Introduction

Telomeres are DNA–protein complexes that cap the end of linear chromosomes and enhance their genetic stability by protecting the DNA ends from recombination, degradation and fusion with other chromosomes (Blasco et al, 1997; Lee et al, 1998; Lingner and Cech, 1998; de Lange and Jacks, 1999). The natural loss of telomeres that occurs during normal cell division is counteracted by the de novo addition of telomeric repeats carried out by telomerase. This enzyme has two core components: a protein (TERT) and an RNA (TER) (Blasco et al, 1999). TERT is the catalytic subunit and shares homology to viral reverse transcriptases (Lingner et al, 1997), whereas TER contains the template (Feng et al, 1995). The holoenzyme contains other accessory proteins (Mitchell et al, 1999; Seto et al, 1999; Witkin and Collins, 2004; O'Connor et al, 2005; Prathapam et al, 2005), but they are less well conserved than TERT and dispensable for basal enzymatic activity. However, many of these proteins are required for activity in vivo as they control ribonucleoprotein assembly and maturation or its subnuclear localization (Mitchell et al, 1999; Wang and Meier, 2004).

TERT protein is highly conserved, whereas TER varies in size, secondary structure and sequence among different organisms (Chen et al, 2000; Zappulla et al, 2005). TER provides the template sequence for reverse transcription as well as a landing pad to organize ribonucleoprotein assembly during RNP maturation and in the fully catalytic mature enzyme (Zappulla and Cech, 2004). Despite variations in sequence and length, some secondary structure elements appear to share common functional roles in ciliates, yeast and vertebrates, suggesting that the basic biochemical activities of telomerase RNA are well conserved (Chen et al, 2000; Zappulla et al, 2005).

All TERs contain three major domains with shared functions that are most clearly defined in vertebrates (Chen et al, 2000). The 5'-end of vertebrate and ciliate TERs contains the template and a pseudoknot that plays key roles in enzymatic activity (Figure 1). The 3'-end of vertebrate telomerase RNAs contains a highly conserved elongated hairpin (CR4–CR5 domain). It is essential for telomerase catalytic activity and defines a second site of interaction with TERT protein (Tesmer et al, 1999; Mitchell and Collins, 2000; Chen et al, 2002). The vertebrate CR4–CR5 domain has a possible functional counterpart in the ciliate stem-loop IV (Mason et al, 2003), the subject of the present investigation. Tetrahymena thermophila provides a well-established model system for understanding how telomerase functions. Its RNA subunit contains four conserved double helices (I–IV) (Figure 1A) (Bhattacharyya and Blackburn, 1994; Lingner et al, 1994). Stem I bridges the 5'- and 3'-ends of the molecule, whereas stems II and IIIa/IIIb are necessary for TERT binding to telomerase RNA, for telomerase activity in vitro and for ribonucleoprotein (RNP) assembly in vivo. Stem-loop IV functions in the correct folding of the stem III pseudoknot and makes a major contribution to telomerase enzyme activity (Sperger and Cech, 2001) and to repeat addition processivity (Lai et al, 2003). Deletion of the complete stem-loop, mutations of conserved residues in the apical loop and deletion of the GA bulge, all dramatically reduce telomerase activity (Autexier and Greider, 1998; Sperger and Cech, 2001). Remarkably, activity can be restored by the addition of stem-loop IV in trans (Mason et al, 2003). The apical loop of stem IV may interact directly with the N-terminus of TERT protein (O'Connor et al, 2005), whereas the region more proximal to the template in secondary structure (Figure 1A), together with the GA bulge, interact with protein p65 (Witkin and Collins, 2004; O'Connor et al, 2005; Prathapam et al, 2005; O'Connor and Collins, 2006). The association of p65 with TER enhances the binding affinity of TERT to TER (Prathapam et al, 2005; O'Connor and Collins, 2006).

The essential roles of T. thermophila stem-loop IV in telomerase enzymatic activity prompted us to study its structure. The NMR structure presented here shows a highly structured distal stem-loop linked to a flexible template-proximal region by a conserved GA bulge that severely kinks the entire RNA. Mutations designed to investigate the structure–function relationship of stem-loop IV identify nucleotides within the apical loop that severely compromise catalytic activity but do not affect recruitment of TERT protein. Based on these results, we propose that the kink induced by the GA bulge, and the flexibility of the lower part of the structure, affect catalytic activity by positioning the apical loop during catalysis.

Results

Structure determination

Initial studies of the full-length 43-nucleotide stem-loop IV (tet43-full) (Figure 2A) revealed substantial spectral overlap even when samples were prepared with selective isotopic labeling of either AU or GC nucleotides. The sequence of stem-loop IV is very AU rich and conformational exchange for the lower part of the structure further complicated data analysis (see below). In order to maximize the quality of the final structure, we adopted the divide-and-conquer method that represents a well-established approach to determine RNA structures by NMR (Greatorex et al, 2002; Lukavsky et al, 2003). The complete 43-nucleotide stem-loop IV was divided into two parts corresponding to the template-proximal (tet43-bot) and distal (tet43-top) parts of the stem-loop (Supplementary Figure 1). A three base-pair overlap was retained to merge the two substructures during structure calculation. In order to establish whether the excised secondary structure elements maintain the same conformation as in the complete stem-loop IV, we compared NMR chemical shifts of common residues in the three RNA constructs. Nucleotides covering the apical part of the structure have virtually identical chemical shifts as those observed for the same positions in the full-length RNA, whereas residues within the template-proximal oligonucleotide are also generally very similar to those observed in the full-length RNA (Supplementary Figure 1D). The small changes are consistent with minor differences due to the conformational exchange observed in this region of the structure (see below). Thus, the same structures are formed in each of the fragments, validating the use of the divide-and-conquer method.

