Postreplication Repair Inhibits CAG {middle dot} CTG Repeat Expansions in Saccharomyces cerevisiae

Trinucleotide repeats (TNRs) are unique DNA microsatellitesthat can expand to cause human disease. Recently, Srs2 was identifiedas a protein that inhibits TNR expansions in Saccharomyces cerevisiae.Here, we demonstrate that Srs2 inhibits CAG · CTG expansionsin conjunction with the error-free branch of postreplicationrepair (PRR). Like srs2 mutants, expansions are elevated inrad18 and rad5 mutants, as well as the PRR-specific PCNA allelespol30-K164R and pol30-K127/164R. Epistasis analysis indicatesthat Srs2 acts upstream of these PRR proteins. Also, like srs2mutants, the pol30-K127/164R phenotype is specific for expansions,as this allele does not alter mutation rates at dinucleotiderepeats, at nonrepeating sequences, or for CAG · CTGrepeat contractions. Our results suggest that Srs2 action andPRR processing inhibit TNR expansions. We also investigatedthe relationship between PRR and Rad27 (Fen1), a well-establishedinhibitor of TNR expansions that acts at 5' flaps. Our resultsindicate that PRR protects against expansions arising from the3' terminus, presumably replication slippage events. This workprovides the first evidence that CAG · CTG expansionscan occur by 3' slippage, and our results help define PRR asa key cellular mechanism that protects against expansions.

INTRODUCTION

Top

INTRODUCTION

Expansions of trinucleotide repeats (TNRs) are the genetic basisfor at least 15 human neurological diseases, including Huntington'sdisease, myotonic dystrophy type 1, and fragile X syndrome (39,42, 60). Inheritance patterns in families afflicted with thesediseases show an ongoing mutational process where the disease-causingallele tends to increase in length in each generation. Thisdynamic mutation pattern (47, 48) results in the non-Mendelianinheritance behavior called anticipation, which is a signatureof this class of diseases. TNR tracts have several unique mutationalproperties in affected families. First, TNR mutations are locusspecific. For example, the HD locus is highly unstable in Huntington'spatients, but mutations are rare elsewhere in the genome, includingother DNA microsatellites (16). DNA sequence is also critical.Fourteen of the diseases in this class are caused by mutationof the repeat sequence CNG (39, 42, 60), where N correspondsto any nucleotide. The sole exception, Friedreich's ataxia,has an unstable GAA sequence (8). This sequence restrictioncorrelates strongly with the ability of these sequences to formhairpins and other unusual secondary structures in vitro (15).Accordingly, these secondary structures are thought to be acritical component of the expansion mechanism (39, 42, 60).Together, locus specificity and sequence restriction indicatethat DNA metabolism at TNR sequences is unusual, compared tothe rest of the genome. Thus, new rules appear to govern howcellular proteins process TNRs to either promote or preventexpansions.

Since TNR expansions require the addition of DNA, clearly themechanism must involve a DNA synthetic event. Various modelsof TNR instability invoke replication and/or repair (28, 39,42, 60), and each model is by no means mutually exclusive. Forproliferating cells, one widely supported model for TNR expansionsinvolves aberrant DNA replication on the lagging strand (18,25, 36, 38, 49). In support of this idea, DNA synthesis of repetitiveelements, such as TNRs, has a high error rate due to polymeraseslippage (44). Furthermore, replication forks often stall whensynthesizing TNR-containing DNA (43, 52). By this model, a DNApolymerase synthesizing through a TNR-containing region inadvertentlyprovokes formation of a single-stranded region (at either the3' or 5' end of the DNA being synthesized) that folds into ahairpin or other secondary structure. One version of the modelenvisions polymerase-mediated strand displacement of the downstream5' terminus to create a flap (18). Left intact, the flap canfold into a hairpin and be ligated, thus sealing extra TNRsinto the newly synthesized strand. The Fen1 flap endonucleasehelps remove single-stranded TNR flaps during Okazaki fragmentprocessing, thereby preventing TNR expansions (20, 33, 34).In vivo, the 5' flap model is supported by high TNR expansionrates in yeast deficient in the Fen1-encoding gene RAD27 (13,34, 54, 57). An alternative view envisions polymerase stallingand 3' terminus slippage (reviewed in reference 60). Slippageat the 3' end allows the polymerase to aberrantly replicatea region of the template more than once. Formation of secondarystructure stabilizes this expanded replication intermediate.As noted earlier, polymerase stalling at TNRs has been documented;however there is little direct experimental evidence to supportthe idea that expansions result from 3' slippage events.

