Proficient Replication of the Yeast Genome by a Viral DNA Polymerase

DNA replication in eukaryotic cells requires minimally three B-family DNA polymerases: Pol α , Pol δ and Pol ε . Pol δ replicates and matures Okazaki fragments on the lagging strand of the replication fork. Saccharomyces cerevisiae Pol δ is a three-subunit enzyme (Pol3-Pol31-Pol32). A small C-terminal domain of the catalytic subunit Pol3 carries both iron-sulfur cluster and zinc binding motifs, which mediate interactions with Pol31, and processive replication with the replication clamp PCNA, respectively. We show that the entire N-terminal domain of Pol3, containing polymerase and proofreading activities, could be effectively replaced by those from bacteriophage RB69, and carry out chromosomal DNA replication in yeast with remarkable high fidelity, provided adaptive mutations in the replication clamp PCNA were introduced. This result is consistent with the model that all essential interactions for DNA replication in yeast are mediated through the small C-terminal domain of Pol3. The chimeric polymerase carries out processive replication with PCNA in vitro , however, in yeast, it requires an increased involvement of the mutagenic translesion DNA polymerase ζ during DNA replication.

ε 2 replicates the leading strand of the replication fork, whereas Pol α-primase initiates Okazaki fragments on the lagging strand that are elongated and matured by Pol δ (6). This simple "division of labor" model is still a matter of debate (7)(8)(9). Furthermore, under certain conditions, such as those of replication restart following DNA recombination, Pol δ carries out substantial DNA synthesis of both strands (10). The fourth B-family enzyme, Pol ζ, is required for translesion synthesis in response to DNA damage, which results in the bulk of damage-induced mutagenesis in eukaryotes (11). Pol ζ also participates in replication past structural blocks when normal replication forks stall (12).
B-family DNA polymerases are ubiquitous; they are found in eukaryotes, bacteria, archaea, and in both bacterial and eukaryotic DNA-based viruses (13). All B-family enzymes contain three conserved domains: a structural N-terminal domain (NTD), a 3'-5' exonuclease domain, and the polymerase domain containing the palm, finger, and thumb subdomains. The NTD is highly conserved, but a specific function for this domain has only been assigned to some archaeal enzymes, in which the NTD recognizes template uracil residues and inhibits continued replication by the DNA polymerase (14,15). The exonuclease domain serves to carry out proofreading of polymerase errors in most enzymes. However, eukaryotic Pol α and Pol ζ, while maintaining this structural domain, lack exonuclease activity. The polymerase domain carries out highfidelity DNA synthesis, with the notable exception of the translesion synthesis enzyme Pol ζ (16)(17)(18).
The cellular eukaryotic members of the Bfamily are structurally more complex in that they are multi-subunit enzymes, and secondly, they uniquely contain an additional, small C-terminal domain (CTD) in the polymerase subunit, which mediates interactions with these accessory subunits (13,19). The CTD sequences of the four eukaryotic enzymes are highly conserved suggesting a common 3D structure of the CTD. Only the structure of the CTD of Pol α has been determined (19,20). It shows an elongated, bilobal form, in which the two lobes are connected by a three-helical bundle. Each lobe contains four conserved cysteines ( Figure 1A). In the Pol α CTD structures, both 4-cysteine lobes bind zinc. However, biochemical studies of Pol δ and Pol ζ have shown that the C-terminal 4-cysteine lobe ligands an iron-sulfur cluster in the [4Fe-4S] 2+ coordination state (21,22). The CTDs of Pol α, Pol δ, and Pol ε each bind a distinct B subunit, called Pol12, Pol31, and Dpb2, respectively, in budding yeast (19,21,23), and these B subunits show both sequence and structural conservation (19,20,24,25). Pol ζ has appropriated the B subunit from Pol δ to elaborate its own 4-subunit assembly (Rev3-Rev7-Pol31-Pol32) (22,(26)(27)(28).
