Pif1p Helicase, a Catalytic Inhibitor of Telomerase in Yeast

See allHide authors and affiliations

Science  04 Aug 2000:
Vol. 289, Issue 5480, pp. 771-774
DOI: 10.1126/science.289.5480.771


Mutations in the yeast Saccharomyces cerevisiae PIF1gene, which encodes a 5′-to-3′ DNA helicase, cause telomere lengthening and a large increase in the formation rate of new telomeres. Here, we show that Pif1p acts by inhibiting telomerase rather than telomere-telomere recombination, and this inhibition requires the helicase activity of Pif1p. Overexpression of enzymatically active Pif1p causes telomere shortening. Thus, Pif1p is a catalytic inhibitor of telomerase-mediated telomere lengthening. Because Pif1p is associated with telomeric DNA in vivo, its effects on telomeres are likely direct. Pif1p-like helicases are found in diverse organisms, including humans. We propose that Pif1p-mediated inhibition of telomerase promotes genetic stability by suppressing telomerase-mediated healing of double-strand breaks.

PIF1 is a nonessential Saccharomyces gene that encodes a 5′-to-3′ DNA helicase (1). Mutations in PIF1 affect telomeres in three ways: telomeres from pif1 mutant cells are longer than wild-type telomeres; healing of double-strand breaks by telomere addition occurs much more often in pif1cells than in wild-type cells; and pif1 cells but not wild-type cells add telomeric DNA to sites that have very little resemblance to telomeric DNA (2). These data suggest that Pif1p is an inhibitor of telomere lengthening. Pif1p also affects mitochondrial (1) and ribosomal DNA (3).

There are two mechanisms that can lengthen the ∼300–base pair (bp) tracts of yeast telomeric C1-3A/TG1-3 DNA: telomerase (4) and telomere-telomere recombination (5). In the absence of genes required for telomerase such asTLC1, which encodes telomerase RNA (6), and EST1, which encodes a telomerase RNA binding protein (7), telomeric DNA gets shorter and shorter, the cultures senesce, and most cells eventually die. Lengthening of telomeres by recombination requires the continued presence of Rad52p (5). If Pif1p inhibits telomere-telomere recombination, telomere lengthening will not occur in a pif1 rad52 strain, and a pif1 tlc1 (or est1) strain might not senesce or would senesce more slowly due to activation of the recombinational pathway for telomere maintenance. If Pif1p inhibits telomerase, telomere lengthening would not occur in pif1 tlc1 orpif1 est1 strains. To distinguish between these possibilities, we constructed singly and doubly mutant strains of the appropriate genotypes and examined telomere lengths (8). Because telomeres were at least as long in a pif1 rad52 as in a pif1 strain (Fig. 1A), the effects of Pif1p did not require Rad52p. In contrast, telomere lengthening did not occur in either a pif1 tlc1 (Fig. 1B) or a pif1 est1 strain (9). In addition, lack of Pif1p did not bypass or even delay the senescence phenotype of cells lacking telomerase (Fig. 1C). Thus, Pif1p inhibits a telomerase-dependent pathway of telomere lengthening.

Figure 1

Pif1p inhibits telomerase, not telomere-telomere recombination. (A) DNA was prepared from three independent transformants from otherwise isogenic strains of the indicated genotypes. The DNA was digested with Xho I and analyzed by Southern hybridization using a C1-3A/TG1-3 telomeric probe here and in (B). The pif1-m2 allele, which affects telomeric but not mitochondrial DNA (2), was used here and in (B) and (C). (B) A diploid strain heterozygous at bothTLC1 and PIF1 was sporulated, tetrads were dissected, and the genotype of the spore products was determined. DNA was isolated from independent spores with the indicated genotypes ∼30 cell divisions after sporulation. (C) Individual spores from tetrads obtained as in (B) were streaked on rich medium and grown to single colonies (∼25 cell divisions). Individual colonies were restreaked repeatedly. The third and fourth restreaks after sporulation are shown for the four spore products from one of nine tetrads examined.

To determine if the helicase function of Pif1p is required to inhibit telomere lengthening, we used site-directed mutagenesis to modify the invariant lysine in the ATP-binding domain to either alanine (K264A) or arginine (K264R) (10), as this residue is essential for the activity of other helicases (11). Both the wild-type and the K264A mutant version of Pif1p were expressed in Sf9 insect cells infected with recombinant baculovirus, purified to near homogeneity (Fig. 2A), and their activities assayed in vitro. Whereas wild-type Pif1p catalyzed unwinding of a 17-base, [32P]end-radiolabeled oligonucleotide annealed to a M13 single-strand circle, the K264A allele had no helicase activity in this assay (Fig. 2B).