In addition to the important cis factors, such as structure-formingcapabilities, trans factors involved in DNA metabolism playimportant roles in mediating TNR instability as well. In humans,evidence for trans-acting factors include the subset of TNRdiseases with a parent-of-origin effect (23, 42, 60). Also,variations in germ line instability between individuals withsimilar Huntington's disease allele lengths (10, 31) suggestthat genetic differences can affect TNR stability quite substantially.As part of the effort to identify relevant trans-acting factorsin model organisms, we recently reported Saccharomyces cerevisiaeSrs2 as a potent inhibitor of TNR expansions (4). Srs2 has twomajor functions. It negatively regulates homologous recombination(1, 27, 35, 40, 45, 51, 53, 61), and it positively modulatesthe RAD6 postreplication repair (PRR) pathway (1, 7, 40, 45,51, 53, 56). Genetic evidence (4) indicated that the majorityof Srs2's inhibitory action for TNR expansions was independentof homologous recombination; therefore, PRR became a focus forinvestigation.

PRR appears to be well conserved from yeast to humans (6, 46).There are well-established human homologs for most PRR proteins,with the exceptions of Srs2 and Rad5. PRR is a damage tolerancepathway that directs gap filling of aborted DNA replicationintermediates arising when a polymerase is unable to proceedpast damage on the template strand (6). Although the PRR pathwaydirects tolerance of both induced and spontaneous forms of DNAdamage, evidence indicates that the role PRR proteins play inbypassing spontaneous damage differs from that played in bypassinginduced damage (9, 14, 32, 37). For example, although Rad18is essential for mutagenesis following induced DNA damage, itis only partially required for spontaneous mutagenesis (14).Thus, there appear to be two similar, but distinct, subpathwaysof PRR: one for bypassing spontaneous lesions and one for bypassinginduced lesions.

Trinucleotide repeat mutations are believed to occur spontaneously,and so for ease we will focus only on PRR at spontaneous lesions.For spontaneous damage, evidence suggests that repair is mediatedby either a Rad6/Rad18 heterodimer or a Rad5/Rad5 homodimerthat assembles on the DNA and directs bypass/tolerance throughthe modification of PCNA (41). These dimers can direct error-prone,translesion synthesis through polymerase ${zeta}$ (Pol ${zeta}$ ). Rad6/Rad18-mediatedmonoubiquitination of PCNA can also direct error-free bypass.This error-free pathway is directed through subsequent polyubiquitinationof PCNA mediated by Ubc13, Mms2, and Rad5. Although the mechanismof this pathway is poorly understood, current evidence in Escherichiacoli suggests that bypass may occur via a pathway analogousto the template switch pathway mediated by DnaK (17). In supportof this, error-free PRR in S. cerevisiae may occur via recombinationbetween partially replicated sister strands in a process thatcompetes with Rad52-mediated homologous recombination (56, 63).Interestingly, error-free PRR has also been linked to completionof replication in response to stresses like hydroxyurea (HU)(5). This evidence suggests that PRR is important for bypassingstalled replication sites as well as damaged templates. Together,the features of PRR and its known relationship to Srs2 madethis pathway relevant for study with respect to TNR instability.

In this study, we utilize genetic means to examine the roleof PRR in TNR instability. By this approach, we have identifiedthat PRR proteins and PRR-specific PCNA modifications inhibitTNR expansions, and this effect is directed by the action ofSrs2. Collectively, this work highlights the importance of adamage bypass pathway in protecting against aberrant DNA replicationat TNRs and provides evidence supporting the 3' slippage modelof TNR expansions.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Strains. BL035, a leu2 version of the wild-type strain MW3317-21A (MAT ${alpha}$ ${Delta}$ trp1 ura3-52 ade2 ${Delta}$ ade8 hom3-10 his3-kpn1 met4 met13) (26), wasutilized. With the exception of the POL30 lysine mutants, allmutants described in this study are deletions. Most mutantswere isogenic derivatives created by targeted deletion techniquesand confirmed by PCR and/or phenotypic traits like UV sensitivity.srs2::LEU2 was a random transposon mutant of the wild-type strainisolated in a screen and confirmed as previously described (4).The endogenous POL30 gene was replaced by integrative replacementplasmids, derived from site-directed mutagenesis of the POL30plasmid pCH1572 (2), to create the pol30-K164R, pol30-K127R,and pol30-K127/164R strains. Integration of the mutant constructswas confirmed by Southern blotting. pol30-K164R and pol30-K164/127Rmutants were further verified by UV sensitivity and the lossof UV-induced mutagenesis. The chromosomal integration of TNR-containingplasmids and confirmation of correct single integrants weredone as described previously (11).