In order to better understand how the multisubunit structures of eukaryotic replicative DNA polymerases are intricately tied to their function, we have used the lagging strand polymerase Pol δ as a model. Budding yeast Pol δ consists of the catalytic subunit Pol3 and the accessory subunits Pol31 and Pol32 (29). Interactions between Pol3 and Pol31 occur through the Pol3 CTD and require an intact iron-sulfur cluster (21). Pol31 binds the third subunit Pol32 to form the complete heterotrimeric polymerase complex (29). This architecture of Pol δ is conserved in other organisms (25,30), except for the presence of an additional, small regulatory subunit in fission yeast and in mammals (31,32). Pol δ alone is a low-processivity enzyme, replicating only a few nucleotides before dissociating from DNA. This problem is overcome through interactions with the replication clamp proliferating cell nuclear antigen (PCNA) (33). PCNA is a donutshaped homotrimeric protein that is loaded onto DNA template-primer termini by Replication Factor C (RFC) in an ATP-dependent manner (34,35). DNA-bound PCNA then recruits Pol δ and increases both the catalytic activity and the processivity of the enzyme, so that it can rapidly replicate hundreds of nucleotides in a single DNA-binding event (36)(37)(38). PCNA-dependent polymerase processivity is vital to efficient genomic DNA replication. Pol δ mutants that are compromised for interactions with PCNA, exhibit in vitro processivity defects that, if severe, are associated with lethality in yeast (21,39,40).
We were interested in understanding what structural domains of Pol δ are required for efficient replication of the budding yeast genome. While mutations that inactivate polymerase activity cause lethality in yeast (41), mutations that abrogate exonuclease activity are viable but cause fidelity defects (42). However, it is possible that structural determinants in the NTD, or in the two catalytic domains may be essential for replisome activity. The overall structure of these three domains is conserved in B-family DNA polymerases, as shown by the superimposition of the structure of bacteriophage Rb69 DNA polymerase with that of the same domains of Pol3 ( Figure 1A). Lacking from the Pol3 structure is its CTD, which mediates interactions with the accessory subunits and, both directly and indirectly, with PCNA. We hypothesized that the essential factors enabling Pol δ to act in a eukaryotic setting are the ability to bind its accessory subunits and PCNA. In order to test this hypothesis, we created a chimeric polymerase subunit by replacing the Pol3 NTD and catalytic core domains with those from the structurally homologous bacteriophage RB69 DNA polymerase. Rb69 and T4 are closely related bacteriophages that use a polymerase processivity model similar to Pol δ, containing a homotrimeric clamp and an ATP-dependent clamp loader (gp45 and gp44/62, respectively) (43).
We found that fusing the 104 kDa RB69 DNA polymerase to the 13 kDa CTD of Pol3 was sufficient to form a three-subunit polymerase complex with Pol31 and Pol32 in yeast. The processivity of this polymerase complex was stimulated by PCNA, but processivity was compromised as compared to Pol δ. We obtained more robust stimulation of this engineered form of Pol δ when we introduced two adaptive mutations in PCNA, and this genetic arrangement conferred growth in yeast that contained the fusion polymerase as only source of Pol δ. Remarkably, when we eliminated fidelity-lowering contributions made by the mutagenic Pol ζ, the fidelity of the engineered Pol δ approximated that of the native enzyme.
Genetic techniques -In order to make yeast strains containing a chromosomal copy of the pol30-rb2 allele, integrating plasmid pBL248-rb2, and pBL248 as control was cut with HpaI, which cuts once in the LEU2 gene, and transformed into the appropriate leu2-3,112 strains to leucine prototrophy. To determine phenotypes of the pol3-69 allele, the appropriate pol3Δ strains, containing pBL304 as complementing plasmid, were transformed with pBL326, or pBL309 as positive control, with Trp selection, and transformants were passed over 5fluoroorotic acid-containing media (5-FOA) to evict complementing plasmid pBL304 (POL3 URA3).
DNA damage sensitivity assays were carried using standard protocols. Fluctuation analyses to determine spontaneous mutation rates were carried out in triplicate with 15-20 independent cultures, and analyzed by the median (46).

Identification of PCNA suppressor mutants -The
POL30 gene in pBL249 was PCR-mutagenized as described (47). The library was transformed into PY227 containing both pBL304 and pBL326, and plated onto SC-Leu media, and after 2 days of growth, replica-plated onto SC-Leu plates containing 5-FOA, to evict the pBL304 plasmid. Plasmid DNA was isolated from positive colonies and re-applied to the same screen. The pBL249 isolates from the second screen that allowed yeast growth without pBL304 were sequenced. The most robust suppressor pol30-rb1 carried six nonsynonymous mutations (F12Y, D17A, Q29H, K31R, I52M, I100T). Each mutation was separately reverted back to wild-type and loss of suppression assessed. From this analysis, we determined that the Q29H mutation was essential for suppression, and K31R increased suppression. Therefore, pol30-rb2 contains only the Q29H and K31R mutations.
