RAD6 gene product of Saccharomyces cerevisiae requires a putative ubiquitin protein ligase (E3) for the ubiquitination of certain proteins.

The RAD6 (UBC2) gene of Saccharomyces cerevisiae which is involved in DNA repair, induced mutagenesis, and sporulation, encodes a ubiquitin-conjugating enzyme (E2). Since the RAD6 gene product can transfer ubiquitin directly to histones in vitro without the participation of a ubiquitin protein ligase (E3), it has been suggested that in vivo it also acts by the unassisted conjugation of ubiquitin to histones or to other target proteins. Here we show that the RAD6 protein can ligate ubiquitin in vitro to a hitherto unknown set of exogenous target proteins (alpha-, beta-, and kappa-casein and beta-lactoglobulin) when supplemented by a putative ubiquitin protein ligase (E3-R) from S. cerevisiae. RAD6 supplemented with E3-R ligates 1 or, sometimes, 2 ubiquitin molecules to the target protein molecule. UBC3 (CDC34) protein in the presence of E3-R has barely detectable activity on the non-histone substrates. Other ubiquitin-conjugating enzymes tested (products of the UBC1 and UBC4 genes) do not cooperate with E3-R in conjugating ubiquitin to the same substrates. Thus, E3-R apparently interacts selectively with RAD6 protein. These findings suggest that some of the in vivo activities of the RAD6 gene may involve E3-R.

The RAD6 (UBCZ) gene plays a key role in one of the three DNA repair epistasis groups of Saccharomyces cerevisiae (reviewed by Friedberg, 1988). Mutants of this gene are extremely sensitive to UV light, x-rays, and alkylating agents and are impaired in meiotic recombination. Diploid, homozygous rad6 mutants are also defective in sporulation. Another recently documented effect of the rad6 mutation is the stimulation of retrotransposition (Picologlou et al., 1990). These observations suggest that the RAD6 gene product performs multiple functions in the cell. The discovery that the RAD6 gene product encodes a ubiquitin-conjugating enzyme (EZ) provides an important clue to its mechanism of action (Jentsch et al., 1987). Since mutation of the ubiquitin acceptor site (CysM) of RAD6 abolishes its biological functions, it is likely that all of its roles depend on its ubiquitin-conjugating activity (Sung et al., 1990).
The covalent attachment of ubiquitin to protein targets is involved in a variety of functions in eukaryotic cells, including * This work was supported by grants from the United States-Israel Binational Science Foundation (BSF), Jerusalem and the Basic Research Foundation administered by the Israel Academy of Sciences and Humanities. 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 U.S.C.
Section 1734 solely to indicate this fact. intracellular proteolysis, heat shock response, cell cycle control, and DNA repair (see Ciechanover and Schwartz, 1989;Hershko, 1988;Rechsteiner, 1988 for reviews). Ligation of ubiquitin to cellular proteins is catalyzed by three enzymatic steps. In the first step, ubiquitin is linked via a thioester bond to the ubiquitin activating enzyme El with the splitting of a molecule of ATP. The second step is the transfer of ubiquitin to the thiol group of one of a family of ubiquitin-conjugating enzymes (Ez) (reviewed by Jentsch et al., 1990). The third step involves one of two alternative mechanisms. 1) Some of the Ez enzymes (such as RAD6 and CDC34 (UBC3) gene products) are able to ligate ubiquitin to target proteins without the participation of additional factors (Pickart and Rose, 1985;Jentsch et al., 1987;Goebl et al., 1988;Haas and Bright, 1988). 2) The conjugation of multiubiquitin chains by Ezs to certain target proteins destined for degradation requires the intervention of an additional class of enzyme known as ubiquitin-protein ligase or E3 (Reiss and Hershko, 1990;Heller and Hershko, 1990;Bartel et al., 1990).
Since the RAD6 (UBCZ) gene product can transfer ubiquitin directly to histones in vitro without the participation of an E3, it has been suggested that in vivo it also acts by the unassisted conjugation of ubiquitin to histones or to other target proteins (Jentsch et al., 1987;Sung et al., 1988). Here we show that the RAD6 protein can ligate ubiquitin in vitro to a hitherto unknown set of exogenous target proteins when supplemented by a factor (E3-R) from S. cerevisiae which may be a ubiquitin protein ligase (E3).

Ubiquitin System Enzymes
Preparation of Ubiquitin-Sepharose Column Fractions-Extracts of late log-phase cells were prepared essentially as described by Jentsch et al. (1987) except that the cell disruption buffer contained 1589 1 50 mM Tris-HC1, pH 7.5, 1 mM DTT' (dithiothreitol), 5 mM EDTA, and 0.5 mM PMSF. Fraction I1 was prepared and chromatographed on ubiquitin-Sepharose columns according to Hershko et al. (1983) with slight modifications (Jentsch et al., 1987). The KC1 wash contained most of the E3 and some E1 and E' . The DTT eluate contained most of the El and E' . The pH 9 eluate contained some E 2 and E,.

