Requirement of Rad18 protein for replication through DNA lesions in mouse and human cells

Abstract

In yeast, the Rad6-Rad18 ubiquitin conjugating enzyme plays a critical role in promoting replication although DNA lesions by translesion synthesis (TLS). In striking contrast, a number of studies have indicated that TLS can occur in the absence of Rad18 in human and other mammalian cells, and also in chicken cells. In this study, we determine the role of Rad18 in TLS that occurs during replication in human and mouse cells, and show that in the absence of Rad18, replication of duplex plasmids containing a cis-syn TT dimer or a (6-4) TT photoproduct is severely inhibited in human cells and that mutagenesis resulting from TLS opposite cyclobutane pyrimidine dimers and (6-4) photoproducts formed at the TT, TC, and CC dipyrimidine sites in the chromosomal cII gene in UV-irradiated mouse cells is abolished. From these and other observations with Rad18, we conclude that the Rad6-Rad18 enzyme plays an essential role in promoting replication through DNA lesions by TLS in mammalian cells. In contrast, the dispensability of Rad18 for TLS in chicken DT40 cells would suggest that the role of the Rad6-Rad18 enzyme complex has diverged considerably between chicken and mammals, raising the possibility that TLS mechanisms differ among them.

DNA lesions in the template strand block the progression of the replication fork. Genetic studies in the yeast Saccharomyces cerevisiae have indicated a critical role for the Rad6-Rad18 ubiquitin-conjugating complex (1, 2) in promoting replication through DNA lesions by translesion DNA synthesis (TLS) and by an alternative pathway that involves template switching (3). In yeast, during replication, TLS through UV-induced cis-syn cyclobutane pyrimidine dimers (CPDs) occurs by the action of DNA polymerases (Pols) η and ζ, in which Polη performs relatively error-free synthesis opposite CPDs, whereas Polζ carries out a more mutagenic mode of TLS (3, 4). Consequently, the incidence of UV-induced mutations is elevated in rad30Δ cells lacking Polη (5) and is greatly reduced in yeast cells lacking Polζ (3). A Rad5-Mms2-Ubc13–dependent template-switching pathway provides an error-free alternative to allow for the passage of the replication fork through DNA lesions (6). Because of the key role of Rad6-Rad18 in advancing replication through DNA lesions, yeast cells lacking Rad6 or Rad18 protein exhibit a very high degree of sensitivity to DNA damaging agents, and because of the requirement of this enzyme complex for both the error-free and error-prone modes of TLS, mutagenesis induced by DNA damaging treatments is inhibited in rad6Δ/rad18Δ yeast cells (3).

Treatment of yeast cells with DNA damaging agents elicits monoubiquitylation of proliferating cell nuclear antigen (PCNA) at lys164 by the Rad6-Rad18 enzyme; subsequently, this PCNA residue is polyubiquitylated via a lys63-linked polyubiquitin chain, for which the Mms2-Ubc13-Rad5 complex is additionally required (7). Genetic studies in yeast have shown that PCNA monoubiquitylation is a necessary prerequisite for TLS and that Rad6-Rad18–dependent template switching requires PCNA polyubiquitylation (7–9). In the Mms2-Ubc13-Rad5 complex, the Mms2-Ubc13 ubiquitin conjugating enzyme (10) functions in conjunction with Rad5 protein, which provides the ubiquitin ligase function (7). Rad5 also possesses a DNA helicase/DNA translocase activity that could function directly in lesion bypass by promoting replication fork regression and template switching (11). Genetic studies in yeast have provided evidence for the requirement of Mms2, Ubc13, and Rad5 for a template-switching mode of lesion bypass and they have shown that both the ubiquitin ligase and DNA translocase activities are essential for Rad5 to carry out its role in lesion bypass (6, 12).

