Telomerase can act as a template- and RNA-independent terminal transferase

Yeast Strains and Plasmids. The pmr1-Δ strain and the isogenic wild-type strain (BY4741) were purchased from Open Biosystems (Huntsville, AL), and the identity of the mutant was confirmed by PCR. The tlc1-Δ strain was constructed in the W303a background by using a PCR-generated disruption cassette that precisely replaces the TLC1 gene with a LEU2 marker. The strain was subsequently transformed with pSE-Est2-C874, which carries a protein A-tagged Est2p and allowed Est2p to be purified on IgG-Sepharose.

Analysis of Yeast Telomeres and Telomerase. The protocols for yeast telomere length analysis, extract preparation, and telomerase activity assays have all been described (25, 26).

Analysis of Human Telomerase. The method for human TERT (hTERT) expression in reticulocyte lysate (RRL) has been described (27). RRL samples were treated with RNase for 30 min before being tested in direct primer extension assays using 2.5 μM biotinylated (G₂T₂AG)₃ primer. Earlier analysis indicates that this RNase treatment is sufficient to abolish all telomerase activity in a conventional assay that contains only magnesium.

We first examined the effect of different divalent cations on partially purified yeast telomerase (Saccharomyces cerevisiae) in primer extension assays, and found that quite a few metals including Ca²⁺, Cd²⁺, Zn²⁺, Cu²⁺, Ni²⁺, and Co²⁺ can significantly repress telomerase activity in vitro at mM concentrations in the presence of 2.5 mM Mg²⁺ (Fig. 5, which is published as supporting information on the PNAS web site). Interestingly, the effect of at least one metal ion appears to be species-specific: Ca²⁺ can severely inhibit telomerase from S. cerevisiae, but had at most a minor effect on the closely related enzyme from Candida albicans (Fig. 5). Thus, not all metals are likely to interact with conserved telomerase residues.

Manganese Altered the Fidelity of Yeast Telomerase. We next assessed the impact of adding Mn²⁺, whose mutagenic effects on other polymerases are most extensively characterized (21, 28). Interestingly, in the presence of the standard combination of 0.2 μM dTTP and 50 μM dGTP, which enables yeast telomerase to add ≈15 nucleotides (29) to the telomeric primer, the inclusion of 3 mM Mn²⁺ resulted in a slight reduction in total DNA synthesis and a mild reduction in the processivity of telomerase (Fig. 1A, lanes 1–4). Remarkably, when either dGTP or dTTP alone was included in the in vitro reaction, telomerase extended the primer by a single nucleotide in the absence of Mn²⁺ (as would be expected based on the primer/template alignment and the template sequence) (Fig. 1 A, lanes 5 and 11), but by up to almost 20 nucleotides in the presence of Mn²⁺ (lanes 6, 7, 12, and 13). Moreover, with the combination of labeled dTTP and unlabeled ddGTP, telomerase faithfully executed the addition of two nucleotides in the absence of Mn²⁺, but became much less active in the presence of Mn²⁺ (lanes 8–10). All of these observations are consistent with a significant loss of fidelity by telomerase in the presence of Mn²⁺. In particular, telomerase appears to be more prone to slippage synthesis or less reliant on template-nucleotide base-pairing during polymerization. The inhibitory effect of ddGTP in the presence of Mn²⁺, for example, may be attributed to competition of unlabeled ddGTP with labeled TTP at the nucleotide binding site that would, under high-fidelity conditions, only accommodate the latter nucleotide.

Telomerase Can Act as a TT. To address the possibility that the requirement for proper base-pairing between template and nucleotide may be substantially relaxed in the presence of Mn²⁺, we pretreated telomerase with RNase A and subjected the resulting preparation to primer extension analysis. As expected, in the presence of Mg²⁺ alone, the RNase pretreatment completely abolished the extension reaction (Fig. 2A, lanes 1 and 2). In contrast, in the presence of Mn²⁺ alone or Mn²⁺ and Mg²⁺, the signals in the dGTP only reactions were reduced by only 2-fold after RNase pretreatment (lanes 3–6). These results indicate that telomerase is at least partially template-independent in the presence of Mn²⁺, and prompted us to examine the enzyme's ability to incorporate labeled dATP and dCTP, two nucleotides that are not encoded by the RNA template. As shown in Fig. 2B, the ability of telomerase to add dCTP and dATP to the starting primer was indeed greatly enhanced by the addition of Mn²⁺ (lanes 1–3 and 5–7). Moreover, the activity was even higher with enzyme that had been pretreated with RNase A (lanes 4 and 8). Thus, even though the RNA template was dispensable for nucleotide addition in the presence of Mn²⁺, it appears still to act as a fidelity factor, mitigating against the incorporation of noncomplementary nucleotides. Titration studies indicate that the dCTP-incorporating activity of telomerase was minimal below 0.3 mM Mn²⁺ but became greatly enhanced at higher concentrations (data not shown). Pretreatment of the telomerase preparation with DNase I had no effect on its RNase-independent, Mn²⁺-dependent activity, suggesting that telomerase did not use a contaminating DNA as the template (Fig. 6, which is published as supporting information on the PNAS web site). In contrast to Mn²⁺, none of the other metal ions mentioned earlier were able to support the incorporation of dCTP (data not shown). Taken together, these results support the notion that, in the presence of Mn²⁺, yeast telomerase can switch to a template-independent mode of DNA synthesis, acting as a TT.