PartMlly Purified Ea-R Was Prepared from Commercial Yeast-
Details of the procedure and additional properties of the E 3 fraction will be published elsewhere.' Briefly, the yeast cake was washed four times with water, and the cells were broken in 25 mM Tris-HC1, pH 7.5, 1 mM DTT, 0.5 mM PMSF, and 5 mM EDTA with a Mautcu-Gaulin homogenizer followed by centrifugation at 25,000 X g. Fraction I1 was prepared by chromatography on DEAE-cellulose (Whatman DE52) as described previously (Hershko et al., 1983). A fraction precipitating between 30 and 90% ammonium sulfate was further purified by ubiquitin-Sepharose affinity chromatography as described above. The KC1 wash from the ubiquitin-Sepharose column was applied to a phenyl-Sepharose column in a buffer containing 50 mM phosphate buffer, pH 7.0, and 300 mM ammonium sulfate. The column was washed with decreasing concentrations of ammonium sulfate in the same buffer. E,-R was eluted with phosphate buffer, pH 7.0, in a range of 15 to 5 mM. This preparation was further purified by fast protein liquid chromatography on a Mono-Q column, and the peak fractions with activity on a-and @-casein and @lactoglobulin eluting between 300 and 340 mM KC1 were pooled. The E3-R used in this investigation when subjected to SDS-polyacrylamide gel electrophoresis followed by silver stain revealed approximately 15 protein bands. 1 microunit of E3-R is defined as the amount of enzyme that conjugates 1 pmol of ubiquitin to a-casein per min.
El was prepared from the SUB325 strain. After preparation of extracts as described above, DE52 and ubiquitin-Sepharose column chromatography was performed as described by Jentsch et al. (1987). The El in the DTT eluate was further purified by fast protein liquid chromatography on a Mono-Q column as described by Sung et al. (1988).
The vectors were propagated in the Escherichia coli strain JM105 (Yanisch-Perron et al., 1985). Midlog phase cells in LB medium plus ampicillin were induced by incubation with 0.5 mM isopropyl-@-Dgalactopyranoside for 4 h. Extracts were prepared by freezing and thawing the cells eight times in buffer containing 50 mM Tris-HC1, pH 7.5, 1 mM DTT, 1 mM EDTA, and 0.25 mM PMSF and centrifuging at 100,000 X g for 30 min.
The activities on histone H2B of extracts of bacteria expressing the RAD6 and CDC34 proteins did not increase linearly with extract concentration. This indicated that some components of the E. coli extract inhibited E' activity. Extracts were, therefore, fractionated to remove inhibiting factors as follows. The preparations were loaded on a Whatman DE52 column equilibrated with 3 mM phosphate buffer, pH 7.0, 1 mM DTT, and 0.25 mM PMSF. Proteins were eluted with successive steps of 0.75 column volume of 20 mM Tris-HC1 buffer, pH 7.2, containing 1 mM DTT, 0.25 mM PMSF, and, respectively, 0.2, 0.25, 0.3, 0.35, 0.40, and 0.45 M KCl. Fractions were concentrated 5-fold on a Centricon 10 ultrafilter. RAD6 and CDC34 protein fractions eluting between 0.30 and 0.35 M KC1 were used. These were essentially free of inhibitory activity. Fractions containing UBCl protein eluting between 0.25 and 0.30 M KC1 and UBC4 protein eluting between 0.20 and 0.25 M KC1 were used.
Amounts of E, in the fraction used were determined as follows. adding SDS electrophoresis buffer without mercaptoethanol, and the samples were run on SDS-polyacrylamide gels without boiling. Controls were stopped with sample buffer containing mercaptoethanol and heated. Under the conditions of the assay, all the counts/min in the band corresponding to the ubiquitinated E' were labile indicating that the band is the "51-~biq~itin-E2 thioester. These countslmin were used to calculate the amount of E' in the preparation.

Ubiguitin Conjugation Assay
HCl, pH 7.6, 5 mM MgCl', 2 mM DTT, 2 mM ATP, 10 mM creatine The assay mixture contained in a volume of 12.5 pl: 50 mM Trisphosphate, 2.5 pg of creatine kinase, 1.5 pmol of lZ5I-ubiquitin (2 X lo6 cpm), 10 pg of protein substrate, and 10 pg/ml each of the following peptide protease inhibitors: leupeptin, pepstatin, chymostatin, bestatin, and antipain. Purified El, E', and E3 were added as indicated. Alternatively in experiments with crude ubiquitin-Sepharose fractions, KC1 eluate, DTT eluate, and pH 9 fraction, each containing about 0.5 pg of protein, were added as indicated. After incubation for 1 h at 30 "C, the reaction was stopped with SDS electrophoresis sample buffer and samples were heated for 3 min in a boiling water bath. Samples were subjected to electrophoresis on 18% SDS-polyacrylamide gels (Thomas and Kornberg, 1975) followed by autoradiography.