Mammalian cells have two RAD6 homologs, RAD6A and RAD6B, and a single RAD18 gene. Importantly, in contrast to the indispensability of the Rad6-Rad18 enzyme and of PCNA ubiquitylation for TLS in yeast cells, several studies have indicated that Rad6-Rad18 and PCNA ubiquitylation may not play as significant a role in TLS in vertebrates as in yeast. For example, in two studies carried out with human cell-free extracts, TLS was shown to occur in the absence of PCNA ubiquitylation. In one study, replication through a site-specific cis-syn TT dimer present on the leading-strand template of a double-stranded circular plasmid was examined in human cell-free extracts using conditions in which origin-dependent initiation and bidirectional replication occur. In this system, replication through the TT dimer carried on the plasmid is absolutely dependent upon Polη; however, when replication opposite the dimer was examined with cell-free extracts using wild-type PCNA or K164R mutant PCNA defective in ubiquitylation, TLS by Polη could occur with K164R mutant PCNA (13). In another study, TLS opposite a cis-syn TT dimer carried on a single-stranded plasmid was analyzed using human cell-free extracts that lacked Rad18 or in which the K164R mutant PCNA was used instead of wild-type PCNA, and TLS was shown to occur in the absence of Rad18 as well as in the absence of PCNA ubiquitylation (14).

In another study, TLS has been examined in mouse cells lacking Rad18 (Rad18^−/−) using a gapped plasmid that carried a site specific cis-syn TT dimer, a (6-4) TT photoproduct, or a cis-Pt GG intrastrand crosslink (15). Opposite all three DNA lesions, TLS occurred with a frequency of ∼30–40% in Rad18^−/− cells compared with that in wild-type cells; and opposite each of these DNA lesions, the relative frequency of error-free and mutagenic TLS remained the same in Rad18^−/− cells as in wild-type cells. Thus, although Rad18 affected the efficiency of TLS, mutagenicity was not affected. Taken together, the above-noted studies have supported the inference that in mammalian cells, TLS Pols can function in lesion bypass in the absence of Rad6-Rad18 enzyme function.

Studies with chicken B lymphocyte DT40 cells have also indicated that TLS can occur in the absence of Rad18. For example, although a deficiency of either Polκ or Rad18 confers an increase in UV sensitivity in DT40 cells, they both impart about the same level of UV sensitivity; and an additive increase in UV sensitivity occurs in the absence of both Polκ and Rad18 (16). An important implication of these observations is that Rad18 is not essential for TLS and that Polκ can function in TLS independently of Rad18. If Rad18 were indispensable for TLS by the various Pols involved in the bypass of UV lesions, then Rad18 deficiency would have conferred a much higher UV sensitivity than that resulting from a deficiency of Polκ or of any other individual Pol involved in TLS opposite UV lesions, and there would have been no further increase in UV sensitivity of Rad18 deficient cells in combination with mutations in any of the TLS Pols.

In another study, the role of Rad18 and of PCNA ubiquitylation was analyzed in UV-irradiated or 4-nitroquinoline-1-oxide (NQO)-treated chicken DT40 cells by DNA fiber labeling (17). Rather surprisingly, the rate of fork progression through damaged DNA was not affected in rad18 mutant cells or in cells carrying the K164R mutation in PCNA. Furthermore, disruption of Rev1 TLS Pol resulted in a reduction in the rate of fork progression through DNA lesions, and sensitivity to UV or NQO was greatly enhanced in DT40 cells carrying the rev1 mutation in combination with the K164R PCNA mutation. These observations have added strong support to the notion that the Rad6-Rad18 enzyme and PCNA ubiquitylation are not as indispensable for TLS in chicken cells as in yeast, and that in chicken cells TLS Pols can carry out lesion bypass in the absence of Rad6-Rad18.