The solution structure of the full stem-loop IV of Tetrahymena telomerase RNA was determined by merging distance and dihedral angle restraints obtained from each of the two substructures (and for the complete RNA) with residual dipolar coupling (RDC) restraints obtained for the full-length RNA, as described in the Materials and methods section. As shown in Table I, over 30 constraints per residue were collected. Of 100 calculated structures, 30 had no NOE violation greater than 0.5 Å or dihedral violations greater than 5°; we chose the 20 structures with the lowest constraint violation to represent the ensemble of conformations of stem-loop IV. Importantly, the inclusion of RDCs in the structure calculation improved significantly the overall precision of the structure; the r.m.s.d. for the ensemble was reduced from 3.65 to 1.96 Å. Structural statistics for the final ensemble of 20 best structures are presented in Table I and a superposition of the 10 lowest energy structures is shown in Figure 2B.

Structure of Tetrahymena telomerase RNA stem-loop IV

The RNA forms a severely kinked structure, with the apical part of the stem-loop rigidly defined and capped by a highly structured 7-nucleotide (CACUAUU) loop (Figure 2B and C). As a single-axis system and common alignment tensor could be used to refine the complete RNA structure using RDCs (see Materials and methods), the kink in stem-loop IV is a rigid distortion and not a dynamic property.

The apical loop is the most conserved region of stem-loop IV, and its conformation is very well defined by a large number of internucleotide NOEs (r.m.s.d. 0.67 Å over 20 structures, Figure 3A). The bases of C132, A133, C134 and U138 continue double helical-like stacking into the loop and a C132–U138 base pair is consistently observed during the structure calculations, with a single hydrogen bond linking C132-NH and U138-O4 (Figure 3B). We note that the structure calculations were conducted without adding any hydrogen-bonding constraint for these two nucleotides. The base-pairing scheme is similar to a CU base pair reported before in A-RNA duplex (Cruse et al, 1994), although the distance between C-N3 and U-N3H is too short (3.2 Å compared to 4.1 Å) to accommodate an additional water-mediated hydrogen-bonding interaction; it is likely that this difference simply reflects a limitation of the NMR method (water molecules cannot be directly observed). The Watson–Crick faces of the A133, C134, U135, A136 and U137 bases are all exposed to solvent (Figure 3A). In contrast, the hydrophobic faces of U137 and U138, as well as the C2 of A133 and A136, point towards each other and form a small hydrophobic patch at the center of the loop. The base of U135 at the very tip of the loop is less precisely defined by the data, yet it is completely exposed in all structures. U137 is on the opposite side of the phosphate backbone to A136, almost perpendicular to A133 and C134 and with the Watson–Crick edge facing the solvent. U138 stacks on U139 and forms a base-pairing interaction with C132. Several residues in the apical loop (U135, A136, U137 and U138) exhibited slightly increased T₁ compared to the double-helical regions (Supplementary Figure 2), indicative of increased ps-ns motion. However, only U135 at the very tip of the loop had a significantly increased T₁ value, consistent with this nucleotide having very few NOE contacts and a conformation that is not well defined in the final structural calculations. Those bases (C132, A133 and C134) that retain stacking in the loop generally have lower T₁ values compared to the rest of the loop bases.

The bulged U127 base interrupts an otherwise perfect eight base-pair helix and bends the RNA but only slightly (by about 10°). We observe two populations of bulged conformers with either U126 or U127 flipping out in different calculated structures, both consistent with the experimental constraints. When U126 is the bulged residue, a single hydrogen-bonding interaction is observed between U127-NH and A143-N1 (2.1 Å) and neither U126 nor U127 forms base stacking interactions. When U127 is bulged out, U126 stacks on U125 instead and no hydrogen-bonding interactions are observed between U126 and A143. The sugars of nucleotides A124 to U127 are all in exchange between N- and S-type conformations, supporting the observation of conformational exchange. The A143 base continues stacking with C142 and A144 in all structures. Owing to the position of A143, U126 can only partially stack on U125 when U127 is flipped out. RNAse V1 (that cleaves stacked or base paired residues) only cleaves weakly near the U127 bulge (Bhattacharyya and Blackburn, 1994), consistent with the formation of a weak base pair and conformation exchange at U127. However, deletion of the C123–U127 region only affected telomerase activity slightly and this bulge is not well conserved (Mason et al, 2003).

The GA bulge in the middle of stem-loop IV sharply kinks the entire structure (about 43°, as calculated using CURVES; Lavery and Sklenar, 1988). The A122 base remains stacked with C123, whereas G121 is completely bulged out generating the kink (Figure 3C). We emphasize that this distortion represents a rigid reorientation of the structure and not dynamic averaging, as we were able to refine the entire structure with RDCs using a single set of orientational parameters (see Materials and methods). The local bulge structure is also consistent with existing chemical and enzymatic mapping data. The bulged G121 makes the groove accessible to RNAse T1 (Bhattacharyya and Blackburn, 1994), an enzyme that cleaves single-stranded guanines. A122 is sensitive to diethylpyrocarbonate (DEPC), a chemical that modifies exposed adenines. The U117 bulged base is extrahelical and further kinks the helix axis.

To our surprise, formation of the two predicted AU base pairs between the bulged U117 and G121–A122 (A118–U149; U119–A148) could not be confirmed by either water NOESY or HNN-COSY data. As a consequence, no hydrogen bonding constraint was introduced in the structural calculation for these nucleotides: the structures presented in the paper are calculated without base-pair constraints for A118–U149 and U119–A148. However, NOESY cross-peaks within each strand are consistent with the bases remaining stacked as in A-form helices. Consistent with the structural instability of this region, correlated experiments indicate that the sugars of U119, C120, G147 and A148 are all in exchange between N- and S-type conformations (data not shown). Conformational exchange is also apparent in the behavior of nonexchangeable resonances. For example, the H5–H6 cross-peak of U149 is extremely broad at 25°C, but becomes much sharper at 35°C. Furthermore, analysis of the base imino resonances indicates that the structural instability of these base pairs also affects nearby base pairs. The imino peak of G150 is broad at low temperature (2°C), but becomes much sharper at temperatures above 10°C.