A Factor (E,-R) Required for the Ubiquitination of Certain
Proteins by the RAD6 Gene Product-We have partially purified a factor from yeast (referred to hereafter as E,-R) which is required, together with El and RAD6 protein (Ez), for the ubiquitination of certain exogenous substrates. Fig. 1 shows the ubiquitination of several protein substrates by various combinations of purified yeast El, E, (RAD6 protein), and E3-R. The ubiquitination of a-, P-, and K-casein and P-lactoglobulin required the presence of E,-R in addition to E, and E?. In contrast, the ligation of ubiquitin to histone H2B required only El and Ea, as has previously been reported (Jentsch et al., 1987;Sung et al., 1988). T h e molecular weights of the ubiquitinated products indicated that a single ubiquitin molecule had been ligated to the substrate molecule. An exception was a-casein which formed a double band apparently corresponding to mono-and diubiquitinated products.
In contrast, UBC4 protein supplemented with El had no detectable activity on core histones (not shown).
The activities of RAD6 and CDC34 proteins on substrates requiring E,-R for ubiquitination were markedly different (Fig. 3, A and B). RAD6 protein plus E,-R rapidly ubiquitinated a-and &casein and P-lactoglobulin, to an extent similar t o that of H2B without E,-R (Fig. 3 A ) . CDC34 protein plus E,-R, on the other hand, had barely detectable activity on these substrates under the same conditions ( Fig. 3B). T h e system containing CDC34 protein ubiquitinated endogenous substrates to form high molecular weight (>66 kDa) conjugates, and the process was apparently enhanced by E,-R (Fig.  3B). UBC4 and UBCl proteins had no activity on the three test substrates with or without E,-R (not shown). Thus, E,-R seems to cooperate selectively with RAD6 protein in the ubiquitination of a-and P-casein and P-lactoglobulin. These findings imply that in vivo E,-R participates in the ubiquitin- ation of unknown targets by the RAD6 gene product.

Ubiquitin Conjugation by
Lack of Detectable Activity on the Test Substrates in a rad6 Null Mutant- Fig. 4 shows that substrates requiring Es-R were not ubiquitinated by ubiquitin system enzymes prepared from a rad6 null mutant (-). Controls (+) showed that the same substrates were uhiquitinated by ubiquitin system enzymes from an isogenic RAD6 strain. When RAD6 prot.ein was added to rad6 mutant extracts, strong ubiquitination of all the test substrates was observed (not shown). Thus, Es-R was present in the rad6 mutant extract but was not detectable unless exogenous RAD6 protein was added. This indicates that En-R has a preference for the RAD6 gene product and does not interact detectably with other major E? species present in growing cells. Activity on the Test Substrates in a ubrl Null Mutant-The question arose whether Ea-R is identical with the product of the URRI gene which encodes the Es involved in "N-end rule" selection (Rartel et a/., 1990). Three of the substrates requiring E:,-R have N-terminal amino acids which are labilizing according to the N-end rule (0-and /&casein, Lys,Arg; @-lactoglobulin, Leu (Eigel et a/., 1984)) and therefore could be targets of the URRI gene product (Rachmair and Varshavsky, 1989;Rachmair et a/., 1986). K-Casein has a blocked Nterminal (pyroglutamic acid (Eigel et a/., 1984)) and therefore is unlikely to be a URRI substrate. In order to test if some or all of the Es activity could be attributed to the URRI gene product, we examined the ubiquitination of substrates requiring Ea-R in extracts of a ubrl null mutant (Fig. 5 ) . Activity on all four substrates was found in this mutant indicating that Ex-R does not correspond with the URRl protein. has been prepared from the uhrl null mutant. providing further support for this conclusion.