Because in yeast cells, the Rad6-Rad18 complex and PCNA ubiquitylation play a key role in lesion bypass by TLS, and because replication through DNA lesions is greatly inhibited in the absence of Rad6 or Rad18 (3, 18), it becomes imperative to know whether during replication, TLS can occur in human cells in the absence of Rad6-Rad18 or whether the Rad6-Rad18 enzyme is as indispensable for TLS in humans as it is in yeast. To establish whether or not the Rad6-Rad18 enzyme plays an essential role in promoting replication through DNA lesions by TLS Pols in mammalian cells, in this study we examine in human cells the effects of Rad18 depletion on the replication of a duplex plasmid harboring a cis-syn TT dimer or a (6-4) TT photoproduct, and we analyze the effects of Rad18 depletion on mutagenesis resulting from TLS opposite CPDs and (6-4) photoproducts formed in UV-irradiated mouse cells. From these studies in human and mouse cells, we conclude that Rad18 is indispensable for TLS in mammalian cells. We discuss the implications of our observations for the role of Rad6-Rad18 enzyme complex and of PCNA ubiquitylation in lesion bypass during replication in human cells.

Results

Epistasis of Rad18 with TLS Pols.

A major role of Rad18 in TLS predicts that the absence of Rad18 would confer a large increase in UV sensitivity and that the UV sensitivity of rad18 mutant cells would not increase upon inactivation of any of the TLS Pols. We have shown previously that in human and mouse cells, Pols η, ζ, and κ provide alternate pathways for replicating through CPDs, the major UV photoproduct, in which Polη carries out highly error-free TLS, and Pols ζ and κ contribute to mutagenic TLS (19). For replicating through the less-frequently formed UV photoproduct, (6-4) PPs, Polζ functions in a predominantly error-free pathway, whereas Polη and Polι provide alternate pathways of mutagenic TLS (20). To determine if an epistatic relationship exists between Rad18 and TLS Pols, we examined the UV sensitivity of human XPA cells, defective in nucleotide excision repair (NER), that have been treated with siRNAs to deplete either Rad18 or any of the TLS Pols known to have a role in TLS opposite UV lesions, or where Rad18 was depleted in combination with depletion of one of the TLS Pols. The high efficiency of Rad18 depletion was verified by RT-PCR and by Western blotting (Fig. S1), and the evidence for high efficiency siRNA depletion of TLS Pols has been provided before (19). As shown in Fig. S2A, compared with control (NC) siRNA-treated XPA cells, Rad18 depletion led to a 65% reduction in UV survival, whereas UV survival of Polη as well as of Polζ (Rev3)-depleted cells was reduced by ∼30%. In accord with a less significant role of Polκ in TLS, UV survival was impacted the least upon Polκ depletion. Importantly, we find that simultaneous depletion of Rad18 with any of the TLS Pols, η, ζ, or κ, conferred no further reduction in UV survival than that seen upon depletion of Rad18 alone. Hence, an epistatic relationship exists between Rad18 and any of the TLS Pols that are required for replicating through UV lesions.

We also verified the epistasis of Rad18 with TLS Pols in mouse cells. As shown in Fig. S2B, compared with control (NC) siRNA-treated cells, Rad18 depletion conferred an ∼60% reduction in UV survival, depletion of either Polη or Polζ resulted in ∼30% reduction in UV survival, whereas simultaneous depletion of Rad18 with any of the TLS Pols conferred no further reduction in UV survival over that observed in Rad18-depleted cells. The epistatic interaction of Rad18 with TLS Pols would suggest that both in human and mouse cells, Rad18 functions in conjunction with TLS Pols in lesion bypass.

Requirement of Rad18 for Replication of Plasmids Containing a cis-syn TT Dimer or a (6-4) TT Photoproduct.

To determine the role of Rad18 in TLS, we examined the frequency of TLS in Rad18-depleted human XPA cells opposite a cis-syn TT dimer or a (6-4) TT photoproduct carried in the leading strand template of a duplex plasmid where bidirectional replication initiates from an SV40 origin (Fig. S3). In this plasmid system, TLS through the DNA lesion results in a blue colony and because the lesion-containing DNA strand carries the wild-type Kan⁺ gene (Fig. S3B), the frequency of TLS is determined from the number of blue colonies among the total Kan⁺ colonies. Hence, the siRNA knockdown of Pols required for TLS opposite the DNA lesion results in a reduction in the frequency of blue colonies. With the undamaged plasmid, however, we observe no significant change in the frequency of blue colonies among Kan⁺ colonies, regardless of which TLS Pols have been depleted; thus, the depletion of TLS Pols affects only the frequency with which the lesion-containing DNA strand replicates, but has no affect on the replication of undamaged plasmid.

