The Dna-binding Protein Hdf1p (a Putative Ku Homologue) Is Required for Maintaining Normal Telomere Length in Saccharomyces Cerevisiae a Yeast Gene (hdf1) Encoding a Putative Homologue of the 70 Kda Subunit of Ku Has Recently Been Identified

In mammalian cells, the Ku autoantigen is an end-binding DNA protein required for the repair of DNA breaks We find that hdf1 mutant strains have substantially shorter telomeres than wild-type strains. We speculate that Hdf1p may bind the natural ends of the chromosome, in addition to binding to the ends of broken DNA molecules. Strains with both an hdf1 mutation and a mutation in TEL1 (a gene related to the human ataxia telangiectasia gene) have extremely short telomeres and grow slowly.


INTRODUCTION
Yeast chromosomes terminate with a simple repeated sequence poly G 1-3 T (3).Although this terminal tract in wild-type strains is usually between 350 and 500 bp in length, mutant strains with shorter or longer tracts have been identified (4).In one such mutant, tel1, the telomeric poly G 1-3 T tracts are shortened to ∼50 bp (5).The TEL1-encoded protein shares homology with phosphatidylinositol (PI)/protein kinases, and is most similar to the human ataxia telangiectasia (AT) protein (6,7).
The mammalian Ku protein includes two subunits (70 and 82 kDa) that bind to the ends of double-stranded DNA (8,9).In addition to interacting with the ends of DNA, Ku also interacts with DNA-dependent protein kinase (p350) (10)(11)(12).Since mammalian cells that are defective in Ku are X-ray sensitive and deficient in VDJ joining (1), Ku is involved in the repair of double-strand DNA breaks.A heterodimeric (70 and 85 kDa) DNA-binding protein has been identified in Saccharomyces cerevisiae and the gene encoding the 70 kDa subunit (HDF1, high affinity DNA-binding factor) has been cloned (2).The yeast protein has ∼22% identity to the human Ku protein over the C-terminal half.Cells with hdf1 null mutations fail to grow at 37_C, but grow and sporulate normally at 30_C (2).
Below, we show that a mutation in HDF1 reduces the length of chromosome telomeres.In addition, the hdf1 mutation has a synthetic interaction with tel1 (another mutation affecting telomere length) resulting in strains that grow slowly and have very short telomeres.

Yeast strains and plasmids
Yeast strains with the hdf1 and/or tel1 mutations were constructed by transformation of the wild-type strains W303a (a ade2 his3 leu2 trp1 ura3) or SLK29-1B (a ade2 ura3) using plasmids containing disruptions of these genes.Strains with the genotype TEL1 hdf1 were derived from SLK29-1B (derivative SPY38) and W303a (derivative W303aU) by transformation of the parental strains with an SspI-HindIII fragment derived from plasmid pGEM4Z S-H/URA (2) (hdf1::URA3); strain W303aU was provided by H. Feldmann (University of Munich).SPY37 is a spontaneous ura3 derivative of W303aU.The tel1 HDF1 strain PG21 was constructed by a two-step transplacement of SLK29-1B with the plasmid pPG25 (tel1::IS1 insertion in YIp5; ref. 7), and the tel1 HDF1 strain SPY40 was derived from W303a by a single-step transplacement with an EcoRI-KpnI fragment of the plasmid pPG47 (tel1::URA3 insertion in Bluescript SK - vector; ref. 7).The tel1 hdf1 strain SPY39 was derived from PG21 by transformation with the plasmid pGEM4Z S-H/URA, and the tel1 hdf1 strain SPY41 was constructed from SPY37 by transformation with the plasmid pPG47.
The plasmid pYT14 contains telomeric repeats and ∼1 kb of the sub-telomeric Y′ element (13).This plasmid was used as a hybridization probe.

Analysis of telomere lengths
The lengths of telomeres were examined by treating DNA with the restriction enzyme XhoI and examining the products by Southern analysis, using pYT14 as a hybridization probe (details of the protocol in refs 7,14).The enzyme XhoI cuts in the conserved Y′ repeat located at the ends of most yeast chromosomes, generating terminal restriction fragments of ∼1.3 kb in wild-type strains.This fragment includes ∼400 bp of the telomeric repeat poly G 1-3 T.

Measurements of X-ray sensitivity
Cultures in the exponential phase of growth were harvested and resuspended in water at a density of ∼10 7 cells/ml.Samples were irradiated with doses of 0, 10, 50, 100 and 220 krad and spread on plates containing rich growth medium in order to monitor cell survival (7).Four plates were analyzed for each X-ray dose in each experiment.The plates containing unirradiated samples had between 100 and 500 colonies.Ninety-five percent confidence intervals on measurements were calculated using the GraphPad INSTAT program on the Macintosh computer.

