Yeast DNA Repair Proteins Rad6 and Rad18 Form a Heterodimer That Has Ubiquitin Conjugating, DNA Binding, and ATP Hydrolytic Activities*

The RAD6 and RAD18 genes of Saccharomyces cerevisiae are required for postreplicative bypass of ultraviolet (UV)-damaged DNA and for UV mutagenesis. The RAD6 encoded protein is a ubiquitin conjugating enzyme, and RAD18 encodes a protein containing a RING finger motif and a nucleotide binding motif. Rad18 can be co-immunoprecipitated with Rad6, indicating that the two proteins exist in a complex in vivo. Here, we co-overproduce the two proteins using a yeast multicopy plasmid, purify the Rad6-Rad18 complex to near homogeneity, and show that the complex is heterodimeric. The Rad6-Rad18 heterodimer has ubiquitin conjugating activity, binds single-stranded DNA, and possesses single-stranded DNA-dependent ATPase activity. The Rad6-Rad18 complex provides the first example wherein a ubiquitin conjugating activity is physically associated with DNA binding and ATPase activities provided by an associated protein factor. The co-existence of these activities should provide the complex with the ability to recognize single-stranded DNA resulting from stalling of the replication machinery at DNA damage sites and to recognize the components of the DNA replication machinery for ubiquitination by Rad6.

Exposure of cells to ultraviolet (UV) light and to many other agents causes the formation of lesions in the DNA. During DNA replication, such lesions located in the template strand block the DNA replication machinery, resulting in a gap in the newly synthesized strand across from the damage site. A variety of postreplicational repair mechanisms have evolved to restore the continuity of the newly synthesized DNA strand (reviewed in Ref. 1).
Genetic studies in the yeast Saccharomyces cerevisiae have been instrumental in identifying the genes involved in postreplicational repair. RAD6 and RAD18, members of the RAD6 epistasis group, play a prominent role in this repair process. Mutations in RAD6 cause extreme sensitivity to UV light and to other DNA damaging agents; rad6 mutants are highly deficient in postreplicational repair of UV-damaged DNA (2) and they exhibit no mutation induction in response to UV (3). RAD6 encodes an ubiquitin conjugating enzyme of 172 residues (4,5). The first 149 amino acids of Rad6 form a globular domain, while the distal 23 residues, which are predominantly acidic, constitute a freely extending tail domain (6). Mutational inactivation of the active site cysteine 88 residue in Rad6 has indicated that the ubiquitin conjugating activity is essential for all the biological functions of Rad6 (7). Mutants of RAD18 resemble those of RAD6 in their high degree of sensitivity to UV, defects in postreplicational repair of UV-damaged DNA (2), and defects in UV mutagenesis (8,9). However, unlike RAD6, which is indispensable for sporulation, mutations in RAD18 do not affect sporulation (10).
Other genes that belong to the RAD6 epistasis group include REV1, REV3, REV7, and RAD5. Although mutants of the REV genes show only a marginal increase in UV sensitivity, like rad6 and rad18 mutants, they are defective in UV mutagenesis (3,11,12). Rev3 and Rev7 together form a DNA polymerase activity (pol) that can bypass a thymine-thymine cis-syn-cyclobutane dimer (13). Mutations in RAD5 enhance UV sensitivity to a greater degree than those in the REV genes; however, the incidence of UV mutagenesis at most loci is not affected (14). From these and other genetic observations, it has been suggested that REV genes and RAD5 function, respectively, in the mutagenic and nonmutagenic modes of RAD6, RAD18-dependent postreplicational repair.
Rad18 can be co-immunoprecipitated with Rad6, indicating physical interaction of the two proteins (15). For delineating the molecular functions of the Rad6-Rad18 complex in postreplicative repair processes, it is essential to purify this complex and to define its biochemical properties. Here, the Rad6-Rad18 complex is purified to near homogeneity from yeast cells genetically tailored to co-overproduce the two proteins. We show that the Rad6-Rad18 complex is heterodimeric and that the Rad6-Rad18 complex has ubiquitin conjugating activity, as well as single-stranded (ss) 1 DNA binding and ssDNA-dependent ATPase activities.

