Mutator alleles of yeast DNA polymerase ζ

Abstract

The yeast REV3 gene encodes the catalytic subunit of DNA polymerase zeta (pol ζ), a B family polymerase that performs mutagenic DNA synthesis in cells. To probe pol ζ mutagenic functions, we generated six mutator alleles of REV3 with amino acid replacements for Leu979, a highly conserved residue inferred to be at the pol ζ active site. Replacing Leu979 with Gly, Val, Asn, Lys, Met or Phe resulted in yeast strains with elevated UV-induced mutant frequencies. While four of these strains had reduced survival following UV irradiation, the rev3-L979F and rev3-L979M strains had normal survival, suggesting retention of pol ζ catalytic activity. UV mutagenesis in the rev3-L979F background was increased when photoproduct bypass by pol η was eliminated by deletion of RAD30. The rev3-L979F mutation had little to no effect on mutagenesis in an ogg1Δ background, which cannot repair 8-oxo-guanine in DNA. UV-induced can1 mutants from rev3-L979F and rad30Δrev3-L979F strains primarily contained base substitutions and complex mutations, suggesting error-prone bypass of UV photoproducts by L979F pol ζ. Spontaneous mutation rates in rev3-L979F and rev3-L979M strains are elevated by about 2-fold overall and by 2- to 8-fold for C to G transversions and complex mutations, both of which are known to be generated by wild-type pol ζ in vitro. These results indicate that Rev3p-Leu979 replacements reduce the fidelity of DNA synthesis by yeast pol ζ in vivo. In conjunction with earlier studies, the data establish that the conserved amino acid at the active site location occupied by Leu979 is critical for the fidelity of all four yeast B family polymerases. Reduced fidelity with retention of robust polymerase activity suggests that the homologous rev3-L979F allele may be useful for analyzing pol ζ functions in mammals, where REV3 deletion is lethal.

1. INTRODUCTION

DNA polymerase ζ(pol ζ) is one of many eukaryotic DNA polymerases [1-3], and it has a particularly important role in mutagenesis [4]. Pol ζ consists of an accessory subunit encoded by the REV7 gene and a polymerase catalytic subunit encoded by the REV3 gene [5]. Cells disrupted for REV3 have reduced levels of DNA damage-induced mutagenesis [4], thereby implicating pol ζ in mutagenic translesion DNA synthesis. REV3 has many functions [4,6,7]. It is needed for the majority of spontaneous mutagenesis [4,8-13], and it contributes to mutagenesis associated with impairment of the major replicative polymerases [11], with defective DNA repair [14-19], with double-strand break repair [20-22], with transcription [23], and with somatic hypermutation of immunoglobulin genes [24,25]. Consistent with these multiple mutagenic functions, pol ζ lacks an intrinsic proofreading exonuclease [5] and it is promiscuous in extending mismatched primer termini [26-29]. Consistent with its role in mutagenic synthesis in vivo, purified yeast pol ζ synthesizes DNA in vitro with lower fidelity than the major replicative polymerases, pol α, pol δ and pol ε[30].

As a member of the B family of DNA polymerases, pol ζ shares homology with pol α, pol δ and pol ε, as well as with enzymes that replicate bacteriophage genomes, such as T4 pol and RB69 pol. Residues in motifs A, B and C that form the polymerase active site are particularly conserved. In the crystal structure of RB69 pol [31], an invariant tyrosine in motif A interacts with the ribose of the incoming dNTP. Adjacent to this tyrosine is a hydrophobic residue, methionine in pol ε or leucine in other Family B members, e.g., Leu979 in pol ζ. In the major replicative polymerases, replacing this residue with certain other amino acids reduces DNA synthesis fidelity in vitro [32-36], due to increased dNTP misinsertion efficiency, increased mismatch extension efficiency and/or decreased proofreading efficiency [see also 37,38]. Consistent with these biochemical properties, strains encoding these mutant polymerases have elevated spontaneous mutation rates that likely reflect reduced DNA replication fidelity in vivo [32-37,39,40]. Based on these studies of other Family B polymerases, we hypothesized that Leu979 in Rev3p may have an important role in determining the fidelity of DNA synthesis by pol ζ. To test this, and ultimately to obtain a more thorough understanding of multiple pol ζ functions in mutagenic DNA transactions in vivo, here we have undertaken a search for amino acid replacements in Rev3p that are likely to retain pol ζ activity yet result in elevated spontaneous and damage-induced mutagenesis in yeast. This approach to study rev3 mutator alleles is intended to add to studies in which pol ζ function has been inferred from rev3 disruption, a strategy has been highly informative in yeast [4] but less informative in mice because rev3 disruption results in embryonic lethality [41-43].

