| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Genetics, Vol. 169, 1939-1955, April 2005, Copyright © 2005
doi:10.1534/genetics.104.033894
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
-Dependent Translesion Synthesis in Saccharomyces cerevisiae
,1
* Biochemistry, Cell and Developmental Biology Program of the Graduate Division of Biological and Biomedical Sciences
Department of Biology, Emory University, Atlanta, Georgia 30322
1 Corresponding author: Department of Biology, Emory University, 1510 Clifton Rd., Atlanta, GA 30322.
E-mail: sue.jinks-robertson{at}emory.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
. In this study, changes in the reversion rate and spectrum of the lys2
A746 –1 frameshift allele have been used to evaluate how the loss of members of the RAD6 epistasis group affects Pol-dependent mutagenesis in response to spontaneous damage. Our data are consistent with a model in which Pol-dependent mutagenesis relies on the presence of either Rad5 or Rad18, which promote two distinct error-prone pathways that partially overlap with respect to lesion specificity. The smallest subunit of Pol
, Pol32, is also required for Pol-dependent spontaneous mutagenesis, suggesting a cooperative role between Pol and Pol for the bypass of spontaneous lesions. A third error-free pathway relies on the presence of Mms2, but may not require PCNA.
Three translesion polymerases have been described in Saccharomyces cerevisiae: Pol
, Rev1, and Pol. Both Pol and Rev1 belong to the Y family of DNA polymerases, whose founding members include polymerases involved in the SOS response of Escherichia coli (OHMORI et al. 1999). In vitro, Pol (encoded by RAD30) can bypass thymine-thymine dimers and 8-oxo-G lesions in an error-free manner (JOHNSON et al. 1999; HARACSKA et al. 2000b; YUAN et al. 2000), while bypassing other lesions in an error-prone manner (YUAN et al. 2000). In vivo, lesion bypass by Pol has a variable effect on spontaneous mutagenesis; rad30 mutants can exhibit an increase, a decrease, or no change in mutation rate, depending upon the assay system used (MCDONALD et al. 1997; ROUSH et al. 1998). REV1 encodes a protein with limited polymerase activity in vitro, specifically incorporating cytosine across from both damaged and undamaged bases (LAWRENCE 2002). This polymerase activity, however, does not appear to be required for the role(s) of Rev1 in in vivo mutagenesis (HARACSKA et al. 2001; S. WILEY and S. JINKS-ROBERTSON, unpublished observations). Rev1 interacts genetically with the third translesion polymerase in yeast, Pol (HARFE and JINKS-ROBERTSON 2000; LAWRENCE 2002). Pol is composed of two subunits: a catalytic subunit encoded by REV3 and an accessory factor encoded by REV7 (NELSON et al. 1996). Like the other translesion polymerases, Pol has no 3'-5' proofreading activity and demonstrates low processivity, with half of the molecules dissociating from a DNA template after the addition of only three to four nucleotides (NELSON et al. 1996). In vitro, the fidelity of Pol when bypassing DNA damage is lesion dependent; Pol bypasses thymine dimers in an error-prone manner (NELSON et al. 1996), but thymine glycols in an error-free manner (JOHNSON et al. 2003). In addition to its capacity to place a nucleotide across from a damaged base, Pol possesses the unusual ability to extend from a primer:template mispair (JOHNSON et al. 2000; HARACSKA et al. 2003), an activity that is absent in replicative DNA polymerases. Given that 50–75% of spontaneous mutagenesis in yeast has been attributed to Pol (QUAH et al. 1980), understanding how the activity of Pol is regulated is an important step in elucidating the complex mechanisms underlying spontaneous mutagenesis.
The use of Pol in bypassing spontaneous DNA lesions appears to be largely regulated by members of the RAD6 epistasis group, which include RAD6, RAD18, RAD5, MMS2, UBC13, and SRS2. This epistasis group was originally defined in relation to induced DNA damage and was termed the postreplication repair (PRR) pathway. This designation specifically reflects the ability of many members of this pathway to convert the low-molecular-weight DNA formed in response to induced damage into larger DNA products without removal of the original lesion (PRAKASH 1981; TORRES-RAMOS et al. 2002). Further work revealed that the PRR pathway is composed of an error-prone and an error-free subpathway, both of which are under the control of Rad6 and Rad18 (PRAKASH 1981). The error-prone pathway depends on Pol, while the error-free PRR pathway is composed of Rad5, Mms2, Ubc13, and Srs2 (ULRICH and JENTSCH 2000; BROOMFIELD et al. 2001; ULRICH 2001; BROOMFIELD and XIAO 2002).
Although the term PRR has been used to describe RAD6-dependent lesion bypass of spontaneous as well as induced damage, use of this acronym may be misleading if used to describe spontaneous lesion bypass, as the roles of these proteins in bypassing spontaneous damage appear to differ from those utilized for bypassing induced damage (LIEFSHITZ et al. 1998; CEJKA et al. 2001). For example, Rad6 and Rad18 are only partially required for spontaneous mutagenesis, whereas they are completely required for mutagenesis in response to induced damage (FRIEDBERG et al. 1995). In addition, the activity of postreplication repair has not been formally demonstrated to occur in the absence of induced damage. For the sake of clarity, we refer to spontaneous lesion bypass promoted by members of the RAD6 epistasis group as the spontaneous lesion bypass (SLB) pathway, to differentiate it from damage-induced lesion bypass.
