Requirement of DNA Polymerase η for Error-Free Bypass of UV-Induced CC and TC Photoproducts

Abstract

The yeast RAD30-encoded DNA polymerase η (Polη) bypasses a cis-syn thymine-thymine dimer efficiently and accurately. Human DNA polymerase η functions similarly in the bypass of this lesion, and mutations in human Polη result in the cancer prone syndrome, the variant form of xeroderma pigmentosum. UV light, however, also elicits the formation of cis-syn cyclobutane dimers and (6-4) photoproducts at 5′-CC-3′ and 5′-TC-3′ sites, and in both yeast and human DNA, UV-induced mutations occur primarily by 3′ C to T transitions. Genetic studies presented here reveal a role for yeast Polη in the error-free bypass of cyclobutane dimers and (6-4) photoproducts formed at CC and TC sites. Thus, by preventing UV mutagenesis at a wide spectrum of dipyrimidine sites, Polη plays a pivotal role in minimizing the incidence of sunlight-induced skin cancers in humans.

The UV component of sunlight is a major epidemiological risk factor for skin cancers that include melanomas, basal cell carcinomas, and squamous cell carcinomas. In the United States, the frequency of skin cancers approaches that of all other cancers combined and is on the rise because of the depletion of the ozone layer (10, 21, 23). UV-induced DNA lesions are removed by nucleotide excision repair, but if left unrepaired, they present a block to normal DNA replication. The yeast RAD30 and human RAD30A genes encode a DNA polymerase, polymerase η (Polη), which has the unique ability to efficiently and accurately replicate through a UV-induced cis-syn thymine-thymine (TT) dimer, and defects in hRAD30A cause the variant form of xeroderma pigmentosum (12, 13, 16, 22, 27). Xeroderma pigmentosum XPV patients suffer from highly elevated levels of sunlight-induced skin cancers.

Because of the efficient insertion of As opposite the TT dimer, it has been suggested that Polη is an A rule polymerase (8). In addition to cyclobutane dimers at two adjacent thymines, UV also induces the formation of lesions at dipyrimidine sites that involve a cytosine, most commonly at 5′-TC-3′ and 5′-CC-3′ sequences. In fact, the 3′ cytosine in both sequence contexts is highly mutagenic, and in both yeast and humans, UV-induced mutations occur primarily by a C→T transition that would result from the insertion of an A opposite the 3′ damaged C residue during DNA replication (1, 3, 6). If Polη were an A rule polymerase which inserts an A residue by default opposite the various lesions, then the bypass of a CC or a TC cyclobutane dimer by Polη would be mutagenic, not error free as for the TT dimer.

In addition to cis-syn cyclobutane pyrimidine dimers, UV induces the formation of pyrimidine (6-4) pyrimidinone photoproducts. The (6-4) photoproduct is formed most frequently at a TC site, whereas the dimer is formed more frequently than the (6-4) photoproduct at a CC site (2, 4, 5). The C of a cis-syn cyclobutane dimer, however, is quite unstable, and in vitro it deaminates to U (24, 25), thus making in vitro bypass studies with TC or CC dimers difficult. Here, we utilize a genetic system designed to test for the role of yeast Polη in the bypass of UV-induced lesions at 5′-TC-3′ and 5′-CC-3′ sites. We find that UV-induced mutations occur at the 3′ C of TC and CC sequences, and importantly, the incidence of these mutations is about fivefold higher in the rad30Δ strain than in the wild-type strain. These studies provide evidence for the requirement of Polη in error-free bypass of cyclobutane dimers and (6-4) photoproducts formed at TC and CC sites.

MATERIALS AND METHODS

Generation of deletions of yeast genes.