Although conformational exchange is observed in the region between bulges U117 and G121–A122, T₁ values were not increased compared to other parts of the structure (Supplementary Figure 2), suggesting that motion occurs on a time scale slower than ns but faster than s-ms. However, when RDCs are plotted against the stem-loop IV secondary structure (Supplementary Figure 3), couplings measured for the bases in the region proximal to the GA bulge (A118–U149, U119–A148) have smaller values than the rest of the structure, indicating that conformational averaging due to local flexibility is occurring (Al-Hashimi et al, 2002). A similar attenuation in RDCs is also observed for the bases in the loop region, except for the CU pair. RDC values attenuated by conformational averaging were not used in structural refinement. Nonetheless, RDCs for nucleotides close to the 5'- and 3'-ends are not attenuated, indicating that the relative orientation of the apical and lowest part of the structure is rigidly defined.

The sequence of the Tetrahymena telomerase stem-loop IV starts with 5'-AAGAC (and complementary bases at the 3'-end) (Figure 1A), whereas our oligonucleotide model starts with a GC (instead of AU) base pair to improve RNA transcription yield. When we prepared a sample starting with GAAGAC (identical to the T. thermophila sequence, except that a G instead of A is present 5' to the first AU base pair), 2D water NOESY spectra showed that the template-proximal stem region adopts the same A-form structure for both sequences (data not shown).

We reasoned that conformational exchange could perhaps be relieved by the addition of divalent metal ions, which often stabilize RNA structures (Allain and Varani, 1995; Cate et al, 1997). However, we only observed general broadening of imino resonances at increased Mg²⁺ concentration (up to 10 mM) due to nonspecific interaction of Mg²⁺ with the phosphate backbone and Mg²⁺-induced RNA aggregation, as is often observed. No new imino peaks appeared and no significant chemical shift changes occurred. As stem-loop IV stimulates nucleotide and repeat addition processivity (Lai et al, 2003), we also tested whether this structure would bind mononucleotides. However, no changes in the imino region were observed when we titrated dGTP and dTTP up to 2.0 mM.

The structure of the apical loop and of the GA bulge is important for telomerase activity

Three specific structural features have been identified for stem-loop IV: (1) the apical loop forms a well-defined structure closed by a CU base pair; (2) the GA bulge flanked by two GC base pairs kinks the entire structure, as previously proposed (Bhattacharyya and Blackburn, 1994; Sperger and Cech, 2001); (3) the two AU base pairs within the double helix proximal to the GA bulge are not formed making the local structure conformationally flexible. The residues contributing to the first two structural features are very highly conserved (Figure 2D). In order to assess the functional importance of these structural properties, we introduced five sets of nucleotide changes into the complete telomerase RNA and compared the activity of telomerase reconstituted with these mutants in enzymatic assays in vitro (Figure 4). The secondary structures of mutants Q1, D2, Q3, S4 and D5 (Figure 4) were established by selective 2'-hydroxyl acylation and primer extension (SHAPE) analysis with N-methylisatoic anhydride (NMIA), which reacts preferentially with the 2'-OH of structurally unconstrained nucleotides (Merino et al, 2005; Wilkinson et al, 2005) (Figure 5), and by NMR (2D water NOESY for the mutant RNAs; Supplementary Figure 4). The SHAPE analysis of wild-type stem-loop IV in full-length TER is consistent with the NMR structure: (1) in the heptaloop, the stacked C132–C134 nucleotides have relatively lower reactivity (weaker bands) to NMIA than U135–U138, which are much more dynamic in the structure; (2) the GA bulge is exposed and highly reactive, and the two GC base pairs next to it are well-defined (almost no reactivity to NMIA for nucleotides C120 and C123, whereas the bands of G146–G147 have the same intensity as those in the DMSO control); (3) the two AU base pairs proximal to the GA bulge have enhanced reactivity toward NMIA, indicating local conformational flexibility; (4) nucleotides U125–U127 are susceptible to modification, consistent with alternative base pairing in the region observed by NMR, whereas the stacked bases A143–U145 are not reactive to NMIA.

Consistent with previous studies (Sperger and Cech, 2001; Lai et al, 2003; Mason et al, 2003), changing the sequence of loop residues has the most profound effect on both telomerase activity and processivity. Two sets of mutations (C132A; S4) and (A133U, U137A; D5) were introduced: S4 investigates the importance of the CU base pair and D5 studies whether a second AU base pair may form and contribute to catalysis. Both mutations drastically reduce telomerase activity (<1%) and are profoundly defective in repeat addition processivity, being almost incapable of the translocation required for addition of more than one repeat (Figure 4B, right panel). Mutant S4 replaced the CU base pair with a more stable AU pair, and base pair formation (U138 imino) was clearly observable in 2D water NOESY (Supplementary Figure 4). However, a possible A133–U137 base pair that could also form in the loop (according to mFOLD) and would significantly alter its conformation is not supported by the NMR nor SHAPE data: the relative intensities of bands corresponding to bases A133 and U137 are very similar to the wild type, indicating equal reactivity in the two structures (Figure 5B). Mutant D5 switches the identities of A133 and U137, and the SHAPE analysis shows that this mutation destabilizes U138 by making this nucleotide highly susceptible to modification.