The TT Activity Is Mediated by TERT. The apparent TT activity of yeast telomerase cannot be attributed to a contaminating enzyme. First, aphidicolin, an inhibitor of the major DNA polymerases in yeast, had no effect on the terminal transferase activity (Fig. 3A). Second, the TT activity was completely absent from a telomerase preparation in which two of the three catalytic aspartic acids of Est2p (30) have been mutated to alanine (Fig. 3B). Another Est2p mutant with a substitution near the RT nucleotide-binding pocket (named R450K; ref. 31) likewise had greatly diminished TT activity (data not shown), suggesting that the TT activity is mediated by the active site of RT. We also addressed the possibility that the telomerase RNA component in our preparation may not have been completely removed by RNase treatment, and that some residual RNA moiety may be important for the TT activity. Telomerase was isolated from both a wild-type and a tlc1-Δ strain (in which the template RNA gene is deleted), and assayed in parallel. As shown in Fig. 3C, in the TT assays, the two enzyme preparations exhibited comparable levels of activity, implying that the entire telomerase RNA is dispensable for the activity. The TT activity of telomerase is apparently conserved in evolution. For instance, the telomerase from Candida albicans also becomes template-independent in the presence of Mn²⁺ (Fig. 7, which is published as supporting information on the PNAS web site). Furthermore, the human telomerase catalytic protein hTERT that was expressed in the absence of telomerase RNA in reticulocyte lysates can also mediate the addition of a nontelomeric nucleotide (dC) in a Mn²⁺-dependent fashion (Fig. 3D). Like the yeast TT activity, the human activity requires at least one of the catalytic Asp residues (D868).

The TT Activity Exhibits Preference for Telomere-Like Substrates. Telomerase has been shown to exhibit a preference for elongating GT-rich telomere-like oligonucleotides. Both the RNA template and some putative DNA-binding domain(s) of the telomerase protein are thought to contribute to this preference (32–35). As shown in Fig. 4, even in the TT mode of DNA synthesis, telomerase manifested a strong preference for extending GT-rich oligonucleotides (Fig. 4 A and B, lanes 4–12). This result further rules out the possibility that a non-sequence-specific contaminating nucleotidyl transferase is responsible for the observed activity. It also indicates that the DNA-binding domain(s) of yeast telomerase has an intrinsic, RNA-independent preference for GT-rich sequences. The protein-mediated sequence recognition appears to be directed toward the 5′ end of the DNA substrate: primers containing GT-rich sequences on the 5′ end consistently supported higher levels of DNA synthesis in the TT assay than comparable primers with GT-rich sequences on the 3′ end (Fig. 4 A, C, and D). As expected, longer GT-rich tracts supported higher levels of activity. These findings provide unequivocal support for the longstanding notion that sequence-dependent recognition of DNA by telomerase is at least bipartite, with the telomerase RNA and protein mediating binding to the 3′ and 5′ end of the DNA, respectively (14). In contrast to Saccharomyces, the Candida telomerase, although capable of terminal transferase activity, did not exhibit a preference for GT-rich primers, consistent with the atypical composition (non GT-rich) of the Candida telomere repeat (Fig. 8, which is published as supporting information on the PNAS web site). Thus, the sequence specificity of the telomerase protein is likely to have evolved in relation to the cognate telomere repeat.

