INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Organisms possess numerous and complex strategies that function with the common purpose of maintaining the integrity of their genetic material, including those devoted to the removal of damage caused by both endogenous and exogenous sources (i.e., excision repair pathways). However, conditions may arise in which the lesion burden is greater than the capacity of the repair process to remove all the lesions efficiently or the lesion itself may not be recognized as a substrate for any given repair pathway (7). Left unrepaired, damage can interfere with normal DNA metabolism, eventually resulting in mutations that, in higher eukaryotes, can contribute to the development of cancers.

Fortunately, cellular strategies exist to protect the genome from the potentially harmful effects of unrepaired damage. These pathways confer the ability to tolerate persistent damage, enabling replication recovery, continuation of the cell cycle, and, ultimately, survival under conditions of genotoxic stress. A subset of the tolerance mechanisms produces mutations which in unicellular organisms may confer an evolutionary advantage by permitting the organism to adapt to a harsh environment (7, 44).

Strategies for tolerating damage include the potentially mutagenic process of translesion synthesis (TLS), in which the information (or lack thereof) provided by the lesion in the damaged template strand is copied during replication. TLS may be continuous, in which the replication complex has or acquires the ability to continue replication directly across the damaged template, or discontinuous, involving the replicative filling-in of a gap left opposite the lesion by the dissociation and reinitiation of the replication complex downstream of the damage (7). Most, if not all, induced mutagenesis is thought to result from this process (7). Fortunately, mutations are rare events, suggesting that mutagenic TLS is a relatively minor component of damage tolerance; indeed, the majority of replication recovery seems to be carried out by largely error-free processes (7, 19, 49). Two essentially error-free mechanisms that share a fundamental feature, the use of the undamaged complementary sequence to accomplish replication, thereby avoiding the lesion (damage avoidance [DA]), have been proposed. Recombinational strand transfer of DNA from the undamaged duplex to the homologous strand of the damaged duplex is one proposed mechanism (41). A second process, referred to as copy choice or strand switch, postulates that the DNA polymerase (Pol) uses the newly synthesized daughter strand of the undamaged complementary sequence temporarily as a template to detour around the lesion (12). Following lesion clearance, the replicating Pol switches back to copying the original template strand. The details of both damage tolerance strategies (TLS and DA) are not well understood, particularly for eukaryotic cells.

In Escherichia coli, TLS is thought to be carried out by DNA Pol III together with at least three accessory proteins, the SOS-controlled umuDC gene products and the activated form of RecA (7, 49). However, TLS also has an umuDC-independent component (28, 31), pointing to the involvement of as-yet-undiscovered factors and underlining the complexity of this recovery strategy. The processes of DA are less well understood for this organism, but they may be the major means of replication recovery in the presence of a blocking lesion (19).

The budding yeast, Saccharomyces cerevisiae, is an excellent model with which to explore the damage tolerance mechanisms in higher eukaryotes, as the proteins involved in DNA metabolic activities are often very similar. The RAD6 pathway, comprised of components of the damage tolerance strategies, is responsible for a substantial fraction of this yeast's resistance to DNA damage and for almost all induced mutagenesis (22). Less than half the genes implicated in the replication of damaged DNA have been cloned, and in only a few cases has it been possible to demonstrate the biochemical activity of the encoded proteins. However, certain protein functions are progressively being uncovered. For example, the multiple activities exhibited by the purified Rad6-Rad18 heterodimer suggest that this complex may recognize single-stranded DNA associated with replication forks stalled at damage sites (1). Once recruited, the complex may proceed to ubiquitinate components of the replication machinery, enabling replication recovery. Additionally, proliferating cell nuclear antigen (47), together with Pol  (46), has been recently suggested to act in an error-free branch of the RAD6-dependent pathways.

An understanding of the potentially mutagenic mechanism of TLS is more advanced, owing mostly to in vitro primer extension studies using damaged templates and purified gene products. The Rev7p and the nonessential Rev3 DNA Pol associate to form a complex designated Pol , whose sole function may be to replicate through damage sites (25). Pol  was able to perform TLS across a pyrimidine dimer-bearing primer template substrate (33). The REV1 gene product has been shown to insert cytosine preferentially across from a template abasic site, creating a lesion terminus that can be readily extended by Pol  (32). Interestingly, human Rev1p and Rev3p homologs may exist (25), rendering facets of yeast TLS potentially applicable to TLS operating in higher eukaryotes. However, in spite of the insights gathered from these in vitro studies, Rev protein-dependent TLS has not yet been demonstrated in vivo.