2. MATERIALS AND METHODS

2.1. Media and growth conditions

Yeast strains were grown nonselectively on YPDA media (1% yeast extract, 2% Bacto-peptone, 2% dextrose, 10 mg/L adenine). Synthetic, selective media contained 2% dextrose and appropriate amino acids. Canavanine resistant mutants were identified by growth on synthetic media supplemented with 60 ug/mL canavanine. All growth was at 30°.

2.2. Construction of yeast strains

All strains used in this study were derived from haploid 8C-YUNI101 (MATa his7-2 leu2-3,112 ura3Δbik1::ura3-29RL trp1-1_UAG ade 2-1_UAA) [44]. The rev3::LEU2 mutant was described previously [44]. All other mutants were constructed by standard one-step (OST) or two-step transplacement (TST) and all strains were verified by PCR and/or sequencing. RAD30 and OGG1 deletions were generated via OST with PCR fragments amplified from pFA6-kanMX4 [45] using primers disRAD30-L and disRAD30-R, or disOGG1-L and disOGG1-R, respectively (for primer sequences see Supplementary Table 1). rev3-L979X alleles were created by site-directed mutagenesis of pREV3Cav2 [44] using the QuickChange Kit (Stratagene). rev3-L979X yeast strains were made by TST with SnaBI-cut mutagenized derivatives of pREV3Cav2. Each of the rev3-L979X yeast strains were verified by sequencing using primers designated REV3seq in Supplementary Table 1.

2.3. Measurements of mutation rates

Spontaneous mutation rates were determined by the method of median [46], using at least 22 independent cultures per experiment. Ultraviolet light (UV)-induced mutation frequencies were measured using 2-4 independent cultures. Cultures were grown to saturation in YPDA liquid, then diluted appropriately and plated onto synthetic media supplemented with canavanine and non-selective synthetic media. Plates were immediately exposed to UVC at the indicated dose and were then incubated at 30°C in the dark. Mutation frequencies were calculated as the average of the number of colonies on canavanine plates divided by the number of colonies on non-selective plates multiplied by the dilution factor.

2.4. Sequence analysis of can1 mutants

UV-induced can1 mutations were isolated following exposure to 20 J/m² under the conditions described above. At least 24 canavanine resistant isolates were sequenced per strain. A 2.0 kb region covering the 3′ end of CAN1 gene was amplified with primers can1-F and can1-R (Supplementary Table 1). The 5′ end of CAN1 gene was amplified using primers can1-F2 and can1-R (Supplementary Table 1). The PCR fragments were subsequently sequenced on both strands using sequencing primers can1-SF1 to can1-SR3 (Supplementary Table 1).

2.5. Statistical analysis

We first estimated the number of can1 mutants expected to be due to an individual type of mutation in each strain by multiplying the proportion of that mutation observed in the mutation spectrum (Table 1) by the total number of can1 survivors obtained after irradiation with 20 J/m² UV. We then estimated mutation frequencies of individual mutation types as the estimated number of can1 mutants of each type divided by the total number of can1 survivors. The mutation frequencies of an individual type in each strain were then compared with the wild-type strain using an independent samples t-test for equality of proportions. The variance of this test was adjusted to reflect the number of mutants sequenced. A complete explanation of the two independent samples test for equality of mutation frequencies can be found in the Supplementary Information.

Table 1.