Figure 2

Pif1p helicase activity is essential for inhibition of telomere lengthening. (A) Recombinant wild-type Pif1p and mutant Pif1p-K264A were purified from Sf9 insect cells infected with recombinant virus and analyzed with silver-stained SDS–polyacrylamide gel electrophoresis (SDS-PAGE) (left panel) or Western blotting with affinity-purified antibodies to Pif1p (right panel). (B) Helicase activity assays were carried out using a 1-μM solution of the partial duplex DNA substrate (a 5′, [32P]end-radiolabeled, 17-nucleotide oligomer annealed to the single-stranded M13mp7 DNA) and 50 ng of purified recombinant protein from (A). Reactions were incubated at 37°C for 30 min and analyzed by 10% PAGE and autoradiography. (C) Genomic DNA from wild-type and mutant strains was analyzed as in Fig. 1A. The lanes contain DNA from otherwise isogenic wild-type (lane 1),pif1Δ (lane 2), pif1-m1 (lane 3),pif1-m2 (lane 4), pif1-K264A (lane 5),pif1-K264R (lane 6), pif1-K264A carrying pVS102, a centromere plasmid bearing wild-type PIF1 (lane 7),pif1-K264R carrying pVS102 (lane 8), and the wild type (lane 9). (D) Western analysis was performed on proteins isolated from the mutant and wild-type cells whose DNA was examined in (C), using affinity-purified anti-Pif1p serum. Lanes are the same as in (C). (E) Genomic DNA was isolated from three independent cultures of PIF1 cells carrying either the multicopy vector YEpFAT7, YEpFAT7 containing the PIF1 gene, or YEpFAT7 containing thePIF1-K264A mutant gene, and then analyzed by Southern blotting as in Fig. 1A.

To determine the phenotype of cells that lacked Pif1p helicase activity, strains with only the K264A or K264R allele were constructed (12). DNA was prepared from cells carrying these mutant alleles, as well as from the wild-type pif1Δ,pif1-m1, or pif1-m2 strains, and examined by Southern analysis (Fig. 2C). The pif1-m1and pif1-m2 alleles are point mutations in, respectively, the first or second AUG of the PIF1 open reading frame (ORF): pif1-m1 cells have mutant mitochondrial function but wild-type telomeres, whereaspif1-m2 cells have wild-type mitochondria but mutant telomeres (2). As expected, pif1Δ (Fig. 2C, lane 2) and pif1-m2 (lane 4) cells had long telomeres, whereas pif1-m1 cells (lane 3) had telomeres of wild-type length (lanes 1 and 9). The strains with the K264A and K264R alleles also had long telomeres (lanes 5 and 6). That this telomere lengthening was due to the point mutations in the ATP-binding pocket was demonstrated by restoration of wild-type telomere length in K264A and K264R cells carrying a plasmid-borne copy of wild-type PIF1 (lanes 7 and 8).

Western analysis established that cells carrying the K264A and K264R alleles produced stable Pif1p (Fig. 2D). Wild-type cells contained two similarly sized proteins of ∼94 kD, the expected size for Pif1p (lane 1). The mitochondrially defectivepif1-m1 strain expressed only the longer form of Pif1p (lane 3), whereas the telomere-impairedpif1-m2 strain expressed only the faster migrating species (lane 4). These data provide direct evidence for the hypothesis (2) that Pif1p is targeted to different subcellular compartments by making two forms of the protein, one localized to mitochondria and the other, whose translation begins at the second AUG in the ORF, destined for the nucleus. (The mitochondrial Pif1p was shorter than nuclear Pif1p due to proteolysis during import into mitochondria). Because the K264A and K264R alleles produced levels of the nuclear form of Pif1p similar to those in the wild type (lanes 5 and 6), their effects on telomere length must be due to loss of Pif1p adenosine triphosphatase/helicase activity.