All yeast strains were derived from EMY74.7 (MATa his3Δ-100 leu2-3,112 trp1Δ ura3-52). The rad30Δ mutation was generated as described previously (14). To generate the rev3Δ mutation, a 4.6-kb DNA fragment containing the entire REV3 gene was first cloned into pUC19. The internal 3.5-kb Bst11075 fragment of REV3 was then replaced by the URA3 gene blaster fragment, generating plasmid pPM292. This plasmid, when cut with EcoRI and BamHI and transformed into yeast, deletes genomic REV3 from nucleotide +436 to +3928 of the 4,152-bp open reading frame (ORF). To generate the ura3Δ mutation, 889-bp and 747-bp PCR products corresponding to the 5′ and 3′ regions, respectively, of the URA3 gene were amplified from yeast genomic DNA and directionally cloned into pUC19. The yeast HIS3 gene was then inserted between these two PCR products, generating plasmid pPM1048. This plasmid, when digested with the restriction enzymes Asp718 and SalI and transformed into yeast, deletes from nucleotide +24 to +774 of the 804-bp URA3 ORF. Deletions were confirmed by PCR analysis of yeast genomic DNA. Loss of the URA3 gene derived from the gene blaster fragment in the various yeast strains was selected for by plating on medium containing 5-fluoro-orotic acid. To determine UV-induced reversion frequencies, yeast strains were transformed to TRP⁺ with plasmids pPM1020 and pPM1021, harboring the ura3-210 and ura3-364 mutant alleles, respectively, which were constructed as described below.

Construction of the ura3-210 and ura3-364 mutations.

Mutations were introduced into the URA3 gene in pUC19-based plasmid YIplac211 (9) using the MORPH system site-specific mutagenesis kit (5 Prime → 3 Prime, Inc., Boulder, Colo.). For the ura3-210 mutation, the oligonucleotide N4675 (5′-GAAGCATTAGGTCCCAAAATTCGTTTACTAAAAACACATGTGG-3′) was used, and for the ura3-364 mutation, the oligonucleotide N4676 (5′-CCGCCAAGTACAATTTTTTACCGTTCGAAGACAGAAAATTTGCTG-3′) was used. The ura3-210 mutation is a T→C missense mutation at position +166 in the URA3 ORF that creates a Cys₅₆→Arg₅₆ change in the encoded protein. The ura3-364 mutation is a T→C missense mutation at position 263 in the URA3 gene. In this case, the mutation results in a Leu₈₈→Pro₈₈ change in Ura3 protein. The ura3-364 allele also contains a C→G silent mutation at position 264 within the proline codon that eliminates a second potential CC site. Both alleles were sequenced and found not to contain any other mutations. Subsequently, the ura3-210 and ura3-364 alleles were used to replace the wild-type URA3 allele in the yeast CEN/ARS plasmid YCplac33 (9) containing the TRP1 gene as a selectable marker, generating plasmids pPM1020 and pPM1021, respectively. The phenotype of the ura3-210 and ura3-364 alleles was confirmed by the inability of yeast strains harboring either pPM1020 or pPM1021 to grow on media lacking uracil and by resistance to 5-fluoro-orotic acid.

UV mutagenesis and sequence analysis of URA3 revertants.

Yeast strains lacking the genomic URA3 gene (ura3Δ::HIS3) and harboring plasmid pPM1020 or pPM1021 were grown overnight at 30°C in synthetic complete medium lacking tryptophan (SC−trp). Cells were harvested by centrifugation, washed, and resuspended to a density of 10⁸ cells/ml in sterile H₂O, and dilutions were plated on the appropriate medium. Plates were irradiated with UV light at a dose of 1 J/m²/s under yellow light and incubated for 3 to 5 days at 30°C in the dark. For UV-induced mutagenesis, cells were plated on SC−trp for viability determinations and plated on SC−trp that also lacked uracil (SC−trp−ura) to determine reversion frequencies of ura3-210 and ura3-364 alleles. The frequency of spontaneous URA3 revertants was subtracted from the frequency of revertants observed following UV irradiation. To determine the sequence of revertants induced by UV light, plasmid DNA was isolated from UV-induced URA3⁺ yeast colonies, and the URA3 gene was amplified by PCR. PCR products were then sequenced in the regions corresponding to the respective mutations using the thermosequenase kit from Amersham and α-³³P-labeled dideoxynucleoside triphosphates.