Plasmid Construct for Overexpressing Rad6 and Rad18 Proteins-
The DNA fragment with the ADC1 promoter and the RAD6 gene from pSCW242 (5) was cloned in a derivative of a 2-m multicopy vector, which contains the GAL1 promoter and the URA3 gene. A DNA fragment containing the RAD18 gene from the ATG initiation codon to nucleotide 342 after the termination codon was cloned under the GAL1 promoter in the vector containing the ADC1-RAD6 insert. The plasmid obtained, pR18.36, is shown in Fig. 1A.
To prepare cell extract, frozen yeast cells (60 gm) were thawed by stirring in 50 mM Tris/HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 10 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstatin, leupeptin, chymostatin, and aprotinin (3 ml of buffer/g of cells) and lysed by passage through a French press at 16,000 p.s.i. The supernatant obtained by centrifugation at 20,000 ϫ g for 30 min was brought to 30% saturation with ammonium sulfate. A second ammonium sulfate precipitation was carried out by bringing the supernatant recovered by high speed centrifugation (100,000 ϫ g for 1 h) to 50% saturation with ammonium sulfate. The pellet was kept on ice overnight and then dissolved in 50 mM Tris/HCl, pH 7.5, 1 mM EDTA, 5 mM DTT, NaCl buffer containing protease inhibitors to give an ionic strength equivalent to that of 0.2 M NaCl in water. The protein solution was applied onto a Q-Sepharose FF (Pharmacia Biotech Inc.) column (60 ml) equilibrated in buffer A containing 0.2 M NaCl (buffer A: 50 mM Tris/HCl, 1 mM EDTA, 2 mM DTT, pH 7.5) and washed with 180 ml of buffer A containing 0.25 M NaCl. The Rad6-Rad18 complex was eluted with buffer A containing 0.4 M NaCl. The sample (30 ml) was applied to a phenyl-Sepharose CL-4B (Pharmacia) column (20 ml) equilibrated with buffer A containing 0.4 M NaCl and washed after sample application with the same buffer. The flow-through and first 20 ml of the wash were pooled (50 ml), and proteins were precipitated with ammonium sulfate to 50% saturation. The pellet was dissolved in buffer A containing 0.15 M NaCl. Insoluble material was removed by centrifugation, and the sample was applied onto a Mono S column (HR5/5; Pharmacia) equilibrated with buffer B containing 0.15 M NaCl (buffer B: 50 mM Tris-HCl, 0.1 mM EDTA, 1 mM DTT, pH 7.5). The column was developed with a 20-ml NaCl gradient from 0.2 M to 0.4 M, and fractions containing the Rad6-Rad18 complex, which elutes at about 0.3 M NaCl, were pooled, diluted twice with buffer C (buffer B containing 10% glycerol), and applied onto a Mono Q column equilibrated with buffer C containing 0.15 M NaCl. Rad6-Rad18 complex was eluted with a 20-ml NaCl gradient from 0.2 M to 0.5 M, and the peak fractions, at about ϳ0.38 M NaCl, were pooled. The concentration of Rad6-Rad18 in the final sample was determined from the absorbance at 280 nm using an extinction coefficient of 56800 M Ϫ1 cm Ϫ1 , calculated according to the formula (number of tryptophan residues ϫ 5700) ϩ (number of tyrosine residues ϫ 1300). Rad6-Rad18 complex obtained from this purification scheme (ϳ5 mg) was stored in small portions at Ϫ70°C.
Purification of Rad6 and Uba1-The EcoRI fragment from YC-pUBA1 (16), which contains the UBA1 gene in the CEN URA3 plasmid YCp50, was cloned into a yeast 2-m multicopy vector, yielding plasmid pPM196. Three chromatographic steps, consisting of Q-Sepharose, ubiquitin-Sepharose, and Mono Q, adapted from (5), were used for purification of Uba1 to apparent homogeneity from yeast strain LY2 carrying plasmid pPM196.