Elevated Intracellular Manganese Induced Telomere Shortening. To determine whether Mn²⁺ can influence telomerase function in vivo, we analyzed the telomeres of a yeast mutant (named pmr1-Δ) with a defect in Mn²⁺ transport, and hence elevated intracellular Mn²⁺ (36). When the strain was placed in 0.25 or 0.5 mM Mn²⁺, its growth was impaired (data not shown) and its telomeres were found to be shorter by ≈50–100 bp than those of the wild-type strain (Fig. 1B), suggesting that Mn²⁺ can affect telomerase function in vivo. Telomere shortening was observed in a second pmr1-Δ clone with a different strain background (data not shown). However, the extent of telomere shortening is relatively small and unlikely to be responsible for the growth impairment (26). In addition, because of the dependence of many cellular enzymes on metal ions, we evidently cannot rule out the possibility that the shortening was due to some other effects of Mn²⁺. The intracellular free Mn²⁺ concentration in yeast has been estimated to be in the low μM range, even for the pmr1-Δ strain (36). Thus, it seems unlikely, given the aforementioned titration study, that yeast telomerase will be able to act as a robust terminal transferase in vivo, even in the mutant. Nevertheless, a low level of activity may be sufficient to induce physiologic consequences.

Aspects of our findings are far from unprecedented. For example, ciliate telomerases have been reported to perform iterative dG addition in the presence of this single nucleotide (although in these studies the RNA requirements were not reported) (37, 38). Furthermore, a variety of DNA and RNA polymerases, including reverse transcriptases, are known to mediate template-independent nucleotide addition under some conditions (39–41). Nevertheless, as discussed below, our discovery of telomerase-mediated terminal transferase activity has both biochemical and physiologic implications.

From the biochemical perspective, our finding addresses a longstanding question in telomerase enzymology, i.e., the role of the RNA in catalysis. It has been argued, based on a substantial body of evidence, that nontemplate regions of the telomerase RNA mediate important catalytic functions (7, 42–45). Our findings indicate that the entire RNA subunit is dispensable for the covalent chemistry of telomerase (although it is entirely plausible that the RNA can modulate protein conformation and/or directly contribute to substrate recognition). Furthermore, our TT assay offers a simple and sensitive method of analyzing the protein-dependent recognition of telomeric DNA free of the confounding effect of telomerase RNA.

From the physiologic perspective, our results suggest a potential mechanism for certain unexplained experimental findings and a potential antineoplastic strategy. There is increasing evidence, for example, that overexpression of mammalian TERT can have physiologic consequences (e.g., inducing senescence and modulating tumorigenesis) that cannot be rationalized by its effect on telomere lengths (16–19). We suggest that, because telomerase RNA is likely to be an important fidelity factor, overexpression of the telomerase protein alone may unmask its normally cryptic TT activity, resulting in the synthesis of nontelomeric sequences and potentially detrimental effects on cell physiology. However, it is not known whether the intracellular concentration of Mn²⁺ in mammalian cells can reach a level to support the TT activity of telomerase in normal or pathologic conditions. On the other hand, our results do indicate an intrinsic ability of hTERT to act as a terminal transferase. Thus, it may be possible to identify small molecules that trigger this activity even in the absence of high Mn²⁺ concentrations. The introduction of abnormal sequences at chromosomal ends through the expression of mutant telomerase RNAs has been shown to induce cell death, and is proposed to be a valid strategy for cancer therapy (24). Our finding suggests an alternative way of introducing abnormal sequences, i.e., by switching on the terminal transferase activity of telomerase. Clearly, further studies will be necessary to determine the physiologic significance and medical potential of this normally cryptic activity of telomerase.

As a final remark, we note that telomerase was initially named “telomere terminal transferase” (46, 47). Our results argue that this original name is prescient in anticipating an inherent attribute of a versatile enzyme.

Author contributions: N.F.L., T.J.M., and C.A. designed research; N.F.L., D.B., T.J.M., B.A., and S.L. performed research; N.F.L., D.B., T.J.M., C.A., B.A., and S.L. analyzed data; and N.F.L. wrote the paper.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: RT, reverse transcriptase; TT, terminal transferase.

We thank Drs. Stewart Shuman and Jerry Hurwitz for comments on the manuscript and Jerry Hurwitz for the provision of polymerase α. This work was supported by National Institutes of Health Grant GM62631 and the Irma T. Hirschl/Monique Weill–Caulier Research Award (to N.F.L.), Canadian Institutes of Health Research (CIHR) Grant MOP68844 and the Fond de la Recherche en Sante du Quebec Chercheur-Boursier Award (to C.A.), and the CIHR doctoral research scholarship (to T.J.M.). The Department of Microbiology and Immunology at Weill Cornell Medical College gratefully acknowledges the support of the William Randolph Hearst Foundation.