Identification of rad27 Mutations That Confer Differential Defects in Mutation Avoidance, Repeat Tract Instability, and Flap Cleavage

Abstract

In eukaryotes, the nuclease activity of Rad27p (Fen1p) is thought to play a critical role in lagging-strand DNA replication by removing ribonucleotides present at the 5′ ends of Okazaki fragments. Genetic analysis of Saccharomyces cerevisiae also has identified a role for Rad27p in mutation avoidance. rad27Δ mutants display both a repeat tract instability phenotype and a high rate of forward mutations to canavanine resistance that result primarily from duplications of DNA sequences that are flanked by direct repeats. These observations suggested that Rad27p activities in DNA replication and repair could be altered by mutagenesis and specifically assayed. To test this idea, we analyzed two rad27 alleles, rad27-G67S and rad27-G240D, that were identified in a screen for mutants that displayed repeat tract instability and mutator phenotypes. In chromosome stability assays, rad27-G67S strains displayed a higher frequency of repeat tract instabilities relative to CAN1 duplication events; in contrast, the rad27-G240D strains displayed the opposite phenotype. In biochemical assays, rad27-G67Sp displayed a weak exonuclease activity but significant single- and double-flap endonuclease activities. In contrast, rad27-G240Dp displayed a significant double-flap endonuclease activity but was devoid of exonuclease activity and showed only a weak single-flap endonuclease activity. Based on these observations, we hypothesize that the rad27-G67S mutant phenotypes resulted largely from specific defects in nuclease function that are important for degrading bubble intermediates, which can lead to DNA slippage events. The rad27-G240D mutant phenotypes were more difficult to reconcile to a specific biochemical defect, suggesting a structural role for Rad27p in DNA replication and repair. Since the mutants provide the means to relate nuclease functions in vitro to genetic characteristics in vivo, they are valuable tools for further analyses of the diverse biological roles of Rad27p.

The Saccharomyces cerevisiae Rad27p belongs to a family of evolutionarily conserved exo- and endonucleases that are involved in DNA replication and repair (12, 13, 36, 41, 46). Biochemical studies, which have included analyses of cell extracts that are competent for simian virus 40 DNA replication, indicated that Fen1p, the mammalian homolog of Rad27p, is required in lagging-strand DNA replication (11, 22, 26, 50, 51, 53). In vitro, Fen1p and Rad27p display a 5′-to-3′ exonuclease activity and an endonuclease activity that is specific to nucleic acid branch structures (1, 14, 20). The 5′-to-3′ exonuclease activity is thought to be required to remove single ribonucleotides that remain at the 5′ ends of Okazaki fragments following cleavage of the RNA primer by RNase H (51). The endonuclease activity is hypothesized to be required for the cleavage of flap structures that are generated as the result of DNA synthesis from an upstream Okazaki fragment that displaces a downstream RNA primer (12, 13). Both activities are thought to be directed by the same catalytic site in Fen1p and Rad27p (19, 21, 29). Subsequent studies have shown that Fen1p and Rad27p track along the length of the single-stranded DNA tail before cleaving at the branch junction (16, 33). This cleavage reaction also is stimulated by polymerase processivity factor PCNA through the enhancement of Fen1p binding at the point where the flap anneals to the template, i.e., the cleavage site (5, 9, 25, 28, 49, 56).

Genetic analyses of S. cerevisiae also have supported a role for RAD27 in DNA replication. rad27Δ mutants display a conditional growth phenotype. At 37°C, rad27Δ cells arrest in S phase and display a single nuclear body (37, 42, 52). Consistent with a role in DNA replication, strains bearing rad27 mutations in conjunction with a pol3 (Polδ) mutation are inviable (7, 27). Some of the defects observed in rad27Δ strains, including conditional viability, are suppressed by overexpression of 5′-to-3′ exonuclease Exo1p, suggesting that some of the Rad27p functions are redundant (47). This idea is also supported by the observation that rad27Δ exo1Δ strains are inviable (35, 47).

In addition to exhibiting DNA replication defects, rad27Δ strains display DNA repair defects and a chromosome instability phenotype. The rad27Δ strains are highly sensitive to the DNA-damaging agent methyl methanesulfonate (MMS) but are only moderately sensitive to UV light and are insensitive to gamma irradiation (37, 52). The rad27Δ mutants also display synthetic lethality with mutants defective in the RAD52 double-strand break repair (DSBR) pathway, suggesting that replication lesions generated in the absence of Rad27p are processed by homologous recombination (45, 48). In assays that measure chromosome instability, rad27Δ strains display a complex mutator phenotype. The rad27Δ mutants have a high rate of forward mutations to canavanine resistance; the mutations primarily consist of duplications within the CAN1 gene of 5- to 100-bp DNA sequences that are flanked by direct repeats (48). Duplication mutations similar to those found in rad27Δ strains also are observed in human tumors and inherited diseases (reviewed in reference 48). In an assay that measures dinucleotide repeat tract instability, rad27Δ strains display instability rates similar to those observed for mismatch repair mutants (23, 42). Double-mutant and mutation spectrum analyses, however, showed that Rad27p plays a role in mutation avoidance that is distinct from mismatch repair (48). Based on these observations and the finding that rad27Δ rad52Δ double mutants are not viable, Tishkoff et al. (48) hypothesized that Rad27p endonuclease activity prevents mutations by efficiently cleaving 5′ flaps generated through the displacement of downstream Okazaki fragments by extension of the upstream fragment. In their model, the absence of Rad27p cleavage results in DNA breakage at the flap junction that is repaired by a nonmutagenic DSBR mechanism or by a slip-pairing DSBR mechanism that results in duplication mutations.

The wide range of defects observed in rad27Δ strains in DNA replication, DNA repair, and chromosome instability assays suggests that Rad27p may perform partly or completely different functions in each of these processes. Alternatively, these defects could be due to the disruption of a single function. To explore this issue, we characterized two rad27 alleles, rad27-G67S and rad27-G240D, that were identified in a screen for mutants that displayed both a repeat tract instability and mutator phenotype. Strains containing either rad27 mutation were viable at 37°C and highly resistant to MMS and did not display synthetic lethality with an exo1Δ mutation. However both rad27 mutations conferred synthetic lethality with the rad52Δ mutation. The two rad27 mutations produced distinct properties in canavanine resistance and repeat tract instability assays. In addition, the Rad27 mutant proteins displayed differences in their biochemical properties. This suggests that distinct structures in Rad27p are responsible for different functions in replication and repair.

MATERIALS AND METHODS

Media and chemicals.

Yeast strains were grown in either yeast extract-peptone-dextrose (YPD) or minimal selective media (39). Sporulation plates were prepared as described previously (6). When required, canavanine was included in minimal selective media at 60 mg/liter (39) and MMS (Aldrich) was included in YPD media at 0.004 to 0.020% (vol/vol). 5-Fluoro-orotic acid (5-FOA) was purchased from U.S. Biologicals and used as described previously (2).

S. cerevisiae strains.

The genotypes of all strains used in these studies are shown in Table 1. All strains were derived from the isogenic FY strain background (55). The msh2Δ::hisG, rad52Δ::URA3, exo1Δ::HIS3, and rad27Δ::HIS3 alleles have complete or nearly complete coding region deletions of their respective genes and were introduced into FY23 and FY86 by single-step transplacement. Double-mutant combinations of the alleles described in Table 1 were made by standard crosses (39). The genotypes of all of the alleles was confirmed by PCR analysis and DNA sequencing of chromosomal DNA isolated from the transformed strains.