In contrast to the two-pathway model proposed for bypass of induced damage, genetic characterization of SLB suggests a three-pathway model, with two error-prone pathways and one error-free pathway (Figure 1). Elimination of Rad18 or Rad5 alone results in an increase in Pol-dependent mutagenesis, while simultaneous loss of Rad18 and Rad5 results in a dramatic decrease in spontaneous mutagenesis, similar to a rev3 mutant (LIEFSHITZ et al. 1998; CEJKA et al. 2001). One error-prone pathway (Rad18-dependent pathway) appears to be composed of Rad6, Rad18, Pol32, and Pol, while the other (Rad5-dependent pathway) is composed of Rad5, Pol32, and Pol. RAD6 encodes a ubiquitin-conjugating enzyme (JENTSCH et al. 1987) that physically associates with Rad18, a single-stranded DNA binding protein (BAILLY et al. 1997). Deletion of RAD6 results in a pleiotrophic phenotype, affecting sporulation (MORRISON et al. 1988), telomere elongation (HUANG et al. 1997), protein degradation (DOHMEN et al. 1991), and damage-induced PRR (PRAKASH 1981), while deletion of RAD18 appears to be specific for PRR (PRAKASH 1981). Rad6 and Rad18 form a stable complex, which can dimerize via the Rad18 subunit to promote Pol-dependent translesion synthesis (BAILLY et al. 1997). The second error-prone branch of the SLB pathway is thought to require a homodimer of the Rad5 protein (ULRICH and JENTSCH 2000; Figure 1), an ssDNA-dependent ATPase with homology to the Swi/Snf family of chromatin remodeling factors (JOHNSON et al. 1992, 1994). Whether the Rad5- and Rad18-dependent error-prone pathways are truly functionally redundant, or whether each exhibits some lesion specificity, is not known. In addition to its self-association, Rad5 can interact with Mms2 and Ubc13 (ULRICH and JENTSCH 2000), which together form a heterodimer possessing K63-linked ubiquitin-conjugating activity (HOFMANN and PICKART 1999). Although Rad5 is capable of interacting with Mms2 and Ubc13, a requirement for these proteins in Rad5-dependent error-prone bypass of spontaneous damage has not been demonstrated.
|
, encoded by POL3, POL31, and POL32, has also been predicted to reside in the error-free arm on the basis of epistasis of rad18 to the pol3-13 allele in response to induced damage (GIOT et al. 1997). Although the catalytic subunit (Pol3) of Pol is implicated in error-free lesion bypass, the smallest subunit of Pol, Pol32, has been associated with mutagenesis. pol32 mutants exhibit decreased spontaneous and damage-induced mutagenesis, similar to a rev3 mutant (HUANG et al. 2000, 2002). However, it remains to be determined if the decrease in mutagenesis seen in a pol32 mutant is the result of loss of Pol activity.
The decision to perform lesion bypass of induced damage via members of the RAD6 epistasis group as opposed to recombination-based bypass appears to rely largely on the activity of Srs2 (reviewed in BROOMFIELD et al. 2001; SMIRNOVA and KLEIN 2003). SRS2 encodes a 3'-5' helicase with antirecombinase activity (RONG and KLEIN 1993) and in vitro can strip the Rad51 protein from the single-stranded nucleoprotein filament that promotes invasion of a duplex DNA molecule (KREJCI et al. 2003; VEAUTE et al. 2003). Thus, Srs2 has been proposed to inhibit Rad51-dependent recombination, thereby allowing members of the RAD6 epistasis group to perform their functions in damage tolerance. Although Srs2 plays a key role in Pol-dependent lesion bypass in response to induced damage (LIEFSHITZ et al. 1998), the role of Srs2 with regard to spontaneous damage is less clear.
Previously, we described a chromosomal-based, frameshift-specific reversion assay that can be used to monitor overall changes in spontaneous mutation rates as well as in patterns of mutagenesis. This system utilizes the lys2A746 –1 frameshift allele that reverts via second-site, compensatory +1 frameshift events. Different classes of net +1 frameshift mutations can be detected within an 
150-bp "reversion window" defined by stop codons in alternate, incorrect reading frames (HARFE and JINKS-ROBERTSON 1999). In most genetic backgrounds, the occurrence of single-nucleotide insertions, termed "simple" events, predominates. However, in genetic backgrounds in which DNA damage repair mechanisms have been compromised, there is a striking increase in the proportion of "complex" events, which are composed of a single-nucleotide insertion coupled with a base-pair substitution (HARFE and JINKS-ROBERTSON 2000). The accumulation of these complex frameshifts requires Pol; deletion of the catalytic subunit of Pol (REV3), inactivation of the catalytic domain of Rev3, or the elimination of the processivity factor Rev7 completely abolishes complex frameshift events (HARFE and JINKS-ROBERTSON 2000; S. YELLUMAHANTI, B. MINESINGER and S. JINKS-ROBERTSON, unpublished results). The complex events detected by the lys2A746 assay system are a distinctive mutational signature of Pol and thus provide a unique opportunity to identify factors affecting its activity in spontaneous mutagenesis. In this study, we use the lys2A746 assay system to analyze Pol-dependent mutagenesis in the absence of key components of the RAD6 epistasis group. The results obtained further refine the role of RAD6 epistasis group members in the regulation of Pol-dependent spontaneous mutagenesis and lend further support to the three-pathway model of spontaneous lesion bypass.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Strain construction:
A complete list of strains is given in Table 1. All strains are congenic derivatives of SJR922 (MAT
ade2-101oc his3200 ura3Nco lys2A746; HARFE and JINKS-ROBERTSON 1999) and were obtained by standard lithium acetate transformation (GIETZ et al. 1995). Ura– derivatives of the rad1 mutant strains SJR1166 and SJR1274 (HARFE and JINKS-ROBERTSON 2000) were identified on 5-FOA medium to create strains SJR1177 and SJR1770, respectively. MMS2 and REV3 were deleted using a PCR-generated kanMX2 cassette (WACH et al. 1994; SWANSON et al. 1999), and transformants were selected on YEPD + G418. RAD5 and POL32 were deleted using a PCR-generated hphMX4 cassette (GOLDSTEIN and MCCUSKER 1999), and transformants were selected on YEPD + hygromycin. All PCR-mediated gene deletions contain a precise deletion of the published open reading frame with the exception of rev3::KAN, in which the first and last 60 nt of the coding region remain (SWANSON et al. 1999). EcoRI-digested pJJ239 (provided by L. Prakash) and SacI/SphI-digested pPM690 (LEE et al. 1999) were used to disrupt RAD18 and SRS2, respectively. SJR2239 (leu2-K pol30::hisG-URA3-hisG pol30-46) and SJR2069 (leu2-R rad1 pol30::hisG-URA3-hisG pol30-46) were constructed as follows. First, AflII-digested pJH188 and KpnI-digested pJH189 (LICHTEN et al. 1987) were used to create leu2-K and leu2-R derivatives of SJR922 and SJR1177, respectively. Following selection for Ura+ transformants, plasmid loss events were selected on 5-FOA and the Leu– phenotype was confirmed. The pol30-46 allele was then introduced by transformation of HpaI-digested pBL248-46 (obtained from P. Burgers), which targets the pol30-46 allele to the LEU2 locus. Finally, the wild-type copy of POL30 was disrupted using KpnI/MluI-digested pBL243 (obtained from P. Burgers). Phenotypic confirmation of gene disruptions was conducted when appropriate. All disruptions were confirmed by PCR.
|
The reversion rate of a specific category of frameshift event was determined by multiplying the percentage of the event in the reversion spectrum by the corresponding overall mutation rate. To determine the 95% confidence intervals for the event, the high and low confidence intervals for the overall mutagenesis rate were multiplied by the percentage of times the specific event occurred in the spectrum.