Plasmids. The pBL94 vector was used to construct all TNR-containing plasmidsas described previously (50). POL30 replacement plasmids wereprovided by Polina Shcherbakova, University of Nebraska MedicalCenter. TRP1-marked pRAD18 was created by blunt end ligationof the 1.1-kb BstUI TRP1 fragment of pRS314 (55) and the 10.6-kbbackbone of FspAI-digested YCp11. Plasmid YCp11 (ClaI insertof RAD18 and its endogenous promoter in YCp50) was generouslyprovided by Francis Fabre, Commissariat a l'Energie Atomique.The dinucleotide repeat-containing plasmid, pSH44, was a generousgift of Tom Petes, Duke University.

Genetic assays and molecular analysis of mutated TNR alleles. Expansion and contraction rates were measured by fluctuationanalysis as described previously (11, 38, 50). Briefly, forexpansions pure TNR tracts were normalized to 25 repeats withrandomized C, T, and G deoxyribonucleotides. For example, (CTG)₁₃tracts have 13 contiguous CTG repeats followed by 36 bp of nonrepeating,randomized C, T, and G deoxyribonucleotides. Normalized TNRtracts were cloned into an S. cerevisiae promoter-reporter constructthat allows spacing-sensitive expression of the downstream URA3reporter. Yeast cells harboring an expansion of four or morerepeats do not express URA3 and are identified by their resistanceto 5-fluoroorotic acid. For contractions, pure TNR tracts arenormalized to 33 repeats, and contractions of 5 or more repeatsallow expression of URA3; cells are identified by their abilityto grow in the absence of uracil. Mutation rates were calculatedby the method of the median (30). Dinucleotide mutation rateswere measured as described previously (19). Forward mutationrates for the CAN1 gene were determined by fluctuation analysiswith selection for canavanine resistance. Three to five independentclones were tested for each of the above assays to ensure reproducibility.Single-colony PCR analysis of expansions and contractions weredone as previously described, and rates were corrected by multiplyingthe percent bona fide expansions/contractions by the apparentmutation rates obtained by fluctuation analysis (11). All statisticalanalyses were performed using the t test (two-tailed distributionand two-sample equal variance) in Microsoft Excel, and P valuesof less than 0.05 were considered statistically significant.Any outliers were determined using the Q-test.

RESULTS

Top

RESULTS

Trinucleotide repeat expansions are exacerbated in PRR mutants. Recently, we identified Srs2 as a potent inhibitor of TNR expansions(4). Srs2 functions both as a negative regulator of homologousrecombination, and it also channels intermediates resultingfrom stalled replication forks into the Rad6-mediated PRR pathway.It was previously shown that elevated TNR expansion rates insrs2 mutants are largely independent of homologous recombination(4). Therefore, we turned our attention to Srs2's role in thePRR pathway. We predicted that Srs2 inhibits TNR expansionsthrough its contribution to PRR. To test this prediction, TNRexpansion rates were examined in various mutants of the PRRpathway.

As a model, we chose PRR-mediated bypass of spontaneous lesions(37). This model is the most relevant, since our assay systemdoes not employ DNA damaging agents; therefore TNR expansionsand contractions occur spontaneously. As described earlier,PRR of spontaneous DNA damage has been genetically characterizedas a three-pathway model, with two error-prone branches (directedby either Rad18 or Rad5) and one error-free branch (mediatedby both Rad18 and Rad5). Thus, if PRR helps prevent TNR expansions,then loss of either RAD18 or RAD5 should lead to higher expansionrates. Both (CAG)₂₅ and (CTG)₁₃ expansions were elevated inrad5 (four- to sixfold) and rad18 (eight- to ninefold) (Table1). Furthermore, this defect in rad18 could be completely rescuedby ectopically expressing RAD18 in the mutant cells, but notby an empty vector control, confirming that loss of RAD18 isresponsible for the increased level of expansions (Table 1).In addition, for each TNR tract examined, the expansion rateswere elevated to a similar extent in both rad18 and rad5, andthere was no significant difference between the respective expansionrates (P = 0.25). To further address the relationship betweenrad18 and rad5, (CTG)₁₃ expansion rates were measured in therad18 rad5 double mutant. The result was indistinguishable fromrad18 and rad5 results (data not shown). While additivity cannotbe completely ruled out, this observation is consistent withthe idea that Rad5 and Rad18 proteins function in the same pathwayto inhibit TNR expansions. Previous work in our lab has illustratedthat (CTG)₁₃ tracts expand by +4 to +11 repeats, but (CAG)₂₅repeats generally have larger expansions (generally +9 to +25repeats). Since both of these tracts have increased expansionrates in our mutants, these results indicate that PRR is importantfor inhibiting CAG · CTG repeat expansions over the rangeof +4 to +25 repeats.

