Yeast Ku as a Regulator of Chromosomal DNA End Structure

See allHide authors and affiliations

Science  01 May 1998:
Vol. 280, Issue 5364, pp. 741-744
DOI: 10.1126/science.280.5364.741


During telomere replication in yeast, chromosome ends acquire an S-phase–specific overhang of the guanosine-rich strand. Here it is shown that in cells lacking Ku, a heterodimeric protein involved in nonhomologous DNA end joining, these overhangs are present throughout the cell cycle. In vivo cross-linking experiments demonstrated that Ku is bound to telomeric DNA. These results show that Ku plays a direct role in establishing a normal DNA end structure on yeast chromosomes, conceivably by functioning as a terminus-binding factor. Because Ku-mediated DNA end joining involving telomeres would result in chromosome instability, our data also suggest that Ku has a distinct function when bound to telomeres.

The Ku protein, comprising two subunits of about 70 and 85 kD, appears to be present in all eukaryotic cells, which suggests that it is involved in a conserved and important function. Originally identified as an autoimmune antigen localized to the nucleus, this protein together with a third component, the catalytic subunit of a DNA-dependent protein kinase (DNA-PKcs), is now known to be critical for nonhomologous DNA double-strand break repair and the site-specific recombination of the V(D)J gene segments (1). In vitro, mammalian Ku binds directly to double-stranded (ds) DNA ends in a sequence-independent fashion (2). Thus, it has been postulated that Ku associates with dsDNA ends produced by DNA damage or during recombination and may recruit additional factors necessary for successful end joining (1).

The yeast Saccharomyces cerevisiae contains two genes,HDF1 and YKU80/HDF2 (hereafter referred to as YKU80), which are homologs of the two mammalian Ku subunits (3-5). Like its mammalian counterpart, yeast Ku binds DNA ends only in the heterodimeric form (5). Consistent with the proposed roles for Ku in mammalian cells, yeast cells lacking Ku activity are severely impaired in Rad52-independent nonhomologous DNA end joining (NHEJ) (4-6). Furthermore, yeast strains devoid of Ku do not grow at elevated temperatures (3, 4) and harbor dramatically shortened telomeres (4, 7). Recently, Hdf1p has been shown to interact with Sir4p in vivo (8). Because Sir4p is a necessary component for transcriptional silencing at several genomic loci, including telomeres, a role for Ku in telomere maintenance has been hypothesized.

The telomeres of virtually all eukaryotic chromosomes are composed of simple direct repeats. Yeast telomeres consist of 250 to 400 base pairs (bp) of TG2–3(TG)1–6 repeats (commonly abbreviated TG1–3 or C1–3A), and the G-rich strand forms the 3′ end of the chromosomes (9,10). During replication of yeast telomeres, chromosomal ends acquire a transient, single-stranded (ss) extension of the G-rich strand that can be detected in native DNA with telomeric probes (11, 12). In addition, these S-phase–specific overhangs can be detected in cells without telomerase, which suggests that they can be generated and processed by other enzymes (12, 13). Because virtually no overhangs can be detected by hybridization in yeast cells not in S phase (Fig.1A, lane 1) (11,12), the terminal DNA configuration in such cells could be blunt or could consist of a short overhang (<20 bases) of either strand.

Figure 1

Altered telomeric end structure caused by mutations in yeast Ku genes. (A) (Top) DNA from strains with the indicated genotypes and grown at 23°C was mock treated (lanes 1, 3, 5, 7, 9, and 11) or digested with E. coli exonuclease I (lanes 2, 4, 6, 8, 10, and 12). All samples were then digested with Xho I and analyzed by nondenaturing in-gel hybridization with a 22-mer oligonucleotide probe composed of telomeric C1–3A repeats (12). About two-thirds of the terminal restriction fragments are of a size indicated by the bar. Additional bands derive from individual telomeres lacking a subtelomeric repeated element Y′. Lanes ss and ds contain DNA with telomeric TG1–3 repeats in ss and ds form, respectively. Lane M, end-labeled 1-kb ladder DNA (Gibco-BRL) serving as size standard. (Bottom) Same gel after denaturation of the DNA and rehybridization with the same telomeric probe. The following strains were used: DWY291 (lanes 1 and 2), DWY290 (lanes 3 and 4), DWY292 (lanes 5 and 6), DWY293 (lanes 7 and 8), RWY737d (lanes 9 and 10), KWRY100 (lanes 11 and 12) (29). (B) A diploid yeast strain formed with haploids KWRY100 (tel1Δ::HIS3, YKU80) and RWY739b (TEL1, yku80) was sporulated and individual tetrads were dissected. All four spores from a tetratype tetrad were grown at 23°C (lanes 1, 3, 5, and 7) or incubated at 37°C (lanes 2, 4, 6, and 8) before DNA isolation and analysis by nondenaturing hybridization (top), followed by denaturation of the DNA and rehybridization (bottom) as in (A).