DNA Binding Assays-DNA binding was examined using the nitrocellulose filter binding method. Nitrocellulose filters (Millipore, HAWP) were pretreated in 0.4 M KOH for 40 min, washed extensively in distilled water, and stored in filter buffer (FB) (25 mM HEPES-KOH, pH 7.0, 15 mM KCl). Filters prepared in this manner could be used over a period of at least 1 month. The buffer for DNA binding assays consisted of FB ϩ 50 g ml Ϫ1 bovine serum albumin and 0.5 mM DTT. Poly(dT) (Pharmacia) was dissolved in TE (10 mM Tris/HCl, pH 7.5, 0.5 mM EDTA) and used as a concentrated stock; the average length of the poly(dT) was 221 nucleotides. The DNA was labeled with 32 P by first treating with calf intestinal alkaline phosphatase and then with [␥-32 P]ATP and polynucleotide kinase. 32 P-Labeled DNA was used at a concentration of 2 M (nucleotides); unless otherwise indicated, MgCl 2 was added to 4 mM when present, and ATP was added to 1 mM when present. Reaction mixtures containing the 32 P-labeled DNA and the Rad6-Rad18 complex were incubated for 10 min at 30°C, and two aliquots (40 l) from each sample were filtered separately on different filters, which were washed with 2 ml of FB, dried, and the associated radioactivity quantitated by scintillation counting. Routinely, the two values for each point varied by Ͻ10%. All experiments were repeated at least twice, but representative data sets obtained on the same day are presented.
ATP Hydrolysis Assay-ATP hydrolysis by the Rad6-Rad18 complex was measured using thin-layer chromatography (TLC) on polyethyleneimine-coated plates. The buffer used for the assay consisted of 25 mM HEPES-KOH, pH 7.0, 50 mM KOAc, 4 mM MgOAc, 50 g/ml bovine serum albumin, and 0.5 mM DTT. Unlabeled ATP was added to a concentration of 200 M unless otherwise indicated; 5 Ci of ␥-32 Plabeled ATP (3000 Ci/mmol; Amersham Corp.) were added to each reaction (10 l). DNA and protein concentrations are indicated in the figure legends. Assays were initiated by the addition of protein; aliquots (2 l) were removed at different times and the reaction stopped by the addition of an equal volume of 20 mM EDTA, 2% SDS. A portion of each sample (1 l) was spotted onto a TLC plate that had been prerun in 0.5 M LiCl, 1 M formic acid, dried, and run again in distilled water and dried again. The plate was developed in the above LiCl/formic acid buffer, dried, and the amount of phosphate released quantified in a phosphor-Imager. These values were converted to M ATP hydrolyzed, plotted, and the resulting rates calculated by linear regression. The reported rates represent the average of at least two independent experiments.

Heterodimer of Rad6 and Rad18
Proteins-To obtain cooverexpression of Rad6 and Rad18 proteins in yeast, we constructed plasmid pR18.36 in which the RAD6 gene is fused to the ADC1 promoter and the RAD18 gene is fused to the GAL1 promoter (Fig. 1A). Plasmid pR18.36 results in 10-and 50-fold overproduction of Rad6 and Rad18 proteins, respectively. Rad6 and Rad18 exist in a complex and were co-purified as such to near homogeneity by a combination of ammonium sulfate precipitation and four column chromatographic steps of Q-Sepharose, phenyl-Sepharose, Mono S, and Mono Q (see "Materials and Methods" and Fig. 1B). The Rad6 and Rad18 proteins remain associated through all these purification steps (Fig. 1, C and D), indicating a high degree of stability of the complex. Densitometric scanning of a Coomassie Blue-stained sample of purified Rad6-Rad18 complex suggested that the proteins were present in equimolar amounts. To further confirm the stoichiometry of the complex, a sample of the Mono Q fraction was subjected to reverse phase chromatography on a C-4 column to separate Rad6 and Rad18 proteins, followed by determination of the absorbance of each protein peak at 214 and 280 nm. The ratios of the absorbance of Rad18 to the absorbance of Rad6 at both wavelengths were compared with the predicted values for a 1:1, 1:2, or 2:1 complex (Table I). The experimental values were closest to the values predicted for a 1:1 complex, again indicating that Rad6 and Rad18 are equimolar in the complex.