Consistent with our suggestion that the absence of the base pairs in the middle helix and the resulting conformational flexibility have a moderate functional role, changing the two AU base pairs to GC (mutant Q1) reduces activity to 40% of wild type, and a decrease of processivity (2-fold) is also observed (Figure 4). These two AU base pairs are highly conserved among Tetrahymena species, and are only replaced by more stable base pairs in Colpidium (Ye and Romero, 2002). NOE interactions linking the imino protons of the three continuous GC base pairs in mutant Q1 demonstrate that these base pairs form in the mutant (Supplementary Figure 4), a result consistent with the SHAPE analysis (Figure 5). Furthermore, abolishing these two base pairs by mutating A148 to G and U149 to C (mutant D2) has a negligible effect on telomerase activity and processivity. The SHAPE reactivity results indicate that mutant D2 contains an unpaired internal loop between U117 and U119, but the G147C120 base pair next to the GA bulge is likely to form as C120 is not susceptible to modification (Figure 5).

The GA bulge in stem-loop IV is critical for telomerase function (Sperger and Cech, 2001), and the two GC base pairs next to the GA bulge are absolutely conserved among 17 Tetrahymenine species (Ye and Romero, 2002). Replacing them with AU base pairs (mutant Q3) reduces telomerase activity 20-fold and decreases telomerase repeat addition processivity 5-fold (Figure 4). These mutations significantly change the conformation of stem-loop IV, as demonstrated by RNA SHAPE analysis (Figure 5) and NMR (Supplementary Figure 4).

A possible explanation for the loss of activity would invoke reduced binding of TERT protein to the mutant RNAs. However, when we measured protein binding directly, none of the mutants were defective in binding to full-length TERT protein (Supplementary Figure 5A), consistent with high-affinity TERT binding being provided by a different part of the RNA (Lai et al, 2001). Because a lower-affinity interaction has been reported between stem-loop IV and the N-terminal region of TERT (Lai et al, 2003; O'Connor et al, 2005), we also measured binding of the mutant stem-loop IV RNAs to a fragment of TERT encompassing amino acids 2–191 (Supplementary Figure 5B). Again, all mutants bound to this region of TERT protein as well as wild-type RNA. The interaction between this fragment of TERT and TER has been reported to be nonspecific (Jacobs et al, 2006), but a construct containing amino acids 1–195 possesses some specificity (O'Connor et al, 2005). Therefore, we cannot rule out the possibility that our mutant RNAs would be defective in binding to the slightly longer TERT fragment. Nevertheless, the protein-binding data demonstrate that these mutants are not defective in TERT recruitment.

Discussion

Stem-loop IV of T. thermophila telomerase RNA is a functionally critical structure: deletion of the complete stem-loop, mutations of conserved residues in the loop and deletion of the conserved GA bulge, all dramatically affect telomerase activity (this work and Sperger and Cech, 2001; Lai et al, 2003; Mason et al, 2003). The NMR structure presented here shows that the apical part of stem IV forms a hepta-nucleotide loop with a well-defined and unique conformation that is closed by a CU base pair. The conserved G121–A122 bulge as well as two GC base pair next to it introduces a sharp kink in the structure, whereas two base pairs predicted to form within the template-proximal region of this RNA are not present, leading to local conformational flexibility.

The structure of the apical loop is critical for telomerase activity

We were surprised to observe a structurally well-defined loop: RNA loops of this size are generally conformationally flexible. Although six of seven nucleotides in the apical loop of stem-loop IV (the exception being C134) are conserved in all Tetrahymenine ciliate species examined (Ye and Romero, 2002), mutational studies (Figure 4 and previous studies (Sperger and Cech, 2001; Mason et al, 2003)) demonstrate that the most important loop residues for telomerase activity are those near the double helical stem. In fact, whereas U135 and U137 are the two most solvent accessible nucleotides in the loop (making them ideal candidates for direct recognition by TERT or to form tertiary interactions), the U135A mutant is only half as active as wild type (Sperger and Cech, 2001). Similarly, the Watson–Crick face of nucleotides A133 and C134 is solvent exposed, yet mutants like A133U and C134G retain nearly wild-type telomerase activity (Sperger and Cech, 2001; O'Connor et al, 2005). In contrast, mutations of either C132 or U138, even to other canonical or mismatched pairs (e.g. AU (mutant S4), GU or CA (Sperger and Cech, 2001)) are not tolerated. Mutant S4 forms a stable AU base pair at the base of the heptaloop, yet is defective in both telomerase activity and processivity, suggesting that a weak base pair at the base of the loop contributes to enzyme activity. Mutant D5 demonstrates that the identity of the two bases (A133 and U137) following the CU base pair cannot be switched; SHAPE analysis demonstrate that this mutation destabilizes the CU base pair, as U138 becomes highly reactive compared to wild type (Figure 5).

A possible role for the apical loop is in binding TERT protein, as both U137 and U138 could be crosslinked to TERT (Lai et al, 2003) and mutation of these two nucleotides reduced binding of the N-terminus of TERT to stem-loop IV (O'Connor et al, 2005). However, we did not observe any significant decrease in affinity of any of our TER mutants with full-length TERT, or with the N-terminal domain of TERT protein. This result is consistent with the observation that high-affinity binding to TERT is provided by a region near the template (Figure 1A) (Lai et al, 2001). Therefore, if these mutants affect telomerase activity through an interaction with TERT, as is certainly possible, they do so by affecting the conformation of the enzyme active site and not just the affinity of the protein–RNA interaction.