UV-induced mutations in wild-type, rad30 and rev3 yeast strains.

							rad30Δ
	wild-type		rev3-L979F		rad30Δ		rev3-L979F
Freq. (×10−6)	88 [59 - 120]		200 [50 - 350]		200 [170 - 230]		540 [470 - 590]
	Events	Freq.	Events	Freq.	Events	Freq.	Events	Freq.
A-T to G-C	4	14	0	0	2	16	0	0
G-C to A-T	8	28	6	48	12	92c	10	220c
A-T to T-A	6	21	2	16	4	32	2	43
A-T to C-G	0	0	0	0	1	8	0	0
G-C to T-A	1	4	2	16	2	16	4	86c
Tandem bps	1	4	3	24	0	0	4	86c
Indelsa	3	11	4	32	0	0	2	43
Complexb	2	7	8	64c	5	48	3	65

Cells were exposed to 20 J/m² of UVC.

Freq. represents mutation frequency. All mutation frequencies are (×10⁻⁶). Numbers in brackets represent one standard deviation.

^aOne and two-base insertions and deletions were categorized as indels.

^bMultiple, closely spaced base substitutions and/or indels were categorized as complex mutations. One of the can1 isolates from the rad30Δ strain contained three single base changes within 30 base-pairs.

^cSignificant difference (p < 0.05) from the REV3 strain. P-values were determined as described in the materials and methods.

3. RESULTS

3.1. Choice of rev3 missense alleles to investigate

Mutator phenotypes had been previously observed for yeast strains containing the following replacements for homologous residues in the three major yeast replicative DNA polymerases: Val, Lys and Met replacements for Leu868 in pol α [32,33], Val, Lys, Met, Phe, Gly and Asn replacements for Leu612 in pol δ [34,39,40] and Phe replacement for Met644 in pol ε [35]. Here we chose to test whether the same replacements for Rev3p-Leu979 might affect the accuracy of DNA synthesis by yeast pol ζ and thereby confer mutator phenotypes characteristic of mutagenic DNA synthesis activity by pol ζ in vivo. Thus, we began this study by constructing yeast strains with Gly, Val, Asn, Lys, Met or Phe replacements for Rev3p-Leu979.

3.2. UV-induced killing and mutagenesis

3.2.1. UV sensitivity and mutagenesis of the six rev3-L979X strains

Because pol ζ has a well-established role in UV-induced survival and mutagenesis, we first examined whether the rev3 missense mutations affected these two parameters. As expected, the rev3Δstrain was more sensitive to UV irradiation than was wild-type yeast (Fig. 1A). The rev3-L979K, rev3-L979G, rev3-L979N and rev3-L979V strains all had intermediate sensitivity, suggesting partial loss of pol ζ activity. In contrast, the rev3-L979F and rev3-L979M strains had wild-type sensitivity to UV irradiation, suggesting greater retention of pol ζ catalytic activity. When UV-induced mutant frequencies at the CAN1 locus were measured, the rev3Δ strain was immutable (Fig. 1B). In contrast, all strains with Rev3p-Leu979 replacements were more UV mutable than wild-type yeast (Fig. 1B), suggesting that these replacements reduce the fidelity of pol ζ as it participates in translesion DNA synthesis (TLS) of UV photoproducts.

Figure 1. UV-induced killing and mutagenesis in yeast strains harboring rev3-L979X alleles.

A. Survival following UV irradiation. B. UV-induced mutagenesis at the CAN1 locus. C. Survival following UV irradiation in rad30 Δstrains. D. UV-induced mutagenesis at the CAN1 locus in rad30 Δ strains. Error bars represent one standard deviation.

3.2.2. UV sensitivity and mutagenesis in the rad30Δrev3-L979F double mutant strain

In wild-type yeast, the majority of UV photoproduct bypass is conducted by pol η, the product of the RAD30 gene. However, in a rad30Δ strain, pol ζ has been implicated in mutagenic bypass [47,48]. Therefore, we next examined UV-induced sensitivity and mutagenesis in a rad30Δrev3-L979F double mutant strain. As expected based on earlier studies [47,49,50], deletion of RAD30 (rad30Δ) slightly reduces survival (Fig. 1C). However, the rad30Δrev3-L979F double mutant strain survives exposure to UV radiation at levels comparable to those for the rad30Δ single mutant (Fig. 1C). This again suggests that L979F pol ζ retains high pol ζ catalytic activity. In addition, the rad30Δrev3-L979F double mutant strain is several-fold more UV mutable than are the single mutant rev3-L979F or rad30Δ strains, and it is much more mutable than the single rev3Δ mutant or the double rad30Δ revΔ mutant strain (Fig. 1B and 1D). These results are consistent with participation of L979F pol ζ in mutagenic TLS of UV photoproducts when pol η is absent.