View this table:
[in this window]
[in a new window]

TABLE 1. TNR mutation rates of PRR mutant strains^a

Based on the observation that srs2 suppresses the sensitivityof rad18 and other PRR mutants to methylmethane sulfonate andUV (7, 29), it is generally believed that Srs2 acts upstreamof the PRR machinery. If this were also true for TNRs, srs2would be epistatic to the other PRR mutants for inhibiting TNRexpansions. To test this hypothesis, we generated an srs2::kanMXmutant, which could be combined with the LEU2-marked PRR mutants.All subsequent experiments in this study used the srs2::kanMXallele, hereafter referred to as srs2. Expansions of a (CTG)₁₃tract were elevated 13-fold in srs2 (Table 1). (We note thatthis rate increase is about threefold less than that previouslyreported for an srs2::LEU2 random transposon mutant [4]. Thebackgrounds of these mutants are identical, so differences inexpansion rates must reflect subtle differences between thealleles.) As predicted, srs2 proved to be epistatic to rad5for TNR expansions, as the srs2 rad5 double mutant gave 11-fold-higher(CTG)₁₃ expansion rates. This value is statistically indistinguishablefrom that of the srs2 single mutant, but it is statisticallydifferent from that for rad5. Although additivity cannot becompletely excluded, these data are consistent with the ideathat Srs2 acts upstream of Rad5 to direct proper maintenanceof the TNR through the action of PRR.

Spontaneous lesions can be repaired through an error-prone branchor an error-free branch of PRR (37). Error-prone bypass by Pol ${zeta}$ is directed by Rad18 or Rad5 homodimers, and error-free bypassoccurs via a pathway involving both Rad6/Rad18 and Rad5. Todistinguish between these possibilities, we measured (CTG)₁₃expansions in a rev7 strain. A rev7 mutant is devoid of theregulatory subunit of Pol ${zeta}$ , rendering the polymerase nonfunctional;therefore, this mutant is only capable of error-free bypass.In contrast to rad5 and rad18, we did not observe any significantchange in (CTG)₁₃ expansion rates in rev7 (10a). Although Rad30(Pol ${eta}$ ) is another bypass polymerase that has been implicatedin PRR, previous work illustrated that a rad30 strain had littleor no detectable change in TNR mutations (12). As a second wayto examine the role of error-free PRR, expansions were measuredin a ubc13 mutant which is deficient in polyubiquitination ofPCNA. The expansion rate of (CAG)₂₅ was elevated sixfold overwild type (Table 1), a phenotype statistically indistinguishablefrom those of rad18 and rad5. These results, along with theabove evidence for PRR inhibition of TNR expansions in rad18and rad5, suggest that error-free bypass, not error-prone, isimportant for inhibiting expansions.

To further examine the role of PRR inhibition of TNR expansions,PCNA modifications were examined genetically. Recent advancesin the PRR field have illustrated that modifications of PCNAare vital for dictating the different outcomes of PRR. Two independentmodifications of PCNA are relevant for PRR, sumoylation andubiquitination. K164 of PCNA is the predominant site of PCNAubiquitination and sumoylation. Sumoylation at this site enhancesSrs2's interaction with PCNA and presumably enhances Srs2'srecruitment to stalled replication forks (40, 45). Notably,sumoylation is not essential for PRR (40), and sumoylated PCNAhas not been detected in human cells. PCNA becomes monoubiquitinated(by Rad6/Rad18) and/or polyubiquitinated (by Rad5 and otherproteins) at K164 to direct the various outcomes of the PRRpathway (22). A second residue, K127, can also be sumoylated.This sumoylation appears to be secondary to K164 sumoylationand is generally not believed to be crucial for PRR (40, 45).

To test whether these modification sites are important for TNRinstability, expansion rates were measured in pol30-K127R, pol30-K164R,and pol30-K127/164R mutants. Mutation of these lysine residuesto arginine abolishes their ability to be sumoylated and ubiquitinated.Results in Table 1 show that elevated levels of (CTG)₁₃ and(CAG)₂₅ expansions were observed in strains harboring pol30-K164R(up to 45-fold) and pol30-K127/K164R (up to 9-fold), but substantialelevations were not observed for pol30-K127R. Modificationsat K127 are not crucial for inhibiting TNR expansions, whereasmodifications at K164 help to inhibit expansions. These resultssupport the idea that PRR inhibits TNR expansions, since PCNAmono- and polyubiquitination at K164 is required for PRR. Interestingly,there is a fivefold difference between the TNR expansion ratesof pol30-K164R and both pol30-K127/164R and srs2. Previous worksuggested that the higher spontaneous mutation rate of pol30-K164Rcompared to pol30-K127/164R is dependent on Pol ${zeta}$ (58). Thisdoes not appear to be the case for TNR expansions, however,because expansion rates of pol30-K164R and the double mutantpol30-K164R rev7 are indistinguishable from one another (datanot shown). Although there is no clear explanation for the higherrate of expansions in pol30-K164R compared to other PRR mutants,a similar situation was observed for sensitivities to DNA damagingagents, where pol30-K164R is more sensitive than either srs2or pol30-K127/164R (40).

