In previous experiments carried out in other labs as well as in the work presented here, the Rev1 Y-family DNA polymerase is required for Pol ζ-dependent lesion bypass, with its loss resulting in a reversionless phenotype identical to that associated with loss of Rev3 (reviewed in [3
]). Deciphering the precise function of Rev1 in lesion bypass has been complicated by the fact that, in addition to its ability to catalyze dCMP insertion opposite lesions in vitro
, it has a DNA-binding BRCT domain and an essential C-terminal scaffolding domain. Whereas the latter two domains are assumed to be universally important, the relevance of the dCMP transferase activity is likely lesion specific [18
]. Here, we focus specifically on the functions of Rev1 during the bypass of endogenously generated AP sites, which is thought to be a multi-polymerase process requiring different polymerases for the insertion and extension steps [13
]. Accumulated evidence indicates that Pol ζ is required for the extension step, but which polymerase is primarily responsible for the insertion step has remained controversial. The controversy revolves around whether dCMP is the predominant nucleotide inserted, which would support a catalytic role for Rev1 during AP-site bypass, or whether dAMP is the preferred nucleotide, which would indicate a primary role for a different DNA polymerase during the insertion step.
In the current study, highly sensitive pTET-lys2
frameshift- and nonsense-reversion assays were used as tools to generate AP sites. The uniqueness of these assays is that the source of the AP sites has been defined: uracil specifically replaces thymine in highly transcribed DNA, and it is the subsequent excision of uracil by Ung1 that creates the AP sites [25
]. Deletion of the C-terminal 72 amino acids of Rev1 resulted in mutation rates and spectra that were indistinguishable from those associated with the rev1Δ
allele, indicating no AP-site bypass. This is in agreement with previous results demonstrating a requirement for the Rev1 C-terminal domain in Pol ζ-mediated UV survival and mutagenesis [16
]. Similar to the yeast protein, the analogous C-terminal domain of human Rev1 also mediates interaction with human Rev7 [41
], and is required for DNA-damage tolerance in the avian DT40 cell line [42
]. In contrast to the null phenotype associated with the rev1-CDEL
allele, mutation of the BRCT domain reduced, but did not eliminate AP-site bypass in the pTET-lys2ΔA746
assay. While this is consistent with previous transformation-based assays using engineered AP sites [23
], the associated changes in frameshift-reversion spectrum suggest that the DNA-binding activity of the BRCT domain affects primer-template slippage during lesion bypass.
In BER (apn1 ntg1 ntg2)- and BER/NER (apn1 rad14)-defective backgrounds the pTET-lys2ΔA746 assay selects for net +1 frameshift mutations that arise during the bypass of uracil-derived AP sites that accumulate when uracil replaces thymine. In particular, complex frameshifts at a 6A hotspot result from dCMP, dGMP or dTMP insertion, while non-mutagenic dAMP insertion yields simple +1 frameshifts within the 6A run (see ). Because the base substitution is not the selected event, we believe that this specific assay allows an unbiased assessment of the frequency with which each dNMP is inserted opposite AP sites, as well as the relevance of Rev1 catalytic activity to the insertion specificity. While it could be argued that concerted misincorporation-slippage during AP-site bypass is rare and, therefore, does not provide an appropriate read-out of insertion specificity, it should be noted that the overall Lys+ rates in the frameshift-reversion assay were within 7-fold of those obtained in the nonsense-reversion assay. This suggests that, if there is a mononucleotide run appropriately positioned near an AP site, slippage subsequent to the insertion step is a frequent event.
In the presence of Rev1, both complex and simple frameshifts were common in the pTET-lys2ΔA746 assay, with almost all complex events containing the T > G or A > C transversion diagnostic of dCMP insertion opposite the AP site. Although introduction of the catalytically inactive rev1-AA allele had only minor effects on the total rate of Lys+ revertants, there were very striking and distinctive changes in mutation types at the 6A hotspot. The complex events that reflect dCMP insertion opposite AP sites were eliminated in the rev1-AA mutants and there was a compensatory increase in the simple frameshifts predicted to result when dAMP is inserted opposite the initiating AP site. These data demonstrate that insertion of dCMP absolutely depends on the catalytic activity of Rev1 and suggest that dCMP, rather than dAMP, is the most frequently inserted dNMP during AP-site bypass in this system.