Reversion spectra:
Spontaneous Lys+ revertants were obtained as described above. To ensure that each reversion event was independent, only one colony from each culture was analyzed. After purifying the revertant colony on SCD-lysine plates, genomic DNA was isolated using standard yeast protocols. Sequencing of an appropriate PCR product was performed as described in HARFE and JINKS-ROBERTSON (1999). Sequences were analyzed using the Sequence Manager Software (DNASTAR, Madison, WI) licensed from BIMCORE at Emory University. A table describing the nature of the complex insertions (cins) in each strain is available upon request.
Statistical analysis:
Monte Carlo statistical analysis (ADAMS and SKOPEK 1987; CARIELLO et al. 1994) was performed to compare the distribution (number and location) of complex and simple frameshift events within two reversion spectra. A complex event is defined as a single-nucleotide insertion accompanied by a base-pair change and a simple frameshift is defined as the insertion of one nucleotide in the reversion window. Briefly, a table was created so that one column contained each nucleotide in the reversion window. In a homopolymer run, one cannot precisely determine where the frameshift occurred; therefore, any repeated sequence was considered as a single nucleotide. Each nucleotide (or nucleotide group) was assigned the designation of "C" for complex or "S" for simple and the number of complex or simple events that occurred at each site was recorded. The location of a given complex frameshift corresponded to the position where the frameshift event occurred, not to where the base substitution occurred. Probability (P) values <0.05 were considered significantly different.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
A746 –1 frameshift allele in the absence of key members of this epistasis group. Previous studies demonstrated that a rad18, rad5, or mms2 mutant exhibits a Pol-dependent mutator phenotype (ROCHE et al. 1995; LIEFSHITZ et al. 1998; XIAO et al. 1999; CEJKA et al. 2001; HUANG et al. 2002). Consistent with these results, elimination of Rad18, Rad5, or Mms2 resulted in a small but significant increase in the lys2A746 spontaneous reversion rate (Table 2). To determine the locations of these compensatory +1 frameshift mutations, we sequenced the "reversion window" of independent, Lys+ revertant colonies and compiled these mutations to create a reversion spectrum. Frameshift events were classified into one of two broad categories: complex frameshifts, in which a simple insertion event was accompanied by a base-pair change, and noncomplex frameshifts, which included all other frameshift events. When necessary, the category of noncomplex frameshift events was further broken down into simple events, composed of the insertion of a single nucleotide; large deletion events in which large portions of the sequence were deleted (see below); and "other" events that did not fit into any previously described category. Our previous work has demonstrated that complex events are a unique mutational signature of Pol-dependent mutagenesis (HARFE and JINKS-ROBERTSON 2000) and, as such, provide a means with which to specifically monitor perturbations in Pol mutagenic activity in vivo. Sequencing of revertants revealed that the percentage of complex frameshift events increased from 6% in a wild-type strain to 23% in the rad18 and mms2 mutants and 28% in the rad5 mutant (Figure 2). This translates into a six- to eightfold increase in the rate of complex events in a rad5, rad18, or mms2 background. While the rate of noncomplex events was not significantly elevated in either a rad5 or an mms2 mutant, there was a slight increase in a rad18 mutant (Table 2).
|
|
0.0006). This is consistent with previous reports in which significant differences in forward mutation spectra were observed in mms2, rad18, and rad5 mutants compared to a wild-type strain (KUNZ et al. 1991; XIAO et al. 1999; HUANG et al. 2002). While the distribution of frameshift events in the mms2 mutant was similar to the distributions in both the rad5 (P = 0.2) and the rad18 (P = 0.09) mutants, the frameshift distributions in the rad5 and rad18 mutants were not similar (P = 0.005).
The function of Mms2 is not required in the Rad5-dependent error-prone subpathway of SLB:
The initial assignment of Mms2 and Ubc13 to the error-free branch of the PRR pathway was primarily based on epistasis analysis with other RAD6-pathway members with regard to sensitivity to various DNA-damaging agents (BROOMFIELD et al. 1998; HOFMANN and PICKART 1999), which may or may not reflect the ways in which the cell deals with spontaneous damage. With regard to spontaneous lesions, the only evidence to place Mms2 and Ubc13 in the error-free pathway is that mms2 and ubc13 mutants have increased spontaneous Pol-dependent mutagenesis (HOFMANN and PICKART 1999; XIAO et al. 2000). However, an increase in spontaneous Pol-dependent mutagenesis is also seen in rad18 or rad5 mutants, which are also defective in the error-prone pathways (HUANG et al. 2000, 2002; CEJKA et al. 2001). Given that the Mms2-Ubc13 complex physically interacts with Rad5 (ULRICH and JENTSCH 2000; ULRICH 2003) and that Rad5 is active in both an error-prone and an error-free pathway of SLB (LIEFSHITZ et al. 1998; CEJKA et al. 2001), it is possible that Mms2-Ubc13 also functions in both pathways in response to spontaneous damage. If Mms2-Ubc13 is required for all functions of Rad5 in SLB, then one would expect the mutation rates and spectra of a rad5 and an mms2 mutant to be the same. Indeed, the overall mutation rates, the complex frameshift rates, and the distribution of frameshift events were indistinguishable between the rad5 and mms2 mutants (Figure 2; Table 2), which would be consistent with Mms2 functioning in both the error-free and the error-prone Rad5-dependent pathways.
To further examine possible differences between Rad5 and Mms2 loss, lys2A746 reversion rates and spectra were obtained in a rad18 background. rad18 rad5 strains have low, rev3-like levels of mutagenesis (LIEFSHITZ et al. 1998; CEJKA et al. 2001). Thus, we hypothesized that if Mms2 was needed for all functions of Rad5, rad18 mms2 mutants would also display low levels of mutagenesis. In agreement with previous studies, rad18 rad5 mutants had a significant decrease in the frameshift rate relative to each of the single mutants in our assay system (Table 2). Strikingly, complex frameshift events were completely absent in the rad18 rad5 double mutant (Figure 2). This represents at least a 38-fold reduction in complex events in the rad18 rad5 double mutant relative to the single mutants and at least a 4.8-fold reduction relative to a wild-type strain. These results are consistent with a model in which Pol-dependent mutagenesis occurs via distinct Rad5- and Rad18-dependent error-prone pathways.