In a number of different experiments in Rad18-depleted cells, we observed a dramatic reduction in the total number of Kan⁺ colonies when the plasmid carried a cis-syn TT dimer or a (6-4) TT photoproduct, whereas the numbers of Kan⁺ colonies that could be recovered from the replication of undamaged plasmid were not affected in Rad18-depleted cells. These observations suggested that the absence of Rad18 has a profound effect on the replication of the damaged plasmid but not on the replication of the undamaged plasmid. To verify these observations, we cotransfected the lesion-containing plasmid together with an undamaged pCDNA 3.1.zeocin plasmid and with Rad18 siRNA into XPA cells that had been pretreated with Rad18 siRNA (Fig. S3C). Hence, with these two plasmids, the replication efficiency of lesion-containing plasmid relative to the replication efficiency of undamaged plasmid can be determined from the relative frequency of Kan⁺ colonies among zeocin-positive colonies.

The replication of undamaged plasmid was not affected by the siRNA knockdown of Rad18 or of any of the TLS Pols, as the numbers of undamaged plasmid recovered were about the same regardless of whether cells were treated with control siRNA or with siRNAs for Rad18 or TLS Pols (Table 1). In XPA cells treated with control (NC) siRNA, opposite both the cis-syn TT dimer and (6-4) TT photoproduct, TLS occurs with a frequency of ∼35% (Table 1), and sequence analyses of blue colonies has shown that mutations occur only opposite the 3′ T or the 5′ T of either photoproduct (19, 20). In addition, from sequence analyses of over 300 white Kan^R colonies from control NC siRNA-treated XPA cells, we have determined that they all contain the SpeI sequence and that no other flanking mutations are present. Thus, any mutations that occur during the replication of damaged plasmid are restricted only to the lesion site, indicating that the blue/white colony assay provides a good measure of TLS frequencies.

Table 1.

Effect of siRNA depletion of Rad18 and TLS Pols on the replication efficiency of plasmid carrying a cis-syn TT or a (6-4) TT photoproduct on the leading strand DNA template in NER-defective XPA human fibroblast cells

		Lesion containing plasmid			Undamaged plasmid	Replication efficiency of lesion-containing plasmid (Kan+/zeocin+)*
DNA lesion	siRNA	No. of Kan+ colonies	No. of blue colonies among Kan+	TLS (%)	No. of zeocin+ colonies
	NC	658	228	34.7	642	1.02
cis-syn	Polη	458	80	17.5	628	0.73
TT dimer	Polη + Rev3	372	44	11.8	656	0.57
	Rad18	65	—†	—†	608	0.11
	NC	612	203	33.2	608	1.00
(6-4)TT	Rev3	451	70	15.5	625	0.72
photoproduct	Polη + Rev3	302	38	12.6	643	0.47
	Rad18	76	—†	—†	614	0.12

*The experiments for determining the replication efficiency of lesion-containing plasmid were repeated five times and similar results were observed in different experiments. The data shown are from one representative experiment.

^†Because of the large reduction in the recovery of lesion-containing plasmid in Rad18-depleted cells, and because only a few blue colonies were observed in each experiment, we could not reliably estimate the TLS frequency for each individual experiment. However, when all of the experiments are considered together, we estimate that TLS frequency was reduced to ∼5%, which likely reflects the residual level of Rad18 that remains in siRNA-depleted cells.