If Pif1p is an inhibitor of telomerase, overexpression of Pif1p might result in telomere shortening. To test this possibility, PIF1 was cloned into the multicopy YEpFAT7 plasmid (13). PIF1 cells carrying YEpFAT7-PIF1 had telomeres that were ∼80 bp shorter than cells carrying YEpFAT7 (Fig. 2E). In contrast, wild-type cells carrying YEpFAT7-K264A PIF1 did not show telomere shortening. The fact that overexpression of wild-type but not helicase-deficient Pif1p caused telomere shortening is consistent with the idea that Pif1p acts enzymatically to inhibit telomerase.

Mutations in certain essential replication proteins also cause telomere lengthening when cells are grown at semipermissive temperatures for loss of function alleles (14). Analysis by fluorescence-activated cell sorting (FACS) shows that the replication mutants with lengthened telomeres had an abnormally large number of S-phase cells. Because telomeric DNA is replicated at the end of S phase (15), changes in telomere length in these mutants might be linked to their genome-wide slowing of DNA replication. In contrast, the FACS profile of pif1 cells revealed that they had a similar fraction of S-phase cells as the isogenic wild-type strain (Fig. 3A). In addition, cells lacking Pif1p have wild-type rates of chromosome loss and mitotic recombination (2), whereas mutants in general replication proteins increase both (16). These data suggest that Pif1p has a direct effect on telomere replication.

Figure 3

Pif1p affects telomeric DNA directly. (A) Log phase pif1-m2 and otherwise isogenic wild-type cells were grown in rich medium, fixed, stained with propidium iodide, and analyzed by fluorescence activated cell sorting (FACS). (B) Chromatin was prepared from otherwise isogenic wild-type orpif1Δ cells that had been cross-linked (X-link +) or not (X-link –) with formaldehyde in vivo. Immunoprecipitation was carried out as described in (3) using either protein A–purified preimmune IgG (rabbit IgG), a polyclonal Rap1p antiserum (α-Rap1p) (17), or affinity purified anti-Pif1p polyclonal antibodies (α-Pif1p). The DNA in the immunoprecipitate was PCR-amplified for 28 cycles using telomeric primers or for 32 cycles using ACT1 primers, separated in an agarose gel, and visualized by staining with ethidium bromide. PCR amplification of the input DNA with telomeric primers is also shown. Although Pif1p association with telomeric DNA did not occur in the absence of cross-linking, the amount of telomeric DNA precipitated with anti-Rap1p was not eliminated in non–cross linked cells.

If Pif1p acts directly to inhibit telomere replication, it should associate physically with telomeric DNA. To assess this possibility, chromatin was cross-linked in vivo with formaldehyde, sheared to ∼1000 bp, and then precipitated with either protein A–purified preimmune antibodies [rabbit immunoglobulin G (IgG)] or affinity-purified anti-Pif1p antibodies (α-Pif1p) (Fig. 3B). As a positive control, chromatin was also precipitated with an anti-Rap1p serum (α-Rap1p): Rap1p, the major structural protein at yeast telomeres is constitutively bound to telomeric DNA (17). The cross-links in the immunoprecipitate were reversed, and the DNA in the immunoprecipitate was amplified with polymerase chain reaction (PCR) using primers that amplified a 233-bp portion of the subtelomeric Y′ element that lies 30 bp upstream of the start of the telomeric repeats or, as a negative control, primers for a 131-bp fragment of the ACT1 gene. Both the anti-Rap1p and anti-Pif1p serum specifically precipitated telomeric DNA. Twofold serial dilutions of immunoprecipitates revealed that telomeric DNA was enriched 5.0 ± 2.0 fold (mean ± SD) in the anti-Pif1p precipitate compared to its presence when preimmune antibodies were used or when the anti-Pif1p antibodies were used to precipitate chromatin from pif1Δ cells. As Pif1p was not telomere associated in the absence of cross-linking (X-link –), its association must have occurred in vivo.

When the sequence of the 859–amino acid Pif1p was compared to the translated DNA data base, several highly similar genes were detected including a second S. cerevisiae gene calledRRM3 (Fig. 4A). None of the other 132 yeast ORFs with helicase motifs (18) had detectable similarity to Pif1p by the criterion of a blast search (19). This search also identified Pif1p-like proteins in Candida maltosa, Caenorhabditis elegans, and Drosophila melanogaster (Fig. 4A). We isolated PIF1-like genes from both Schizosaccharomyces pombe (rph1 +,RRM3/PIF1 homolog) and Homo sapiens(hPif1p, human Pif1p) (20). Thus, Pif1p is the prototype member of a helicase subfamily, conserved from yeasts to humans. Pif1 subfamily members encode proteins with 30 to 50% identity in all pair-wise combinations over a region of more than 300 amino acids (Fig. 4A). This degree of relatedness is similar to or greater than that seen within other helicase subfamilies (21). Because all of the Pif1p-like proteins had very high identity to Pif1p, a known DNA helicase, within each of the helicase motif regions (22), the other members of the Pif1p subfamily are also likely to be helicases. Like Pif1p, the Saccharomyces Rrm3p and the S. pombe Rph1p also affect telomeric DNA (19).

