Roles for RNA in Telomerase Nucleotide and Repeat Addition Processivity

The 3′ Half of Tetrahymena Telomerase RNA Influences Elongation Processivity

Previous studies of recombinant Tetrahymena telomerase have identified specific functions for three sequence elements within the Tetrahymena telomerase RNA (Figure 1B). Located 5′ of the template is the TBE motif, designating its function as both TERT binding element and template 5′ boundary definition element (Lai et al., 2002). Immediately 3′ of the template is a template recognition element motif (TRE) that positions adjacent sequence in the active site for use as template (Miller and Collins, 2002). The TBE, template, and TRE are all near the 5′ end of the RNA and are predicted by phylogenetic comparison to be mostly single stranded (Romero and Blackburn, 1991). Truncation of the RNA 3′ end to position 107 (removing the last 52 nucleotides) deletes stem IV but leaves the TBE, template, and TRE intact. When combined with TERT, this RNA formed an active RNP (Figure 1C, lane 2). However, both the product profile and overall level of telomerase activity were different from when full-length telomerase RNA was used (Figure 1C, lane 1). More extensive 3′ truncation to position 63 also leaves intact all three of the telomerase RNA elements with template-related functions (Figure 1B). When combined with TERT, the 1-63 RNA formed an RNP with detectable telomerase activity (Figure 1C, lane 3) although activity was weaker than with 1-107 RNA. Truncation further, removing the TRE, reduced activity to an undetectable level (Figure 1C, lane 4), despite the fact that this truncated RNA could still bind to TERT (see below). These results are consistent with previous experiments using RNA sequence swaps and deletions in vivo (Bhattacharyya and Blackburn, 1997) and in vitro (Licht and Collins, 1999), which suggested that regions of the telomerase RNA outside of the TBE, template, and TRE play important but uncharacterized roles in enzyme activity.

The patterns of activity observed with the 1-63 and 1-107 RNAs suggested defects in nucleotide and repeat addition processivity. Tetrahymena telomerase typically produces a product profile with prominent products at six-nucleotide intervals (Figure 1C, lane 1). The primer used in the assays described here positions uniquely at the template such that the first nucleotide added is at the template +3 position (Figure 1A). In assays with wild-type telomerase RNA, 64% of the ³²P-dGTP incorporation in the first repeat was +6 in length, whereas in assays with the 1-107 and 1-63 RNAs only 29% and 24% of the ³²P-dGTP incorporation in the first repeat was +6 in length, respectively (Figure 1C). This +3 to +6 product profile was established within the first few minutes of an activity assay and did not change with additional reaction time, indicating that reaction time does not limit development of a product profile (Figure S1). These results suggest that stem IV plays some role in promoting nucleotide addition processivity. Telomerase RNPs with the 1-107 and 1-63 RNAs also did not efficiently elongate a primer by addition of more than one repeat (Figure 1C, lanes 2 and 3). In the standard assay conditions, products longer than +6 must derive from processive repeat addition due to the vast excess of unreacted primer. Therefore, stem IV of telomerase RNA seems to influence repeat addition processivity as well.

Stem IV Stimulates Nucleotide and Repeat Addition Processivity

The changes observed in telomerase activity could indicate that the 3′ end of the telomerase RNA plays specific roles in telomerase active site function. Alternatively, truncation of the 3′ end of the RNA could act in a nonspecific negative manner, for example by altering the global folding of the RNA to make the template less accessible. To differentiate between these alternatives, we took a complementation approach. We tested whether an RNA fragment containing only the 3′ end could restore processivity when added back to RNP assembled with 3′ truncated RNA. We reasoned that if we could add back a defined region of the telomerase RNA in trans to restore a specific aspect of catalytic specificity, we could establish which regions of the RNA are required for telomerase function independent of any potential role in cis-linked template definition.

