Expansion and Length-Dependent Fragility of CTG Repeats in Yeast

See allHide authors and affiliations

Science  06 Feb 1998:
Vol. 279, Issue 5352, pp. 853-856
DOI: 10.1126/science.279.5352.853


Expansion of DNA trinucleotide repeats (TNRs) is the causative mutation in a growing number of human genetic diseases. Large expansions of a CTG tract were obtained and shown by genetic and physical assays to be length-dependent sites of chromosome breakage inSaccharomyces cerevisiae. Deletion of RAD27, which encodes a nuclease involved in Okazaki fragment processing, caused length-dependent destabilization of CTG tracts and a substantial increase in expansion frequency. The genetic assay described here can be used to evaluate other factors that induce TNR expansion or chromosome fragility in humans.

Expansion of TNRs has been shown to cause human genetic diseases, including myotonic dystrophy, Huntington's disease, several ataxias, and fragile X syndrome (1). As a TNR tract expands, the probability of further expansion increases. For example, at the myotonic dystrophy locus, the normal allele size of 5 to 37 CTG/CAG repeats (hereafter called CTG) is relatively stable, but when tracts reach ∼50 repeats, they are much more likely to expand in the next generation (1). TNR DNA forms stable hairpins that cause replication pausing in vitro (2, 3); it has been proposed that replication pausing promotes expansion or contraction in vivo by replication slippage (4). The five known loci of expanded CCG repeats in the human genome all share the additional feature of being fragile sites. Fragile sites, defined as gaps or breaks in condensed human chromosomes, are typically assayed by exposing tissue culture cells to reagents that slow replication or bind DNA (5). At least some fragile sites also cause chromosome breakage in humans (6).

Expansion and fragility of TNRs are thought to result from perturbations of DNA replication or chromatin structure or both, which are features that can be mimicked in yeast and other organisms. However, TNRs in model systems, including yeast, have so far failed to exhibit the dramatic and frequent expansions that occur in human disease, perhaps because expansions are rare in wild-type cells (7, 8). We wanted to identify a length-dependent phenotype for CTG tracts so that genes or conditions that affect TNR expansion could be isolated. To determine whether CTG tracts are fragile sites in yeast, we took advantage of the fact that a double-strand break (DSB) between two direct repeats increases the rate of recombination between the repeats, with concomitant deletion of the intervening sequences (9). Strains were constructed with a CTG tract and a URA3 gene between two direct repeats (Fig.1A). Recombinants were identified by elimination of URA3, which generates a cell that can grow in the presence of the drug 5-flouro-orotic acid (FOA). If the CTG tract increases chromosome breakage, strains with a CTG tract between the repeated regions will generate FOA-resistant (FOAR) cells at an elevated rate as compared with a strain without the CTG tract (Fig. 1A).

Figure 1

Assay for chromosome fragility. (A) Structure of the modified LYS2 locus is drawn approximately to scale except for 5′ LYS2. TheURA3 gene (1.17 kb) and CTG tract (390 to 750 bp) were placed between a full-length copy of the LYS2 gene and a 700-bp 3′ fragment of the LYS2 gene to create a 700-bp nontandem duplication of 3′ LYS2. Starting strains are FOA-sensitive (FOAS) because they contain theURA3 gene. If breakage occurs in the 2.1-kb region between the duplicated segments of LYS2, recombination can occur between them, thereby eliminating the intervening sequences to generate a FOAR cell. A control strain lacking the CTG tract was constructed as above, except that the duplicated segments ofLYS2 are 1.1 kb and the distance between them is 1.7 kb. Because of these differences, the recombination rates in the control strain are probably overestimated compared to the strains containing CTG tracts. (B) DNA was prepared from Lys+transformants (10) for the control strain, a CTG-130 tract strain, and transformants with expanded CTG tracts (tract size is indicated above each lane). DNA was digested with Eco R I and Hind III or Dde I, which are expected to yield fragments of 590 bp and 850 bp,respectively, for CTG-130; run on a 1% agarose gel; blotted to nylon membrane; and hybridized to a CTG probe. Recognition sites for Eco RI and Hind III are indicated in (A); Dde I cuts the region at 22 locations. (C) Fluctuation analysis was performed (12) to determine the rate of FOARcolony formation for the control strain and four CTG-tract strains (diamonds). The average of two (CTG-180 and CTG-210) or three (CTG-0, CTG-130, and CTG-250) experiments is shown. Error bars indicate the range of values obtained. Fluctuation analysis was also done using colonies grown on media containing 0.1 M HU (squares). The CTG-250 + HU sample value of 670 × 10–5 is off the chart. (D) The rate of generation of FOARcolonies is shown for CTG-0, CTG-130, and CTG-250 strains in the wild-type, rad50Δ, and rad52Δ backgrounds, and for CTG-0 and CTG-130 strains in therad1Δ background. Values are the average of two (rad50Δ, rad52Δ, and rad1Δ) or three [wild-type (WT); Fig. 1C] experiments (12). Error bars indicate the range of values obtained. For wild-type CTG-250, the range is 202 to 294 × 10–5. Double hatch marks indicate a discontinuity in the y axis.