To determine the size of the Rad6-Rad18 complex, a fraction of the Mono Q sample was analyzed by sedimentation in a glycerol gradient; we observed that Rad6-Rad18 sedimented with an apparent molecular mass of ϳ90 kDa. Taken together, the results indicate that Rad6 and Rad18 combine to form a highly stable heterodimer.
Although Rad6 protein can be purified alone (5), we have not yet succeeded in purifying Rad18 protein by itself because little or no enrichment of Rad18 could be obtained when yeast extract containing overproduced Rad18 protein was subjected to fractionation in various chromatographic matrices, including DEAE-Sephacel, Q-Sepharose, Bio-Rex 70, and single-stranded DNA agarose (data not shown). The poor chromatographic behavior of free Rad18 protein suggests that proper folding of the protein may be effected via its association with Rad6 protein.
The Rad6-Rad18 Heterodimer Has Ubiquitin Conjugating Activity-Rad6 protein has ubiquitin conjugating (E2) activity (4,5). We examined if the Rad6 ubiquitin conjugating activity was affected because of the association with Rad18 by testing the ability of Rad18-bound Rad6 protein to form a thioester conjugate with ubiquitin ( Fig. 2A) and to transfer the conjugated ubiquitin to a protein substrate (Fig. 2B). Since a thioester linkage can be easily disrupted by treatment with a thiol reducing agent such as dithiothreitol or ␤-mercaptoethanol, to detect formation of a thioester adduct between ubiquitin and Rad6, the results of two electrophoreses without and with prior boiling in the presence of dithiothreitol were compared ( Fig.  2A, compare lanes 1-3 to lanes 4 -6). In this assay, GST-ubiquitin was used because it could be easily radiolabeled with 32 P (18). In the presence of ATP and the ubiquitin-activating (E1) enzyme Uba1, Rad6 forms a thioester adduct with GST-32 Pubiquitin (compare lanes 2 and 5, Fig. 2A). Rad18-associated Rad6 also forms a thioester with GST-32 P-ubiquitin in presence of Uba1 and ATP, and the reaction occurs to the same extent seen with free Rad6 (Fig. 2A, compare lanes 3 and 6). Thus, the binding of Rad18 to Rad6 does not prevent the latter from interacting with Uba1 and forming a thioester conjugate with ubiquitin.
Histone H2B (5) was used as the test substrate to examine the ability of Rad18-associated Rad6 to catalyze the formation of ubiquitin conjugates with protein substrates. The results in Fig. 2B show that H2B is a substrate for Rad18-associated Rad6, thus indicating that the Rad6-Rad18 complex retains ubiquitin conjugating activity.
Rad6-Rad18 Complex Binds DNA-We used nitrocellulose filter binding to examine the interaction between the Rad6-Rad18 heterodimer and 32 P-labeled DNA. In this assay, protein bound DNA molecules are retained on the nitrocellulose filters,

FIG. 2. Rad18-associated Rad6 protein has ubiquitin conjugating activity.