3.2.3. Specificity of UV-induced mutations in rev3-L979F mutant strains

In order to determine if the spectrum of mutations resulting from in vivo TLS by L979F pol ζ is similar to that of wild-type pol ζ, we determined the types of UV-induced mutations that arose at CAN1 in wild-type and rev3-L979F strains. We also sequenced can1 isolates from isogenic rad30Δ and rev3-L979F rad30Δ strains, which lack TLS by pol η. Yeast cells were irradiated with 20 J/m² of UVC, a dose that induces mutants in all four strains examined at frequencies that are much higher than the spontaneous frequencies (Fig. 1 and see below).

The UV-induced mutation spectrum in the wild-type strain contained primarily single base substitutions (76%), all of which occurred at dipyrimidine sequences. Of these, eight are consistent with misincorporation opposite the 3′ base of a photodimer and two are consistent with misincorporation opposite the 5′ base of a photodimer. An additional nine single base substitutions arose at tripyrimidine sequences and are thus ambiguous with respect to site of the misincorporation. Among transitions, G-C to A-T substitutions predominated, and among transversions, A-T to T-A changes predominated (Table 1). This specificity is consistent with earlier studies [see also 49 and references therein], and suggests that UV light induces mutations predominantly via C•A and T•T mispairs that preferentially arise at the 3′ base of photodimers. Additionally, the frequencies of UV-induced insertion/deletion (indel) and complex mutations are at least 100-fold higher than the spontaneous mutation rates for these events (compare Tables 1 and 2), suggesting that these mutations also result from mutagenic bypass of UV photoproducts.

Table 2.

Spontaneous mutations in wild-type and rev3 yeast strains.

	wild-type		rev3-L979F		rev3-L979M		rev3Δ
Rate (× 10−8)	26 [23 – 34]		42 [32 – 60]		46 [39 – 88]		20 [11-24]
	Events	Rate	Events	Rate	Events	Rate	Events	Rate
A-T to G-C	1	1	1	2	1	2	2	2
G-C to A-T	6	6	8	13	5	10	10	8
A-T to T-A	3	3	0	≤2	2	4	1	1
A-T to C-G	2	2	0	≤2	1	2	1	1
G-C to T-A	5	5	6	10	2	4	1	1
G-C to C-G	1	1	5	8	3	6	3	2
Tandem bps	1	1	0	≤2	1	2	0	≤1
Indelsa	5	5	0	≤2	8	15	5	4
Duplications	1	1	1	2	0	≤2	2	2
Complexb	0	≤1	4	7	1	2	0	≤1

All rates are (× 10⁻⁸). Numbers in brackets are 95% confidence intervals.

^aOne to twelve-base insertions and deletions were categorized as indels.

^bMultiple, closely spaced base substitutions and/or indels were categorized as complex mutations.

The UV-induced mutation spectrum of the rev3-L979F single mutant strain (Table 1 and Fig. 2) differed from that of wild-type yeast in several ways. Foremost was the presence of eight complex mutations (32% of can1 mutations; for sequences see Supplementary Table 2), which contain multiple, closely spaced base substitutions and/or indel mutations. The frequency of complex mutations was significantly higher in the rev3-L979F mutant compared to the wild-type strain (9-fold, p = 0.0046). This increase is consistent with mutagenic TLS of UV photoproducts by L979F pol ζ that is less accurate than TLS by wild-type pol ζ. The frequencies of tandem double base substitutions and G-C to T-A transversions were also elevated 5-7 fold in the rev3-L979F mutant. Because only 25 clones were sequenced, the number of each type of event is small, such that these increases are not statistically significant. All seven of single base substitutions from the rev3-L979F mutant strain that occurred at dipyrimidines are consistent with misincorporation opposite the 3′ base of the dipyrimidine. This specificity suggests that although L979F pol ζ is less accurate than wild-type pol ζ, it participates in a bypass reaction that retains the preference for misincorporation opposite the 3′ base of a photodimer.