RESULTS

To determine whether Polη is involved in the error-free bypass of UV-induced TC and CC photoproducts, we designed a genetic assay for Saccharomyces cerevisiae that utilizes the ura3-210 and ura3-364 mutant alleles, which completely inactivate URA3 function (19). The ura3-210 allele has a T to C mutation at position 166 in the URA3 gene, and this mutation is in a 5′-TC-3′ sequence. The resulting mutant protein contains an arginine at position 56 instead of the normal cysteine residue (Fig. 1). Reversion of this mutant allele to the wild-type URA3 gene occurs by the incorporation of an A residue opposite the 3′ C of the TC UV-induced lesion, restoring the cysteine codon. The ura3-364 allele is also a T to C mutation, but at position 263 in the URA3 gene, and this mutation is in a CC sequence. The mutant protein contains a proline at position 88 instead of leucine (Fig. 1). The ura3-364 allele also reverts to the wild type by the incorporation of an A opposite the 3′ C of the CC UV-induced lesion, but reversion of this allele can also occur by the incorporation of two As opposite the two Cs in the CC sequence.

FIG. 1.

Reversion assay for ura3-210 and ura3-364. The region of the URA3 gene corresponding to the ura3-210 and ura3-364 mutations is shown. The asterisk indicates the position of a T→C mutation. The amino acid substitution in each mutant is circled. Following UV irradiation, a TC photoproduct in ura3-210 or a CC photoproduct in ura3-364 can be formed and is shown in bold with a caret over it. Replication across either lesion by the insertion of an A residue opposite the 3′ C results in reversion of the alleles to a Ura⁺ phenotype.

The ura3-210 and ura3-364 alleles were incorporated into a low-copy-number yeast plasmid containing the TRP1 gene for selection. Wild-type, rad30Δ, rev3Δ, and rad30Δ rev3Δ yeast strains lacking the genomic URA3 gene but harboring the mutant ura3 gene on a plasmid were irradiated with UV light, and the frequency of URA3 revertants was determined. As shown in Fig. 2, UV sensitivity was increased in rev3Δ and rad30Δ single mutants, and the UV sensitivity of the rad30Δ rev3Δ double mutant was greater than that of the single mutants, consistent with the involvement of the RAD30 and REV3 genes in alternate pathways for bypassing UV lesions. UV irradiation induces the reversion of the ura3-210 allele in both the wild-type and rad30Δ strains. However, at all UV doses, the reversion frequency was about fivefold higher in the rad30Δ strain than in the wild type (Fig. 3), indicating a role for Polη in the incorporation of the correct G residue opposite the 3′ C residue in TC photoproducts. These UV-induced mutations depend upon the REV3 gene, since few or no mutations occurred in the rad30Δ rev3Δ strain (see the legend to Fig. 3). Sequence analysis of 32 independent revertants from the wild-type and rad30Δ strains revealed that all UV-induced mutations occur by a 3′ C to T change (Table 1), indicating the incorporation of an A opposite the 3′ C.

FIG. 2.

Survival after UV irradiation of various yeast strains lacking the genomic URA3 gene and harboring the plasmid pPM1020, which carries the ura3-210 mutation. Cells were UV irradiated and plated on SC−trp for viability determinations. Each curve represents the average of three independent experiments. Symbols: ●, wild type; ■, rev3Δ; ○, rad30Δ; □, rad30Δ rev3Δ.

FIG. 3.

UV-induced reversion of the ura3-210 allele in wild-type (white bar), rad30Δ (gray bar), and rad30Δ rev3Δ yeast strains. Yeast strains lacking the genomic URA3 gene and harboring plasmid pPM1020, which carries the ura3-210 mutation, were UV irradiated, and the average frequency of URA3 revertants was determined from results of three independent experiments. Spontaneous mutation frequencies ranged from 0.1 × 10⁻⁷ to 1 × 10⁻⁷ for the various experiments and were similar for the different strains. The standard error is shown for each determination. The frequency of UV-induced revertants at 5, 10, 15, and 20 J/m² for the rad30Δ rev3Δ strain was ≤1 in 10⁷. Also, few or no UV-induced revertants were observed for the rev3Δ strain.