Telomerase RNP with RNA lacking stem IV (1-107 RNA) exhibited weak nucleotide addition processivity and little if any repeat addition processivity (Figure 2B, lane 1). Addition in trans of a second telomerase RNA fragment containing stem IV restored both nucleotide addition processivity and repeat addition processivity to wild-type levels (Figure 2B, lanes 2 and 3). Telomerase RNP with an RNA that could not form stem IV or stem I (1-102 RNA) also was not repeat addition processive and was less nucleotide addition processive than RNP with 1-107 RNA (Figure 2B, lane 4). Addition of stem IV in trans to RNP with 1-102 RNA also restored substantial nucleotide and repeat addition processivity (Figure 2B, lanes 5 and 6). RNP with 1-102 RNA required the addition of higher concentrations of stem IV in trans to optimally stimulate repeat addition processivity (Figure 2B; data not shown). These results demonstrate that stem IV is required for elongation processivity and can function when cis-linked on the same RNA molecule as the template or when present in the activity assay as a separate RNA molecule.

The Distal Loop of Stem IV Stimulates Both Forms of Processivity

We next investigated whether a particular subregion within stem IV was sufficient to stimulate processivity. In the full-length RNA context, truncation of the 3′ end to position 146 did not alter the pattern of telomerase activity (Figure 2C, lanes 1–3). Further truncation to position 131, removing the distal loop of stem IV, eliminated repeat addition processivity and reduced nucleotide addition processivity (Figure 2C, lane 4). Truncation of the remaining sequence of stem IV had little if any additional impact (Figure 2C, lanes 5 and 6). To test the importance of the loop sequence alone, we created an RNA with a UUCG tetraloop in place of positions 132-138. The tetraloop-substitution RNA produced a similar level and pattern of activity as did removal of the entire stem IV (Figure 2C, lane 7). This finding is consistent with previous results demonstrating that sequence changes in the loop of stem IV affect the level of telomerase activity (Sperger and Cech, 2001).

Because stem IV stimulated nucleotide and repeat addition processivity when added in trans to a template-containing RNA (Figure 2B) and the sequence of the loop of stem IV appeared to be critical (Figure 2C), it seemed possible that just the distal end of stem IV could function in trans to rescue nucleotide and repeat addition processivity. We indeed found that RNA oligonucleotides representing positions 123-146 or 128-142 (Figure 2D, lanes 2–5) complemented defective processivity of RNP with 1-107 RNA as effectively as the entire stem IV (Figure 2D, lanes 6 and 7). We conclude that the distal loop of stem IV mediates stem IV function in nucleotide and repeat addition processivity.

The Stem III Pseudoknot and Stem IV Cooperate to Generate Repeat Addition Processivity

Previously, the distal loop of stem IV was shown to be important for stem III pseudoknot folding, suggesting a possible tertiary interaction between these telomerase RNA elements (Sperger and Cech, 2001). Based on this and other findings, we examined whether the stem III region is also important for nucleotide and/or repeat addition processivity. Truncation of the wild-type telomerase RNA 3′ end to position 63, removing stem III, decreased the level of RNP activity substantially relative to RNP with 1-107 RNA (Figure 1C, lanes 2 and 3). A role of stem III in nucleotide or repeat addition processivity was not possible to discern from this progressive 3′ truncation analysis, however, because 1-107 RNA lacking only stem IV was already compromised for both.

To investigate the role of stem III in the presence of stem IV, we tested the activity of template-containing RNAs with or without stem III when stem IV was present in trans. The low processivity of RNP with 1-102 RNA was rescued by trans complementation with an RNA containing stems III and IV (64-159 RNA) or an RNA containing stem IV alone (103-159 RNA) (Figure 3B, lanes 1–5). Either combination improved both nucleotide and repeat addition processivity. The same stem IV RNAs were also added in trans to RNP with 1-63 RNA. Addition of stem IV alone to 1-63 RNA rescued nucleotide addition processivity but not repeat addition processivity, resulting in a strong +6 product but no products longer than +6 (Figure 3B, lanes 7 and 8). In contrast, addition of the stem III-stem IV fusion (64-159 RNA) to RNP with 1-63 RNA rescued both nucleotide and repeat addition processivity (Figure 3B, lanes 9 and 10). These results demonstrate that stem III is not essential for nucleotide addition processivity but is important in conjunction with stem IV for repeat addition processivity. Addition of stem III alone (64-102 RNA) in trans to RNAs lacking stem III did not have an impact on nucleotide or repeat addition processivity (data not shown). In order to function, stem III must be covalently connected to another region of the RNA.