We reasoned that cells that lack the ability to re-form a normal terminal DNA structure would arrest in late S/G2 because of an abnormal end structure that might be recognized by DNA damage checkpoints. Thus, a bank of randomly mutagenized and temperature-sensitive (ts) yeast strains was screened for those that arrested in G2/M at the restrictive temperature; then they were analyzed for their terminal DNA structure by in-gel hybridization to native DNA (12). Analysis of an outcrossed strain (RWY737d) that carries a previously uncharacterized mutation is shown (Fig. 1A, lanes 9 and 10) (14). In addition to the strong telomeric signals for ss G strands observed in native gels, the overall length of the telomeric repeats was reduced by about 200 bp in this strain (Fig. 1A, bottom; compare lanes 1 and 9).

Three lines of evidence indicate that the mutation in strain RWY737d resides in the YKU80 gene. First, strains that carry a deletion of the YKU80 gene did not complement the ts, telomeric length, or DNA end-structure phenotypes, whereas strains that carry a deletion of HDF1 did complement (15). Second, a plasmid expressing a Myc epitope-tagged Yku80 protein complemented all these phenotypes (16). Third, when assayed for telomeric repeat length and ss extensions of the G-rich strand, the phenotype of a strain that carries a deletion in YKU80 was indistinguishable from that of strains that carry the mutation identified here (Fig. 1A, lanes 5 and 6). In addition, strains that carry a deletion of the HDF1 gene or strains with a deletion of both HDF1 and YKU80 genes also displayed the same phenotypes (Fig. 1A, lanes 3, 4, 7, and 8). Digestion of the genomic DNA with Escherichia coli exonuclease I, a 3′-specific ssDNA exonuclease (12, 17), eliminated all signals in the native gels (Fig. 1A, top), reduced the hybridization intensity of the terminal restriction fragments after denaturation (Fig. 1A, bottom), and slightly reduced the sizes of these fragments. We estimate that the overhangs are about 50 to 90 bases long (18). Thus, the telomeric signals observed for Ku cells in native gels correspond to unusually long terminal extensions of the G-rich strand.

For at least some end-joining reactions in mammalian cells, an association of Ku with DNA-PKcs is required (1). Yeast encodes a number of genes such as TEL1 andMEC1 that share some sequence homologies with the gene for DNA-PKcs (19). Most notably, yeast strains with a deletion of TEL1 also display shortened telomeres (Fig.1A, bottom, lane 11; Fig. 1B, bottom, lanes 5 and 6) (20). However, the terminal DNA in yeast strains with a deletion ofTEL1 had no detectable overhangs (Fig. 1A, top, lanes 11 and 12; Fig. 1B, top, lanes 5 and 6). In addition, strains with a mutation in both YKU80 and TEL1 had even shorter telomeric TG1–3 tracts than either single mutant and displayed ss extensions (Fig. 1B, lanes 7 and 8) (7). Thus, the altered DNA end structure observed for strains with mutations in one of the Ku genes is not simply a consequence of a short telomeric repeat tract but a specific phenotype associated with the lack of Ku. Consistent with this conclusion, the signals for ss G-rich extensions were equally visible whether Ku cells were grown at permissive or restrictive temperatures (Fig. 1B, top, lanes 3 and 4). These results also support the notion that Ku and Tel1p affect telomere length through different genetic pathways (7).