The current investigation was undertaken in an attempt to determine the frequencies with which the two major damage tolerance strategies (TLS and DA) are used in a eukaryotic cell faced with the problem of having to replicate a lesion-bearing, double-stranded DNA molecule. Replication of unmodified and heteroduplex plasmids carrying a single N-2-acetylaminofluorene (AAF) adduct situated at the third guanine of the sequence 5' GGG^AAF 3' was examined in REV3 and rev3 strains deficient in both nucleotide excision (rad10; NER) and mismatch (msh2) repair, respectively. Once the mechanisms used to complete damaged-plasmid replication were determined, the proposed role of the Rev3 component of Pol  in in vivo TLS was investigated. We present biochemical evidence of what appears to be an absolute requirement for Rev3p activity in TLS of the strong replication-blocking AAF adduct in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Yeast strains and media. Yeast strains are derived from FL200 strains MAT ura3 trp1-4 leu2-1 (40). Both the REV3 and rev3 strains were in a rad10 msh2 background to avoid repair of the AAF adduct (48) and the 4-bp sequence heterology (36), respectively. All strains were constructed by the one-step gene disruption method (39). The rad10::LEU2 deletion was made by cloning the BamHI-HindIII LEU2 gene fragment of pFL33 (constructed by A. Bresson-Roy) into the XbaI site of the SalI-EcoRV RAD10 fragment (from pTB215; kindly provided by L. Prakash) previously cloned into the EcoRI site of plasmid pBR322. Inactivation of the MSH2 gene was accomplished by transforming rad10::LEU2 cells with the 9.5-kb SpeI fragment containing msh2::Tn10LUK obtained from restriction enzyme digestion of pII-2-Tn10LUK7-7 (generously provided by R. Kolodner). The rev3 strain was obtained by electroporating rad10::LEU2 msh2::Tn10LUK cells with the KpnI restriction fragment of the vector pDG347 (a generous gift from R. D. Gietz) containing the rev3::hisG-URA3-hisG cassette. Chromosomal integration and gene inactivation were confirmed by spore segregation analysis, mutant phenotype characterization, and Southern blot analysis (data not shown).

Bacterial strains and media. The E. coli strains used were either TB1 (F ara[lac-proAB] rpsL 80[lac(lacZ)M15] thi hsdR or XL1-Blue [F'::Tn10 proA⁺ B⁺ lacI^q (lacZ)M15/recA1 endA1 gyrA96 thi hsdR17 supE44 relA1 lac] (42). Plasmids isolated from yeast for further analysis in E. coli (13) (see below) were transformed by electroporation into the above strains with the Bio-Rad gene pulser. Transformants were plated onto Luria-Bertani medium containing 100 µg of ampicillin per ml (for transformant selection) as well as isopropyl -D-thiogalactopyranoside (IPTG) and 5-bromo-4-chloro-3-indolyl--D-galactoside (X-Gal) to detect Lac⁺ colonies indicative of a 1G frameshift mutation having occurred in the sequence 5' GGG^AAF 3' during plasmid replication in yeast (described in reference 19).

Plasmids. The E. coli-S. cerevisiae shuttle vector pKBL10 was derived from pUC8. The HpaI-NdeI fragment of the 2µm plasmid digested from plasmid pFL45L (6), containing both the replication origin (ARS consensus sequence) and REP3 sequences, was cloned into the NdeI site of pUC8. The yeast TRP1 selective marker (Klenow fragment-filled BglII fragment from pFL39 [6]) was introduced into the Klenow fragment-filled AflIII site of pUC8. The pKBL(Helper), pKBL(3G), and pKBL(3G+3) parental plasmids were made by replacing the EcoRI-HindIII fragment of pKBL10 with duplex oligonucleotides containing the helper, 3G, or 3G+3 sequences, respectively (Fig. 1A).

Construction of heteroduplexes containing a single AAF adduct. The strategy used to construct plasmids containing a single AAF adduct on the 3' guanine of the sequence 5' GGG 3', itself located within a sequence heterology, involved the formation of gapped-duplex molecules described previously (19). The only modification was that the resulting adduct-containing heteroduplex plasmids were treated with Plasmid-Safe ATP-dependent DNase (Epicentre Technologies) after recovery from CsCl gradients in order to eliminate contaminating linear, double-stranded, homoduplex DNA which can transform yeast, albeit with a lower efficiency than that of closed, circular molecules (4, 39). Potential contamination by undigested pKBL(Helper) plasmids was minimized by HincII digestion prior to DNase treatment.