Bending of the structure by the GA bulge in the middle of stem IV affects catalytic activity

The GA bulge and the two GC base pairs on either side of it are phylogenetically absolutely conserved in telomerase RNAs from 17 Tetrahymenine ciliates (Ye and Romero, 2002) (Figure 2D). Deletion of this bulge reduces telomerase activity nearly 20-fold, but its substitution with UU or CU restores activity (Autexier and Greider, 1998; Sperger and Cech, 2001): these results make it highly likely that the function of these nucleotides is structural. Bulged nucleotides often distort the helical axis of RNA helices (Bhattacharyya and Blackburn, 1994) and indeed we observe that the GA bulge opens up the major groove and causes a 40–45° kink in the double helix axis (Figure 2). The distortion is caused by A122 remaining stacked within the double helix while G121 is completely bulged out. The two conserved GC base pairs next to the GA bulge are required for the correct formation of the bulge structure, as changing them to AU pairs (mutant Q3) reduces activity 20-fold (Figure 4) and dramatically alters the central and distal stem-loop IV structure (Figure 5). This mutation is as deleterious as removing the GA bulge altogether. Deletion of the GA bulge would generate a perfectly straight and base paired helix in the middle of stem IV.

In contrast to the structurally rigid distal part of stem-loop IV, two AU base pairs (A118–U149, U119–A148) proposed to form near the GA bulge are not observed. However, the pattern of NOE interactions suggests that base stacking still exists in this region, explaining why RNAse V1 cleaves these nucleotides poorly and DEP modifies A118 and A148 inefficiently (Bhattacharyya and Blackburn, 1994; Sperger and Cech, 2001). Mutations that abolish base-pairing for the two AU pairs and open up the region between bulges U117 and U119 (mutant D2, U149C and A148G) have very little effect on enzyme activity and processivity, but rigidifying this region through the conservative introduction of two GC pairs (mutant Q1) affects activity and processivity 2–3-fold. It is quite possible that these mutations would have more significant effects in vivo by affecting the interaction of TER with p65, a holoenzyme protein that recognizes the proximal region of stem-loop IV including the GA bulge and cooperatively improves the TERT-TER affinity (O'Connor and Collins, 2006).

Role of stem-loop IV in telomere synthesis

The catalytic cycle of telomeric repeat synthesis by telomerase involves a complex set of conformational rearrangements to position the template, substrates and product into the enzyme active site, and to reposition the enzyme onto the template following each repeat addition. It was suggested that a dynamic structural rearrangement of the stem III pseudoknot pairing could be part of the process (Lai et al, 2003), and that the folding of the pseudoknot region could be influenced by interactions with stem-loop IV and with TERT (Sperger and Cech, 2001). Our results confirm that the highly structured stem IV loop closed by a unique CU base pair is functionally very important and that the combination of rigid kinking by the GA bulge and structural flexibility in the nearby proximal stem region contribute to enzymatic activity. Activity and processivity are not significantly affected by mutations such as D2 that disrupt base pair formation completely, but are reduced by mutations that rigidify the template-proximal region of the structure. We propose that the bending at the GA bulge and the flexibility in the nearby region enables the apical loop to be repositioned during the nucleotide and repeat addition process, but they may also contribute to RNP assembly. Deletion of the GA bulge strongly (30-fold) reduces p65 enhancement of TERT RNP assembly (O'Connor and Collins, 2006; Prathapam et al, 2005). However, this possibility was not investigated here because the in vitro activity assays were conducted in crude reticulocyte lysates that lack p65.

Ciliate stem-loop IVs have been proposed to be functionally analogous to the CR4–CR5 region of vertebrate telomerases (Mason et al, 2003) and to a hairpin from the telomerase RNA of the budding yeast Kluyveromyces lactis (Roy et al, 1998). Are the ciliate and human telomerase RNA structures also structurally related? A stem-loop called P6.1 within human CR4–CR5 is required for TERT binding and is critical for telomerase activity in vitro and in vivo (Mitchell and Collins, 2000). Just like the apical loop of T. thermophila stem-loop IV, loop P6.1 has a rigid and well-defined conformation stabilized by a GU wobble pair formed by two of the five unpaired loop residues (Leeper et al, 2003). It is also near a three-way junction that we have observed to be in conformational exchange by NMR (data not shown). We propose that this junction plays the same structural and functional role as the template-proximal section of stem-loop IV, by providing a flexible joint to position loop P6.1.

Materials and methods

RNA preparation

Three oligonucleotides were prepared for structural determination (Supplementary Figure 1): (1) Tet43-full, nucleotides 113–153 plus an extra GC for synthetic reasons; (2) Tet43-top, nucleotides 123–146 plus two GC base pairs; (3) Tet43-bot, nucleotides 113–125 and 144–153, capped by a CUUCGG tetraloop for stability, plus an extra GC base pair. All RNAs were synthesized by in vitro transcription using T7 RNA polymerase and synthetic DNA templates (IDT) with commercially available unlabeled or ¹³C/¹⁵N labeled nucleotides (Silantes). All RNA samples were purified as described (Price et al, 1998). Dialysis was used to bring the final buffer to 10 mM Na-phosphate (pH 6.0) with 0.1 mM EDTA. Freeze-dried RNAs were redissolved in 5% D₂O/95% H₂O or 100% D₂O. For the tet43-full, selective labeling (¹³C/¹⁵N labels only for A and C or G and U) was used to reduce spectral overlap. Samples used for RDC measurements were dialyzed into 10 mM Na-succinate (pH 6.0) and mixed with Pf1 phage (ASLA Ltd.) dialyzed against the same buffer at a final phage concentration of 12–15 mg/ml.

Data collection and analysis

Most NMR data were acquired on Bruker 500 MHz DRX or 750 MHz DMX spectrometers equipped with conventional HCN probes. Three-dimensional NOESY-HSQC experiments were collected at PNNL (Richland, WA) on a Varian 600 MHz spectrometer equipped with a cryo-probe. Data were processed with nmrPipe (Delaglio et al, 1995) and analyzed with Sparky (Goddard and Kneller).