Figure 2. Relative frequencies of UV-induced mutagenesis for various classes of mutations.

White, hatched, grey and solid bars correspond to wild-type, rev3-L979F, rad30 Δ and rev3-L979F rad30 Δ strains, respectively. Mutation frequencies (and one standard deviation) for each class were calculated by multiplying the proportion of the specific type of event in the corresponding mutation spectrum (Table 1) by the total mutation frequency. These data were collected after a dose of 20 J/m². As the data from the rad30 Δ mutant were more variable than those from the other strains, the error bars for the rad30 Δ mutant are comparatively large.

The rad30Δ single mutant (Table 1 and Fig. 2) had a significantly elevated frequency of UV-induced G-C to A-T transitions compared to the wild-type strain (3-fold, p = 0.019). Also slightly elevated were G-C to T-A transversions (4-fold) and complex mutations containing multiple, closely-spaced nucleotide changes (7-fold). Twenty of 21 single base substitutions from the rad30Δ strain occurred at dipyrimidine or tripyrimidine sequences. Of these, 11 are consistent with misincorporation opposite the 3′ base of a photodimer, two are consistent with misincorporation opposite the 5′ base of a photodimer and seven are ambiguous, as they occurred at tripyrimidine sites. Since pol ζ is present in the rad30Δ single mutant strain, these data are consistent with a contribution of pol ζ to mutagenic TLS of photoproducts that yields single base mispairs preferentially located at the 3′ base of photodimers, as well as complex errors (Table 1, Fig. 2). The complex mutations observed when pol ζ is present and pol η is absent are reminiscent of those generated by purified yeast pol ζ as it copies undamaged DNA in vitro [30].

The UV-induced mutation spectrum of the rad30Δ rev3-L979F double mutant showed combined characteristics of the spectra for the rad30Δ and rev3-L979F single mutant strains (Table 1, Fig. 2). The frequencies of three types of UV-induced mutations were significantly elevated in the rad30Δ rev3-L979F double mutant compared than those of wild-type: G-C to A-T transitions (8-fold, p = 0.009), G-C to T-A transversions (26-fold, p = 0.04), and tandem, double base substitutions (26-fold, p = 0.04). UV-induced complex mutations were also elevated 9-fold, although this elevation was not statistically significant. Of the sixteen base substitution errors in the rad30Δ rev3-L979F double mutant, eleven occurred at dipyrimidines and four at tripyrimidines. All of the mutations that occurred at dipyrimidines are consistent with misincorporation opposite the 3′ base, again implicating misinsertions opposite photodimers as the mutagenic event in rev3-L979F strains. This suggests that the lesions responsible for most of these substitutions in the double mutant strain are photodimers that are normally copied by pol η, such that in its absence (rad30Δ), L979F pol ζ participates in mutagenic TLS. The situation may be slightly different for indels and complex mutations, whose mutant frequencies are similar in the rad30Δ rev3-L979F double mutant and the rev3-L979F single mutant (Table 1, Fig. 2).

3.3. Minimal effect of the rev3-L979F allele on ogg1-dependent mutagenesis

To determine if the rev3-L979F allele has an effect on mutagenesis involving a spontaneously arising oxidative lesion, we examined mutation rates in ogg1Δ mutants, which are deficient for the glycosylase that removes 8-oxo-guanine from DNA. In REV3 strains, ogg1Δ results in an elevated mutation rate at CAN1, and this rate is further elevated in an ogg1Δ rad30Δ double mutant (Fig. 3A). These results are consistent with earlier studies that revealed the roles of Ogg1 and pol η in preventing G-C to T-A substitutions resulting from spontaneous oxidative stress [51,52]. Interestingly, rates are not further elevated in ogg1Δ or ogg1Δ rad30Δ mutant strains that harbor the rev3-L979F allele (Fig. 3A). Similar results were obtained for reversion of the ura3-29 reporter (Fig. 3B), which in an ogg1Δ background is a highly sensitive marker for G-C to T-A transversions [53]. These results imply that, unlike the situation for UV photoproducts, L979F pol ζ has little if any role in mutagenic bypass of spontaneously arising 8-oxo-guanine in DNA in vivo. This is consistent with an earlier study indicating that wild-type pol ζ does not have a major role in mutagenesis in an ogg1Δ strain [51].