TABLE 1.

Strains used in this study

Strain	Relevant genotype
FY23	MATaura3-52 leu2Δ1 trp1Δ63
FY86	MATα ura3-52 leu2Δ1 his3Δ200
EAY281	MATaura3-52 leu2Δ1 trp1Δ63 msh2Δ::hisG
EAY432	MATaura3-52 leu2Δ1 trp1Δ63 rad52::URA3
EAY545	MATα leu2Δ1 his3Δ-200 ura3-52 rad27Δ::HIS3
EAY595	MATaleu2Δ1 ade8 rad27-G67S
EAY596	MATaleu2Δ1 ade8 rad27-G240D
EAY597	MATα leu2Δ1 his3Δ-200 rad27-G240D
EAY607	MATα leu2Δ1 his3Δ200 ura3-52 lys2BglII trp1Δ63 ade8 rad27-G67S
EAY608	MATaleu2Δ1 his3Δ200 ura3-52 lys2BglII, trp1Δ63 ade8 rad27-G67S msh2Δ::hisG
EAY611	MATaleu2Δ1 ura3-52 lys2BglII trp1Δ63 rad27-G240D
EAY613	MATaleu2Δ1 ura3-52 lys2BglII trp1Δ63 rad27-G240D msh2Δ::hisG
EAY615	MATα leu2Δ1 ura3-52 lys2BglII trp1Δ63 his3Δ200 rad27Δ::HIS3
EAY616	MATaleu2Δ1 ura3-52 lys2BglII trp1Δ63 his3Δ200 rad27Δ::HIS3 msh2Δ::hisG
EAY618	MATα leu2Δ1 his3Δ200 ura3-52 exo1Δ::HIS3
EAY752	MATα ura3-52 leu2Δ1 his3Δ200 RAD27::LEU2
EAY753	MATα ura3-52 leu2Δ1 his3Δ200 rad27-G67S::LEU2
EAY754	MATα ura3-52 leu2Δ1 his3Δ200 rad27-G240D::LEU2
EAY755	MATα ura3-52 leu2Δ1 trp1Δ63 RAD27::LEU2
EAY756	MATaura3-52 leu2Δ1 trp1Δ63 his3Δ200 rad27-G67S::LEU2
EAY757	MATaura3-52 leu2Δ1 trp1Δ63 his3Δ200 rad27-G240D::LEU2

Plasmids.

pEAA67 (MLH1 PMS1 MSH2 ARSH4 CEN6 URA3) and pK5 ([GT]₁₄G-LACZ 2μm LEU2; kindly provided by Richard Kolodner) were introduced into wild-type strain FY86 prior to mutagenesis (see below) to avoid identifying mutators resulting from defects in the previously characterized MLH1, MSH2, and PMS1 genes. The MSH2, MLH1, and PMS1 genes in pEAA67 are each expressed under their native promoters. pK5 contains an out-of-frame GT repeat sequence inserted into the LACZ open reading frame. Previous studies indicated that nearly 100% of mismatch repair-defective colonies containing pK5 display a blue colony phenotype due to DNA slippage events in the GT repeat tract; in contrast, less than 0.5% of wild-type colonies show a blue colony phenotype (44).

pR2.14 (RAD27 2μm URA3) and pRDK480 (EXO1 2μm LEU2) were kindly provided by Satya Prakash and Richard Kolodner, respectively. pEAI137 is a pUC19-based plasmid that contains the pR2.14-derived RAD27 gene on a 2.3-kb EcoRI-FspI fragment. Overlapping PCR mutagenesis (17) was performed using pEAI137 as a template to introduce the rad27-G67S (pEAI139) and rad27-G240D (pEAI140) mutations. pEAI141, pEAI143, and pEAI144 are derivatives of pEAI137, pEAI139, and pEAI140, respectively. These plasmids each contain a 2.2-kb LEU2 fragment inserted 250 bp upstream of the RAD27 start codon. The RAD27::LEU2, rad27-G67S::LEU2, and rad27-G240D::LEU2 markers in pEAI141, -143, and -144, respectively, were introduced into the FY strains by digesting these plasmids with BglII prior to transformation. The resulting strains, which were confirmed by DNA sequencing, displayed the same phenotype as RAD27 and rad27 strains that lacked the LEU2 insertion. pSH44 ([GT]₁₆T-URA3 ARSH4 CEN6 TRP1), kindly provided by Tom Petes, was introduced into wild-type and rad27 strains to assess repeat tract instability (15, 43) (see Table 5).

TABLE 5.

Frequency and distribution of dinucleotide repeat tract instability events in wild-type, rad27, and msh2 strains^a

Strain genotype	Instability		Tracts sequenced	No. of tracts with indicated bp change
	Avg (individual expt) (10−5)	Relative		+2	−2	+4	−14	+14	Other
Wild type	2.5 (3.0, 2.3, 2.0)	1.0	11	9	1	1	0	0	0
msh2Δ::hisG	660 (880, 600, 500)	264	28	7	21	0	0	0	0
rad27Δ::HIS3	4,600 (1,000, 7,000, 5,800, 4,600)	1,800	35	29	0	5	1	0	0
rad27-G67S	290 (390, 340, 210, 220)	116	19	16	0	3	0	0	0
rad27-G240D	58 (74, 58, 43)	23	20	17	0	0	0	1	2b

^aWild-type (FY23), msh2Δ::hisG (EAY281), rad27Δ::HIS3 (EAY615), rad27-G67S (EAY607), and rad27-G240D (EAY611) strains were tested in the 5-FOA resistance assay as described in Materials and Methods. In each experiment 11 independent colonies were tested. The average median frequency in each assay relative to the wild-type frequency is presented. The distribution in base pairs of insertions and deletions in the repeat tract is shown. DNA sequencing was performed on plasmids derived from the following 5-FOA-resistant strains containing pSH44 or a derivative plasmid: wild-type (FY23) (57), msh2Δ (24), rad27Δ (23), rad27-G67S (EAY756 [this study]), and rad27-G240D (EAY757 [this study]) (24) strains. 

^bFor both events, a 19-bp duplication ([CA]₁₆ TGTCGACGATCCCCTGGCAAAACGACGATCCCCTGGCAAAACGACGATCTTCTTAG) occurred in the sequence immediately following the GT repeat. The duplicated sequence is in boldface, and the microhomology flanking the duplication is underlined. 

Rad27p substrates.

Oligonucleotides were designed to form nicked or flap substrates. For the nicked substrate, two primers were annealed to a template such that the upstream and downstream primers formed a nick. Flap substrates were generated by including sequences not complementary to the template at the 5′ end of the downstream primer. These sequences form the unannealed 5′ tail or flap. Upstream primers were annealed to the template such that they formed a nick at the base of the flap. Oligomer sequences are listed in Table 2.

TABLE 2.