Strand segregation analysis (SSA). The following technique enables sensitive and reproducible detection of the hybridization between a short, 5'-³²P-labeled oligonucleotide and target plasmid sequences isolated from yeast transformants. Individual colonies issued from the transformation with heteroduplex constructions were picked up from SCTrp plates and subsequently grown for approximately 24 h in 150 µl of SCTrp in 96-well microtitration plates covered with sterile adhesive tape. Prior to lysis, cells were replica plated with a 48-prong manual replicator onto SCTrp master plates so that plasmids having undergone TLS could be isolated (13) and transformed into E. coli for mutational analysis. The culture dishes were centrifuged at 3,000 rpm for 5 to 10 min in a Jouan GR2000S centrifuge to pellet the cells. Culture medium was aspirated off prior to spheroplast production to prevent inhibition of zymolyase activity. Spheroplasts were generated by enzymatic digestion for >30 min at 37°C by resuspending the above cell pellets in 100 µl of a digestion buffer containing 1 M sorbitol, 100 mM Na citrate, 50 mM EDTA, 14 mM -mercaptoethanol, and 1 mg of Zymolyase 100T (Seikaguku) per ml. Spheroplast lysis was performed in the culture plaques by first adding, to the 100-µl spheroplast mixture, a 2× solution yielding a final concentration of 400 mM NaOH and 20 mM EDTA; this was followed by heating for approximately 3 min at 100°C. A microfiltration slot blot apparatus (Bio-Dot; Bio-Rad) was used to isolate total DNA (genomic and plasmid) onto a nylon membrane (Dupont/NEN) under vacuum pressure. The DNA was fixed to the filters with a solution of 400 mM NaOH, after which the membranes were briefly rinsed in 2× SSC. Hybridizations were performed according to the membrane manufacturer's instructions (Dupont) with 20-mer oligonucleotide probes 3G (specific for the lesion-containing strand) and 3G+3 (specific for the strand marker) (Fig. 1C) at a hybridization temperature of 60°C. Dehybridization was done by placing the membranes in a boiling solution of 0.5% sodium dodecyl sulfate for 5 to 10 min followed by cooling them to room temperature. Both hybridization patterns and dehybridization efficiency were evaluated with a Molecular Dynamics PhosphorImager. Hybridization signals of the negative controls (i.e., cells transformed with 3G+3 parental plasmids for 3G hybridization and vice versa for 3G+3 hybridization) were quantified by using the Image Quant function of Molecular Dynamics. The average of these measurements and its variance were calculated. All hybridization signals greater than the average negative control value plus two times the variance were considered positive. SSAs were systematically performed on clones arising from a minimum of four independent transformations per plasmid for each yeast strain.

Mutational analysis. An initial screen for the 1G mutation in yeast was undertaken by hybridizing filters having previously undergone SSA (see above) with the 2G mutant-specific probe (Fig. 1C) at 62°C. A secondary analysis consisted of transforming E. coli with plasmids recovered from all yeast clones (grown on SCTrp replica plates) having responded positively to both or either of the 3G and 2G probes. Sequence analysis (42) was performed on approximately 10 blue (mutant) and 10 white (nonmutant) bacterial colonies in order both to characterize the mutation and to confirm SSA in yeast, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

SSA. The present analysis in yeast cells was undertaken in a manner similar to that reported previously for E. coli (19) and is based on the following strategy: under normal circumstances, a single round of replication on a double-stranded DNA molecule yields two daughter molecules containing the genetic information derived from each original strand of the parent duplex. If the parent is a homoduplex, the daughters will be identical to each other and to the parent. However, if the parent molecule is a heteroduplex, the resulting progeny molecules will contain the genetic information of each original strand and will no longer be similar to each other nor to the original heteroduplex. This principle was used to study the damage tolerance strategies proposed to function in a eukaryote system when replication is forced to occur on a damaged DNA template. To this end, plasmids which contained, in the complementary strand directly opposite the adduct site, four extrahelical bases were constructed (Fig. 1B). Short 5'-³²P-labeled oligonucleotides specific to each strand of the original, monomodified heteroduplex (probes 3G and 3G+3 [Fig. 1C]) were used to probe the progeny plasmids resulting from its replication. With this approach, the fate of each strand of a modified and unmodified heteroduplex could be monitored at the level of the strand marker after replication in individual yeast colonies. This enabled the determination of frequencies with which TLS and DA are used to complete replication of a bulky lesion-containing plasmid.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Vector design and hybridization strategy used in SSA. (A) Parental plasmids used to construct monomodified heteroduplexes and as hybridization controls. (B) Location of the strand marker with respect to the AAF adduct (asterisk). (C) Oligonucleotide probes to analyze strand segregation (3G and 3G+3) and 1G mutagenesis (2G).