Figure 4

PIF1 is the prototype of a subfamily of putative helicases. (A) The predicted sequences of thePIF1-like proteins were compared using the TBLASTN 2.06 program (29). The top line for each pair shows the expectation value, a measure of the probability that the match occurred by chance. The number of amino acids of homology shared between the two proteins and percentage identity within the helicase region is shown below. These numbers were obtained by aligning the helicase region using the MacVector 6.0 (Oxford Molecular) implementation of the ClustalW program (30). (B) Models for Pif1p inhibition of telomerase lengthening. The left telomere of a single chromosome is shown. The solid circle is Pif1p. (Left) A 3′-to-5′ helicase could unwind chromosomes from their ends to create a substrate for a nuclease that inhibits telomerase by destroying its substrate. Because Pif1p is a 5′-to-3′ helicase, this model can not explain the effects of Pif1p. (Middle) Pif1p might dissociate the last Okazaki fragment to generate a ∼100 base TG1-3 tail. This single-strand TG1-3 tail could inhibit telomerase either by forming a higher-order structure, such as a G-quartet (24) or t-loop (25), or by serving as a substrate for an exo-nuclease. The open rectangle represents the 8- to 12-base RNA that primes Okazaki fragments. (Right) Pif1p might inhibit telomerase directly by promoting dissociation of the telomerase RNA-telomeric DNA hybrid that is an intermediate in telomere replication. The structure base paired to the 3′ single-strand tail is telomerase RNA.

We demonstrate that the Saccharomyces Pif1p inhibits telomerase lengthening of telomeric DNA (Fig. 1). Telomere length was inversely proportional to the amount of Pif1p in cells. Loss of Pif1p led to telomere lengthening (Fig. 2C), and overexpression of Pif1p caused telomere shortening (Fig. 2E). The catalytic activity of Pif1p was required for both of these effects (Fig. 2, C and E). Although helicases are required for transcription, RNA processing, and translation, as well as for DNA replication, the association of Pif1p with telomeric DNA in vivo (Fig. 3B) argues strongly that its effects on telomeres are direct.

How might a 5′-to-3′ DNA helicase counter telomerase activity? Because yeast chromosomes have 3′ single-strand tails (23), Pif1p does not have the right polarity to unwind chromosomes from their ends (Fig. 4B, left). However, Pif1p could dissociate the last Okazaki fragment to generate a long single-strand TG1-3 tail (Fig. 4B, middle). Although a priori, G-tails seem more likely to stimulate than inhibit telomerase, TG1-3 tails might fold into a structure that prevents telomerase lengthening (24, 25). Alternatively, the helicase activity of Pif1p might dissociate telomeric DNA from telomerase RNA (Fig. 4B, right). If Pif1p preferentially dissociates RNA-DNA hybrids held together by a very small number of base pairs, it would explain the reduced specificity of telomere addition seen in the absence of Pif1p (2).

Is there an advantage to inhibiting telomerase? Cells lacking the nuclear form of Pif1p had no cell-cycle defect (Fig. 3A) and wild-type chromosome stability (2). Thus, Pif1p is not important in the normal mitotic cell cycle. However, Pif1p might be critical after DNA damage. When a yeast chromosome loses a telomere, the broken chromosome is either lost or a new telomere is gained by homologous recombination (26). Although wild-type cells very rarely add telomeres de novo to broken chromosomes (2, 27), the rate of de novo telomere addition in pif1 cells is elevated as much as 600-fold (2). Adding a telomere to a double-strand break results in aneuploidy for sequences distal to the site of telomere addition. By inhibiting such events, Pif1p could promote genetic stability.

  • * These authors contributed equally to this work.

  • Present address: Genaissance Pharmaceuticals, Five Science Park, New Haven, CT 06511, USA.

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


Stay Connected to Science

Navigate This Article