As shown in Table 1, with the TT dimer-containing plasmid, the frequency of Kan⁺ colonies relative to zeocin-positive colonies was reduced by ∼25% upon knockdown of Polη alone and by ∼40% upon knockdown of both Polη and Polζ. Rad18 depletion, however, led to a much more drastic reduction in the replication of lesion-containing plasmid, as indicated from the recovery of only ∼10% Kan⁺ colonies relative to zeocin-positive colonies. These effects of Rad18 depletion on the replication of undamaged and damaged plasmids have been verified in a number of different experiments (Table 1). Similar results were obtained with (6-4) TT photoproduct-containing plasmid. Relative to zeocin-positive colonies, the frequency of Kan⁺ colonies was reduced by ∼30% in Rev3-depleted cells, and by ∼50% in cells depleted for both Polη and Rev3, whereas in Rad18 depleted cells, the recovery of Kan⁺ colonies was reduced by ∼90%. Hence, opposite both the UV lesions, Rad18 depletion has a very severe effect on the replication of lesion containing plasmid.

As we have shown previously, and determined again here for the effects of Polη and Polζ (Rev3) depletion on TLS opposite both the DNA lesions, the TLS frequency was reduced to ∼12% in cells depleted for both the Pols (Table 1). TLS was reduced even further in Rad18-depleted cells, as opposite both the lesions, TLS occurred at a frequency of only ∼5% (Table 1), which presumably results from the residual Rad18 that remains in cells treated with Rad18 siRNA. We infer from these observations that all of the TLS pathways are inhibited in the absence of Rad18.

Requirement of Rad18 for UV Mutagenesis Resulting from TLS Opposite CPDs in the Chromosomal cII Gene in Mouse Cells.

To provide further evidence that Rad18 plays an indispensable role in TLS, we examined the effects of Rad18 depletion on UV mutagenesis in mouse cells. For this purpose, we separately determined the effects of Rad18 depletion on UV mutagenesis resulting from TLS through CPDs and (6-4) photoproducts in the cII transgene that has been integrated into the genome in a Big Blue mouse embryonic fibroblast (BBMEF) cell line (21). In a number of different experiments, this system has been shown to exhibit similar mutational patterns as those observed with endogenous chromosomal genes (21–23).

To analyze mutagenesis resulting specifically from TLS opposite CPDs, the (6-4) photoproducts were removed from the genome by expressing a (6-4) photolyase gene in BBMEF cells using the experimental protocol that allows for the complete removal of (6-4) photoproducts from UV-irradiated cells (21). As shown in Table 2, the frequency of mutations in unirradiated cells treated with control (NC) siRNA was ∼13 × 10⁻⁵, and spontaneous mutation frequency was not affected in Rad18-depleted cells. In UV-irradiated mouse cells expressing the (6-4) photoproduct photolyase gene and exposed to photoreactivating light, the mutation frequency increased to ∼48 × 10⁻⁵. Because the (6-4) photoproducts have been removed with the experimental protocol being used, this mutation frequency represents the contributions of spontaneous mutations (∼13 × 10⁻⁵) plus those resulting from mutagenic TLS opposite CPDs (∼35 × 10⁻⁵).

Table 2.

Effects of Rad18 depletion on UV induced mutation frequencies in the cII gene in mouse cells (BBMEF) expressing a (6-4) PP photolyase or a CPD photolyase and exposed to photoreactivating light

Photolyase	siRNA	UV*	PR†	Mutation frequency‡ (×10−5)
	NC§	—	+	13.2 ± 2.4
	Rad18	—	+	12.6 ± 3.1
	NC	+	+	48.2 ± 4.6
	Rad18	+	+	17.2 ± 3.6
(6-4)PP	Polη	+	+	104.6 ± 4.2
Photolyase	Polκ	+	+	32.6 ± 2.8
	Rev3	+	+	30.8 ± 3.6
	Polκ + Rev3	+	+	15.8 ± 2.7
	Polη + Rad18	+	+	19.2 ± 3.8
	Polκ + Rad18	+	+	15.6 ± 2.8
	Rev3 + Rad18	+	+	16.1 ± 2.3
	NC	—	+	12.6 ± 2.2
	Rad18	—	+	11.4 ± 2.6
CPD	NC	+	+	26.3 ± 3.2
Photolyase	Rad18	+	+	13.6 ± 3.8
	Rev3	+	+	39.2 ± 4.3
	Rev3 + Rad18	+	+	15.3 ± 3.1

*5 J/m² of UVC (254 nm) light.