Oligonucleotides used in this study to create Rad27p substrates

Primer	Size	Sequencea
Nicked and flap substrate primers
Dnick	18-mer	5′-GTAAAACGACGGCCAGTG
D6nt	24-mer	5′-TTCCAAGTAAAACGACGGCCAGTG
D15nt	33-mer	5′-TAGAGCTGTTTCCAAGTAAAACGACGGCCAGTG
T44	44-mer	3′-GCGGTCCCAAAAGGGTCAGTGCTGGCATTTTGCTGCCGGTCACG
U26	26-mer	5′-CGCCAGGGTTTTCCCAGTCACGACCA
U25	25-mer	5′-CGCCAGGGTTTTCCCAGTCACGACC
U24	24-mer	5′-CGCCAGGGTTTTCCCAGTCACGAC
Bubble substrate primers
Dbubble	80-mer	5′-GACTCTCGACTCACGTAGAGCTGTTTTTTTTTTTTTTTTTTTTTTTTTTTTCCAAGTAAAACGACGGCCAGTGCTACGAG
Tbubble	66-mer	3′-CTGAGAGCTGCTGAGAGCTGAGTGCATCTCGACAA............GCTGG.............CATTTTGCTGCCGGTCACGATGCTCG
Additional primer
T1	22-mer	5′-CATTTTGCTGCCGGTCACGTCA

^aAnnealed residues are underlined, and unannealed residues are in boldface. Thick and thin underlines indicate upstream and downstream regions, respectively. 

Prior to annealing, downstream primers were radiolabeled at either the 5′ or 3′ end. Downstream primers (10 pmol) were 5′ end radiolabeled with [γ-³²P]ATP (New England Nuclear) by T4 polynucleotide kinase in accordance with the manufacturer's instructions. For 3′ end-radiolabeled primers, downstream primers (10 pmol) were annealed to template T₁ (25 pmol) for nicked and flap substrates or T_bubble for the bubble substrate shown in Fig. 7. This results in a 5′ template overhang. The downstream primers were extended with [α-³²P]dCTP (New England Nuclear) by Klenow polymerase (Roche Molecular Biochemicals) at 37°C for 2 h. After removal of unincorporated radionucleotides by a Micro Bio-Spin 30 chromatography column (Bio-Rad), all radiolabeled primers were purified by gel isolation from a 12% polyacrylamide–7 M urea denaturing gel.

FIG. 7.

Cleavage of intermediates of repeat expansion by rad27-G67Sp and rad27-G240Dp. Increasing amounts of either wild-type or mutant Rad27p (10, 20, 40, 60, 80, and 100 fmol) were incubated with a 3′ radiolabeled bubble substrate containing primers D_bubble and T_bubble. Each primer contained both 5′ and 3′ regions of complementarity resulting in the formation of a bubble. Reaction mixtures contained Rad27p (lanes 2 to 7), rad27-G67Sp (G67S; lanes 9 to 14), or rad27-G240Dp (G240D; lanes 16 to 21). Reaction mixtures in lanes 1, 8, and 15 contained only substrate. Schematic diagrams of the substrates are shown. The lengths of substrates and products are in nucleotides. Asterisk, position of radiolabeled nucleotide.

Substrates were generated by annealing a downstream primer, template, and upstream primer at a molar ratio of 1:5:20, respectively. The high molar ratio of primers ensures complete formation of the final substrate. A downstream primer and template were placed in 50 μl of Tris-EDTA with 50 mM KCl and 1 mM dithiothreitol and heated to 100°C for 5 min. The reaction mixture was placed at 70°C and allowed to slowly cool to 25°C. After an upstream primer was added, the mixture was incubated at 37°C for 30 min. The nicked substrate contains downstream primer D_nick and upstream primer U₂₅ annealed to template T₄₄. Flaps of 6 or 15 nucleotides contain downstream primer D₆ or D₁₅, respectively, and upstream primer U₂₅ annealed to template T₄₄. The gap substrate was formed by annealing primer U₂₄ in place of U₂₅, which results in a one-nucleotide gap between the upstream and downstream primers. The double-flap substrate contained downstream primers D₆ and U₂₆ annealed to template T₄₄. Primers D_bubble and T_bubble were annealed to form the bubble substrate. All primers were synthesized by Integrated DNA Technologies (Coralville, Iowa).

Genetic techniques.

Yeast was transformed with DNA using the lithium acetate method as described by Geitz and Schiestl (8). Tetrads were dissected on YPD plates immediately after Zymolyase treatment using previously established methods (39). All tetrads that yielded four, three, and sometimes two and one viable spores were examined for relevant genetic markers by PCR or by segregation of a linked marker (e.g., rad27Δ::HIS3 rad52Δ::URA3 exo1Δ::HIS3).

To isolate the rad27 alleles described in this paper, several independent mid-log cultures of FY86 containing pK5 and pEAA67 were mutagenized with UV light to 20% viability. UV-mutagenized cells, grown on leucine-uracil-threonine minimal-dropout plates containing 2% glucose, were replica plated to corresponding X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) plates containing 2% galactose and 2% sucrose (39). Blue colonies were streaked to single colonies, and 11 colonies from each candidate were patched to minimal plates containing canavanine. Blue colonies that also showed a higher median frequency of resistance to canavanine were backcrossed to wild type three times and then retested in both the X-Gal–DNA slippage and canavanine mutator assays. Three candidates from 220,000 UV-mutagenized cells that displayed a consistent phenotype in both assays were identified. The defects in all three strains could be corrected by transformation with the pR2.14 plasmid. For each strain, DNA spanning the RAD27 gene was amplified by PCR from chromosomal DNA and then sequenced to reveal the rad27-G67S, rad27-G240D, and rad27-324 mutations. The rad27-324 allele, which contains a frameshift mutation in amino acid 324 of the RAD27 open reading frame, produced phenotypes similar to those produced by a null allele. The MMS, DNA slippage, and mutator phenotypes observed in the original rad27-G67S and rad27-G240D strains were also observed in strains in which the rad27-G67S and rad27-G240D mutations had been introduced by gene replacement (see above).

Mutation frequencies in the yeast strains shown in Table 3 were determined by measuring the frequency of forward mutation to canavanine resistance (38). Repeat tract instability frequencies (see Table 5) were determined by measuring frameshift events that resulted in resistance to 5-FOA in strains containing pSH44 (15). In both the mutator and repeat tract instability studies, tested strains were streaked to form single colonies on selective minimal plates. Eleven independent colonies were suspended in water, and appropriate dilutions were then plated onto minimal media with or without canavanine or 5-FOA. The median frequencies of canavanine and 5-FOA resistance were determined for each experiment, and the averages of at least three independent experiments are presented for each strain. The lengths of GT repeat tracts were determined by sequencing pSH44-derived plasmids recovered from independently isolated 5-FOA-resistant colonies. Plasmids were sequenced using the −40 primer described by Henderson and Petes (15). To examine large insertion-deletion alterations in the CAN1 gene, the complete open reading frame of the CAN1 gene was amplified by PCR from chromosomal DNA isolated from independently identified Can^r colonies. Amplified DNA was digested with HphI and then separated by electrophoresis on a 2% Tris-acetate-EDTA–agarose gel. The genetic data presented in Tables 3 to 5 were analyzed using the Mann-Whitney test statistic, where P values of <0.05 are considered significant (34).

TABLE 3.