To address the relationship between PCNA modifications and Srs2,comparisons were made between (CTG)₁₃ expansion rates of srs2,pol30-K164R, and the double mutant srs2 pol30-K164R. The comparisonsare consistent with an epistatic relationship between the twomutants (Table 1). Since the double mutant is not significantlydifferent from srs2 (11-fold versus 13-fold, respectively),it implies Srs2 acts upstream of K164 modification to inhibitTNR expansions.

To gain insight into the types of TNR mutations occurring inPRR mutants, expansion sizes were measured. Previous reportswith srs2 illustrated that (CTG)₁₃ tracts usually expand by5 to 10 repeats with a bell-shaped spectrum, which is similarto a wild-type spectrum (+4 to 11 repeats) (4). Likewise, expansionsfrom (CTG)₁₃ starting tracts from rad18 and pol30-K164R werein the range of +4 to 11 repeats (Fig. 1). Comparisons betweenthe expansion spectra of the wild type, srs2 (4), rad18, andpol30-K164R demonstrate that the expansion ranges overlap, suggestingthe same types of expansion events are occurring and no uniqueexpansion events have emerged. Thus, as with srs2, the sheernumber of expansions accounts for the increased expansion rateobserved in PRR mutants.

View larger version (26K):
[in this window]
[in a new window]

FIG. 1. Spectra of (CTG)₁₃ expansion events for wild type, srs2, rad18, pol30-K164R, rad27, rad18 rad27, and pol30-K164R rad27 strains. The x axis denotes the number of repeats added, e.g., +7 denotes a final allele length of 20 CTG repeats. The y axis indicates the number of expansions of each size. The panel for rad27 is intentionally duplicated in both columns for ease of comparison. Data for all expansions from srs2 and some expansions from wild-type cells are from reference 4. One +31 expansion from the srs2 mutant (4) was omitted here to facilitate comparisons of expansions up to +25 repeats.

Ubiquitination of PCNA is important for inhibiting TNR expansions. The results presented above clearly illustrate that modificationof PCNA, largely at residue K164, is important for inhibitingTNR expansions. Additionally, the enhancement of TNR expansionsin rad18 and rad5 mutants indicates a clear role for PCNA ubiquitination.However, these results do not reveal whether sumoylation isimportant. Where ubiquitination is important for directing PRR,sumoylation of PCNA enhances the recruitment of Srs2 to thereplication fork (40, 45). These two modifications appear tobe largely independent of one another, since polyubiquitinationof PCNA can occur in the absence of sumoylation and vice versa(40). Additionally, these modifications do not compete, sinceneither modification is enhanced in the absence of the other(40). To determine which modification of PCNA is required forinhibiting TNR expansions, we planned to knock out SIZ1, whichencodes a protein required for sumoylation of PCNA at residueK164, and then assay TNR expansions. During the course of thiswork it became clear that our parental strain was already SIZ1deficient. SIZ1 resides on chromosome IV adjacent to ADE8, anda deletion introduced into ADE8 (26) extends into SIZ1 as well(T. Mertz, personal communication). Biochemical analysis revealedthat PCNA sumoylation at K164 is absent in our parental strain(H. D. Ulrich, personal communication). Since TNR expansionsare elevated in PRR mutants in the absence of sumoylation ofPCNA, we conclude that ubiquitination of PCNA is indeed importantfor inhibiting TNR expansions. To address K164 sumoylation,we restored SIZ1 in our strain, verified by biochemical analysisthat PCNA sumoylation was restored (H. D. Ulrich, personal communication),and then reassayed for TNR expansions. Rates of TNR expansionwere unchanged when SIZ1 was restored, suggesting that Siz1has no role in mediating TNR expansions (Table 2). To verifythe SIZ1 result, we measured TNR expansion rates in a SIZ1-proficientbackground, which is completely independent from other strainsused in these studies. This wild-type strain and an isogenicrad18 single mutant had rates of expansion that were elevatedto a similar extent as in our Siz1-deficient background (Table2). Differences in strain backgrounds likely explain the variationbetween the expansion rates of the two rad18 strains. Theseresults indicate that Siz1p has no detectable role in CAG ·CTG repeat expansions in our system.