We next analyzed the rad18 mms2 strain. If Mms2 is required for all functions of Rad5 in SLB, then a rad18 mms2 mutant should be equivalent to a rad18 rad5 mutant. Although the overall reversion rate of a rad18 mms2 mutant was not significantly different from that of the rad18 rad5 mutant, the rate of complex frameshift accumulation in the rad18 mms2 background was dramatically different from that in a rad18 rad5 background (Figure 2; Table 2). The rate of complex frameshift events was at least 43-fold higher in a rad18 mms2 mutant than in the rad18 rad5 mutant, with complex events accounting for 30% of the reversion events in the rad18 mms2 strain (Figure 2). A comparison of the distribution of frameshift events in a rad18 rad5 mutant vs. a rad18 mms2 mutant by Monte Carlo analysis revealed highly significant differences (P < 0.0001). These data provide a clear distinction between the functions of Rad5 and Mms2 in response to spontaneous damage and are consistent with the existence of a Rad5-dependent, Mms2-independent error-prone bypass pathway.
Pol-dependent mutagenesis in rad18, mms2, and rad5 mutants in the absence of nucleotide excision repair:
To more thoroughly investigate the role of Rad5, Rad18, and Mms2 in spontaneous mutagenesis, the lys2A746 reversion system was sensitized by deleting the corresponding genes in a nucleotide excision repair (NER)-defective (rad1) background. We anticipated that the resulting increase in the amount of unrepaired spontaneous damage would create more opportunities for Pol-dependent bypass of DNA lesions, thus allowing enhanced detection of subtle changes in frameshift accumulation. Indeed, elimination of rad5, rad18, or mms2 in a rad1 background significantly increased the lys2A746 reversion rate two- to threefold when compared to either the corresponding NER-proficient parent or the rad1 single mutant (Table 3). While the proportion of complex frameshift events did not change when RAD5 or RAD18 was deleted in a rad1 background (Figure 3; 29% in rad1 rad5 vs. 28% in rad5; 29% in rad1 rad18 vs. 23% in rad18; 28% in rad1), the percentage of complex frameshift events nearly doubled in the rad1 mms2 double mutant (51%) relative to the mms2 or rad1 single mutant (23% or 28%, respectively). These results are consistent with the involvement of Mms2 in only an error-free pathway. Multiplying the percentage of complex frameshift events of the double-mutant strains by their respective mutation rates revealed that the double mutants had complex frameshift rates three- to sevenfold greater than those of the corresponding single mutants (Tables 2 and 3).
|
|
-dependent activity, we eliminated REV3 from the rad1 rad18 mutant. In agreement with earlier observations (HUANG et al. 2002), the increased mutation rate of the rad1 rad18 strain was dependent upon Rev3 (Table 3). Accordingly, the rad1 rad18 rev3 strain was devoid of complex frameshift events (Figure 3). In the triple mutant, we estimate that the rate of complex insertions was at least 25-fold lower than that in a rad1 mutant and at least 120-fold lower than that in a rad1 rad18 mutant. In addition to determining the rates of different classes of frameshifts, the overall distribution of frameshift events in the double mutants was compared to those of the appropriate SLB-defective or NER-defective single mutants. By Monte Carlo analysis, neither the rad1 rad18 nor the rad1 mms2 double mutant was similar to the relevant single-mutant strains (data not shown). Although the distribution of complex and simple events in a rad1 rad5 mutant differed significantly from the distribution in a rad1 strain (P = 0.002), it was not significantly different from that of a rad5 strain (P = 0.05).
Work in a NER-proficient background indicated that Mms2 was not required for all functions of Rad5, and we were interested in determining if this held true in response to an increased load of unrepaired spontaneous damage. Similar to their NER-proficient counterparts, the overall mutation rates and rates of complex frameshift formation of the rad1 rad5 and rad1 mms2 mutants were not statistically different. However, the distribution of frameshift events by Monte Carlo analysis revealed significant differences between rad1 mms2 and rad1 rad5 strains (P < 0.0001).
In the absence of NER, Rad18 and Rad5 are required for accumulation of complex frameshift events at discrete locations in the lys2A746 reversion window:
One striking feature of the rad1 spectrum is the accumulation of complex frameshift events primarily in two distinct locations, termed hotspots 1 and 2. These locations are thought to be sites of a specific type (or class) of spontaneous damage that remains in the genome in the absence of NER, thus making it a substrate for Pol-dependent translesion synthesis (HARFE and JINKS-ROBERTSON 2000; B. MINESINGER, A. ABDULOVIC and S. JINKS-ROBERTSON, unpublished results). In stark contrast to the rad1 mms2 and the rad1 rad5 spectra, each of which contained complex frameshifts at hotspots 1 and 2, there were no events at these locations in the rad1 rad18 spectrum (Figure 3). Relative to a rad1 strain, both the rad1 mms2 and the rad1 rad5 mutants had an 2-fold increase in the rate of complex events at hotspots 1 and 2, while the rad1 rad18 strain had at least a 6.7-fold decrease in the rate of these events (Table 4). The lack of complex events at hotspots 1 and 2 in the rad1 rad18 mutant (as well as a rad1 rad18 mms2 mutant; see Figure 3) suggests that Pol requires the presence of Rad18 to bypass the relevant lesion(s) that accumulates near hotspots 1 and 2 in the absence of NER. Alternatively, Pol may still bypass the lesion(s) in the absence of Rad18, but does so in a manner undetectable in our assay system.
|
11% of the total reversion events in the rad1 rad18 and rad1 mms2 strains, whereas these events composed <1.2% of the rad1 rad5 spectrum (assuming one event). This corresponds to at least an eightfold greater rate of complex events at the 6A run in the rad1 rad18 or the rad1 mms2 strain compared to the rad1 rad5 mutant (Table 4). Interestingly, the discrepancy between the rates of complex frameshifts at the 6A run of rad5, rad18, and mms2 mutants was evident only in the absence of NER, suggesting that in the absence of NER, lesion(s) that accumulate near the 6A run require Rad5 for efficient Pol-dependent mutagenesis.