Addition of 64-159 RNA to telomerase RNP assembled with 1-63 RNA provided the stem III pseudoknot and stem IV but also allowed formation of stem I. Stem I pairing does not appear essential for repeat addition processivity, because addition of stem IV without stem I in trans stimulated repeat addition processivity of RNPs assembled with either 1-107 or 1-102 RNA (Figure 2B). To assess a possible contribution of stem I to repeat addition processivity more directly, we assayed the activity of telomerase RNPs assembled with stem I variant RNAs (Figure S2A). Removing the bottom strand of stem I by 5′ truncation of the RNA or unpairing stem I by mutagenesis of the top or bottom strand decreased overall telomerase activity substantially (Figure S2B). The decrease in activity of the stem I variant RNAs was not simply a consequence of reduced TERT binding (Figure S2C). Although decreased in specific activity, RNPs with stem I variant RNAs retained high nucleotide and repeat addition processivity. These findings confirm that stem I is not necessary for nucleotide or repeat addition processivity, and that a decrease in overall activity does not necessarily impose a processivity defect.

Pairing of Both Helices of the Stem III Pseudoknot Is Important for Repeat Addition Processivity

Tetrahymena telomerase RNA stem III is believed to form a pseudoknot structure (Figure 4A) (ten Dam et al., 1991; Lingner et al., 1994). A pseudoknot structure 3′ of the template is also predicted to be present in vertebrate telomerase RNAs (Chen et al., 2000). Unlike other paired elements of telomerase RNA, the pseudoknot may not be stably or uniquely folded (Bhattacharyya and Blackburn, 1994; Sperger and Cech, 2001; Antal et al., 2002). It seems possible that the stem III region adopts alternate conformations at different times during the catalytic cycle of telomeric repeat synthesis.

To probe the roles of the two stems that compose the pseudoknot, we assayed the activity of RNPs assembled with pseudoknot variant RNAs. Deletion of large portions of the pseudoknot resulted in a decrease in overall activity as noted previously (Licht and Collins, 1999) and also a partial loss of repeat addition processivity (Figure 4B, lanes 2 and 3). The RNA variants AC70-71UG and GU85-86CA both disrupt the same two base pairs within the shorter helix of the pseudoknot (Figure 4A). RNPs assembled with either RNA possessed reduced activity and reduced repeat addition processivity (Figure 4B, lanes 4 and 5). Similarly, disrupting two base pairs in the longer helix of the pseudoknot also decreased activity and repeat addition processivity (Figure 4B, lanes 7 and 8). Compensatory changes that restored base pairing potential increased repeat addition processivity to wild-type levels (Figure 4B, lanes 6 and 9). RNPs with RNAs containing the CC77-78GG substitution had a lower level of activity than RNPs with other stem III variants regardless of the overall base pairing potential and the extent of repeat addition processivity (compare unpaired stem III variants in lanes 7 and 8 and paired stem III variants in lanes 1 and 9). These results suggest that pseudoknot pairing and stem sequence are both important for activity, while the pairing potential but not the sequence of both stems of the pseudoknot is required for optimal repeat addition processivity. Stem III variant RNAs had little if any impact on RNP nucleotide addition processivity, consistent with our finding that the pseudoknot region is dispensable for rescue of the nucleotide addition processivity defect of RNP with 1-63 RNA by stem IV addition in trans (Figure 3).

The Distal Loop of Stem IV Binds TERT Directly

One potential mechanism by which RNA elements could stimulate RNP catalytic function is by interaction with TERT. We examined whether progressively 3′ truncated telomerase RNAs were compromised for binding to TERT by assaying RNA coimmunoprecipitation with TERT. Truncations of the telomerase RNA 3′ end up to position 63 did not result in a dramatic decrease in TERT binding (Figure 5B, lanes 2 and 3; data not shown). Removing additional regions, we found that both the TRE (positions 52 to 63) and the template (positions 43 to 51) were important for optimal TERT binding (Figure 5B, lanes 3–5). A finer 3′ truncation series suggested that within the TRE, positions 59 to 63 were the most critical (Figure S3). An RNA including only positions 1 to 42, containing an intact TBE but lacking both template and TRE, showed little binding to TERT at the sub-micromolar RNA concentration used in the coimmunoprecipitation assay (Figure 5B, lane 5). These results suggest that an interaction occurs between the TRE and TERT. This TERT-TRE interaction must occur with substantially lower affinity than the TERT-TBE interaction, because only TBE substitutions have a dramatic impact on TERT association in the context of full-length RNA (Lai et al., 2002). A weak interaction is consistent with the requirement for a micromolar concentration of RNA oligonucleotide containing TRE and template sequence for function in trans with the rest of the RNA body (Miller and Collins, 2002).