View this table:
[in this window]
[in a new window]

TABLE 2. Siz1p does not affect (CTG)₁₃ expansions

pol30 modification mutations specifically inhibit expansions of mutable TNRs. Defects in Srs2 specifically exacerbate TNR expansions but haveno substantial effect on TNR contractions, mutations of dinucleotiderepeats, or forward mutation rates (4). If Srs2 achieves thisspecificity in conjunction with its role in PRR, then othermutants in PRR should show a similar specificity. We testedthis prediction by examining mutation rates for several differentmutational reporters in pol30-K127/164R. Defects in PCNA modificationsdo not substantially increase contractions of (CTG)₂₅ repeats(Table 1) or of (CAG)₂₀ repeats (data not shown). These resultsindicate that the protective action of Srs2 and PRR are directedtowards the newly synthesized strand, where expansions occur,rather than the template strand, which is thought to be thesite where contractions primarily arise. TNR mutations are believedto occur through a mechanism that is distinct from other spontaneousmutations, such as frameshifts and point mutations. To furtheraddress the specificity of PRR for inhibiting TNR expansions,we used PCNA mutants to assess mutation rates of other sequences.The frameshift mutation rate of a dinucleotide repeat tract,(GT)_16.5, was unchanged for pol30-K127/164R compared to wild-typecontrols (Table 3). This suggests that error-free PRR does notplay a discernible role in inhibiting dinucleotide repeat mutationsand that the decrease in poly(GT) tract mutation rates in arad5 mutant reported previously (24) might be due to an error-pronefunction of RAD5. Forward mutations that inactivate the CAN1gene occurred at similar rates for the wild type, pol30-K164R,and pol30-K127/164R (Table 3), in accordance with previouslypublished results for pol30-K127/164R (45). Collectively, theseexperiments illustrate that PRR-specific PCNA modificationsare important for inhibiting TNR expansions but not other spontaneoussequence mutations.

View this table:
[in this window]
[in a new window]

TABLE 3. Mutation rates of non-TNR sequences are unaffected in the pol30 modification mutants tested

TNR expansions are believed to be mediated through a stablesecondary structure and, in support of this model, non-hairpin-formingTNRs (15) are genetically stable in our system (50). We testedwhether the inhibition of TNR expansion by PCNA modificationcorrelated with the ability of the sequence to form a stablehairpin structure by measuring the expansion rate of the geneticallystable (CTA)₂₅ repeat and a scrambled repeat sequence, (C, T,G)₂₅, each of which is incapable of significant hairpin formation.As expected, these repeat tracts did not detectably undergoexpansions in either wild-type or pol30-K127/164R strains (Table1). Together, these results suggest that, like Srs2, the PRRpathway specifically inhibits expansions of TNR sequences capableof hairpin formation.

Synergistic interaction between Rad27 and PRR for TNR expansions. Rad27 is a well-characterized trans-acting factor for inhibitingTNR expansions (13, 20, 33, 34, 54, 57). These findings arein excellent agreement with models for TNR expansions resultingfrom 5' flaps which then fold into hairpin structures (18).An alternative expansion model posits that 3' strand slippageoccurs if DNA polymerases stall or pause in a TNR tract (reviewedin reference 60). By this model, the slipped strand folds intoa hairpin with a 3' terminus in the duplex region, which canthen be extended by a polymerase. To date, however, data todirectly support expansions from 3' slippage have not been available.PRR processing is predicted to be recruited to an intermediatecontaining a 3' end where replication has been stalled (59);therefore, our data with PRR mutants may indicate expansionsfrom 3' slippage. If so, inhibition of TNR expansions by Rad27and PRR would be exclusive of one another and should be geneticallydistinct. A model to further clarify this hypothesis is shownin Fig. 2.

View larger version (17K):
[in this window]
[in a new window]

FIG. 2. 5' and 3' models of TNR expansions. A. At the 5' end of an Okazaki fragment, Rad27 (Fen1) cleavage inhibits TNR expansion. In the absence of Rad27, TNR-containing flaps adopt stable TNR hairpins, which can be ligated to form an expanded strand. B. At the 3' end, polymerase stalling and/or hairpin formation at the TNR tract recruits PRR. Possible unwinding by Srs2, and also resolution of the stalled replication fork by error-free PRR, inhibits TNR expansions.