A, ubiquitin thioester formation with Rad6. GST-32 Pubiquitin was incubated with Uba1 (E1) and Rad6 (lanes 2 and 5) or Rad6- Rad18 (lanes 1, 3, 4, and 6), in the absence (lanes 1 and 4) or presence of ATP (lanes 2, 3, 5, and 6). The reaction was stopped, and the proteins were denatured with SDS and urea in the absence (lanes 1-3) or presence of DTT (lanes 4 -6). The samples were analyzed by electrophoresis and autoradiography. B, histone H2B ubiquitination. Histone H2B was incubated with Uba1 (E1), 125 I-ubiquitin, ATP (lane 3), containing Rad6 (lane 2) or Rad6-Rad18 complex (lane 1). 0.5 g of either Rad6 or Rad6-Rad18 were used in these assays (see "Materials and Methods"). The samples were analyzed by electrophoresis and autoradiography. Rad6-Rad18 protein complex was separated on a Vydac C 4 reverse phase column; the elution of the two proteins was monitored by measuring UV absorbance at 214 and 280 nm and by analyzing an aliquot by polyacrylamide gel electrophoresis; the area under the peaks corresponding to Rad6 and Rad18 and the ratios were calculated (observed); the number of amino acids in each protein was used to calculate the expected ratio at 214 nm. The formula A 280 ϭ (no. of Trp ϫ 5700) ϩ (no. of Tyr ϫ 1300) was used to calculate the expected ratio at 280 nm.
whereas free DNA flows through the filters, thus allowing determination of the amount of radiolabeled DNA bound by Rad6-Rad18 complex by measuring the radioactivity associated with the filters. The data presented in Fig. 3 show that the Rad6-Rad18 complex binds the single-stranded polynucleotide substrate poly(dT) in a protein concentration-dependent manner (Fig. 3A). Binding of the polynucleotide reaches the maximal level at 0.25 M Rad6-Rad18 complex.
We also determined the relative affinities of the Rad6-Rad18 complex for poly d(T) and double-stranded DNA. This was carried out by examining the ability of unlabeled linear doublestranded DNA to compete with radiolabeled poly(dT) for binding to the Rad6-Rad18 complex. The data presented in Fig. 3B were obtained from experiments wherein 0.1 M Rad6-Rad18 complex was first incubated with 2.0 M 32 P-labeled poly(dT) for 10 min, and subsequently, varying amounts of unlabeled linear dsDNA or unlabeled poly(dT) were added as competitor. After a further 10-min incubation, the reaction mixtures were applied to nitrocellulose filters, and the amount of radiolabeled poly(dT) remaining complexed with Rad6-Rad18 complex quantitated. While the inclusion of 5 M unlabeled poly(dT) reduced binding to the radiolabeled poly(dT) by 70%, as much as 40 M double-stranded DNA did not inhibit binding of the radiolabeled poly(dT) appreciably. These observations indicate that Rad6-Rad18 complex binds poly(dT), but binds dsDNA only weakly. Similar nitrocellulose filter binding studies employing ss and ds M13 DNAs also indicate that Rad6-Rad18 has much higher affinity for ssDNA (data not shown). When the competition is carried out in the reverse manner, beginning with the Rad6-Rad18 complex bound to 32 P-labeled dsDNA, the addition of poly(dT) competes the binding to dsDNA (Fig. 3C). The amount of dsDNA retained is much less (ϳ10%) than that observed for poly(dT) (data not shown), consistent with the apparently lower affinity for dsDNA displayed by the Rad6-Rad18 complex. Thus, we conclude that the Rad6-Rad18 complex has a strong preference for binding to poly(dT), and to ssDNA in general, as the same behavior is observed when ss M13 DNA is used in place of poly dT (data not shown).
Since the Rad6-Rad18 complex exhibits a ssDNA-dependent ATPase activity (see results below), it was of considerable interest to determine whether ATP affects the DNA binding properties of the Rad6-Rad18 complex. However, ATP, ADP, or ATP␥S have no effect on the binding of poly(dT) (data not shown). We also tested the effect of ATP in two other assays which were designed to reveal minor differences in DNA binding affinity. Using radiolabeled poly(dT) as DNA substrate, we first looked for differences in the sensitivity to displacement by increasing concentrations of NaCl, and secondly, at the kinetics FIG. 3. Rad6-Rad18 complex binds ssDNA. A, binding of Rad6-Rad18 complex to poly(dT). Rad6-Rad18 complex binding to poly(dT) was measured by retention of 32 P-labeled polynucleotide on nitrocellulose filters as described under "Material and Methods." The reaction buffer contained Mg 2ϩ at a concentration of 4 mM; the concentration of poly(dT) was 2 M. B and C, preferential binding of Rad6-Rad18 to poly(dT) over dsDNA. B, Rad6-Rad18 complex (0.1 M) was incubated with 2 M 32 P-labeled poly(dT) for 10 min before addition of increasing amounts of unlabeled competitor dsDNA (linearized pUC19) or poly(dT) and further incubation for another 10 min. Each aliquot was filtered and the amount of 32 P-labeled poly(dT) remaining bound determined and expressed as a percentage of the binding observed in the absence of added competitor. C, Rad6-Rad18 complex (0.1 M) was incubated with a 32 P-labeled 300-base pair ds fragment as in B. Unlabeled poly(dT) was added as the competing polynucleotide.
