Template Boundary in a Yeast Telomerase Specified by RNA Structure

See allHide authors and affiliations

Science  05 May 2000:
Vol. 288, Issue 5467, pp. 863-867
DOI: 10.1126/science.288.5467.863


The telomerase ribonucleoprotein has a phylogenetically divergent RNA subunit, which contains a short template for telomeric DNA synthesis. To understand how telomerase RNA participates in mechanistic aspects of telomere synthesis, we studied a conserved secondary structure adjacent to the template. Disruption of this structure caused DNA synthesis to proceed beyond the normal template boundary, resulting in altered telomere sequences, telomere shortening, and cellular growth defects. Compensatory mutations restored normal telomerase function. Thus, the RNA structure, rather than its sequence, specifies the template boundary. This study reveals a specific function for an RNA structure in the enzymatic action of telomerase.

Telomerase, a ribonucleoprotein reverse transcriptase (RT), replenishes telomeric DNA that would otherwise be lost with each round of eukaryotic DNA replication (1). The telomerase complex contains an RNA subunit (TER), a catalytic RT protein (TERT), and several additional protein components (2). Telomerase is activated in most human cancers, and its ectopic expression can greatly extend the life-span of normal human cells in culture (3).

Telomerase RNAs are extremely divergent in sequence and vary in length from 146 nucleotides (nt) in the ciliate Tetrahymena paravorax (4) to 1544 nt in the budding yeastCandida albicans (5). Unlike other RTs, which perform extensive genome copying, telomerase copies only a small portion (termed the “template”) of an intrinsic RNA moiety (6). This feature allows telomerase to synthesize onto telomeres a species-specific, 5- to 26-base-long repeated sequence (7). How telomerase specifies its template boundaries (where DNA synthesis initiates and where it ends on the TER sequence) is not understood.

Nontemplate regions have been previously shown to be required for telomerase activity (8, 9) and ribonucleoprotein (RNP) assembly (9, 10). To further investigate the participation of telomerase RNA in the enzymatic function of telomerase, we searched for conserved sequences and structural elements in budding yeast telomerase RNAs. We cloned and analyzed TER genes from four Kluyveromyces species closely related to K. lactis (11). The mature RNAs ranged in length from 930 nt in K. aestuarii to 1320 nt in K. dobzhanskii. Sequence identity between any given pair of genes ranged from insignificant to about 70% overall identity. The computer programmfold (12) predicted extensive secondary structures for these RNA sequences, including a common feature shared by all five TERs: base pairing of the sequence immediately upstream of the template (pairing element B) (Fig. 1A) with a sequence 200 to 350 nt further upstream (pairing element A), located near the 5′ end of the RNA. The region between the pairing elements (indicated by the dashed line in Fig. 1A and the dashed loop in Fig. 1B) was shown previously to be dispensable in K. lactis (9). The proximity of this conserved putative pairing region to the 5′ end of the template led us to hypothesize that its function is to limit DNA synthesis, thereby defining the downstream boundary of the template.

Figure 1

Prediction of a pairing region in budding yeast telomerase RNAs. (A) A linear map of K. lactisTER illustrating the location of the template and the pairing elements A and B. (B) A pairing region, predicted by the computer program mfold (12), is located immediately upstream of the telomerase RNA template in five Kluyveromycesspecies. For each species, the horizontal gray line denotes the maximal putative template sequence, including the short repeated sequence at both ends (thick gray lines) thought to be required for realignment and synthesis of multiple telomeric repeats. Arrows indicate the direction of DNA polymerization along the template (solid arrow for the known template of K. lactis; dashed arrows for the putative templates of the newly cloned genes). Boxed nucleotides, sequence variation from the K. lactis TER gene; shaded boxed nucleotides, variations that retain the pairing potential. (C) Substitution mutations designed to test the pairing hypothesis in K. lactis. Shaded boxes, wild-type sequences that are substituted in the mutations. A and B are the pairing elements indicated in (A). (D) Effects of mutations within the pairing region, as predicted by the mfold program (12). Thick lines represent mutated sequences.