Replication of the unmodified heteroduplex vector. SSA was initially carried out on the REV3 and rev3 strains transformed with unmodified heteroduplex vectors as a control to compare with replication of the lesion-containing vector. Neither genomic DNA nor the endogenous 2µm plasmid interfered with hybridization analysis, as untransformed cells subjected to SSA did not hybridize with either the 3G or the 3G+3 probe (data not shown). The results are summarized in Table 1, and an example hybridization analysis performed on one filter for each strain is shown in Fig. 2. In both the REV3 and rev3 strains, an average of 80% of the transformants hybridized with both the 3G and 3G+3 probes (mixed colonies [Table 1]), indicating that both strands of the original heteroduplex molecule had been replicated to approximately the same extent, confirming the nonessential role of Rev3p in plasmid replication (37, 38). Of the remaining 20%, approximately half hybridized with only the 3G or the 3G+3 probe, respectively (Table 1) (3G and 3G+3 pure colonies). The presence of only one strand of the original heteroduplex in these clones could reflect prereplicative MSH2-independent mismatch repair. An activity that binds to heteroduplexes containing four to nine extra bases has been previously observed in cellular extracts prepared from yeast containing a mutation in the MSH2 gene (29). However, it is also possible that these clones arise from the preferential replication of one strand relative to the other. Regardless of their origin, the activity giving rise to the pure colonies was not dependent on functional Rev3p, as their total proportion (i.e., pure versus mixed) remained essentially the same in both strains (Table 1).

View this table: [in this window] [in a new window]   TABLE 1.   SSA of the unmodified heteroduplex plasmid in REV3 and rev3 strainsa

View larger version (57K): [in this window] [in a new window]   FIG. 2.   SSA of strains transformed with the nonmodified heteroduplex plasmids. (A) Molecular Dynamics PhosphorImager image of an analysis performed on REV3 clones. At left is the hybridization pattern observed with the 3G probe. To the right is the same filter hybridized with the 3G+3-specific probe. (B) The same analysis as shown in panel A for the rev3 strain. Hybridization controls (parental plasmids 3G and 3G+3 [Fig. 1A]) are indicated at the top of each filter (rectangles).

Replication of the AAF-modified plasmid. AAF-monomodified heteroduplex plasmids were transformed into the above yeast strains and analyzed as described for the unmodified vector. AAF was the lesion of choice, as the chemical adducts it forms in DNA are stable and well characterized and have been shown to be strong blocks to replication (5, 14, 19, 27, 31, 45). Moreover, AAF-monomodified plasmids have been successfully used as tools to investigate damage tolerance strategies in E. coli (19, 31). The total number of transformants obtained with both the unmodified and the modified plasmid preparations was the same for both strains (data not shown), thereby eliminating a possible bias in the observations resulting from adduct toxicity. As can be seen in Table 2 and Fig. 3A, the presence of a single AAF lesion had a marked effect on plasmid replication in the REV3 strain (compare with Table 1 and Fig. 2A). An average of 92% of the transformant colonies contained only the 3G+3 strand (versus 14% observed with the unmodified plasmid [Table 1]), while the proportion of colonies responding positively to both probes was reduced from 78% in the absence of the lesion to 6% in its presence. Interestingly, 2% of the colonies appear to be 3G pure (Table 2).

View this table: [in this window] [in a new window]   TABLE 2.   SSA of REV3 and rev3 strains transformed with monomodified heteroduplexesa

View larger version (64K): [in this window] [in a new window]   FIG. 3.   SSA of clones transformed with monomodified heteroduplex plasmids. (A) Example filter of REV3 clones analyzed, at left, with the 3G probe specific to the lesion-containing strand and, at right, with the 3G+3 probe. (B) As shown for panel A, with the rev3 strain. Circles represent those clones in which TLS has occurred. Controls are as described for Fig. 2. Note that the 2G control cross-hybridizes with the 3G probe.