Median frequencies of forward mutations in wild type, rad27, and msh2Δ strains^a

Relevant genotype	Avg forward mutation frequency to Canr (10−7) (individual expt results)	Relative frequency
Wild type	4.3 (5.6, 3.4, 2.4, 7.2, 2.8)	1.0
msh2Δ::hisG	320 (360, 310, 300)	74
rad27Δ::HIS3	667 (400, 200, 660, 780, 1,100, 860)	155
rad27-G67S	93 (90, 45, 130, 120, 110, 70, 110, 70)	22
rad27-G240D	263 (500, 140, 170, 180, 330, 270, 250, 260)	61
rad27Δ::HIS3 msh2Δ::hisG	2,300 (1,700, 2,000, 3,100)	535
rad27-G67S msh2Δ::hisG	950 (1,000, 1,200, 790, 820)	221
rad27-G240D, msh2Δ::hisG	1,200 (1,200, 2,200, 250)	279
Wild type + pEXO1 2μm	5.6 (3.6, 4.2, 9.0)	1.3
rad27Δ::HIS3 + pEXO1 2μm	68 (42, 34, 140, 56)	16
rad27-G67S + pEXO1 2μm	10 (6.6, 9.0, 15, 11)	2.3
rad27-G240D + pEXO1 2μm	60 (90, 64, 26)	14

^aWild-type (FY86), msh2Δ (EAY281), rad27Δ::HIS3 (EAY615 or EAY545), rad27-G67S (EAY607 or EAY595), rad27-G240D (EAY611 or EAY595), rad27Δ::HIS3 msh2Δ::hisG (EAY616), rad27-G67S msh2Δ::hisG (EAY608), and rad27-G240D msh2Δ::hisG (EAY613) strains were tested in the Can^r assay as described in Materials and Methods. In each experiment, 11 independent colonies were tested. The average median frequency in each assay is presented relative to the wild-type frequency. Strains were transformed with a 2μm plasmid containing EXO1 (pRDK480) as indicated. 

Nucleic acid techniques.

All restriction endonucleases were purchased from New England Biolabs (Beverly, Mass.) and used according to the manufacturer's specifications. Taq DNA polymerase was purchased from Perkin-Elmer Cetus. Yeast chromosomal DNA was prepared as described by Holm et al. (18). PCR was performed as described previously (40), and amplification conditions and primer sequences for the different reactions are available upon request. The DNA primer synthesis and DNA sequencing was performed at the Cornell Biotechnology Analytical/Synthesis facility.

Expression and purification of wild-type and mutant Rad27 proteins.

The expression and purification of wild-type and mutant Rad27p will be described in detail elsewhere. In brief, Rad27, rad27-G67S, and rad27-G240D proteins were expressed in Escherichia coli using T7 expression vector pET-24b. After transformation into E. coli strain BL21(DE3) codon plus (Stratagene, Inc.), cultures were grown until the optical density at 600 nm reached ∼0.5. The addition of IPTG (isopropyl-β-d-thiogalactopyranoside) results in the expression of wild-type or mutant Rad27p with the addition of a six-His tag at the C terminus. After lysis by a French press, proteins were separated in successive chromatographic columns containing Ni⁺-agarose (Qiagen, Inc.), carboxymethyl-Sepharose, Mono-S, hydroxyapatite, and phenyl-Sepharose resins to obtain highly purified (>95%) enzyme fractions.

Nuclease assays.

Amounts of substrate indicated in the figure legends and wild-type or mutant Rad27p were incubated in reaction buffer (30 mM HEPES [pH 7.6] diluted from a 1 M stock, 40 mM KCl, 8 mM MgCl_2, 0.01% NP-40, and 0.1 mg of bovine serum albumin/ml) in a final volume of 20 μl. Assay mixtures were incubated at 30°C for 15 min, and reactions were stopped by the addition of 10 μl of termination dye (95% formamide [vol/vol] with bromophenol blue and xylene cyanole). After a 95°C incubation for 5 min, samples were separated on a 12 or 18% polyacrylamide–7 M urea denaturing gel. The gels were quantified using a PhosphorImager (Molecular Dynamics) and analyzed using Imagequant, version 1.2, software from Molecular Dynamics. In all studies, the amounts of substrate and product(s) were quantitated and the percentages of product formed were determined by the product/(substrate plus product) ratio. This method allowed for the correction of any loading errors between lanes. All assays were performed at least in triplicate, and representative assays are shown.

RESULTS

Isolation of rad27 alleles.

The rad27-G67S and rad27-G240D mutations were acquired in a two-part screen to identify mutants that displayed both mutator and DNA slippage phenotypes (see Materials and Methods) (57). The rad27 mutations map to the N (G67) and I (G240) nuclease domains. These domains are highly conserved in eukaryotic, prokaryotic, and viral 5′ nucleases, which include the Fen1p family of DNA flap endonucleases, XPG endonuclease, T4 phage RNase H, and E. coli DNA polymerase I (7, 41). The glycine 67 residue is conserved only in the DNA flap endonuclease family, while the glycine 240 residue is conserved among all family members (41). Other mutations in these nuclease domains, none of which correspond to the G67S and G240D mutations described here, affect Rad27p nuclease activity on DNA flap substrates but do not necessarily disrupt DNA binding (7, 41).

rad27Δ strains display a temperature-sensitive growth phenotype at 37°C and are sensitive to the alkylating agent MMS (48). Unlike the rad27Δ strains, the rad27-G67S and rad27-G240D strains displayed wild-type colony sizes, were highly resistant to MMS, and did not show a temperature-sensitive phenotype (Fig. 1; data not shown). Previous studies had shown that rad27Δ rad52Δ and rad27Δ exo1Δ strains were inviable (35, 45). Together, this information suggests that the defects observed in rad27-G67S and rad27-G240D strains are subtle and might be altered or magnified in an informative manner in exo1Δ or rad52Δ strain backgrounds. Haploid strains containing the rad27Δ , rad27-G67S, and rad27-G240D alleles were mated to rad52Δ and exo1Δ strains, and tetrads from the resulting diploids were examined for spore viability, segregation of markers, and mutator phenotypes. Spore clones containing both rad27 and exo1Δ or rad52Δ alleles were classified as viable based on the genotyping of four-spore viable tetrads. Double-mutant combinations were classified as inviable based on genotyping and the detection of inviable spore segregation patterns, consistent with two genes segregating independently (parental ditype, tetratype, and nonparental ditype in the proportion of 1:4:1). In cases of synthetic lethality, no spore clones that contained both mutations were identified. This analysis confirmed that rad27Δ rad52Δ and rad27Δ exo1Δ strains were inviable and showed that the rad27-G67S rad52Δ and rad27-G240D rad52Δ double mutants were also inviable. The rad27-G67S exo1Δ and rad27-G240D exo1Δ strains, however, were viable and displayed colony sizes that were indistinguishable from those for the wild type (data not shown).

FIG. 1.

Resistance of rad27 strains to MMS. Saturated cultures of wild-type (FY86), rad27Δ (EAY545), rad27-G67S (EAY595), and rad27-G240D (EAY597) strains were diluted in water and spotted in 10-fold serial dilutions (10⁻¹ to 10⁻⁵) onto YPD media containing 0 to 0.020% MMS. The plates were photographed after a 3-day incubation at 30°C.

rad27-G67S and -G240D mutations confer distinct phenotypes in forward mutation and DNA slippage assays.

rad27Δ strains display a strong mutator phenotype that was assessed by measuring resistance to toxic arginine analog canavanine. In strains bearing mismatch repair mutations such as the msh2Δ strain, canavanine resistance primarily results from forward mutations in CAN1 which include nucleotide misincorporation and single-nucleotide-deletion events (30). In rad27Δ strains, these forward mutations primarily result from duplication events that occur between short repeated sequences in the CAN1 gene (48). Compared to the wild type, rad27Δ strains showed a 150-fold increase in canavanine resistance frequency, a value that is twofold higher than those observed in msh2Δ strains. The rad27-G67S and rad27-G240D strains displayed canavanine resistance frequencies that were 20- and 61-fold higher, respectively, than that for the wild type. The difference in frequency between the rad27-G67S and rad27-G240D strains was statistically significant (P = 0.0008). A previous double-mutant analysis indicated a nonepistatic relationship between rad27Δ and mismatch repair mutations (48). As shown in Table 3, the frequency of canavanine resistance in rad27Δ msh2Δ strains (535-fold increase) was greater than the frequencies of the single mutant strains (msh2Δ, 74-fold increase; rad27Δ, 155-fold increase). A similar observation was made in the analysis of msh2Δ rad27-G67S and -G240D double mutants.