To make the direct repeat substrate, we transformed a yeast strain containing a tract of 130 CTG repeats (CTG-130) within the yeastLYS2 locus on chromosome II (8, 10). Five of the 33 Lys+ transformants had expanded CTG tracts with tract sizes ranging from ∼170 to ∼250 repeats (Fig. 1B) (10, 11). To establish that the increase in band size was due to expansion of the CTG tract, we digested the DNA with 14 different restriction enzymes, including enzymes with sites withinLYS2 and URA3 and enzymes that cut variants of CTG. In each case, only fragments expected for expansion of the CTG tract were obtained (Fig. 1B) (11). Digestion with Bbv I, which cuts CTG repeats, eliminated all hybridization (11). In previous studies of TNRs in model systems such as Escherichia coli, yeast, human cells in culture, and mice, deletions were much more common than expansions, and expansions that occurred were usually small and present in only a subset of cells (7,8). Thus, some aspect of the transformation process, such as recombination near the tract, appeared to increase the generation or recovery of expansions.

To evaluate whether a long CTG tract could act as a fragile site in yeast, we determined the rate of recombination between the repeatedLYS2 sequences in five strains with tract sizes of 0 to 250 repeats, using fluctuation analysis (12). A substantial tract length–dependent increase in the rate of generating FOAR colonies was observed: Compared to the control, the rate was ∼5-fold greater for CTG-130, ∼9-fold greater for CTG-180, ∼27-fold greater for CTG-200, and ∼53-fold greater for CTG-250 (Fig. 1C). Southern (DNA) blot analysis showed that the structure of the LYS2 locus in FOAR colonies was as expected (11). To ensure that the increase in recombination was not due to a second site mutation, we introduced another recombination substrate of repeated sequences flanking a URA3 gene at a different chromosomal location into an FOAR derivative of the CTG-250 strain and into the parent strain. The rate of generation of FOAR cells was the same and was very low in both strains (11).

In human cells, growth in folate- or thymidine-deprived media slows replication and increases the breakage of expanded CCG tracts (5). In yeast, hydroxyurea (HU) slows DNA replication by lowering nucleotide pools (13). A fluctuation analysis of strains grown on plates containing HU showed that the rate of FOAR colony formation was increased about threefold for the control and for each CTG tract strain (Fig. 1C). Therefore, as with human cells, perturbation of DNA replication in yeast increases the likelihood of chromosome breakage. In E. coli, stalled replication forks are preferential sites for DSBs (14). If long TNR tracts cause pausing during DNA synthesis, breakage of stalled replication forks could explain the observed chromosome fragility. The effect of HU shows that the recombination assay can be used to detect conditions that increase TNR-induced chromosome breakage.