of dissociation from a radioactively labeled substrate. The presence of ATP has no effect on either the sensitivity to salt (Fig.  4A) or on the rate of dissociation from poly(dT) (Fig. 4B). Thus, we conclude that the DNA binding behavior of the Rad6-Rad18 complex is not modulated significantly by ATP.
Rad6-Rad18 Complex Has ssDNA-dependent ATPase Activity-The Rad18 protein sequence contains a Walker type A nucleotide binding motif GKS found in a variety of proteins that bind and hydrolyze ATP (10). Thus, it was important to determine whether the Rad6-Rad18 complex has ATPase activity. We found that purified Rad6-Rad18 complex hydrolyzes ATP with a dependence on a DNA cofactor, with singlestranded DNA being much more effective than double-stranded DNA in stimulating ATPase activity. For example, Rad6-Rad18 ATPase activity was stimulated 20-fold by M13 ssDNA but only ϳ2-fold by M13 ds DNA.
When the fractions from the last chromatographic purification step in Mono Q were assayed for their ATPase activity, co-elution of the ATPase activity with the Rad6-Rad18 complex was observed (Fig. 5). We also subjected a sample of the Mono Q-purified Rad6-Rad18 complex to molecular sizing in a Superose 12 column and once again found the precise co-elution of this ATPase activity with the Rad6-Rad18 protein (data not shown). The co-elution observed and the high degree of purity of the Rad6-Rad18 complex indicate that the ATPase activity is an intrinsic property of this protein complex.
The ATP concentration dependence of the ATP hydrolysis activity (Fig. 6A) shows Michaelis-Menten behavior. Since in this assay, the amount of poly(dT) is in excess over the Rad6-Rad18 complex, the V max is also equal to the k cat of 0.18 min Ϫ1 . The rates of ATP hydrolysis are linear over a 1 h time period, where the concentration of accumulated ADP reaches 50 -60 M (ϳ25% of the total nucleotide concentration), suggesting there is no significant inhibition by ADP. Finally, there is no trace of pyrophosphate released, suggesting that the Rad6-Rad18 complex catalyzes the hydrolysis of ATP to ADP and inorganic phosphate.
We carried out a protein titration in the presence of a constant amount of ssDNA while monitoring the ATP hydrolysis rate. The data in Fig. 6B show that the rate of ATP hydrolysis increases linearly with added protein until an inflection point is reached where all of the available binding sites are occupied. Assuming that all of the available sites on the DNA co-factor are occupied, the site size (n) calculated for the Rad6-Rad18 complex (reflecting the binding site size) is four nucleotides/ Rad6-Rad18 molecule. The k cat calculated from these data (the maximum rate divided by the protein concentration at the inflection point) is in good agreement with the value calculated previously, where the amount of DNA is well in excess of what was needed to allow all the protein to bind. The pH dependence of the ATP hydrolysis activity shows that between pH 6.0 and pH 8.8 there is no significant difference in the observed rate of hydrolysis (data not shown).

DISCUSSION
In this study, we show that Rad18 exists in a stable complex with Rad6. The two proteins co-purify through sequential chromatographic fractionation in columns of Q-Sepharose, phenyl-Sepharose, Mono S, and Mono Q. As judged by Coomassie Blue staining, reverse phase chromatography, and glycerol gradient sedimentation, the purified complex contains an equimolar ratio of Rad6 and Rad18. The Rad6-Rad18 complex represents the first example wherein a ubiquitin conjugating enzyme is physically associated with a DNA-binding protein. This association provides an explanation for the very similar roles of RAD6 and RAD18 in DNA repair and damage-induced mutagenesis.