Fitness defects are apparent when both RAD18 and RAD5 are deleted in a NER-defective background:
To further characterize the roles of Rad18, Rad5, and Mms2 in the presence of an increased spontaneous damage load, we created rad1 rad18 mms2 and rad1 rad18 rad5 strains. Although the rad1 rad18 mms2 mutant demonstrated a nearly fivefold increase in the overall and complex lys2A746 reversion rate relative to the rad1 single mutant, these rates were not statistically different from those of the rad1 mms2 or rad1 rad18 strain (Table 3). Complex reversion events in the triple mutant composed 25% of the spectrum (Figure 3), similar to that of the rad1 rad18 strain (29%) and less than half that of the rad1 mms2 strain (51%). By Monte Carlo analysis, the distribution of simple and complex frameshift events in a rad1 rad18 mms2 reversion spectrum differed significantly from the rad1 and rad1 mms2 spectra (P 0.0001), but was not different from the rad1 rad18 spectrum (P = 0.055). Notably, complex reversion events were completely absent at hotspots 1 and 2 in rad1 rad18 mms2 triple mutant, similar to the rad1 rad18 spectrum (Figure 3), strengthening the argument that Rad18 is required for bypass of a lesion(s) accumulating near these locations.
Examination of a rad1 rad18 rad5 strain revealed an unexpected fitness defect. Plating efficiencies (percentage of colonies formed per total cell number) of the rad1 rad18 rad5 mutant were at least 5-fold lower than that of a rad1 strain and at least 2.5-fold lower than that of a rad1 rad18 strain (data not shown). The precise plating efficiency and reversion rate were difficult to quantify accurately in this mutant, as the percentage of viable cells varied considerably between cultures. In addition to inconsistent plating efficiency, colony sizes were variable in the rad1 rad18 rad5 mutant background (data not shown). This defect in fitness was observed only in the rad18 rad5 mutant in the absence of NER, which suggests that when NER and all three branches of the SLB pathway are disabled simultaneously, the remaining DNA repair/bypass pathways are unable to fully compensate for the loss, leading to an increase in cell death or loss of cell proliferation. Despite the decrease in fitness, we examined the lys2A746 reversion spectrum of the rad1 rad18 rad5 strain. Given that the rad18 rad5 mutant did not accumulate complex reversion events, we were surprised to detect complex events when RAD18 and RAD5 were deleted in a rad1 background (Figure 3). The unexpected presence of complex events suggests that Pol may be able to promote translesion synthesis without Rad18 and Rad5 in response to an increased load of spontaneous damage. However, given the fitness defect of this strain, it is possible that other repair or replication-associated processes may be perturbed, resulting in the complex events detected in the rad1 rad18 rad5 mutant spectrum. An investigation of whether Pol was responsible for the complex events seen in the rad1 rad18 rad5 mutation spectrum would require the creation of a rad1 rad18 rad5 rev3 quadruple-mutant strain, which was not attempted due to the poor growth of the triple-mutant strain.
pol30-46 mutants have increased overall spontaneous mutation rates, but pol30-46 spectra are not equivalent to mms2 spectra:
On the basis of epistasis analysis with mms2 and pol30-46 mutants in response to exogenous damage, ULRICH (2001) and BROOMFIELD and XIAO (2002) have placed PCNA downstream of MMS2 in the error-free bypass pathway. pol30-46 mutants demonstrate an increase in spontaneous mutagenesis of 2- to 7-fold relative to a wild-type strain (AYYAGARI et al. 1995; TORRES-RAMOS et al. 1996); whether Pol contributes to the pol30-46 mutator phenotype has not been determined. Therefore, we examined the lys2A746 reversion rates and spectra of pol30-46 strains in the presence and absence of NER. If PCNA resides downstream of MMS2 in the error-free SLB pathway, we hypothesized that spontaneous mutagenesis in a pol30-46 mutant would be similar to that in an mms2 mutant; namely, pol30-46 strains should have an overall increase in mutation rate and a specific increase in the rate of complex frameshift events similar to that of mms2 strains. Indeed, pol30-46 mutants had an increased overall mutation rate that was slightly, but significantly (1.4-fold), greater than that of a wild-type strain, similar to previous reports (AYYAGARI et al. 1995; TORRES-RAMOS et al. 1996; Table 2). This rate was indistinguishable from that of an mms2 strain. While the rate of complex frameshift events increased by 2.4-fold in the pol30-46 mutant relative to a wild-type strain, the rate of complex frameshift events in the pol30-46 mutant decreased by >2-fold relative to the mms2 strain. Using Monte Carlo analysis, we found that the distribution of frameshifts in the pol30-46 spectrum was significantly different from that in the mms2 spectrum (P = 0.0023), but not from that in the wild-type spectrum (P = 0.24). Similar results were obtained when mutation rates and spectra were compared between the NER-defective rad1 pol30-46 and rad1 mms2 strains (Table 3; Figure 3). While the overall mutation rates of the rad1 pol30-46 and the rad1 mms2 strains were not significantly different, the rate of complex events was 7-fold lower in the rad1 pol30-46 strain than in the rad1 mms2 strain (Table 3). Monte Carlo analysis revealed significant differences between the distribution of frameshift events when comparing the rad1 mms2 and rad1 pol30-46 strains (P < 0.0001), but not when comparing the rad1 and the rad1 pol30-46 strains (P = 0.36). Taken together, these data indicate that even though POL30 and MMS2 both play a role in response to spontaneous damage, they are not equivalently required for the same process (see DISCUSSION).
Additional analysis of the pol30-46 mutation spectra in both the presence and the absence of NER revealed an unusually high number of large deletion events, in which 95 or 131 bp were deleted between either a 10- or a 7-bp direct repeat, respectively. The rate of large deletion events in a pol30-46 strain (8.5 x 10–10) was nearly fivefold greater than that in a wild-type strain (1.8 x 10–10). These results were similar to the fourfold increase in large deletion events detected in the rad1 pol30-46 strain (16 x 10–10) relative to the rad1 strain (4.1 x 10–10) and suggest that processive DNA replication may be perturbed in pol30-46 mutants.