Wild-type, rad27Δ, rad27-G67S, and rad27-G240D strains were examined for the presence of insertion-deletion mutations in CAN1 (Table 4; Fig. 2) (48). In the wild type, none of the Can^r colonies (0 of 17) displayed an insertion-deletion event at CAN1; in rad27Δ strains, an increase in the size of a single CAN1-derived fragment was observed for 73% (11 of 15) of the Can^r colonies. While the proportion and distribution of rearrangements in CAN1 were similar to those observed in rad27-G240D (71%; 15 of 21), a lower proportion (41%; 18 of 44) but a similar distribution were observed in rad27-G67S.

TABLE 4.

rad27-G240D and rad27-G67S strains display chromosomal rearrangement phenotypes^a

Relevant genotypes	% Insertion-deletion (no. analyzed)	No. of insertions (fragment size)	No. of deletions (fragment size)
Wild type	0b (17)	0	0
rad27Δ	73 (15)	3 (490), 2 (411), 2 (252 or 249),  4 (207)	0
rad27-G67S	41 (44)	9 (490), 1 (411), 3 (252 or 249),  4 (207)	1 (411)
rad27-G240D	71 (21)	7 (490), 4 (411), 1 (252 or 249),  3 (207)	0

^aIndependent Can^r colonies were obtained from wild-type (FY86), rad27Δ (EAY545), rad27-G67S (EAY595), and rad27-G240D (EAY597) strains. The CAN1 open reading frame was amplified by PCR from each of these resistant colonies and digested with HphI, resulting in 490-, 411-, 314-, 252-, 249-, and 207-bp fragments that could be detected by gel electrophoresis (55). 

^bOne Can^r colony displayed a mutation in CAN1 that eliminated an HphI restriction site, and the other displayed a mutation that created an HphI site. 

FIG. 2.

rad27-G240D and rad27-G67S strains display different insertion-deletion phenotypes at the CAN1 locus. Independent Can^r colonies were obtained from EAY595 (rad27-G67S), EAY597 (rad27-G240D), EAY545 (rad27Δ), and FY86 (wild-type) strains, and a DNA fragment containing the CAN1 open reading frame was amplified by PCR from each of these resistant colonies and digested with HphI, resulting in 490-, 411-, 314-, 252-, 249-, and 207-bp fragments, which could be detected by gel electrophoresis (57). Two smaller bands of 87 and 46 bp could not be detected. Lanes A and B, DNA fragments from the wild-type CAN1 locus in FY86; lanes 1 to 12 and 13 to 25, HphI-digested CAN1 DNA from Can^r rad27-G67S and rad27-G240D, respectively. Asterisks, lanes containing bands that differ from the wild-type CAN1 pattern.

Studies by Tishkoff et al. (47) showed that overexpression of EXO1 on 2μm plasmids resulted in suppression of the rad27Δ temperature-sensitive growth phenotype and partial suppression of the forward mutation phenotype. As shown in Table 3, rad27Δ (155 to 16; P = 0.01) and rad27-G240D (61 to 14; P = 0.01) strains overexpressing EXO1 each displayed a mutation frequency that was only about 15-fold higher than that for the wild type. The rad27-G67S strains overexpressing EXO1 (22 to 2.3; P = 0.007) displayed a frequency that was only twofold higher.

The modest mutator phenotype and lower proportion of insertion-deletion events observed for rad27-G67S strains in the canavanine resistance assay were surprising, because both the rad27-G67S and rad27-G240D mutations conferred inviability in the presence of the rad52Δ mutation. One way to account for this difference is that rad27-G67S strains displayed a frameshifting phenotype that could not be efficiently detected using the canavanine resistance assay. This was tested by examining the frequency of frameshifting events that occur in a polynucleotide tract present within the open reading frame of URA3 reporter construct pSH44 (15). As shown in Table 5, rad27-G240D, rad27-G67S, and rad27Δ strains showed 23-, 116-, and 1,800-fold increases, respectively, in the frequency of frameshift events that were detected as resistance to 5-FOA. The fivefold-higher level of frameshift events in rad27-G67S strains compared to that in rad27-G240D strains was statistically significant (P = 0.03).

pSH44 plasmids obtained from independent wild-type, rad27, and msh2 5-FOA-resistant colonies were sequenced to examine repeat tract sequence changes (Table 5). As shown previously, the vast majority of frameshift events in rad27Δ strains resulted from the insertion of a single repeat unit (29 of 35) (23). The vast majority of tract alterations in rad27-G67S (16 of 19) and rad27-G240D (17 of 20) strains were also single-repeat (+2-bp) mutations. In rad27-G67S, the remaining three alterations were +4-bp (two-repeat) mutations. In rad27-G240D, one of the three remaining alterations was a +14-bp (seven-repeat) insertion and the other two were duplications of the DNA sequence spanning the junction of the dinucleotide repeat and URA3 sequences. The spectra of repeat tract insertion-deletion events in the rad27 strains differed from that observed in msh2Δ strains, where only single-repeat insertions- deletions were observed, with the majority of these events consisting of deletions (24).

Biochemical analysis of Rad27 mutant proteins.

To examine the biochemical properties of rad27-G67S and rad27-G240D proteins, we expressed Rad27p and the two mutant proteins in E. coli. Rad27p and rad27-G67Sp were expressed and purified similarly, with final preparations at more than 95% homogeneity (data not shown). Expression of rad27-G240Dp resulted in a number of truncation products whose presence was not reduced by the addition of protease inhibitors. Through additional chromatographic steps it was found that the final protein fraction contained full-length rad27-G240Dp, active for catalysis, and one truncation product representing ∼20% of the preparation that did not contain any detectable activity (data not shown). Using these recombinant enzymes, we examined the activity of each mutant on a set of standard nicked and flap substrates to determine the effect of the point mutations (16).

The ability of the Rad27p to cleave a 5′-end flap substrate at or near the base of the flap is the hallmark of this family of nucleases (1). Therefore, we tested rad27-G67Sp and rad27-G240Dp for any alteration in their ability to cleave a flap substrate with a six-nucleotide unannealed 5′ tail. In gel mobility shift assays, Rad27p and the two mutant proteins bound to a six-nucleotide flap substrate with nearly identical affinities (Fig. 3). This suggests that the DNA binding properties of the mutant proteins had not been significantly altered. Incubation of the 5′-end-labeled substrate with increasing amounts of Rad27p resulted in the formation of products five to seven nucleotides in length corresponding to cleavage at the base of the flap (Fig. 4, lanes 2 to 7). Similar products also were observed following incubation with rad27-G67Sp (lanes 9 to 14) and rad27-G240Dp (lanes 16 to 21), indicating that the specificity of the two mutants for this substrate is unchanged. However, we consistently observed that the level of cleavage of each was reduced compared to that for Rad27p.