Spectral assignments initiated by standard 2D NOESY were extended using 3D NOESY-HSQC experiments (Varani et al, 1996). Base pairs were established from the 2D water NOESY collected with Watergate water suppression and by HNN-COSY experiments (Dingley and Grzesiek, 1998; Pervushin et al, 1998). Dihedral angle restraints were obtained from 2D and 3D-TOCSY, ¹H/³¹P HETCOR and 3D HCP spectra (Varani et al, 1996). In-Phase-Anti-Phase HSQC spectra (IPAP) (Andersson et al, 1998) for RDC measurement were collected at 500 MHz and 25°C on the AC- and GU-labeled full-length samples. ¹³C T₁ of the C6 of pyrimidines and C8 of purine were recorded as a series of 2D spectra at 500 MHz with constant time acquisition, and calculated as described (Shajani and Varani, 2005).

Structure determination

NOEs involving nonexchangeable protons were classified as strong (1.8–3.2 Å), medium (2.2–4.2 Å), weak (2.5–5.5 Å) or very weak (3.0–7.0 Å) based on cross-peak intensities at 100 and 200 ms mixing times (2D NOESYs) and 120 ms (3D NOESYs). NOEs involving exchangeable protons were characterized as strong (1.8–3.4 Å), medium (1.8–4.5 Å), weak (1.8–6.0 Å) and very weak (1.8–6.5 Å) based on their NOE cross-peak intensities at 100 and 200 ms mixing times. Each experimentally determined GC base pair was constrained by six hydrogen bonding distance restraints, and each AU base pair by four. Weak base-pair planarity restraints that allow propeller twist and standard hydrogen-bonding restraints were used for unambiguously established base pairs. The ribose conformations and the other backbone torsion angles were established and restrained by using methods as described (Varani et al, 1996). Constraint statistics are shown in Table I.

RDCs were calculated using the method as described (Leeper and Varani, 2005). However, separate grid searches for D_a and R were carried out for the apical part, the proximal stem and the complete molecule, to establish whether there was significant interdomain motion. The final values of the orientation parameters were very close (within 5%) for each of the three fragments: therefore, the extent of dynamic reorientation between the top and bottom part of the RNA structure is small and a single-axis system and common alignment tensor were used to refine the final RNA full-length structure.

A single extended starting structure was generated within Xplor-NIH (Schwieters et al, 2003) and initial velocities were randomized for each of 100 distinct structures that were subjected to torsion angle dynamics and simulated annealing. RDC-derived restraints were added only after the majority of converged structures were found to be consistent with the NOE and dihedral constraints. Only RDCs obtained from the well-ordered residues, 112–116, 123–125, 128–131, 139–142, 143–146, 150–154, were used in the refinement.

Telomerase activity

A plasmid (pTet-telo) containing the Tetrahymena telomerase RNA gene was a gift from Art Zaug (Zaug and Cech, 1995). Plasmids encoding mutant telomerase RNAs were constructed by site-directed mutagenesis of pTet-telo and confirmed by sequencing. Wild-type and mutant plasmids were transcribed in vitro and the RNAs were isolated by gel-purification as described (Bryan et al, 2000).

FLAG-tagged Tetrahymena TERT (Bryan et al, 2003) was translated in a rabbit reticulocyte lysate reaction using the TnT Quick for PCR kit (Promega), and 9 l of the translation reaction were added to 1 l wild-type or mutant telomerase RNA (final concentrations of 2 and 20 nM). The reaction was incubated at 30°C for 10 min to allow for complex formation. Telomerase activity was initiated by addition of reaction buffer (Bryan et al, 2000), 2.5 M of the indicated DNA primer, 100 M dTTP and 10 M [-³²P]dGTP at 80 Ci/mmol (Perkin-Elmer). The reaction was incubated at 30°C for 60 min and then electrophoresed on a 10% polyacrylamide/8 M urea gel. A 100-mer DNA oligonucleotide labeled with T4 polynucleotide kinase and [-³²P]ATP was added to the reaction before phenol/chloroform extraction as a recovery and loading control.

Telomerase activity assays were conducted in crude reticulocyte lysates (as above) in order to circumvent quantitation problems due to varying immunoprecipitation efficiencies. However, as there are telomerase processivity inhibitors in rabbit reticulocyte lysates (Bryan et al, 2000), we also measured activity of the TER mutants using immunopurified telomerase. For these reactions, the plasmid encoding FLAG-tagged TERT was translated in the presence of 20 nM telomerase RNA in 50 l rabbit reticulocyte lysate reactions using the TnT Quick for PCR kit (Promega) at 30°C for 60 min. The telomerase complexes were then immunopurified on anti-FLAG M2 affinity gel (Sigma-Aldrich) as described (Bryan et al, 2000), with the exception that the buffers contained no Nonidet P-40. The immunoprecipitation efficiency was determined by electrophoresis of 5 l of the resulting bead slurry on an 8% SDS–PAGE gel, and an equal amount of precipitated protein was included in telomerase activity assays as described above, with the exception that these reactions contained 1 M each of DNA primer. The relative activities of wild-type and mutant telomerases did not differ significantly between crude and purified telomerase.

Relative activity levels were quantified using ImageQuant software (GE). The total intensity of bands in each lane was normalized to the intensity of the labeled 100-mer oligonucleotide used as a loading control. Repeat addition processivity of the reactions using primer (G₄T₂)₃ was quantified as described (Bryan et al, 2000).

Binding of telomerase RNA to TERT protein

Available as Supplementary data at The EMBO Journal online.