It was formally possible that a fraction of the Ku cell population would delay in late S phase and therefore have telomeres with long overhangs, but for the rest of the cell cycle have normal terminal DNA structures (21). To investigate this possibility, we induced yku80 mutant cells to arrest in G1, early S, or M phase and then assayed for G-strand overhangs (Fig. 2). Although wild-type strains did not have ss G strands at the telomeres at any of the arrest points, the arrested yku80 mutant cells always yielded signals that were indistinguishable from those obtained with nonsynchronized mutant cells (Fig. 2, A and C). Thus, whereas the telomeres in wild-type cells acquire G-strand extensions only transiently in late S phase, the telomeres in Ku cells have such overhangs throughout the cell cycle.

Figure 2

Analysis of G-strand overhangs throughout the cell cycle. (A) Strains RWY738a (yku80), RWY738d (YKU80), and DWY293 (hdf1Δ::TRP1, yku80Δ::URA3) were grown to logarithmic phase in rich medium (lanes 1, 3, and 5, marked C) or arrested in G1 phase by incubation with 0.75 μM α factor for 2 hours (lanes 2, 4, and 6, marked G1). DNA was isolated and analyzed for G-strand overhangs on terminal restriction fragments as in Fig. 1. (B) Before DNA isolation, part of the culture analyzed in (A) was stained for DNA with propidium iodide and was analyzed for DNA content by standard flow cytometry. Relative fluorescence is indicated on the x axis in arbitrary units. (C) Strains DWY291 (YKU80), DWY292 (yku80Δ::URA3), and RWY737d (yku80) were grown to logarithmic phase (lanes 1, 4, and 7, marked C), arrested with 0.4 M hydroxyurea in early S phase (lanes 2, 5, and 8, marked S), or arrested with nocodazole at 20 μg/ml in M phase (lanes 3, 6, and 9, marked M). Flow cytometric DNA analysis showed that >90% of the cells were at the expected arrest points (23). Analysis for G-strand overhangs in these samples was as in (A). Denaturation and rehybridization of the gel with the telomeric oligonucleotide showed that all lanes contained about equal amounts of DNA (23).

To determine whether yeast Ku modulates DNA end structures by binding to chromosomal ends, we performed in vivo cross-linking studies. A yku80 mutant strain was transformed with a plasmid encoding a Myc epitope-tagged Yku80p and protein-DNA cross-links were induced in vivo (Fig. 3) (22). After we immunoprecipitated whole cell extracts with the Myc-specific antibody, we analyzed DNA from the pellets and supernatants by Southern blotting. DNA fragments containing telomeric repeats were recovered in immunoprecipitates from the strain harboring the pKU80-myc plasmid (Fig. 3A, lanes 5 and 6) but not from strains lacking Yku80p (Fig. 3A, lane 1). These DNA fragments were not immunoprecipitated in the absence of antibody (Fig. 3A, lane 2) or in the absence of formaldehyde-induced cross-linking (Fig. 3A, lanes 3 and 4). Finally, when a probe specific for the yeast ribosomal DNA repeat was used on the same blot, no signal above background was detectable after immunoprecipitation (23).

Figure 3

Immunoprecipitation of telomeric repeat DNA with an antibody specific for a tagged Yku80p after in vivo cross-linking. (A) Strain RWY737d (yku80) was transformed with pRS316 (lane 1) or with pKU80-myc (lanes 2 to 6) (16). Protein-DNA cross-linking, preparation of whole-cell extracts, and immunoprecipitation with the 9E10 antibody were as in (22), except that protein G-agarose (Boehringer) was used instead of protein A–Sepharose. DNA from the pellets was deproteinized and analyzed by Southern blotting with a fragment containing 280 bp of TG1–3 repeats as probe (11). The controls were as follows: lane 1, control plasmid without the YKU80-myc gene; lane 2, no antibodies added; lanes 3 and 4, no cross-linking (X-link) induced. In addition, 0.2 μg of a linearized plasmid containing 280 bp of telomeric repeats (24) was added to the cells represented in lanes 4 and 6 after cross-linking (lane 6) or before extract preparation (lane 4). In lane 7, linearized pVZ1 DNA was loaded as a control. Bar indicates the signal for telomeric repeat-containing fragments. IP, immunoprecipitate; Comp. DNA, competitor DNA. (B) Blot in (A) was rehybridized to a randomly labeled pVZ1 probe. (C) As a control for DNA input, 4% of the DNA that remained in the supernatants and washes after immunoprecipitation was analyzed by Southern blotting with the pVZ1 DNA probe. Thick arrow indicates the linearized pKU80-myc. Small arrow indicates the linearized pRS316 DNA. Heavy smears below 3 kb in lanes 4 and 6 are due to the added competitor DNA.