FIG. 3.

Ability of rad27-G67Sp and rad27-G240Dp to bind to a flap substrate. Wild-type and mutant Rad27p was incubated with a substrate containing a six-nucleotide flap radiolabeled at the 3′ end of the downstream primer. Substrate (5 fmol) containing primers D_6nt,T₄₄, and U₂₅ was incubated with increasing amounts of enzyme, as indicated, in reaction buffer (see Materials and Methods) without MgCl₂ at 25°C for 8 min in a total volume of 20 μl. After the addition of 2 μl of 50% glycerol, DNA and DNA-enzyme complexes were separated by electrophoresis on a 1% agarose–0.5% polyacrylamide gel in 0.25× Tris-borate-EDTA at 4°C. The gel was dried onto DE81 Whatman filter paper, and products were detected using a PhosphorImager. Reactions mixtures contained either Rad27p, rad27-G67Sp (G67S), or rad27-G240Dp (G240D), as noted. Lanes “boiled,” substrate incubated at 100°C prior to loading to indicate the location of unannealed substrate. A schematic diagram of the substrate is at the top. The lengths of substrates and products are in nucleotides. Asterisk, position of the radiolabeled nucleotide.

FIG. 4.

Endonucleolytic cleavage of flap substrates by rad27-G67Sp and rad27-G240Dp. (A) Wild-type or mutant Rad27p was incubated with a substrate containing a six-nucleotide flap. Substrate (5 fmol) containing primers D_6nt, T₄₄, and U₂₅ was incubated with increasing amounts of enzyme (10, 25, 50, 100, 500, and 1,000 fmol) at 30°C as described in Materials and Methods and separated by electrophoresis on a denaturing 12% polyacrylamide gel. Reaction mixtures contained either Rad27p (lanes 2 to 7), rad27-G67Sp (G67S; lanes 9 to 14), or rad27-G240Dp (G240D; lanes 16 to 21). Reaction mixtures in lanes 1, 8, and 15 contained only substrate. A schematic diagram of the substrate is at the top. The lengths of substrates and products are in nucleotides. Asterisk, position of radiolabeled nucleotide. (B) Increasing amounts of Rad27p (solid lines), rad27-G67Sp (G67S; dotted lines), and rad27-G240Dp (G240D; dashed lines) were incubated with substrates containing a nick (no flap) (circle) or a 6-nucleotide (square) or 15-nucleotide (triangle) flap as described for panel A. Results of the experiments with the six-nucleotide flap substrate are graphical representations of the gels presented in panel A. The products were detected using a PhosphorImager, quantitated, and presented as the percentages of substrate converted to product versus amounts of enzyme. The nicked substrate contains primers D_nick, T₄₄, and U₂₅, and the 15- nucleotide flap contains primers D_15nt, T₄₄, and U₂₅.

To determine whether the reduced cleavage activity exhibited by the mutant proteins reflects an altered preference for the length of the flap, we tested their activity on substrates containing a nick (no flap) or a longer 15-nucleotide flap. Incubation of these substrates with Rad27p results in the release of either a single nucleotide for the nicked substrate or a 15- to 16-nucleotide product for the longer flap. The data for each of these substrates and the six-nucleotide flap in Fig. 4A were quantified with respect to both the starting substrate and products and were expressed as the percentages of substrate converted to product (products/[substrate + products] × 100) versus the amount of enzyme (Fig. 4B). On all substrates, the amounts of cleavage by rad27-G67Sp (30 to 60%) and rad27-G240Dp (∼5%) were substantially reduced compared to the amount of cleavage by Rad27p (90%). Similar results were obtained when the experiments were done at 37°C (data not shown). The reduction in cleavage by the mutants might have reflected a change in the stability of these enzymes over time at the reaction temperature. Therefore, we did a similar experiment but preincubated the enzymes at 30°C for 15 min prior the addition of substrate. The levels of cleavage observed were unchanged for all proteins (data not shown).

In addition to its endonuclease activity at the base of flap structures, Rad27p displays a 5′ exonuclease activity. This activity is thought to be required to remove single ribonucleotides that remain at the 5′ ends of Okazaki fragments following cleavage of the RNA primer by RNase H (51). Exonucleolytic activity can be measured by examining the ability of Rad27p to continue degrading the annealed portion of the downstream primer following endonucleolytic removal of the flap (20). Wild-type and mutant Rad27p exonuclease activities were examined on six-nucleotide flap, nicked, and gapped substrates (Fig. 5). In these assays, the DNA substrate was 3′ end labeled and incubated with increasing amounts of wild-type or mutant protein.

FIG. 5.

Endo- and exonucleolytic cleavage of 3′-end labeled substrates by rad27-G67Sp and rad27-G240Dp. (A) Wild-type or mutant Rad27p was incubated with a substrate containing a six-nucleotide flap radiolabeled at the 3′ end of the downstream primer. Substrate (5 fmol) containing primers D_6nt, T₄₄, and U₂₅ was incubated with increasing amounts of enzyme (10, 25, 50, 100, 500, and 1,000 fmol) at 30°C as described in Materials and Methods and separated by electrophoresis on a denaturing 18% polyacrylamide gel. Reaction mixtures contained either Rad27p (lanes 2 to 7), rad27-G67Sp (G67S; lanes 9 to 14), or rad27-G240Dp (G240D; lanes 16 to 21). Reaction mixtures in lanes 1, 8, and 15 contained only substrate. A schematic diagram of the substrate is at the top. The lengths of substrates and products are in nucleotides. Asterisk, position of radiolabeled nucleotide. (B) Rad27p or rad27-G67Sp was incubated with a substrate containing a nick (lanes 1 to 14) or one-nucleotide gap (lanes 15 to 28) radiolabeled at the 3′ end of the downstream primer. Substrate (5 fmol) was incubated with increasing amounts of enzyme (10, 20, 40, 60, 80, and 100 fmol) at 30°C as described in Materials and Methods and separated by electrophoresis on a denaturing 18% polyacrylamide gel. Reaction mixtures contained either Rad27p (lanes 2 to 7 and 16 to 21) or rad27-G67Sp (G67S; lanes 9 to 14 and 22 to 28). Reaction mixtures in lanes 1, 8, 15, and 22 contained only substrate. Schematic diagrams of the substrates are shown. The nicked substrate contains primers D_nick, T₄₄, and U₂₅, and the one-nucleotide gap substrate contains primers D_nick, T₄₄, and U₂₄.

Incubation of the six-nucleotide flap substrate with 10 fmol of Rad27p resulted primarily in endonucleolytic cleavage of the substrate at the base of the flap (Fig. 5A, lane 1). At higher Rad27p concentrations, smaller reaction products, which represent progressive exonuclease cleavage within the annealed region of the downstream primer, were observed (Fig. 5A, lanes 2 to 7). The cleavage activity of both rad27-G67Sp (lane 9 to 14) and rad27-G240Dp (lanes 16 to 21) was greatly reduced relative to that of the wild type, with the latter mutant protein showing a particularly strong defect. Both mutant proteins were virtually unable to cleave within the annealed portion of the downstream primer. This characteristic was particularly striking with rad27-G67Sp, which displayed substantial endonucleolytic activity. One explanation is that, because of the overall reduction of cleavage, cuts subsequent to flap removal are simply not detected. We performed similar assays with greatly increased levels (10 pmol) of rad27-G67Sp but were unable to detect further degradation of the substrate (data not shown).