RESULTS
To determine whether Hdf1p, a putative homologue of the Ku protein, is involved in the control of telomere length in yeast, we examined isogenic strains of four types: TEL1 HDF1, tel1 HDF1, TEL1 hdf1, and tel1 hdf1.Two different genetic backgrounds were used, W303 (2) and SLK29-1B (7).Mutations were introduced into these strain backgrounds by transformation; since the introduced DNA contain disruptions of the coding sequence of TEL1 or HDF1, the mutant alleles are likely to represent null mutations.We examined the phenotype of the tel1 hdf1 double mutant, in addition to the phenotypes of the single mutants, in order to determine whether the Tel1p and Hdf1p gene products acted in a single pathway.If both mutants affect telomere length and the double mutant has the same phenotype as the single mutants, it is likely that Tel1p and Hdf1p act in the same pathway.Alternatively, if both mutants affect telomere length, but the double mutant has a different phenotype than the single mutants, Tel1p and Hdf1p act in different pathways affecting telomere length.One reason that Tel1p and Hdf1p might be expected to act in the same pathway is that Tel1p shares homology with the mammalian protein DNA-PK, a DNA-dependent protein kinase that interacts with Ku (15).
The restriction enzyme XhoI cuts yeast DNA in the sub-telomeric Y′ repeat, generating a terminal restriction fragment in wild-type yeast strains of ∼1.3 kb, ∼350-500 bp representing the terminal poly G 1-3 T tract (14).When DNA was isolated from strains that were grown without extensive sub-culturing (lanes 5-8 and 13-16), the wild-type strains had the longest telomeres, the single tel1 or hdf1 strains had intermediate length telomeres, and the double mutant tel1 hdf1 strains had the shortest telomeres; since all strains have a distribution of telomere lengths and these distributions often overlap, these conclusion are based on the average telomere length which represents the center of the distribution.We estimate that the terminal poly G 1-3 T tract in the double mutant strains is <50 bp.
The full expression of the mutant phenotype in yeast strains with mutations affecting telomere length often requires multiple (∼100) generations (5).We, therefore, vegetatively sub-cultured Table 1.X-ray resistance of wild-type, hdg1, tel1 and tel1 hdg1 strains 1 The first mumbers represent the average per cent survival (based on two independent experiments), normalized to 100% survival at 0 dose of X-rays.The numbers in parentheses are the per cent survival measured in the individual experiments (additional details in Materials and Methods) with 95% confidence limits.
the mutant strains for 100 generations and re-examined telomere length (Fig. 1, lanes 1-4 and 9-12).In both genetic backgrounds, wild-type strains had the longest telomeres, and strains with the hdf1 mutation had telomeres that were shorter than wild-type, but slightly longer than strains with the tel1 mutation.In the SLK29-1B genetic background, the tel1 hdf1 strain had the shortest telomeres, whereas, in the W303 genetic background, the tel1 hdf1 strain had telomeres that were approximately the same size as those observed in the tel1 HDF1 strain.In summary, in both genetic backgrounds, the rate at which telomeric sequences were lost was greater in the tel1 hdf1 double mutant strains than in either of the single mutant strains.In addition, in one of the genetic backgrounds, the telomeres at equilibrium (after sub-culturing) were shorter in the double mutant strain than in the single mutant strains.
Strains with either the tel1 mutation or the hdf1 mutation grow at approximately the same rate as wild-type strains at 30_C.The double mutant strains, however, had a distinctly slower growth rate in both genetic backgrounds than the rates of the single mutant strains (Fig. 2).Possible reasons for the slow growth of the double mutant will be discussed below.
Since mammalian cells that lack Ku are sensitive to X-rays (1), we examined the X-ray sensitivities of strains with hdf1, tel1 and tel1 hdf1 genotypes.As shown in Table 1, none of these strains were substantially more sensitive to X-rays than the isogenic wild-type strain.

DISCUSSION
The results described above indicate that Hdf1p has a role in telomere metabolism.Since the tel1 hdf1 double mutant has phenotypes that are different from the single mutants, the effect of Hdf1p on telomere length is likely to involve a different pathway from that involving the TEL1 gene product.Given the known end-binding activity of Hdf1p, one obvious possibility is that this protein protects the telomeres from cellular exonucleases, and the mutant phenotype reflects an altered balance between telomere elongation and degradation.Since Tel1p shares homology with yeast genes involved in checkpoint control of DNA damage, Tel1p may be involved in checkpoint control of telomere length (7).
Given the role of Hdf1p in telomere metabolism in yeast, it would be of interest to examine the lengths of telomeres in Ku - mammalian cell lines.
It is also possible that Hdf1p affects telomere length without binding to the chromosomal ends.Strains with the hdf1 mutation are temperature-sensitive and, at the restrictive temperature, accumulate elevated amounts of DNA (2).Feldmann and Winnacker (2) suggested, therefore, that Hdf1p may be involved in the control of the initiation of DNA replication.Although our experiments were done at the permissive temperature for growth of hdf1 strains, it is possible that the regulation of DNA replication is somewhat aberrant.Thus, hdf1 strains could have an imbalance between cycles of 'normal' DNA replication and telomere replication, leading to shortened telomeres.
As described above, the tel1 hdf1 strains grew slowly relative to the single mutant strains.One possibility is that this slow rate reflects chromosome loss from some cells in the population whose chromosomal telomeres have been completely eliminated.Although the tel1 mutation increases chromosome loss rate only 3-4-fold, the double mutant may have a stronger effect.Whatever the cause of the slow growth, by isolating mutants that restore the wild-type growth rate, one may be able to identify other yeast gene products that interact with Tel1p or Hdf1p.
We found that strains with hdf1, tel1 or tel1 hdf1 genotypes were not more X-ray sensitive than wild-type strains.These results confirm studies, done in other genetic backgrounds, indicating that tel1 (7) and hdf1 (16) strains are not unusually X-ray sensitive.Since the double mutant also has wild-type sensitivity to X-rays, we conclude that the two gene products are not functionally redundant with each other in the repair of X-ray damage.Interestingly, however, both gene products appear to be functionally redundant with other gene products involved in the cellular response to X-ray damage.Strains with the tel1 mec1 genotype are much more sensitive to X-rays than strains with either single mutation, and the X-ray sensitivity of mec1 strains can be complemented with a single extra dose of the TEL1 gene product (17).In addition, hdf1 rad52 genotype are more sensitive to X-ray damage than strains with either single mutation (16).In summary, our data and those of other investigators indicate that both the Tel1p and Hdf1p proteins have dual roles, a unique function in telomere metabolism and a functionally redundant role in the repair of X-ray damage.