RNA SHAPE analysis

RNAs for SHAPE analysis were synthesized with wild-type 5' ends, a 3' linker and 3' RT primer-binding site from templates generated by PCR (Merino et al, 2005; Wilkinson et al, 2005). The 43 nucleotide 3' extension has been shown to fold independently of TER (unreported data). RNA (1 pmol) was snap annealed in 7 l of deionized water at 95°C for 2 min and placed on ice for 5 min before 2 l of 5 TER Hit Buffer (250 mM Hepes pH 8.0, 10 mM MgCl₂) was added. The RNA solution was then incubated at 30°C for 5 min, treated immediately with 1 l of 100 mM N-methylisatoic anhydride (NMIA-Molecular Probes) in DMSO or DMSO only as a control, incubated at 30°C for 90 min, precipitated with ethanol in the presence of 0.2 M NaCl and 200 g/ml glycogen, and reconstituted in 5 l of TE (pH 8.0). Hit RNA (5 l) was mapped by reverse transcription using 5'-³²P-labeled DNA primer (1 pmol, 5'-GAACCGGACCGAAGCCCG). The primer was annealed by heating to 95°C for 1 min, 65°C for 6 min, 35°C for 10 min, and on ice for 5 min followed by the addition of 2 l of 5 First-Strand Buffer (Invitrogen) reverse transcription buffer (250 mM Tris-Cl pH 8.3, 375 mM KCl, 15 mM MgCl₂), 0.5 l 10 mM dNTP mix and 0.5 l 100 mM DTT. The solution was heated to 52°C for 1 min, Superscript III reverse transcriptase (100 units; Invitrogen) was immediately added, and allowed to extend for exactly 3 min at 52°C. The reaction was quenched by the addition of 2.5 l of 1 M NaOH, heated at 95°C for 5 min, neutralized by the addition of 2.5 l of 1 M HCl, ethanol precipitated and resuspended in 5 l denaturing formamide loading buffer (75% formamide, 45 mM Tris/borate, 5 mM EDTA, 0.01% bromophenol blue and xylene cyanol). Dideoxythymidine sequencing ladders were generated by incorporating 0.5 mM ddTTP in the reverse transcription reaction of unmodified TERs. The radiolabeled extension products were separated by electrophoresis on 8% denaturing sequencing gels and visualized by phosphorimaging using ImageQuant 5.1. Individual band intensities of NMIA and DMSO lanes were integrated using SAFA (Das et al, 2005).

Accession codes

Coordinates have been deposited in the Protein Data Bank with the following accession code 2FEY.

Supplementary data

Supplementary data are available at The EMBO Journal Online.

Acknowledgements

We thank EMSL-PNNL for access to NMR instruments and Dr Nancy Isern (PNNL) for help with data acquisition; Dr Thomas Leeper for assistance with the NMR spectroscopy and structural calculation; Ms Zahra Shajani for help with the T₁ measurements; Julie Jurczyluk for technical assistance on the telomerase activity assay; Drs Steven Jacobs, Elaine Podell, Art Zaug and Thomas Cech for the kind gifts of the TEN protein and TER plasmid. This work was supported in part by a grant from the NSF to MBJ, the Wellcome Trust (TMB, Senior Research Fellowship GR066727MA) and the NIH-NCI and the Human Frontier of Science to GV.

References

Al-Hashimi HM , Gosser Y , Gorin A , Hu W , Majumdar A , Patel DJ (2002) Concerted motions in HIV-1 TAR RNA may allow access to bound state conformations: RNA dynamics from NMR residual dipolar couplings. J Mol Biol 315: 95–102  | Article | PubMed | ChemPort |

Allain FHT , Varani G (1995) Divalent metal ion binding to a conserved wobble pair defining the upstream site of cleavage of group I self-splicing introns. Nucleic Acids Res 23: 341–350  | PubMed | ISI | ChemPort |

Andersson P , Nordstrand K , Sunnerhagen M , Liepinsh E , Turovskis I , Otting G (1998) Heteronuclear correlation experiments for the determination of one-bond coupling constants. J Biomol NMR 11: 445–450  | Article | PubMed | ChemPort |

Autexier C , Greider CW (1998) Mutational analysis of the Tetrahymena telomerase RNA: identification of residues affecting telomerase activity in vitro. Nucleic Acids Res 26: 787–795  | Article | PubMed | ISI | ChemPort |

Bhattacharyya A , Blackburn EH (1994) Architecture of telomerase RNA. EMBO J 23: 5721–5731

Blasco MA , Gasser SM , Lingner J (1999) Telomeres and telomerase. Genes Dev 13: 2353–2359  | Article | PubMed | ChemPort |

Blasco MA , Lee HW , Hande MP , Samper E , Lansdorp PM , DePinho RA , Greider CW (1997) Telomere shortening and tumor formation by mouse cells Lacking telomerase RNA. Cell 91: 25–34  | Article | PubMed | ISI | ChemPort |

Bryan TM , Goodrich KJ , Cech TR (2000) A mutant of Tetrahymena telomerase reverse transcriptase with increased processivity. J Biol Chem 275: 24199–24207  | Article | PubMed | ISI | ChemPort |

Bryan TM , Goodrich KJ , Cech TR (2003) Tetrahymena telomerase is active as a monomer. Mol Biol Cell 14: 4794–4804  | Article | PubMed | ISI | ChemPort |

Cate JH , Hanna RL , Doudna JA (1997) A magnesium ion core at the heart of a ribozyme domain. Nat Struct Biol 4: 553–558  | Article | PubMed | ISI | ChemPort |

Chen JL , Blasco MA , Greider CW (2000) Secondary Structure of vertebrate telomerase RNA. Cell 100: 503–514  | Article | PubMed | ISI | ChemPort |

Chen JL , Keyer Opperman K , Greider CW (2002) A critical stem-loop structure in the CR4–CR5 domain of mammalian telomerase RNA. Nucleic Acids Res 30: 592–597  | Article | PubMed | ISI | ChemPort |

Cruse WBT , Saludijan P , Biala E , Strazewski P , Prangé T , Kennard O (1994) Structure of a mispaired RNA double helix at 1.6 Å resolution and implications for the prediction of RNA secondary structure. Proc Natl Acad Sci USA 91: 4160–4164  | Article | PubMed | ChemPort |