TABLE 1.

Sequence analysis of UV-induced (20 J/m²) revertants for wild-type (RAD⁺) and rad30Δ yeast strains

Strain	Genotype	ura3 allele	Reversion type	No. of revertants/total
YSL903	Wild type	ura3-210	C→T	32/32
YSL905	rad30Δ	ura3-210	C→T	32/32
YSL911	Wild type	ura3-364	CC→CT	30/32
			CC→TT	2/32
YSL913	rad30Δ	ura3-364	CC→CT	27/28
			CC→TT	1/28

The ura3-364 allele containing the CC sequence also reverts in a UV dose-dependent manner (Fig. 4). In this case, however, the frequency of revertants was about fivefold lower than that for the TC sequence in ura3-210. In both yeast and mammalian cells, UV-induced TC photoproducts are often much more mutagenic than the CC photoproducts (1, 3, 6). This may be due to the more frequent formation of (6-4) photoproduct at the TC site than at the CC site (2, 4, 5). UV-induced reversion of ura3-364 was four- to fivefold higher in the rad30Δ strain than in the wild-type strain (Fig. 4), indicating a role for Polη in the accurate bypass of CC photoproducts. Again, few or no UV-induced revertants were formed in the rad30Δ rev3Δ strain (Fig. 4). In both the wild-type and rad30Δ strains, as was observed for the TC sequence, reversion of the CC sequence also occurs by a 3′ C→T change, indicating the incorporation of an A opposite the 3′ C (Table 1). In two cases for the wild-type strain, however, a C→T mutation was additionally observed at the 5′ C in the CC sequence. The occurrence of a tandem CC→TT mutation was also observed for the rad30Δ strain (Table 1).

FIG. 4.

UV-induced reversion of the ura3-364 allele in wild-type (white bar), rad30Δ (gray bar), and rad30Δ rev3Δ (hatched bar) strains. Yeast strains lacking the genomic URA3 gene and harboring plasmid pPM1021, which carries the ura3-364 allele, were UV irradiated, and the average frequency of URA3 revertants was determined from results of three independent experiments. Spontaneous mutation frequencies ranged from 0.05 × 10⁻⁷ to 0.2 × 10⁻⁷ for the various experiments and were similar for the different strains. The standard error is shown for each determination. No UV-induced revertants were observed for the rad30Δ rev3Δ strain at 5, 10, and 15 J/m². Also, few or no UV-induced revertants were observed for the rev3Δ strain.

DISCUSSION

We have developed a genetic system to measure the frequency of UV-induced mutations at the 3′ C of 5′-TC-3′ and 5′-CC-3′ sequences and find that the incidence of these mutations is about fivefold higher for the rad30Δ strain than for the wild-type strain. These data provide strong evidence for a role of Polη in the error-free bypass of UV lesions at TC and CC sites. The (6-4) photoproduct is formed more frequently at the TC site, and the cis-syn cyclobutane dimer is formed more frequently at the CC site (2, 4, 5); thus, these results indicate a role for Polη in the error-free bypass of cyclobutane dimers and of (6-4) photoproducts at these sites. Hence, the ability of Polη for error-free bypass of UV lesions is not restricted to the TT dimer.

These observations also have a bearing on the role of cytosine deamination in UV mutagenesis of CC and TC sites in eukaryotic cells. In Escherichia coli, such deamination has been proposed to account for the C→T transitions which then occur because of accurate insertion of adenines opposite uracils resulting from cytosine deamination (26). If Polη were to replicate through uracil-containing dimers by inserting an A opposite a U, as it does so efficiently for the TT dimer, in that case, inactivation of Polη should reduce the frequency of UV-induced mutations at the TC and CC sites. The fact that 3′ C→T transitions rise approximately fivefold in the rad30Δ strain indicates that Polη bypasses UV lesions at these sites by inserting the correct nucleotide G opposite the Cs. Also, the almost absolute requirement of REV3-encoded Polζ for C→T transitions at these sites points to the active involvement of this polymerase in the mutagenic bypass of TC and CC photoproducts wherein an A is incorporated opposite the 3′ C. Thus, in eukaryotic cells, cytosines in dimers must persist long enough for Polη- and Polζ-dependent damage bypass to occur.

