Mms2-like role in the cytosolic IB␣ kinase activation pathway (28).
We also mutated a residue of Mms2 that is predicted to be more centrally located in the acceptor site. The Mms2-S33A mutation (see Fig. 1) had no detectable effect on chain synthesis in vitro or UV sensitivity in vivo (data not shown). These results exclude an important contribution of the Ser-33-hydroxyl to acceptor Ub affinity. It remains possible that this side chain helps to generate shape complementarity.
We proposed previously that Ubc13-Asp-81 resides at the border of the acceptor site (Fig. 1), because mutating this residue inhibited catalytic activity severely without impeding Ub occupancy of the donor site (30). Although competition studies revealed detectably weakened binding of the acceptor Ub to the Mms2⅐Ubc13⅐D81A complex (see Fig. 5C of Ref. 30), this effect was modest relative to the effects of the Mms2-I57A and Ub-I44A mutations in the present study (Figs. 3C and 4B). The new results presented here suggest that Ubc13-Asp-81 does not play a major role in determining the affinity of the acceptor Ub. Further work is needed to explain why mutating this residue inhibits catalysis so strongly. Perhaps Ubc13-Asp-81 helps to position Ub-Lys-63 correctly in the Ubc13 active site.
We observed a perfect correlation between in vitro and in vivo effects of the mutations studied here (Figs. 2-4), suggesting that the Mms2/Ubc13-catalyzed reaction contributes to rate limitation in vivo in our experimental system. (This may not be true in all yeast strains (60).) This outcome is most significant in the case of the Mms2-I57A mutation, which has no known effect besides reducing the affinity of the acceptor Ub. Thus, the UV light sensitivity conferred by this mutation (Fig. 4C) strongly argues that the Ub binding site of Mms2 remains relevant when Rad5 participates in chain synthesis. The location of the Rad5 binding site at the top of Ubc13 (as oriented in Fig. 1 (30, 48, 60)), distant from the acceptor site of Mms2, is consistent with this model. The precise role of Rad5 is uncertain, but it is likely that its interaction with the Rad18⅐Rad6 complex (19) helps to guide the Mms2/Ubc13 heterodimer to monoubiquitinated PCNA. Rad18 possesses autonomous affinity for PCNA, and the two proteins associate in mammalian cells (25,26). Such a bivalent interaction with PCNA could help to overcome the weak intrinsic affinity of Ub for the acceptor site of Mms2.
Understanding the principles that enforce linkage specificity in polyUb chain synthesis, recognition, and disassembly is an important challenge, because these principles underlie signaling specificity and in some cases, may guide pharmacologic intervention (61). Here we have identified molecular determinants that are important for binding of the acceptor Ub to the UEV protein Mms2. Our finding that Ub-Ile-44 is an important determinant reinforces the notion that linkage specificity in this particular reaction arises through an indirect mechanism (30). The Ub-Ile-44 determinant is distant from the site of chemical reaction, but contact with this side chain acts to exclude lysine residues other than Lys-63 from the active site of Ubc13. In marked contrast, the small ubiquitin-like modifier-specific conjugating enzyme Ubc9 directly binds side chains immediately adjacent to the target lysine, generating specificity for lysines within a particular consensus sequence (62). Therefore, target lysine specificity in the ligation of Ub family proteins can arise through at least two different mechanisms (63).
Interestingly, the other well characterized UEV protein, Tsg101/Vps23, interacts with mono-Ub through a different UEV surface that is mainly contributed by a unique insertion in the primary structure of Tsg101 (64,65). (Although Ub-Ile-44 is centrally located in the Tsg101/Ub interface, its quantitative contribution to affinity has not been determined.) The outcome of the Ub/UEV interaction is also differs between the two proteins. Tsg101 promotes the entry of monoubiquitinated cargo proteins into multivesicular bodies, whereas Mms2 acts in conjunction with Ubc13 to catalyze a specific chemical reaction (63). Diverse modes of Ub recognition by these and other Ub/polyUb receptors will contribute to the selectivity of Ub-dependent signaling.