To test this hypothesis, we constructed a series of mutations in the putative pairing region of the K. lactis TER gene (Fig. 1, C and D). We replaced the wild-type TER gene in K. lactiswith the mutant genes by a vector-shuffling system described previously (9) and analyzed their effects in vitro and in vivo. In each of four disruption mutations (D1, D1′, D2, and D2′) (Fig. 1, C and D), a trinucleotide sequence within either strand of the pairing region was substituted with its complementary sequence. The D1 and D1′ mutations, in pairing elements A and B, respectively, were designed to unwind the first 3 base pairs (bp) of the putative pairing region, adjacent to the template. The D2 and D2′ mutations targeted 3 bp in the middle of pairing elements A and B, respectively, and were predicted to cause a more extensive disruption of the pairing. Pairing potential was restored in the double mutants D1/D1′ and D2/D2′. In a full replacement mutant, R1, 10 nt on each strand of the pairing region were replaced with unrelated sequences that maintained the original base composition and pairing potential (Fig. 1, C and D).

We next studied the effects of the pairing mutations on telomerase function, by assaying partially purified extracts from wild-type or mutant cells for telomerase activity in vitro (13). The characteristic pattern of elongation products synthesized by telomerase with a wild-type pairing region (Fig. 2A, lane 2) was described previously (14). It includes a faint band, corresponding to the longest product that can result from one round of synthesis along the maximal potential template (open arrowhead), and stronger bands, which are 1 to 3 nt shorter. The activities from each of the various cell extracts were ribonuclease A–sensitive (Fig. 2A, lane 1) (15), a hallmark of telomerase activity.

Figure 2

Base-pairing disruption results in polymerization beyond the normal template boundary. Kluyveromyces lactis telomerase activity was assayed in vitro (13). All telomerases assayed contained a silent Bcl I mutation that is used to mark their action in vivo (see Fig. 3). (A) Telomerase reactions were incubated in the presence or absence of dATP, as indicated below the lanes. The predicted pairing region structures are illustrated above the lanes for each mutant. In lane 1 (R+), cell extract was pretreated with ribonuclease A (14). Open arrowheads, products ending at the last position of a template with a normal boundary; double arrowhead, a product of the D2 mutant enzyme observed upon omitting dATP; closed arrowheads, read-through products. (B) Reactions were incubated in the presence or absence of dCTP, as indicated below the lanes. V-shaped arrowheads highlight read-through products that disappear upon omitting dCTP from the reaction. Schematics on the sides of the panels illustrate polymerization along the template (thick gray line) with vertical arrows ending at the positions where the longest corresponding wild-type (A) or mutant (B) telomerase products were detected. Thick black lines, mutated TER sequences. The schematics show the nucleotides predicted to be incorporated at each position, starting at the first nucleotide added to the primer terminus. Read-through nucleotides are shown in bold. Nucleotides resulting from the predicted incorporation of the TER mutations are boxed.

Strikingly, all the pairing-disruption mutants synthesized longer products than wild-type telomerase (Fig. 2A, compare lanes 4, 6, and 12 to lane 2; see also Fig. 2B, lanes 5 and 7). The D1 and D1′ mutations each resulted in detectable read-through of 2 nt (Fig. 2A, lanes 4 and 6; Fig. 2B, lanes 3 and 1). Because extracts prepared from the D2 and D2′ mutants had reduced telomerase activity (Fig. 2A, lanes 10 to 13), we integrated the D2 and D2′ mutant TER genes into the genome by replacing the endogenous gene (16). The resulting strains, iD2 and iD2′, exhibited stronger telomerase activity in vitro, generating detectable products with up to seven and four read-through nucleotides (Fig. 2B, lanes 7 and 5), respectively. For all strains tested, the polymerization activity observed was specific to the telomerase template sequence, as indicated by experiments in which deoxyadenylate triphosphate (dATP) was omitted. The resulting –dATP products were of the length expected if telomerase elongated the correctly aligned primer and stopped just before the uridine that is the last nucleotide in the maximal possible template (Fig. 2A, –dATP lanes). The read-through products were correctly copied from the TER sequence 5′ of the template, as indicated by the pattern of bands when deoxycytidine triphosphate (dCTP) was omitted from the reactions. Telomerase activity of the D1′ but not the D1 mutant was limited by the omission of dCTP, as expected if the G-containing D1′ mutation was to be copied (Fig. 2B, see disappearance of a band in lane 2 compared with that indicated by V-shaped arrowhead in lane 1). The difference in mobility between the D1 and D1′ read-through products (Fig. 2B, lanes 1 and 3) was another indication that these mutant enzymes copied different nucleotides. Synthesis by the D2 mutant was also limited by the omission of dCTP, as expected from copying the wild-type sequence adjacent to the template (Fig. 2B, see disappearance of bands in lane 6 compared with those indicated by V-shaped arrowheads in lane 7).