Pol32, a subunit of Pol, is required for complex frameshift events:
In addition to PCNA, another protein involved in DNA replication, Pol32, has been placed in the RAD6 epistasis group on the basis of sensitivities of pol32 mutants to DNA-damaging agents (HUANG et al. 2000). Pol32 is a small, nonessential subunit of the replicative DNA polymerase Pol (GERIK et al. 1998) and appears to help tether Pol to PCNA (BURGERS and GERIK 1998; LU et al. 2002). POL32 has been implicated in Pol-dependent mutagenesis, as pol32 mutants display defects in damage-induced mutagenesis and have reduced levels of spontaneous mutagenesis (GERIK et al. 1998; HUANG et al. 2000, 2002). However, it is not clear whether the defects in spontaneous mutagenesis are the direct result of a perturbation in Pol-dependent mutagenesis or the result of a decrease in another type of mutagenesis. We, therefore, created a rad1 pol32 strain and compared the lys2A746 reversion rate and spectrum to those of the rad1 parent strain. Although the overall reversion rate was not significantly affected (Table 3), there was a dramatic change in the reversion spectrum upon deletion of POL32. Of 94 reversion events sequenced, only 1 event was complex (1.1%) in the rad1 pol32 double mutant compared to 36 complex events among 129 sequenced (28%) in the rad1 parent strain (Figure 4). This corresponds to an 30-fold decrease in the rate of accumulation of complex events in the rad1 pol32 double mutant relative to the rad1 strain. In addition to a decrease in complex insertions and consistent with the findings of HUANG et al. (2002), there was also a 2.4-fold increase in the rate of large deletion events in the absence of POL32. The significant increase in large deletion events detected in the rad1 pol32 strain was not a reflection of DNA damage repair capacity of the cell, as NER-proficient pol32 strains also demonstrated a significant 30-fold increase in the rate of large deletion events relative to a wild-type strain (Figure 2). These data indicate that in the absence of POL32, Pol-dependent mutagenesis is compromised and that replication processivity is perturbed.
|
-dependent spontaneous mutagenesis at hotspots 1 and 2 in a rad1 strain:-dependent mutagenesis in srs2 mutants may be too subtle to detect by conventional mutation rate calculations. We, therefore, examined both the overall and the complex frameshift mutation rate of the lys2A746 allele in an srs2 strain.
If lesions accumulating in a rad1 background were substrates for Rad51-dependent recombination as well as Pol-dependent translesion synthesis, one would predict that loss of Srs2 should lead to a decrease in the accumulation of complex frameshift events in an srs2 mutant, as more lesions would be bypassed by a recombination-based mechanism. Consistent with this idea, the rate of complex frameshift events in the rad1 srs2 strain was slightly but significantly less than that in a rad1 strain (Table 3). A much larger decrease in rate of complex frameshift events was seen at hotspots 1 and 2, with a fivefold decrease in the rate of complex frameshifts in the rad1 srs2 strain relative to the rad1 strain (Figure 4; Table 4). This suggests that in the absence of NER, loss of Srs2 prevents Pol-dependent bypass of at least a subset of the lesion(s) accumulating at or near hotspots 1 and 2.
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
A746 allele occurs via a wide variety of net +1 frameshifts within an 150-bp reversion window. Complex reversion events, in which one or more base substitutions accompany the selected frameshift, are of particular significance, as our previous studies have demonstrated that these events are completely dependent on the activity of the Pol translesion DNA polymerase (HARFE and JINKS-ROBERTSON 2000). In this work, we have used this unique mutational signature to probe the relationship between members of the RAD6 epistasis group and spontaneous Pol-dependent mutagenesis.
The current three-pathway model for SLB by members of this epistasis group is based primarily on rates of CAN1 forward mutations, which are assumed to be mostly base substitution events. Two distinct error-prone arms have been proposed—one controlled by the Rad6-Rad18 heterodimer and the other by a Rad5 homodimer—as well as an error-free arm composed of Rad6, Rad18, Rad5, Mms2, Ubc13, PCNA, and Pol (LIEFSHITZ et al. 1998; CEJKA et al. 2001). Both error-prone pathways appear to involve translesion synthesis by Pol, while the error-free pathway is speculated to involve a strand-switching mechanism in which the newly synthesized strand of the sister chromatid is used as a template to bypass the lesion. As predicted by this model, we observed an increase in the lys2A746 reversion rate in the absence of Mms2, Rad18, or Rad5, each of which is in the error-free arm. Although the increase in overall reversion rate was twofold at most, the types of mutations that occurred in a rad18, mms2, or rad5 mutant were dramatically different from those in a wild-type strain. Specifically, there were proportionally many more complex events in the mutant spectra than in the wild-type spectrum. Relative to wild type, the rates of Pol-dependent complex frameshift events increased six- to eightfold, while the rates of noncomplex frameshifts changed little, if any, in the mutants. Whereas elimination of rad5 or rad18 alone caused a striking increase in complex frameshifts, simultaneous deletion of rad5 and rad18 resulted in the complete disappearance of these events. The frameshift spectra thus clearly demonstrate not only that Rad5 and Rad18 participate in an error-free pathway, but also that each protein individually regulates a Pol-dependent error-prone bypass mechanism.
The lys2A746 reversion rate increases observed when Rad18, Rad5, or Mms2 was eliminated in a NER-compromised rad1 background were also consistent with a three-pathway model for spontaneous lesion bypass. In contrast to the RAD1 background, however, where only the rates of complex events were elevated, both complex and noncomplex events were elevated to the same extent in the rad1 rad18 and rad1 rad5 double mutants. At least in the case of the rad1 rad18 mutant, the rate elevation of both types of events was dependent on the presence of Rev3, indicating that Pol can generate noncomplex as well as complex frameshifts in the lys2A746 assay. Relative to the rad1 single mutant and unlike the rad1 rad5 or rad1 rad18 mutant, complex events were elevated more than noncomplex events in the rad1 mms2 mutant, consistent with the involvement of Mms2 only in an error-free bypass pathway (see below for further discussion).