^†Photoreactivation with UVA (360 nm) light for 3 h.

^‡Mutation frequencies were determined from averages of six to seven independent experiments.

^§NC, negative control siRNA.

The siRNA knockdown of Rad18 conferred a large reduction in mutation frequency from ∼48 × 10⁻⁵ in UV-irradiated control (NC) siRNA treated cells to ∼17 × 10⁻⁵ in Rad18-depleted cells (Table 2). Similar to the results we have reported previously (19), the siRNA knockdown of Polη results in a highly elevated frequency of ∼105 × 10⁻⁵, whereas the knockdown of either Polκ or Polζ confers a reduction in mutation frequencies to ∼30 × 10⁻⁵, and the simultaneous knockdown of Polκ and Polζ reduces mutation frequencies to a level (∼16 × 10⁻⁵) similar to that seen in unirradiated cells (Table 2). The indispensability of Rad18 for UV mutagenesis resulting from TLS opposite CPDs would predict that the elevated mutagenesis in Polη-depleted cells will not occur in Rad18-depleted cells and that UV mutagenesis in Rad18-depleted cells will not be affected when either Polκ or Polζ have also been depleted. As shown in Table 2, UV induced mutation frequencies in Rad18-depleted cells remained the same, regardless of whether Polη, Polκ, or Polζ was depleted along with Rad18, and in all cases, the frequency of UV-induced mutations resulting from TLS opposite CPDs declined to levels similar to those in unirradiated cells.

UV-induced mutations occur predominantly by C-to-T transitions in human cells (23–25) and a similar mutational spectrum is observed in the mouse cII gene (21). To determine the effects of Rad18 depletion on the mutational pattern resulting from TLS opposite CPDs, we analyzed the types of mutations that were formed in the cII gene in UV-irradiated mouse cells expressing the (6-4) photolyase and exposed to photoreactivating light. As we have shown previously (19), in control BBMEF cells treated with NC siRNA, a large majority of UV-induced mutations resulting from TLS opposite CPDs are clustered at particular dipyrimidine sites labeled with numbers 1–11 (Fig. 1A). Importantly, we find that none of the hot spots characteristic of TLS opposite CPDs remain in UV-irradiated Rad18-depleted cells (Fig. 1A), and the pattern of mutations that occur becomes similar to that in unirradiated cells treated with NC siRNA or with Rad18 siRNA (Fig. 1B). Thus, we conclude that Rad18 is essential for mutagenesis resulting from TLS opposite CPDs.

Fig. 1.

Mutational spectra opposite CPDs in UV irradiated mouse (BBMEF) cells treated with Rad18 siRNA. Mutational spectrum was determined for the cII gene in cells exposed to 5 J/m² of UV light; only the sequence for nucleotide positions 25–288 is shown because the UV-induced mutations are clustered in this region. (A) Mutations resulting from TLS opposite CPDs in UV irradiated cells treated with control (NC) siRNA (above the sequence) or treated with Rad18 siRNA (below the sequence). The (6-4) photoproducts were removed from the genome by expressing a (6-4) photolyase gene and exposing cells to photoreactivating light. The positions of mutational hot spots opposite CPDs (numbers 1–11) are indicated. (B) Mutations in non-UV–irradiated cell line expressing the (6-4) photolyase, and treated with control (NC) siRNA (above the sequence) or treated with Rad18 siRNA (below the sequence).

Requirement of Rad18 for UV Mutagenesis Resulting from TLS Opposite (6-4) Photoproducts in the Chromosomal cII Gene in Mouse Cells.