The presence of sequence covariation in the pairing elements of the TER genes (Fig. 1B) suggested that base pairing, rather than a specific sequence, is required to specify the template boundary. Two different experiments support this prediction. First, the D1/D1′ and D2/D2′ compensatory mutations re-established a normal, wild-type–like boundary (Fig. 2A, lanes 8 and 14). Second, R1 telomerase, with complementary but scrambled pairing sequences, also retained a normal boundary (Fig. 2A, lane 16). Together, the results showed that disrupting the secondary structure causes template read-through, and that restoring base pairing with complementary mutations re-establishes the normal boundary.

We also tested the effects of these mutations in vivo. To distinguish telomeric repeats added by the mutant telomerases from repeats synthesized previously by the wild-type enzyme, we used an additional single-nucleotide mutation producing a Bcl I restriction site within the template sequence. This mutation is phenotypically silent but results in the incorporation of telomeric repeats containing a Bcl I restriction site onto telomeres (9), thus marking the action of the mutant enzyme. All the pairing-mutant telomerases were active in vivo, which is evident by the incorporation of Bcl I site containing telomeric repeats. These repeats were detected by differential hybridization of a Bcl I–specific probe to Southern blotted genomic DNA cut with Eco RI restriction endonuclease (Fig. 3A). Secondary digestion with Bcl I endonuclease shortened the Eco RI telomeric restriction fragments, as detected by a wild-type probe (Fig. 3B), and eliminated the Bcl I–specific hybridization signal (Fig. 3A). The D1 and D1′ mutations, as well as the D1/D1′ double mutation, did not appear to affect telomere length or colony morphology (Fig. 3A). However, the D2 and D2′ telomeres (Fig. 3, A and B) and the iD2 and iD2′ telomeres (17) were considerably shortened, containing significantly fewer total telomeric repeats than wild-type telomeres, as indicated by the reduced hybridization intensity of the telomeric bands. This shortening correlated with rough colony appearance and longer population doubling times than that of the wild-type strain (15, 18). In addition, two telomeric restriction fragments disappeared in each of the D2 and D2′ mutants (arrowheads,Fig. 3B) most likely through recombination in the subtelomeric region, which has been shown to be associated with impaired telomere maintenance (18).

Figure 3

Base-pairing disruption causes impaired telomere maintenance in vivo. Genomic DNA was prepared from K. lactis strains (deleted for the chromosomal TER gene and carrying the different TER alleles on a plasmid) at their 15th passage. The control strains carry on a plasmid a wild-type TER gene (WT), a Bcl I–marked (WT-Bcl I) TER gene, or no insert (Δ). DNA was digested with Eco RI restriction endonuclease (– lanes) or double-digested with Eco RI and Bcl I (+ lanes) and was then separated on 1% agarose gel and vacuum blotted. Blots were hybridized (9) with a Bcl I–specific oligonucleotide probe (A) and then with a wild-type telomeric sequence probe (B). Arrowheads indicate the expected location of telomeric restriction fragments that disappeared during cell divisions. The predicted pairing region structures are illustrated above the lanes in (A) for each mutant. Yeast colonies of the corresponding strains were examined under the microscope. Strains exhibiting the smooth wild-type colony phenotype are labeled WT; mutants exhibiting rough colony phenotype are labeled M (see text). The observed phenotype remained unchanged from the third passage (60 to 75 cell divisions) to the 15th passage (300 to 375 cell divisions). (C) Incorporation of Bcl I–marked repeats and read-through sequences onto telomeres in vivo. Sequence examples of telomere clones (19) are shown. Black, telomerase RNA sequences; blue, wild-type repeats; green, Bcl I–marked repeats; red, read-through sequences. Boxed, the Bcl I and the D2′ mutations in the RNA and the corresponding incorporated DNA sequences.