FIG. 5. ATPase activity co-elutes with the Rad6-Rad18 complex. Top, Rad18 in the Mono Q peak fractions of Rad6-Rad18; proteins were analyzed by SDS-polyacrylamide gel electrophoresis. Bottom, the amount of ATP hydrolyzed in 10 min was measured using 1 l of each fraction.
FIG. 4. ATP has no effect on the DNA binding affinity of the Rad6-Rad18 complex. A, ATP does not affect sensitivity to salt. The sensitivity of Rad6-Rad18 DNA binding to increasing salt concentrations was measured by the nitrocellulose filter binding assay. 32 P-Labeled poly(dT) (2 M) was incubated with 0.25 M Rad6-Rad18 for 10 min at various salt concentrations and processed as described under "Materials and Methods." When present, ATP (1 mM) and Mg 2ϩ (4 mM) were used. B, dissociation from poly(dT) is unaffected by ATP. The kinetics of Rad6-Rad18 complex dissociating from poly(dT) was measured by incubating 0.25 M Rad6-Rad18 protein with 2 M 32 P-poly(dT) for 10 min followed by the addition of 40 M unlabeled poly(dT) and filtering aliquots on nitrocellulose filters at the indicated times. The amount remaining bound was determined and expressed as a percentage of the zero time point. The concentration of ATP when present was 1 mM, with Mg 2ϩ at 4 mM.
The Rad6-Rad18 complex forms a thioester adduct with ubiquitin and conjugates ubiquitin to histone H2B. Thus, the complex retains the ubiquitin conjugating activity of Rad6. In addition, we demonstrate that the Rad6-Rad18 complex binds preferentially to ssDNA. Since Rad6 contains no DNA binding motifs and displays no DNA binding ability, the ssDNA binding ability of the Rad6-Rad18 complex very likely derives from Rad18. Consistent with this, Rad18 contains a C 3 HC 4 sequence motif (10), also known as the RING finger motif (20), as well as a CX 2 CX 12 HX 3 C (C 2 HC) motif (10), either or both of which could be utilized in DNA binding.
The Rad6-Rad18 complex exhibits an ssDNA-dependent ATPase activity with a k cat of 0.18 min Ϫ1 . The high degree of purity of the complex and the precise co-elution of the ATPase activity with the Rad6-Rad18 complex suggest that this activity is intrinsic to this complex. The presence of the "GKS" Walker type A nucleotide binding motif in Rad18 (10), and the absence of any nucleotide binding motif or of any NTPase activity in Rad6 implies that the ATPase activity of the complex resides in Rad18. The fact that the ATPase activity is relatively weak may indicate that the observed activity is a basal rate and only in the presence of an as yet unidentified partner is the rate increased, in a manner analogous to that seen with the stimulation of RF-C ATPase activity by proliferating cell nuclear antigen (21).
How might the DNA binding, ATPase, and ubiquitin conjugating activities of the Rad6-Rad18 complex function in the postreplicative bypass of damaged DNA? The DNA binding activity could target the complex to sites of ssDNA where the DNA replication machinery has been blocked by DNA lesions. As the ATPase activity seems to have no significant effect on the DNA binding activity of the Rad6-Rad18 complex, it is possible that this activity is utilized in Rad18's role as a molecular matchmaker (22), wherein Rad18 recognizes the protein substrates for ubiquitination by Rad6, and ATP binding and hydrolysis modulate these protein-protein interactions.
The availability of the purified Rad6-Rad18 complex, and the identification of biochemical activities contained within it, now presents the opportunity to identify the components of the replication machinery that are ubiquitinated by the complex and to determine if ubiquitin-dependent degradation of some of the replicative proteins is required for the assembly of the postreplicative bypass DNA repair machinery.