Analogous experiments were done with BER- and BER/NER-defective strains containing the pTET-lys2-TAA
nonsense allele, a system that only allows detection of dCMP, dGMP or dTMP insertion opposite uracil-derived AP sites (dAMP insertion does change the stop codon). Consistent with Rev1-catalyzed dCMP insertion opposite AP sites, the reversion rate of the pTET-lys2-TAA
allele was reduced 5–10 fold upon introduction of the rev1-AA
allele. Reversion spectra confirmed that dCMP insertion is strongly preferred to that of dGMP or dTMP, and that dCMP insertion absolutely requires the catalytic activity of Rev1. Because compensatory dAMP insertion is not detected in the pTET
reversion assay, we also were able to observe what would normally be very minor insertion of dGMP or dTMP opposite AP sites. Our data suggest that the nucleotide selectivity during AP site bypass in yeast follows the pattern of dCMP > dAMP
dGMP ~ dTMP.
The role of Rev1 in inserting the nucleotides other than dCMP is presumably structural, and the identity of the DNA polymerase that steps in to accomplish insertion opposite AP sites in rev1-AA
strains is not known. While Pol η can accomplish the insertion step in vitro
], its loss in the rev1-AA
background had very little, if any, effect on reversion rates or spectra in either assay used here. Among the three yeast replicative DNA polymerases, both Pol δ and Pol α have been reported to insert nucleotides opposite AP sites in biochemical assays [13
]; Pol ε has not been examined. Given that Pol δ has a strong preference for dAMP insertion opposite AP sites in vitro
], and the requirement for its noncatalytic, Pol32 subunit for Pol ζ-dependent mutagenesis in vivo
], this replicative polymerase may be the most likely candidate.
Finally, it should be noted that both of our reporters were located close to the strong ARS306
origin of replication on Chromosome III. Because AP sites accumulate on complementary strands in BER- versus BER/NER-defective backgrounds [26
], our systems have the potential to detect replication-related differences in Rev1 activity during AP-site bypass. AP sites remain mainly on the nontranscribed strand in BER-defective strains, which corresponds to the lagging-strand template for a fork originating from the ARS306
]. Conversely, in BER/NER-defective strains, AP lesions persist on the transcribed strand, which is the leading-strand template during replication. Although there were generally no differences in the mutagenic effects associated with various rev1
alleles tested in these two strain backgrounds, there was a difference in the compensatory insertion of dAMP in the rev1-AA
mutants. In the BER-defective background, there was complete compensation so that there was no associated drop in the overall reversion rate; in the BER/NER-defective background, there was little compensation, resulting in a 3-fold drop in Lys+
rate. One interpretation of this difference is that the dAMP-inserting polymerase has better access to AP sites encountered during lagging-strand than during leading-strand synthesis. Given that Pol δ is responsible for most lagging-strand synthesis in yeast [50
], this would be consistent with it being the major polymerase responsible for dAMP insertion opposite AP sites.
By demonstrating the significance of Rev1 dCMP transferase activity during AP-site bypass, our data further support a model in which engagement of the catalytic activity of this protein depends on the specific lesion being bypassed [18
]. The dCMP insertion specificity of Rev1 would make biological sense if most spontaneous AP sites occur at guanines, as this would serve to limit AP-associated mutagenesis. Available data suggest, however, that direct incorporation of uracil is the major source of endogenous AP sites in yeast DNA [52
], which would result in the TA > GC mutations of the type documented here. It still remains possible, however, that the most biologically relevant lesion, especially outside a laboratory environment, is a damaged form of guanine, which may or may not lead to AP site formation. If this were the case, then the dCMP transferase activity of Rev1 would provide an error-free bypass mechanism.
In higher eukaryotes, the TLS of uracil-derived AP sites is a critical step in the somatic hypermutation of immunoglobulin genes [4
]. In mammalian cells, Rev1 appears to play an important role in this process, with its loss affecting the spectrum of mutations [55
]. Similar shifts in mutation spectra have been observed in mouse or chicken cells expressing catalytically inactive Rev1 [56
]. Finally, the dCMP transferase activity of Rev1 documented here might be especially relevant when the cellular ratio of dUTP to dTTP is perturbed. Inhibitors of thymidylate synthase and anti-folates, for example, are commonly used chemotherapeutics that elevate uracil in DNA [58
]. The presence and/or catalytic activity of Rev1 might have implications for cell survival and especially mutagenesis following treatment with such drugs.