To examine the relationship between these two pathways, we generatedrad18 rad27 and pol30-K164R rad27 double mutants and measuredtheir expansion rates. (srs2 rad27 double mutants are inviableand could not be tested.) Interestingly, double mutant expansionrates of a (CTG)₁₃ repeat revealed a synergistic increase inexpansion rates relative to single mutants (Table 4). Specifically,the expansion rate in the rad18 rad27 strain was elevated 230-foldover wild type, compared to 8-fold for rad18 and 93-fold forrad27. Similarly, the pol30-K164R rad27 double mutant showeda 930-fold increase over wild type, compared to values of 45-foldand 93-fold for the pol30-K164R and rad27 single mutants. Bothdouble mutant rates were statistically different from the singlemutant values (Table 4). These data clearly indicate that thetwo pathways act independently but that they synergize to inhibita subset of expansion events. By inference, expansions of (CTG)₁₃in wild-type yeast occur by both 5' flap and 3' slippage mechanisms.

View this table:
[in this window]
[in a new window]

TABLE 4. Synergistic interactions between PRR and Rad27 for (CTG)₁₃ expansions

To further test the separability of Rad27 and PRR activity onblocking expansions, we examined the expansion size profilesof single and double mutant strains. The histograms in Fig.1 illustrate the unique features of each mutant. With the PRRmutants rad18 and pol30-K164R, TNRs expanded by +4 to +11 repeats.In contrast, TNRs in rad27 expanded by +7 to +10 repeats withinfrequent, but characteristic, large expansions (+15 repeats).The accumulation of large expansions in a rad27 strain has beennoted previously (57). Statistical analysis indicates that thesePRR mutant and rad27 spectra are significantly different fromone another (rad18 versus rad27 and pol30-K164R versus rad27;P < 0.001). The expansion spectra from the two double mutantsrad18 rad27 and pol30-K164R rad27 approximate a merge of therespective single mutants. Both double mutants showed a rangeof expansion sizes that encompassed the range of the PRR mutantsand rad27 (+4 to +11), and both double mutants resulted in occasionallarge expansions (+20 to +25). These mutation spectra providefurther evidence that the two pathways of inhibiting TNR expansionsare separable and that expansions occurring at the 3' end tendto be smaller than those occurring at the 5' end. Together,the rate data and these mutational spectra support the ideathat expansions in our system can arise either from 5' flapsor from 3' slippage events.

DISCUSSION

Top

DISCUSSION

This study identified the error-free PRR pathway as a novelprotective mechanism for inhibiting TNR expansions. Mutationof RAD18 or RAD5, which are required for ubiquitination of PCNA,or mutation of the ubiquitin acceptor site at PCNA residue lysine164 resulted in 4- to 45-fold increases in expansions of (CTG)₁₃and (CAG)₂₅ repeat tracts. This mutator effect was specificfor expansions, as no mutator phenotype was detected for TNRcontractions, dinucleotide repeat mutations, or forward mutationsat CAN1 in PRR mutants. This unusual mutator signature matcheswell to that seen previously for srs2 mutants, and epistasisanalysis indicated that SRS2 functions with PRR to help avoidexpansions. Additional analysis of PRR mutants indicated thatmost of the expansion inhibition can be attributed to the error-freepathway of PRR. This is apparent from the similar rates of TNRexpansions in rad18, rad5, and ubc13 mutants and the lack ofa phenotype when the error-prone pathway was blocked by a rev7mutation (10a). Synergy between mutants of PRR and the 5' flapendonuclease encoded by RAD27, as well as the mutational spectra,suggest that PRR works from the 3' side of the putative hairpinintermediate. Together, these results help clarify the roleof Srs2 and PRR proteins in enhancing faithful maintenance ofthe TNR tract in wild-type yeast. PRR is thought to be largelyconserved between yeast and humans (6, 46). If this functionalconservation includes action to stabilize trinucleotide repeats,our yeast results suggest that PRR in mammalian cells helpsoffset the high mutability associated with trinucleotide repeats.

Based on these findings, we propose a model for spontaneousPRR of TNRs (Fig. 3) that is based largely on the spontaneouslesion bypass pathway described by Jinks-Robertson and colleagues(37). We propose that DNA synthesis through TNRs results inhairpin formation and/or polymerase stalling, and this signalsrecruitment of Srs2 and the PRR machinery. Action by Srs2 andPRR helps to surpass the replicative stress (i.e., fork stallingand/or hairpin formation) through faithful maintenance of theTNR tract in a manner dependent on mono- and polyubiquitinationof PCNA. Since the mechanism of error-free PRR is still poorlyunderstood, the precise mechanism that allows error-free synthesisof the TNR tract remains enigmatic. This model is supportedby several pieces of evidence. First, in this study we haveshown that TNR expansions are elevated in mutants that blockeach step shown in Fig. 2. Second, it is well established thatreplication forks often stall at TNR tracts (43, 52). However,it is unknown whether stalling is caused by the TNR tract aloneor following hairpin formation. Third, the HU sensitivity ofPRR mutants suggests that PRR proteins are recruited to helpresolve stalled replication forks (5), and so it is reasonableto suggest that PRR helps relieve stress associated with DNAsynthesis through TNR tracts. What happens to a TNR hairpinin the absence of PRR is an open question. Either the hairpinpersists and leads directly to an expansion, or some alternativeprocessing event actively promotes expansion in an error-pronemanner. Additional work will be necessary to distinguish betweenthese two possibilities.