Figure 3. Spontaneous mutation rates in ogg1, rad30 and rev3 single, double and triple mutant strains.

Mutation rates were measured for (A) the can1 forward and (b) ura3-29 reversion reporters. Mutation rates shown were normalized relative to the values observed in the wild-type strain, which were 1.72 × 10⁻⁷ (95% confidence limits: 1.10 – 2.19 × 10⁻⁷) for can1 and 6.57 × 10⁻⁹(95% confidence limits: 4.34 – 8.04 × 10⁻⁹) for ura3-29. Error bars represent 95% confidence limits.

3.4. Spontaneous mutation rates in rev3-L979F and rev3-L979M strains

In the above experiments, the spontaneous mutation rate at CAN1 for the rev3-L979F strain (3.1 × 10⁻⁷, 95% CI = 2.6 – 4.2) was 1.8-fold higher than for the wild-type strain (1.7 × 10⁻⁷, 95% CI = 1.1 – 2.2). In a second experiment, the rev3-L979F and rev3-L979M strains had spontaneous mutation rates that were 1.6-fold and 1.8-fold higher, respectively, than in a wild-type strain (Table 2), and 2.1-fold and 2.3-fold higher, respectively, than in the rev3Δ strain. Thus, the rev3-L979F and rev3-L979M alleles confer a weak spontaneous mutator phenotype to yeast. To determine if this effect was specific to any particular types of mutations, independent can1 mutants were sequenced. The results (Table 2) indicate that the rates of G-C to C-G transversions and of complex mutations involving multiple base changes in the rev3-L979F and rev3-L979M strains were higher than in the wild-type or rev3Δ strains. The rev3-L979M strain also had a higher rate of indels, all but one of which were single base deletions, and all of which occurred in non-iterated sequences or sequences with only two repeat units.

4. DISCUSSION

Our current understanding of the cellular functions of DNA polymerase ζ largely derives from comparing the phenotypes of normal cells to those in which the REV3 gene is disrupted or its expression is reduced. These approaches typically reduce both spontaneous and DNA damage-induced mutagenesis, indicating that DNA polymerase ζ contributes to mutagenic DNA synthesis in cells. Here we have taken a different and complementary approach by demonstrating that yeast cells expressing REV3 derivatives with amino acid replacements in the pol ζ active site have mutator phenotypes. The results are informative when compared to studies of other Family B polymerases, and also regarding the role of pol ζ in mutagenic bypass of UV photoproducts and in spontaneous mutagenesis.

4.1. Inferred activity of pol ζ L979X mutants

The reduced survival of rev3-L979V, rev3-L979N, rev3-L979G and rev3-L979K mutant strains following UV irradiation (Fig. 1A) suggests that pol ζ harboring any of these four amino acid replacements is less active than wild-type pol ζ. The inferred reduction in polymerase activity resulting from these replacements is consistent with the observation that yeast strains containing homologous L612N, L612G and L612K replacements in pol δ [40] progress more slowly through S phase and have increased sensitivity to DNA damaging agents. Nonetheless, the rev3-L979V, rev3-L979N, rev3-L979G and rev3-L979K mutant strains apparently retain some pol ζ activity, because their survival (Fig. 1A) and mutant frequencies (Fig. 1B) following UV irradiation are greater than those of a rev3Δstrain. An even higher level of pol ζ activity is apparently retained by the rev3-L979F and rev3-L979M strains, since their survival remains relatively normal following UV radiation (Fig. 1A), even when pol η-dependant bypass of UV-induced cyclobutane pyrimidine dimers is eliminated by deletion of RAD30 (Fig. 1C). Retention of robust polymerase activity by pol ζL979F and L979Mmutants is consistent with previous studies showing nearly wild-type polymerase activity in vitro for pol α L868F and L868M mutants [32,33], a pol δ L612M mutant [34] and a pol ε M644F mutant [35].