Our demonstration that stem IV is functional when added in trans to an RNP preassembled from TERT and a template-containing RNA (Figure 2) suggested that stem IV might also engage in an interaction with TERT. If so, the affinity provided by association of stem IV and TERT could be masked when the TBE, template, and TRE are all present on the same RNA molecule. To test whether stem IV contains a low-affinity TERT interaction element similar to the TRE, we generated an RNA containing stem IV and the TBE but lacking the TRE and template. To do this, we synthesized a circularly per- muted (cp) RNA with 5′ end at position 107 and 3′ end at position 42 (Figure 5A) with the wild-type 5′ and 3′ ends connected through a linker (Miller and Collins, 2002). The cp107-42 RNA bound to TERT much more strongly than 1-42 RNA lacking stem IV (Figure 5B, compare lanes 5 and 6), indicating that stem IV can increase the association of telomerase RNA with TERT in combination with the TBE motif. In order to delineate the position of the TERT binding element within stem IV, other circularly permuted telomerase RNAs that lacked the template and TRE were tested. Substitutions of the distal end of stem IV with a tetraloop (Figure 5C, lanes 2 and 3) or removal of most of the upper strand of stem IV without deletion of the loop (Figure 5C, lane 5) each slightly but reproducibly decreased TERT interaction. These findings suggest that both the paired region and the loop of stem IV contribute to TERT binding. This implies that a larger region of stem IV is involved in promoting TERT association than participates in establishing nucleotide and repeat addition processivity. Removal of either individual stem III region hairpin or disruption of either pseudoknot stem pairing had no substantial impact on TERT interaction in any context examined (data not shown).

The coimmunoprecipitation data above strongly suggests but does not prove that stem IV binds directly to TERT. An RNA-RNA interaction could instead have recruited stem IV to TERT and could be the mechanism of stem IV loop function in elongation processivity. To determine if the region of stem IV critical for influencing RNP catalytic function contacts TERT directly, an RNA with photoreactive thiouracil substitutions in the stem IV loop was utilized in a TERT crosslinking assay. RNA with thiouracil substitutions was created by ligation of a synthetic oligonucleotide to a circularly permuted RNA lacking the stem IV loop (Figure 6A). RNP assembled with the unsubstituted control cp143-142 RNA produced a pattern of product DNA similar to that of the wild-type enzyme (Figure 6B, lane 1). RNP with 4-thiouracil substitutions in RNA stem IV had wild-type nucleotide addition processivity and at least some repeat addition processivity (Figure 6B, lane 2). Radiolabeled telomerase RNA with thiouracil substitutions at positions 137 and 138 crosslinked to TERT in a UV-dependent manner (Figure 6C, lanes 3 and 4), while RNA without thiouracil substitutions did not crosslink to TERT under the same conditions (Figure 6C, lanes 1 and 2). The crosslink analyzed by SDS-PAGE was verified as a covalent association with TERT using several criteria. Most notably, mock reactions omitting TERT did not produce the crosslink and truncated TERT proteins changed the crosslink mobility (data not shown). This crosslinking analysis supports the coimmunoprecipitation assay in demonstrating that the loop of stem IV makes a direct contact to TERT. A similar crosslinking analysis with RNA bearing a thiouracil substitution at TRE position 57 or 62 demonstrated that TERT also directly contacts the TRE motif (Figure S4).

RNA Functions in the Telomerase Catalytic Cycle

The catalytic cycle of telomeric repeat synthesis by telomerase involves a complex transit of template, substrates, and product through the active site (Figure 7). Product synthesis is initiated by appropriate positioning of a primer 3′ end, template, and dNTP in the active site (Figure 7, top). Correct placement of the stem IV distal loop, involving its association with TERT (Figure 7, right), stimulates processive nucleotide addition to the template 5′ boundary (Figure 7, bottom). Stem IV RNA could have a direct role in the positioning of primer, dNTP, or template. Alternately, stem IV RNA could function as an allosteric activator of TERT, favoring the productive positioning of a template-paired dNTP or primer 3′ end in the active site.