To verify that the fragments immunoprecipitated were due to in vivo cross-links and not associations that occurred during extract preparation, we added a fivefold excess of a linearized plasmid to the cells before we prepared the extract (24). Fragments containing telomeric DNA were immunoprecipitated as efficiently as in the absence of competitor DNA (Fig. 3A, compare lanes 5 and 6), but virtually no competitor DNA was recovered in the pellet (Fig.3B). These results demonstrate that yeast Yku80p can be cross-linked in vivo to chromosomal DNA composed of telomeric repeats. The exact binding site for Ku on the telomeres is unknown, but the in vitro properties of Ku and the altered terminal DNA structure in Ku cells strongly suggest that Ku binds at or near the most distal telomeric repeats on yeast chromosomes.

These results show that yeast Ku is involved in establishing the proper terminal DNA structure on yeast chromosomes and that this effect is likely mediated by the direct binding of Ku at or near the chromosome ends. We have also observed a dramatic reduction in the transcriptional repression of URA3, a gene required for uracil biosynthesis, when this gene is located near a telomere in Ku cells (25); this result reinforces the conclusion that Ku is present in normal telomeric chromatin. Surprisingly, Ku cells are viable at the permissive temperature, although they have an altered terminal DNA structure. However, yeast cells harboring both a deletion of the TLC1gene, which encodes the RNA component of telomerase (26), and a yku80 mutation die after only about 10 generations (Fig. 4). Similar results were obtained when thecdc13-2est mutation, which affects telomerase function in vivo (27), was combined with a yku80mutation (Fig. 4). These results show that defects in components associated with the yeast telomerase are deleterious in strains that do not maintain a normal terminal DNA end structure. It is therefore possible that in rapidly dividing cells grown at elevated temperatures, Ku-mediated establishment of a proper terminal DNA structure would be required for maintaining a functional block of telomeric repeats via telomerase, whereas in cells grown at 23°C normal telomerase activity or alternative recruitment of telomerase by ss G-strand binding proteins would suffice. Alternatively, some aspect of the telomerase-mediated maintenance of telomeric repeats could be inherently sensitive to elevated temperatures.

Figure 4

Accelerated death of cells when a yku80mutation is combined with mutations affecting telomerase. (Top) Diploid strain RWY80 (cdc13-2est/CDC13, YKU80/yku80) (31) was sporulated and individual tetrads were dissected. Colonies of a single tetrad were restreaked onto rich medium and incubated at 23°C (left). Cells harboring the yku80mutation alone did not grow at 37°C (right) but did not senesce after five successive restreaks at 23°C (23). Cells with thecdc13-2est mutation were not ts (right), but they ceased to grow after two successive restreaks (23). Cells labeled WT (wild type) were not ts and did not show senescence. (Bottom) Diploid RWY85 (tlc1Δ/TLC1, YKU80/yku80) (31) was sporulated, tetrads were dissected, and cell phenotypes were identified as described above. For both crosses, each tetratype tetrad analyzed in this way (six of six) contained one colony from which viable cells could not be recovered and the genetic analysis of the other three colonies indicated that these dead cells were the double mutants. The number of times the cells were restreaked at the indicated temperatures is indicated by 1× and 2×, which correspond to about 30 and 50 generations of growth, respectively.

In addition to its role in DNA end joining (1, 5, 6), we show here that Ku also binds to chromosome ends, affects the terminal DNA configuration, and is required in the absence of a functional telomerase. However, telomere-to-telomere end-joining reactions would lead to dicentric chromosomes and genomic instability. Therefore, the presence of Ku at yeast telomeres is surprising and suggests that this protein is involved in at least two distinct mechanisms, depending on its binding site. We speculate that interactions of Ku with telomerase-associated proteins or with proteins involved in NHEJ may play a crucial role in distinguishing chromosome ends, where end-to-end fusions are not desirable, from dsDNA breaks within a chromosome. Because the predominant ds break repair mechanism in mammalian cells is NHEJ and because telomeres in these cells are maintained by telomerase, it will be interesting to investigate the roles of Ku in telomere maintenance in such systems.

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


Stay Connected to Science

Navigate This Article