To examine the effects of Rad18 depletion on mutagenesis resulting from replicative bypass of (6-4) photoproducts formed at various dipyrimidine sites in the mouse cII gene, CPDs from UV-irradiated cells were selectively removed from the genome by expressing a CPD photolyase gene and exposing cells to photoreactivating light (21). As shown in Table 2, in control siRNA-treated UV-irradiated mouse cells expressing the CPD photolyase gene and exposed to photoreactivating light, mutation frequency increased to ∼26 × 10⁻⁵ compared with the spontaneous mutation frequency of ∼13 × 10⁻⁵ in unirradiated cells. In Rad18-depleted cells, the frequency of UV-induced mutations arising from replicative bypass of (6-4) photoproducts declined to ∼14 × 10⁻⁵, a level similar to that observed in unirradiated cells.

As we have shown previously, during replication in mouse and human cells, TLS opposite (6-4) photoproducts occurs via a predominantly error-free pathway dependent upon Polζ, and by a more mutagenic pathway that requires Polη or Polι (20). Hence, in cells depleted of both Polη and Polι, mutagenic TLS opposite (6-4) photoproducts is abrogated because, in the absence of both these Pols, only the Polζ-controlled error-free pathway operates. On the other hand, in the absence of Polζ, the frequency of mutagenic TLS opposite (6-4) photoproducts is elevated because then TLS occurs via the Polη- and Polι-dependent error-prone pathways. As would be expected from the indispensability of Rad18 for mutagenic TLS opposite (6-4) photoproducts, simultaneous depletion of Rev3 with Rad18 led to a drastic reduction in the elevated mutation frequencies that occur upon the depletion of Rev3 alone (∼39 × 10⁻⁵), and mutation frequencies declined to a level (∼15 × 10⁻⁵) similar to that observed for unirradiated cells treated with control (NC) siRNA (∼13 × 10⁻⁵) (Table 2).

As has been shown previously (20), mutations at (6-4) photoproducts in the cII gene occur primarily at five dipyrimidine sites (numbers 1–5), and in the sequence data shown in Fig. 2A, clustering of mutations is seen at four of these five hot spot sites, numbers 1, 2, 4, and 5, but not at site number 3, which is the least frequently mutated site among the five sites. Importantly, in Rad18 depleted cells, none of the UV induced mutational hot spots remain (Fig. 2A), and the random pattern of mutations becomes similar to that in unirradiated cells (Fig. 2B). From these observations, we conclude an essential role of Rad18 in mutagenesis resulting from TLS opposite (6-4) photoproducts.

Fig. 2.

Mutational spectra opposite (6-4) photoproducts in UV irradiated mouse (BBMEF) cells treated with Rad18 siRNA. (A) Mutations in the cII gene resulting from TLS opposite (6-4) photoproducts in UV-irradiated (5 J/m²) cells treated with control (NC) siRNA (above the sequence) or treated with Rad18 siRNA (below the sequence). CPDs were removed from the genome by expressing a CPD photolyase gene and by exposing cells to photoreactivating light. The positions of mutational hot spots opposite (6-4) photoproducts (numbers1–5) are indicated. (B) Mutations in the non-UV–irradiated cell line expressing the CPD photolyase, and treated with control (NC) siRNA (above the sequence) or treated with Rad18 siRNA (below the sequence).

Discussion

Our observations indicate an essential role for Rad18 in TLS during replication in mouse and human cells. Briefly, we find that: (i) The UV sensitivity of Rad18-depleted mouse and human cells is not enhanced upon the simultaneous depletion of any of the TLS Pols involved in UV lesion bypass. The epistasis of Rad18 with TLS Pols is in accord with a role for Rad18 in promoting lesion bypass by TLS Pols. (ii) In Rad18-depleted human cells, replication of undamaged plasmid is not affected but the replication efficiency of a plasmid carrying a cis-syn TT dimer or a (6-4) TT photoproduct is greatly reduced. (iii) Opposite both the UV lesions, the TLS frequency is greatly reduced in cells depleted for Rad18, indicating that all of the TLS pathways are inhibited. (iv) As determined from the analyses of mutation frequencies and mutational spectra resulting from TLS opposite CPDs and opposite (6-4) photoproducts in mouse cells, Rad18 is essential for mutagenic TLS opposite both these UV induced DNA lesions. Because mutagenic TLS opposite CPDs depends upon Pols κ and ζ (19), and mutagenic TLS opposite (6-4) photoproducts requires Pols η and ι (20), Rad18 controls TLS by all these Pols.