When assayed in vitro, telomerase enzymes from some organisms direct many rounds of repeat synthesis per primer binding event. Recent evidence suggests that at least in mammalian cells, processive repeat addition is a critical feature of telomere length maintenance (Pascolo et al., 2002). To accomplish reiterative repeat addition, a product 3′ end extended by synthesis to the template 5′ end must dissociate from the template, reestablish hybridization with a repositioned template 3′ end, and return to productive engagement in the active site (Figure 7). The stem III pseudoknot pairing could undergo a dynamic structural rearrangement as a part of this process (Figure 7, left). Analysis of the protection of Tetrahymena telomerase RNA residues from chemical modification in vitro revealed that stem IV and TERT influence the folding of the pseudoknot region (Sperger and Cech, 2001). Transient distortion of the pseudoknot induced by template threading through the active site could create a conformational change in the RNP that promotes product dissociation from the template. The template, now free from product hybridization, would return to default positioning. Pseudoknot refolding and formation of a repositioned primer-template hybrid could then again favor proper placement of the product 3′ end in the active site.

Telomerase enzymes from different organisms are surprisingly divergent in TERT and telomerase RNA sequence as well as in the composition of proteins that direct the physiological pathways of RNP assembly, activation, and regulation (Weilbaecher and Lundblad, 1999). The lack of conservation of telomerase RNA primary sequence makes it difficult to assign functions described here for the Tetrahymena telomerase RNA to telomerase RNAs from yeasts or mammals. The human telomerase RNA contains a region 3′ of the template (positions 253 to 318) that like Tetrahymena stem IV can interact directly with TERT to complement the activity defect of a 3′ truncated, template-bearing RNA fragment (Mitchell and Collins, 2000). In addition, all telomerase RNAs with mapped secondary structure are predicted by phylogenetic comparison to possess a pseudoknot 3′ of the template. Pseudoknot pairing is critical for ciliate telomerase RNA function in vivo (Gilley and Blackburn, 1999) and is important for in vivo reconstitution of mammalian telomerase activity (Martin-Rivera and Blasco, 2001). Investigation of the putative functional conservation of nontemplate telomerase RNA motifs will require future studies with telomerase enzymes from many additional species.

A Model for the Evolutionary Specialization of Telomerase RNP

Our present understanding of TERT and telomerase RNA roles in the RNP suggests a possible step-wise process of telomerase evolutionary specialization. Assuming that TERT evolved from an ancestral reverse transcriptase protein (Nakamura and Cech, 1998), its ability to synthesize telomeric repeats may first have required a mechanism to specify what RNA should serve as template. The TRE element could have served this role, correctly positioning any adjacent sequence as intended template (Miller and Collins, 2002). To improve the accuracy of repeat synthesis, a sharply defined 5′ template boundary could have evolved next, conferred by RNA sequences 5′ of the template (Tzfati et al., 2000; Lai et al., 2002). Acquisition of the TBE motif could have simultaneously established the 5′ template boundary and produced a more stable RNP. This telomerase RNP predecessor would have been specialized for accurate synthesis of a specific sequence repeat. The RNA component, initially only serving template-related functions, could next have acquired additional motifs that stimulated RNP nucleotide and repeat addition processivity. Nucleotide and repeat addition processivity are both critical for telomerase elongation efficiency. In addition, these aspects of catalytic function may also be critical for establishing the proper permutation of the most distal telomeric repeat and thereby the structure of the chromosome 3′ terminus.

Expanded Roles for RNA in the Modern RNP World

RNPs formed with non-protein coding (nc) RNAs have been implicated in almost every aspect of cellular function, including DNA replication, chromatin organization, transcription, translation, intracellular and intercellular trafficking, signal transduction, and oncogenic transformation. Bioinformatic methods fall short of readily identifying ncRNA genomic loci, implying that many additional ncRNA transcripts remain to be identified (Eddy, 2001). Our demonstration here that the RNA subunit of telomerase RNP has roles in specifying the catalytic features of DNA synthesis expands the known repertoire of ncRNA function. It is likely that the versatile roles of telomerase RNA will be echoed in future studies of noncoding RNA functions in other RNPs.