For both the nicked and gapped substrates, Rad27p removed the 5′ nucleotide and then continued to cleave within the annealed region of the downstream primer (Fig. 5B, lanes 2 to 7 and 16 to 21). The cleavage of the nicked substrate was more efficient. In contrast, rad27-G67Sp showed reduced cleavage of the first nucleotide of the nicked substrate and essentially no further cleavage (lanes 9 to 14). The mutant enzyme was virtually inert on the gapped substrate (lanes 22 to 28). The poor cleavage of the 5′ nucleotide of the 3′-labeled nick substrate by rad27-G67Sp is consistent with the results presented in Fig. 4B using a 5′-labeled substrate. Failure to cleave either the nicked substrate after creation of a one-nucleotide gap or the initially gapped substrate indicates that rad27-G67Sp exonuclease activity requires an immediately adjacent upstream primer. We failed to detect any activity on these substrates using rad27-G240Dp (data not shown).

Activity of wild-type and mutant Rad27 proteins on double-flap and bubble substrates.

Double-flap structures have been hypothesized to reflect an intermediate formed in vivo in end-joining and homologous-recombination pathways (14). These structures might also be an intermediate in nick translation reactions widely used in DNA replication and repair. Several homologs of Rad27p from Eucarya and Archaea were shown to display an enhanced endonuclease activity on substrates containing a double-flap structure. This enhanced activity was identified by altering the length of the upstream primer (14, 31, 49). For the experiments presented in Fig. 6, the upstream primer for the six-nucleotide flap substrate (Fig. 4A and 5A) was extended one nucleotide to form a 3′ flap adjacent to the 5′ flap formed by the downstream primer. We tested the activity of each mutant on such a substrate (Fig. 6). Both the wild type (lanes 2 to 6) and rad27-G67Sp (lanes 7 to 11) easily cleaved this substrate. In fact, rad27-G67Sp cleaved the double flap (87%) more efficiently than the conventional flap, almost reaching the cleavage level of Rad27p (95%). Although rad27-G240Dp effectively failed to cleave standard flap and nick substrates, the double-flap structure was cleaved (60%) (lanes 12 to 15). However, after endonucleolytic cleavage, neither rad27-G67Sp nor rad27-G240Dp was able to carry out subsequent exonucleolytic cleavage (data not shown).

FIG. 6.

Cleavage of a double-flap substrate by rad27-G67Sp and rad27-G240Dp. Wild-type or mutant Rad27p was incubated with a substrate containing a six-nucleotide 5′ flap and a one-nucleotide 3′ flap radiolabeled at the 5′ end of the downstream primer. Substrate (5 fmol) containing primers D_6nt, T₄₄, and U₂₆ was incubated with increasing amounts of enzyme (10, 50, 100, 500, and 1,000 fmol) at 30°C as described in Materials and Methods and separated by electrophoresis on a denaturing 12% polyacrylamide gel. Reaction mixtures contained either Rad27p (lanes 2 to 6), rad27-G67Sp (G67S; lanes 7 to 11), or rad27-G240Dp (G240D; lanes 12 to 16). Lane 1 contained only substrate. A schematic diagram of the substrate is shown. The lengths of substrates and products are in nucleotides. Asterisk, position of radiolabeled nucleotide.

A current model describing the role of Rad27p in expansion of repeat sequences involves the formation of intermediate structures that promote expansion (10, 16, 48). These include structures in which the 5′-end region of the flap anneals to the template (bubbles) or to itself (foldbacks). Both structures inhibit endonucleolytic flap cleavage. The loss of exonucleolytic activity of rad27-G67Sp and rad27-G240Dp could favor persistence of these structures in vivo. To examine this possibility, we constructed a bubble substrate and compared the abilities of the wild-type and mutant nucleases to effect endonucleolytic cleavage (Fig. 7). We previously demonstrated that progressive exonucleolytic cleavage of the 5′ annealed region of the flap precedes endonucleolytic cleavage at the conventional flap cleavage site (16). Presumably the 5′-end region of the flap must be degraded sufficiently to melt away from the template to allow proper tracking of the nuclease to the endonucleolytic cleavage site. Incubation of the bubble substrate with Rad27p resulted in the expected pattern, with degradation of the annealed region occurring first (Fig. 7, lanes 2 to 7). Neither rad27-G67Sp nor rad27-G240Dp was able to resolve this structure (lanes 9 to 14 and 16 to 21).

Finally, a substrate with a flap foldback having 18 complementary nucleotides was tested. This substrate was cleaved more slowly by Rad27p than a conventional flap substrate, and no cleavage activity was observed for either mutant protein (data not shown).

DISCUSSION

We isolated rad27 alleles that displayed specific properties in chromosome stability assays and in their ability to cleave substrates that mimic structures formed during lagging-strand DNA synthesis. These alleles contain mutations that map to distinct nuclease domains (N for rad27-G67Sp, I for rad27-G240Dp) that were previously identified in RAD27 (7, 41). Like the rad27Δ allele, the rad27-G67S and -G240D alleles were inviable in the recombination-deficient rad52Δ strain background. Unlike rad27Δ strains, rad27-G67S and -G240D strains were resistant to MMS, were viable in an exo1Δ strain background, and were also viable at 37°C. The rad27-G67S and -G240D strains, however, could be distinguished from each other with respect to their chromosome instability phenotype and flap cleavage activities. These properties have made the mutant strains particularly useful in probing the role of Rad27p in maintaining chromosomal integrity. In chromosome stability assays, rad27-G67S strains displayed a higher frequency of repeat tract instabilities than of CAN1 insertion-deletion events; in contrast, the rad27-G240D strains displayed the opposite phenotype. In biochemical assays, rad27-G67Sp and rad27-G240Dp were found to have distinct defects in nuclease function, suggesting that catalytic deficiency is the basic cause of the biological defects in the mutant strains. rad27-G67Sp displayed a weak exonuclease activity but showed a double-flap endonuclease activity that was similar to that for the wild type and a single-flap endonuclease activity that was reduced only twofold. In contrast, rad27-G240Dp was devoid of exonuclease activity, displayed a double-flap endonuclease activity that was 60% of that for the wild type, and showed an extremely weak single-flap endonuclease activity.

Models to explain the mutator and DNA slippage phenotypes observed in rad27Δ strains.

Models have been developed to correlate the biochemical and genetic phenotypes exhibited by rad27 mutants. Tishkoff et al. (48) suggested that Rad27p endonuclease activity plays an important role in preventing insertion-deletion mutations that are observed in rad27Δ strains by cleaving 5′ flaps generated through the displacement of downstream Okazaki fragments by extension of the upstream fragment. Any delays in this cleavage increase the probability of DNA breakage at the flap junction. DSBR mechanisms are then thought to act on these break sites. Support of this model was provided by genetic studies which showed that rad27 mutants are inviable in recombination-defective strain backgrounds and also display a duplication mutator phenotype that is thought to result from DSBR (45, 48).