DNA replication assays -Assays contained 20 mM Tris-HCl pH 7.8, 1 mM DTT, 100 µg/ml bovine serum albumin, 8 mM magnesium acetate, 0.5 mM ATP, 100 µM each of dCTP, dGTP, and dTTP, 10 mM of [α-32 P]dATP, 100 mM NaCl, 3.5 nM singlestranded bluescript DNA, singly primed at positions 592-621, either with a 30-mer DNA primer or 5'-RNA 8 DNA 22 primer, 400 nM RPA, and PCNA or pcna-rb2 as indicated. PCNA was loaded onto the primed DNA by incubation with 7 nM RFC at 30 °C for 1 min prior to reaction initiation. Reactions were initiated by addition of 7 nM Pol δ or RbPol δ. In the assays in Figure 2D, 7 nM FEN1 and 14 nM DNA ligase I were added together with the polymerase. Aliquots were taken at various time points and stopped with 50 mM EDTA and 0.2% SDS, final concentration. Reactions were either resolved on a 1% alkaline agarose gel ( Figure 2B) or a 1% neutral agarose gel containing 0.5 µg/ml ethidium bromide. Gels were dried and documented by PhosphorImager analysis (GE Healthcare). Alternatively, 1 ml of 10% trichloroacetic acid was added to stopped replication samples. After 10 min on ice, the mixture was filtered over a GF/C filter, The filter was washed twice with 2 ml of 1M HCl and 0.05 M sodium pyrophosphate, rinsed with ethanol, dried, and counted in counting fluid in a liquid scintillation counter. All assays were carried out in duplicate or triplicate, and either representative gels are presented or standard errors are shown ( Figure 2C).

RESULTS AND DISCUSSION
Designing the Rb69-Pol3 polymerase fusion gene -Bacteriophage T4 expresses a replication elongation apparatus consisting of a B-family DNA polymerase, a homotrimeric replication clamp gp45, which is the ortholog of eukaryotic PCNA, and an ATPdependent clamp loader. While extensive biochemical and genetic DNA replication studies are available for the T4 system (48,49), we focused our attention on the highly related bacteriophage Rb69, because its DNA polymerase has been the subject of detailed structural characterization (3,50). Rb69 DNA polymerase can efficiently substitute for T4 DNA polymerase in faithfully replicating the T4 genome (51). The closest eukaryotic homologue to these bacteriophage enzymes is Pol3, the catalytic subunit of Pol δ. T4 and Rb69 DNA polymerase (Rb-Pol) not only carry out high-fidelity DNA replication, but are also responsible for the proper maturation of Okazaki fragments during phage DNA replication. The latter activity is allocated solely to Pol δ in eukaryotic cells (52). Figure 1A shows a structural comparison between Rb69-Pol and aa95-985 of the 1097aa yeast Pol3 (3,53). The structures of both enzymes were solved in a complex with template-primer and a base-paired dNTP. The Pol3 structure comprises the structured NTD and the exonuclease and polymerase domains, but lacks the unstructured N-terminal tail and its CTD. The CTD of Pol α serves as a structural model for this domain in the other B-family DNA polymerases (19,20).
We decided to fuse Rb69-Pol (1-896), which lacks only the C-terminal 7aa that mediate interactions with its gp45 clamp (54), to the CTD (981-1097) of Pol3 ( Figures 1A and 2A). This CTD contains a putative PCNA-binding motif (996-1005) (40), and the two 4-cysteine cluster metal binding sites, starting at aa1009 (21). The fusion gene is designated as pol3-69 and the resulting three-subunit variant of Pol δ as RbPol δ. First, we established that the fusion polypeptide contained the necessary determinants for expressing a stable 3-subunit enzyme in yeast, which it does (Figure 2A). Preliminary biochemical studies showed that the replication activity by the fusion enzyme was stimulated by PCNA, but much less so than wildtype Pol δ (see below). Therefore, it was not surprising that the pol3-69 fusion gene failed to complement the lethality of a pol3Δ mutant ( Figure  1B). Among the potential reasons for this failure to complement could be: (i) that the fusion protein lacked essential interactions with other replication proteins, e.g. though its NTD; (ii) that either the fidelity or rate of replication by the RB69 catalytic domains was incompatible with yeast genome replication; (iii) that, for structural reasons, the fusion protein failed to properly present its PCNAbinding domains to PCNA for highly processive DNA replication. We pursued the latter possibility, particularly because we noted that the PCNA binding motifs on the CTD of the catalytic subunit are located close to the fusion point. We therefore tested whether we could select for PCNA mutations that might ameliorate the processivity defect and thereby allow growth of pol3-69. A yeast pol3Δ strain containing both POL3 and pol3-69 on separate plasmids was transformed with a heavily mutagenized POL30 library, encoding PCNA. Transformants were replica-plated onto 5fluoroorotic acid-containing media (5-FOA), which evicted the wild-type POL3 plasmid, enforcing viability of the pol3-69 mutant for cell growth. We isolated two PCNA suppressor mutants of which only one, designated pol30-rb1, showed robust growth. The pcna-rb1 mutant carried six amino acid changes. By subsequent elimination analysis, we determined that the Q29H mutation was essential for suppression of lethality, while the additional K31R mutation increased the efficiency of suppression to that of the pol30-rb1 suppressor containing all six mutations ( Figure 1B and data not shown). These two mutations are localized adjacent to each other on the outer rim of the PCNA donut, close to the interaction pocket of many PCNA-interacting proteins ( Figure 1C). All further studies were carried with this double mutant, which we designate as pol30-rb2.