Our observations have provided clear evidence that Rad18 plays a crucial role in TLS in mammalian cells. Because Rad18 associates with Rad6 and because the human Rad6-Rad18 complex monoubiquitylates PCNA at K164 (26), we presume that PCNA ubiquitylation is a prerequisite for TLS in mammalian cells as well. The observations with human cell-free extracts that replication through a cis-syn TT dimer could occur in the absence of Rad18 as well as with the K164R mutant PCNA (13, 14) might reflect the possibility that the complexity of cellular mechanisms that operate when the replication fork stalls at the lesion site and TLS occurs is not recapitulated in these in vitro studies.

Our observations that Rad18 is required for UV-induced mutations resulting from replication through CPDs and (6-4) photoproducts induced in the chromosomal cII gene in mouse cells have indicated the indispensability of Rad18 for mutagenic TLS by Pols κ and ζ opposite CPDs, and for mutagenic TLS opposite (6-4) photoproducts by Pols η and ι. Thus, Rad18 is required for promoting the function of each of the Pols involved in TLS opposite these DNA lesions. Furthermore, our observation that replication through a cis-syn TT dimer or a (6-4) TT photoproduct carried on a duplex plasmid is severely curtailed in Rad18-depleted human cells, and that the frequency of TLS opposite both the UV lesions is greatly reduced, have provided additional and direct evidence for the indispensability of Rad18 for lesion bypass during replication. These observations differ from the results that have been reported for TLS in a gapped plasmid carried in mouse Rad18^−/− cells, because in that case, only the efficiency of TLS was affected but the relative proportion of error-free vs. mutagenic TLS remained the same as in wild-type cells (15), indicating that TLS via the various pathways could still occur during gap repair in the absence of Rad18. The indispensability of Rad18 for TLS during replication but not for TLS in a gapped plasmid adds further evidence that the genetic control of TLS that occurs during replication differs in many ways from that in gap repair. In this regard, in a previous study, we provided evidence that TLS opposite a (6-4) TT photoproduct carried on the leading or the lagging strand DNA template of a duplex plasmid in human cells incurs in less than 2% of mutational events, and that opposite this UV lesion, Polζ functions in TLS in a nonmutagenic manner in both human and mouse cells (20). In contrast, TLS opposite a (6-4) TT photoproduct carried in a gapped plasmid in mouse cells occurred in a very highly mutagenic manner, such that ∼75% of TLS events were mutagenic, and Polζ was responsible for mediating this highly error-prone mode of TLS (27). Hence, we infer from these various observations that the results obtained from TLS analyses using a gapped plasmid are not informative for understanding how TLS operates during replication.

In conclusion, from analyses of replication of plasmids carrying a cis-syn TT dimer or a (6-4) TT photoproduct in human cells, and from determination of mutation frequencies and mutational spectra formed opposite CPDs or (6-4) photoproducts in the chromosomal cII gene in mouse cells, we show that Rad18 is indispensable for TLS and damage-induced mutagenesis. This conclusion implies that the observations that have been reported for the much less important role of Rad18 in lesion bypass in chicken DT40 cells (17) do not pertain to human or mouse cells.

Materials and Methods

TLS was examined in human cells using SV40 origin based duplex plasmids, which carried a cis-syn TT dimer or a (6-4) TT photoproduct on the leading strand DNA template. Frequencies of UV induced mutations that resulted specifically from replication through CPDs or (6-4) photoproducts formed at TT, TC, and CC dipyrimidine sites were examined in the cII gene carried in big blue mouse embryonic fibroblast (BBMEF) cells. The details for construction of lesion containing plasmid and for lesion bypass assays in human cells and for mutational studies in mouse cells have been described previously (19, 20), and the details that deal more specifically with Rad18 are provided in the SI Text.