A number of physical studies have indicated that a cis-syn TT dimer has only a modest effect on DNA structure, and this distortion does not affect the ability of the two Ts in the dimer to base pair with As (7, 11, 18). By contrast, a (6-4) TT photoproduct induces a large structural distortion in the DNA helix, and the 3′ T in this lesion is perpendicular to the 5′ T (17, 20). Nuclear magnetic resonance studies have also indicated that the O2 carbonyl of the 3′ T in the (6-4) TT lesion can form a stable hydrogen bond with the imino and amino protons of an opposed G residue (20). The (6-4) lesion at the TC site is expected to be structurally very similar to that at the TT site, and the O2 carbonyl of the 3′ C in the TC (6-4) lesion is also predicted to form a hydrogen bond with the G residue. Incorporation of a G opposite the 3′ C by Polη would result in error-free bypass of a TC (6-4) lesion. In support of this premise, our steady-state kinetic studies with purified yeast and human Polη indicate that they both insert a G opposite the 3′ T of the (6-4) TT photoproduct fairly efficiently, but Polη is unable to insert the subsequent nucleotide; Polζ then inserts an A opposite the 5′ T of the lesion, thereby completing the bypass process (R. E. Johnson, L. Haracska, S. Prakash, and L. Prakash, unpublished observations). Similarly, opposite a (6-4) TC photoproduct, we expect Polη to insert a G opposite the 3′ C and Polζ to extend by inserting an A opposite the 5′ T of this lesion. The sequential action of Polη and Polζ will then coordinate the error-free bypass of (6-4) TC photoproducts. Thus, although Polη would function in an error-prone manner in the bypass of the rare (6-4) TT photoproduct, it would promote error-free bypass of the more frequently formed (6-4) photoproducts at the TC site and those formed at the CC site. The insertion of a nucleotide by Polη opposite the 3′ site in the (6-4) photoproduct suggests that its active site can conform to the severe distortion conferred upon DNA by this lesion.

By analogy to the efficient and accurate bypass of a cis-syn TT dimer, we expect Polη to bypass a cis-syn CC or TC cyclobutane dimer by inserting the correct nucleotides opposite the two residues of the dimer. The ability of Polη to promote accurate bypass of a cis-syn cyclobutane dimer or a (6-4) photoproduct at the TC and CC sites shows that Polη is not an A rule polymerase, inserting an A opposite UV lesions by default. Further, our observation that UV-induced reversion at TC and CC sites in both the wild-type and rad30Δ strains occurs by a 3′ C→T mutation indicates that a DNA polymerase other than Polη is responsible for the incorporation of an A residue opposite the 3′ C of photoproducts. Although, and as expected, UV-induced mutagenesis at both the TC and CC sites is nearly abolished in the rev3Δ or the rad30Δ rev3Δ strains, REV3-encoded DNA polymerase ζ may not be the enzyme which inserts the A residue opposite the 3′ C in these photoproducts, because Polζ is highly inefficient at inserting nucleotides opposite the 3′ site of these lesions (15). Instead, Polζ functions in the subsequent step of elongating from the base incorporated opposite the lesion by another DNA polymerase (15).

Yeast and human Polη resemble each other in structure and function, and steady-state kinetic analyses have indicated that they both bypass a cis-syn TT dimer with the same efficiency and fidelity as through undamaged Ts (16, 27). Genetic studies presented here support a role for yeast Polη in the error-free bypass of TC and CC photoproducts, and they suggest a similar role for human Polη. Thus, by preventing UV mutagenesis at a wide spectrum of dipyrimidine sites, Polη plays an indispensable role in minimizing the incidence of sunlight-induced skin cancers in humans.