View larger version (12K):
[in this window]
[in a new window]

FIG. 3. Model for PRR inhibition of TNR expansions. Polymerase stalling and/or hairpin formation within the TNR tract triggers the PRR pathway. Action from the 3' end allows the TNR to be properly maintained through error-free PRR. Gray shading indicates the factors, at each step of the model, which have been shown here to be important for inhibiting expansions.

The expansion rates and expansion spectra of single and doublerad27 and PRR mutants indicate that these two pathways act throughlargely independent pathways to inhibit expansions. It is wellaccepted that Rad27 helps to prevent TNR expansions at the 5'end of Okazaki fragments through the cleavage of single-strandedTNR flaps (13, 20, 33, 34, 54, 57). On the other hand, PRR ispredicted to act from the 3' end, based on the model of PRR(59) and the behavior of PRR mutants when treated with HU (5).Additionally, this model for PRR action is consistent with ourexpansion results. Thus, these two pathways afford cells theopportunity to inhibit TNR expansions at both the 5' and 3'ends of newly synthesized DNA (Fig. 2) and, combined, they offerdramatic protection for CTG tracts near the crucial thresholdlength in yeast. We note that the expansion rate observed inthe pol30-K164R rad27 double mutant, 1.3 x 10^–4 per generation(Table 4), makes the (CTG)₁₃ tract over 10-fold more unstablethan a (CTG)₂₅ repeat in wild-type cells (38, 50). Thus, the(CTG)₁₃ allele in the double mutant background takes on theexpansion behavior of a much longer triplet repeat.

A synergistic interaction between PRR and Rad27 suggests thatthese pathways are not completely independent and each may compensatefor the loss of the other. One possible explanation for thisresult is that loss of either Rad27 or PRR increases the formationof TNR structures that are substrates for the remaining pathway.For example, in the absence of Rad27, 5' flaps can reequilibrateinto various alternative structures (21, 62) which may be substratesfor PRR. One possible intermediate would arise if the 5' flapreannealed to the template and displaced the upstream Okazakifragment (62), resulting in a 3' TNR slip-out. This seems likea reasonable substrate that may recruit PRR. Alternatively,in the absence of PRR, TNR loop-outs at the 3' end would persistand might also reequilibrate with downstream Okazaki fragmentsto generate 5' flap substrates for Rad27.

In addition to directing PRR to the triplet repeat tract, Srs2may play a second role by unwinding the putative hairpin intermediate.This possibility is supported by biochemical experiments showingthat purified Srs2 is unusually active at unwinding oligonucleotidesubstrates containing TNR repeats (3). If so, one might predictthat sumoylation of PCNA by Siz1 might help recruit Srs2 tothe hairpin in vivo, based on studies from the antirecombinationproperties of Srs2 (40, 45). Interestingly, a siz1 mutationdid not appreciably affect expansions (Table 2), suggestingthat PCNA sumoylation is not essential for inhibiting CAG ·CTG expansions in this system. If Srs2 does have a second roleto unwind the hairpin intermediate, it does not seem to requirerecruitment by PCNA sumoylated at residue K164.

ACKNOWLEDGMENTS

This work was supported by the National Science Foundation GraduateResearch Fellowship Program (to D.L.D.), NCI training grantT32 CA09476 (to D.L.D.), and National Institutes of Health grantGM61961 (to R.S.L.).

We thank Frances Fabre for providing YCp11, Wei Xiao for theDBY747 strains, Tom Petes for pSH44, Polina Shcherbakova forPOL30 replacement plasmids, and Helle Ulrich for biochemicalconfirmation of PCNA sumoylation.

FOOTNOTES

* Corresponding author. Mailing address: Eppley Institute for Research in Cancer and Allied Diseases, University of Nebraska Medical Center, Box 986805, Omaha, NE 68198. Phone: (402) 559-4619. Fax: (402) 559-8270. E-mail: rlahue{at}unmc.edu.

Published ahead of print on 23 October 2006.

REFERENCES

Top