Biochemical activities of RbPol δ -We next investigated the replication properties of RbPol δ with either wild-type PCNA or the double mutant pcna-rb2 ( Figure 2B). While wild-type PCNA stimulated the replication activity of RbPol δ ( Figure  2B, compare lanes 6,7 with 5), it did not replicate as efficiently as Pol δ. The defect was somewhat suppressed at higher concentrations of PCNA (lanes 8,9; Figure 2C), consistent with an impaired stability of the the DNA-PCNA-RbPol δ complex. Significantly, the mutant pcna-rb2 clamp largely suppressed this processivity defect, allowing more rapid DNA synthesis at lower concentrations than wild-type PCNA did ( Figure 2B,C). Rb69 DNA polymerase itself showed no processive DNA synthesis with either wild-type PCNA or pcna-rb2.
In addition to the elongation of Okazaki fragments, another essential function of Pol δ is the maturation of these fragments (55). During this process, Pol δ coordinates with the flap endonuclease FEN1 to remove a 7-10 nt RNA primer and replace it with DNA during a process called nick translation, in order to generate a DNA-DNA nick that can be sealed by DNA ligase I. In our biochemical assay, the polymerizing complex encounters an 8 nt RNA primer when it has completely replicated around the 3 kb DNA circle as shown in Figure 1D. The RNA is degraded by iterative steps of Pol δ-mediated strand displacement synthesis of 1-2 ribonucleotides, followed by FEN1 cutting of the emerging 5'-flap (56). Finally, after all RNA has been degraded, DNA ligation is mediated by DNA ligase I. With wild-type Pol δ and PCNA, this reaction is essentially complete after 3 min, and substituting pcna-rb2 did not affect the kinetics ( Figure 2D). In contrast, Rb-Pol δ only completed replication and subsequent Okazaki fragment maturation when the suppressor pcna-rb2 was present, and not with wild-type PCNA. These data suggest that the lethality of the pol3-69 fusion mutant may result not just from inefficient elongation of replication, but perhaps even more from the inability to perform efficient Okazaki fragment maturation, with the suppressor mutant pol30-rb2 largely overcoming these deficiencies.
Fidelity defects associated with Rb69 polymerase activity -Having established that the suppressor pcna-rb2 largely restored processive functionality to RbPol δ in vitro, we next asked which potential defects were associated with the genome being replicated by RbPol δ. All genetic studies were carried out in a POL30/pol30-rb2 heterozygous background, comparing the phenotypes of pol3-69 with that of POL3. While the pol3-69 fusion allele showed robust growth at 30 ˚C, it was cold-sensitive for growth at 15 ˚C ( Figure 3B). Secondly, the Pol32 subunit is non-essential in yeast, even though many phenotypic defects are associated with pol32Δ mutants (29,(56)(57)(58). However, pol32Δ showed synthetic lethality with pol3-69, suggesting that the activity of RbPol δ lacking Pol32 was unacceptably compromised ( Figure 3A). The pol30-69 mutant was sensitive to the replication inhibitor hydroxyurea ( Figure 3B), but not to the topoisomerase inhibitor camptothecin, which induces double stranded breaks (data not shown). However, the mutant was significantly more sensitive to UV irradiation than wild-type POL3.
We combined the pol3-69 allele with a deletion of REV3, the catalytic subunit of Pol ζ, and/or with a deletion of RAD30, which encodes Pol η. Pol ζ is responsible for the bulk of damageinduced mutagenesis in the cell (11,59), and Pol η mediates mostly error-free bypass of pyrimidine dimers (60). While defects in these damage-response mechanisms showed a slight increase in damage sensitivity, it was not profound, suggesting that no specific pathway was inactivated in pol3-69 ( Figure  3B).