Fidelity of TLS. The dG-C8-AAF located at the 3' end of a run of three consecutive guanines (5' GGG^AAF 3') has been previously characterized as inducing 1G frameshift mutations in both E. coli (21) and yeast (4, 40) at high frequencies. However, in this particular experimental system, phenotypic detection of mutagenic events arising in yeast transformants was not possible. An initial hybridization screen of yeast filters previously subjected to SSA was performed with the 1 frameshift mutation-specific 2G probe, but cross-hybridization of this probe with the wild-type 3G sequence rendered positive mutation identification uncertain (data not shown). However, the guanine run is located in the lacZ' gene in a +1 reading frame, enabling easy phenotypic detection of 1 frameshift events in E. coli via the so-called lacZ complementation assay. Since mutations are thought to arise during TLS (21), all plasmids hybridizing with the 3G probe (mixed and 3G pure [Table 2]), regardless of their hybridization response to the mutant probe, were subjected to further analysis in E. coli (see Materials and Methods). Of the TLS events observed in the REV3 strain (122 of 1,558, or 8%), only 3% were mutagenic (4 of 122) (Fig. 4). Among the four mutants recovered, three were the expected 1G; the remaining mutant was an untargeted 1C (previously observed in this sequence context in E. coli [21] and yeast [4]). Interestingly, the four mutants recovered were identified in SSA as being mixed (3G and 3G+3 positive [Fig. 3]) and gave variable but inconclusive hybridization signals with the 2G mutation-specific probe (data not shown). Secondary transformation of E. coli with plasmids isolated from these mutants gave rise to transformant populations in which the total number of blue (mutant) colonies was different for each yeast clone but was always smaller than the white (nonmutant) population, consisting of 3G and 3G+3 homoduplex plasmids (confirmed by sequencing analysis [data not shown]).

View larger version (44K): [in this window] [in a new window]   FIG. 4.   Comparison of the proportions of DA and error-free and mutagenic TLS observed in yeast (rad10 msh2) and SOS-induced E. coli (uvrA mutS) strains transformed with the AAF-modified heteroduplex vector. Results for E. coli are taken from reference 19. The percentages represent the contribution of each strategy to the total events detected with SSA. DA includes all processes giving rise to 3G+3 pure clones, including lesion-induced strand loss (see text for details).

Dependence of TLS on the Rev3 Pol component of Pol . The possible involvement of the Rev3 Pol in the TLS observed in the REV3 strain was examined by performing SSA on the rev3 mutant strain transformed with the AAF-modified heteroduplex plasmid. The results show that, in the absence of functional Rev3 protein, all TLS, indicated by mixed and 3G pure clones, is abolished (Table 2 and Fig. 3B). Completion of plasmid replication in the rev3 mutant occurred 100% of the time by a mechanism employing, at least locally at the adduct site, the complementary, lesion-free strand. As the unmodified plasmid is replicated in a similar manner in both REV3 and rev3 strains (Table 1), these results are not due to a general Rev3-dependent defect in plasmid replication.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

TLS of the AAF-modified sequence (5' GGG^AAF 3') is infrequent and relatively error free. Replication of the AAF-monomodified plasmid in the REV3 strain was completed by a mechanism using, at least locally at the lesion site, the information provided by the undamaged strand in more than 90% of the clones analyzed (Table 2; Fig. 3A). Mechanisms avoiding direct copying of unrepaired damage include recombinational daughter strand gap filling (41) and Pol strand switch (12). Either one of these mechanisms could conceivably give rise to the large proportion of 3G+3 pure clones observed in both REV3 and rev3 strains. However, an alternative explanation relevant to the plasmid situation that could equally account for the high proportion of 3G+3 pure clones might be the preferential replication and concomitant amplification of the undamaged strand. Lesion-induced strand loss was proposed to explain the prevalence of the undamaged strand in transformants arising from the replication of vectors containing either AAF adducts (20, 45) or a single cis-syn cyclobutane thymine dimer (15) in E. coli. It is not possible, with the present constructs, to distinguish between true DA events and lesion-induced strand loss.

Dependence of TLS on Rev3p. The hypomutator phenotype exhibited by rev3 mutants with respect to both spontaneous (37, 38) and certain agent-induced (26, 30, 34, 35) mutagenesis, together with it being a nonessential, distant member of the  family of DNA Pols (30), suggested that yeast, unlike E. coli, required a specialized Pol to replicate across damage sites. This has been substantiated with the demonstration of Rev3p-dependent TLS in vitro (32, 33). In this study, elimination of the Rev3p component of Pol  abolished all TLS (Table 2; Fig. 3B), identifying Rev3p, and probably by extension Pol , as being required for the TLS observed at the dG-AAF located in the sequence 5' GGG^AAF 3' on a double-stranded plasmid.