Telomerase RNA Synthesis

Template DNAs for 5′ and 3′ telomerase RNA truncations, stem IV variants, and stem III and stem IV RNAs added in trans were generated by PCR from pT7159 DNA. Circularly permuted RNAs were generated by PCR from TER perm A1 DNA, which encodes for two telomerase RNA molecules in tandem with their 5′ and 3′ ends connected by a 5′-TTTTGGATCC-3′ linker sequence (Miller and Collins, 2002). Stem I and stem III truncations and sequence variants were generated by site-specific mutagenesis of pT7159 DNA. Wild-type telomerase RNA and all telomerase RNA derivatives contain a 5′ guanosine leader to increase transcription efficiency.

Telomerase RNAs were synthesized in vitro by T7 transcription and gel-purified. RNA concentration was determined by UV spectrophotometry and/or fluorometry. The position 128-142 and position 123-146 stem IV RNAs were synthesized chemically (Dharmacon, Lafayette, CO).

Telomerase Activity Assay

For each activity assay, 4 to 10 μl of rabbit reticulocyte lysate (RRL) coupled transcription and translation reaction (Promega) expressing HA-tagged TERT was assembled with 4 pmole of telomerase RNA at 30°C for 15 min. The complex was immunopurified on Protein G Sepharose (Amersham Pharmacia) with HA antibody. The activity assay contained as buffer 50 mM Tris-acetate (pH 8.0), 10 mM spermidine, 5 mM β-mercaptoethanol, and 2 mM MgCl₂. To initiate the reaction, 250 μM TTP, 3.5 μM unlabeled dGTP, 1.2 μM ³²P-dGTP (800 Ci/mmol), and 2 μM of the primer (TG)₈TTGGG were added. The reactions were incubated at 30°C for 20 min followed by phenol:chloroform:isoamyl alcohol extraction and ethanol precipitation. The radiolabeled DNA products were resolved by denaturing gel electrophoresis. For the activity assays using telomerase RNA truncations added in trans, TERT was assembled with the template-containing RNA and immunopurified as described above and then mixed with the indicated amount of trans-complementing RNA in the activity assay reaction. Repeat addition processivity was calculated by determining the percentage of product extended past the first repeat on a molar basis (approximately 18% for wild-type enzyme under the reaction conditions used here).

TERT-Telomerase RNA Binding Assay

Ten microliters of RRL expressing HA-tagged TERT was mixed with 0.56 pmole of telomerase RNA, 5 μg of BSA, and 5 μg of tRNA, and then incubated at 30°C for 20 min. Samples were incubated with 15 μl of Protein G Sepharose with HA antibody in 400 μl of T2MG (20 mM Tris-HCl [pH 8.0], 1 mM MgCl₂, and 10% glycerol) with 100 mM NaCl for 4 hr at 4°C. TERT complex bound to the beads was washed with T2MG + 100 mM NaCl, extracted with phenol:chloroform:isoamyl alcohol, and ethanol precipitated. Bound RNA was then separated by denaturing gel electrophoresis and transferred to Hybond N+ (Amersham Pharmacia) for hybridization analysis. The RNA was probed with a ³²P-labeled DNA oligonucleotide complementary to the RNA sequence 5′ of the template. Quantification was performed by PhosphorImager analysis.

Crosslinking of TERT-Telomerase RNA Complexes

The 4-thiouracil substituted telomerase RNAs were made by enzymatic ligation of a longer T7 RNA polymerase transcript and a shorter, chemically synthesized oligonucleotide using standard methods (Moore and Query, 2000). Full-length, ³²P-labeled ligation product was gel-purified from the ligation reaction. HA-tagged TERT protein was synthesized in RRL and assembled with telomerase RNA at 30°C. The complex was immunopurified on Protein G Sepharose with HA antibody in T2MG with no salt. Crosslinking by UV irradiation was done at approximately 366 nm for 15 min. Sample buffer was then added followed by SDS-PAGE.