Das R , Laederach A , Pearlman SM , Herschlag D , Altman RB (2005) SAFA: semi-automated footprinting analysis software for high-throughput quantification of nucleic acid footprinting experiments. RNA 11: 344–354  | Article | PubMed | ISI | ChemPort |

Delaglio F , Grzesiek S , Vuister GW , Zhu G , Pfeifer J , Bax A (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR 6: 277–293  | Article | PubMed | ISI | ChemPort |

de Lange T , Jacks T (1999) For better or worse? Telomerase inhibition and cancer. Cell 98: 273–275  | Article | PubMed | ISI | ChemPort |

De Lano WL (2002) The PyMOL Molecular Graphics System. San Carlos USA: Delano Scientific

Dingley AJ , Grzesiek S (1998) Direct observation of hydrogen bonds in nucleic acid base pairs by internucleotide ²J_NN couplings. J Am Chem Soc 120: 8293–8297  | Article | ISI | ChemPort |

Feng J , Funk WD , Wang SS , Weinrich SL , Avilion AA , Chiu CP , Adams RR , Chang E , Allsopp RC , Yu Y , Le S , West MD , Harley CB , Andrews WH , Greider CW , Villeponteau B (1995) The RNA component of human telomerase. Science 269: 1236–1241  | PubMed | ISI | ChemPort |

Goddard TD , Kneller DG (2003) Sparky 3. San Francisco: University of California

Greatorex J , Gallego J , Varani G , Lever A (2002) Structure and stability of wild-type and mutant RNA internal loops from the SL-1 domain of the HIV-1 packaging signal. J Mol Biol 322: 543–557  | Article | PubMed | ChemPort |

Jacobs SA , Podell ER , Cech TR (2006) Crystal structure of the essential N-terminal domain of telomerase reverse transcriptase. Nat Struct Mol Biol 13: 218–225  | Article | PubMed | ChemPort |

Koradi R , Billeter M , Wüthrich K (1996) MOLMOL: a program for display and analysis of macromolecular structures. J Mol Graph 14: 51–55  | Article | PubMed | ISI | ChemPort |

Lai CK , Miller MC , Collins K (2003) Roles for RNA in telomerase nucleotide and repeat addition processivity. Mol Cell 11: 1673–1683  | Article | PubMed | ChemPort |

Lai CK , Mitchell JR , Collins K (2001) RNA binding domain of telomerase reverse transcriptase. Mol Cell Biol 21: 990–1000  | Article | PubMed | ISI | ChemPort |

Lavery R , Sklenar H (1988) The definition of generalized helicoidal parameters and of axis curvature for irregular nucleic acids. J Biomol Struct Dyn 6: 63–91  | PubMed | ISI | ChemPort |

Lee HW , Blasco MA , Gottlieb GJ , Horner JWI , Greider CW , DePinho RA (1998) Essential role of mouse telomerase in highly proliferative organs. Nature 392: 569–574  | Article | PubMed | ISI | ChemPort |

Leeper T , Leulliot N , Varani G (2003) The solution structure of an essential stem-loop of human telomerase RNA. Nucleic Acids Res 31: 2614–2621  | Article | PubMed | ChemPort |

Leeper TC , Varani G (2005) The structure of an enzyme-activating fragment of human telomerase RNA. RNA 11: 394–403  | Article | PubMed | ChemPort |

Lingner J , Cech TR (1998) Telomerase and chromosome end maintenance. Curr Opin Gen Dev 8: 226–232  | ChemPort |

Lingner J , Hendrick LL , Cech TR (1994) Telomerase RNAs of different ciliates have a common secondary structure and a permuted template. Genes Dev 8: 1984–1998  | PubMed | ISI | ChemPort |

Lingner J , Hughes TR , Shevchenko A , Mann M , Lundblad V , Cech TR (1997) Reverse transcriptase motifs in the catalytic subunit of telomerase. Science 276: 561–567  | Article | PubMed | ISI | ChemPort |

Lukavsky PJ , Kim I , Otto GA , Puglisi JD (2003) Structure of HCV IRES domain II determined by NMR. Nat Struct Biol 10: 1033–1038  | Article | PubMed | ISI | ChemPort |

Mason DX , Goneska E , Greider CW (2003) Stem-loop IV of Tetrahymena telomerase RNA stimulates processivity in trans. Mol Cell Biol 23: 5606–5613  | Article | PubMed | ChemPort |

Merino EJ , Wilkinson KA , Coughlan JL , Weeks KM (2005) RNA structure analysis at single nucleotide resolution by selective 2'-hydroxyl acylation and primer extension (SHAPE). J Am Chem Soc 127: 4223–4231  | Article | PubMed | ChemPort |

Mitchell JR , Cheng J , Collins K (1999) A Box H/ACA small nucleolar RNA-like domain at the human telomerase RNA 3' end. Mol Cell Biol 19: 567–576  | PubMed | ISI | ChemPort |

Mitchell JR , Collins K (2000) Human telomerase action requires two independent interactions between telomerase RNA and telomerase reverse transcriptase. Mol Cell 6: 361–371  | Article | PubMed | ISI | ChemPort |

O'Connor CM , Collins K (2006) A novel RNA binding domain in Tetrahymena telomerase p65 initiates hierarchical assembly of Telomerase holoenzyme. Mol Cell Biol 26: 2029–2036  | Article | PubMed | ChemPort |

O'Connor CM , Lai CK , Collins K (2005) Two purified domains of telomerase reverse transcriptase reconstitute sequence-specific interactions with RNA. J Biol Chem 280: 17533–17539  | Article | PubMed | ChemPort |

Pervushin K , Ono A , Fernandez C , Szyperski T , Kainosho M , Wüthrich K (1998) NMR scalar coupling across Watson–Crick base pair hydrogen bonds in DNA observed by transverse relaxation-optimized spectroscopy. Proc Natl Acad Sci USA 95: 14147–14151  | Article |