The examination of mutation spectra when RAD6 epistasis group genes were deleted in a NER-defective background proved to be particularly informative. Normally in a rad1 background, complex events compose 28% of the mutant spectrum and accumulate at two distinct hotspots (hotspots 1 and 2). These hotspots are believed to correspond to sites of a discrete type of spontaneous damage (most likely an oxidative lesion; B. MINESINGER, A. ABDULOVIC and S. JINKS-ROBERTSON, unpublished observations) that is specifically removed by the NER pathway. In the absence of NER, the damage presumably is bypassed by a mechanism that leads to the observed accumulation of the Pol-dependent complex events. (However, it is formally possible that these mutations reflect the primary extension of base mispairs by Pol, rather than by a Pol-dependent lesion bypass.) The presence of the distinct hotspots in a rad1 background allowed not only these specific complex events to be monitored but also additional complex hotspots to become evident upon deletion of RAD18, RAD5, or MMS2. In the NER-defective background, rad18 mutants failed to accumulate any complex events at hotspot 1 or 2, while the rate of these events increased twofold upon elimination of Rad5 or Mms2. These data indicate that complex events at hotspots 1 and 2 are generated specifically by the Rad18-dependent error-prone pathway; either the Rad5-dependent error-prone pathway does not bypass the underlying lesion or bypass occurs in a manner not detected by the lys2A746 assay system. In contrast to the rad1 rad18 mutant, the rad1 rad5 strain did not accumulate complex events at a novel hotspot within a 6A run, implying that the Rad18 error-prone pathway does not generate these events. Thus, although both the Rad5- and the Rad18-dependent error-prone pathways promote Pol-dependent lesion bypass, the distinct mutational signatures indicate that these pathways are not simply functionally redundant. In E. coli, the sequence context of a defined lesion appears to determine the specific constellation of proteins required for translesion synthesis (WAGNER et al. 2002). Our data might reflect a similar context-specific bypass of a common lesion or could reflect the bypass of different lesions by the Rad5- vs. Rad18-dependent error-prone pathway. Alternatively, the changes in the distributions of complex events could reflect distinct roles for Rad5 and/or Rad18 when bypassing lesions during normal DNA replication vs. gap filling or during leading- vs. lagging-strand DNA synthesis.
The third branch of the three-pathway model of SLB is an error-free arm regulated by a heterodimer of Rad6-Rad18 and a complex of Rad5-Ubc13-Mms2. Although Ubc13-Mms2 forms a stable complex with Rad5 (ULRICH and JENTSCH 2000; ULRICH 2003), the possibility that Ubc13-Mms2 might also be involved in the Rad5-, Pol-dependent error-prone bypass pathway has not been addressed. The indistinguishable lys2A746 reversion rates of rad5 and mms2 mutants (Tables 2 and 3) would be consistent with a coincident requirement for Rad5 and Ubc13-Mms2 in both error-free and error-prone pathways of SLB. Examination of the corresponding reversion spectra, however, does not support a role for Ubc13-Mms2 in the Rad5-dependent error-prone pathway. This is most evident in a rad18 background, where deletion of MMS2 vs. RAD5 generated reversion spectra containing 30% complex events vs. 0% complex events, respectively (Figure 2). In addition to the clear separation of the roles of Rad5 and Mms2 with regard to spontaneous mutagenesis, strains completely lacking Rad5 are more sensitive to UV than are strains harboring a mutant Rad5 protein that is unable to physically interact with the Mms2-Ubc13 heterodimer (ULRICH 2003). This suggests that there also may be an Mms2-Ubc13 independent role of Rad5 in induced DNA damage tolerance. Given the homology of Rad5 to the Swi/Snf family of chromatin remodeling factors (JOHNSON et al. 1994) and its genetic interaction with histone H2B (MARTINI et al. 2002), it is possible that the Mms2-Ubc13 independent activity of Rad5 may involve chromatin remodeling.
Although there were no complex events among the 94 lys2A746 revertants analyzed in the rad18 rad5 double-mutant background, 7% (7/95) of the revertants isolated in the rad1 rad18 rad5 triple-mutant background harbored a complex frameshift. These data suggest either that repair/bypass factors other than Pol can generate these events or that Pol retains some function even in the absence of both Rad18 and Rad5. In relation to the latter possibility, it has been reported that deletion of RAD5 and RAD18 in a recombination-defective strain (rad51 mutant) does not reduce the CAN1 mutation rate to that of a rad51 rev3 strain, suggesting that a portion of Pol activity remains (LIEFSHITZ et al. 1998). The facilitator (if any) of this Pol-dependent activity in the absence of Rad18 and Rad5 is not known, but may depend upon Rad6, independent of its association with Rad18 (CEJKA et al. 2001). While the relationship between the complex events detected in the triple-mutant strain and Rev3 was not determined due to the poor health of the rad1 rad18 rad5 strain, it should be noted that in every spectrum examined to date, complex events completely disappear upon deletion of REV3.
PCNA interacts with multiple proteins involved in DNA synthesis and DNA repair, presumably targeting/recruiting these proteins as needed to sites of nascent DNA synthesis (MAGA and HUBSCHER 2003). It recently has been shown that post-translation ubiquitination and/or sumoylation of PCNA is important for activating the error-prone and error-free arms of the RAD6 pathway (HOEGE et al. 2002; STELTER and ULRICH 2003; HARACSKA et al. 2004). Current data suggest that either Ubc9-Siz1-dependent sumoylation or Rad6-Rad18-dependent monoubiquitination of amino acid K164 of PCNA (together with sumoylation of K127) leads to Pol-dependent translesion synthesis. Error-free lesion bypass appears to be triggered when the Rad5-Ubc13-Mms2 complex adds multiple K63-linked ubiquitin proteins to the initial monoubiquitin moiety. These data have led to the proposal of a two-pathway model for lesion bypass in response to induced DNA lesions: an error-prone, Pol-dependent pathway regulated by Rad6-Rad18 and an error-free pathway that requires the Rad5-Ubc13-Mms2 complex in addition to the Rad6-Rad18 complex. While a similar model has been proposed to explain the bypass of spontaneous lesions, a two-pathway model does not take into account the Rad5-dependent error prone pathway, which, suggested by others (LIEFSHITZ et al. 1998; CEJKA et al. 2001) and clearly shown here, does not involve the Mms2-Ubc13 complex. Incorporating the data of STELTER and ULRICH (2003) into a three-pathway model, we suggest that simultaneous sumoylation of PCNA at K127 and K164 in the absence of RAD18 (disabling the error-free pathway) leads to Rad5-dependent error-prone bypass, whereas simultaneous sumoylation at K127 and monubiquitination at K164 (i.e., in an mms2, ubc13, or rad5 mutant) leads to Rad18-dependent error-prone bypass (Figure 5). An important area of future studies will be to determine the relevance of the various PCNA modifications to the Rad5- vs. Rad18-dependent error-prone pathways of spontaneous lesion bypass.
|
A746 allele was similar in the pol30-46 and mms2 mutant strains, and we expected the reversion spectra to also be similar. They were significantly different, however, leading to an estimated twofold reduction in the overall rate of complex frameshifts in the pol30-46 mutant relative to the mms2 mutant. Significant differences were also detected in the rate of complex frameshift events in the pol30-46 mutant and the mms2 mutant when NER was compromised, with a sevenfold decrease in complex frameshift events in the rad1 pol30-46 strain relative to the rad1 mms2 strain. The pol30-46 allele thus clearly affects the bypass of spontaneous and induced lesions in fundamentally different ways. It should also be noted that pol30-46 strains may have replication processivity defects (an attribute that had not been detected previously for this allele), as large deletion events increased significantly in pol30-46 strains relative to the parental strains.