The D2/D2′ and R1 mutants exhibited normal colony morphology and wild-type telomere length (Fig. 3A), consistent with their normal template boundary observed in vitro (Fig. 2, lanes 14 and 16). The lack of any apparent effect of the 20-base R1 substitution is striking in light of the severe defects in telomere maintenance and cell growth caused by the 3-base D2 or D2′ substitutions. Thus, base pairing at this region, rather than sequence, is important for telomere maintenance in vivo, supporting the results obtained in vitro.

To test whether nontelomeric read-through sequences were incorporated onto telomeres in vivo, we cloned and sequenced telomeres from each of the TER mutants with a new method designed to preserve any 3′ overhang of the telomere (19). At least 10 telomere clones were sequenced for each strain. In these clones, the average telomere length and the extent of Bcl I incorporation (15) matched the observations by Southern analyses (Fig. 3B), indicating that the polymerase chain reaction (PCR) products were representative of the telomere population.

In two of the D1 telomere clones (Fig. 3C) composed of the expected Bcl I–marked telomeric repeats (in green) distal to wild-type repeats (in blue), an extra A residue (in red) was found, embedded between two Bcl I repeats. Such a single additional A residue was unique to the D1 clones and was not found in any of the other (>130) wild-type and mutant telomere clones. The position of this A residue is consistent with DNA synthesis proceeding 1 nt beyond the normal template boundary in vivo. In vitro read-through occurred in 44% of the product molecules (see arrowheads, Fig. 2A, lane 4). The paucity of read-through sequences in the telomere clones may reflect less reading through in vivo, or removal of nontelomeric nucleotides (20). The D2 and D2′ telomere clones contained up to 11 read-through nucleotides incorporated onto telomeric termini (in red,Fig. 3C). In each case, the additional sequence was that predicted from copying by the mutant enzyme beyond the normal template boundary, including the D2′ mutation (boxed nucleotides). In contrast to the D1, D2, and D2′ clones, no read-through sequences were detected in any of the D1′ (21), D1/D1′, D2/D2′, and R1 clones (at least 10 different clones of each). Together, these results indicate that disrupting the pairing region causes template read-through not only in vitro, but also in vivo.

In summary, we have demonstrated that a phylogenetically conserved, long-range base-pairing interaction adjacent to the template in a yeast telomerase RNA specifies one boundary of the telomerase template, thus determining the end of the telomeric repeat synthesized. It is not known whether this mechanism of demarcating the template boundary is used in other telomerases. In ciliate TERs, a conserved sequence immediately upstream of the template has been proposed to play a role in specifying the template boundary (22). In T. thermophila, mutations at this region resulted in alteration of the template boundary, as revealed by an in vitro activity reconstitution assay (23). In Saccharomyces cerevisiae, a trinucleotide substitution introduced adjacent to the template caused telomerase to copy 1 nt beyond the normal boundary in vitro (24). Secondary structure predictions reveal a putative double-stranded element adjacent to the template of telomerase RNA in several other species, including S. cerevisiae and human (15). Whether these putative structures function similarly to the one described here is yet to be explored.

Local RNA secondary structures have been shown to cause pausing in DNA synthesis by retroviral RTs (25). Here, we report the utilization of a long-range RNA-RNA base-pairing interaction as a barrier for reverse transcription, specifying the telomerase template boundary. Hence, the special feature of telomerase—precise limitation of polymerization to the template—is achieved by an RNA-directed mechanism of the RNP enzyme and is not an inherent property of the telomerase RT protein. Such a direct function for TER in the enzymatic action of telomerase is consistent with an evolutionary scheme in which RNA enzymes, in an archaic RNA world, acquired protein components evolving into RNP enzymes (26). The RNA components then gradually lost their functional roles in catalysis and were subsequently dispensable. Telomerase RNP may represent an evolutionary relic—an intermediate in the transition from RNA replicases to protein reverse transcriptases (27). Further study of functional elements in TER will shed more light on telomerase evolution and function.

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


View Abstract

Navigate This Article