If CTG tract-mediated recombination proceeds via a DSB, then recombination should depend on genes required for DSB repair. The fluctuation analysis was repeated using strains lacking genes involved in DSB repair (12, 15): Rad52p is required for repair of DSBs by either gene conversion or single-strand annealing, Rad1p is required for single-strand annealing, and DSBs are processed more slowly in rad50Δ strains (9). The rate of generating FOAR colonies was decreased substantially and to about the same extent in both the rad52Δ andrad1Δ strains (Fig. 1D). The rate was also decreased in the rad50Δ background, although less dramatically. These data indicate that the FOAR colonies arise by repair of a DSB that occurs near or at the CTG tract, and that repair occurs mainly by single-strand annealing.

The structure of the chromosome with the CTG tract was determined by pulsed-field gel electrophoresis (16). To increase the fraction of broken chromosomes, a rad50Δ strain grown in HU was used. Southern hybridization revealed that therad50Δ CTG-250 strain contained intact chromosome II (825 kb) as well as two fragments with the size (∼470 and ∼355 kb) and hybridization properties expected for breakage in the vicinity ofLYS2 (Fig. 2). The 470- and 355-kb fragments were not detected in the control strain (Fig. 2). These data provide direct physical evidence that a CTG-250 tract increased breakage of chromosome II within a few kilobases of the tract. The CTG tract at the myotonic dystrophy locus in humans has not been observed to be a fragile site (17). However, given the considerable conservation in both chromatin structure and enzymes involved in DNA replication and repair between yeast and humans, it is likely that long CTG tracts also act as fragile sites in humans. Because conditions that increase chromosome fragility are different for different expanded sequences (5), the appropriate growth conditions for inducing breaks at CTG tracts may not have been tested, or CTG tracts may break at a frequency lower than that detectable by standard cytological techniques.

Figure 2

Pulsed-field gel analysis of chromosome breakage. rad50Δ CTG-0 and CTG-250 strains were grown in media containing 0.05 M HU, and chromosomes were isolated and separated by pulsed-field gel electrophoresis (16). The gel was blotted to nylon membrane and hybridized to probes either to the left (II-L) or right (II-R) (black boxes) of the CTG tract (white box). The centromere is represented by a black circle. Breakage of the 825-kb chromosome II near the LYS2 locus will produce bands of ∼470 kb (left fragment) and ∼355 kb (right fragment). The gel was probed sequentially with TLC1 (∼160 kb left ofLYS2), RAD16 (1.5 kb left of LYS2),TKL2 (1.5 kb right of LYS2), andRIF1(∼280 kb right of LYS2). The data for theTLC1 (left) and RIF1 (right) probes are shown. Size markers in kilobases were determined from the positions of intact yeast chromosomes on the ethidium bromide–stained gel. Broken chromosomes were also observed in the rad50Δ CTG-250 strain in the absence of HU, but the fraction was smaller (11). Longer exposures revealed additional bands that appear to be fusions of the breakage products (11).

The length-dependent breakage of CTG tracts in yeast provides a genetic assay for genes or conditions that increase tract expansions or fragility by testing their effects on the generation of FOAR colonies. In yeast, deletion of RAD27/RTH1causes decreased stability of dinucleotide repeats and an increase in duplication mutations between short direct repeats (18). The mammalian homolog of RAD27, FEN1 (flap endonuclease), is involved in processing the 5′ ends of Okazaki fragments (19). Comparison of wild-type andrad27Δ strains by means of the direct repeat recombination assay suggested an increase in the frequency of generating FOAR colonies in the rad27Δ background (11). To assess tract length quantitatively, tract size was determined with the polymerase chain reaction (PCR) in individual colonies grown at 30oC from both wild-type andrad27Δ strains carrying either a CTG-40 or CTG-70 tract (Table 1) (15, 20) and was confirmed by Southern hybridization in a subset of colonies (Fig. 3A) (20). Expansions and contractions were rare in the wild-type background. However, in therad27Δ strains, expansions and contractions were frequent (Table 1; see Fig. 3A for six examples). Compared to the wild type, the frequency of expansions and contractions of CTG-40 in therad27Δ strain were 36- and 10-fold greater, respectively. For CTG-70, the expansion frequency was at least 33-fold greater and the contraction frequency was 8-fold greater. Expansions were also detected in a CTG-130 rad27Δ strain by Southern hybridization (Fig. 3B, lane 4), but the frequency was not determined because tracts longer than CTG-130 were difficult to amplify with PCR.