In addition to examining the role of PCNA in response to spontaneous Pol-dependent mutagenesis, we have also investigated the role of Pol in this process. Pol is composed of one nonessential (Pol32) and two essential (Pol3 and Pol31) subunits (GERIK et al. 1998). Mutations in POL3 and POL32 are epistatic with rad6 mutations in terms of DNA damage sensitivity and cause defects in induced mutagenesis in response to UV and MNNG damage (GERIK et al. 1998; HARACSKA et al. 2000a; HUANG et al. 2000). Similar decreases in spontaneous mutagenesis have been detected using the CAN1 forward mutation assay, where deletion of POL32 resulted in a weak antimutator phenotype in a wild-type background and completely eliminated the spontaneous mutator phenotype of strains defective in NER (rad1 and rad10) and in lesion bypass (rad6, rad18, and mms2), but not in recombination (rad51 and rad52; HUANG et al. 2000, 2002). Deletion of REV3 conferred similar phenotypes in these backgrounds, suggesting an essential role for Pol32 in the Pol-dependent bypass of spontaneous DNA lesions. Although loss of POL32 did not decrease the spontaneous reversion rate of the lys2A746 allele in a rad1 background, it did completely eliminate complex frameshifts in this assay, indicating that it is indeed essential for Pol-dependent lesion bypass. Given that the Pol32 subunit of Pol interacts with PCNA, Pol, and Srs2 (HUANG et al. 2000; JOHANSSON et al. 2004), Pol32 could be involved in sensing, signaling, or coordinating the replacement of Pol with Pol at sites of blocking damage. Alternatively, Pol32 could be required for Pol to insert a nucleotide opposite a lesion, with the resulting mispaired terminus then being extended by Pol (see HARACSKA et al. 2001, 2003).
The results reported here highlight the complexity of DNA damage tolerance pathways in yeast and underscore differences between lesion bypass in response to spontaneous vs. induced DNA damage. Data obtained using the lys2A746 reversion assay are consistent with a three-pathway model for the bypass of spontaneous lesions by the RAD6 epistasis group (Figure 5). The analysis of reversion spectra in appropriate mutant strains not only has confirmed functional differences between Rad5 and Mms2, but also has revealed clear differences between the Rad18- and Rad5-dependent error-prone bypass pathways. Given that many of the proteins examined in this study are highly conserved, these results will likely be relevant to understanding the role of lesion bypass in spontaneous damage tolerance in higher eukaryotes.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
ADAMS, W. T., and T. R. SKOPEK, 1987 Statistical test for the comparison of samples from mutational spectra. J. Mol. Biol. 194: 391–396.[CrossRef][Medline]
ALTMAN, D. G., 1990 Practical Statistics for Medical Research. CRC Press, New York.
AYYAGARI, R., K. J. IMPELLIZZERI, B. L. YODER, S. L. GARY and P. M. BURGERS, 1995 A mutational analysis of the yeast proliferating cell nuclear antigen indicates distinct roles in DNA replication and DNA repair. Mol. Cell. Biol. 15: 4420–4429.[Abstract]
BAILLY, V., S. LAUDER, S. PRAKASH and L. PRAKASH, 1997 Yeast DNA repair proteins Rad6 and Rad18 form a heterodimer that has ubiquitin conjugating, DNA binding, and ATP hydrolytic activities. J. Biol. Chem. 272: 23360–23365.
BOEKE, J. D., J. TRUEHEART, G. NATSOULIS and G. R. FINK, 1987 5-Fluoroorotic acid as a selective agent in yeast molecular genetics. Methods Enzymol. 154: 164–175.[Medline]
BROOMFIELD, S., and W. XIAO, 2002 Suppression of genetic defects within the RAD6 pathway by srs2 is specific for error-free post-replication repair but not for damage-induced mutagenesis. Nucleic Acids Res. 30: 732–739.
BROOMFIELD, S., B. L. CHOW and W. XIAO, 1998 MMS2, encoding a ubiquitin-conjugating-enzyme-like protein, is a member of the yeast error-free postreplication repair pathway. Proc. Natl. Acad. Sci. USA 95: 5678–5683.
BROOMFIELD, S., T. HRYCIW and W. XIAO, 2001 DNA postreplication repair and mutagenesis in Saccharomyces cerevisiae. Mutat. Res. 486: 167–184.[Medline]
BURGERS, P. M., and K. J. GERIK, 1998 Structure and processivity of two forms of Saccharomyces cerevisiae DNA polymerase . J. Biol. Chem. 273: 19756–19762.
CARIELLO, N. F., W. W. PIEGORSCH, W. T. ADAMS and T. R. SKOPEK, 1994 Computer program for the analysis of mutational spectra: application to p53 mutations. Carcinogenesis 15: 2281–2285.
CEJKA, P., V. VONDREJS and Z. STORCHOVA, 2001 Dissection of the functions of the Saccharomyces cerevisiae RAD6 postreplicative repair group in mutagenesis and UV sensitivity. Genetics 159: 953–963.
DOHMEN, R. J., K. MADURA, B. BARTEL and A. VARSHAVSKY, 1991 The N-end rule is mediated by the UBC2 (Rad6) ubiquitin-conjugating enzyme. Proc. Natl. Acad. Sci. USA 88: 7351–7355.
FRIEDBERG, E. C., G. C. WALKER and W. SIEDE, 1995 DNA Repair and Mutagenesis. ASM Press, Washington, DC.