Despite being responsible for the replication of a relatively small genome, Rb69 DNA polymerase shows a remarkably high replication fidelity (61). We determined whether this high fidelity phenotype was preserved in yeast, using the CAN1 gene as a target for forward mutagenesis. In the pol3-69 mutant, canavanine-resistant mutations occurred at an 8-fold increased rate compared to wild-type ( Figure 3C). However, defects in the stability of replication complexes can induce the recruitment of Pol ζ, which results in an increased accumulation of mutations (12,(62)(63)(64). This process is called DRIM (defective replisome-induced mutagenesis). DRIM is under analogous genetic control as damageinduced mutagenesis (62,65). Therefore, we repeated the fluctuation analysis in a rev3Δ strain, defective for Pol ζ. Indeed, the pol3-69 rev3Δ mutant showed a strongly reduced mutator phenotype, being only ~3-fold higher than that of POL3 rev3Δ. An analysis of the spectrum of mutations obtained showed that by far the largest class of mutations in the pol3-69 single mutant are GCàCG transversion mutations that are a classical signature of Pol ζ-and Rev1-dependent activity (Table 1) (17,41,66). Indeed, they are not observed in the pol3-69 rev3Δ double mutant. Other types of mutations that are substantially enhanced in pol3-69 compared to pol3-69 rev3Δ are ATàTA transversions and complex mutations, also consistent with Pol ζ-and Rev1dependent activity (62,66). When the mutation spectrum of the pol3-69 rev3Δ strain is compared to that of POL3 rev3Δ (63,64,67), substitution mutations in all classes are somewhat enhanced, but the largest increase attributable to RbPol δ are in deletion formation.
Half of the mutants in pol3-69 rev3Δ are due to intermediate size deletions (11-64 nt) between direct repeats, 4-8 nt in length (Table I). These deletions are caused by primer misalignment during lagging strand replication by RbPol δ. When Pol ζ is functional, the rate of formation of these deletions is not significantly altered, suggesting that the misaligned primer does not provoke a TLS response by Pol ζ. Interestingly, the same 4-8 nt direct repeats that cause deletion formation in pol3-69, induce duplications in a rad27Δ strain that is defective for FEN1 flap endonuclease, and therefore compromised in Okazaki fragment maturation (68).
Our analysis has shown that the catalytic polymerase and domains of Pol δ can be substituted with those from a bacteriophage DNA polymerase, provided adaptive mutations are made in PCNA. The N-terminal domain is structurally conserved in all Bfamily DNA polymerases, and in archaea serves a specific function in the recognition of template uracil residues (15). The function of the NTD in other organisms remains to be determined, but our analysis shows that this NTD does not specify organismspecific essential functions.  Alkaline agarose gel electrophoresis of replication products with purified proteins as indicated. Schematic is shown and described in Experimental Procedures. (C) PCNA titration; replication assays were performed with the indicated proteins, as in B, for 60 sec. Incorporation of [α-32 P] dNTPs was determined by scintillation counting. Activity is represented relative to that of Pol δ with saturating PCNA. Rb69-Pol, RB69 DNA polymerase (D) Okazaki fragment maturation assay; replication products were resolved on an agarose gel containing 0.5 µg/ml ethidium bromide. Replication assays were performed as in B, except for the addition of both FEN1 and DNA ligase I along with polymerase and dNTPs upon reaction initiation (see Experimental Procedures). Labels at left indicate positions of nicked double-stranded DNA and closed circular double-stranded DNA. The latter has a high mobility in an ethidium bromide-containing gel. The pol3-69 mutation shows synthetic lethality with pol32Δ. Growth of either POL3 or pol3-69 in PY236 (POL30/pol30-rb2) and PY243 (POL30/pol30-rb2 pol32Δ) on 5-FOA media, which evicts complementing plasmid pBL304 (POL3 URA3), was monitored. (B) Sensitivity of the pol3-69 POL30/pol30-rb2 strain to low-temperature growth and to DNA damaging agents. Serial ten-fold dilutions of strains PY236 (REV3 RAD30), PY237 (rev3Δ), PY238 (rad30Δ), or PY239 (rev3Δ rad30Δ), containing either POL3 or pol3-69. All strains contain pol30-rb2 integrated into the chromosomal LEU2 locus. HU, hydroxyurea. (C) Spontaneous forward mutation rates (with 95% confidence intervals) to canavanine resistance, of PY236 and PY237, containing either POL3 or pol3-69.

Figure 1
by guest on March 22, 2020