A second model was developed to explain the role of Rad27p (Fen1p) in preventing DNA slippage events (16) (Fig. 8). In this model, an intermediate in repeat sequence expansion involves the generation of a flap containing sequence repeats (Fig. 8, step 1). The reannealing of its 5′-end region to adjacent repeats in the template creates a structure called a bubble (Fig. 8, step 2). We previously have shown that this structure resists cleavage by Fen1p, because the annealed 5′-end region of the flap inhibits the tracking mechanism of the nuclease (16). Analysis of the kinetics of cleavage indicates that Fen1p attacks the bubble substrate with initial exonucleolytic cleavage of the annealed 5-end region. When flap length is reduced sufficiently so that the Fen1p can no longer bind effectively to the template, the nuclease tracks to the base of the flap and cleaves. A defect in exonuclease function would be expected to reduce the degradation of the 5′-annealed region, thereby prolonging the lifetime of the bubble intermediate. These delays in flap cleavage can allow for ligation of the bubble intermediate, leading to repeat sequence expansion (Fig. 8, steps 3b and 4b).

FIG. 8.

Proposed model for duplications and expansions. Displacement from DNA synthesis causes the formation of a flap structure (steps 1 and 2). Some sequences are able to adopt a secondary structure that forms either foldbacks or bubble intermediates (arrows). Rad27p binds to the 5′ end of the foldback or bubble and exonucleolytically degrade the annealed portions, resulting in the formation a more preferred flap structure (steps 2a and 3a). Step 4a, cleavage of the flap followed by synthesis and ligation complete synthesis. In this model, absence of or reduction in Fen1p activity (cross) permits the ligation of the bubble intermediate to an upstream primer (step 3b). Resolution of this structure by repair DNA synthesis or a subsequent round of DNA replication leads to expansion (step 4b).

Correlations between the genetic and biochemical defects in rad27-G67S strains.

We hypothesize that the high ratio of DNA slippage to Can^r mutator events observed in rad27-G67S strains resulted largely from defects in nuclease function that are important for degrading bubble intermediates, which can lead to DNA slippage events. As shown in Fig. 5 and 7, rad27-G67Sp displayed an extremely weak exonuclease activity on bubble, nicked, and gapped substrates. This defect correlates well with the relatively high level of DNA slippage events observed in rad27-G67S strains (Table 5). Genetic studies involving the Exo1 double-stranded 5′-to-3′ exonuclease were also consistent with the conclusion that the exonuclease defect is a major cause of the mutant phenotype in rad27-G67S strains. As shown in Table 3, the CAN1 mutator defect observed in rad27-G67S strains was almost completely suppressed by Exo1p overexpression, whereas the rad27Δ and rad27-G240D alleles were only partially suppressed. In contrast to its weak exonuclease activity, rad27-G67Sp displayed a flap endonuclease activity that was reduced only about twofold compared to that for the wild type for single-flap structures (Fig. 4B) and was similar to that for the wild type for double-flap structures (Fig. 6). This relatively active endonuclease activity correlates well with the low frequency of Can^r mutations and the weak insertion-deletion phenotype that was observed in rad27-G67S strains (Table 3).

Exonuclease activity of Rad27p is thought to reflect a specialized case of endonucleolytic cleavage, in which the nuclease takes advantage of transient denaturation of the 5′ end of a primer to recognize the 5′ nucleotide as a short flap (29). Although the same active site is thought to perform both endonucleolytic and exonucleolytic processes, one can envision how a mutation would specifically inhibit the ability of Rad27p to take advantage of the short lived substrate. For example, a distortion in the protein caused by the mutation might not allow the active site to bind the transient intermediate with sufficient affinity to allow catalysis.

Analysis of rad27-G240Dp.

Our results demonstrate that the catalytic defect in rad27-G240Dp is distinctly different. As shown in Fig. 4, rad27-G240Dp displayed an extremely weak flap endonuclease activity on a six-nucleotide flap substrate and a moderate activity on a double-flap substrate. In contrast to what was found for rad27-G67S, Exo1p overexpression only partially suppressed the CAN1 mutator phenotype of rad27-G240D. The level of suppression, to ∼15-fold higher than the wild-type level, was similar to that observed for rad27Δ strains overexpressing Exo1p (Table 3). This genetic observation, in combination with the relatively high frequency of the Can^r and strong-insertion-deletion phenotype observed in rad27-G240D strains, is consistent with a substantial defect in rad27-G240Dp endonuclease activity.

We also were unable to observe a significant exonuclease activity for rad27-G240Dp (Fig. 5 and 7; data not shown). In fact, the exonucleolytic defect of rad27-G240Dp is even more profound than that of rad27-G67Sp. With rad27-G67Sp, the presence of an upstream primer allows detectable exonucleolytic cleavage (Fig. 5B, lanes 9 to 14), whereas rad27-G240Dp is not stimulated by an upstream primer. These observations show that the catalytic defect of rad27-G240Dp is not limited to a particular class of substrate. Since rad27-G240Dp is defective for cleavage of a wider range of substrates, one might expect its phenotypic defect to be worse than that of rad27-G67Sp, but in DNA slippage assays it is less severe (Table 5). This suggests that the conditions present in vivo allow rad27-G240Dp to be sufficiently effective as a nuclease to carry out essential cleavage reactions in a timely manner. One possibility is that the double-flap substrate, on which the rad27-G240Dp displayed substantial activity, is a key substrate in vivo. Another possibility is that the nuclease interacts with other DNA replication proteins that stimulate its activity. Previous studies have shown that Rad27p interacts with replication factors including Dna2 helicase and proliferating cell nuclear antigen (PCNA), and yeast PCNA has been shown to stimulate Rad27p activity in vitro (3, 28, 49). Although yeast PCNA was able to activate rad27-G240Dp, increased cleavage was limited to the double-flap substrate. PCNA, however, did not restore the activity of rad27-G240Dp to the level seen for the wild type (data not shown).

Genetic and biochemical studies are consistent with a structural role for Rad27p.

The biochemical characteristics of the two mutant forms of Rad27p are not able to completely explain the relative differences of the mutant strains in canavanine resistance and repeat tract instability assays as well as their resistance to MMS. A likely explanation is that Rad27p serves a structural role that is also perturbed by the mutations and that Rad27p has distinct structural and catalytic functions. In this scenario, a combination of Rad27p structural and catalytic defects results in the observed phenotypes. In support of this idea, Rad27p has been shown to interact with other replication factors (see above) and recent reports have suggested that large complexes of proteins exist in eukaryotic cells that contain both DNA repair and replication factors. The absence of key components could therefore disrupt these complexes (54).

Current and future analyses of Rad27p mutants.

Although the crystal structure of S. cerevisiae Rad27p has not been solved, the structure of several homologs including T5 exonuclease (4), T4 RNase H (32), Methanococcus jannaschii Fen1p (21), and Pyrococcus furiosus Fen1p (PfFen1p) (19) are available. Based on the crystal structure of PfFen1p, the glycines at positions 67 and 240 are located within different regions of the enzyme. Glycine 67 is part of an α helix proposed to form one side of the catalytic groove of the enzyme. Mutation of this residue is consistent with the observed reduction in catalytic activity. Glycine 240 is located within a helix-turn-helix motif. This region is proposed to participate in the binding of Fen1p to the template strand of the substrate. In our studies, we do not detect major changes in the binding of the rad27-G240Dp to a flap substrate. However, the mutation may have more-subtle effects that alter the substrate sequence or structure requirements for efficient cleavage. We are currently performing extensive measurements of substrate specificity with this mutant.

In conclusion, the genetic and biochemical approaches described in this paper provide a powerful way to explore the specific cellular functions of Rad27p. These approaches can be effectively used in combination with structural analyses of Fen1p (19) to delineate the roles of Rad27p in DNA replication and in mutation avoidance.