4.2. Mutator effect of pol ζL979X mutants compared to other Family B polymerases

All six amino acid replacements for Leu979 in yeast pol ζ result in increased UV-induced mutagenesis (Fig 1B) and the L979F and L979M replacements increase spontaneous mutagenesis (Table 2). Spontaneous mutator phenotypes have also been reported in yeast strains with similar replacements for the homologous amino acids in active site motif A of yeast pol α Leu868 [32,33], pol δ Leu612 [34,37,40]), and pol ε Met644 [35,36]. Those studies further revealed reduced DNA synthesis fidelity by the mutant polymerases, leading us to infer that this is likely to be the case for the pol ζ mutants studied here. If so, the identity of the side chain at this location is a determinant of the fidelity of DNA synthesis in vivo by each of the four eukaryotic Family B DNA polymerases. In the crystal structure of prokaryotic Family B member RB69 pol [31], Leu415 at this position interacts with an adjacent, invariant tyrosine (Tyr416), which in turn interacts with the ribose of the incoming dNTP. Given this location, amino acid replacements for Leu979 in pol ζ, or for homologous residues in pols α, δ and ε, may alter the geometry of the nascent base pair binding pocket and/or the chemistry of the reaction. This in turn may reduce selectivity against incorrect dNTP insertion and/or increase the probability of extending mismatched termini. Indeed, both parameters are affected by replacements for Leu868 in yeast pol α [32,33], L614 in yeast pol δ [34] and Met644 in yeast pol ε [35,36]. Attempts are currently underway to express and purify yeast pol ζ with the L979F replacement, in order to examine its properties in comparison to the wild-type pol ζ.

4.3. Mutagenic bypass of UV photoproducts by pol ζ L979X mutants

The spontaneous mutator phenotypes of the pol α, δ and ε mutant strains reported earlier revealed the importance of this conserved residue in the fidelity of replication of undamaged DNA. The UV-induced mutator phenotype of the pol ζ L979X mutants (Fig. 1) extends this by indicating that Leu979 is a determinant of the fidelity of pol ζ dependent translesion DNA synthesis in vivo. The mutational specificity data imply that the majority of the lesions responsible for UV-induced mutagenesis are dipyrimidine photoproducts. In the wild-type and rev3-L979F strains, where cyclobutane pyrimidine dimers (CPDs) can be efficiently bypassed by pol η, the mutagenic lesions may primarily be 6-4 photoproducts. In the rad30 Δ and rad30 Δ rev3-L979F strains, which lack pol η, a greater contribution of CPDs to pol ζ-dependent mutagenesis seems likely. In all four strains, there is a bias for UV-induced base substitutions at the 3′ base of dipyrimidines. Given the increased frequency of UV-induced base substitutions in rev3-L979F strains (Table 1, Fig. 2) compared to REV3 strains, synthesis by L979F pol ζ is inferred to contribute to mutagensis opposite the 3′ base during photodimer bypass in vivo. Whether this reflects L979F pol ζ misinsertions opposite the damaged bases, extension of mismatched primers, or both, awaits biochemical analysis. However, this in vivo specificity is consistent with kinetic data showing that wild-type yeast pol ζ preferentially misinserts nucleotides opposite the 3′ T as compared to the 5′ T of a cis-syn cyclobutane TT dimer [28,54]. This same bias is shared by human pol η [55] and may reflect differences in base-base hydrogen bonding potential for the 3′ versus 5′ base of a photodimer bound in the polymerase active site, as has been observed for Sulfolobus sulfataricus Dpo4 [56].