DISCUSSION
The above experiments show that the E ? enzyme encoded by the RAD6 gene can act in two different modes: with or without an additional factor, &R. Without E,'.,-R, it can transfer ubiquitin directly to histones, A S was earlier reported (Jentsch et a/., 1987;Sung ~t al., 1988). In the presence of E , -R, it can transfer ubiquitin to a broader spectrum of target RAD6 -+ -+ -+ -+ -. I  strain. Lanes 1 , 3, 5, 7, 9, fractions from rad6 null mutant; lanes 2, 4, 6, 8, 10, fractions from RAD6 strain. Substrates: none 1 and 2), a-casein (lanes 3 and 4 ) , @-lactoglobulin (lanes 5 and 6 ) , K-casein (lanes 7 and 8), @-casein (lanes 9 and I O ) . proteins than without it. E,-R resembles the ubiquitin-protein ligases (E3s) described previously in that it facilitates the transfer of ubiquitin from an E2 to specific target proteins (Reiss and Heller and Hershko, 1990). Unlike the known E3s, it usually ligates 1, or sometimes 2, ubiquitin molecules to the target molecules tested rather than a multiubiquitin chain. It remains to be determined whether E3-R plays a similar role in target protein selection to the previously studied E3s. The presence of E3-R activity in a ubrl null mutant shows that this activity is not due to the known E3 of yeast encoded by the UBRl gene (Bartel et al., 1990). Since the E3-R preparation used in these experiments was not homogeneous, it is possible that it contained more than one type of E3 molecule with different specificities. This should become clear upon further purification.

20-
The question arises whether E3-R, like the E3s described previously, is involved in protein degradation (Hershko, 1988). This is uncertain for the following reasons. E3-R apparently does not interact with the UBCl and UBC4 gene products which have been clearly implicated in protein break-down Seufert and Jentsch, 1990). Also, the attachment of multiubiquitin chains has been shown to be a prerequisite for the degradation of some proteins (Chau et al., 1989), whereas E3-R produces monoubiquitinated proteins ( Figs. 1 and 3A). However, the existence of an unknown pathway degrading specific monoubiquitinated proteins cannot at present be excluded. The most interesting feature of E3-R is that it seems to cooperate preferentially with the RAD6 protein. The specificity of the interaction is emphasized by the fact that even the product of the CDC34 gene, which is closely related to RAD6 (Goebl et al., 1988), collaborates only weakly with E,-R in the ubiquitination of the non-histone substrates. No activity of UBCl and UBC4 gene products on the model substrates was detected in the presence of E3-R. Since so far only four of the seven cloned UBC gene products have been tested with E3-R in vitro, other E2s not tested so far may interact with it. However, the absence of activity on E,-R substrates in a rad6 null mutant (Fig. 4) suggests that none of the major E2 species present in growing cells cooperates with E3-R.
A reasonable assumption is that E3-R is involved in the function of the RAD6 gene which plays a key role in one of the DNA repair pathways of yeast (Friedberg, 1988). rad6 mutants, in addition to being extremely sensitive to UV light, x-rays, and chemical mutagens have low induced mutagenesis, increased frequencies of retrotransposition, and are defective in sporulation. It has been suggested that the multiple activities of the RAD6 gene product are related to its known in vitro activity, the direct transfer of ubiquitin to bovine histones (Jentsch et al., 1987Sung et al., 1988). However, ubiquitinated histones have not been detected in S. cereuisiae (Swerdlow et al., 1990). Moreover, genetic modifications of yeast histones H2A and H2B in a manner which would prevent ubiquitination at the sites ubiquitinated in higher eukaryotes had no obvious effects on cellular function (Swerdlow et al., 1990;Schuster et al., 1986). Therefore, there is no good evidence to support the notion that histones are in vivo targets of RAD6 Our experiments show that RAD6 protein can function in vitro both with or without E3-R and that the involvement of E3-R can extend the spectrum of its target proteins. It is, thus, tempting to suggest that, although some RAD6 protein functions may involve a direct transfer of ubiquitin to target molecules, others may require the mediation of E,-R or similar factors. This assumption could explain the pleiotropic nature of rud6 mutations as well as the apparent division of RAD6 epistasis group genes into subtypes (Friedberg, 1988). It is, therefore, possible that known or unknown genes of the RAD6 epistasis group encode factors such as E3-R which modify RAD6 target specificity, while others encode corresponding target proteins.
Much attention has been given to the highly acidic Cterminal region of the RAD6 protein (Reynolds et al., 1985). This acidic tail has been found to be required for optimal ubiquitination activity on histones in vitro (Sung et aL, 1988). Since S. cerevisiae mutants lacking all 23 amino acids of this C-terminal have impaired sporulation but are capable of normal DNA repair, it was suggested that sporulation requires histone ubiquitination whereas other functions involve the ubiquitination of target proteins not requiring the RAD6 acidic tail (Morrison et al., 1988). A possible explanation of these results, suggested by our findings, is that sporulation involves the direct ubiquitination of target proteins whereas other RAD6 activities are mediated by E3-R or similar factors. Another possibility is that all the RAD6 functions require factors of the E3-R type, with a different specificity for each function. The last hypothesis might explain why the product of the Schizosaccharomyces pombe rhp6+ gene, which corresponds with the S. cereuisiae RAD6 gene, lacks a polyacidic C-terminal but functions in both DNA repair and sporulation (Reynolds et al., 1990). Hopefully, further characterization of the E3s interacting with the RAD6 gene product will help to clarify some of the questions raised above.