Figure 3

Decrease in CTG tract stability byRAD27 deletion. Genomic DNA was prepared fromrad27Δ strains of the indicated tract length, digested with Eco RI and Hind III, separated on a 1% agarose gel, blotted to nylon membrane, and hybridized to a CTG probe. Size markers are given in base pairs. (A) Three representative expansions of the CTG-40 and CTG-70 tracts in rad27Δ strains are shown. In each panel, lane 2 is the CTG tract from wild-type strains (arrows). Estimated tract sizes are 83, 70, and 77 repeats for the CTG-40 expansions (lanes 1, 3, and 4), and 100, 100, and 83 repeats for the CTG-70 expansions (lanes 1, 3, and 4). For CTG-70, lane 1 shows a mixed population in which about half of the cells had an expansion and about half did not. (B) DNA was prepared fromrad27Δ CTG-130 strains grown at 25oC (lane 1) or 30oC (lanes 2 and 4). Lane 3 contains DNA from a wild-type CTG-130 strain grown at 30oC. The expanded tract in lane 4 is ∼177 repeats. The spot of low-molecular-weight hybridization at the bottom of the gel for the rad27Δ strains grown at 30oC (lanes 2 and 4) is probably due to degradation of the tract, because it was also present in undigested DNA (11).

Table 1

Increased frequency of expansions and contractions in a rad27Δ strain. PCR (20) was performed on individual colonies to determine the percentage of expanded, contracted, or unchanged tract lengths for both CTG-40 and CTG-70 strains in wild-type or rad27Δ backgrounds after growth at 30°C. Expansion or contraction was defined as having ≳30% of the tract expanded or shortened, respectively. Because cells grew for ∼25 generations, a minimum estimate of the expansion rate for therad27Δ CTG-70 strain is 10−2 per colony per generation. Examples of expansions are shown in Fig. 3.

View this table:

The instability of the CTG tract in rad27Δ strains increased with increasing tract length, because the CTG-hybridizing bands became more heterogeneous in length as tract size increased (Fig.3) (11). In addition, the instability of CTG tracts in therad27Δ strain increased with higher temperature: For CTG-130 at 25oC there was little change in tract length (Fig. 3B, lane 1), whereas tract heterogeneity was evident at 30oC (Fig. 3B, lanes 2 and 4). The increased frequency of expansions and contractions seen in rad27Δ cells was specific, because large or frequent expansions (or both) were not observed when other genes involved in DNA metabolism such asMLH1 (mismatch repair), REV3 (repair polymerase), or RAD1, RAD50, RAD51, orRAD52 genes (recombination) were mutated (11). Because Rad27p is normally involved in processing the 5′ ends of Okazaki fragments, these data suggest that unprocessed Okazaki fragments can lead to tract expansion and contraction. Extension at the 3′ end of an Okazaki fragment is predicted to displace the 5′ end of the adjacent Okazaki fragment to generate a 5′ single-stranded tail or flap that is normally cleaved by Rad27p (19). A flap of CTG repeats could fold into a hairpin structure, which might then be ligated to the 3′ end of the preceding Okazaki fragment to create an expansion (21).

Both genetic and physical assays indicated that CTG tracts are length-dependent sites of breakage in yeast and that a condition that slows DNA replication (the presence of HU) increased fragility. We have identified a mutation (rad27Δ) and a condition (transformation) that allow the recovery of large and homogeneous expansions of CTG repeats in yeast, similar in size to disease-causing expansions in humans. The increased frequency of CTG tract expansion in a strain lacking Rad27p suggests that humans with mutations in theRAD27 homolog FEN1 might be predisposed to expansion of TNRs, because FEN1 is highly conserved between yeast and humans (22). The recombination assay described here could be used to test other sequences, conditions, or gene products suspected of inducing chromosome fragility or TNR expansion.

  • * To whom correspondence should be addressed. E-mail: vzakian{at}


Stay Connected to Science

Navigate This Article