Also of note are the higher frequencies of UV-induced tandem double base substitutions in the rev3-L979F and rad30 Δ rev3-L979F mutant strains in comparison to their REV3 counterparts (Fig. 2). None of these were a CC to TT change that could have resulted from “correct” insertion of two dAMPs opposite uracils created by deamination of cytosine in photodimers. Thus, the L979F replacement renders pol ζ more capable of contributing to mutagenesis that requires an initial misinsertion, a second misinsertion from a mismatched primer terminus, and extension of the resulting doubly mismatched terminus. This remarkable promiscuity extends to an even more diverse array of complex UV-induced mutations containing two or more substitutions, deletions and/or insertions (Table 1, Fig 2). These mutations, which are more frequent in the rev3-L979F strains (Table 1, Fig. 2), require formation and extension of aberrant primer-templates having multiple mismatches of different types. The results presented in Table 1 are consistent with the possibility that pol ζ may contribute to the UV-induced triplet mutation formation described in a recent report [57].

4.4. Contribution of pol ζ L979F/M mutants to spontaneous mutagenesis

The remarkable promiscuity of L979F pol ζ revealed in the UV-induced mutation spectra is also apparent during spontaneous mutagenesis, in the form of an increased rate of complex mutations (Table 2). These complex events could result from mutagenic bypass of endogenously arising lesions in DNA. Supporting this possibility is a previous study showing that wild-type pol ζ contributes to formation of spontaneous, complex mutations in a yeast strain defective in nucleotide excision repair [9]. That study provided evidence that the responsible lesion was a modified guanine resulting from spontaneous oxidative stress. This is particularly interesting in relation to our observation that the rev3-L979F allele does not contribute strongly to mutagenesis in ogg1 Δ strains (Fig. 3). This implies that, unlike the situation for UV photoproducts, L979F pol ζ may have little if any role in mutagenic bypass of spontaneously arising 8-oxo-guanine in DNA in vivo. This does not exclude a role for L979F pol ζ in bypass of other oxidative lesions, e.g., those that may be subject to nucleotide excision repair [9] and/or toxic oxidation products of 8-oxo-guanuine [58]. Indeed, some secondary oxidation products of 8-oxo-guanuine result in GC to C-G transversions [59], which is interesting because G-C to C-G transversions are elevated in rev3-L979F and rev3-L979M strains (Table 2).

An additional, nonexclusive possibility is that some of the spontaneous complex mutations and G-C to C-G substitutions seen in the rev3-L979F and rev3-L979M strains result from L979F/M pol ζ-dependent copying of undamaged DNA. This possibility is consistent with the error specificity of wild-type yeast pol ζ as it copies an undamaged DNA template in vitro [30]. Compared to the major replicative DNA polymerases, wild type pol ζ has lower fidelity, yet it can perform processive DNA synthesis [60]. Interestingly, wild type pol ζ generates lacZ mutants in vitro [30] and L979F pol ζ generates can1 mutants in vivo that contain multiple sequence changes separated by several correct base pairs (Table 2; including a mutant with three substitutions in a stretch of 18 base pairs). This combination of inaccurate but processive synthesis is shared only by DNA polymerase κ [61], making these two TLS polymerases prime candidates for generating mutational “showers” in genomes [62]. Such showers of multiple, dispersed sequence changes within a single mutant clone have important implications for the evolutionary fitness of organisms and for the origins of diseases [63].

4.5. Utility of pol ζ mutator alleles

As mentioned in the Introduction, inactivation of the REV3 gene in mice is lethal [41-43,64]. Thus, the present search for mutator alleles of yeast pol ζ that retain robust pol ζ catalytic activity was partly motivated by the desire to study the function of mammalian pol ζ using homologous mouse rev3 missense mutants that are predicted to be viable. The genetic results presented here suggest that mutations equivalent to yeast rev3-L979F and rev3-L979M may be ideal, in that polymerases with these changes appear to retain high polymerase activity but they appear to have lower fidelity. If the equivalent mouse pol ζ mutants are viable and confer mutator effects, then given the multiple functions of pol ζ mentioned in the Introduction, it will be interesting to determine if mice harboring these rev3 alleles have altered sensitivity to environmental stress, altered rates of spontaneous and/or DNA damage induced mutagenesis, altered susceptibility to environmental diseases (e.g., cancer), or altered somatic hypermutation of immunoglobulin genes.